Home | History | Annotate | Line # | Download | only in netinet
raw_ip.c revision 1.169
      1 /*	$NetBSD: raw_ip.c,v 1.169 2018/02/26 09:04:29 maxv 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) 1982, 1986, 1988, 1993
     34  *	The Regents of the University of California.  All rights reserved.
     35  *
     36  * Redistribution and use in source and binary forms, with or without
     37  * modification, are permitted provided that the following conditions
     38  * are met:
     39  * 1. Redistributions of source code must retain the above copyright
     40  *    notice, this list of conditions and the following disclaimer.
     41  * 2. Redistributions in binary form must reproduce the above copyright
     42  *    notice, this list of conditions and the following disclaimer in the
     43  *    documentation and/or other materials provided with the distribution.
     44  * 3. Neither the name of the University nor the names of its contributors
     45  *    may be used to endorse or promote products derived from this software
     46  *    without specific prior written permission.
     47  *
     48  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     49  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     50  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     51  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     52  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     53  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     54  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     55  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     56  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     57  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     58  * SUCH DAMAGE.
     59  *
     60  *	@(#)raw_ip.c	8.7 (Berkeley) 5/15/95
     61  */
     62 
     63 /*
     64  * Raw interface to IP protocol.
     65  */
     66 
     67 #include <sys/cdefs.h>
     68 __KERNEL_RCSID(0, "$NetBSD: raw_ip.c,v 1.169 2018/02/26 09:04:29 maxv Exp $");
     69 
     70 #ifdef _KERNEL_OPT
     71 #include "opt_inet.h"
     72 #include "opt_ipsec.h"
     73 #include "opt_mrouting.h"
     74 #include "opt_net_mpsafe.h"
     75 #endif
     76 
     77 #include <sys/param.h>
     78 #include <sys/sysctl.h>
     79 #include <sys/mbuf.h>
     80 #include <sys/socket.h>
     81 #include <sys/protosw.h>
     82 #include <sys/socketvar.h>
     83 #include <sys/errno.h>
     84 #include <sys/systm.h>
     85 #include <sys/proc.h>
     86 #include <sys/kauth.h>
     87 
     88 #include <net/if.h>
     89 
     90 #include <netinet/in.h>
     91 #include <netinet/in_systm.h>
     92 #include <netinet/ip.h>
     93 #include <netinet/ip_var.h>
     94 #include <netinet/ip_private.h>
     95 #include <netinet/ip_mroute.h>
     96 #include <netinet/ip_icmp.h>
     97 #include <netinet/in_pcb.h>
     98 #include <netinet/in_proto.h>
     99 #include <netinet/in_var.h>
    100 
    101 #ifdef IPSEC
    102 #include <netipsec/ipsec.h>
    103 #include <netipsec/ipsec_var.h>
    104 #include <netipsec/ipsec_private.h>
    105 #endif	/* IPSEC */
    106 
    107 struct inpcbtable rawcbtable;
    108 
    109 int	 rip_pcbnotify(struct inpcbtable *, struct in_addr,
    110     struct in_addr, int, int, void (*)(struct inpcb *, int));
    111 static int	 rip_connect_pcb(struct inpcb *, struct sockaddr_in *);
    112 static void	 rip_disconnect1(struct inpcb *);
    113 
    114 static void sysctl_net_inet_raw_setup(struct sysctllog **);
    115 
    116 /*
    117  * Nominal space allocated to a raw ip socket.
    118  */
    119 #define	RIPSNDQ		8192
    120 #define	RIPRCVQ		8192
    121 
    122 static u_long		rip_sendspace = RIPSNDQ;
    123 static u_long		rip_recvspace = RIPRCVQ;
    124 
    125 /*
    126  * Raw interface to IP protocol.
    127  */
    128 
    129 /*
    130  * Initialize raw connection block q.
