xref: /netbsd-src/sys/netinet/in.c (revision 6cf6fe02a981b55727c49c3d37b0d8191a98c0ee)
1 /*	$NetBSD: in.c,v 1.148 2014/09/09 20:16:12 rmind Exp $	*/
2 
3 /*
4  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the project nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  */
31 
32 /*-
33  * Copyright (c) 1998 The NetBSD Foundation, Inc.
34  * All rights reserved.
35  *
36  * This code is derived from software contributed to The NetBSD Foundation
37  * by Public Access Networks Corporation ("Panix").  It was developed under
38  * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon.
39  *
40  * Redistribution and use in source and binary forms, with or without
41  * modification, are permitted provided that the following conditions
42  * are met:
43  * 1. Redistributions of source code must retain the above copyright
44  *    notice, this list of conditions and the following disclaimer.
45  * 2. Redistributions in binary form must reproduce the above copyright
46  *    notice, this list of conditions and the following disclaimer in the
47  *    documentation and/or other materials provided with the distribution.
48  *
49  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
50  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
51  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
52  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
53  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
54  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
55  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
56  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
57  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
58  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
59  * POSSIBILITY OF SUCH DAMAGE.
60  */
61 
62 /*
63  * Copyright (c) 1982, 1986, 1991, 1993
64  *	The Regents of the University of California.  All rights reserved.
65  *
66  * Redistribution and use in source and binary forms, with or without
67  * modification, are permitted provided that the following conditions
68  * are met:
69  * 1. Redistributions of source code must retain the above copyright
70  *    notice, this list of conditions and the following disclaimer.
71  * 2. Redistributions in binary form must reproduce the above copyright
72  *    notice, this list of conditions and the following disclaimer in the
73  *    documentation and/or other materials provided with the distribution.
74  * 3. Neither the name of the University nor the names of its contributors
75  *    may be used to endorse or promote products derived from this software
76  *    without specific prior written permission.
77  *
78  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
79  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
80  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
81  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
82  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
83  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
84  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
85  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
86  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
87  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
88  * SUCH DAMAGE.
89  *
90  *	@(#)in.c	8.4 (Berkeley) 1/9/95
91  */
92 
93 #include <sys/cdefs.h>
94 __KERNEL_RCSID(0, "$NetBSD: in.c,v 1.148 2014/09/09 20:16:12 rmind Exp $");
95 
96 #include "opt_inet.h"
97 #include "opt_inet_conf.h"
98 #include "opt_mrouting.h"
99 
100 #include <sys/param.h>
101 #include <sys/ioctl.h>
102 #include <sys/errno.h>
103 #include <sys/malloc.h>
104 #include <sys/socket.h>
105 #include <sys/socketvar.h>
106 #include <sys/sysctl.h>
107 #include <sys/systm.h>
108 #include <sys/proc.h>
109 #include <sys/syslog.h>
110 #include <sys/kauth.h>
111 
112 #include <sys/cprng.h>
113 
114 #include <net/if.h>
115 #include <net/route.h>
116 #include <net/pfil.h>
117 
118 #include <net/if_ether.h>
119 
120 #include <netinet/in_systm.h>
121 #include <netinet/in.h>
122 #include <netinet/in_var.h>
123 #include <netinet/ip.h>
124 #include <netinet/ip_var.h>
125 #include <netinet/in_ifattach.h>
126 #include <netinet/in_pcb.h>
127 #include <netinet/if_inarp.h>
128 #include <netinet/ip_mroute.h>
129 #include <netinet/igmp_var.h>
130 
131 #ifdef IPSELSRC
132 #include <netinet/in_selsrc.h>
133 #endif
134 
135 static u_int	in_mask2len(struct in_addr *);
136 static void	in_len2mask(struct in_addr *, u_int);
137 static int	in_lifaddr_ioctl(struct socket *, u_long, void *,
138 	struct ifnet *);
139 
140 static int	in_addprefix(struct in_ifaddr *, int);
141 static int	in_scrubprefix(struct in_ifaddr *);
142 static void	in_sysctl_init(struct sysctllog **);
143 
144 #ifndef SUBNETSARELOCAL
145 #define	SUBNETSARELOCAL	1
146 #endif
147 
148 #ifndef HOSTZEROBROADCAST
149 #define HOSTZEROBROADCAST 1
150 #endif
151 
152 /* Note: 61, 127, 251, 509, 1021, 2039 are good. */
153 #ifndef IN_MULTI_HASH_SIZE
154 #define IN_MULTI_HASH_SIZE	509
155 #endif
156 
157 static int			subnetsarelocal = SUBNETSARELOCAL;
158 static int			hostzeroisbroadcast = HOSTZEROBROADCAST;
159 
160 /*
161  * This list is used to keep track of in_multi chains which belong to
162  * deleted interface addresses.  We use in_ifaddr so that a chain head
163  * won't be deallocated until all multicast address record are deleted.
164  */
165 
166 LIST_HEAD(in_multihashhead, in_multi);		/* Type of the hash head */
167 
168 static struct pool		inmulti_pool;
169 static u_int			in_multientries;
170 static struct in_multihashhead *in_multihashtbl;
171 static u_long			in_multihash;
172 static krwlock_t		in_multilock;
173 
174 #define IN_MULTI_HASH(x, ifp) \
175     (in_multihashtbl[(u_long)((x) ^ (ifp->if_index)) % IN_MULTI_HASH_SIZE])
176 
177 struct in_ifaddrhashhead *	in_ifaddrhashtbl;
178 u_long				in_ifaddrhash;
179 struct in_ifaddrhead		in_ifaddrhead;
180 
181 void
182 in_init(void)
183 {
184 	pool_init(&inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl",
185 	    NULL, IPL_SOFTNET);
186 	TAILQ_INIT(&in_ifaddrhead);
187 
188 	in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true,
189 	    &in_ifaddrhash);
190 	in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true,
191 	    &in_multihash);
192 	rw_init(&in_multilock);
193 
194 	in_sysctl_init(NULL);
195 }
196 
197 /*
198  * Return 1 if an internet address is for a ``local'' host
199  * (one to which we have a connection).  If subnetsarelocal
200  * is true, this includes other subnets of the local net.
201  * Otherwise, it includes only the directly-connected (sub)nets.
202  */
203 int
204 in_localaddr(struct in_addr in)
205 {
206 	struct in_ifaddr *ia;
207 
208 	if (subnetsarelocal) {
209 		TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list)
210 			if ((in.s_addr & ia->ia_netmask) == ia->ia_net)
211 				return (1);
212 	} else {
213 		TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list)
214 			if ((in.s_addr & ia->ia_subnetmask) == ia->ia_subnet)
215 				return (1);
216 	}
217 	return (0);
218 }
219 
220 /*
221  * Determine whether an IP address is in a reserved set of addresses
222  * that may not be forwarded, or whether datagrams to that destination
223  * may be forwarded.
