1 /*	$NetBSD: in.c,v 1.177 2016/07/28 09:03:50 ozaki-r 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 1.177 2016/07/28 09:03:50 ozaki-r Exp $");
95 
96 #include "arp.h"
97 
98 #ifdef _KERNEL_OPT
99 #include "opt_inet.h"
100 #include "opt_inet_conf.h"
101 #include "opt_mrouting.h"
102 #endif
103 
104 #include <sys/param.h>
105 #include <sys/ioctl.h>
106 #include <sys/errno.h>
107 #include <sys/kernel.h>
108 #include <sys/malloc.h>
109 #include <sys/socket.h>
110 #include <sys/socketvar.h>
111 #include <sys/sysctl.h>
112 #include <sys/systm.h>
113 #include <sys/proc.h>
114 #include <sys/syslog.h>
115 #include <sys/kauth.h>
116 #include <sys/kmem.h>
117 
118 #include <sys/cprng.h>
119 
120 #include <net/if.h>
121 #include <net/route.h>
122 #include <net/pfil.h>
123 
124 #include <net/if_arp.h>
125 #include <net/if_ether.h>
126 #include <net/if_types.h>
127 #include <net/if_llatbl.h>
128 #include <net/if_dl.h>
129 
130 #include <netinet/in_systm.h>
131 #include <netinet/in.h>
132 #include <netinet/in_var.h>
133 #include <netinet/ip.h>
134 #include <netinet/ip_var.h>
135 #include <netinet/in_ifattach.h>
136 #include <netinet/in_pcb.h>
137 #include <netinet/in_selsrc.h>
138 #include <netinet/if_inarp.h>
139 #include <netinet/ip_mroute.h>
140 #include <netinet/igmp_var.h>
141 
142 #ifdef IPSELSRC
143 #include <netinet/in_selsrc.h>
144 #endif
145 
146 static u_int	in_mask2len(struct in_addr *);
147 static void	in_len2mask(struct in_addr *, u_int);
148 static int	in_lifaddr_ioctl(struct socket *, u_long, void *,
149 	struct ifnet *);
150 
151 static int	in_addprefix(struct in_ifaddr *, int);
152 static int	in_scrubprefix(struct in_ifaddr *);
153 static void	in_sysctl_init(struct sysctllog **);
154 
155 #ifndef SUBNETSARELOCAL
156 #define	SUBNETSARELOCAL	1
157 #endif
158 
159 #ifndef HOSTZEROBROADCAST
160 #define HOSTZEROBROADCAST 0
161 #endif
162 
163 /* Note: 61, 127, 251, 509, 1021, 2039 are good. */
164 #ifndef IN_MULTI_HASH_SIZE
165 #define IN_MULTI_HASH_SIZE	509
166 #endif
167 
168 static int			subnetsarelocal = SUBNETSARELOCAL;
169 static int			hostzeroisbroadcast = HOSTZEROBROADCAST;
170 
171 /*
172  * This list is used to keep track of in_multi chains which belong to
173  * deleted interface addresses.  We use in_ifaddr so that a chain head
174  * won't be deallocated until all multicast address record are deleted.
175  */
176 
177 LIST_HEAD(in_multihashhead, in_multi);		/* Type of the hash head */
178 
179 static struct pool		inmulti_pool;
180 static u_int			in_multientries;
181 static struct in_multihashhead *in_multihashtbl;
182 static u_long			in_multihash;
183 static krwlock_t		in_multilock;
184 
185 #define IN_MULTI_HASH(x, ifp) \
186     (in_multihashtbl[(u_long)((x) ^ (ifp->if_index)) % IN_MULTI_HASH_SIZE])
187 
188 /* XXX DEPRECATED. Keep them to avoid breaking kvm(3) users. */
189 struct in_ifaddrhashhead *	in_ifaddrhashtbl;
190 u_long				in_ifaddrhash;
191 struct in_ifaddrhead		in_ifaddrhead;
192 
193 struct pslist_head *		in_ifaddrhashtbl_pslist;
194 u_long				in_ifaddrhash_pslist;
195 struct pslist_head		in_ifaddrhead_pslist;
196 
197 void
in_init(void)198 in_init(void)
199 {
200 	pool_init(&inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl",
201 	    NULL, IPL_SOFTNET);
202 	TAILQ_INIT(&in_ifaddrhead);
203 	PSLIST_INIT(&in_ifaddrhead_pslist);
204 
205 	in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true,
206 	    &in_ifaddrhash);
207 	in_ifaddrhashtbl_pslist = hashinit(IN_IFADDR_HASH_SIZE, HASH_PSLIST,
208 	    true, &in_ifaddrhash_pslist);
209 	in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true,
210 	    &in_multihash);
211 	rw_init(&in_multilock);
212 
213 	in_sysctl_init(NULL);
214 }
215 
216 /*
217  * Return 1 if an internet address is for a ``local'' host
218  * (one to which we have a connection).  If subnetsarelocal
219  * is true, this includes other subnets of the local net.
220  * Otherwise, it includes only the directly-connected (sub)nets.
221  */
222 int
in_localaddr(struct in_addr in)223 in_localaddr(struct in_addr in)
224 {
225 	struct in_ifaddr *ia;
226 
227 	if (subnetsarelocal) {
228 		IN_ADDRLIST_READER_FOREACH(ia) {
229 			if ((in.s_addr & ia->ia_netmask) == ia->ia_net)
230 				return (1);
231 		}
232 	} else {
233 		IN_ADDRLIST_READER_FOREACH(ia) {
234 			if ((in.s_addr & ia->ia_subnetmask) == ia->ia_subnet)
235 				return (1);
236 		}
237 	}
238 	return (0);
239 }
240 
241 /*
242  * Determine whether an IP address is in a reserved set of addresses
243  * that may not be forwarded, or whether datagrams to that destination
244  * may be forwarded.
245  */
246 int
in_canforward(struct in_addr in)247 in_canforward(struct in_addr in)
248 {
249 	u_int32_t net;
250 
251 	if (IN_EXPERIMENTAL(in.s_addr) || IN_MULTICAST(in.s_addr))
252 		return (0);
253 	if (IN_CLASSA(in.s_addr)) {
254 		net = in.s_addr & IN_CLASSA_NET;
255 		if (net == 0 || net == htonl(IN_LOOPBACKNET << IN_CLASSA_NSHIFT))
256 			return (0);
257 	}
258 	return (1);
259 }
260 
261 /*
262  * Trim a mask in a sockaddr
263  */
264 void
in_socktrim(struct sockaddr_in * ap)265 in_socktrim(struct sockaddr_in *ap)
266 {
267 	char *cplim = (char *) &ap->sin_addr;
268 	char *cp = (char *) (&ap->sin_addr + 1);
269 
270 	ap->sin_len = 0;
271 	while (--cp >= cplim)
272 		if (*cp) {
273 			(ap)->sin_len = cp - (char *) (ap) + 1;
274 			break;
275 		}
276 }
277 
278 /*
279  *  Routine to take an Internet address and convert into a
280  *  "dotted quad" representation for printing.
281  */
282 const char *
in_fmtaddr(struct in_addr addr)283 in_fmtaddr(struct in_addr addr)
284 {
285 	static char buf[sizeof("123.456.789.123")];
286 
287 	addr.s_addr = ntohl(addr.s_addr);
288 
289 	snprintf(buf, sizeof(buf), "%d.%d.%d.%d",
290 		(addr.s_addr >> 24) & 0xFF,
291 		(addr.s_addr >> 16) & 0xFF,
292 		(addr.s_addr >>  8) & 0xFF,
293 		(addr.s_addr >>  0) & 0xFF);
294 	return buf;
295 }
296 
297 /*
298  * Maintain the "in_maxmtu" variable, which is the largest
299  * mtu for non-local interfaces with AF_INET addresses assigned
300  * to them that are up.
301  */
302 unsigned long in_maxmtu;
303 
304 void
in_setmaxmtu(void)305 in_setmaxmtu(void)
306 {
307 	struct in_ifaddr *ia;
308 	struct ifnet *ifp;
309 	unsigned long maxmtu = 0;
310 
311 	IN_ADDRLIST_READER_FOREACH(ia) {
312 		if ((ifp = ia->ia_ifp) == 0)
313 			continue;
314 		if ((ifp->if_flags & (IFF_UP|IFF_LOOPBACK)) != IFF_UP)
315 			continue;
316 		if (ifp->if_mtu > maxmtu)
317 			maxmtu = ifp->if_mtu;
318 	}
319 	if (maxmtu)
320 		in_maxmtu = maxmtu;
321 }
322 
323 static u_int
in_mask2len(struct in_addr * mask)324 in_mask2len(struct in_addr *mask)
325 {
326 	u_int x, y;
327 	u_char *p;
328 
329 	p = (u_char *)mask;
330 	for (x = 0; x < sizeof(*mask); x++) {
331 		if (p[x] != 0xff)
332 			break;
333 	}
334 	y = 0;
335 	if (x < sizeof(*mask)) {
336 		for (y = 0; y < NBBY; y++) {
337 			if ((p[x] & (0x80 >> y)) == 0)
338 				break;
339 		}
340 	}
341 	return x * NBBY + y;
342 }
343 
344 static void
in_len2mask(struct in_addr * mask,u_int len)345 in_len2mask(struct in_addr *mask, u_int len)
346 {
347 	u_int i;
348 	u_char *p;
349 
350 	p = (u_char *)mask;
351 	memset(mask, 0, sizeof(*mask));
352 	for (i = 0; i < len / NBBY; i++)
353 		p[i] = 0xff;
354 	if (len % NBBY)
355 		p[i] = (0xff00 >> (len % NBBY)) & 0xff;
356 }
357 
358 /*
359  * Generic internet control operations (ioctl's).
360  * Ifp is 0 if not an interface-specific ioctl.
361  */
362 /* ARGSUSED */
363 static int
in_control0(struct socket * so,u_long cmd,void * data,struct ifnet * ifp)364 in_control0(struct socket *so, u_long cmd, void *data, struct ifnet *ifp)
365 {
366 	struct ifreq *ifr = (struct ifreq *)data;
367 	struct in_ifaddr *ia = NULL;
368 	struct in_aliasreq *ifra = (struct in_aliasreq *)data;
369 	struct sockaddr_in oldaddr;
370 	int error, hostIsNew, maskIsNew;
371 	int newifaddr = 0;
372 	bool run_hook = false;
373 	bool need_reinsert = false;
374 
375 	switch (cmd) {
376 	case SIOCALIFADDR:
377 	case SIOCDLIFADDR:
378 	case SIOCGLIFADDR:
379 		if (ifp == NULL)
380 			return EINVAL;
381 		return in_lifaddr_ioctl(so, cmd, data, ifp);
382 	case SIOCGIFADDRPREF:
383 	case SIOCSIFADDRPREF:
384 		if (ifp == NULL)
385 			return EINVAL;
386 		return ifaddrpref_ioctl(so, cmd, data, ifp);
387 	}
388 
389 	/*
390 	 * Find address for this interface, if it exists.