    131  */
    132 void
    133 rip_init(void)
    134 {
    135 
    136 	sysctl_net_inet_raw_setup(NULL);
    137 	in_pcbinit(&rawcbtable, 1, 1);
    138 }
    139 
    140 static void
    141 rip_sbappendaddr(struct inpcb *last, struct ip *ip, const struct sockaddr *sa,
    142     int hlen, struct mbuf *opts, struct mbuf *n)
    143 {
    144 	if (last->inp_flags & INP_NOHEADER)
    145 		m_adj(n, hlen);
    146 	if (last->inp_flags & INP_CONTROLOPTS
    147 	    || SOOPT_TIMESTAMP(last->inp_socket->so_options))
    148 		ip_savecontrol(last, &opts, ip, n);
    149 	if (sbappendaddr(&last->inp_socket->so_rcv, sa, n, opts) == 0) {
    150 		/* should notify about lost packet */
    151 		m_freem(n);
    152 		if (opts)
    153 			m_freem(opts);
    154 	} else
    155 		sorwakeup(last->inp_socket);
    156 }
    157 
    158 /*
    159  * Setup generic address and protocol structures
    160  * for raw_input routine, then pass them along with
    161  * mbuf chain.
    162  */
    163 void
    164 rip_input(struct mbuf *m, ...)
    165 {
    166 	int hlen, proto;
    167 	struct ip *ip = mtod(m, struct ip *);
    168 	struct inpcb_hdr *inph;
    169 	struct inpcb *inp;
    170 	struct inpcb *last = NULL;
    171 	struct mbuf *n, *opts = NULL;
    172 	struct sockaddr_in ripsrc;
    173 	va_list ap;
    174 
    175 	va_start(ap, m);
    176 	(void)va_arg(ap, int);		/* ignore value, advance ap */
    177 	proto = va_arg(ap, int);
    178 	va_end(ap);
    179 
    180 	sockaddr_in_init(&ripsrc, &ip->ip_src, 0);
    181 
    182 	/*
    183 	 * XXX Compatibility: programs using raw IP expect ip_len
    184 	 * XXX to have the header length subtracted, and in host order.
    185 	 * XXX ip_off is also expected to be host order.
    186 	 */
    187 	hlen = ip->ip_hl << 2;
    188 	ip->ip_len = ntohs(ip->ip_len) - hlen;
    189 	NTOHS(ip->ip_off);
    190 
    191 	TAILQ_FOREACH(inph, &rawcbtable.inpt_queue, inph_queue) {
    192 		inp = (struct inpcb *)inph;
    193 		if (inp->inp_af != AF_INET)
    194 			continue;
    195 		if (inp->inp_ip.ip_p && inp->inp_ip.ip_p != proto)
    196 			continue;
    197 		if (!in_nullhost(inp->inp_laddr) &&
    198 		    !in_hosteq(inp->inp_laddr, ip->ip_dst))
    199 			continue;
    200 		if (!in_nullhost(inp->inp_faddr) &&
    201 		    !in_hosteq(inp->inp_faddr, ip->ip_src))
    202 			continue;
    203 		if (last == NULL)
    204 			;
    205 #if defined(IPSEC)
    206 		/* check AH/ESP integrity. */
    207 		else if (ipsec_used && ipsec_in_reject(m, last)) {
    208 			IPSEC_STATINC(IPSEC_STAT_IN_POLVIO);
    209 			/* do not inject data to pcb */
    210 		}
    211 #endif /*IPSEC*/
    212 		else if ((n = m_copypacket(m, M_DONTWAIT)) != NULL) {
    213 			rip_sbappendaddr(last, ip, sintosa(&ripsrc), hlen, opts,
    214 			    n);
    215 			opts = NULL;
    216 		}
    217 		last = inp;
    218 	}
    219 #if defined(IPSEC)
    220 	/* check AH/ESP integrity. */
    221 	if (ipsec_used && last != NULL && ipsec_in_reject(m, last)) {
    222 		m_freem(m);
    223 		IPSEC_STATINC(IPSEC_STAT_IN_POLVIO);
    224 		IP_STATDEC(IP_STAT_DELIVERED);
    225 		/* do not inject data to pcb */
    226 	} else
    227 #endif /*IPSEC*/
