Home | History | Annotate | Line # | Download | only in netinet
raw_ip.c revision 1.171.2.4
      1 /*	$NetBSD: raw_ip.c,v 1.171.2.4 2018/05/21 04:36:16 pgoyette 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 1.171.2.4 2018/05/21 04:36:16 pgoyette 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 #endif
    104 
    105 struct inpcbtable rawcbtable;
    106 
    107 int	 rip_pcbnotify(struct inpcbtable *, struct in_addr,
    108     struct in_addr, int, int, void (*)(struct inpcb *, int));
    109 static int	 rip_connect_pcb(struct inpcb *, struct sockaddr_in *);
    110 static void	 rip_disconnect1(struct inpcb *);
    111 
    112 static void sysctl_net_inet_raw_setup(struct sysctllog **);
    113 
    114 /*
    115  * Nominal space allocated to a raw ip socket.
    116  */
    117 #define	RIPSNDQ		8192
    118 #define	RIPRCVQ		8192
    119 
    120 static u_long		rip_sendspace = RIPSNDQ;
    121 static u_long		rip_recvspace = RIPRCVQ;
    122 
    123 /*
    124  * Raw interface to IP protocol.
    125  */
    126 
    127 /*
    128  * Initialize raw connection block q.
    129  */
    130 void
    131 rip_init(void)
    132 {
    133 
    134 	sysctl_net_inet_raw_setup(NULL);
    135 	in_pcbinit(&rawcbtable, 1, 1);
    136 }
    137 
    138 static void
    139 rip_sbappendaddr(struct inpcb *last, struct ip *ip, const struct sockaddr *sa,
    140     int hlen, struct mbuf *n)
    141 {
    142 	struct mbuf *opts = NULL;
    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 		soroverflow(last->inp_socket);
    151 		m_freem(n);
    152 		if (opts)
    153 			m_freem(opts);
    154 	} else {
    155 		sorwakeup(last->inp_socket);
    156 	}
    157 }
    158 
    159 /*
    160  * Setup generic address and protocol structures
    161  * for raw_input routine, then pass them along with
    162  * mbuf chain.
    163  */
    164 void
    165 rip_input(struct mbuf *m, ...)
    166 {
    167 	int hlen, proto;
    168 	struct ip *ip = mtod(m, struct ip *);
    169 	struct inpcb_hdr *inph;
    170 	struct inpcb *inp;
    171 	struct inpcb *last = NULL;
    172 	struct mbuf *n;
    173 	struct sockaddr_in ripsrc;
    174 	va_list ap;
    175 
    176 	va_start(ap, m);
    177 	(void)va_arg(ap, int);		/* ignore value, advance ap */
    178 	proto = va_arg(ap, int);
    179 	va_end(ap);
    180 
    181 	sockaddr_in_init(&ripsrc, &ip->ip_src, 0);
    182 
    183 	/*
    184 	 * XXX Compatibility: programs using raw IP expect ip_len
    185 	 * XXX to have the header length subtracted, and in host order.
    186 	 * XXX ip_off is also expected to be host order.