391 	 */
392 	if (ifp != NULL)
393 		ia = in_get_ia_from_ifp(ifp);
394 
395 	hostIsNew = 1;		/* moved here to appease gcc */
396 	switch (cmd) {
397 	case SIOCAIFADDR:
398 	case SIOCDIFADDR:
399 	case SIOCGIFALIAS:
400 	case SIOCGIFAFLAG_IN:
401 		if (ifra->ifra_addr.sin_family == AF_INET) {
402 			IN_ADDRHASH_READER_FOREACH(ia,
403 			    ifra->ifra_addr.sin_addr.s_addr) {
404 				if (ia->ia_ifp == ifp &&
405 				    in_hosteq(ia->ia_addr.sin_addr,
406 				    ifra->ifra_addr.sin_addr))
407 					break;
408 			}
409 		}
410 		if ((cmd == SIOCDIFADDR ||
411 		    cmd == SIOCGIFALIAS ||
412 		    cmd == SIOCGIFAFLAG_IN) &&
413 		    ia == NULL)
414 			return (EADDRNOTAVAIL);
415 
416 		if (cmd == SIOCDIFADDR &&
417 		    ifra->ifra_addr.sin_family == AF_UNSPEC) {
418 			ifra->ifra_addr.sin_family = AF_INET;
419 		}
420 		/* FALLTHROUGH */
421 	case SIOCSIFADDR:
422 		if (ia == NULL || ia->ia_addr.sin_family != AF_INET)
423 			;
424 		else if (ifra->ifra_addr.sin_len == 0) {
425 			ifra->ifra_addr = ia->ia_addr;
426 			hostIsNew = 0;
427 		} else if (in_hosteq(ia->ia_addr.sin_addr,
428 		           ifra->ifra_addr.sin_addr))
429 			hostIsNew = 0;
430 		/* FALLTHROUGH */
431 	case SIOCSIFDSTADDR:
432 		if (ifra->ifra_addr.sin_family != AF_INET)
433 			return (EAFNOSUPPORT);
434 		/* FALLTHROUGH */
435 	case SIOCSIFNETMASK:
436 		if (ifp == NULL)
437 			panic("in_control");
438 
439 		if (cmd == SIOCGIFALIAS || cmd == SIOCGIFAFLAG_IN)
440 			break;
441 
442 		if (ia == NULL &&
443 		    (cmd == SIOCSIFNETMASK || cmd == SIOCSIFDSTADDR))
444 			return (EADDRNOTAVAIL);
445 
446 		if (kauth_authorize_network(curlwp->l_cred, KAUTH_NETWORK_INTERFACE,
447 		    KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp, (void *)cmd,
448 		    NULL) != 0)
449 			return (EPERM);
450 
451 		if (ia == NULL) {
452 			ia = malloc(sizeof(*ia), M_IFADDR, M_WAITOK|M_ZERO);
453 			if (ia == NULL)
454 				return (ENOBUFS);
455 			ia->ia_ifa.ifa_addr = sintosa(&ia->ia_addr);
456 			ia->ia_ifa.ifa_dstaddr = sintosa(&ia->ia_dstaddr);
457 			ia->ia_ifa.ifa_netmask = sintosa(&ia->ia_sockmask);
458 #ifdef IPSELSRC
459 			ia->ia_ifa.ifa_getifa = in_getifa;
460 #else /* IPSELSRC */
461 			ia->ia_ifa.ifa_getifa = NULL;
462 #endif /* IPSELSRC */
463 			ia->ia_sockmask.sin_len = 8;
464 			ia->ia_sockmask.sin_family = AF_INET;
465 			if (ifp->if_flags & IFF_BROADCAST) {
466 				ia->ia_broadaddr.sin_len = sizeof(ia->ia_addr);
467 				ia->ia_broadaddr.sin_family = AF_INET;
468 			}
469 			ia->ia_ifp = ifp;
470 			ia->ia_idsalt = cprng_fast32() % 65535;
471 			LIST_INIT(&ia->ia_multiaddrs);
472 			IN_ADDRHASH_ENTRY_INIT(ia);
473 			IN_ADDRLIST_ENTRY_INIT(ia);
474 
475 			newifaddr = 1;
476 		}
477 		break;
478 
479 	case SIOCSIFBRDADDR:
480 		if (kauth_authorize_network(curlwp->l_cred, KAUTH_NETWORK_INTERFACE,
481 		    KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp, (void *)cmd,
482 		    NULL) != 0)
483 			return (EPERM);
484 		/* FALLTHROUGH */
485 
486 	case SIOCGIFADDR:
487 	case SIOCGIFNETMASK:
488 	case SIOCGIFDSTADDR:
489 	case SIOCGIFBRDADDR:
490 		if (ia == NULL)
491 			return (EADDRNOTAVAIL);
492 		break;
493 	}
494 	error = 0;
495 	switch (cmd) {
496 
497 	case SIOCGIFADDR:
498 		ifreq_setaddr(cmd, ifr, sintocsa(&ia->ia_addr));
499 		break;
500 
501 	case SIOCGIFBRDADDR:
502 		if ((ifp->if_flags & IFF_BROADCAST) == 0)
503 			return (EINVAL);
504 		ifreq_setdstaddr(cmd, ifr, sintocsa(&ia->ia_broadaddr));
505 		break;
506 
507 	case SIOCGIFDSTADDR:
508 		if ((ifp->if_flags & IFF_POINTOPOINT) == 0)
509 			return (EINVAL);
510 		ifreq_setdstaddr(cmd, ifr, sintocsa(&ia->ia_dstaddr));
511 		break;
512 
513 	case SIOCGIFNETMASK:
514 		/*
515 		 * We keep the number of trailing zero bytes the sin_len field
516 		 * of ia_sockmask, so we fix this before we pass it back to
517 		 * userland.
518 		 */
519 		oldaddr = ia->ia_sockmask;
520 		oldaddr.sin_len = sizeof(struct sockaddr_in);
521 		ifreq_setaddr(cmd, ifr, (const void *)&oldaddr);
522 		break;
523 
524 	case SIOCSIFDSTADDR:
525 		if ((ifp->if_flags & IFF_POINTOPOINT) == 0)
526 			return (EINVAL);
527 		oldaddr = ia->ia_dstaddr;
528 		ia->ia_dstaddr = *satocsin(ifreq_getdstaddr(cmd, ifr));
529 		if ((error = if_addr_init(ifp, &ia->ia_ifa, false)) != 0) {
530 			ia->ia_dstaddr = oldaddr;
531 			return error;
532 		}
533 		if (ia->ia_flags & IFA_ROUTE) {
534 			ia->ia_ifa.ifa_dstaddr = sintosa(&oldaddr);
535 			rtinit(&ia->ia_ifa, RTM_DELETE, RTF_HOST);
536 			ia->ia_ifa.ifa_dstaddr = sintosa(&ia->ia_dstaddr);
537 			rtinit(&ia->ia_ifa, RTM_ADD, RTF_HOST|RTF_UP);
538 		}
539 		break;
540 
541 	case SIOCSIFBRDADDR:
542 		if ((ifp->if_flags & IFF_BROADCAST) == 0)
543 			return EINVAL;
544 		ia->ia_broadaddr = *satocsin(ifreq_getbroadaddr(cmd, ifr));
545 		break;
546 
547 	case SIOCSIFADDR:
548 		if (!newifaddr) {
549 			LIST_REMOVE(ia, ia_hash);
550 			IN_ADDRHASH_WRITER_REMOVE(ia);
551 			need_reinsert = true;
552 		}
553 		error = in_ifinit(ifp, ia, satocsin(ifreq_getaddr(cmd, ifr)),
554 		    1, hostIsNew);
555 
556 		run_hook = true;
557 		break;
558 
559 	case SIOCSIFNETMASK:
560 		in_ifscrub(ifp, ia);
561 		ia->ia_sockmask = *satocsin(ifreq_getaddr(cmd, ifr));
562 		ia->ia_subnetmask = ia->ia_sockmask.sin_addr.s_addr;
563 		if (!newifaddr) {
564 			LIST_REMOVE(ia, ia_hash);
565 			IN_ADDRHASH_WRITER_REMOVE(ia);
566 			need_reinsert = true;
567 		}
568 		error = in_ifinit(ifp, ia, NULL, 0, 0);
569 		break;
570 
571 	case SIOCAIFADDR:
572 		maskIsNew = 0;
573 		if (ifra->ifra_mask.sin_len) {
574 			/* Only scrub if we control the prefix route,
575 			 * otherwise userland gets a bogus message */
576 			if ((ia->ia_flags & IFA_ROUTE))
577 				in_ifscrub(ifp, ia);
578 			ia->ia_sockmask = ifra->ifra_mask;
579 			ia->ia_subnetmask = ia->ia_sockmask.sin_addr.s_addr;
580 			maskIsNew = 1;
581 		}
582 		if ((ifp->if_flags & IFF_POINTOPOINT) &&
583 		    (ifra->ifra_dstaddr.sin_family == AF_INET)) {
584 			/* Only scrub if we control the prefix route,
585 			 * otherwise userland gets a bogus message */
586 			if ((ia->ia_flags & IFA_ROUTE))
587 				in_ifscrub(ifp, ia);
588 			ia->ia_dstaddr = ifra->ifra_dstaddr;
589 			maskIsNew  = 1; /* We lie; but the effect's the same */
590 		}
591 		if (ifra->ifra_addr.sin_family == AF_INET &&
592 		    (hostIsNew || maskIsNew)) {
593 			if (!newifaddr) {
594 				LIST_REMOVE(ia, ia_hash);
595 				IN_ADDRHASH_WRITER_REMOVE(ia);
596 				need_reinsert = true;
597 			}
598 			error = in_ifinit(ifp, ia, &ifra->ifra_addr, 0,
599 			    hostIsNew);
600 		}
601 		if ((ifp->if_flags & IFF_BROADCAST) &&
602 		    (ifra->ifra_broadaddr.sin_family == AF_INET))
603 			ia->ia_broadaddr = ifra->ifra_broadaddr;
604 		run_hook = true;
605 		break;
606 
607 	case SIOCGIFALIAS:
608 		ifra->ifra_mask = ia->ia_sockmask;
609 		if ((ifp->if_flags & IFF_POINTOPOINT) &&
610 		    (ia->ia_dstaddr.sin_family == AF_INET))
611 			ifra->ifra_dstaddr = ia->ia_dstaddr;
612 		else if ((ifp->if_flags & IFF_BROADCAST) &&
613 		    (ia->ia_broadaddr.sin_family == AF_INET))
614 			ifra->ifra_broadaddr = ia->ia_broadaddr;
615 		else
616 			memset(&ifra->ifra_broadaddr, 0,
617 			      sizeof(ifra->ifra_broadaddr));
618 		break;
619 
620 	case SIOCGIFAFLAG_IN:
621 		ifr->ifr_addrflags = ia->ia4_flags;
622 		break;
623 
624 	case SIOCDIFADDR:
625 		in_purgeaddr(&ia->ia_ifa);
626 		run_hook = true;
627 		break;
628 
629 #ifdef MROUTING
630 	case SIOCGETVIFCNT:
631 	case SIOCGETSGCNT:
632 		error = mrt_ioctl(so, cmd, data);
633 		break;
634 #endif /* MROUTING */
635 
636 	default:
637 		return ENOTTY;
638 	}
639 
640 	/*
641 	 * XXX insert regardless of error to make in_purgeaddr below work.