    228 	if (last != NULL)
    229 		rip_sbappendaddr(last, ip, sintosa(&ripsrc), hlen, opts, m);
    230 	else if (inetsw[ip_protox[ip->ip_p]].pr_input == rip_input) {
    231 		uint64_t *ips;
    232 
    233 		icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PROTOCOL,
    234 		    0, 0);
    235 		ips = IP_STAT_GETREF();
    236 		ips[IP_STAT_NOPROTO]++;
    237 		ips[IP_STAT_DELIVERED]--;
    238 		IP_STAT_PUTREF();
    239 	} else
    240 		m_freem(m);
    241 	return;
    242 }
    243 
    244 int
    245 rip_pcbnotify(struct inpcbtable *table,
    246     struct in_addr faddr, struct in_addr laddr, int proto, int errno,
    247     void (*notify)(struct inpcb *, int))
    248 {
    249 	struct inpcb_hdr *inph, *ninph;
    250 	int nmatch;
    251 
    252 	nmatch = 0;
    253 	TAILQ_FOREACH_SAFE(inph, &table->inpt_queue, inph_queue, ninph) {
    254 		struct inpcb *inp = (struct inpcb *)inph;
    255 		if (inp->inp_af != AF_INET)
    256 			continue;
    257 		if (inp->inp_ip.ip_p && inp->inp_ip.ip_p != proto)
    258 			continue;
    259 		if (in_hosteq(inp->inp_faddr, faddr) &&
    260 		    in_hosteq(inp->inp_laddr, laddr)) {
    261 			(*notify)(inp, errno);
    262 			nmatch++;
    263 		}
    264 	}
    265 
    266 	return nmatch;
    267 }
    268 
    269 void *
    270 rip_ctlinput(int cmd, const struct sockaddr *sa, void *v)
    271 {
    272 	struct ip *ip = v;
    273 	void (*notify)(struct inpcb *, int) = in_rtchange;
    274 	int errno;
    275 
    276 	if (sa->sa_family != AF_INET ||
    277 	    sa->sa_len != sizeof(struct sockaddr_in))
    278 		return NULL;
    279 	if ((unsigned)cmd >= PRC_NCMDS)
    280 		return NULL;
    281 	errno = inetctlerrmap[cmd];
    282 	if (PRC_IS_REDIRECT(cmd))
    283 		notify = in_rtchange, ip = 0;
    284 	else if (cmd == PRC_HOSTDEAD)
    285 		ip = 0;
    286 	else if (errno == 0)
    287 		return NULL;
    288 	if (ip) {
    289 		rip_pcbnotify(&rawcbtable, satocsin(sa)->sin_addr,
    290 		    ip->ip_src, ip->ip_p, errno, notify);
    291 
    292 		/* XXX mapped address case */
    293 	} else
    294 		in_pcbnotifyall(&rawcbtable, satocsin(sa)->sin_addr, errno,
    295 		    notify);
    296 	return NULL;
    297 }
    298 
    299 /*
    300  * Generate IP header and pass packet to ip_output.
    301  * Tack on options user may have setup with control call.
    302  */
    303 int
    304 rip_output(struct mbuf *m, struct inpcb *inp, struct mbuf *control,
    305     struct lwp *l)
    306 {
    307 	struct ip *ip;
    308 	struct mbuf *opts;
    309 	struct ip_pktopts pktopts;
    310 	kauth_cred_t cred;
    311 	int error, flags;
    312 
    313 	flags = (inp->inp_socket->so_options & SO_DONTROUTE) |
    314 	    IP_ALLOWBROADCAST | IP_RETURNMTU;
    315 
    316 	if (l == NULL)
    317 		cred = NULL;
    318 	else
    319 		cred = l->l_cred;
    320 
    321 	/* Setup IP outgoing packet options */
    322 	memset(&pktopts, 0, sizeof(pktopts));
    323 	error = ip_setpktopts(control, &pktopts, &flags, inp, cred);
    324 	if (control != NULL)
    325 		m_freem(control);
    326 	if (error != 0)
    327 		goto release;
    328 
    329 	/*
    330 	 * If the user handed us a complete IP packet, use it.
