Home | History | Annotate | Line # | Download | only in netinet
raw_ip.c revision 1.161.2.1
      1 /*	$NetBSD: raw_ip.c,v 1.161.2.1 2017/04/21 16:54:06 bouyer 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 1.161.2.1 2017/04/21 16:54:06 bouyer Exp $");
     69 
     70 #ifdef _KERNEL_OPT
     71 #include "opt_inet.h"
     72 #include "opt_compat_netbsd.h"
     73 #include "opt_ipsec.h"
     74 #include "opt_mrouting.h"
     75 #include "opt_net_mpsafe.h"
     76 #endif
     77 
     78 #include <sys/param.h>
     79 #include <sys/sysctl.h>
     80 #include <sys/mbuf.h>
     81 #include <sys/socket.h>
     82 #include <sys/protosw.h>
     83 #include <sys/socketvar.h>
     84 #include <sys/errno.h>
     85 #include <sys/systm.h>
     86 #include <sys/proc.h>
     87 #include <sys/kauth.h>
     88 
     89 #include <net/if.h>
     90 
     91 #include <netinet/in.h>
     92 #include <netinet/in_systm.h>
     93 #include <netinet/ip.h>
     94 #include <netinet/ip_var.h>
     95 #include <netinet/ip_private.h>
     96 #include <netinet/ip_mroute.h>
     97 #include <netinet/ip_icmp.h>
     98 #include <netinet/in_pcb.h>
     99 #include <netinet/in_proto.h>
    100 #include <netinet/in_var.h>
    101 
    102 #ifdef IPSEC
    103 #include <netipsec/ipsec.h>
    104 #include <netipsec/ipsec_var.h>
    105 #include <netipsec/ipsec_private.h>
    106 #endif	/* IPSEC */
    107 
    108 #ifdef COMPAT_50
    109 #include <compat/sys/socket.h>
    110 #endif
    111 
    112 struct inpcbtable rawcbtable;
    113 
    114 int	 rip_pcbnotify(struct inpcbtable *, struct in_addr,
    115     struct in_addr, int, int, void (*)(struct inpcb *, int));
    116 static int	 rip_connect_pcb(struct inpcb *, struct sockaddr_in *);
    117 static void	 rip_disconnect1(struct inpcb *);
    118 
    119 static void sysctl_net_inet_raw_setup(struct sysctllog **);
    120 
    121 /*
    122  * Nominal space allocated to a raw ip socket.
    123  */
    124 #define	RIPSNDQ		8192
    125 #define	RIPRCVQ		8192
    126 
    127 static u_long		rip_sendspace = RIPSNDQ;
    128 static u_long		rip_recvspace = RIPRCVQ;
    129 
    130 /*
    131  * Raw interface to IP protocol.
    132  */
    133 
    134 /*
    135  * Initialize raw connection block q.
    136  */
    137 void
    138 rip_init(void)
    139 {
    140 
    141 	sysctl_net_inet_raw_setup(NULL);
    142 	in_pcbinit(&rawcbtable, 1, 1);
    143 }
    144 
    145 static void
    146 rip_sbappendaddr(struct inpcb *last, struct ip *ip, const struct sockaddr *sa,
    147     int hlen, struct mbuf *opts, struct mbuf *n)
    148 {
    149 	if (last->inp_flags & INP_NOHEADER)
    150 		m_adj(n, hlen);
    151 	if (last->inp_flags & INP_CONTROLOPTS
    152 #ifdef SO_OTIMESTAMP
    153 	    || last->inp_socket->so_options & SO_OTIMESTAMP
    154 #endif
    155 	    || last->inp_socket->so_options & SO_TIMESTAMP)
    156 		ip_savecontrol(last, &opts, ip, n);
    157 	if (sbappendaddr(&last->inp_socket->so_rcv, sa, n, opts) == 0) {
    158 		/* should notify about lost packet */
    159 		m_freem(n);
    160 		if (opts)
    161 			m_freem(opts);
    162 	} else
    163 		sorwakeup(last->inp_socket);
    164 }
    165 
    166 /*
    167  * Setup generic address and protocol structures
    168  * for raw_input routine, then pass them along with
    169  * mbuf chain.
    170  */
    171 void
    172 rip_input(struct mbuf *m, ...)