642 	 * Need to improve.
643 	 */
644 	if (newifaddr) {
645 		TAILQ_INSERT_TAIL(&in_ifaddrhead, ia, ia_list);
646 		ifaref(&ia->ia_ifa);
647 		ifa_insert(ifp, &ia->ia_ifa);
648 		IN_ADDRLIST_WRITER_INSERT_TAIL(ia);
649 		LIST_INSERT_HEAD(&IN_IFADDR_HASH(ia->ia_addr.sin_addr.s_addr),
650 		    ia, ia_hash);
651 		IN_ADDRHASH_WRITER_INSERT_HEAD(ia);
652 	} else if (need_reinsert) {
653 		LIST_INSERT_HEAD(&IN_IFADDR_HASH(ia->ia_addr.sin_addr.s_addr),
654 		    ia, ia_hash);
655 		IN_ADDRHASH_WRITER_INSERT_HEAD(ia);
656 	}
657 
658 	if (error == 0) {
659 		if (run_hook)
660 			(void)pfil_run_hooks(if_pfil,
661 			    (struct mbuf **)cmd, ifp, PFIL_IFADDR);
662 	} else if (newifaddr) {
663 		KASSERT(ia != NULL);
664 		in_purgeaddr(&ia->ia_ifa);
665 	}
666 
667 	return error;
668 }
669 
670 int
in_control(struct socket * so,u_long cmd,void * data,struct ifnet * ifp)671 in_control(struct socket *so, u_long cmd, void *data, struct ifnet *ifp)
672 {
673 	int error;
674 
675 	mutex_enter(softnet_lock);
676 	error = in_control0(so, cmd, data, ifp);
677 	mutex_exit(softnet_lock);
678 
679 	return error;
680 }
681 
682 /* Add ownaddr as loopback rtentry. */
683 static void
in_ifaddlocal(struct ifaddr * ifa)684 in_ifaddlocal(struct ifaddr *ifa)
685 {
686 	struct in_ifaddr *ia;
687 
688 	ia = (struct in_ifaddr *)ifa;
689 	if (ia->ia_addr.sin_addr.s_addr == INADDR_ANY ||
690 	    (ia->ia_ifp->if_flags & IFF_POINTOPOINT &&
691 	    in_hosteq(ia->ia_dstaddr.sin_addr, ia->ia_addr.sin_addr)))
692 	{
693 		rt_newaddrmsg(RTM_NEWADDR, ifa, 0, NULL);
694 		return;
695 	}
696 
697 	rt_ifa_addlocal(ifa);
698 }
699 
700 /* Remove loopback entry of ownaddr */
701 static void
in_ifremlocal(struct ifaddr * ifa)702 in_ifremlocal(struct ifaddr *ifa)
703 {
704 	struct in_ifaddr *ia, *p;
705 	struct ifaddr *alt_ifa = NULL;
706 	int ia_count = 0;
707 
708 	ia = (struct in_ifaddr *)ifa;
709 	/* Delete the entry if exactly one ifaddr matches the
710 	 * address, ifa->ifa_addr. */
711 	IN_ADDRLIST_READER_FOREACH(p) {
712 		if (!in_hosteq(p->ia_addr.sin_addr, ia->ia_addr.sin_addr))
713 			continue;
714 		if (p->ia_ifp != ia->ia_ifp)
715 			alt_ifa = &p->ia_ifa;
716 		if (++ia_count > 1 && alt_ifa != NULL)
717 			break;
718 	}
719 
720 	if (ia_count == 0)
721 		return;
722 
723 	rt_ifa_remlocal(ifa, ia_count == 1 ? NULL : alt_ifa);
724 }
725 
726 void
in_purgeaddr(struct ifaddr * ifa)727 in_purgeaddr(struct ifaddr *ifa)
728 {
729 	struct ifnet *ifp = ifa->ifa_ifp;
730 	struct in_ifaddr *ia = (void *) ifa;
731 
732 	/* stop DAD processing */
733 	if (ia->ia_dad_stop != NULL)
734 		ia->ia_dad_stop(ifa);
735 
736 	in_ifscrub(ifp, ia);
737 	in_ifremlocal(ifa);
738 	LIST_REMOVE(ia, ia_hash);
739 	IN_ADDRHASH_WRITER_REMOVE(ia);
740 	IN_ADDRHASH_ENTRY_DESTROY(ia);
741 	ifa_remove(ifp, &ia->ia_ifa);
742 	TAILQ_REMOVE(&in_ifaddrhead, ia, ia_list);
743 	IN_ADDRLIST_WRITER_REMOVE(ia);
744 	IN_ADDRLIST_ENTRY_DESTROY(ia);
745 
746 	if (ia->ia_allhosts != NULL)
747 		in_delmulti(ia->ia_allhosts);
748 	ifafree(&ia->ia_ifa);
749 	in_setmaxmtu();
750 }
751 
752 void
in_purgeif(struct ifnet * ifp)753 in_purgeif(struct ifnet *ifp)		/* MUST be called at splsoftnet() */
754 {
755 	if_purgeaddrs(ifp, AF_INET, in_purgeaddr);
756 	igmp_purgeif(ifp);		/* manipulates pools */
757 #ifdef MROUTING
758 	ip_mrouter_detach(ifp);
759 #endif
760 }
761 
762 /*
763  * SIOC[GAD]LIFADDR.
764  *	SIOCGLIFADDR: get first address. (???)
765  *	SIOCGLIFADDR with IFLR_PREFIX:
766  *		get first address that matches the specified prefix.
767  *	SIOCALIFADDR: add the specified address.
768  *	SIOCALIFADDR with IFLR_PREFIX:
769  *		EINVAL since we can't deduce hostid part of the address.
770  *	SIOCDLIFADDR: delete the specified address.
771  *	SIOCDLIFADDR with IFLR_PREFIX:
772  *		delete the first address that matches the specified prefix.