    331 	 * Otherwise, allocate an mbuf for a header and fill it in.
    332 	 */
    333 	if ((inp->inp_flags & INP_HDRINCL) == 0) {
    334 		if ((m->m_pkthdr.len + sizeof(struct ip)) > IP_MAXPACKET) {
    335 			error = EMSGSIZE;
    336 			goto release;
    337 		}
    338 		M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
    339 		if (!m) {
    340 			error = ENOBUFS;
    341 			goto release;
    342 		}
    343 		ip = mtod(m, struct ip *);
    344 		ip->ip_tos = 0;
    345 		ip->ip_off = htons(0);
    346 		ip->ip_p = inp->inp_ip.ip_p;
    347 		ip->ip_len = htons(m->m_pkthdr.len);
    348 		ip->ip_src = pktopts.ippo_laddr.sin_addr;
    349 		ip->ip_dst = inp->inp_faddr;
    350 		ip->ip_ttl = MAXTTL;
    351 		opts = inp->inp_options;
    352 	} else {
    353 		if (m->m_pkthdr.len > IP_MAXPACKET) {
    354 			error = EMSGSIZE;
    355 			goto release;
    356 		}
    357 		ip = mtod(m, struct ip *);
    358 
    359 		/*
    360 		 * If the mbuf is read-only, we need to allocate
    361 		 * a new mbuf for the header, since we need to
    362 		 * modify the header.
    363 		 */
    364 		if (M_READONLY(m)) {
    365 			int hlen = ip->ip_hl << 2;
    366 
    367 			m = m_copyup(m, hlen, (max_linkhdr + 3) & ~3);
    368 			if (m == NULL) {
    369 				error = ENOMEM;	/* XXX */
    370 				goto release;
    371 			}
    372 			ip = mtod(m, struct ip *);
    373 		}
    374 
    375 		/* XXX userland passes ip_len and ip_off in host order */
    376 		if (m->m_pkthdr.len != ip->ip_len) {
    377 			error = EINVAL;
    378 			goto release;
    379 		}
    380 		HTONS(ip->ip_len);
    381 		HTONS(ip->ip_off);
    382 		if (ip->ip_id != 0 || m->m_pkthdr.len < IP_MINFRAGSIZE)
    383 			flags |= IP_NOIPNEWID;
    384 		opts = NULL;
    385 		/* XXX prevent ip_output from overwriting header fields */
    386 		flags |= IP_RAWOUTPUT;
    387 		IP_STATINC(IP_STAT_RAWOUT);
    388 	}
    389 
    390 	/*
    391 	 * IP output.  Note: if IP_RETURNMTU flag is set, the MTU size
    392 	 * will be stored in inp_errormtu.
    393 	 */
    394 	return ip_output(m, opts, &inp->inp_route, flags, pktopts.ippo_imo,
    395 	    inp);
    396 
    397  release:
    398 	if (m != NULL)
    399 		m_freem(m);
    400 	return error;
    401 }
    402 
    403 /*
    404  * Raw IP socket option processing.