    173 {
    174 	int hlen, proto;
    175 	struct ip *ip = mtod(m, struct ip *);
    176 	struct inpcb_hdr *inph;
    177 	struct inpcb *inp;
    178 	struct inpcb *last = NULL;
    179 	struct mbuf *n, *opts = NULL;
    180 	struct sockaddr_in ripsrc;
    181 	va_list ap;
    182 
    183 	va_start(ap, m);
    184 	(void)va_arg(ap, int);		/* ignore value, advance ap */
    185 	proto = va_arg(ap, int);
    186 	va_end(ap);
    187 
    188 	sockaddr_in_init(&ripsrc, &ip->ip_src, 0);
    189 
    190 	/*
    191 	 * XXX Compatibility: programs using raw IP expect ip_len
    192 	 * XXX to have the header length subtracted, and in host order.
    193 	 * XXX ip_off is also expected to be host order.
    194 	 */
    195 	hlen = ip->ip_hl << 2;
    196 	ip->ip_len = ntohs(ip->ip_len) - hlen;
    197 	NTOHS(ip->ip_off);
    198 
    199 	TAILQ_FOREACH(inph, &rawcbtable.inpt_queue, inph_queue) {
    200 		inp = (struct inpcb *)inph;
    201 		if (inp->inp_af != AF_INET)
    202 			continue;
    203 		if (inp->inp_ip.ip_p && inp->inp_ip.ip_p != proto)
    204 			continue;
    205 		if (!in_nullhost(inp->inp_laddr) &&
    206 		    !in_hosteq(inp->inp_laddr, ip->ip_dst))
    207 			continue;
    208 		if (!in_nullhost(inp->inp_faddr) &&
    209 		    !in_hosteq(inp->inp_faddr, ip->ip_src))
    210 			continue;
    211 		if (last == NULL)
    212 			;
    213 #if defined(IPSEC)
    214 		/* check AH/ESP integrity. */
    215 		else if (ipsec_used &&
    216 		    ipsec4_in_reject(m, last)) {
    217 			IPSEC_STATINC(IPSEC_STAT_IN_POLVIO);
    218 			/* do not inject data to pcb */
    219 		}
    220 #endif /*IPSEC*/
    221 		else if ((n = m_copypacket(m, M_DONTWAIT)) != NULL) {
    222 			rip_sbappendaddr(last, ip, sintosa(&ripsrc), hlen, opts,
    223 			    n);
    224 			opts = NULL;
    225 		}
    226 		last = inp;
    227 	}
    228 #if defined(IPSEC)
    229 	/* check AH/ESP integrity. */
    230 	if (ipsec_used && last != NULL
    231 	    && ipsec4_in_reject(m, last)) {
    232 		m_freem(m);
    233 		IPSEC_STATINC(IPSEC_STAT_IN_POLVIO);
    234 		IP_STATDEC(IP_STAT_DELIVERED);
    235 		/* do not inject data to pcb */
    236 	} else
    237 #endif /*IPSEC*/
    238 	if (last != NULL)
    239 		rip_sbappendaddr(last, ip, sintosa(&ripsrc), hlen, opts, m);
    240 	else if (inetsw[ip_protox[ip->ip_p]].pr_input == rip_input) {
    241 		uint64_t *ips;
    242 
    243 		icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PROTOCOL,
    244 		    0, 0);
    245 		ips = IP_STAT_GETREF();
    246 		ips[IP_STAT_NOPROTO]++;
    247 		ips[IP_STAT_DELIVERED]--;
    248 		IP_STAT_PUTREF();
    249 	} else
    250 		m_freem(m);
    251 	return;
    252 }
    253 
    254 int
    255 rip_pcbnotify(struct inpcbtable *table,
    256     struct in_addr faddr, struct in_addr laddr, int proto, int errno,
    257     void (*notify)(struct inpcb *, int))
    258 {
    259 	struct inpcb_hdr *inph, *ninph;
    260 	int nmatch;
    261 
    262 	nmatch = 0;
    263 	TAILQ_FOREACH_SAFE(inph, &table->inpt_queue, inph_queue, ninph) {
    264 		struct inpcb *inp = (struct inpcb *)inph;
    265 		if (inp->inp_af != AF_INET)
    266 			continue;
    267 		if (inp->inp_ip.ip_p && inp->inp_ip.ip_p != proto)
    268 			continue;
    269 		if (in_hosteq(inp->inp_faddr, faddr) &&
    270 		    in_hosteq(inp->inp_laddr, laddr)) {
    271 			(*notify)(inp, errno);
    272 			nmatch++;
    273 		}
    274 	}
    275 