773  * return values:
774  *	EINVAL on invalid parameters
775  *	EADDRNOTAVAIL on prefix match failed/specified address not found
776  *	other values may be returned from in_ioctl()
777  */
778 static int
in_lifaddr_ioctl(struct socket * so,u_long cmd,void * data,struct ifnet * ifp)779 in_lifaddr_ioctl(struct socket *so, u_long cmd, void *data,
780     struct ifnet *ifp)
781 {
782 	struct if_laddrreq *iflr = (struct if_laddrreq *)data;
783 	struct ifaddr *ifa;
784 	struct sockaddr *sa;
785 
786 	/* sanity checks */
787 	if (data == NULL || ifp == NULL) {
788 		panic("invalid argument to in_lifaddr_ioctl");
789 		/*NOTRECHED*/
790 	}
791 
792 	switch (cmd) {
793 	case SIOCGLIFADDR:
794 		/* address must be specified on GET with IFLR_PREFIX */
795 		if ((iflr->flags & IFLR_PREFIX) == 0)
796 			break;
797 		/*FALLTHROUGH*/
798 	case SIOCALIFADDR:
799 	case SIOCDLIFADDR:
800 		/* address must be specified on ADD and DELETE */
801 		sa = (struct sockaddr *)&iflr->addr;
802 		if (sa->sa_family != AF_INET)
803 			return EINVAL;
804 		if (sa->sa_len != sizeof(struct sockaddr_in))
805 			return EINVAL;
806 		/* XXX need improvement */
807 		sa = (struct sockaddr *)&iflr->dstaddr;
808 		if (sa->sa_family != AF_UNSPEC && sa->sa_family != AF_INET)
809 			return EINVAL;
810 		if (sa->sa_len != 0 && sa->sa_len != sizeof(struct sockaddr_in))
811 			return EINVAL;
812 		break;
813 	default: /*shouldn't happen*/
814 #if 0
815 		panic("invalid cmd to in_lifaddr_ioctl");
816 		/*NOTREACHED*/
817 #else
818 		return EOPNOTSUPP;
819 #endif
820 	}
821 	if (sizeof(struct in_addr) * NBBY < iflr->prefixlen)
822 		return EINVAL;
823 
824 	switch (cmd) {
825 	case SIOCALIFADDR:
826 	    {
827 		struct in_aliasreq ifra;
828 
829 		if (iflr->flags & IFLR_PREFIX)
830 			return EINVAL;
831 
832 		/* copy args to in_aliasreq, perform ioctl(SIOCAIFADDR). */
833 		memset(&ifra, 0, sizeof(ifra));
834 		memcpy(ifra.ifra_name, iflr->iflr_name,
835 			sizeof(ifra.ifra_name));
836 
837 		memcpy(&ifra.ifra_addr, &iflr->addr,
838 			((struct sockaddr *)&iflr->addr)->sa_len);
839 
840 		if (((struct sockaddr *)&iflr->dstaddr)->sa_family) {	/*XXX*/
841 			memcpy(&ifra.ifra_dstaddr, &iflr->dstaddr,
842 				((struct sockaddr *)&iflr->dstaddr)->sa_len);
843 		}
844 
845 		ifra.ifra_mask.sin_family = AF_INET;
846 		ifra.ifra_mask.sin_len = sizeof(struct sockaddr_in);
847 		in_len2mask(&ifra.ifra_mask.sin_addr, iflr->prefixlen);
848 
849 		return in_control(so, SIOCAIFADDR, &ifra, ifp);
850 	    }
851 	case SIOCGLIFADDR:
852 	case SIOCDLIFADDR:
853 	    {
854 		struct in_ifaddr *ia;
855 		struct in_addr mask, candidate, match;
856 		struct sockaddr_in *sin;
857 		int cmp;
858 
859 		memset(&mask, 0, sizeof(mask));
860 		memset(&match, 0, sizeof(match));	/* XXX gcc */
861 		if (iflr->flags & IFLR_PREFIX) {
862 			/* lookup a prefix rather than address. */
863 			in_len2mask(&mask, iflr->prefixlen);
864 
865 			sin = (struct sockaddr_in *)&iflr->addr;
866 			match.s_addr = sin->sin_addr.s_addr;
867 			match.s_addr &= mask.s_addr;
868 
869 			/* if you set extra bits, that's wrong */
870 			if (match.s_addr != sin->sin_addr.s_addr)
871 				return EINVAL;
872 
873 			cmp = 1;
874 		} else {
875 			if (cmd == SIOCGLIFADDR) {
876 				/* on getting an address, take the 1st match */
877 				cmp = 0;	/*XXX*/
878 			} else {
879 				/* on deleting an address, do exact match */
880 				in_len2mask(&mask, 32);
881 				sin = (struct sockaddr_in *)&iflr->addr;
882 				match.s_addr = sin->sin_addr.s_addr;
883 
884 				cmp = 1;
885 			}
886 		}
887 
888 		IFADDR_READER_FOREACH(ifa, ifp) {
889 			if (ifa->ifa_addr->sa_family != AF_INET)
890 				continue;
891 			if (cmp == 0)
892 				break;
893 			candidate.s_addr = ((struct sockaddr_in *)ifa->ifa_addr)->sin_addr.s_addr;
894 			candidate.s_addr &= mask.s_addr;
895 			if (candidate.s_addr == match.s_addr)
896 				break;
897 		}
898 		if (ifa == NULL)
899 			return EADDRNOTAVAIL;
900 		ia = (struct in_ifaddr *)ifa;
901 
902 		if (cmd == SIOCGLIFADDR) {
903 			/* fill in the if_laddrreq structure */
904 			memcpy(&iflr->addr, &ia->ia_addr, ia->ia_addr.sin_len);
905 
906 			if ((ifp->if_flags & IFF_POINTOPOINT) != 0) {
907 				memcpy(&iflr->dstaddr, &ia->ia_dstaddr,
908 					ia->ia_dstaddr.sin_len);
909 			} else
910 				memset(&iflr->dstaddr, 0, sizeof(iflr->dstaddr));
911 
912 			iflr->prefixlen =
913 				in_mask2len(&ia->ia_sockmask.sin_addr);
914 
915 			iflr->flags = 0;	/*XXX*/
916 
917 			return 0;
918 		} else {
919 			struct in_aliasreq ifra;
920 
921 			/* fill in_aliasreq and do ioctl(SIOCDIFADDR) */
922 			memset(&ifra, 0, sizeof(ifra));
923 			memcpy(ifra.ifra_name, iflr->iflr_name,
924 				sizeof(ifra.ifra_name));
925 
926 			memcpy(&ifra.ifra_addr, &ia->ia_addr,
927 				ia->ia_addr.sin_len);
928 			if ((ifp->if_flags & IFF_POINTOPOINT) != 0) {
929 				memcpy(&ifra.ifra_dstaddr, &ia->ia_dstaddr,
930 					ia->ia_dstaddr.sin_len);
931 			}
932 			memcpy(&ifra.ifra_dstaddr, &ia->ia_sockmask,
933 				ia->ia_sockmask.sin_len);
934 
935 			return in_control(so, SIOCDIFADDR, &ifra, ifp);
936 		}
937 	    }
938 	}
939 
940 	return EOPNOTSUPP;	/*just for safety*/
941 }
942 
943 /*
944  * Delete any existing route for an interface.
945  */
946 void
in_ifscrub(struct ifnet * ifp,struct in_ifaddr * ia)947 in_ifscrub(struct ifnet *ifp, struct in_ifaddr *ia)
948 {
949 
950 	in_scrubprefix(ia);
951 }
952 
953 /*
954  * Initialize an interface's internet address
955  * and routing table entry.
956  */
957 int
in_ifinit(struct ifnet * ifp,struct in_ifaddr * ia,const struct sockaddr_in * sin,int scrub,int hostIsNew)958 in_ifinit(struct ifnet *ifp, struct in_ifaddr *ia,
959     const struct sockaddr_in *sin, int scrub, int hostIsNew)
960 {
961 	u_int32_t i;
962 	struct sockaddr_in oldaddr;
963 	int s = splnet(), flags = RTF_UP, error;
964 
965 	if (sin == NULL)
966 		sin = &ia->ia_addr;
967 
968 	/*
969 	 * Set up new addresses.
970 	 */
971 	oldaddr = ia->ia_addr;
972 	ia->ia_addr = *sin;
973 
974 	/* Set IN_IFF flags early for if_addr_init() */
975 	if (hostIsNew && if_do_dad(ifp) && !in_nullhost(ia->ia_addr.sin_addr)) {
976 		if (ifp->if_link_state == LINK_STATE_DOWN)
977 			ia->ia4_flags |= IN_IFF_DETACHED;
978 		else
979 			/* State the intent to try DAD if possible */
980 			ia->ia4_flags |= IN_IFF_TRYTENTATIVE;
981 	}
982 
983 	/*
984 	 * Give the interface a chance to initialize
985 	 * if this is its first address,
986 	 * and to validate the address if necessary.
987 	 */
988 	if ((error = if_addr_init(ifp, &ia->ia_ifa, true)) != 0)
989 		goto bad;
990 	/* Now clear the try tentative flag, it's job is done. */
991 	ia->ia4_flags &= ~IN_IFF_TRYTENTATIVE;
992 	splx(s);
993 
994 	if (scrub) {
995 		ia->ia_ifa.ifa_addr = sintosa(&oldaddr);
996 		in_ifscrub(ifp, ia);
997 		ia->ia_ifa.ifa_addr = sintosa(&ia->ia_addr);
998 	}
999 
1000 	/* Add the local route to the address */
1001 	in_ifaddlocal(&ia->ia_ifa);
1002 
1003 	i = ia->ia_addr.sin_addr.s_addr;
1004 	if (IN_CLASSA(i))
1005 		ia->ia_netmask = IN_CLASSA_NET;
1006 	else if (IN_CLASSB(i))
1007 		ia->ia_netmask = IN_CLASSB_NET;
1008 	else
1009 		ia->ia_netmask = IN_CLASSC_NET;
1010 	/*
1011 	 * The subnet mask usually includes at least the standard network part,
1012 	 * but may may be smaller in the case of supernetting.
1013 	 * If it is set, we believe it.
1014 	 */
1015 	if (ia->ia_subnetmask == 0) {
1016 		ia->ia_subnetmask = ia->ia_netmask;
1017 		ia->ia_sockmask.sin_addr.s_addr = ia->ia_subnetmask;
1018 	} else
1019 		ia->ia_netmask &= ia->ia_subnetmask;
1020 
1021 	ia->ia_net = i & ia->ia_netmask;
1022 	ia->ia_subnet = i & ia->ia_subnetmask;
1023 	in_socktrim(&ia->ia_sockmask);
1024 	/* re-calculate the "in_maxmtu" value */
1025 	in_setmaxmtu();
1026 	/*
1027 	 * Add route for the network.
1028 	 */
1029 	ia->ia_ifa.ifa_metric = ifp->if_metric;
1030 	if (ifp->if_flags & IFF_BROADCAST) {
1031 		ia->ia_broadaddr.sin_addr.s_addr =
1032 			ia->ia_subnet | ~ia->ia_subnetmask;
1033 		ia->ia_netbroadcast.s_addr =
1034 			ia->ia_net | ~ia->ia_netmask;
1035 	} else if (ifp->if_flags & IFF_LOOPBACK) {
1036 		ia->ia_dstaddr = ia->ia_addr;
1037 		flags |= RTF_HOST;
1038 	} else if (ifp->if_flags & IFF_POINTOPOINT) {
1039 		if (ia->ia_dstaddr.sin_family != AF_INET)
1040 			return (0);
1041 		flags |= RTF_HOST;
1042 	}
1043 	error = in_addprefix(ia, flags);
1044 	/*
1045 	 * If the interface supports multicast, join the "all hosts"
1046 	 * multicast group on that interface.
1047 	 */
1048 	if ((ifp->if_flags & IFF_MULTICAST) != 0 && ia->ia_allhosts == NULL) {
1049 		struct in_addr addr;
1050 
1051 		addr.s_addr = INADDR_ALLHOSTS_GROUP;
1052 		ia->ia_allhosts = in_addmulti(&addr, ifp);
1053 	}
1054 
1055 	if (hostIsNew && if_do_dad(ifp) &&
1056 	    !in_nullhost(ia->ia_addr.sin_addr) &&
1057 	    ia->ia4_flags & IN_IFF_TENTATIVE)
1058 		ia->ia_dad_start((struct ifaddr *)ia);
1059 
1060 	return (error);
1061 bad:
1062 	splx(s);
1063 	ia->ia_addr = oldaddr;
1064 	return (error);
1065 }
1066 
1067 #define rtinitflags(x) \
1068 	((((x)->ia_ifp->if_flags & (IFF_LOOPBACK | IFF_POINTOPOINT)) != 0) \
1069 	    ? RTF_HOST : 0)
1070 
1071 /*
1072  * add a route to prefix ("connected route" in cisco terminology).
1073  * does nothing if there's some interface address with the same prefix already.