    405  */
    406 int
    407 rip_ctloutput(int op, struct socket *so, struct sockopt *sopt)
    408 {
    409 	struct inpcb *inp = sotoinpcb(so);
    410 	int error = 0;
    411 	int optval;
    412 
    413 	if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_NOHEADER) {
    414 		if (op == PRCO_GETOPT) {
    415 			optval = (inp->inp_flags & INP_NOHEADER) ? 1 : 0;
    416 			error = sockopt_set(sopt, &optval, sizeof(optval));
    417 		} else if (op == PRCO_SETOPT) {
    418 			error = sockopt_getint(sopt, &optval);
    419 			if (error)
    420 				goto out;
    421 			if (optval) {
    422 				inp->inp_flags &= ~INP_HDRINCL;
    423 				inp->inp_flags |= INP_NOHEADER;
    424 			} else
    425 				inp->inp_flags &= ~INP_NOHEADER;
    426 		}
    427 		goto out;
    428 	} else if (sopt->sopt_level != IPPROTO_IP)
    429 		return ip_ctloutput(op, so, sopt);
    430 
    431 	switch (op) {
    432 
    433 	case PRCO_SETOPT:
    434 		switch (sopt->sopt_name) {
    435 		case IP_HDRINCL:
    436 			error = sockopt_getint(sopt, &optval);
    437 			if (error)
    438 				break;
    439 			if (optval)
    440 				inp->inp_flags |= INP_HDRINCL;
    441 			else
    442 				inp->inp_flags &= ~INP_HDRINCL;
    443 			break;
    444 
    445 #ifdef MROUTING
    446 		case MRT_INIT:
    447 		case MRT_DONE:
    448 		case MRT_ADD_VIF:
    449 		case MRT_DEL_VIF:
    450 		case MRT_ADD_MFC:
    451 		case MRT_DEL_MFC:
    452 		case MRT_ASSERT:
    453 		case MRT_API_CONFIG:
    454 		case MRT_ADD_BW_UPCALL:
    455 		case MRT_DEL_BW_UPCALL:
    456 			error = ip_mrouter_set(so, sopt);
    457 			break;
    458 #endif
    459 
    460 		default:
    461 			error = ip_ctloutput(op, so, sopt);
    462 			break;
    463 		}
    464 		break;
    465 
    466 	case PRCO_GETOPT:
    467 		switch (sopt->sopt_name) {
    468 		case IP_HDRINCL:
    469 			optval = inp->inp_flags & INP_HDRINCL;
    470 			error = sockopt_set(sopt, &optval, sizeof(optval));
    471 			break;
    472 
    473 #ifdef MROUTING
    474 		case MRT_VERSION:
    475 		case MRT_ASSERT:
    476 		case MRT_API_SUPPORT:
    477 		case MRT_API_CONFIG:
    478 			error = ip_mrouter_get(so, sopt);
    479 			break;
    480 #endif
    481 
    482 		default:
    483 			error = ip_ctloutput(op, so, sopt);
    484 			break;
    485 		}
    486 		break;
    487 	}
    488  out:
    489 	return error;
    490 }
    491 
    492 int
    493 rip_connect_pcb(struct inpcb *inp, struct sockaddr_in *addr)
    494 {
    495 
    496 	if (IFNET_READER_EMPTY())
    497 		return (EADDRNOTAVAIL);
    498 	if (addr->sin_family != AF_INET)
    499 		return (EAFNOSUPPORT);
    500 	inp->inp_faddr = addr->sin_addr;
    501 	return (0);
    502 }
    503 
    504 static void
    505 rip_disconnect1(struct inpcb *inp)
    506 {
    507 
    508 	inp->inp_faddr = zeroin_addr;
    509 }
    510 
    511 static int
    512 rip_attach(struct socket *so, int proto)
    513 {
    514 	struct inpcb *inp;
    515 	int error;
    516 
    517 	KASSERT(sotoinpcb(so) == NULL);
    518 	sosetlock(so);
    519 
    520 	if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
    521 		error = soreserve(so, rip_sendspace, rip_recvspace);
    522 		if (error) {
    523 			return error;
    524 		}
    525 	}
    526 
    527 	error = in_pcballoc(so, &rawcbtable);
    528 	if (error) {
    529 		return error;
    530 	}
    531 	inp = sotoinpcb(so);
    532 	inp->inp_ip.ip_p = proto;
