Home | History | Annotate | Line # | Download | only in netinet
in_pcb.c revision 1.87
      1 /*	$NetBSD: in_pcb.c,v 1.87 2003/08/15 03:42:01 jonathan 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  * 3. All advertising materials mentioning features or use of this software
     49  *    must display the following acknowledgement:
     50  *	This product includes software developed by the NetBSD
     51  *	Foundation, Inc. and its contributors.
     52  * 4. Neither the name of The NetBSD Foundation nor the names of its
     53  *    contributors may be used to endorse or promote products derived
     54  *    from this software without specific prior written permission.
     55  *
     56  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
     57  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     58  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     59  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
     60  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     61  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     62  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     63  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     64  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     65  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     66  * POSSIBILITY OF SUCH DAMAGE.
     67  */
     68 
     69 /*
     70  * Copyright (c) 1982, 1986, 1991, 1993, 1995
     71  *	The Regents of the University of California.  All rights reserved.
     72  *
     73  * Redistribution and use in source and binary forms, with or without
     74  * modification, are permitted provided that the following conditions
     75  * are met:
     76  * 1. Redistributions of source code must retain the above copyright
     77  *    notice, this list of conditions and the following disclaimer.
     78  * 2. Redistributions in binary form must reproduce the above copyright
     79  *    notice, this list of conditions and the following disclaimer in the
     80  *    documentation and/or other materials provided with the distribution.
     81  * 3. Neither the name of the University nor the names of its contributors
     82  *    may be used to endorse or promote products derived from this software
     83  *    without specific prior written permission.
     84  *
     85  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     86  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     87  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     88  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     89  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     90  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     91  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     92  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     93  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     94  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     95  * SUCH DAMAGE.
     96  *
     97  *	@(#)in_pcb.c	8.4 (Berkeley) 5/24/95
     98  */
     99 
    100 #include <sys/cdefs.h>
    101 __KERNEL_RCSID(0, "$NetBSD: in_pcb.c,v 1.87 2003/08/15 03:42:01 jonathan Exp $");
    102 
    103 #include "opt_ipsec.h"
    104 
    105 #include <sys/param.h>
    106 #include <sys/systm.h>
    107 #include <sys/malloc.h>
    108 #include <sys/mbuf.h>
    109 #include <sys/protosw.h>
    110 #include <sys/socket.h>
    111 #include <sys/socketvar.h>
    112 #include <sys/ioctl.h>
    113 #include <sys/errno.h>
    114 #include <sys/time.h>
    115 #include <sys/pool.h>
    116 #include <sys/proc.h>
    117 
    118 #include <net/if.h>
    119 #include <net/route.h>
    120 
    121 #include <netinet/in.h>
    122 #include <netinet/in_systm.h>
    123 #include <netinet/ip.h>
    124 #include <netinet/in_pcb.h>
    125 #include <netinet/in_var.h>
    126 #include <netinet/ip_var.h>
    127 
    128 #ifdef IPSEC
    129 #include <netinet6/ipsec.h>
    130 #include <netkey/key.h>
    131 #elif FAST_IPSEC