1074  */
1075 static int
in_addprefix(struct in_ifaddr * target,int flags)1076 in_addprefix(struct in_ifaddr *target, int flags)
1077 {
1078 	struct in_ifaddr *ia;
1079 	struct in_addr prefix, mask, p;
1080 	int error;
1081 
1082 	if ((flags & RTF_HOST) != 0)
1083 		prefix = target->ia_dstaddr.sin_addr;
1084 	else {
1085 		prefix = target->ia_addr.sin_addr;
1086 		mask = target->ia_sockmask.sin_addr;
1087 		prefix.s_addr &= mask.s_addr;
1088 	}
1089 
1090 	IN_ADDRLIST_READER_FOREACH(ia) {
1091 		if (rtinitflags(ia))
1092 			p = ia->ia_dstaddr.sin_addr;
1093 		else {
1094 			p = ia->ia_addr.sin_addr;
1095 			p.s_addr &= ia->ia_sockmask.sin_addr.s_addr;
1096 		}
1097 
1098 		if (prefix.s_addr != p.s_addr)
1099 			continue;
1100 
1101 		/*
1102 		 * if we got a matching prefix route inserted by other
1103 		 * interface address, we don't need to bother
1104 		 *
1105 		 * XXX RADIX_MPATH implications here? -dyoung
1106 		 */
1107 		if (ia->ia_flags & IFA_ROUTE)
1108 			return 0;
1109 	}
1110 
1111 	/*
1112 	 * noone seem to have prefix route.  insert it.
1113 	 */
1114 	error = rtinit(&target->ia_ifa, RTM_ADD, flags);
1115 	if (error == 0)
1116 		target->ia_flags |= IFA_ROUTE;
1117 	else if (error == EEXIST) {
1118 		/*
1119 		 * the fact the route already exists is not an error.
1120 		 */
1121 		error = 0;
1122 	}
1123 	return error;
1124 }
1125 
1126 /*
1127  * remove a route to prefix ("connected route" in cisco terminology).
1128  * re-installs the route by using another interface address, if there's one
1129  * with the same prefix (otherwise we lose the route mistakenly).
1130  */
1131 static int
in_scrubprefix(struct in_ifaddr * target)1132 in_scrubprefix(struct in_ifaddr *target)
1133 {
1134 	struct in_ifaddr *ia;
1135 	struct in_addr prefix, mask, p;
1136 	int error;
1137 
1138 	/* If we don't have IFA_ROUTE we should still inform userland */
1139 	if ((target->ia_flags & IFA_ROUTE) == 0)
1140 		return 0;
1141 
1142 	if (rtinitflags(target))
1143 		prefix = target->ia_dstaddr.sin_addr;
1144 	else {
1145 		prefix = target->ia_addr.sin_addr;
1146 		mask = target->ia_sockmask.sin_addr;
1147 		prefix.s_addr &= mask.s_addr;
1148 	}
1149 
1150 	IN_ADDRLIST_READER_FOREACH(ia) {
1151 		if (rtinitflags(ia))
1152 			p = ia->ia_dstaddr.sin_addr;
1153 		else {
1154 			p = ia->ia_addr.sin_addr;
1155 			p.s_addr &= ia->ia_sockmask.sin_addr.s_addr;
1156 		}
1157 
1158 		if (prefix.s_addr != p.s_addr)
1159 			continue;
1160 
1161 		/*
1162 		 * if we got a matching prefix route, move IFA_ROUTE to him
1163 		 */
1164 		if ((ia->ia_flags & IFA_ROUTE) == 0) {
1165 			rtinit(&target->ia_ifa, RTM_DELETE,
1166 			    rtinitflags(target));
1167 			target->ia_flags &= ~IFA_ROUTE;
1168 
1169 			error = rtinit(&ia->ia_ifa, RTM_ADD,
1170 			    rtinitflags(ia) | RTF_UP);
1171 			if (error == 0)
1172 				ia->ia_flags |= IFA_ROUTE;
1173 			return error;
1174 		}
1175 	}
1176 
1177 	/*
1178 	 * noone seem to have prefix route.  remove it.
1179 	 */
1180 	rtinit(&target->ia_ifa, RTM_DELETE, rtinitflags(target));
1181 	target->ia_flags &= ~IFA_ROUTE;
1182 	return 0;
1183 }
1184 
1185 #undef rtinitflags
1186 
1187 /*
1188  * Return 1 if the address might be a local broadcast address.
1189  */
1190 int
in_broadcast(struct in_addr in,struct ifnet * ifp)1191 in_broadcast(struct in_addr in, struct ifnet *ifp)
1192 {
1193 	struct ifaddr *ifa;
1194 
1195 	if (in.s_addr == INADDR_BROADCAST ||
1196 	    in_nullhost(in))
1197 		return 1;
1198 	if ((ifp->if_flags & IFF_BROADCAST) == 0)
1199 		return 0;
1200 	/*
1201 	 * Look through the list of addresses for a match
1202 	 * with a broadcast address.
1203 	 */
1204 #define ia (ifatoia(ifa))
1205 	IFADDR_READER_FOREACH(ifa, ifp)
1206 		if (ifa->ifa_addr->sa_family == AF_INET &&
1207 		    !in_hosteq(in, ia->ia_addr.sin_addr) &&
1208 		    (in_hosteq(in, ia->ia_broadaddr.sin_addr) ||
1209 		     in_hosteq(in, ia->ia_netbroadcast) ||
1210 		     (hostzeroisbroadcast &&
1211 		      /*
1212 		       * Check for old-style (host 0) broadcast.
1213 		       */
1214 		      (in.s_addr == ia->ia_subnet ||
1215 		       in.s_addr == ia->ia_net))))
1216 			return 1;
1217 	return (0);
1218 #undef ia
1219 }
1220 
1221 /*
1222  * perform DAD when interface becomes IFF_UP.
1223  */
1224 void
in_if_link_up(struct ifnet * ifp)1225 in_if_link_up(struct ifnet *ifp)
1226 {
1227 	struct ifaddr *ifa;
1228 	struct in_ifaddr *ia;
1229 
1230 	/* Ensure it's sane to run DAD */
1231 	if (ifp->if_link_state == LINK_STATE_DOWN)
1232 		return;
1233 	if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
1234 		return;
1235 
1236 	IFADDR_READER_FOREACH(ifa, ifp) {
1237 		if (ifa->ifa_addr->sa_family != AF_INET)
1238 			continue;
1239 		ia = (struct in_ifaddr *)ifa;
1240 
1241 		/* If detached then mark as tentative */
1242 		if (ia->ia4_flags & IN_IFF_DETACHED) {
1243 			ia->ia4_flags &= ~IN_IFF_DETACHED;
1244 			if (if_do_dad(ifp) && ia->ia_dad_start != NULL)
1245 				ia->ia4_flags |= IN_IFF_TENTATIVE;
1246 			else if ((ia->ia4_flags & IN_IFF_TENTATIVE) == 0)
1247 				rt_newaddrmsg(RTM_NEWADDR, ifa, 0, NULL);
1248 		}
1249 
1250 		if (ia->ia4_flags & IN_IFF_TENTATIVE) {
1251 			/* Clear the duplicated flag as we're starting DAD. */
1252 			ia->ia4_flags &= ~IN_IFF_DUPLICATED;
1253 			ia->ia_dad_start(ifa);
1254 		}
1255 	}
1256 }
1257 
1258 void
in_if_up(struct ifnet * ifp)1259 in_if_up(struct ifnet *ifp)
1260 {
1261 
1262 	/* interface may not support link state, so bring it up also */
1263 	in_if_link_up(ifp);
1264 }
1265 
1266 /*
1267  * Mark all addresses as detached.
1268  */
1269 void
in_if_link_down(struct ifnet * ifp)1270 in_if_link_down(struct ifnet *ifp)
1271 {
1272 	struct ifaddr *ifa;
1273 	struct in_ifaddr *ia;
1274 
1275 	IFADDR_READER_FOREACH(ifa, ifp) {
1276 		if (ifa->ifa_addr->sa_family != AF_INET)
1277 			continue;
1278 		ia = (struct in_ifaddr *)ifa;
1279 
1280 		/* Stop DAD processing */
1281 		if (ia->ia_dad_stop != NULL)
1282 			ia->ia_dad_stop(ifa);
1283 
1284 		/*
1285 		 * Mark the address as detached.
1286 		 */
1287 		if (!(ia->ia4_flags & IN_IFF_DETACHED)) {
1288 			ia->ia4_flags |= IN_IFF_DETACHED;
1289 			ia->ia4_flags &=
1290 			    ~(IN_IFF_TENTATIVE | IN_IFF_DUPLICATED);
1291 			rt_newaddrmsg(RTM_NEWADDR, ifa, 0, NULL);
1292 		}
1293 	}
1294 }
1295 
1296 void
in_if_down(struct ifnet * ifp)1297 in_if_down(struct ifnet *ifp)
1298 {
1299 
1300 	in_if_link_down(ifp);
1301 }
1302 
1303 void
in_if_link_state_change(struct ifnet * ifp,int link_state)1304 in_if_link_state_change(struct ifnet *ifp, int link_state)
1305 {
1306 
1307 	switch (link_state) {
1308 	case LINK_STATE_DOWN:
1309 		in_if_link_down(ifp);
1310 		break;
1311 	case LINK_STATE_UP:
1312 		in_if_link_up(ifp);
1313 		break;
1314 	}
1315 }
1316 
1317 /*
1318  * in_lookup_multi: look up the in_multi record for a given IP
1319  * multicast address on a given interface.  If no matching record is
1320  * found, return NULL.
1321  */
1322 struct in_multi *
in_lookup_multi(struct in_addr addr,ifnet_t * ifp)1323 in_lookup_multi(struct in_addr addr, ifnet_t *ifp)
1324 {
1325 	struct in_multi *inm;
1326 
1327 	KASSERT(rw_lock_held(&in_multilock));
1328 
1329 	LIST_FOREACH(inm, &IN_MULTI_HASH(addr.s_addr, ifp), inm_list) {
1330 		if (in_hosteq(inm->inm_addr, addr) && inm->inm_ifp == ifp)
1331 			break;
1332 	}
1333 	return inm;
1334 }
1335 
1336 /*
1337  * in_multi_group: check whether the address belongs to an IP multicast
1338  * group we are joined on this interface.  Returns true or false.
1339  */
1340 bool
in_multi_group(struct in_addr addr,ifnet_t * ifp,int flags)1341 in_multi_group(struct in_addr addr, ifnet_t *ifp, int flags)
1342 {
1343 	bool ingroup;
1344 
1345 	if (__predict_true(flags & IP_IGMP_MCAST) == 0) {
1346 		rw_enter(&in_multilock, RW_READER);
1347 		ingroup = in_lookup_multi(addr, ifp) != NULL;
1348 		rw_exit(&in_multilock);
1349 	} else {
1350 		/* XXX Recursive call from ip_output(). */
1351 		KASSERT(rw_lock_held(&in_multilock));
1352 		ingroup = in_lookup_multi(addr, ifp) != NULL;
1353 	}
1354 	return ingroup;
1355 }
1356 
1357 /*
1358  * Add an address to the list of IP multicast addresses for a given interface.