    533 	KASSERT(solocked(so));
    534 
    535 	return 0;
    536 }
    537 
    538 static void
    539 rip_detach(struct socket *so)
    540 {
    541 	struct inpcb *inp;
    542 
    543 	KASSERT(solocked(so));
    544 	inp = sotoinpcb(so);
    545 	KASSERT(inp != NULL);
    546 
    547 #ifdef MROUTING
    548 	extern struct socket *ip_mrouter;
    549 	if (so == ip_mrouter) {
    550 		ip_mrouter_done();
    551 	}
    552 #endif
    553 	in_pcbdetach(inp);
    554 }
    555 
    556 static int
    557 rip_accept(struct socket *so, struct sockaddr *nam)
    558 {
    559 	KASSERT(solocked(so));
    560 
    561 	panic("rip_accept");
    562 
    563 	return EOPNOTSUPP;
    564 }
    565 
    566 static int
    567 rip_bind(struct socket *so, struct sockaddr *nam, struct lwp *l)
    568 {
    569 	struct inpcb *inp = sotoinpcb(so);
    570 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
    571 	int error = 0;
    572 	int s, ss;
    573 	struct ifaddr *ifa;
    574 
    575 	KASSERT(solocked(so));
    576 	KASSERT(inp != NULL);
    577 	KASSERT(nam != NULL);
    578 
    579 	if (addr->sin_len != sizeof(*addr))
    580 		return EINVAL;
    581 
    582 	s = splsoftnet();
    583 	if (IFNET_READER_EMPTY()) {
    584 		error = EADDRNOTAVAIL;
    585 		goto release;
    586 	}
    587 	if (addr->sin_family != AF_INET) {
    588 		error = EAFNOSUPPORT;
    589 		goto release;
    590 	}
    591 	ss = pserialize_read_enter();
    592 	if ((ifa = ifa_ifwithaddr(sintosa(addr))) == NULL &&
    593 	    !in_nullhost(addr->sin_addr))
    594 	{
    595 		pserialize_read_exit(ss);
    596 		error = EADDRNOTAVAIL;
    597 		goto release;
    598 	}
    599         if (ifa && (ifatoia(ifa))->ia4_flags & IN6_IFF_DUPLICATED) {
    600 		pserialize_read_exit(ss);
    601 		error = EADDRNOTAVAIL;
    602 		goto release;
    603 	}
    604 	pserialize_read_exit(ss);
    605 
    606 	inp->inp_laddr = addr->sin_addr;
    607 
    608 release:
    609 	splx(s);
    610 	return error;
    611 }
    612 
    613 static int
    614 rip_listen(struct socket *so, struct lwp *l)
    615 {
    616 	KASSERT(solocked(so));
    617 
    618 	return EOPNOTSUPP;
    619 }
    620 
    621 static int
    622 rip_connect(struct socket *so, struct sockaddr *nam, struct lwp *l)
    623 {
    624 	struct inpcb *inp = sotoinpcb(so);
    625 	int error = 0;
    626 	int s;
    627 
    628 	KASSERT(solocked(so));
    629 	KASSERT(inp != NULL);
    630 	KASSERT(nam != NULL);
    631 
    632 	s = splsoftnet();
    633 	error = rip_connect_pcb(inp, (struct sockaddr_in *)nam);
    634 	if (! error)
    635 		soisconnected(so);
    636 	splx(s);
    637 
    638 	return error;
    639 }
    640 
    641 static int
    642 rip_connect2(struct socket *so, struct socket *so2)
    643 {
    644 	KASSERT(solocked(so));
    645 
    646 	return EOPNOTSUPP;
    647 }
    648 
    649 static int
    650 rip_disconnect(struct socket *so)
    651 {
    652 	struct inpcb *inp = sotoinpcb(so);
    653 	int s;
    654 
    655 	KASSERT(solocked(so));
    656 	KASSERT(inp != NULL);
    657 
    658 	s = splsoftnet();
    659 	soisdisconnected(so);
    660 	rip_disconnect1(inp);
    661 	splx(s);
    662 
    663 	return 0;
    664 }
    665 
    666 static int
    667 rip_shutdown(struct socket *so)
    668 {
    669 	int s;
    670 
    671 	KASSERT(solocked(so));
    672 
    673 	/*
    674 	 * Mark the connection as being incapable of further input.