224  */
225 int
226 in_canforward(struct in_addr in)
227 {
228 	u_int32_t net;
229 
230 	if (IN_EXPERIMENTAL(in.s_addr) || IN_MULTICAST(in.s_addr))
231 		return (0);
232 	if (IN_CLASSA(in.s_addr)) {
233 		net = in.s_addr & IN_CLASSA_NET;
234 		if (net == 0 || net == htonl(IN_LOOPBACKNET << IN_CLASSA_NSHIFT))
235 			return (0);
236 	}
237 	return (1);
238 }
239 
240 /*
241  * Trim a mask in a sockaddr
242  */
243 void
244 in_socktrim(struct sockaddr_in *ap)
245 {
246 	char *cplim = (char *) &ap->sin_addr;
247 	char *cp = (char *) (&ap->sin_addr + 1);
248 
249 	ap->sin_len = 0;
250 	while (--cp >= cplim)
251 		if (*cp) {
252 			(ap)->sin_len = cp - (char *) (ap) + 1;
253 			break;
254 		}
255 }
256 
257 /*
258  *  Routine to take an Internet address and convert into a
259  *  "dotted quad" representation for printing.
260  */
261 const char *
262 in_fmtaddr(struct in_addr addr)
263 {
264 	static char buf[sizeof("123.456.789.123")];
265 
266 	addr.s_addr = ntohl(addr.s_addr);
267 
268 	snprintf(buf, sizeof(buf), "%d.%d.%d.%d",
269 		(addr.s_addr >> 24) & 0xFF,
270 		(addr.s_addr >> 16) & 0xFF,
271 		(addr.s_addr >>  8) & 0xFF,
272 		(addr.s_addr >>  0) & 0xFF);
273 	return buf;
274 }
275 
276 /*
277  * Maintain the "in_maxmtu" variable, which is the largest
278  * mtu for non-local interfaces with AF_INET addresses assigned
279  * to them that are up.
280  */
281 unsigned long in_maxmtu;
282 
283 void
284 in_setmaxmtu(void)
285 {
286 	struct in_ifaddr *ia;
287 	struct ifnet *ifp;
288 	unsigned long maxmtu = 0;
289 
290 	TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) {
291 		if ((ifp = ia->ia_ifp) == 0)
292 			continue;
293 		if ((ifp->if_flags & (IFF_UP|IFF_LOOPBACK)) != IFF_UP)
294 			continue;
295 		if (ifp->if_mtu > maxmtu)
296 			maxmtu = ifp->if_mtu;
297 	}
298 	if (maxmtu)
299 		in_maxmtu = maxmtu;
300 }
301 
302 static u_int
303 in_mask2len(struct in_addr *mask)
304 {
305 	u_int x, y;
306 	u_char *p;
307 
308 	p = (u_char *)mask;
309 	for (x = 0; x < sizeof(*mask); x++) {
310 		if (p[x] != 0xff)
311 			break;
312 	}
313 	y = 0;
314 	if (x < sizeof(*mask)) {
315 		for (y = 0; y < NBBY; y++) {
316 			if ((p[x] & (0x80 >> y)) == 0)
317 				break;
318 		}
319 	}
320 	return x * NBBY + y;
321 }
322 
323 static void
324 in_len2mask(struct in_addr *mask, u_int len)
325 {
326 	u_int i;
327 	u_char *p;
328 
329 	p = (u_char *)mask;
330 	memset(mask, 0, sizeof(*mask));
331 	for (i = 0; i < len / NBBY; i++)
332 		p[i] = 0xff;
333 	if (len % NBBY)
334 		p[i] = (0xff00 >> (len % NBBY)) & 0xff;
335 }
336 
337 /*
338  * Generic internet control operations (ioctl's).
339  * Ifp is 0 if not an interface-specific ioctl.
340  */
341 /* ARGSUSED */
342 int
343 in_control(struct socket *so, u_long cmd, void *data, struct ifnet *ifp)
344 {
345 	struct ifreq *ifr = (struct ifreq *)data;
346 	struct in_ifaddr *ia = NULL;
347 	struct in_aliasreq *ifra = (struct in_aliasreq *)data;
348 	struct sockaddr_in oldaddr;
349 	int error, hostIsNew, maskIsNew;
350 	int newifaddr = 0;
351 
352 	switch (cmd) {
353 	case SIOCALIFADDR:
354 	case SIOCDLIFADDR:
355 	case SIOCGLIFADDR:
356 		if (ifp == NULL)
357 			return EINVAL;
358 		return in_lifaddr_ioctl(so, cmd, data, ifp);
359 	case SIOCGIFADDRPREF:
360 	case SIOCSIFADDRPREF:
361 		if (ifp == NULL)
362 			return EINVAL;
363 		return ifaddrpref_ioctl(so, cmd, data, ifp);
364 	}
365 
366 	/*
367 	 * Find address for this interface, if it exists.