    276 	return nmatch;
    277 }
    278 
    279 void *
    280 rip_ctlinput(int cmd, const struct sockaddr *sa, void *v)
    281 {
    282 	struct ip *ip = v;
    283 	void (*notify)(struct inpcb *, int) = in_rtchange;
    284 	int errno;
    285 
    286 	if (sa->sa_family != AF_INET ||
    287 	    sa->sa_len != sizeof(struct sockaddr_in))
    288 		return NULL;
    289 	if ((unsigned)cmd >= PRC_NCMDS)
    290 		return NULL;
    291 	errno = inetctlerrmap[cmd];
    292 	if (PRC_IS_REDIRECT(cmd))
    293 		notify = in_rtchange, ip = 0;
    294 	else if (cmd == PRC_HOSTDEAD)
    295 		ip = 0;
    296 	else if (errno == 0)
    297 		return NULL;
    298 	if (ip) {
    299 		rip_pcbnotify(&rawcbtable, satocsin(sa)->sin_addr,
    300 		    ip->ip_src, ip->ip_p, errno, notify);
    301 
    302 		/* XXX mapped address case */
    303 	} else
    304 		in_pcbnotifyall(&rawcbtable, satocsin(sa)->sin_addr, errno,
    305 		    notify);
    306 	return NULL;
    307 }
    308 
    309 /*
    310  * Generate IP header and pass packet to ip_output.
    311  * Tack on options user may have setup with control call.
    312  */
    313 int
    314 rip_output(struct mbuf *m, struct inpcb *inp)
    315 {
    316 	struct ip *ip;
    317 	struct mbuf *opts;
    318 	int flags;
    319 
    320 	flags =
    321 	    (inp->inp_socket->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST
    322 	    | IP_RETURNMTU;
    323 
    324 	/*
    325 	 * If the user handed us a complete IP packet, use it.
    326 	 * Otherwise, allocate an mbuf for a header and fill it in.
    327 	 */
    328 	if ((inp->inp_flags & INP_HDRINCL) == 0) {
    329 		if ((m->m_pkthdr.len + sizeof(struct ip)) > IP_MAXPACKET) {
    330 			m_freem(m);
    331 			return (EMSGSIZE);
    332 		}
    333 		M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
    334 		if (!m)
    335 			return (ENOBUFS);
    336 		ip = mtod(m, struct ip *);
    337 		ip->ip_tos = 0;
    338 		ip->ip_off = htons(0);
    339 		ip->ip_p = inp->inp_ip.ip_p;
    340 		ip->ip_len = htons(m->m_pkthdr.len);
    341 		ip->ip_src = inp->inp_laddr;
    342 		ip->ip_dst = inp->inp_faddr;
    343 		ip->ip_ttl = MAXTTL;
    344 		opts = inp->inp_options;
    345 	} else {
    346 		if (m->m_pkthdr.len > IP_MAXPACKET) {
    347 			m_freem(m);
    348 			return (EMSGSIZE);
    349 		}
    350 		ip = mtod(m, struct ip *);
    351 
    352 		/*
    353 		 * If the mbuf is read-only, we need to allocate
    354 		 * a new mbuf for the header, since we need to
    355 		 * modify the header.
    356 		 */
    357 		if (M_READONLY(m)) {
    358 			int hlen = ip->ip_hl << 2;
    359 
    360 			m = m_copyup(m, hlen, (max_linkhdr + 3) & ~3);
    361 			if (m == NULL)
    362 				return (ENOMEM);	/* XXX */
    363 			ip = mtod(m, struct ip *);
    364 		}
    365 
    366 		/* XXX userland passes ip_len and ip_off in host order */
    367 		if (m->m_pkthdr.len != ip->ip_len) {
    368 			m_freem(m);
    369 			return (EINVAL);
    370 		}
    371 		HTONS(ip->ip_len);
    372 		HTONS(ip->ip_off);
    373 		if (ip->ip_id != 0 || m->m_pkthdr.len < IP_MINFRAGSIZE)
    374 			flags |= IP_NOIPNEWID;
    375 		opts = NULL;
    376 		/* XXX prevent ip_output from overwriting header fields */
    377 		flags |= IP_RAWOUTPUT;
    378 		IP_STATINC(IP_STAT_RAWOUT);
    379 	}
    380 
    381 	/*
    382 	 * IP output.  Note: if IP_RETURNMTU flag is set, the MTU size
    383 	 * will be stored in inp_errormtu.