    675 	 */
    676 	s = splsoftnet();
    677 	socantsendmore(so);
    678 	splx(s);
    679 
    680 	return 0;
    681 }
    682 
    683 static int
    684 rip_abort(struct socket *so)
    685 {
    686 	KASSERT(solocked(so));
    687 
    688 	panic("rip_abort");
    689 
    690 	return EOPNOTSUPP;
    691 }
    692 
    693 static int
    694 rip_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp)
    695 {
    696 	return in_control(so, cmd, nam, ifp);
    697 }
    698 
    699 static int
    700 rip_stat(struct socket *so, struct stat *ub)
    701 {
    702 	KASSERT(solocked(so));
    703 
    704 	/* stat: don't bother with a blocksize. */
    705 	return 0;
    706 }
    707 
    708 static int
    709 rip_peeraddr(struct socket *so, struct sockaddr *nam)
    710 {
    711 	int s;
    712 
    713 	KASSERT(solocked(so));
    714 	KASSERT(sotoinpcb(so) != NULL);
    715 	KASSERT(nam != NULL);
    716 
    717 	s = splsoftnet();
    718 	in_setpeeraddr(sotoinpcb(so), (struct sockaddr_in *)nam);
    719 	splx(s);
    720 
    721 	return 0;
    722 }
    723 
    724 static int
    725 rip_sockaddr(struct socket *so, struct sockaddr *nam)
    726 {
    727 	int s;
    728 
    729 	KASSERT(solocked(so));
    730 	KASSERT(sotoinpcb(so) != NULL);
    731 	KASSERT(nam != NULL);
    732 
    733 	s = splsoftnet();
    734 	in_setsockaddr(sotoinpcb(so), (struct sockaddr_in *)nam);
    735 	splx(s);
    736 
    737 	return 0;
    738 }
    739 
    740 static int
    741 rip_rcvd(struct socket *so, int flags, struct lwp *l)
    742 {
    743 	KASSERT(solocked(so));
    744 
    745 	return EOPNOTSUPP;
    746 }
    747 
    748 static int
    749 rip_recvoob(struct socket *so, struct mbuf *m, int flags)
    750 {
    751 	KASSERT(solocked(so));
    752 
    753 	return EOPNOTSUPP;
    754 }
    755 
    756 static int
    757 rip_send(struct socket *so, struct mbuf *m, struct sockaddr *nam,
    758     struct mbuf *control, struct lwp *l)
    759 {
    760 	struct inpcb *inp = sotoinpcb(so);
    761 	int error = 0;
    762 	int s;
    763 
    764 	KASSERT(solocked(so));
    765 	KASSERT(inp != NULL);
    766 	KASSERT(m != NULL);
    767 
    768 	/*
    769 	 * Ship a packet out.  The appropriate raw output
    770 	 * routine handles any massaging necessary.
    771 	 */
    772 	s = splsoftnet();
    773 	if (nam) {
    774 		if ((so->so_state & SS_ISCONNECTED) != 0) {
    775 			error = EISCONN;
    776 			goto die;
    777 		}
    778 		error = rip_connect_pcb(inp, (struct sockaddr_in *)nam);
    779 		if (error)
    780 			goto die;
    781 	} else {
    782 		if ((so->so_state & SS_ISCONNECTED) == 0) {
    783 			error = ENOTCONN;
    784 			goto die;
    785 		}
    786 	}
    787 	error = rip_output(m, inp, control, l);
    788 	m = NULL;
    789 	control = NULL;
    790 	if (nam)
    791 		rip_disconnect1(inp);
    792  die:
    793 	if (m != NULL)
    794 		m_freem(m);
    795 	if (control != NULL)
    796 		m_freem(control);
    797 
    798 	splx(s);
    799 	return error;
    800 }
    801 
    802 static int
    803 rip_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control)
    804 {
    805 	KASSERT(solocked(so));
    806 