368 	 */
369 	if (ifp != NULL)
370 		IFP_TO_IA(ifp, ia);
371 
372 	switch (cmd) {
373 	case SIOCAIFADDR:
374 	case SIOCDIFADDR:
375 	case SIOCGIFALIAS:
376 		if (ifra->ifra_addr.sin_family == AF_INET)
377 			LIST_FOREACH(ia,
378 			    &IN_IFADDR_HASH(ifra->ifra_addr.sin_addr.s_addr),
379 			    ia_hash) {
380 				if (ia->ia_ifp == ifp &&
381 				    in_hosteq(ia->ia_addr.sin_addr,
382 				    ifra->ifra_addr.sin_addr))
383 					break;
384 			}
385 		if ((cmd == SIOCDIFADDR || cmd == SIOCGIFALIAS) && ia == NULL)
386 			return (EADDRNOTAVAIL);
387 
388 		if (cmd == SIOCDIFADDR &&
389 		    ifra->ifra_addr.sin_family == AF_UNSPEC) {
390 			ifra->ifra_addr.sin_family = AF_INET;
391 		}
392 		/* FALLTHROUGH */
393 	case SIOCSIFADDR:
394 	case SIOCSIFDSTADDR:
395 		if (ifra->ifra_addr.sin_family != AF_INET)
396 			return (EAFNOSUPPORT);
397 		/* FALLTHROUGH */
398 	case SIOCSIFNETMASK:
399 		if (ifp == NULL)
400 			panic("in_control");
401 
402 		if (cmd == SIOCGIFALIAS)
403 			break;
404 
405 		if (ia == NULL &&
406 		    (cmd == SIOCSIFNETMASK || cmd == SIOCSIFDSTADDR))
407 			return (EADDRNOTAVAIL);
408 
409 		if (kauth_authorize_network(curlwp->l_cred, KAUTH_NETWORK_INTERFACE,
410 		    KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp, (void *)cmd,
411 		    NULL) != 0)
412 			return (EPERM);
413 
414 		if (ia == NULL) {
415 			ia = malloc(sizeof(*ia), M_IFADDR, M_WAITOK|M_ZERO);
416 			if (ia == NULL)
417 				return (ENOBUFS);
418 			TAILQ_INSERT_TAIL(&in_ifaddrhead, ia, ia_list);
419 			ifaref(&ia->ia_ifa);
420 			ifa_insert(ifp, &ia->ia_ifa);
421 			ia->ia_ifa.ifa_addr = sintosa(&ia->ia_addr);
422 			ia->ia_ifa.ifa_dstaddr = sintosa(&ia->ia_dstaddr);
423 			ia->ia_ifa.ifa_netmask = sintosa(&ia->ia_sockmask);
424 #ifdef IPSELSRC
425 			ia->ia_ifa.ifa_getifa = in_getifa;
426 #else /* IPSELSRC */
427 			ia->ia_ifa.ifa_getifa = NULL;
428 #endif /* IPSELSRC */
429 			ia->ia_sockmask.sin_len = 8;
430 			if (ifp->if_flags & IFF_BROADCAST) {
431 				ia->ia_broadaddr.sin_len = sizeof(ia->ia_addr);
432 				ia->ia_broadaddr.sin_family = AF_INET;
433 			}
434 			ia->ia_ifp = ifp;
435 			ia->ia_idsalt = cprng_fast32() % 65535;
436 			LIST_INIT(&ia->ia_multiaddrs);
437 			newifaddr = 1;
438 		}
439 		break;
440 
441 	case SIOCSIFBRDADDR:
442 		if (kauth_authorize_network(curlwp->l_cred, KAUTH_NETWORK_INTERFACE,
443 		    KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp, (void *)cmd,
444 		    NULL) != 0)
445 			return (EPERM);
446 		/* FALLTHROUGH */
447 
448 	case SIOCGIFADDR:
449 	case SIOCGIFNETMASK:
450 	case SIOCGIFDSTADDR:
451 	case SIOCGIFBRDADDR:
452 		if (ia == NULL)
453 			return (EADDRNOTAVAIL);
454 		break;
455 	}
456 	error = 0;
457 	switch (cmd) {
458 
459 	case SIOCGIFADDR:
460 		ifreq_setaddr(cmd, ifr, sintocsa(&ia->ia_addr));
461 		break;
462 
463 	case SIOCGIFBRDADDR:
464 		if ((ifp->if_flags & IFF_BROADCAST) == 0)
465 			return (EINVAL);
466 		ifreq_setdstaddr(cmd, ifr, sintocsa(&ia->ia_broadaddr));
467 		break;
468 
469 	case SIOCGIFDSTADDR:
470 		if ((ifp->if_flags & IFF_POINTOPOINT) == 0)
471 			return (EINVAL);
472 		ifreq_setdstaddr(cmd, ifr, sintocsa(&ia->ia_dstaddr));
473 		break;
474 
475 	case SIOCGIFNETMASK:
476 		ifreq_setaddr(cmd, ifr, sintocsa(&ia->ia_sockmask));
477 		break;
478 
479 	case SIOCSIFDSTADDR:
480 		if ((ifp->if_flags & IFF_POINTOPOINT) == 0)
481 			return (EINVAL);
482 		oldaddr = ia->ia_dstaddr;
483 		ia->ia_dstaddr = *satocsin(ifreq_getdstaddr(cmd, ifr));
484 		if ((error = if_addr_init(ifp, &ia->ia_ifa, false)) != 0) {
485 			ia->ia_dstaddr = oldaddr;
486 			return error;
487 		}
488 		if (ia->ia_flags & IFA_ROUTE) {
489 			ia->ia_ifa.ifa_dstaddr = sintosa(&oldaddr);
490 			rtinit(&ia->ia_ifa, RTM_DELETE, RTF_HOST);
491 			ia->ia_ifa.ifa_dstaddr = sintosa(&ia->ia_dstaddr);
492 			rtinit(&ia->ia_ifa, RTM_ADD, RTF_HOST|RTF_UP);
493 		}
494 		break;
495 
496 	case SIOCSIFBRDADDR:
497 		if ((ifp->if_flags & IFF_BROADCAST) == 0)
498 			return EINVAL;
499 		ia->ia_broadaddr = *satocsin(ifreq_getbroadaddr(cmd, ifr));
500 		break;
501 
502 	case SIOCSIFADDR:
503 		error = in_ifinit(ifp, ia, satocsin(ifreq_getaddr(cmd, ifr)),
504 		    1);
505 		if (error == 0) {
506 			(void)pfil_run_hooks(if_pfil,
507 			    (struct mbuf **)SIOCSIFADDR, ifp, PFIL_IFADDR);
508 		}
509 		break;
510 
511 	case SIOCSIFNETMASK:
512 		in_ifscrub(ifp, ia);
513 		ia->ia_sockmask = *satocsin(ifreq_getaddr(cmd, ifr));
514 		ia->ia_subnetmask = ia->ia_sockmask.sin_addr.s_addr;
515 		error = in_ifinit(ifp, ia, NULL, 0);
516 		break;
517 
518 	case SIOCAIFADDR:
519 		maskIsNew = 0;
520 		hostIsNew = 1;
521 		if (ia->ia_addr.sin_family != AF_INET)
522 			;
523 		else if (ifra->ifra_addr.sin_len == 0) {
524 			ifra->ifra_addr = ia->ia_addr;
525 			hostIsNew = 0;
526 		} else if (in_hosteq(ia->ia_addr.sin_addr,
527 		           ifra->ifra_addr.sin_addr))
528 			hostIsNew = 0;
529 		if (ifra->ifra_mask.sin_len) {