    132 #include <netipsec/ipsec.h>
    133 #include <netipsec/key.h>
    134 #endif /* IPSEC */
    135 
    136 struct	in_addr zeroin_addr;
    137 
    138 #define	INPCBHASH_BIND(table, laddr, lport) \
    139 	&(table)->inpt_bindhashtbl[ \
    140 	    ((ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_bindhash]
    141 #define	INPCBHASH_CONNECT(table, faddr, fport, laddr, lport) \
    142 	&(table)->inpt_connecthashtbl[ \
    143 	    ((ntohl((faddr).s_addr) + ntohs(fport)) + \
    144 	     (ntohl((laddr).s_addr) + ntohs(lport))) & (table)->inpt_connecthash]
    145 
    146 struct inpcb *
    147 	in_pcblookup_port __P((struct inpcbtable *,
    148 	    struct in_addr, u_int, int));
    149 
    150 int	anonportmin = IPPORT_ANONMIN;
    151 int	anonportmax = IPPORT_ANONMAX;
    152 int	lowportmin  = IPPORT_RESERVEDMIN;
    153 int	lowportmax  = IPPORT_RESERVEDMAX;
    154 
    155 struct pool inpcb_pool;
    156 
    157 void
    158 in_pcbinit(table, bindhashsize, connecthashsize)
    159 	struct inpcbtable *table;
    160 	int bindhashsize, connecthashsize;
    161 {
    162 	static int inpcb_pool_initialized;
    163 
    164 	if (inpcb_pool_initialized == 0) {
    165 		pool_init(&inpcb_pool, sizeof(struct inpcb), 0, 0, 0,
    166 		    "inpcbpl", NULL);
    167 		inpcb_pool_initialized = 1;
    168 	}
    169 
    170 	CIRCLEQ_INIT(&table->inpt_queue);
    171 	table->inpt_bindhashtbl = hashinit(bindhashsize, HASH_LIST, M_PCB,
    172 	    M_WAITOK, &table->inpt_bindhash);
    173 	table->inpt_connecthashtbl = hashinit(connecthashsize, HASH_LIST,
    174 	    M_PCB, M_WAITOK, &table->inpt_connecthash);
    175 	table->inpt_lastlow = IPPORT_RESERVEDMAX;
    176 	table->inpt_lastport = (u_int16_t)anonportmax;
    177 }
    178 
    179 int
    180 in_pcballoc(so, v)
    181 	struct socket *so;
    182 	void *v;
    183 {
    184 	struct inpcbtable *table = v;
    185 	struct inpcb *inp;
    186 	int s;
    187 #if defined(IPSEC) || defined(FAST_IPSEC)
    188 	int error;
    189 #endif
    190 
    191 	inp = pool_get(&inpcb_pool, PR_NOWAIT);
    192 	if (inp == NULL)
    193 		return (ENOBUFS);
    194 	bzero((caddr_t)inp, sizeof(*inp));
    195 	inp->inp_table = table;
    196 	inp->inp_socket = so;
    197 	inp->inp_errormtu = -1;
    198 #if defined(IPSEC) || defined(FAST_IPSEC)
    199 	error = ipsec_init_pcbpolicy(so, &inp->inp_sp);
    200 	if (error != 0) {
    201 		pool_put(&inpcb_pool, inp);
    202 		return error;
    203 	}
    204 #endif
    205 	so->so_pcb = inp;
    206 	s = splnet();
    207 	CIRCLEQ_INSERT_HEAD(&table->inpt_queue, inp, inp_queue);
    208 	in_pcbstate(inp, INP_ATTACHED);
    209 	splx(s);
    210 	return (0);
    211 }
    212 
    213 int
    214 in_pcbbind(v, nam, p)
    215 	void *v;
    216 	struct mbuf *nam;
    217 	struct proc *p;
    218 {
    219 	struct in_ifaddr *ia = NULL;
    220 	struct inpcb *inp = v;
    221 	struct socket *so = inp->inp_socket;
    222 	struct inpcbtable *table = inp->inp_table;
    223 	struct sockaddr_in *sin;
    224 	u_int16_t lport = 0;
    225 	int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
    226 
    227 	if (TAILQ_FIRST(&in_ifaddr) == 0)
    228 		return (EADDRNOTAVAIL);
    229 	if (inp->inp_lport || !in_nullhost(inp->inp_laddr))
    230 		return (EINVAL);
    231 	if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
    232 		wild = 1;
    233 	if (nam == 0)
    234 		goto noname;
    235 	sin = mtod(nam, struct sockaddr_in *);
    236 	if (nam->m_len != sizeof (*sin))
    237 		return (EINVAL);
    238 	if (sin->sin_family != AF_INET)
    239 		return (EAFNOSUPPORT);
    240 	lport = sin->sin_port;
    241 	if (IN_MULTICAST(sin->sin_addr.s_addr)) {
    242 		/*
    243 		 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
    244 		 * allow complete duplication of binding if
    245 		 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
    246 		 * and a multicast address is bound on both
    247 		 * new and duplicated sockets.