1359  */
1360 struct in_multi *
in_addmulti(struct in_addr * ap,ifnet_t * ifp)1361 in_addmulti(struct in_addr *ap, ifnet_t *ifp)
1362 {
1363 	struct sockaddr_in sin;
1364 	struct in_multi *inm;
1365 
1366 	/*
1367 	 * See if address already in list.
1368 	 */
1369 	rw_enter(&in_multilock, RW_WRITER);
1370 	inm = in_lookup_multi(*ap, ifp);
1371 	if (inm != NULL) {
1372 		/*
1373 		 * Found it; just increment the reference count.
1374 		 */
1375 		inm->inm_refcount++;
1376 		rw_exit(&in_multilock);
1377 		return inm;
1378 	}
1379 
1380 	/*
1381 	 * New address; allocate a new multicast record.
1382 	 */
1383 	inm = pool_get(&inmulti_pool, PR_NOWAIT);
1384 	if (inm == NULL) {
1385 		rw_exit(&in_multilock);
1386 		return NULL;
1387 	}
1388 	inm->inm_addr = *ap;
1389 	inm->inm_ifp = ifp;
1390 	inm->inm_refcount = 1;
1391 
1392 	/*
1393 	 * Ask the network driver to update its multicast reception
1394 	 * filter appropriately for the new address.
1395 	 */
1396 	sockaddr_in_init(&sin, ap, 0);
1397 	if (if_mcast_op(ifp, SIOCADDMULTI, sintosa(&sin)) != 0) {
1398 		rw_exit(&in_multilock);
1399 		pool_put(&inmulti_pool, inm);
1400 		return NULL;
1401 	}
1402 
1403 	/*
1404 	 * Let IGMP know that we have joined a new IP multicast group.
1405 	 */
1406 	if (igmp_joingroup(inm) != 0) {
1407 		rw_exit(&in_multilock);
1408 		pool_put(&inmulti_pool, inm);
1409 		return NULL;
1410 	}
1411 	LIST_INSERT_HEAD(
1412 	    &IN_MULTI_HASH(inm->inm_addr.s_addr, ifp),
1413 	    inm, inm_list);
1414 	in_multientries++;
1415 	rw_exit(&in_multilock);
1416 
1417 	return inm;
1418 }
1419 
1420 /*
1421  * Delete a multicast address record.
1422  */
1423 void
in_delmulti(struct in_multi * inm)1424 in_delmulti(struct in_multi *inm)
1425 {
1426 	struct sockaddr_in sin;
1427 
1428 	rw_enter(&in_multilock, RW_WRITER);
1429 	if (--inm->inm_refcount > 0) {
1430 		rw_exit(&in_multilock);
1431 		return;
1432 	}
1433 
1434 	/*
1435 	 * No remaining claims to this record; let IGMP know that
1436 	 * we are leaving the multicast group.
1437 	 */
1438 	igmp_leavegroup(inm);
1439 
1440 	/*
1441 	 * Notify the network driver to update its multicast reception
1442 	 * filter.
1443 	 */
1444 	sockaddr_in_init(&sin, &inm->inm_addr, 0);
1445 	if_mcast_op(inm->inm_ifp, SIOCDELMULTI, sintosa(&sin));
1446 
1447 	/*
1448 	 * Unlink from list.
1449 	 */
1450 	LIST_REMOVE(inm, inm_list);
1451 	in_multientries--;
1452 	rw_exit(&in_multilock);
1453 
1454 	pool_put(&inmulti_pool, inm);
1455 }
1456 
1457 /*
1458  * in_next_multi: step through all of the in_multi records, one at a time.
1459  * The current position is remembered in "step", which the caller must
1460  * provide.  in_first_multi(), below, must be called to initialize "step"
1461  * and get the first record.  Both macros return a NULL "inm" when there
1462  * are no remaining records.
1463  */
1464 struct in_multi *
in_next_multi(struct in_multistep * step)1465 in_next_multi(struct in_multistep *step)
1466 {
1467 	struct in_multi *inm;
1468 
1469 	KASSERT(rw_lock_held(&in_multilock));
1470 
1471 	while (step->i_inm == NULL && step->i_n < IN_MULTI_HASH_SIZE) {
1472 		step->i_inm = LIST_FIRST(&in_multihashtbl[++step->i_n]);
1473 	}
1474 	if ((inm = step->i_inm) != NULL) {
1475 		step->i_inm = LIST_NEXT(inm, inm_list);
1476 	}
1477 	return inm;
1478 }
1479 
1480 struct in_multi *
in_first_multi(struct in_multistep * step)1481 in_first_multi(struct in_multistep *step)
1482 {
1483 	KASSERT(rw_lock_held(&in_multilock));
1484 
1485 	step->i_n = 0;
1486 	step->i_inm = LIST_FIRST(&in_multihashtbl[0]);
1487 	return in_next_multi(step);
1488 }
1489 
1490 void
in_multi_lock(int op)1491 in_multi_lock(int op)
1492 {
1493 	rw_enter(&in_multilock, op);
1494 }
1495 
1496 void
in_multi_unlock(void)1497 in_multi_unlock(void)
1498 {
1499 	rw_exit(&in_multilock);
1500 }
1501 
1502 int
in_multi_lock_held(void)1503 in_multi_lock_held(void)
1504 {
1505 	return rw_lock_held(&in_multilock);
1506 }
1507 
1508 struct sockaddr_in *
in_selectsrc(struct sockaddr_in * sin,struct route * ro,int soopts,struct ip_moptions * mopts,int * errorp)1509 in_selectsrc(struct sockaddr_in *sin, struct route *ro,
1510     int soopts, struct ip_moptions *mopts, int *errorp)
1511 {
1512 	struct rtentry *rt = NULL;
1513 	struct in_ifaddr *ia = NULL;
1514 
1515 	/*
1516          * If route is known or can be allocated now, take the
1517          * source address from the interface.  Otherwise, punt.
1518 	 */
1519 	if ((soopts & SO_DONTROUTE) != 0)
1520 		rtcache_free(ro);
1521 	else {
1522 		union {
1523 			struct sockaddr		dst;
1524 			struct sockaddr_in	dst4;
1525 		} u;
1526 
1527 		sockaddr_in_init(&u.dst4, &sin->sin_addr, 0);
1528 		rt = rtcache_lookup(ro, &u.dst);
1529 	}
1530 	/*
1531 	 * If we found a route, use the address
1532 	 * corresponding to the outgoing interface
1533 	 * unless it is the loopback (in case a route
1534 	 * to our address on another net goes to loopback).
1535 	 *
1536 	 * XXX Is this still true?  Do we care?
1537 	 */
1538 	if (rt != NULL && (rt->rt_ifp->if_flags & IFF_LOOPBACK) == 0)
1539 		ia = ifatoia(rt->rt_ifa);
1540 	if (ia == NULL) {
1541 		u_int16_t fport = sin->sin_port;
1542 
1543 		sin->sin_port = 0;
1544 		ia = ifatoia(ifa_ifwithladdr(sintosa(sin)));
1545 		sin->sin_port = fport;
1546 		if (ia == NULL) {
1547 			/* Find 1st non-loopback AF_INET address */
1548 			IN_ADDRLIST_READER_FOREACH(ia) {
1549 				if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK))
1550 					break;
1551 			}
1552 		}
1553 		if (ia == NULL) {
1554 			*errorp = EADDRNOTAVAIL;
1555 			return NULL;
1556 		}
1557 	}
1558 	/*
1559 	 * If the destination address is multicast and an outgoing
1560 	 * interface has been set as a multicast option, use the
1561 	 * address of that interface as our source address.
1562 	 */
1563 	if (IN_MULTICAST(sin->sin_addr.s_addr) && mopts != NULL) {
1564 		struct ip_moptions *imo;
1565 
1566 		imo = mopts;
1567 		if (imo->imo_multicast_if_index != 0) {
1568 			struct ifnet *ifp;
1569 			int s = pserialize_read_enter();
1570 
1571 			ifp = if_byindex(imo->imo_multicast_if_index);
1572 			if (ifp != NULL) {
1573 				ia = in_get_ia_from_ifp(ifp);		/* XXX */
1574 			} else
1575 				ia = NULL;
1576 			if (ia == NULL || ia->ia4_flags & IN_IFF_NOTREADY) {
1577 				pserialize_read_exit(s);
1578 				*errorp = EADDRNOTAVAIL;
1579 				return NULL;
1580 			}
1581 			pserialize_read_exit(s);
1582 		}
1583 	}
1584 	if (ia->ia_ifa.ifa_getifa != NULL) {
1585 		ia = ifatoia((*ia->ia_ifa.ifa_getifa)(&ia->ia_ifa,
1586 		                                      sintosa(sin)));
1587 		if (ia == NULL) {
1588 			*errorp = EADDRNOTAVAIL;
1589 			return NULL;
1590 		}
1591 	}
1592 #ifdef GETIFA_DEBUG
1593 	else
1594 		printf("%s: missing ifa_getifa\n", __func__);
1595 #endif
1596 	return &ia->ia_addr;
1597 }
1598 
1599 #if NARP > 0
1600 
1601 struct in_llentry {
1602 	struct llentry		base;
1603 };
1604 
1605 #define	IN_LLTBL_DEFAULT_HSIZE	32
1606 #define	IN_LLTBL_HASH(k, h) \
1607 	(((((((k >> 8) ^ k) >> 8) ^ k) >> 8) ^ k) & ((h) - 1))
1608 
1609 /*
1610  * Do actual deallocation of @lle.
1611  * Called by LLE_FREE_LOCKED when number of references
1612  * drops to zero.
1613  */
1614 static void
in_lltable_destroy_lle(struct llentry * lle)1615 in_lltable_destroy_lle(struct llentry *lle)
1616 {
1617 
1618 	LLE_WUNLOCK(lle);
1619 	LLE_LOCK_DESTROY(lle);
1620 	kmem_intr_free(lle, sizeof(*lle));
1621 }
1622 
1623 static struct llentry *
in_lltable_new(struct in_addr addr4,u_int flags)1624 in_lltable_new(struct in_addr addr4, u_int flags)
1625 {
1626 	struct in_llentry *lle;
1627 
1628 	lle = kmem_intr_zalloc(sizeof(*lle), KM_NOSLEEP);
1629 	if (lle == NULL)		/* NB: caller generates msg */
1630 		return NULL;
1631 
1632 	/*
1633 	 * For IPv4 this will trigger "arpresolve" to generate
1634 	 * an ARP request.