530 			/* Only scrub if we control the prefix route,
531 			 * otherwise userland gets a bogus message */
532 			if ((ia->ia_flags & IFA_ROUTE))
533 				in_ifscrub(ifp, ia);
534 			ia->ia_sockmask = ifra->ifra_mask;
535 			ia->ia_subnetmask = ia->ia_sockmask.sin_addr.s_addr;
536 			maskIsNew = 1;
537 		}
538 		if ((ifp->if_flags & IFF_POINTOPOINT) &&
539 		    (ifra->ifra_dstaddr.sin_family == AF_INET)) {
540 			/* Only scrub if we control the prefix route,
541 			 * otherwise userland gets a bogus message */
542 			if ((ia->ia_flags & IFA_ROUTE))
543 				in_ifscrub(ifp, ia);
544 			ia->ia_dstaddr = ifra->ifra_dstaddr;
545 			maskIsNew  = 1; /* We lie; but the effect's the same */
546 		}
547 		if (ifra->ifra_addr.sin_family == AF_INET &&
548 		    (hostIsNew || maskIsNew)) {
549 			error = in_ifinit(ifp, ia, &ifra->ifra_addr, 0);
550 		}
551 		if ((ifp->if_flags & IFF_BROADCAST) &&
552 		    (ifra->ifra_broadaddr.sin_family == AF_INET))
553 			ia->ia_broadaddr = ifra->ifra_broadaddr;
554 		if (error == 0)
555 			(void)pfil_run_hooks(if_pfil,
556 			    (struct mbuf **)SIOCAIFADDR, ifp, PFIL_IFADDR);
557 		break;
558 
559 	case SIOCGIFALIAS:
560 		ifra->ifra_mask = ia->ia_sockmask;
561 		if ((ifp->if_flags & IFF_POINTOPOINT) &&
562 		    (ia->ia_dstaddr.sin_family == AF_INET))
563 			ifra->ifra_dstaddr = ia->ia_dstaddr;
564 		else if ((ifp->if_flags & IFF_BROADCAST) &&
565 		    (ia->ia_broadaddr.sin_family == AF_INET))
566 			ifra->ifra_broadaddr = ia->ia_broadaddr;
567 		else
568 			memset(&ifra->ifra_broadaddr, 0,
569 			      sizeof(ifra->ifra_broadaddr));
570 		break;
571 
572 	case SIOCDIFADDR:
573 		in_purgeaddr(&ia->ia_ifa);
574 		(void)pfil_run_hooks(if_pfil, (struct mbuf **)SIOCDIFADDR,
575 		    ifp, PFIL_IFADDR);
576 		break;
577 
578 #ifdef MROUTING
579 	case SIOCGETVIFCNT:
580 	case SIOCGETSGCNT:
581 		error = mrt_ioctl(so, cmd, data);
582 		break;
583 #endif /* MROUTING */
584 
585 	default:
586 		return ENOTTY;
587 	}
588 
589 	if (error != 0 && newifaddr) {
590 		KASSERT(ia != NULL);
591 		in_purgeaddr(&ia->ia_ifa);
592 	}
593 
594 	return error;
595 }
596 
597 void
598 in_purgeaddr(struct ifaddr *ifa)
599 {
600 	struct ifnet *ifp = ifa->ifa_ifp;
601 	struct in_ifaddr *ia = (void *) ifa;
602 
603 	in_ifscrub(ifp, ia);
604 	LIST_REMOVE(ia, ia_hash);
605 	ifa_remove(ifp, &ia->ia_ifa);
606 	TAILQ_REMOVE(&in_ifaddrhead, ia, ia_list);
607 	if (ia->ia_allhosts != NULL)
608 		in_delmulti(ia->ia_allhosts);
609 	ifafree(&ia->ia_ifa);
610 	in_setmaxmtu();
611 }
612 
613 void
614 in_purgeif(struct ifnet *ifp)		/* MUST be called at splsoftnet() */
615 {
616 	if_purgeaddrs(ifp, AF_INET, in_purgeaddr);
617 	igmp_purgeif(ifp);		/* manipulates pools */
618 #ifdef MROUTING
619 	ip_mrouter_detach(ifp);
620 #endif
621 }
622 
623 /*
624  * SIOC[GAD]LIFADDR.
625  *	SIOCGLIFADDR: get first address. (???)
626  *	SIOCGLIFADDR with IFLR_PREFIX:
627  *		get first address that matches the specified prefix.
628  *	SIOCALIFADDR: add the specified address.
629  *	SIOCALIFADDR with IFLR_PREFIX:
630  *		EINVAL since we can't deduce hostid part of the address.
631  *	SIOCDLIFADDR: delete the specified address.
632  *	SIOCDLIFADDR with IFLR_PREFIX:
633  *		delete the first address that matches the specified prefix.
634  * return values:
635  *	EINVAL on invalid parameters
636  *	EADDRNOTAVAIL on prefix match failed/specified address not found
637  *	other values may be returned from in_ioctl()
638  */
639 static int
640 in_lifaddr_ioctl(struct socket *so, u_long cmd, void *data,
641     struct ifnet *ifp)
642 {
643 	struct if_laddrreq *iflr = (struct if_laddrreq *)data;
644 	struct ifaddr *ifa;
645 	struct sockaddr *sa;
646 
647 	/* sanity checks */
648 	if (data == NULL || ifp == NULL) {
649 		panic("invalid argument to in_lifaddr_ioctl");
650 		/*NOTRECHED*/
651 	}
652 
653 	switch (cmd) {
654 	case SIOCGLIFADDR:
655 		/* address must be specified on GET with IFLR_PREFIX */
656 		if ((iflr->flags & IFLR_PREFIX) == 0)
657 			break;
658 		/*FALLTHROUGH*/
659 	case SIOCALIFADDR:
660 	case SIOCDLIFADDR:
661 		/* address must be specified on ADD and DELETE */
662 		sa = (struct sockaddr *)&iflr->addr;
663 		if (sa->sa_family != AF_INET)
664 			return EINVAL;
665 		if (sa->sa_len != sizeof(struct sockaddr_in))
666 			return EINVAL;
667 		/* XXX need improvement */
668 		sa = (struct sockaddr *)&iflr->dstaddr;
669 		if (sa->sa_family != AF_UNSPEC && sa->sa_family != AF_INET)
670 			return EINVAL;
671 		if (sa->sa_len != 0 && sa->sa_len != sizeof(struct sockaddr_in))
672 			return EINVAL;
673 		break;
674 	default: /*shouldn't happen*/
675 #if 0
676 		panic("invalid cmd to in_lifaddr_ioctl");
677 		/*NOTREACHED*/
678 #else
679 		return EOPNOTSUPP;
680 #endif
681 	}
682 	if (sizeof(struct in_addr) * NBBY < iflr->prefixlen)
683 		return EINVAL;
684 
685 	switch (cmd) {
686 	case SIOCALIFADDR:
687 	    {
688 		struct in_aliasreq ifra;
689 
690 		if (iflr->flags & IFLR_PREFIX)
691 			return EINVAL;