    248 		 */
    249 		if (so->so_options & SO_REUSEADDR)
    250 			reuseport = SO_REUSEADDR|SO_REUSEPORT;
    251 	} else if (!in_nullhost(sin->sin_addr)) {
    252 		sin->sin_port = 0;		/* yech... */
    253 		INADDR_TO_IA(sin->sin_addr, ia);
    254 		/* check for broadcast addresses */
    255 		if (ia == NULL)
    256 			ia = ifatoia(ifa_ifwithaddr(sintosa(sin)));
    257 		if (ia == NULL)
    258 			return (EADDRNOTAVAIL);
    259 	}
    260 	if (lport) {
    261 		struct inpcb *t;
    262 #ifndef IPNOPRIVPORTS
    263 		/* GROSS */
    264 		if (ntohs(lport) < IPPORT_RESERVED &&
    265 		    (p == 0 || suser(p->p_ucred, &p->p_acflag)))
    266 			return (EACCES);
    267 #endif
    268 		if (so->so_uid && !IN_MULTICAST(sin->sin_addr.s_addr)) {
    269 			t = in_pcblookup_port(table, sin->sin_addr, lport, 1);
    270 		/*
    271 		 * XXX:	investigate ramifications of loosening this
    272 		 *	restriction so that as long as both ports have
    273 		 *	SO_REUSEPORT allow the bind
    274 		 */
    275 			if (t &&
    276 			    (!in_nullhost(sin->sin_addr) ||
    277 			     !in_nullhost(t->inp_laddr) ||
    278 			     (t->inp_socket->so_options & SO_REUSEPORT) == 0)
    279 			    && (so->so_uid != t->inp_socket->so_uid)) {
    280 				return (EADDRINUSE);
    281 			}
    282 		}
    283 		t = in_pcblookup_port(table, sin->sin_addr, lport, wild);
    284 		if (t && (reuseport & t->inp_socket->so_options) == 0)
    285 			return (EADDRINUSE);
    286 	}
    287 	if (!in_nullhost(inp->inp_laddr)) {
    288 		KASSERT(inp->inp_ia != NULL);
    289 		LIST_REMOVE(inp, inp_ialink);
    290 		IFAFREE(&inp->inp_ia->ia_ifa);
    291 		inp->inp_ia = NULL;
    292 	}
    293 	inp->inp_laddr = sin->sin_addr;
    294 	if (ia != NULL) {
    295 		inp->inp_ia = ia;
    296 		LIST_INSERT_HEAD(&ia->ia_inpcbs, inp, inp_ialink);
    297 		IFAREF(&ia->ia_ifa);
    298 	}
    299 
    300 
    301 noname:
    302 	if (lport == 0) {
    303 		int	   cnt;
    304 		u_int16_t  min, max;
    305 		u_int16_t *lastport;
    306 
    307 		if (inp->inp_flags & INP_LOWPORT) {
    308 #ifndef IPNOPRIVPORTS
    309 			if (p == 0 || suser(p->p_ucred, &p->p_acflag))
    310 				return (EACCES);
    311 #endif
    312 			min = lowportmin;
    313 			max = lowportmax;
    314 			lastport = &table->inpt_lastlow;
    315 		} else {
    316 			min = anonportmin;
    317 			max = anonportmax;
    318 			lastport = &table->inpt_lastport;
    319 		}
    320 		if (min > max) {	/* sanity check */
    321 			u_int16_t swp;
    322 
    323 			swp = min;
    324 			min = max;
    325 			max = swp;
    326 		}
    327 
    328 		lport = *lastport - 1;
    329 		for (cnt = max - min + 1; cnt; cnt--, lport--) {
    330 			if (lport < min || lport > max)
    331 				lport = max;
    332 			if (!in_pcblookup_port(table, inp->inp_laddr,
    333 			    htons(lport), 1))
    334 				goto found;
    335 		}
    336 		if (!in_nullhost(inp->inp_laddr)) {
    337 		       if (inp->inp_ia != NULL) {
    338 				LIST_REMOVE(inp, inp_ialink);
    339 				IFAFREE(&inp->inp_ia->ia_ifa);
    340 				inp->inp_ia = NULL;
    341 			}
    342 			inp->inp_laddr.s_addr = INADDR_ANY;
    343 		}
    344 		return (EAGAIN);
    345 	found:
    346 		inp->inp_flags |= INP_ANONPORT;
    347 		*lastport = lport;
    348 		lport = htons(lport);
    349 	}
    350 	inp->inp_lport = lport;
    351 	in_pcbstate(inp, INP_BOUND);
    352 	return (0);
    353 }
    354 
    355 /*
    356  * Connect from a socket to a specified address.
    357  * Both address and port must be specified in argument sin.
    358  * If don't have a local address for this socket yet,
    359  * then pick one.
    360  */
    361 int
    362 in_pcbconnect(v, nam)
    363 	void *v;
    364 	struct mbuf *nam;
    365 {
    366 	struct inpcb *inp = v;
    367 	struct in_ifaddr *ia = NULL;
    368 	struct sockaddr_in *ifaddr = NULL;
    369 	struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *);
    370 	int error;
    371 
    372 	if (nam->m_len != sizeof (*sin))
    373 		return (EINVAL);
    374 	if (sin->sin_family != AF_INET)
    375 		return (EAFNOSUPPORT);
    376 	if (sin->sin_port == 0)
    377 		return (EADDRNOTAVAIL);
    378 	if (TAILQ_FIRST(&in_ifaddr) != 0) {
    379 		/*
    380 		 * If the destination address is INADDR_ANY,
    381 		 * use any local address (likely loopback).