    384 	 */
    385 	return ip_output(m, opts, &inp->inp_route, flags, inp->inp_moptions,
    386 	     inp);
    387 }
    388 
    389 /*
    390  * Raw IP socket option processing.
    391  */
    392 int
    393 rip_ctloutput(int op, struct socket *so, struct sockopt *sopt)
    394 {
    395 	struct inpcb *inp = sotoinpcb(so);
    396 	int error = 0;
    397 	int optval;
    398 
    399 	if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_NOHEADER) {
    400 		if (op == PRCO_GETOPT) {
    401 			optval = (inp->inp_flags & INP_NOHEADER) ? 1 : 0;
    402 			error = sockopt_set(sopt, &optval, sizeof(optval));
    403 		} else if (op == PRCO_SETOPT) {
    404 			error = sockopt_getint(sopt, &optval);
    405 			if (error)
    406 				goto out;
    407 			if (optval) {
    408 				inp->inp_flags &= ~INP_HDRINCL;
    409 				inp->inp_flags |= INP_NOHEADER;
    410 			} else
    411 				inp->inp_flags &= ~INP_NOHEADER;
    412 		}
    413 		goto out;
    414 	} else if (sopt->sopt_level != IPPROTO_IP)
    415 		return ip_ctloutput(op, so, sopt);
    416 
    417 	switch (op) {
    418 
    419 	case PRCO_SETOPT:
    420 		switch (sopt->sopt_name) {
    421 		case IP_HDRINCL:
    422 			error = sockopt_getint(sopt, &optval);
    423 			if (error)
    424 				break;
    425 			if (optval)
    426 				inp->inp_flags |= INP_HDRINCL;
    427 			else
    428 				inp->inp_flags &= ~INP_HDRINCL;
    429 			break;
    430 
    431 #ifdef MROUTING
    432 		case MRT_INIT:
    433 		case MRT_DONE:
    434 		case MRT_ADD_VIF:
    435 		case MRT_DEL_VIF:
    436 		case MRT_ADD_MFC:
    437 		case MRT_DEL_MFC:
    438 		case MRT_ASSERT:
    439 		case MRT_API_CONFIG:
    440 		case MRT_ADD_BW_UPCALL:
    441 		case MRT_DEL_BW_UPCALL:
    442 			error = ip_mrouter_set(so, sopt);
    443 			break;
    444 #endif
    445 
    446 		default:
    447 			error = ip_ctloutput(op, so, sopt);
    448 			break;
    449 		}
    450 		break;
    451 
    452 	case PRCO_GETOPT:
    453 		switch (sopt->sopt_name) {
    454 		case IP_HDRINCL:
    455 			optval = inp->inp_flags & INP_HDRINCL;
    456 			error = sockopt_set(sopt, &optval, sizeof(optval));
    457 			break;
    458 
    459 #ifdef MROUTING
    460 		case MRT_VERSION:
    461 		case MRT_ASSERT:
    462 		case MRT_API_SUPPORT:
    463 		case MRT_API_CONFIG:
    464 			error = ip_mrouter_get(so, sopt);
    465 			break;
    466 #endif
    467 
    468 		default:
    469 			error = ip_ctloutput(op, so, sopt);
    470 			break;
    471 		}
    472 		break;
    473 	}
    474  out:
    475 	return error;
    476 }
    477 
    478 int
    479 rip_connect_pcb(struct inpcb *inp, struct sockaddr_in *addr)
    480 {
    481 
    482 	if (IFNET_READER_EMPTY())
    483 		return (EADDRNOTAVAIL);
    484 	if (addr->sin_family != AF_INET)
    485 		return (EAFNOSUPPORT);
    486 	inp->inp_faddr = addr->sin_addr;
    487 	return (0);
    488 }
    489 
    490 static void
    491 rip_disconnect1(struct inpcb *inp)
    492 {
    493 
    494 	inp->inp_faddr = zeroin_addr;
    495 }
    496 
    497 static int
    498 rip_attach(struct socket *so, int proto)
    499 {
    500 	struct inpcb *inp;
    501 	int error;
    502 
    503 	KASSERT(sotoinpcb(so) == NULL);
    504 	sosetlock(so);
    505 
    506 	if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
    507 		error = soreserve(so, rip_sendspace, rip_recvspace);
    508 		if (error) {
    509 			return error;
    510 		}
    511 	}
    512 
    513 	error = in_pcballoc(so, &rawcbtable);
    514 	if (error) {