692 
693 		/* copy args to in_aliasreq, perform ioctl(SIOCAIFADDR). */
694 		memset(&ifra, 0, sizeof(ifra));
695 		memcpy(ifra.ifra_name, iflr->iflr_name,
696 			sizeof(ifra.ifra_name));
697 
698 		memcpy(&ifra.ifra_addr, &iflr->addr,
699 			((struct sockaddr *)&iflr->addr)->sa_len);
700 
701 		if (((struct sockaddr *)&iflr->dstaddr)->sa_family) {	/*XXX*/
702 			memcpy(&ifra.ifra_dstaddr, &iflr->dstaddr,
703 				((struct sockaddr *)&iflr->dstaddr)->sa_len);
704 		}
705 
706 		ifra.ifra_mask.sin_family = AF_INET;
707 		ifra.ifra_mask.sin_len = sizeof(struct sockaddr_in);
708 		in_len2mask(&ifra.ifra_mask.sin_addr, iflr->prefixlen);
709 
710 		return in_control(so, SIOCAIFADDR, &ifra, ifp);
711 	    }
712 	case SIOCGLIFADDR:
713 	case SIOCDLIFADDR:
714 	    {
715 		struct in_ifaddr *ia;
716 		struct in_addr mask, candidate, match;
717 		struct sockaddr_in *sin;
718 		int cmp;
719 
720 		memset(&mask, 0, sizeof(mask));
721 		memset(&match, 0, sizeof(match));	/* XXX gcc */
722 		if (iflr->flags & IFLR_PREFIX) {
723 			/* lookup a prefix rather than address. */
724 			in_len2mask(&mask, iflr->prefixlen);
725 
726 			sin = (struct sockaddr_in *)&iflr->addr;
727 			match.s_addr = sin->sin_addr.s_addr;
728 			match.s_addr &= mask.s_addr;
729 
730 			/* if you set extra bits, that's wrong */
731 			if (match.s_addr != sin->sin_addr.s_addr)
732 				return EINVAL;
733 
734 			cmp = 1;
735 		} else {
736 			if (cmd == SIOCGLIFADDR) {
737 				/* on getting an address, take the 1st match */
738 				cmp = 0;	/*XXX*/
739 			} else {
740 				/* on deleting an address, do exact match */
741 				in_len2mask(&mask, 32);
742 				sin = (struct sockaddr_in *)&iflr->addr;
743 				match.s_addr = sin->sin_addr.s_addr;
744 
745 				cmp = 1;
746 			}
747 		}
748 
749 		IFADDR_FOREACH(ifa, ifp) {
750 			if (ifa->ifa_addr->sa_family != AF_INET)
751 				continue;
752 			if (cmp == 0)
753 				break;
754 			candidate.s_addr = ((struct sockaddr_in *)ifa->ifa_addr)->sin_addr.s_addr;
755 			candidate.s_addr &= mask.s_addr;
756 			if (candidate.s_addr == match.s_addr)
757 				break;
758 		}
759 		if (ifa == NULL)
760 			return EADDRNOTAVAIL;
761 		ia = (struct in_ifaddr *)ifa;
762 
763 		if (cmd == SIOCGLIFADDR) {
764 			/* fill in the if_laddrreq structure */
765 			memcpy(&iflr->addr, &ia->ia_addr, ia->ia_addr.sin_len);
766 
767 			if ((ifp->if_flags & IFF_POINTOPOINT) != 0) {
768 				memcpy(&iflr->dstaddr, &ia->ia_dstaddr,
769 					ia->ia_dstaddr.sin_len);
770 			} else
771 				memset(&iflr->dstaddr, 0, sizeof(iflr->dstaddr));
772 
773 			iflr->prefixlen =
774 				in_mask2len(&ia->ia_sockmask.sin_addr);
775 
776 			iflr->flags = 0;	/*XXX*/
777 
778 			return 0;
779 		} else {
780 			struct in_aliasreq ifra;
781 
782 			/* fill in_aliasreq and do ioctl(SIOCDIFADDR) */
783 			memset(&ifra, 0, sizeof(ifra));
784 			memcpy(ifra.ifra_name, iflr->iflr_name,
785 				sizeof(ifra.ifra_name));
786 
787 			memcpy(&ifra.ifra_addr, &ia->ia_addr,
788 				ia->ia_addr.sin_len);
789 			if ((ifp->if_flags & IFF_POINTOPOINT) != 0) {
790 				memcpy(&ifra.ifra_dstaddr, &ia->ia_dstaddr,
791 					ia->ia_dstaddr.sin_len);
792 			}
793 			memcpy(&ifra.ifra_dstaddr, &ia->ia_sockmask,
794 				ia->ia_sockmask.sin_len);
795 
796 			return in_control(so, SIOCDIFADDR, &ifra, ifp);
797 		}
798 	    }
799 	}
800 
801 	return EOPNOTSUPP;	/*just for safety*/
802 }
803 
804 /*
805  * Delete any existing route for an interface.
806  */
807 void
808 in_ifscrub(struct ifnet *ifp, struct in_ifaddr *ia)
809 {
810 
811 	in_scrubprefix(ia);
812 }
813 
814 /*
815  * Initialize an interface's internet address
816  * and routing table entry.
817  */
818 int
819 in_ifinit(struct ifnet *ifp, struct in_ifaddr *ia,
820     const struct sockaddr_in *sin, int scrub)
821 {
822 	u_int32_t i;
823 	struct sockaddr_in oldaddr;
824 	int s = splnet(), flags = RTF_UP, error;
825 
826 	if (sin == NULL)
827 		sin = &ia->ia_addr;
828 
829 	/*
830 	 * Set up new addresses.
831 	 */
832 	oldaddr = ia->ia_addr;
833 	if (ia->ia_addr.sin_family == AF_INET)
834 		LIST_REMOVE(ia, ia_hash);
835 	ia->ia_addr = *sin;
836 	LIST_INSERT_HEAD(&IN_IFADDR_HASH(ia->ia_addr.sin_addr.s_addr), ia, ia_hash);
837 
838 	/*
839 	 * Give the interface a chance to initialize
840 	 * if this is its first address,
841 	 * and to validate the address if necessary.
842 	 */
843 	if ((error = if_addr_init(ifp, &ia->ia_ifa, true)) != 0)
844 		goto bad;
845 	splx(s);
846 	if (scrub) {
847 		ia->ia_ifa.ifa_addr = sintosa(&oldaddr);
848 		in_ifscrub(ifp, ia);
849 		ia->ia_ifa.ifa_addr = sintosa(&ia->ia_addr);
850 	}
851 
852 	i = ia->ia_addr.sin_addr.s_addr;
853 	if (IN_CLASSA(i))
854 		ia->ia_netmask = IN_CLASSA_NET;
855 	else if (IN_CLASSB(i))
856 		ia->ia_netmask = IN_CLASSB_NET;
857 	else
858 		ia->ia_netmask = IN_CLASSC_NET;
859 	/*
860 	 * The subnet mask usually includes at least the standard network part,
861 	 * but may may be smaller in the case of supernetting.