1635 	 */
1636 	lle->base.la_expire = time_uptime; /* mark expired */
1637 	lle->base.r_l3addr.addr4 = addr4;
1638 	lle->base.lle_refcnt = 1;
1639 	lle->base.lle_free = in_lltable_destroy_lle;
1640 	LLE_LOCK_INIT(&lle->base);
1641 	callout_init(&lle->base.la_timer, CALLOUT_MPSAFE);
1642 
1643 	return (&lle->base);
1644 }
1645 
1646 #define IN_ARE_MASKED_ADDR_EQUAL(d, a, m)	(			\
1647 	    (((ntohl((d).s_addr) ^ (a)->sin_addr.s_addr) & (m)->sin_addr.s_addr)) == 0 )
1648 
1649 static int
in_lltable_match_prefix(const struct sockaddr * prefix,const struct sockaddr * mask,u_int flags,struct llentry * lle)1650 in_lltable_match_prefix(const struct sockaddr *prefix,
1651     const struct sockaddr *mask, u_int flags, struct llentry *lle)
1652 {
1653 	const struct sockaddr_in *pfx = (const struct sockaddr_in *)prefix;
1654 	const struct sockaddr_in *msk = (const struct sockaddr_in *)mask;
1655 
1656 	/*
1657 	 * (flags & LLE_STATIC) means deleting all entries
1658 	 * including static ARP entries.
1659 	 */
1660 	if (IN_ARE_MASKED_ADDR_EQUAL(lle->r_l3addr.addr4, pfx, msk) &&
1661 	    ((flags & LLE_STATIC) || !(lle->la_flags & LLE_STATIC)))
1662 		return (1);
1663 
1664 	return (0);
1665 }
1666 
1667 static void
in_lltable_free_entry(struct lltable * llt,struct llentry * lle)1668 in_lltable_free_entry(struct lltable *llt, struct llentry *lle)
1669 {
1670 	struct ifnet *ifp __diagused;
1671 	size_t pkts_dropped;
1672 
1673 	LLE_WLOCK_ASSERT(lle);
1674 	KASSERT(llt != NULL);
1675 
1676 	/* Unlink entry from table if not already */
1677 	if ((lle->la_flags & LLE_LINKED) != 0) {
1678 		ifp = llt->llt_ifp;
1679 		IF_AFDATA_WLOCK_ASSERT(ifp);
1680 		lltable_unlink_entry(llt, lle);
1681 	}
1682 
1683 	/* cancel timer */
1684 	if (callout_halt(&lle->lle_timer, &lle->lle_lock))
1685 		LLE_REMREF(lle);
1686 
1687 	/* Drop hold queue */
1688 	pkts_dropped = llentry_free(lle);
1689 	arp_stat_add(ARP_STAT_DFRDROPPED, (uint64_t)pkts_dropped);
1690 }
1691 
1692 static int
in_lltable_rtcheck(struct ifnet * ifp,u_int flags,const struct sockaddr * l3addr)1693 in_lltable_rtcheck(struct ifnet *ifp, u_int flags, const struct sockaddr *l3addr)
1694 {
1695 	struct rtentry *rt;
1696 	int error = EINVAL;
1697 
1698 	KASSERTMSG(l3addr->sa_family == AF_INET,
1699 	    "sin_family %d", l3addr->sa_family);
1700 
1701 	rt = rtalloc1(l3addr, 0);
1702 	if (rt == NULL)
1703 		return error;
1704 
1705 	/*
1706 	 * If the gateway for an existing host route matches the target L3
1707 	 * address, which is a special route inserted by some implementation
1708 	 * such as MANET, and the interface is of the correct type, then
1709 	 * allow for ARP to proceed.
1710 	 */
1711 	if (rt->rt_flags & RTF_GATEWAY) {
1712 		if (!(rt->rt_flags & RTF_HOST) || !rt->rt_ifp ||
1713 		    rt->rt_ifp->if_type != IFT_ETHER ||
1714 #ifdef __FreeBSD__
1715 		    (rt->rt_ifp->if_flags & (IFF_NOARP | IFF_STATICARP)) != 0 ||
1716 #else
1717 		    (rt->rt_ifp->if_flags & IFF_NOARP) != 0 ||
1718 #endif
1719 		    memcmp(rt->rt_gateway->sa_data, l3addr->sa_data,
1720 		    sizeof(in_addr_t)) != 0) {
1721 			goto error;
1722 		}
1723 	}
1724 
1725 	/*
1726 	 * Make sure that at least the destination address is covered
1727 	 * by the route. This is for handling the case where 2 or more
1728 	 * interfaces have the same prefix. An incoming packet arrives
1729 	 * on one interface and the corresponding outgoing packet leaves
1730 	 * another interface.
1731 	 */
1732 	if (!(rt->rt_flags & RTF_HOST) && rt->rt_ifp != ifp) {
1733 		const char *sa, *mask, *addr, *lim;
1734 		int len;
1735 
1736 		mask = (const char *)rt_mask(rt);
1737 		/*
1738 		 * Just being extra cautious to avoid some custom
1739 		 * code getting into trouble.
1740 		 */
1741 		if (mask == NULL)
1742 			goto error;
1743 
1744 		sa = (const char *)rt_getkey(rt);
1745 		addr = (const char *)l3addr;
1746 		len = ((const struct sockaddr_in *)l3addr)->sin_len;
1747 		lim = addr + len;
1748 
1749 		for ( ; addr < lim; sa++, mask++, addr++) {
1750 			if ((*sa ^ *addr) & *mask) {
1751 #ifdef DIAGNOSTIC
1752 				log(LOG_INFO, "IPv4 address: \"%s\" is not on the network\n",
1753 				    inet_ntoa(((const struct sockaddr_in *)l3addr)->sin_addr));
1754 #endif
1755 				goto error;
1756 			}
1757 		}
1758 	}
1759 
1760 	error = 0;
1761 error:
1762 	rtfree(rt);
1763 	return error;
1764 }
1765 
1766 static inline uint32_t
in_lltable_hash_dst(const struct in_addr dst,uint32_t hsize)1767 in_lltable_hash_dst(const struct in_addr dst, uint32_t hsize)
1768 {
1769 
1770 	return (IN_LLTBL_HASH(dst.s_addr, hsize));
1771 }
1772 
1773 static uint32_t
in_lltable_hash(const struct llentry * lle,uint32_t hsize)1774 in_lltable_hash(const struct llentry *lle, uint32_t hsize)
1775 {
1776 
1777 	return (in_lltable_hash_dst(lle->r_l3addr.addr4, hsize));
1778 }
1779 
1780 static void
in_lltable_fill_sa_entry(const struct llentry * lle,struct sockaddr * sa)1781 in_lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa)
1782 {
1783 	struct sockaddr_in *sin;
1784 
1785 	sin = (struct sockaddr_in *)sa;
1786 	memset(sin, 0, sizeof(*sin));
1787 	sin->sin_family = AF_INET;
1788 	sin->sin_len = sizeof(*sin);
1789 	sin->sin_addr = lle->r_l3addr.addr4;
1790 }
1791 
1792 static inline struct llentry *
in_lltable_find_dst(struct lltable * llt,struct in_addr dst)1793 in_lltable_find_dst(struct lltable *llt, struct in_addr dst)
1794 {
1795 	struct llentry *lle;
1796 	struct llentries *lleh;
1797 	u_int hashidx;
1798 
1799 	hashidx = in_lltable_hash_dst(dst, llt->llt_hsize);
1800 	lleh = &llt->lle_head[hashidx];
1801 	LIST_FOREACH(lle, lleh, lle_next) {
1802 		if (lle->la_flags & LLE_DELETED)
1803 			continue;
1804 		if (lle->r_l3addr.addr4.s_addr == dst.s_addr)
1805 			break;
1806 	}
1807 
1808 	return (lle);
1809 }
1810 
1811 static int
in_lltable_delete(struct lltable * llt,u_int flags,const struct sockaddr * l3addr)1812 in_lltable_delete(struct lltable *llt, u_int flags,
1813     const struct sockaddr *l3addr)
1814 {
1815 	const struct sockaddr_in *sin = (const struct sockaddr_in *)l3addr;
1816 	struct ifnet *ifp __diagused = llt->llt_ifp;
1817 	struct llentry *lle;
1818 
1819 	IF_AFDATA_WLOCK_ASSERT(ifp);
1820 	KASSERTMSG(l3addr->sa_family == AF_INET,
1821 	    "sin_family %d", l3addr->sa_family);
1822 
1823 	lle = in_lltable_find_dst(llt, sin->sin_addr);
1824 	if (lle == NULL) {
1825 #ifdef DIAGNOSTIC
1826 		log(LOG_INFO, "interface address is missing from cache = %p  in delete\n", lle);
1827 #endif
1828 		return (ENOENT);
1829 	}
1830 
1831 	LLE_WLOCK(lle);
1832 	lle->la_flags |= LLE_DELETED;
1833 #ifdef DIAGNOSTIC
1834 	log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle);
1835 #endif
1836 	if ((lle->la_flags & (LLE_STATIC | LLE_IFADDR)) == LLE_STATIC)
1837 		llentry_free(lle);
1838 	else
1839 		LLE_WUNLOCK(lle);
1840 
1841 	return (0);
1842 }
1843 
1844 static struct llentry *
in_lltable_create(struct lltable * llt,u_int flags,const struct sockaddr * l3addr)1845 in_lltable_create(struct lltable *llt, u_int flags, const struct sockaddr *l3addr)
1846 {
1847 	const struct sockaddr_in *sin = (const struct sockaddr_in *)l3addr;
1848 	struct ifnet *ifp = llt->llt_ifp;
1849 	struct llentry *lle;
1850 
1851 	IF_AFDATA_WLOCK_ASSERT(ifp);
1852 	KASSERTMSG(l3addr->sa_family == AF_INET,
1853 	    "sin_family %d", l3addr->sa_family);
1854 
1855 	lle = in_lltable_find_dst(llt, sin->sin_addr);
1856 
1857 	if (lle != NULL) {
1858 		LLE_WLOCK(lle);
1859 		return (lle);
1860 	}
1861 
1862 	/* no existing record, we need to create new one */
1863 
1864 	/*
1865 	 * A route that covers the given address must have
1866 	 * been installed 1st because we are doing a resolution,
1867 	 * verify this.