    382 		 * If the supplied address is INADDR_BROADCAST,
    383 		 * use the broadcast address of an interface
    384 		 * which supports broadcast. (loopback does not)
    385 		 */
    386 
    387 		if (in_nullhost(sin->sin_addr)) {
    388 			sin->sin_addr =
    389 			    TAILQ_FIRST(&in_ifaddr)->ia_addr.sin_addr;
    390 		} else if (sin->sin_addr.s_addr == INADDR_BROADCAST) {
    391 			TAILQ_FOREACH(ia, &in_ifaddr, ia_list) {
    392 				if (ia->ia_ifp->if_flags & IFF_BROADCAST) {
    393 					sin->sin_addr =
    394 					    ia->ia_broadaddr.sin_addr;
    395 					break;
    396 				}
    397 			}
    398 		}
    399 	}
    400 	/*
    401 	 * If we haven't bound which network number to use as ours,
    402 	 * we will use the number of the outgoing interface.
    403 	 * This depends on having done a routing lookup, which
    404 	 * we will probably have to do anyway, so we might
    405 	 * as well do it now.  On the other hand if we are
    406 	 * sending to multiple destinations we may have already
    407 	 * done the lookup, so see if we can use the route
    408 	 * from before.  In any case, we only
    409 	 * chose a port number once, even if sending to multiple
    410 	 * destinations.
    411 	 */
    412 	if (in_nullhost(inp->inp_laddr)) {
    413 		int error;
    414 		ifaddr = in_selectsrc(sin, &inp->inp_route,
    415 			inp->inp_socket->so_options, inp->inp_moptions, &error);
    416 		if (ifaddr == NULL) {
    417 			if (error == 0)
    418 				error = EADDRNOTAVAIL;
    419 			return error;
    420 		}
    421 		INADDR_TO_IA(ifaddr->sin_addr, ia);
    422 		if (ia == NULL)
    423 			return (EADDRNOTAVAIL);
    424 	}
    425 	if (in_pcblookup_connect(inp->inp_table, sin->sin_addr, sin->sin_port,
    426 	    !in_nullhost(inp->inp_laddr) ? inp->inp_laddr : ifaddr->sin_addr,
    427 	    inp->inp_lport) != 0)
    428 		return (EADDRINUSE);
    429 	if (in_nullhost(inp->inp_laddr)) {
    430 		if (inp->inp_lport == 0) {
    431 			error = in_pcbbind(inp, (struct mbuf *)0,
    432 			    (struct proc *)0);
    433 			/*
    434 			 * This used to ignore the return value
    435 			 * completely, but we need to check for
    436 			 * ephemeral port shortage.
    437 			 * XXX Should we check for other errors, too?
    438 			 */
    439 			if (error == EAGAIN)
    440 				return (error);
    441 		}
    442 		KASSERT(inp->inp_ia == NULL);
    443 		inp->inp_laddr = ia->ia_addr.sin_addr;
    444 		inp->inp_ia = ia;
    445 		LIST_INSERT_HEAD(&ia->ia_inpcbs, inp, inp_ialink);
    446 		IFAREF(&ia->ia_ifa);
    447 		inp->inp_laddr = ifaddr->sin_addr;
    448 	}
    449 	inp->inp_faddr = sin->sin_addr;
    450 	inp->inp_fport = sin->sin_port;
    451 	in_pcbstate(inp, INP_CONNECTED);
    452 #if defined(IPSEC) /*|| defined(FAST_IPSEC)*/ /*XXX*/
    453 	if (inp->inp_socket->so_type == SOCK_STREAM)
    454 		ipsec_pcbconn(inp->inp_sp);
    455 #endif
    456 	return (0);
    457 }
    458 
    459 void
    460 in_pcbdisconnect(v)
    461 	void *v;
    462 {
    463 	struct inpcb *inp = v;
    464 
    465 	inp->inp_faddr = zeroin_addr;
    466 	inp->inp_fport = 0;
    467 	in_pcbstate(inp, INP_BOUND);
    468 	if (inp->inp_socket->so_state & SS_NOFDREF)
    469 		in_pcbdetach(inp);
    470 #if defined(IPSEC) /*|| defined(FAST_IPSEC)*/ /*XXX*/
    471 	ipsec_pcbdisconn(inp->inp_sp);
    472 #endif
    473 }
    474 
    475 void
    476 in_pcbdetach(v)
    477 	void *v;
    478 {
    479 	struct inpcb *inp = v;
    480 	struct socket *so = inp->inp_socket;
    481 	int s;
    482 
    483 #if defined(IPSEC) || defined(FAST_IPSEC)
    484 	ipsec4_delete_pcbpolicy(inp);
    485 #endif /*IPSEC*/
    486 	so->so_pcb = 0;
    487 	sofree(so);
    488 	if (inp->inp_options)
    489 		(void)m_free(inp->inp_options);
    490 	if (inp->inp_route.ro_rt)
    491 		rtfree(inp->inp_route.ro_rt);
    492 	ip_freemoptions(inp->inp_moptions);
    493 	if (inp->inp_ia != NULL) {
    494 		LIST_REMOVE(inp, inp_ialink);