862 	 * If it is set, we believe it.
863 	 */
864 	if (ia->ia_subnetmask == 0) {
865 		ia->ia_subnetmask = ia->ia_netmask;
866 		ia->ia_sockmask.sin_addr.s_addr = ia->ia_subnetmask;
867 	} else
868 		ia->ia_netmask &= ia->ia_subnetmask;
869 
870 	ia->ia_net = i & ia->ia_netmask;
871 	ia->ia_subnet = i & ia->ia_subnetmask;
872 	in_socktrim(&ia->ia_sockmask);
873 	/* re-calculate the "in_maxmtu" value */
874 	in_setmaxmtu();
875 	/*
876 	 * Add route for the network.
877 	 */
878 	ia->ia_ifa.ifa_metric = ifp->if_metric;
879 	if (ifp->if_flags & IFF_BROADCAST) {
880 		ia->ia_broadaddr.sin_addr.s_addr =
881 			ia->ia_subnet | ~ia->ia_subnetmask;
882 		ia->ia_netbroadcast.s_addr =
883 			ia->ia_net | ~ia->ia_netmask;
884 	} else if (ifp->if_flags & IFF_LOOPBACK) {
885 		ia->ia_dstaddr = ia->ia_addr;
886 		flags |= RTF_HOST;
887 	} else if (ifp->if_flags & IFF_POINTOPOINT) {
888 		if (ia->ia_dstaddr.sin_family != AF_INET)
889 			return (0);
890 		flags |= RTF_HOST;
891 	}
892 	error = in_addprefix(ia, flags);
893 	/*
894 	 * If the interface supports multicast, join the "all hosts"
895 	 * multicast group on that interface.
896 	 */
897 	if ((ifp->if_flags & IFF_MULTICAST) != 0 && ia->ia_allhosts == NULL) {
898 		struct in_addr addr;
899 
900 		addr.s_addr = INADDR_ALLHOSTS_GROUP;
901 		ia->ia_allhosts = in_addmulti(&addr, ifp);
902 	}
903 	return (error);
904 bad:
905 	splx(s);
906 	LIST_REMOVE(ia, ia_hash);
907 	ia->ia_addr = oldaddr;
908 	if (ia->ia_addr.sin_family == AF_INET)
909 		LIST_INSERT_HEAD(&IN_IFADDR_HASH(ia->ia_addr.sin_addr.s_addr),
910 		    ia, ia_hash);
911 	return (error);
912 }
913 
914 #define rtinitflags(x) \
915 	((((x)->ia_ifp->if_flags & (IFF_LOOPBACK | IFF_POINTOPOINT)) != 0) \
916 	    ? RTF_HOST : 0)
917 
918 /*
919  * add a route to prefix ("connected route" in cisco terminology).
920  * does nothing if there's some interface address with the same prefix already.
921  */
922 static int
923 in_addprefix(struct in_ifaddr *target, int flags)
924 {
925 	struct in_ifaddr *ia;
926 	struct in_addr prefix, mask, p;
927 	int error;
928 
929 	if ((flags & RTF_HOST) != 0)
930 		prefix = target->ia_dstaddr.sin_addr;
931 	else {
932 		prefix = target->ia_addr.sin_addr;
933 		mask = target->ia_sockmask.sin_addr;
934 		prefix.s_addr &= mask.s_addr;
935 	}
936 
937 	TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) {
938 		if (rtinitflags(ia))
939 			p = ia->ia_dstaddr.sin_addr;
940 		else {
941 			p = ia->ia_addr.sin_addr;
942 			p.s_addr &= ia->ia_sockmask.sin_addr.s_addr;
943 		}
944 
945 		if (prefix.s_addr != p.s_addr)
946 			continue;
947 
948 		/*
949 		 * if we got a matching prefix route inserted by other
950 		 * interface address, we don't need to bother
951 		 *
952 		 * XXX RADIX_MPATH implications here? -dyoung
953 		 *
954 		 * But we should still notify userland of the new address
955 		 */
956 		if (ia->ia_flags & IFA_ROUTE) {
957 			rt_newaddrmsg(RTM_NEWADDR, &target->ia_ifa, 0, NULL);
958 			return 0;
959 		}
960 	}
961 
962 	/*
963 	 * noone seem to have prefix route.  insert it.
964 	 */
965 	error = rtinit(&target->ia_ifa, RTM_ADD, flags);
966 	if (error == 0)
967 		target->ia_flags |= IFA_ROUTE;
968 	else if (error == EEXIST) {
969 		/*
970 		 * the fact the route already exists is not an error.
971 		 */
972 		error = 0;
973 	}
974 	return error;
975 }
976 
977 /*
978  * remove a route to prefix ("connected route" in cisco terminology).
979  * re-installs the route by using another interface address, if there's one
980  * with the same prefix (otherwise we lose the route mistakenly).
981  */
982 static int
983 in_scrubprefix(struct in_ifaddr *target)
984 {
985 	struct in_ifaddr *ia;
986 	struct in_addr prefix, mask, p;
987 	int error;
988 
989 	/* If we don't have IFA_ROUTE we should still inform userland */
990 	if ((target->ia_flags & IFA_ROUTE) == 0) {
991 		rt_newaddrmsg(RTM_DELADDR, &target->ia_ifa, 0, NULL);
992 		return 0;
993 	}
994 
995 	if (rtinitflags(target))
996 		prefix = target->ia_dstaddr.sin_addr;
997 	else {
998 		prefix = target->ia_addr.sin_addr;
999 		mask = target->ia_sockmask.sin_addr;
1000 		prefix.s_addr &= mask.s_addr;
1001 	}
1002 
1003 	TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) {
1004 		if (rtinitflags(ia))
1005 			p = ia->ia_dstaddr.sin_addr;
1006 		else {
1007 			p = ia->ia_addr.sin_addr;
1008 			p.s_addr &= ia->ia_sockmask.sin_addr.s_addr;
1009 		}
1010 
1011 		if (prefix.s_addr != p.s_addr)
1012 			continue;
1013 
1014 		/*
1015 		 * if we got a matching prefix route, move IFA_ROUTE to him
1016 		 */
1017 		if ((ia->ia_flags & IFA_ROUTE) == 0) {
1018 			rtinit(&target->ia_ifa, RTM_DELETE,
1019 			    rtinitflags(target));
1020 			target->ia_flags &= ~IFA_ROUTE;
1021 
1022 			error = rtinit(&ia->ia_ifa, RTM_ADD,
1023 			    rtinitflags(ia) | RTF_UP);
1024 			if (error == 0)
1025 				ia->ia_flags |= IFA_ROUTE;
1026 			return error;
1027 		}
1028 	}
1029 
1030 	/*
1031 	 * noone seem to have prefix route.  remove it.
