Home | History | Annotate | Line # | Download | only in netinet
raw_ip.c revision 1.177.2.1
      1 /*	$NetBSD: raw_ip.c,v 1.177.2.1 2019/06/10 22:09:47 christos 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 1.177.2.1 2019/06/10 22:09:47 christos 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, int off, int proto)
    166 {
    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;
    172 	struct sockaddr_in ripsrc;
    173 	int hlen;
    174 
    175 	sockaddr_in_init(&ripsrc, &ip->ip_src, 0);
    176 
    177 	/*
    178 	 * XXX Compatibility: programs using raw IP expect ip_len
    179 	 * XXX to have the header length subtracted, and in host order.
    180 	 * XXX ip_off is also expected to be host order.
    181 	 */
    182 	hlen = ip->ip_hl << 2;
    183 	ip->ip_len = ntohs(ip->ip_len) - hlen;
    184 	NTOHS(ip->ip_off);
    185 
    186 	TAILQ_FOREACH(inph, &rawcbtable.inpt_queue, inph_queue) {
    187 		inp = (struct inpcb *)inph;
    188 		if (inp->inp_af != AF_INET)
    189 			continue;
    190 		if (inp->inp_ip.ip_p && inp->inp_ip.ip_p != proto)
    191 			continue;
    192 		if (!in_nullhost(inp->inp_laddr) &&
    193 		    !in_hosteq(inp->inp_laddr, ip->ip_dst))
    194 			continue;
    195 		if (!in_nullhost(inp->inp_faddr) &&
    196 		    !in_hosteq(inp->inp_faddr, ip->ip_src))
    197 			continue;
    198 
    199 		if (last == NULL) {
    200 			;
    201 		}
    202 #if defined(IPSEC)
    203 		else if (ipsec_used && ipsec_in_reject(m, last)) {
    204 			/* do not inject data into pcb */
    205 		}
    206 #endif
    207 		else if ((n = m_copypacket(m, M_DONTWAIT)) != NULL) {
    208 			rip_sbappendaddr(last, ip, sintosa(&ripsrc), hlen, n);
    209 		}
    210 
    211 		last = inp;
    212 	}
    213 
    214 #if defined(IPSEC)
    215 	if (ipsec_used && last != NULL && ipsec_in_reject(m, last)) {
    216 		m_freem(m);
    217 		IP_STATDEC(IP_STAT_DELIVERED);
    218 		/* do not inject data into pcb */
    219 	} else
    220 #endif
    221 	if (last != NULL) {
    222 		rip_sbappendaddr(last, ip, sintosa(&ripsrc), hlen, m);
    223 	} else if (inetsw[ip_protox[ip->ip_p]].pr_input == rip_input) {
    224 		uint64_t *ips;
    225 
    226 		icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PROTOCOL,
    227 		    0, 0);
    228 		ips = IP_STAT_GETREF();
    229 		ips[IP_STAT_NOPROTO]++;
    230 		ips[IP_STAT_DELIVERED]--;
    231 		IP_STAT_PUTREF();
    232 	} else {
    233 		m_freem(m);
    234 	}
    235 
    236 	return;
    237 }
    238 
    239 int
    240 rip_pcbnotify(struct inpcbtable *table,
    241     struct in_addr faddr, struct in_addr laddr, int proto, int errno,
    242     void (*notify)(struct inpcb *, int))
    243 {
    244 	struct inpcb_hdr *inph, *ninph;
    245 	int nmatch;
    246 
    247 	nmatch = 0;
    248 	TAILQ_FOREACH_SAFE(inph, &table->inpt_queue, inph_queue, ninph) {
    249 		struct inpcb *inp = (struct inpcb *)inph;
    250 		if (inp->inp_af != AF_INET)
    251 			continue;
    252 		if (inp->inp_ip.ip_p && inp->inp_ip.ip_p != proto)
    253 			continue;
    254 		if (in_hosteq(inp->inp_faddr, faddr) &&
    255 		    in_hosteq(inp->inp_laddr, laddr)) {
    256 			(*notify)(inp, errno);
    257 			nmatch++;
    258 		}
    259 	}
    260 
    261 	return nmatch;
    262 }
    263 
    264 void *
    265 rip_ctlinput(int cmd, const struct sockaddr *sa, void *v)
    266 {
    267 	struct ip *ip = v;
    268 	void (*notify)(struct inpcb *, int) = in_rtchange;
    269 	int errno;
    270 
    271 	if (sa->sa_family != AF_INET ||
    272 	    sa->sa_len != sizeof(struct sockaddr_in))
    273 		return NULL;
    274 	if ((unsigned)cmd >= PRC_NCMDS)
    275 		return NULL;
    276 	errno = inetctlerrmap[cmd];
    277 	if (PRC_IS_REDIRECT(cmd))
    278 		notify = in_rtchange, ip = 0;
    279 	else if (cmd == PRC_HOSTDEAD)
    280 		ip = 0;
    281 	else if (errno == 0)
    282 		return NULL;
    283 	if (ip) {
    284 		rip_pcbnotify(&rawcbtable, satocsin(sa)->sin_addr,
    285 		    ip->ip_src, ip->ip_p, errno, notify);
    286 
    287 		/* XXX mapped address case */
    288 	} else
    289 		in_pcbnotifyall(&rawcbtable, satocsin(sa)->sin_addr, errno,
    290 		    notify);
    291 	return NULL;
    292 }
    293 
    294 /*
    295  * Generate IP header and pass packet to ip_output.
    296  * Tack on options user may have setup with control call.