    495 		IFAFREE(&inp->inp_ia->ia_ifa);
    496 		inp->inp_ia = NULL;
    497 	}
    498 	s = splnet();
    499 	in_pcbstate(inp, INP_ATTACHED);
    500 	CIRCLEQ_REMOVE(&inp->inp_table->inpt_queue, inp, inp_queue);
    501 	splx(s);
    502 	pool_put(&inpcb_pool, inp);
    503 }
    504 
    505 void
    506 in_setsockaddr(inp, nam)
    507 	struct inpcb *inp;
    508 	struct mbuf *nam;
    509 {
    510 	struct sockaddr_in *sin;
    511 
    512 	nam->m_len = sizeof (*sin);
    513 	sin = mtod(nam, struct sockaddr_in *);
    514 	bzero((caddr_t)sin, sizeof (*sin));
    515 	sin->sin_family = AF_INET;
    516 	sin->sin_len = sizeof(*sin);
    517 	sin->sin_port = inp->inp_lport;
    518 	sin->sin_addr = inp->inp_laddr;
    519 }
    520 
    521 void
    522 in_setpeeraddr(inp, nam)
    523 	struct inpcb *inp;
    524 	struct mbuf *nam;
    525 {
    526 	struct sockaddr_in *sin;
    527 
    528 	nam->m_len = sizeof (*sin);
    529 	sin = mtod(nam, struct sockaddr_in *);
    530 	bzero((caddr_t)sin, sizeof (*sin));
    531 	sin->sin_family = AF_INET;
    532 	sin->sin_len = sizeof(*sin);
    533 	sin->sin_port = inp->inp_fport;
    534 	sin->sin_addr = inp->inp_faddr;
    535 }
    536 
    537 /*
    538  * Pass some notification to all connections of a protocol
    539  * associated with address dst.  The local address and/or port numbers
    540  * may be specified to limit the search.  The "usual action" will be
    541  * taken, depending on the ctlinput cmd.  The caller must filter any
    542  * cmds that are uninteresting (e.g., no error in the map).
    543  * Call the protocol specific routine (if any) to report
    544  * any errors for each matching socket.
    545  *
    546  * Must be called at splsoftnet.
    547  */
    548 int
    549 in_pcbnotify(table, faddr, fport_arg, laddr, lport_arg, errno, notify)
    550 	struct inpcbtable *table;
    551 	struct in_addr faddr, laddr;
    552 	u_int fport_arg, lport_arg;
    553 	int errno;
    554 	void (*notify) __P((struct inpcb *, int));
    555 {
    556 	struct inpcbhead *head;
    557 	struct inpcb *inp, *ninp;
    558 	u_int16_t fport = fport_arg, lport = lport_arg;
    559 	int nmatch;
    560 
    561 	if (in_nullhost(faddr) || notify == 0)
    562 		return (0);
    563 
    564 	nmatch = 0;
    565 	head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport);
    566 	for (inp = LIST_FIRST(head); inp != NULL; inp = ninp) {
    567 		ninp = LIST_NEXT(inp, inp_hash);
    568 		if (in_hosteq(inp->inp_faddr, faddr) &&
    569 		    inp->inp_fport == fport &&
    570 		    inp->inp_lport == lport &&
    571 		    in_hosteq(inp->inp_laddr, laddr)) {
    572 			(*notify)(inp, errno);
    573 			nmatch++;
    574 		}
    575 	}
    576 	return (nmatch);
    577 }
    578 
    579 void
    580 in_pcbnotifyall(table, faddr, errno, notify)
    581 	struct inpcbtable *table;
    582 	struct in_addr faddr;
    583 	int errno;
    584 	void (*notify) __P((struct inpcb *, int));
    585 {
    586 	struct inpcb *inp, *ninp;
    587 
    588 	if (in_nullhost(faddr) || notify == 0)
    589 		return;
    590 
    591 	for (inp = CIRCLEQ_FIRST(&table->inpt_queue);
    592 	    inp != (void *)&table->inpt_queue;
    593 	    inp = ninp) {
    594 		ninp = CIRCLEQ_NEXT(inp, inp_queue);
    595 		if (in_hosteq(inp->inp_faddr, faddr))
    596 			(*notify)(inp, errno);
    597 	}
    598 }
    599 
    600 void
    601 in_pcbpurgeif0(table, ifp)
    602 	struct inpcbtable *table;
    603 	struct ifnet *ifp;
    604 {
    605 	struct inpcb *inp, *ninp;
    606 	struct ip_moptions *imo;
    607 	int i, gap;
    608 
    609 	for (inp = CIRCLEQ_FIRST(&table->inpt_queue);
    610 	    inp != (void *)&table->inpt_queue;
    611 	    inp = ninp) {
    612 		ninp = CIRCLEQ_NEXT(inp, inp_queue);
    613 		imo = inp->inp_moptions;
    614 		if (imo != NULL) {
    615 			/*
    616 			 * Unselect the outgoing interface if it is being
    617 			 * detached.