    187 	 */
    188 	hlen = ip->ip_hl << 2;
    189 	ip->ip_len = ntohs(ip->ip_len) - hlen;
    190 	NTOHS(ip->ip_off);
    191 
    192 	TAILQ_FOREACH(inph, &rawcbtable.inpt_queue, inph_queue) {
    193 		inp = (struct inpcb *)inph;
    194 		if (inp->inp_af != AF_INET)
    195 			continue;
    196 		if (inp->inp_ip.ip_p && inp->inp_ip.ip_p != proto)
    197 			continue;
    198 		if (!in_nullhost(inp->inp_laddr) &&
    199 		    !in_hosteq(inp->inp_laddr, ip->ip_dst))
    200 			continue;
    201 		if (!in_nullhost(inp->inp_faddr) &&
    202 		    !in_hosteq(inp->inp_faddr, ip->ip_src))
    203 			continue;
    204 
    205 		if (last == NULL) {
    206 			;
    207 		}
    208 #if defined(IPSEC)
    209 		else if (ipsec_used && ipsec_in_reject(m, last)) {
    210 			/* do not inject data into pcb */
    211 		}
    212 #endif
    213 		else if ((n = m_copypacket(m, M_DONTWAIT)) != NULL) {
    214 			rip_sbappendaddr(last, ip, sintosa(&ripsrc), hlen, n);
    215 		}
    216 
    217 		last = inp;
    218 	}
    219 
    220 #if defined(IPSEC)
    221 	if (ipsec_used && last != NULL && ipsec_in_reject(m, last)) {
    222 		m_freem(m);
    223 		IP_STATDEC(IP_STAT_DELIVERED);
    224 		/* do not inject data into pcb */
    225 	} else
    226 #endif
    227 	if (last != NULL) {
    228 		rip_sbappendaddr(last, ip, sintosa(&ripsrc), hlen, m);
    229 	} else if (inetsw[ip_protox[ip->ip_p]].pr_input == rip_input) {
    230 		uint64_t *ips;
    231 
    232 		icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PROTOCOL,
    233 		    0, 0);
    234 		ips = IP_STAT_GETREF();
    235 		ips[IP_STAT_NOPROTO]++;
    236 		ips[IP_STAT_DELIVERED]--;
    237 		IP_STAT_PUTREF();
    238 	} else {
    239 		m_freem(m);
    240 	}
    241 
    242 	return;
    243 }
    244 
    245 int
    246 rip_pcbnotify(struct inpcbtable *table,
    247     struct in_addr faddr, struct in_addr laddr, int proto, int errno,
    248     void (*notify)(struct inpcb *, int))
    249 {
    250 	struct inpcb_hdr *inph, *ninph;
    251 	int nmatch;
    252 
    253 	nmatch = 0;
    254 	TAILQ_FOREACH_SAFE(inph, &table->inpt_queue, inph_queue, ninph) {
    255 		struct inpcb *inp = (struct inpcb *)inph;
    256 		if (inp->inp_af != AF_INET)
    257 			continue;
    258 		if (inp->inp_ip.ip_p && inp->inp_ip.ip_p != proto)
    259 			continue;
    260 		if (in_hosteq(inp->inp_faddr, faddr) &&
    261 		    in_hosteq(inp->inp_laddr, laddr)) {
    262 			(*notify)(inp, errno);
    263 			nmatch++;
    264 		}
    265 	}
    266 
    267 	return nmatch;
    268 }
    269 
    270 void *
    271 rip_ctlinput(int cmd, const struct sockaddr *sa, void *v)
    272 {
    273 	struct ip *ip = v;
    274 	void (*notify)(struct inpcb *, int) = in_rtchange;
    275 	int errno;
    276 
    277 	if (sa->sa_family != AF_INET ||
    278 	    sa->sa_len != sizeof(struct sockaddr_in))
    279 		return NULL;
    280 	if ((unsigned)cmd >= PRC_NCMDS)
    281 		return NULL;
    282 	errno = inetctlerrmap[cmd];
    283 	if (PRC_IS_REDIRECT(cmd))
    284 		notify = in_rtchange, ip = 0;
    285 	else if (cmd == PRC_HOSTDEAD)
    286 		ip = 0;
    287 	else if (errno == 0)
    288 		return NULL;
    289 	if (ip) {
    290 		rip_pcbnotify(&rawcbtable, satocsin(sa)->sin_addr,
    291 		    ip->ip_src, ip->ip_p, errno, notify);
    292 
    293 		/* XXX mapped address case */
    294 	} else
    295 		in_pcbnotifyall(&rawcbtable, satocsin(sa)->sin_addr, errno,
    296 		    notify);
    297 	return NULL;
    298 }
    299 
    300 /*
    301  * Generate IP header and pass packet to ip_output.
    302  * Tack on options user may have setup with control call.