    807 	m_freem(m);
    808 	m_freem(control);
    809 
    810 	return EOPNOTSUPP;
    811 }
    812 
    813 static int
    814 rip_purgeif(struct socket *so, struct ifnet *ifp)
    815 {
    816 	int s;
    817 
    818 	s = splsoftnet();
    819 	mutex_enter(softnet_lock);
    820 	in_pcbpurgeif0(&rawcbtable, ifp);
    821 #ifdef NET_MPSAFE
    822 	mutex_exit(softnet_lock);
    823 #endif
    824 	in_purgeif(ifp);
    825 #ifdef NET_MPSAFE
    826 	mutex_enter(softnet_lock);
    827 #endif
    828 	in_pcbpurgeif(&rawcbtable, ifp);
    829 	mutex_exit(softnet_lock);
    830 	splx(s);
    831 
    832 	return 0;
    833 }
    834 
    835 PR_WRAP_USRREQS(rip)
    836 #define	rip_attach	rip_attach_wrapper
    837 #define	rip_detach	rip_detach_wrapper
    838 #define	rip_accept	rip_accept_wrapper
    839 #define	rip_bind	rip_bind_wrapper
    840 #define	rip_listen	rip_listen_wrapper
    841 #define	rip_connect	rip_connect_wrapper
    842 #define	rip_connect2	rip_connect2_wrapper
    843 #define	rip_disconnect	rip_disconnect_wrapper
    844 #define	rip_shutdown	rip_shutdown_wrapper
    845 #define	rip_abort	rip_abort_wrapper
    846 #define	rip_ioctl	rip_ioctl_wrapper
    847 #define	rip_stat	rip_stat_wrapper
    848 #define	rip_peeraddr	rip_peeraddr_wrapper
    849 #define	rip_sockaddr	rip_sockaddr_wrapper
    850 #define	rip_rcvd	rip_rcvd_wrapper
    851 #define	rip_recvoob	rip_recvoob_wrapper
    852 #define	rip_send	rip_send_wrapper
    853 #define	rip_sendoob	rip_sendoob_wrapper
    854 #define	rip_purgeif	rip_purgeif_wrapper
    855 
    856 const struct pr_usrreqs rip_usrreqs = {
    857 	.pr_attach	= rip_attach,
    858 	.pr_detach	= rip_detach,
    859 	.pr_accept	= rip_accept,
    860 	.pr_bind	= rip_bind,
    861 	.pr_listen	= rip_listen,
    862 	.pr_connect	= rip_connect,
    863 	.pr_connect2	= rip_connect2,
    864 	.pr_disconnect	= rip_disconnect,
    865 	.pr_shutdown	= rip_shutdown,
    866 	.pr_abort	= rip_abort,
    867 	.pr_ioctl	= rip_ioctl,
    868 	.pr_stat	= rip_stat,
    869 	.pr_peeraddr	= rip_peeraddr,
    870 	.pr_sockaddr	= rip_sockaddr,
    871 	.pr_rcvd	= rip_rcvd,
    872 	.pr_recvoob	= rip_recvoob,
    873 	.pr_send	= rip_send,
    874 	.pr_sendoob	= rip_sendoob,
    875 	.pr_purgeif	= rip_purgeif,
    876 };
    877 
    878 static void
    879 sysctl_net_inet_raw_setup(struct sysctllog **clog)
    880 {
    881 
    882 	sysctl_createv(clog, 0, NULL, NULL,
    883 		       CTLFLAG_PERMANENT,
    884 		       CTLTYPE_NODE, "inet", NULL,
    885 		       NULL, 0, NULL, 0,
    886 		       CTL_NET, PF_INET, CTL_EOL);
    887 	sysctl_createv(clog, 0, NULL, NULL,
    888 		       CTLFLAG_PERMANENT,
    889 		       CTLTYPE_NODE, "raw",
    890 		       SYSCTL_DESCR("Raw IPv4 settings"),
    891 		       NULL, 0, NULL, 0,
    892 		       CTL_NET, PF_INET, IPPROTO_RAW, CTL_EOL);
    893 
    894 	sysctl_createv(clog, 0, NULL, NULL,
    895 		       CTLFLAG_PERMANENT,
    896 		       CTLTYPE_STRUCT, "pcblist",
    897 		       SYSCTL_DESCR("Raw IPv4 control block list"),
    898 		       sysctl_inpcblist, 0, &rawcbtable, 0,
    899 		       CTL_NET, PF_INET, IPPROTO_RAW,
    900 		       CTL_CREATE, CTL_EOL);
    901 }
    902