    515 		return error;
    516 	}
    517 	inp = sotoinpcb(so);
    518 	inp->inp_ip.ip_p = proto;
    519 	KASSERT(solocked(so));
    520 
    521 	return 0;
    522 }
    523 
    524 static void
    525 rip_detach(struct socket *so)
    526 {
    527 	struct inpcb *inp;
    528 
    529 	KASSERT(solocked(so));
    530 	inp = sotoinpcb(so);
    531 	KASSERT(inp != NULL);
    532 
    533 #ifdef MROUTING
    534 	extern struct socket *ip_mrouter;
    535 	if (so == ip_mrouter) {
    536 		ip_mrouter_done();
    537 	}
    538 #endif
    539 	in_pcbdetach(inp);
    540 }
    541 
    542 static int
    543 rip_accept(struct socket *so, struct sockaddr *nam)
    544 {
    545 	KASSERT(solocked(so));
    546 
    547 	panic("rip_accept");
    548 
    549 	return EOPNOTSUPP;
    550 }
    551 
    552 static int
    553 rip_bind(struct socket *so, struct sockaddr *nam, struct lwp *l)
    554 {
    555 	struct inpcb *inp = sotoinpcb(so);
    556 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
    557 	int error = 0;
    558 	int s, ss;
    559 	struct ifaddr *ifa;
    560 
    561 	KASSERT(solocked(so));
    562 	KASSERT(inp != NULL);
    563 	KASSERT(nam != NULL);
    564 
    565 	if (addr->sin_len != sizeof(*addr))
    566 		return EINVAL;
    567 
    568 	s = splsoftnet();
    569 	if (IFNET_READER_EMPTY()) {
    570 		error = EADDRNOTAVAIL;
    571 		goto release;
    572 	}
    573 	if (addr->sin_family != AF_INET) {
    574 		error = EAFNOSUPPORT;
    575 		goto release;
    576 	}
    577 	ss = pserialize_read_enter();
    578 	if ((ifa = ifa_ifwithaddr(sintosa(addr))) == NULL &&
    579 	    !in_nullhost(addr->sin_addr))
    580 	{
    581 		pserialize_read_exit(ss);
    582 		error = EADDRNOTAVAIL;
    583 		goto release;
    584 	}
    585         if (ifa && (ifatoia(ifa))->ia4_flags & IN6_IFF_DUPLICATED) {
    586 		pserialize_read_exit(ss);
    587 		error = EADDRNOTAVAIL;
    588 		goto release;
    589 	}
    590 	pserialize_read_exit(ss);
    591 
    592 	inp->inp_laddr = addr->sin_addr;
    593 
    594 release:
    595 	splx(s);
    596 	return error;
    597 }
    598 
    599 static int
    600 rip_listen(struct socket *so, struct lwp *l)
    601 {
    602 	KASSERT(solocked(so));
    603 
    604 	return EOPNOTSUPP;
    605 }
    606 
    607 static int
    608 rip_connect(struct socket *so, struct sockaddr *nam, struct lwp *l)
    609 {
    610 	struct inpcb *inp = sotoinpcb(so);
    611 	int error = 0;
    612 	int s;
    613 
    614 	KASSERT(solocked(so));
    615 	KASSERT(inp != NULL);
    616 	KASSERT(nam != NULL);
    617 
    618 	s = splsoftnet();
    619 	error = rip_connect_pcb(inp, (struct sockaddr_in *)nam);
    620 	if (! error)
    621 		soisconnected(so);
    622 	splx(s);
    623 
    624 	return error;
    625 }
    626 
    627 static int
    628 rip_connect2(struct socket *so, struct socket *so2)
    629 {
    630 	KASSERT(solocked(so));
    631 
    632 	return EOPNOTSUPP;
    633 }
    634 
    635 static int
    636 rip_disconnect(struct socket *so)
    637 {
    638 	struct inpcb *inp = sotoinpcb(so);
    639 	int s;
    640 
    641 	KASSERT(solocked(so));
    642 	KASSERT(inp != NULL);
    643 
    644 	s = splsoftnet();
    645 	soisdisconnected(so);
    646 	rip_disconnect1(inp);
    647 	splx(s);
    648 
    649 	return 0;
    650 }
    651 
    652 static int
    653 rip_shutdown(struct socket *so)
    654 {
    655 	int s;
    656 
    657 	KASSERT(solocked(so));
    658 
    659 	/*
    660 	 * Mark the connection as being incapable of further input.