    303  */
    304 int
    305 rip_output(struct mbuf *m, struct inpcb *inp, struct mbuf *control,
    306     struct lwp *l)
    307 {
    308 	struct ip *ip;
    309 	struct mbuf *opts;
    310 	struct ip_pktopts pktopts;
    311 	kauth_cred_t cred;
    312 	int error, flags;
    313 
    314 	flags = (inp->inp_socket->so_options & SO_DONTROUTE) |
    315 	    IP_ALLOWBROADCAST | IP_RETURNMTU;
    316 
    317 	if (l == NULL)
    318 		cred = NULL;
    319 	else
    320 		cred = l->l_cred;
    321 
    322 	/* Setup IP outgoing packet options */
    323 	memset(&pktopts, 0, sizeof(pktopts));
    324 	error = ip_setpktopts(control, &pktopts, &flags, inp, cred);
    325 	if (control != NULL)
    326 		m_freem(control);
    327 	if (error != 0)
    328 		goto release;
    329 
    330 	/*
    331 	 * If the user handed us a complete IP packet, use it.
    332 	 * Otherwise, allocate an mbuf for a header and fill it in.
    333 	 */
    334 	if ((inp->inp_flags & INP_HDRINCL) == 0) {
    335 		if ((m->m_pkthdr.len + sizeof(struct ip)) > IP_MAXPACKET) {
    336 			error = EMSGSIZE;
    337 			goto release;
    338 		}
    339 		M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
    340 		if (!m) {
    341 			error = ENOBUFS;
    342 			goto release;
    343 		}
    344 		ip = mtod(m, struct ip *);
    345 		ip->ip_tos = 0;
    346 		ip->ip_off = htons(0);
    347 		ip->ip_p = inp->inp_ip.ip_p;
    348 		ip->ip_len = htons(m->m_pkthdr.len);
    349 		ip->ip_src = pktopts.ippo_laddr.sin_addr;
    350 		ip->ip_dst = inp->inp_faddr;
    351 		ip->ip_ttl = MAXTTL;
    352 		opts = inp->inp_options;
    353 	} else {
    354 		if (m->m_pkthdr.len > IP_MAXPACKET) {
    355 			error = EMSGSIZE;
    356 			goto release;
    357 		}
    358 		if (m->m_pkthdr.len < sizeof(struct ip)) {
    359 			error = EINVAL;
    360 			goto release;
    361 		}
    362 		ip = mtod(m, struct ip *);
    363 
    364 		/*
    365 		 * If the mbuf is read-only, we need to allocate
    366 		 * a new mbuf for the header, since we need to
    367 		 * modify the header.
    368 		 */
    369 		if (M_READONLY(m)) {
    370 			int hlen = ip->ip_hl << 2;
    371 
    372 			m = m_copyup(m, hlen, (max_linkhdr + 3) & ~3);
    373 			if (m == NULL) {
    374 				error = ENOMEM;
    375 				goto release;
    376 			}
    377 			ip = mtod(m, struct ip *);
    378 		}
    379 
    380 		/* XXX userland passes ip_len and ip_off in host order */
    381 		if (m->m_pkthdr.len != ip->ip_len) {
    382 			error = EINVAL;
    383 			goto release;
    384 		}
    385 		HTONS(ip->ip_len);
    386 		HTONS(ip->ip_off);
    387 
    388 		if (ip->ip_id != 0 || m->m_pkthdr.len < IP_MINFRAGSIZE)
    389 			flags |= IP_NOIPNEWID;
    390 		opts = NULL;
    391 
    392 		/* Prevent ip_output from overwriting header fields. */
    393 		flags |= IP_RAWOUTPUT;
    394 
    395 		IP_STATINC(IP_STAT_RAWOUT);
    396 	}
    397 
    398 	/*
    399 	 * IP output.  Note: if IP_RETURNMTU flag is set, the MTU size
    400 	 * will be stored in inp_errormtu.
    401 	 */
    402 	return ip_output(m, opts, &inp->inp_route, flags, pktopts.ippo_imo,
    403 	    inp);
    404 
    405  release:
    406 	if (m != NULL)
    407 		m_freem(m);
    408 	return error;
    409 }
    410 
    411 /*
    412  * Raw IP socket option processing.