    618 			 */
    619 			if (imo->imo_multicast_ifp == ifp)
    620 				imo->imo_multicast_ifp = NULL;
    621 
    622 			/*
    623 			 * Drop multicast group membership if we joined
    624 			 * through the interface being detached.
    625 			 */
    626 			for (i = 0, gap = 0; i < imo->imo_num_memberships;
    627 			    i++) {
    628 				if (imo->imo_membership[i]->inm_ifp == ifp) {
    629 					in_delmulti(imo->imo_membership[i]);
    630 					gap++;
    631 				} else if (gap != 0)
    632 					imo->imo_membership[i - gap] =
    633 					    imo->imo_membership[i];
    634 			}
    635 			imo->imo_num_memberships -= gap;
    636 		}
    637 	}
    638 }
    639 
    640 void
    641 in_pcbpurgeif(table, ifp)
    642 	struct inpcbtable *table;
    643 	struct ifnet *ifp;
    644 {
    645 	struct inpcb *inp, *ninp;
    646 
    647 	for (inp = CIRCLEQ_FIRST(&table->inpt_queue);
    648 	    inp != (void *)&table->inpt_queue;
    649 	    inp = ninp) {
    650 		ninp = CIRCLEQ_NEXT(inp, inp_queue);
    651 		if (inp->inp_route.ro_rt != NULL &&
    652 		    inp->inp_route.ro_rt->rt_ifp == ifp)
    653 			in_rtchange(inp, 0);
    654 	}
    655 }
    656 
    657 /*
    658  * Check for alternatives when higher level complains
    659  * about service problems.  For now, invalidate cached
    660  * routing information.  If the route was created dynamically
    661  * (by a redirect), time to try a default gateway again.
    662  */
    663 void
    664 in_losing(inp)
    665 	struct inpcb *inp;
    666 {
    667 	struct rtentry *rt;
    668 	struct rt_addrinfo info;
    669 
    670 	if ((rt = inp->inp_route.ro_rt)) {
    671 		inp->inp_route.ro_rt = 0;
    672 		bzero((caddr_t)&info, sizeof(info));
    673 		info.rti_info[RTAX_DST] = &inp->inp_route.ro_dst;
    674 		info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
    675 		info.rti_info[RTAX_NETMASK] = rt_mask(rt);
    676 		rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0);
    677 		if (rt->rt_flags & RTF_DYNAMIC)
    678 			(void) rtrequest(RTM_DELETE, rt_key(rt),
    679 				rt->rt_gateway, rt_mask(rt), rt->rt_flags,
    680 				(struct rtentry **)0);
    681 		else
    682 		/*
    683 		 * A new route can be allocated
    684 		 * the next time output is attempted.
    685 		 */
    686 			rtfree(rt);
    687 	}
    688 }
    689 
    690 /*
    691  * After a routing change, flush old routing
    692  * and allocate a (hopefully) better one.
    693  */
    694 void
    695 in_rtchange(inp, errno)
    696 	struct inpcb *inp;
    697 	int errno;
    698 {
    699 
    700 	if (inp->inp_route.ro_rt) {
    701 		rtfree(inp->inp_route.ro_rt);
    702 		inp->inp_route.ro_rt = 0;
    703 		/*
    704 		 * A new route can be allocated the next time
    705 		 * output is attempted.