    661 	 */
    662 	s = splsoftnet();
    663 	socantsendmore(so);
    664 	splx(s);
    665 
    666 	return 0;
    667 }
    668 
    669 static int
    670 rip_abort(struct socket *so)
    671 {
    672 	KASSERT(solocked(so));
    673 
    674 	panic("rip_abort");
    675 
    676 	return EOPNOTSUPP;
    677 }
    678 
    679 static int
    680 rip_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp)
    681 {
    682 	return in_control(so, cmd, nam, ifp);
    683 }
    684 
    685 static int
    686 rip_stat(struct socket *so, struct stat *ub)
    687 {
    688 	KASSERT(solocked(so));
    689 
    690 	/* stat: don't bother with a blocksize. */
    691 	return 0;
    692 }
    693 
    694 static int
    695 rip_peeraddr(struct socket *so, struct sockaddr *nam)
    696 {
    697 	int s;
    698 
    699 	KASSERT(solocked(so));
    700 	KASSERT(sotoinpcb(so) != NULL);
    701 	KASSERT(nam != NULL);
    702 
    703 	s = splsoftnet();
    704 	in_setpeeraddr(sotoinpcb(so), (struct sockaddr_in *)nam);
    705 	splx(s);
    706 
    707 	return 0;
    708 }
    709 
    710 static int
    711 rip_sockaddr(struct socket *so, struct sockaddr *nam)
    712 {
    713 	int s;
    714 
    715 	KASSERT(solocked(so));
    716 	KASSERT(sotoinpcb(so) != NULL);
    717 	KASSERT(nam != NULL);
    718 
    719 	s = splsoftnet();
    720 	in_setsockaddr(sotoinpcb(so), (struct sockaddr_in *)nam);
    721 	splx(s);
    722 
    723 	return 0;
    724 }
    725 
    726 static int
    727 rip_rcvd(struct socket *so, int flags, struct lwp *l)
    728 {
    729 	KASSERT(solocked(so));
    730 
    731 	return EOPNOTSUPP;
    732 }
    733 
    734 static int
    735 rip_recvoob(struct socket *so, struct mbuf *m, int flags)
    736 {
    737 	KASSERT(solocked(so));
    738 
    739 	return EOPNOTSUPP;
    740 }
    741 
    742 static int
    743 rip_send(struct socket *so, struct mbuf *m, struct sockaddr *nam,
    744     struct mbuf *control, struct lwp *l)
    745 {
    746 	struct inpcb *inp = sotoinpcb(so);
    747 	int error = 0;
    748 	int s;
    749 
    750 	KASSERT(solocked(so));
    751 	KASSERT(inp != NULL);
    752 	KASSERT(m != NULL);
    753 
    754 	/*
    755 	 * Ship a packet out.  The appropriate raw output
    756 	 * routine handles any massaging necessary.