1032 	 */
1033 	rtinit(&target->ia_ifa, RTM_DELETE, rtinitflags(target));
1034 	target->ia_flags &= ~IFA_ROUTE;
1035 	return 0;
1036 }
1037 
1038 #undef rtinitflags
1039 
1040 /*
1041  * Return 1 if the address might be a local broadcast address.
1042  */
1043 int
1044 in_broadcast(struct in_addr in, struct ifnet *ifp)
1045 {
1046 	struct ifaddr *ifa;
1047 
1048 	if (in.s_addr == INADDR_BROADCAST ||
1049 	    in_nullhost(in))
1050 		return 1;
1051 	if ((ifp->if_flags & IFF_BROADCAST) == 0)
1052 		return 0;
1053 	/*
1054 	 * Look through the list of addresses for a match
1055 	 * with a broadcast address.
1056 	 */
1057 #define ia (ifatoia(ifa))
1058 	IFADDR_FOREACH(ifa, ifp)
1059 		if (ifa->ifa_addr->sa_family == AF_INET &&
1060 		    !in_hosteq(in, ia->ia_addr.sin_addr) &&
1061 		    (in_hosteq(in, ia->ia_broadaddr.sin_addr) ||
1062 		     in_hosteq(in, ia->ia_netbroadcast) ||
1063 		     (hostzeroisbroadcast &&
1064 		      /*
1065 		       * Check for old-style (host 0) broadcast.
1066 		       */
1067 		      (in.s_addr == ia->ia_subnet ||
1068 		       in.s_addr == ia->ia_net))))
1069 			return 1;
1070 	return (0);
1071 #undef ia
1072 }
1073 
1074 /*
1075  * in_lookup_multi: look up the in_multi record for a given IP
1076  * multicast address on a given interface.  If no matching record is
1077  * found, return NULL.
1078  */
1079 struct in_multi *
1080 in_lookup_multi(struct in_addr addr, ifnet_t *ifp)
1081 {
1082 	struct in_multi *inm;
1083 
1084 	KASSERT(rw_lock_held(&in_multilock));
1085 
1086 	LIST_FOREACH(inm, &IN_MULTI_HASH(addr.s_addr, ifp), inm_list) {
1087 		if (in_hosteq(inm->inm_addr, addr) && inm->inm_ifp == ifp)
1088 			break;
1089 	}
1090 	return inm;
1091 }
1092 
1093 /*
1094  * in_multi_group: check whether the address belongs to an IP multicast
1095  * group we are joined on this interface.  Returns true or false.
1096  */
1097 bool
1098 in_multi_group(struct in_addr addr, ifnet_t *ifp, int flags)
1099 {
1100 	bool ingroup;
1101 
1102 	if (__predict_true(flags & IP_IGMP_MCAST) == 0) {
1103 		rw_enter(&in_multilock, RW_READER);
1104 		ingroup = in_lookup_multi(addr, ifp) != NULL;
1105 		rw_exit(&in_multilock);
1106 	} else {
1107 		/* XXX Recursive call from ip_output(). */
1108 		KASSERT(rw_lock_held(&in_multilock));
1109 		ingroup = in_lookup_multi(addr, ifp) != NULL;
1110 	}
1111 	return ingroup;
1112 }
1113 
1114 /*
1115  * Add an address to the list of IP multicast addresses for a given interface.
1116  */
1117 struct in_multi *
1118 in_addmulti(struct in_addr *ap, ifnet_t *ifp)
1119 {
1120 	struct sockaddr_in sin;
1121 	struct in_multi *inm;
1122 
1123 	/*
1124 	 * See if address already in list.
1125 	 */
1126 	rw_enter(&in_multilock, RW_WRITER);
1127 	inm = in_lookup_multi(*ap, ifp);
1128 	if (inm != NULL) {
1129 		/*
1130 		 * Found it; just increment the reference count.
1131 		 */
1132 		inm->inm_refcount++;
1133 		rw_exit(&in_multilock);
1134 		return inm;
1135 	}
1136 
1137 	/*
1138 	 * New address; allocate a new multicast record.
1139 	 */
1140 	inm = pool_get(&inmulti_pool, PR_NOWAIT);
1141 	if (inm == NULL) {
1142 		rw_exit(&in_multilock);
1143 		return NULL;
1144 	}
1145 	inm->inm_addr = *ap;
1146 	inm->inm_ifp = ifp;
1147 	inm->inm_refcount = 1;
1148 
1149 	/*
1150 	 * Ask the network driver to update its multicast reception
1151 	 * filter appropriately for the new address.
1152 	 */
1153 	sockaddr_in_init(&sin, ap, 0);
1154 	if (if_mcast_op(ifp, SIOCADDMULTI, sintosa(&sin)) != 0) {
1155 		rw_exit(&in_multilock);
1156 		pool_put(&inmulti_pool, inm);
1157 		return NULL;
1158 	}
1159 
1160 	/*
1161 	 * Let IGMP know that we have joined a new IP multicast group.
1162 	 */
1163 	if (igmp_joingroup(inm) != 0) {
1164 		rw_exit(&in_multilock);
1165 		pool_put(&inmulti_pool, inm);
1166 		return NULL;
1167 	}
1168 	LIST_INSERT_HEAD(
1169 	    &IN_MULTI_HASH(inm->inm_addr.s_addr, ifp),
1170 	    inm, inm_list);
1171 	in_multientries++;
1172 	rw_exit(&in_multilock);
1173 
1174 	return inm;
1175 }
1176 
1177 /*
1178  * Delete a multicast address record.
1179  */
1180 void
1181 in_delmulti(struct in_multi *inm)
1182 {
1183 	struct sockaddr_in sin;
1184 
1185 	rw_enter(&in_multilock, RW_WRITER);
1186 	if (--inm->inm_refcount > 0) {
1187 		rw_exit(&in_multilock);
1188 		return;
1189 	}
1190 
1191 	/*
1192 	 * No remaining claims to this record; let IGMP know that
1193 	 * we are leaving the multicast group.
1194 	 */
1195 	igmp_leavegroup(inm);
1196 
1197 	/*
1198 	 * Notify the network driver to update its multicast reception
1199 	 * filter.
1200 	 */
1201 	sockaddr_in_init(&sin, &inm->inm_addr, 0);
1202 	if_mcast_op(inm->inm_ifp, SIOCDELMULTI, sintosa(&sin));
1203 
1204 	/*
1205 	 * Unlink from list.