    706 		 */
    707 	}
    708 	/* XXX SHOULD NOTIFY HIGHER-LEVEL PROTOCOLS */
    709 }
    710 
    711 struct inpcb *
    712 in_pcblookup_port(table, laddr, lport_arg, lookup_wildcard)
    713 	struct inpcbtable *table;
    714 	struct in_addr laddr;
    715 	u_int lport_arg;
    716 	int lookup_wildcard;
    717 {
    718 	struct inpcb *inp, *match = 0;
    719 	int matchwild = 3, wildcard;
    720 	u_int16_t lport = lport_arg;
    721 
    722 	CIRCLEQ_FOREACH(inp, &table->inpt_queue, inp_queue) {
    723 		if (inp->inp_lport != lport)
    724 			continue;
    725 		wildcard = 0;
    726 		if (!in_nullhost(inp->inp_faddr))
    727 			wildcard++;
    728 		if (in_nullhost(inp->inp_laddr)) {
    729 			if (!in_nullhost(laddr))
    730 				wildcard++;
    731 		} else {
    732 			if (in_nullhost(laddr))
    733 				wildcard++;
    734 			else {
    735 				if (!in_hosteq(inp->inp_laddr, laddr))
    736 					continue;
    737 			}
    738 		}
    739 		if (wildcard && !lookup_wildcard)
    740 			continue;
    741 		if (wildcard < matchwild) {
    742 			match = inp;
    743 			matchwild = wildcard;
    744 			if (matchwild == 0)
    745 				break;
    746 		}
    747 	}
    748 	return (match);
    749 }
    750 
    751 #ifdef DIAGNOSTIC
    752 int	in_pcbnotifymiss = 0;
    753 #endif
    754 
    755 struct inpcb *
    756 in_pcblookup_connect(table, faddr, fport_arg, laddr, lport_arg)
    757 	struct inpcbtable *table;
    758 	struct in_addr faddr, laddr;
    759 	u_int fport_arg, lport_arg;
    760 {
    761 	struct inpcbhead *head;
    762 	struct inpcb *inp;
    763 	u_int16_t fport = fport_arg, lport = lport_arg;
    764 
    765 	head = INPCBHASH_CONNECT(table, faddr, fport, laddr, lport);
    766 	LIST_FOREACH(inp, head, inp_hash) {
    767 		if (in_hosteq(inp->inp_faddr, faddr) &&
    768 		    inp->inp_fport == fport &&
    769 		    inp->inp_lport == lport &&
    770 		    in_hosteq(inp->inp_laddr, laddr))
    771 			goto out;
    772 	}
    773 #ifdef DIAGNOSTIC
    774 	if (in_pcbnotifymiss) {
    775 		printf("in_pcblookup_connect: faddr=%08x fport=%d laddr=%08x lport=%d\n",
    776 		    ntohl(faddr.s_addr), ntohs(fport),
    777 		    ntohl(laddr.s_addr), ntohs(lport));
    778 	}
    779 #endif
    780 	return (0);
    781 
    782 out:
    783 	/* Move this PCB to the head of hash chain. */
    784 	if (inp != LIST_FIRST(head)) {
    785 		LIST_REMOVE(inp, inp_hash);
    786 		LIST_INSERT_HEAD(head, inp, inp_hash);
    787 	}
    788 	return (inp);
    789 }
    790 
    791 struct inpcb *
    792 in_pcblookup_bind(table, laddr, lport_arg)
    793 	struct inpcbtable *table;
    794 	struct in_addr laddr;
    795 	u_int lport_arg;
    796 {
    797 	struct inpcbhead *head;
    798 	struct inpcb *inp;
    799 	u_int16_t lport = lport_arg;
    800 
    801 	head = INPCBHASH_BIND(table, laddr, lport);
    802 	LIST_FOREACH(inp, head, inp_hash) {
    803 		if (inp->inp_lport == lport &&
    804 		    in_hosteq(inp->inp_laddr, laddr))
    805 			goto out;
    806 	}
    807 	head = INPCBHASH_BIND(table, zeroin_addr, lport);
    808 	LIST_FOREACH(inp, head, inp_hash) {
    809 		if (inp->inp_lport == lport &&
    810 		    in_hosteq(inp->inp_laddr, zeroin_addr))
    811 			goto out;
    812 	}
    813 #ifdef DIAGNOSTIC
    814 	if (in_pcbnotifymiss) {
    815 		printf("in_pcblookup_bind: laddr=%08x lport=%d\n",
    816 		    ntohl(laddr.s_addr), ntohs(lport));
    817 	}
    818 #endif
    819 	return (0);
    820 
    821 out:
    822 	/* Move this PCB to the head of hash chain. */
    823 	if (inp != LIST_FIRST(head)) {
    824 		LIST_REMOVE(inp, inp_hash);
    825 		LIST_INSERT_HEAD(head, inp, inp_hash);
    826 	}
    827 	return (inp);
    828 }
    829 
    830 void
    831 in_pcbstate(inp, state)
    832 	struct inpcb *inp;
    833 	int state;
    834 {
    835 
    836 	if (inp->inp_state > INP_ATTACHED)
    837 		LIST_REMOVE(inp, inp_hash);
    838 
    839 	switch (state) {
    840 	case INP_BOUND:
    841 		LIST_INSERT_HEAD(INPCBHASH_BIND(inp->inp_table,
    842 		    inp->inp_laddr, inp->inp_lport), inp, inp_hash);