    757 	 */
    758 	if (control && control->m_len) {
    759 		m_freem(control);
    760 		m_freem(m);
    761 		return EINVAL;
    762 	}
    763 
    764 	s = splsoftnet();
    765 	if (nam) {
    766 		if ((so->so_state & SS_ISCONNECTED) != 0) {
    767 			error = EISCONN;
    768 			goto die;
    769 		}
    770 		error = rip_connect_pcb(inp, (struct sockaddr_in *)nam);
    771 		if (error) {
    772 		die:
    773 			m_freem(m);
    774 			splx(s);
    775 			return error;
    776 		}
    777 	} else {
    778 		if ((so->so_state & SS_ISCONNECTED) == 0) {
    779 			error = ENOTCONN;
    780 			goto die;
    781 		}
    782 	}
    783 	error = rip_output(m, inp);
    784 	if (nam)
    785 		rip_disconnect1(inp);
    786 
    787 	splx(s);
    788 	return error;
    789 }
    790 
    791 static int
    792 rip_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control)
    793 {
    794 	KASSERT(solocked(so));
    795 
    796 	m_freem(m);
    797 	m_freem(control);
    798 
    799 	return EOPNOTSUPP;
    800 }
    801 
    802 static int
    803 rip_purgeif(struct socket *so, struct ifnet *ifp)
    804 {
    805 	int s;
    806 
    807 	s = splsoftnet();
    808 	mutex_enter(softnet_lock);
    809 	in_pcbpurgeif0(&rawcbtable, ifp);
    810 #ifdef NET_MPSAFE
    811 	mutex_exit(softnet_lock);
    812 #endif
    813 	in_purgeif(ifp);
    814 #ifdef NET_MPSAFE
    815 	mutex_enter(softnet_lock);
    816 #endif
    817 	in_pcbpurgeif(&rawcbtable, ifp);
    818 	mutex_exit(softnet_lock);
    819 	splx(s);
    820 
    821 	return 0;
    822 }
    823 
    824 PR_WRAP_USRREQS(rip)
    825 #define	rip_attach	rip_attach_wrapper
    826 #define	rip_detach	rip_detach_wrapper
    827 #define	rip_accept	rip_accept_wrapper
    828 #define	rip_bind	rip_bind_wrapper
    829 #define	rip_listen	rip_listen_wrapper
    830 #define	rip_connect	rip_connect_wrapper
    831 #define	rip_connect2	rip_connect2_wrapper
    832 #define	rip_disconnect	rip_disconnect_wrapper
    833 #define	rip_shutdown	rip_shutdown_wrapper
    834 #define	rip_abort	rip_abort_wrapper
    835 #define	rip_ioctl	rip_ioctl_wrapper
    836 #define	rip_stat	rip_stat_wrapper
    837 #define	rip_peeraddr	rip_peeraddr_wrapper
    838 #define	rip_sockaddr	rip_sockaddr_wrapper
    839 #define	rip_rcvd	rip_rcvd_wrapper
    840 #define	rip_recvoob	rip_recvoob_wrapper
    841 #define	rip_send	rip_send_wrapper
    842 #define	rip_sendoob	rip_sendoob_wrapper
    843 #define	rip_purgeif	rip_purgeif_wrapper
    844 
    845 const struct pr_usrreqs rip_usrreqs = {
    846 	.pr_attach	= rip_attach,
    847 	.pr_detach	= rip_detach,
    848 	.pr_accept	= rip_accept,
    849 	.pr_bind	= rip_bind,
    850 	.pr_listen	= rip_listen,
    851 	.pr_connect	= rip_connect,
    852 	.pr_connect2	= rip_connect2,
    853 	.pr_disconnect	= rip_disconnect,
    854 	.pr_shutdown	= rip_shutdown,
    855 	.pr_abort	= rip_abort,
    856 	.pr_ioctl	= rip_ioctl,
    857 	.pr_stat	= rip_stat,
    858 	.pr_peeraddr	= rip_peeraddr,
    859 	.pr_sockaddr	= rip_sockaddr,
    860 	.pr_rcvd	= rip_rcvd,
    861 	.pr_recvoob	= rip_recvoob,
    862 	.pr_send	= rip_send,
    863 	.pr_sendoob	= rip_sendoob,
    864 	.pr_purgeif	= rip_purgeif,
    865 };
    866 
    867 static void
    868 sysctl_net_inet_raw_setup(struct sysctllog **clog)
    869 {
    870 
    871 	sysctl_createv(clog, 0, NULL, NULL,
    872 		       CTLFLAG_PERMANENT,
    873 		       CTLTYPE_NODE, "inet", NULL,
    874 		       NULL, 0, NULL, 0,
    875 		       CTL_NET, PF_INET, CTL_EOL);
    876 	sysctl_createv(clog, 0, NULL, NULL,
    877 		       CTLFLAG_PERMANENT,
    878 		       CTLTYPE_NODE, "raw",
    879 		       SYSCTL_DESCR("Raw IPv4 settings"),
    880 		       NULL, 0, NULL, 0,
    881 		       CTL_NET, PF_INET, IPPROTO_RAW, CTL_EOL);
    882 
    883 	sysctl_createv(clog, 0, NULL, NULL,
    884 		       CTLFLAG_PERMANENT,
    885 		       CTLTYPE_STRUCT, "pcblist",
    886 		       SYSCTL_DESCR("Raw IPv4 control block list"),
    887 		       sysctl_inpcblist, 0, &rawcbtable, 0,
    888 		       CTL_NET, PF_INET, IPPROTO_RAW,
    889 		       CTL_CREATE, CTL_EOL);
    890 }
    891