    413  */
    414 int
    415 rip_ctloutput(int op, struct socket *so, struct sockopt *sopt)
    416 {
    417 	struct inpcb *inp = sotoinpcb(so);
    418 	int error = 0;
    419 	int optval;
    420 
    421 	if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_NOHEADER) {
    422 		if (op == PRCO_GETOPT) {
    423 			optval = (inp->inp_flags & INP_NOHEADER) ? 1 : 0;
    424 			error = sockopt_set(sopt, &optval, sizeof(optval));
    425 		} else if (op == PRCO_SETOPT) {
    426 			error = sockopt_getint(sopt, &optval);
    427 			if (error)
    428 				goto out;
    429 			if (optval) {
    430 				inp->inp_flags &= ~INP_HDRINCL;
    431 				inp->inp_flags |= INP_NOHEADER;
    432 			} else
    433 				inp->inp_flags &= ~INP_NOHEADER;
    434 		}
    435 		goto out;
    436 	} else if (sopt->sopt_level != IPPROTO_IP)
    437 		return ip_ctloutput(op, so, sopt);
    438 
    439 	switch (op) {
    440 
    441 	case PRCO_SETOPT:
    442 		switch (sopt->sopt_name) {
    443 		case IP_HDRINCL:
    444 			error = sockopt_getint(sopt, &optval);
    445 			if (error)
    446 				break;
    447 			if (optval)
    448 				inp->inp_flags |= INP_HDRINCL;
    449 			else
    450 				inp->inp_flags &= ~INP_HDRINCL;
    451 			break;
    452 
    453 #ifdef MROUTING
    454 		case MRT_INIT:
    455 		case MRT_DONE:
    456 		case MRT_ADD_VIF:
    457 		case MRT_DEL_VIF:
    458 		case MRT_ADD_MFC:
    459 		case MRT_DEL_MFC:
    460 		case MRT_ASSERT:
    461 		case MRT_API_CONFIG:
    462 		case MRT_ADD_BW_UPCALL:
    463 		case MRT_DEL_BW_UPCALL:
    464 			error = ip_mrouter_set(so, sopt);
    465 			break;
    466 #endif
    467 
    468 		default:
    469 			error = ip_ctloutput(op, so, sopt);
    470 			break;
    471 		}
    472 		break;
    473 
    474 	case PRCO_GETOPT:
    475 		switch (sopt->sopt_name) {
    476 		case IP_HDRINCL:
    477 			optval = inp->inp_flags & INP_HDRINCL;
    478 			error = sockopt_set(sopt, &optval, sizeof(optval));
    479 			break;
    480 
    481 #ifdef MROUTING
    482 		case MRT_VERSION:
    483 		case MRT_ASSERT:
    484 		case MRT_API_SUPPORT:
    485 		case MRT_API_CONFIG:
    486 			error = ip_mrouter_get(so, sopt);
    487 			break;
    488 #endif
    489 
    490 		default:
    491 			error = ip_ctloutput(op, so, sopt);
    492 			break;
    493 		}
    494 		break;
    495 	}
    496  out:
    497 	return error;
    498 }
    499 
    500 int
    501 rip_connect_pcb(struct inpcb *inp, struct sockaddr_in *addr)
    502 {
    503 
    504 	if (IFNET_READER_EMPTY())
    505 		return (EADDRNOTAVAIL);
    506 	if (addr->sin_family != AF_INET)
    507 		return (EAFNOSUPPORT);
    508 	inp->inp_faddr = addr->sin_addr;
    509 	return (0);
    510 }
    511 
    512 static void
    513 rip_disconnect1(struct inpcb *inp)
    514 {
    515 
    516 	inp->inp_faddr = zeroin_addr;
    517 }
    518 
    519 static int
    520 rip_attach(struct socket *so, int proto)
    521 {
    522 	struct inpcb *inp;
    523 	int error;
    524 
    525 	KASSERT(sotoinpcb(so) == NULL);
    526 	sosetlock(so);
    527 
    528 	if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
    529 		error = soreserve(so, rip_sendspace, rip_recvspace);
    530 		if (error) {
    531 			return error;
    532 		}
    533 	}
    534 
    535 	error = in_pcballoc(so, &rawcbtable);
    536 	if (error) {
    537 		return error;
    538 	}
    539 	inp = sotoinpcb(so);
    540 	inp->inp_ip.ip_p = proto;