    843 		break;
    844 	case INP_CONNECTED:
    845 		LIST_INSERT_HEAD(INPCBHASH_CONNECT(inp->inp_table,
    846 		    inp->inp_faddr, inp->inp_fport,
    847 		    inp->inp_laddr, inp->inp_lport), inp, inp_hash);
    848 		break;
    849 	}
    850 
    851 	inp->inp_state = state;
    852 }
    853 
    854 struct rtentry *
    855 in_pcbrtentry(inp)
    856 	struct inpcb *inp;
    857 {
    858 	struct route *ro;
    859 
    860 	ro = &inp->inp_route;
    861 
    862 	if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 ||
    863 	    !in_hosteq(satosin(&ro->ro_dst)->sin_addr, inp->inp_faddr))) {
    864 		RTFREE(ro->ro_rt);
    865 		ro->ro_rt = (struct rtentry *)NULL;
    866 	}
    867 	if (ro->ro_rt == (struct rtentry *)NULL &&
    868 	    !in_nullhost(inp->inp_faddr)) {
    869 		bzero(&ro->ro_dst, sizeof(struct sockaddr_in));
    870 		ro->ro_dst.sa_family = AF_INET;
    871 		ro->ro_dst.sa_len = sizeof(ro->ro_dst);
    872 		satosin(&ro->ro_dst)->sin_addr = inp->inp_faddr;
    873 		rtalloc(ro);
    874 	}
    875 	return (ro->ro_rt);
    876 }
    877 
    878 struct sockaddr_in *
    879 in_selectsrc(sin, ro, soopts, mopts, errorp)
    880 	struct sockaddr_in *sin;
    881 	struct route *ro;
    882 	int soopts;
    883 	struct ip_moptions *mopts;
    884 	int *errorp;
    885 {
    886 	struct in_ifaddr *ia;
    887 
    888 	ia = (struct in_ifaddr *)0;
    889 	/*
    890 	 * If route is known or can be allocated now,
    891 	 * our src addr is taken from the i/f, else punt.
    892 	 * Note that we should check the address family of the cached
    893 	 * destination, in case of sharing the cache with IPv6.
    894 	 */
    895 	if (ro->ro_rt &&
    896 	    (ro->ro_dst.sa_family != AF_INET ||
    897 	    !in_hosteq(satosin(&ro->ro_dst)->sin_addr, sin->sin_addr) ||
    898 	    soopts & SO_DONTROUTE)) {
    899 		RTFREE(ro->ro_rt);
    900 		ro->ro_rt = (struct rtentry *)0;
    901 	}
    902 	if ((soopts & SO_DONTROUTE) == 0 && /*XXX*/
    903 	    (ro->ro_rt == (struct rtentry *)0 ||
    904 	     ro->ro_rt->rt_ifp == (struct ifnet *)0)) {
    905 		/* No route yet, so try to acquire one */
    906 		bzero(&ro->ro_dst, sizeof(struct sockaddr_in));
    907 		ro->ro_dst.sa_family = AF_INET;
    908 		ro->ro_dst.sa_len = sizeof(struct sockaddr_in);
    909 		satosin(&ro->ro_dst)->sin_addr = sin->sin_addr;
    910 		rtalloc(ro);
    911 	}
    912 	/*
    913 	 * If we found a route, use the address
    914 	 * corresponding to the outgoing interface
    915 	 * unless it is the loopback (in case a route
    916 	 * to our address on another net goes to loopback).
    917 	 *
    918 	 * XXX Is this still true?  Do we care?
    919 	 */
    920 	if (ro->ro_rt && !(ro->ro_rt->rt_ifp->if_flags & IFF_LOOPBACK))
    921 		ia = ifatoia(ro->ro_rt->rt_ifa);
    922 	if (ia == NULL) {
    923 		u_int16_t fport = sin->sin_port;
    924 
    925 		sin->sin_port = 0;
    926 		ia = ifatoia(ifa_ifwithladdr(sintosa(sin)));
    927 		sin->sin_port = fport;
    928 		if (ia == 0) {
    929 			/* Find 1st non-loopback AF_INET address */
    930 			TAILQ_FOREACH(ia, &in_ifaddr, ia_list) {
    931 				if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK))
    932 					break;
    933 			}
    934 		}
    935 		if (ia == NULL) {
    936 			*errorp = EADDRNOTAVAIL;
    937 			return NULL;
    938 		}
    939 	}
    940 	/*
    941 	 * If the destination address is multicast and an outgoing
    942 	 * interface has been set as a multicast option, use the
    943 	 * address of that interface as our source address.
    944 	 */
    945 	if (IN_MULTICAST(sin->sin_addr.s_addr) && mopts != NULL) {
    946 		struct ip_moptions *imo;
    947 		struct ifnet *ifp;
    948 
    949 		imo = mopts;
    950 		if (imo->imo_multicast_ifp != NULL) {
    951 			ifp = imo->imo_multicast_ifp;
    952 			IFP_TO_IA(ifp, ia);		/* XXX */
    953 			if (ia == 0) {
    954 				*errorp = EADDRNOTAVAIL;
    955 				return NULL;
    956 			}
    957 		}
    958 	}
    959 	return satosin(&ia->ia_addr);
    960 }
    961