1206 	 */
1207 	LIST_REMOVE(inm, inm_list);
1208 	in_multientries--;
1209 	rw_exit(&in_multilock);
1210 
1211 	pool_put(&inmulti_pool, inm);
1212 }
1213 
1214 /*
1215  * in_next_multi: step through all of the in_multi records, one at a time.
1216  * The current position is remembered in "step", which the caller must
1217  * provide.  in_first_multi(), below, must be called to initialize "step"
1218  * and get the first record.  Both macros return a NULL "inm" when there
1219  * are no remaining records.
1220  */
1221 struct in_multi *
1222 in_next_multi(struct in_multistep *step)
1223 {
1224 	struct in_multi *inm;
1225 
1226 	KASSERT(rw_lock_held(&in_multilock));
1227 
1228 	while (step->i_inm == NULL && step->i_n < IN_MULTI_HASH_SIZE) {
1229 		step->i_inm = LIST_FIRST(&in_multihashtbl[++step->i_n]);
1230 	}
1231 	if ((inm = step->i_inm) != NULL) {
1232 		step->i_inm = LIST_NEXT(inm, inm_list);
1233 	}
1234 	return inm;
1235 }
1236 
1237 struct in_multi *
1238 in_first_multi(struct in_multistep *step)
1239 {
1240 	KASSERT(rw_lock_held(&in_multilock));
1241 
1242 	step->i_n = 0;
1243 	step->i_inm = LIST_FIRST(&in_multihashtbl[0]);
1244 	return in_next_multi(step);
1245 }
1246 
1247 void
1248 in_multi_lock(int op)
1249 {
1250 	rw_enter(&in_multilock, op);
1251 }
1252 
1253 void
1254 in_multi_unlock(void)
1255 {
1256 	rw_exit(&in_multilock);
1257 }
1258 
1259 int
1260 in_multi_lock_held(void)
1261 {
1262 	return rw_lock_held(&in_multilock);
1263 }
1264 
1265 struct sockaddr_in *
1266 in_selectsrc(struct sockaddr_in *sin, struct route *ro,
1267     int soopts, struct ip_moptions *mopts, int *errorp)
1268 {
1269 	struct rtentry *rt = NULL;
1270 	struct in_ifaddr *ia = NULL;
1271 
1272 	/*
1273          * If route is known or can be allocated now, take the
1274          * source address from the interface.  Otherwise, punt.
1275 	 */
1276 	if ((soopts & SO_DONTROUTE) != 0)
1277 		rtcache_free(ro);
1278 	else {
1279 		union {
1280 			struct sockaddr		dst;
1281 			struct sockaddr_in	dst4;
1282 		} u;
1283 
1284 		sockaddr_in_init(&u.dst4, &sin->sin_addr, 0);
1285 		rt = rtcache_lookup(ro, &u.dst);
1286 	}
1287 	/*
1288 	 * If we found a route, use the address
1289 	 * corresponding to the outgoing interface
1290 	 * unless it is the loopback (in case a route
1291 	 * to our address on another net goes to loopback).
1292 	 *
1293 	 * XXX Is this still true?  Do we care?
1294 	 */
1295 	if (rt != NULL && (rt->rt_ifp->if_flags & IFF_LOOPBACK) == 0)
1296 		ia = ifatoia(rt->rt_ifa);
1297 	if (ia == NULL) {
1298 		u_int16_t fport = sin->sin_port;
1299 
1300 		sin->sin_port = 0;
1301 		ia = ifatoia(ifa_ifwithladdr(sintosa(sin)));
1302 		sin->sin_port = fport;
1303 		if (ia == NULL) {
1304 			/* Find 1st non-loopback AF_INET address */
1305 			TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) {
1306 				if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK))
1307 					break;
1308 			}
1309 		}
1310 		if (ia == NULL) {
1311 			*errorp = EADDRNOTAVAIL;
1312 			return NULL;
1313 		}
1314 	}
1315 	/*
1316 	 * If the destination address is multicast and an outgoing
1317 	 * interface has been set as a multicast option, use the
1318 	 * address of that interface as our source address.
1319 	 */
1320 	if (IN_MULTICAST(sin->sin_addr.s_addr) && mopts != NULL) {
1321 		struct ip_moptions *imo;
1322 		struct ifnet *ifp;
1323 
1324 		imo = mopts;
1325 		if (imo->imo_multicast_ifp != NULL) {
1326 			ifp = imo->imo_multicast_ifp;
1327 			IFP_TO_IA(ifp, ia);		/* XXX */
1328 			if (ia == 0) {
1329 				*errorp = EADDRNOTAVAIL;
1330 				return NULL;
1331 			}
1332 		}
1333 	}
1334 	if (ia->ia_ifa.ifa_getifa != NULL) {
1335 		ia = ifatoia((*ia->ia_ifa.ifa_getifa)(&ia->ia_ifa,
1336 		                                      sintosa(sin)));
1337 	}
1338 #ifdef GETIFA_DEBUG
1339 	else
1340 		printf("%s: missing ifa_getifa\n", __func__);
1341 #endif
1342 	return satosin(&ia->ia_addr);
1343 }
1344 
1345 static void
1346 in_sysctl_init(struct sysctllog **clog)
1347 {
1348 	sysctl_createv(clog, 0, NULL, NULL,
1349 		       CTLFLAG_PERMANENT,
1350 		       CTLTYPE_NODE, "inet",
1351 		       SYSCTL_DESCR("PF_INET related settings"),
1352 		       NULL, 0, NULL, 0,
1353 		       CTL_NET, PF_INET, CTL_EOL);
1354 	sysctl_createv(clog, 0, NULL, NULL,
1355 		       CTLFLAG_PERMANENT,
1356 		       CTLTYPE_NODE, "ip",
1357 		       SYSCTL_DESCR("IPv4 related settings"),
1358 		       NULL, 0, NULL, 0,
1359 		       CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL);
1360 
1361 	sysctl_createv(clog, 0, NULL, NULL,
1362 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1363 		       CTLTYPE_INT, "subnetsarelocal",
1364 		       SYSCTL_DESCR("Whether logical subnets are considered "
1365 				    "local"),
1366 		       NULL, 0, &subnetsarelocal, 0,
1367 		       CTL_NET, PF_INET, IPPROTO_IP,
1368 		       IPCTL_SUBNETSARELOCAL, CTL_EOL);
1369 	sysctl_createv(clog, 0, NULL, NULL,
1370 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1371 		       CTLTYPE_INT, "hostzerobroadcast",
1372 		       SYSCTL_DESCR("All zeroes address is broadcast address"),
1373 		       NULL, 0, &hostzeroisbroadcast, 0,
1374 		       CTL_NET, PF_INET, IPPROTO_IP,
1375 		       IPCTL_HOSTZEROBROADCAST, CTL_EOL);
1376 }
1377