    541 	KASSERT(solocked(so));
    542 
    543 	return 0;
    544 }
    545 
    546 static void
    547 rip_detach(struct socket *so)
    548 {
    549 	struct inpcb *inp;
    550 
    551 	KASSERT(solocked(so));
    552 	inp = sotoinpcb(so);
    553 	KASSERT(inp != NULL);
    554 
    555 #ifdef MROUTING
    556 	extern struct socket *ip_mrouter;
    557 	if (so == ip_mrouter) {
    558 		ip_mrouter_done();
    559 	}
    560 #endif
    561 	in_pcbdetach(inp);
    562 }
    563 
    564 static int
    565 rip_accept(struct socket *so, struct sockaddr *nam)
    566 {
    567 	KASSERT(solocked(so));
    568 
    569 	panic("rip_accept");
    570 
    571 	return EOPNOTSUPP;
    572 }
    573 
    574 static int
    575 rip_bind(struct socket *so, struct sockaddr *nam, struct lwp *l)
    576 {
    577 	struct inpcb *inp = sotoinpcb(so);
    578 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
    579 	int error = 0;
    580 	int s, ss;
    581 	struct ifaddr *ifa;
    582 
    583 	KASSERT(solocked(so));
    584 	KASSERT(inp != NULL);
    585 	KASSERT(nam != NULL);
    586 
    587 	if (addr->sin_len != sizeof(*addr))
    588 		return EINVAL;
    589 
    590 	s = splsoftnet();
    591 	if (IFNET_READER_EMPTY()) {
    592 		error = EADDRNOTAVAIL;
    593 		goto release;
    594 	}
    595 	if (addr->sin_family != AF_INET) {
    596 		error = EAFNOSUPPORT;
    597 		goto release;
    598 	}
    599 	ss = pserialize_read_enter();
    600 	if ((ifa = ifa_ifwithaddr(sintosa(addr))) == NULL &&
    601 	    !in_nullhost(addr->sin_addr))
    602 	{
    603 		pserialize_read_exit(ss);
    604 		error = EADDRNOTAVAIL;
    605 		goto release;
    606 	}
    607         if (ifa && (ifatoia(ifa))->ia4_flags & IN6_IFF_DUPLICATED) {
    608 		pserialize_read_exit(ss);
    609 		error = EADDRNOTAVAIL;
    610 		goto release;
    611 	}
    612 	pserialize_read_exit(ss);
    613 
    614 	inp->inp_laddr = addr->sin_addr;
    615 
    616 release:
    617 	splx(s);
    618 	return error;
    619 }
    620 
    621 static int
    622 rip_listen(struct socket *so, struct lwp *l)
    623 {
    624 	KASSERT(solocked(so));
    625 
    626 	return EOPNOTSUPP;
    627 }
    628 
    629 static int
    630 rip_connect(struct socket *so, struct sockaddr *nam, struct lwp *l)
    631 {
    632 	struct inpcb *inp = sotoinpcb(so);
    633 	int error = 0;
    634 	int s;
    635 
    636 	KASSERT(solocked(so));
    637 	KASSERT(inp != NULL);
    638 	KASSERT(nam != NULL);
    639 
    640 	s = splsoftnet();
    641 	error = rip_connect_pcb(inp, (struct sockaddr_in *)nam);
    642 	if (! error)
    643 		soisconnected(so);
    644 	splx(s);
    645 
    646 	return error;
    647 }
    648 
    649 static int
    650 rip_connect2(struct socket *so, struct socket *so2)
    651 {
    652 	KASSERT(solocked(so));
    653 
    654 	return EOPNOTSUPP;
    655 }
    656 
    657 static int
    658 rip_disconnect(struct socket *so)
    659 {
    660 	struct inpcb *inp = sotoinpcb(so);
    661 	int s;
    662 
    663 	KASSERT(solocked(so));
    664 	KASSERT(inp != NULL);
    665 
    666 	s = splsoftnet();
    667 	soisdisconnected(so);
    668 	rip_disconnect1(inp);
    669 	splx(s);
    670 
    671 	return 0;
    672 }
    673 
    674 static int
    675 rip_shutdown(struct socket *so)
    676 {
    677 	int s;
    678 
    679 	KASSERT(solocked(so));
    680 
    681 	/*
    682 	 * Mark the connection as being incapable of further input.