1868 	 */
1869 	if (!(flags & LLE_IFADDR) &&
1870 	    in_lltable_rtcheck(ifp, flags, l3addr) != 0)
1871 		return (NULL);
1872 
1873 	lle = in_lltable_new(sin->sin_addr, flags);
1874 	if (lle == NULL) {
1875 		log(LOG_INFO, "lla_lookup: new lle malloc failed\n");
1876 		return (NULL);
1877 	}
1878 	lle->la_flags = flags;
1879 	if ((flags & LLE_IFADDR) == LLE_IFADDR) {
1880 		memcpy(&lle->ll_addr, CLLADDR(ifp->if_sadl), ifp->if_addrlen);
1881 		lle->la_flags |= (LLE_VALID | LLE_STATIC);
1882 	}
1883 
1884 	lltable_link_entry(llt, lle);
1885 	LLE_WLOCK(lle);
1886 
1887 	return (lle);
1888 }
1889 
1890 /*
1891  * Return NULL if not found or marked for deletion.
1892  * If found return lle read locked.
1893  */
1894 static struct llentry *
in_lltable_lookup(struct lltable * llt,u_int flags,const struct sockaddr * l3addr)1895 in_lltable_lookup(struct lltable *llt, u_int flags, const struct sockaddr *l3addr)
1896 {
1897 	const struct sockaddr_in *sin = (const struct sockaddr_in *)l3addr;
1898 	struct llentry *lle;
1899 
1900 	IF_AFDATA_LOCK_ASSERT(llt->llt_ifp);
1901 	KASSERTMSG(l3addr->sa_family == AF_INET,
1902 	    "sin_family %d", l3addr->sa_family);
1903 
1904 	lle = in_lltable_find_dst(llt, sin->sin_addr);
1905 
1906 	if (lle == NULL)
1907 		return NULL;
1908 
1909 	if (flags & LLE_EXCLUSIVE)
1910 		LLE_WLOCK(lle);
1911 	else
1912 		LLE_RLOCK(lle);
1913 
1914 	return lle;
1915 }
1916 
1917 static int
in_lltable_dump_entry(struct lltable * llt,struct llentry * lle,struct rt_walkarg * w)1918 in_lltable_dump_entry(struct lltable *llt, struct llentry *lle,
1919     struct rt_walkarg *w)
1920 {
1921 	struct sockaddr_in sin;
1922 
1923 	LLTABLE_LOCK_ASSERT();
1924 
1925 	/* skip deleted entries */
1926 	if (lle->la_flags & LLE_DELETED)
1927 		return 0;
1928 
1929 	sockaddr_in_init(&sin, &lle->r_l3addr.addr4, 0);
1930 
1931 	return lltable_dump_entry(llt, lle, w, sintosa(&sin));
1932 }
1933 
1934 #endif /* NARP > 0 */
1935 
1936 static int
in_multicast_sysctl(SYSCTLFN_ARGS)1937 in_multicast_sysctl(SYSCTLFN_ARGS)
1938 {
1939 	struct ifnet *ifp;
1940 	struct ifaddr *ifa;
1941 	struct in_ifaddr *ifa4;
1942 	struct in_multi *inm;
1943 	uint32_t tmp;
1944 	int error;
1945 	size_t written;
1946 	struct psref psref;
1947 	int bound;
1948 
1949 	if (namelen != 1)
1950 		return EINVAL;
1951 
1952 	bound = curlwp_bind();
1953 	ifp = if_get_byindex(name[0], &psref);
1954 	if (ifp == NULL) {
1955 		curlwp_bindx(bound);
1956 		return ENODEV;
1957 	}
1958 
1959 	if (oldp == NULL) {
1960 		*oldlenp = 0;
1961 		IFADDR_FOREACH(ifa, ifp) {
1962 			if (ifa->ifa_addr->sa_family != AF_INET)
1963 				continue;
1964 			ifa4 = (void *)ifa;
1965 			LIST_FOREACH(inm, &ifa4->ia_multiaddrs, inm_list) {
1966 				*oldlenp += 2 * sizeof(struct in_addr) +
1967 				    sizeof(uint32_t);
1968 			}
1969 		}
1970 		if_put(ifp, &psref);
1971 		curlwp_bindx(bound);
1972 		return 0;
1973 	}
1974 
1975 	error = 0;
1976 	written = 0;
1977 	IFADDR_FOREACH(ifa, ifp) {
1978 		if (ifa->ifa_addr->sa_family != AF_INET)
1979 			continue;
1980 		ifa4 = (void *)ifa;
1981 		LIST_FOREACH(inm, &ifa4->ia_multiaddrs, inm_list) {
1982 			if (written + 2 * sizeof(struct in_addr) +
1983 			    sizeof(uint32_t) > *oldlenp)
1984 				goto done;
1985 			error = sysctl_copyout(l, &ifa4->ia_addr.sin_addr,
1986 			    oldp, sizeof(struct in_addr));
1987 			if (error)
1988 				goto done;
1989 			oldp = (char *)oldp + sizeof(struct in_addr);
1990 			written += sizeof(struct in_addr);
1991 			error = sysctl_copyout(l, &inm->inm_addr,
1992 			    oldp, sizeof(struct in_addr));
1993 			if (error)
1994 				goto done;
1995 			oldp = (char *)oldp + sizeof(struct in_addr);
1996 			written += sizeof(struct in_addr);
1997 			tmp = inm->inm_refcount;
1998 			error = sysctl_copyout(l, &tmp, oldp, sizeof(tmp));
1999 			if (error)
2000 				goto done;
2001 			oldp = (char *)oldp + sizeof(tmp);
2002 			written += sizeof(tmp);
2003 		}
2004 	}
2005 done:
2006 	if_put(ifp, &psref);
2007 	curlwp_bindx(bound);
2008 	*oldlenp = written;
2009 	return error;
2010 }
2011 
2012 static void
in_sysctl_init(struct sysctllog ** clog)2013 in_sysctl_init(struct sysctllog **clog)
2014 {
2015 	sysctl_createv(clog, 0, NULL, NULL,
2016 		       CTLFLAG_PERMANENT,
2017 		       CTLTYPE_NODE, "inet",
2018 		       SYSCTL_DESCR("PF_INET related settings"),
2019 		       NULL, 0, NULL, 0,
2020 		       CTL_NET, PF_INET, CTL_EOL);
2021 	sysctl_createv(clog, 0, NULL, NULL,
2022 		       CTLFLAG_PERMANENT,
2023 		       CTLTYPE_NODE, "multicast",
2024 		       SYSCTL_DESCR("Multicast information"),
2025 		       in_multicast_sysctl, 0, NULL, 0,
2026 		       CTL_NET, PF_INET, CTL_CREATE, CTL_EOL);
2027 	sysctl_createv(clog, 0, NULL, NULL,
2028 		       CTLFLAG_PERMANENT,
2029 		       CTLTYPE_NODE, "ip",
2030 		       SYSCTL_DESCR("IPv4 related settings"),
2031 		       NULL, 0, NULL, 0,
2032 		       CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL);
2033 
2034 	sysctl_createv(clog, 0, NULL, NULL,
2035 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2036 		       CTLTYPE_INT, "subnetsarelocal",
2037 		       SYSCTL_DESCR("Whether logical subnets are considered "
2038 				    "local"),
2039 		       NULL, 0, &subnetsarelocal, 0,
2040 		       CTL_NET, PF_INET, IPPROTO_IP,
2041 		       IPCTL_SUBNETSARELOCAL, CTL_EOL);
2042 	sysctl_createv(clog, 0, NULL, NULL,
2043 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2044 		       CTLTYPE_INT, "hostzerobroadcast",
2045 		       SYSCTL_DESCR("All zeroes address is broadcast address"),
2046 		       NULL, 0, &hostzeroisbroadcast, 0,
2047 		       CTL_NET, PF_INET, IPPROTO_IP,
2048 		       IPCTL_HOSTZEROBROADCAST, CTL_EOL);
2049 }
2050 
2051 #if NARP > 0
2052 
2053 static struct lltable *
in_lltattach(struct ifnet * ifp)2054 in_lltattach(struct ifnet *ifp)
2055 {
2056 	struct lltable *llt;
2057 
2058 	llt = lltable_allocate_htbl(IN_LLTBL_DEFAULT_HSIZE);
2059 	llt->llt_af = AF_INET;
2060 	llt->llt_ifp = ifp;
2061 
2062 	llt->llt_lookup = in_lltable_lookup;
2063 	llt->llt_create = in_lltable_create;
2064 	llt->llt_delete = in_lltable_delete;
2065 	llt->llt_dump_entry = in_lltable_dump_entry;
2066 	llt->llt_hash = in_lltable_hash;
2067 	llt->llt_fill_sa_entry = in_lltable_fill_sa_entry;
2068 	llt->llt_free_entry = in_lltable_free_entry;
2069 	llt->llt_match_prefix = in_lltable_match_prefix;
2070 	lltable_link(llt);
2071 
2072 	return (llt);
2073 }
2074 
2075 #endif /* NARP > 0 */
2076 
2077 void *
in_domifattach(struct ifnet * ifp)2078 in_domifattach(struct ifnet *ifp)
2079 {
2080 	struct in_ifinfo *ii;
2081 
2082 	ii = kmem_zalloc(sizeof(struct in_ifinfo), KM_SLEEP);
2083 	KASSERT(ii != NULL);
2084 
2085 #if NARP > 0
2086 	ii->ii_llt = in_lltattach(ifp);
2087 #endif
2088 
2089 #ifdef IPSELSRC
2090 	ii->ii_selsrc = in_selsrc_domifattach(ifp);
2091 	KASSERT(ii->ii_selsrc != NULL);
2092 #endif
2093 
2094 	return ii;
2095 }
2096 
2097 void
in_domifdetach(struct ifnet * ifp,void * aux)2098 in_domifdetach(struct ifnet *ifp, void *aux)
2099 {
2100 	struct in_ifinfo *ii = aux;
2101 
2102 #ifdef IPSELSRC
2103 	in_selsrc_domifdetach(ifp, ii->ii_selsrc);
2104 #endif
2105 #if NARP > 0
2106 	lltable_free(ii->ii_llt);
2107 #endif
2108 	kmem_free(ii, sizeof(struct in_ifinfo));
2109 }
2110