    683 	 */
    684 	s = splsoftnet();
    685 	socantsendmore(so);
    686 	splx(s);
    687 
    688 	return 0;
    689 }
    690 
    691 static int
    692 rip_abort(struct socket *so)
    693 {
    694 	KASSERT(solocked(so));
    695 
    696 	panic("rip_abort");
    697 
    698 	return EOPNOTSUPP;
    699 }
    700 
    701 static int
    702 rip_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp)
    703 {
    704 	return in_control(so, cmd, nam, ifp);
    705 }
    706 
    707 static int
    708 rip_stat(struct socket *so, struct stat *ub)
    709 {
    710 	KASSERT(solocked(so));
    711 
    712 	/* stat: don't bother with a blocksize. */
    713 	return 0;
    714 }
    715 
    716 static int
    717 rip_peeraddr(struct socket *so, struct sockaddr *nam)
    718 {
    719 	int s;
    720 
    721 	KASSERT(solocked(so));
    722 	KASSERT(sotoinpcb(so) != NULL);
    723 	KASSERT(nam != NULL);
    724 
    725 	s = splsoftnet();
    726 	in_setpeeraddr(sotoinpcb(so), (struct sockaddr_in *)nam);
    727 	splx(s);
    728 
    729 	return 0;
    730 }
    731 
    732 static int
    733 rip_sockaddr(struct socket *so, struct sockaddr *nam)
    734 {
    735 	int s;
    736 
    737 	KASSERT(solocked(so));
    738 	KASSERT(sotoinpcb(so) != NULL);
    739 	KASSERT(nam != NULL);
    740 
    741 	s = splsoftnet();
    742 	in_setsockaddr(sotoinpcb(so), (struct sockaddr_in *)nam);
    743 	splx(s);
    744 
    745 	return 0;
    746 }
    747 
    748 static int
    749 rip_rcvd(struct socket *so, int flags, struct lwp *l)
    750 {
    751 	KASSERT(solocked(so));
    752 
    753 	return EOPNOTSUPP;
    754 }
    755 
    756 static int
    757 rip_recvoob(struct socket *so, struct mbuf *m, int flags)
    758 {
    759 	KASSERT(solocked(so));
    760 
    761 	return EOPNOTSUPP;
    762 }
    763 
    764 static int
    765 rip_send(struct socket *so, struct mbuf *m, struct sockaddr *nam,
    766     struct mbuf *control, struct lwp *l)
    767 {
    768 	struct inpcb *inp = sotoinpcb(so);
    769 	int error = 0;
    770 	int s;
    771 
    772 	KASSERT(solocked(so));
    773 	KASSERT(inp != NULL);
    774 	KASSERT(m != NULL);
    775 
    776 	/*
    777 	 * Ship a packet out.  The appropriate raw output
    778 	 * routine handles any massaging necessary.
    779 	 */
    780 	s = splsoftnet();
    781 	if (nam) {
    782 		if ((so->so_state & SS_ISCONNECTED) != 0) {
    783 			error = EISCONN;
    784 			goto die;
    785 		}
    786 		error = rip_connect_pcb(inp, (struct sockaddr_in *)nam);
    787 		if (error)
    788 			goto die;
    789 	} else {
    790 		if ((so->so_state & SS_ISCONNECTED) == 0) {
    791 			error = ENOTCONN;
    792 			goto die;
    793 		}
    794 	}
    795 	error = rip_output(m, inp, control, l);
    796 	m = NULL;
    797 	control = NULL;
    798 	if (nam)
    799 		rip_disconnect1(inp);
    800  die:
    801 	if (m != NULL)
    802 		m_freem(m);
    803 	if (control != NULL)
    804 		m_freem(control);
    805 
    806 	splx(s);
    807 	return error;
    808 }
    809 
    810 static int
    811 rip_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control)
    812 {
    813 	KASSERT(solocked(so));
    814 
    815 	m_freem(m);
    816 	m_freem(control);
    817 
    818 	return EOPNOTSUPP;
    819 }
    820 
    821 static int
    822 rip_purgeif(struct socket *so, struct ifnet *ifp)
    823 {
    824 	int s;
    825 
    826 	s = splsoftnet();
    827 	mutex_enter(softnet_lock);
    828 	in_pcbpurgeif0(&rawcbtable, ifp);
    829 #ifdef NET_MPSAFE
    830 	mutex_exit(softnet_lock);
    831 #endif
    832 	in_purgeif(ifp);
    833 #ifdef NET_MPSAFE
    834 	mutex_enter(softnet_lock);
    835 #endif
    836 	in_pcbpurgeif(&rawcbtable, ifp);
    837 	mutex_exit(softnet_lock);
    838 	splx(s);
    839 
    840 	return 0;
    841 }
    842 
    843 PR_WRAP_USRREQS(rip)
    844 #define	rip_attach	rip_attach_wrapper
    845 #define	rip_detach	rip_detach_wrapper
    846 #define	rip_accept	rip_accept_wrapper
    847 #define	rip_bind	rip_bind_wrapper
    848 #define	rip_listen	rip_listen_wrapper
    849 #define	rip_connect	rip_connect_wrapper
    850 #define	rip_connect2	rip_connect2_wrapper
    851 #define	rip_disconnect	rip_disconnect_wrapper
    852 #define	rip_shutdown	rip_shutdown_wrapper
    853 #define	rip_abort	rip_abort_wrapper
    854 #define	rip_ioctl	rip_ioctl_wrapper
    855 #define	rip_stat	rip_stat_wrapper
    856 #define	rip_peeraddr	rip_peeraddr_wrapper
    857 #define	rip_sockaddr	rip_sockaddr_wrapper
    858 #define	rip_rcvd	rip_rcvd_wrapper
    859 #define	rip_recvoob	rip_recvoob_wrapper
    860 #define	rip_send	rip_send_wrapper
    861 #define	rip_sendoob	rip_sendoob_wrapper
    862 #define	rip_purgeif	rip_purgeif_wrapper
    863 
    864 const struct pr_usrreqs rip_usrreqs = {
    865 	.pr_attach	= rip_attach,
    866 	.pr_detach	= rip_detach,
    867 	.pr_accept	= rip_accept,
    868 	.pr_bind	= rip_bind,
    869 	.pr_listen	= rip_listen,
    870 	.pr_connect	= rip_connect,
    871 	.pr_connect2	= rip_connect2,
    872 	.pr_disconnect	= rip_disconnect,
    873 	.pr_shutdown	= rip_shutdown,
    874 	.pr_abort	= rip_abort,
    875 	.pr_ioctl	= rip_ioctl,
    876 	.pr_stat	= rip_stat,
    877 	.pr_peeraddr	= rip_peeraddr,
    878 	.pr_sockaddr	= rip_sockaddr,
    879 	.pr_rcvd	= rip_rcvd,
    880 	.pr_recvoob	= rip_recvoob,
    881 	.pr_send	= rip_send,
    882 	.pr_sendoob	= rip_sendoob,
    883 	.pr_purgeif	= rip_purgeif,
    884 };
    885 
    886 static void
    887 sysctl_net_inet_raw_setup(struct sysctllog **clog)
    888 {
    889 
    890 	sysctl_createv(clog, 0, NULL, NULL,
    891 		       CTLFLAG_PERMANENT,
    892 		       CTLTYPE_NODE, "inet", NULL,
    893 		       NULL, 0, NULL, 0,
    894 		       CTL_NET, PF_INET, CTL_EOL);
    895 	sysctl_createv(clog, 0, NULL, NULL,
    896 		       CTLFLAG_PERMANENT,
    897 		       CTLTYPE_NODE, "raw",
    898 		       SYSCTL_DESCR("Raw IPv4 settings"),
    899 		       NULL, 0, NULL, 0,
    900 		       CTL_NET, PF_INET, IPPROTO_RAW, CTL_EOL);
    901 
    902 	sysctl_createv(clog, 0, NULL, NULL,
    903 		       CTLFLAG_PERMANENT,
    904 		       CTLTYPE_STRUCT, "pcblist",
    905 		       SYSCTL_DESCR("Raw IPv4 control block list"),
    906 		       sysctl_inpcblist, 0, &rawcbtable, 0,
    907 		       CTL_NET, PF_INET, IPPROTO_RAW,
    908 		       CTL_CREATE, CTL_EOL);
    909 }
    910