    297  */
    298 int
    299 rip_output(struct mbuf *m, struct inpcb *inp, struct mbuf *control,
    300     struct lwp *l)
    301 {
    302 	struct ip *ip;
    303 	struct mbuf *opts;
    304 	struct ip_pktopts pktopts;
    305 	kauth_cred_t cred;
    306 	int error, flags;
    307 
    308 	flags = (inp->inp_socket->so_options & SO_DONTROUTE) |
    309 	    IP_ALLOWBROADCAST | IP_RETURNMTU;
    310 
    311 	if (l == NULL)
    312 		cred = NULL;
    313 	else
    314 		cred = l->l_cred;
    315 
    316 	/* Setup IP outgoing packet options */
    317 	memset(&pktopts, 0, sizeof(pktopts));
    318 	error = ip_setpktopts(control, &pktopts, &flags, inp, cred);
    319 	if (control != NULL)
    320 		m_freem(control);
    321 	if (error != 0)
    322 		goto release;
    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 			error = EMSGSIZE;
    331 			goto release;
    332 		}
    333 		M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
    334 		if (!m) {
    335 			error = ENOBUFS;
    336 			goto release;
    337 		}
    338 		ip = mtod(m, struct ip *);
    339 		ip->ip_tos = 0;
    340 		ip->ip_off = htons(0);
    341 		ip->ip_p = inp->inp_ip.ip_p;
    342 		ip->ip_len = htons(m->m_pkthdr.len);
    343 		ip->ip_src = pktopts.ippo_laddr.sin_addr;
    344 		ip->ip_dst = inp->inp_faddr;
    345 		ip->ip_ttl = MAXTTL;
    346 		opts = inp->inp_options;
    347 	} else {
    348 		if (m->m_pkthdr.len > IP_MAXPACKET) {
    349 			error = EMSGSIZE;
    350 			goto release;
    351 		}
    352 		if (m->m_pkthdr.len < sizeof(struct ip)) {
    353 			error = EINVAL;
    354 			goto release;
    355 		}
    356 		ip = mtod(m, struct ip *);
    357 
    358 		/*
    359 		 * If the mbuf is read-only, we need to allocate
    360 		 * a new mbuf for the header, since we need to
    361 		 * modify the header.
    362 		 */
    363 		if (M_READONLY(m)) {
    364 			int hlen = ip->ip_hl << 2;
    365 
    366 			m = m_copyup(m, hlen, (max_linkhdr + 3) & ~3);
    367 			if (m == NULL) {
    368 				error = ENOMEM;
    369 				goto release;
    370 			}
    371 			ip = mtod(m, struct ip *);
    372 		}
    373 
    374 		/* XXX userland passes ip_len and ip_off in host order */
    375 		if (m->m_pkthdr.len != ip->ip_len) {
    376 			error = EINVAL;
    377 			goto release;
    378 		}
    379 		HTONS(ip->ip_len);
    380 		HTONS(ip->ip_off);
    381 
    382 		if (ip->ip_id != 0 || m->m_pkthdr.len < IP_MINFRAGSIZE)
    383 			flags |= IP_NOIPNEWID;
    384 		opts = NULL;
    385 
    386 		/* Prevent ip_output from overwriting header fields. */
    387 		flags |= IP_RAWOUTPUT;
    388 
    389 		IP_STATINC(IP_STAT_RAWOUT);
    390 	}
    391 
    392 	/*
    393 	 * IP output.  Note: if IP_RETURNMTU flag is set, the MTU size
    394 	 * will be stored in inp_errormtu.
    395 	 */
    396 	return ip_output(m, opts, &inp->inp_route, flags, pktopts.ippo_imo,
    397 	    inp);
    398 
    399  release:
    400 	if (m != NULL)
    401 		m_freem(m);
    402 	return error;
    403 }
    404 
    405 /*
    406  * Raw IP socket option processing.
    407  */
    408 int
    409 rip_ctloutput(int op, struct socket *so, struct sockopt *sopt)
    410 {
    411 	struct inpcb *inp = sotoinpcb(so);
    412 	int error = 0;
    413 	int optval;
    414 
    415 	if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_NOHEADER) {
    416 		if (op == PRCO_GETOPT) {
    417 			optval = (inp->inp_flags & INP_NOHEADER) ? 1 : 0;
    418 			error = sockopt_set(sopt, &optval, sizeof(optval));
    419 		} else if (op == PRCO_SETOPT) {
    420 			error = sockopt_getint(sopt, &optval);
    421 			if (error)
    422 				goto out;
    423 			if (optval) {
    424 				inp->inp_flags &= ~INP_HDRINCL;
    425 				inp->inp_flags |= INP_NOHEADER;
    426 			} else
    427 				inp->inp_flags &= ~INP_NOHEADER;
    428 		}
    429 		goto out;
    430 	} else if (sopt->sopt_level != IPPROTO_IP)
    431 		return ip_ctloutput(op, so, sopt);
    432 
    433 	switch (op) {
    434 
    435 	case PRCO_SETOPT:
    436 		switch (sopt->sopt_name) {
    437 		case IP_HDRINCL:
    438 			error = sockopt_getint(sopt, &optval);
    439 			if (error)
    440 				break;
    441 			if (optval)
    442 				inp->inp_flags |= INP_HDRINCL;
    443 			else
    444 				inp->inp_flags &= ~INP_HDRINCL;
    445 			break;
    446 
    447 #ifdef MROUTING
    448 		case MRT_INIT:
    449 		case MRT_DONE:
    450 		case MRT_ADD_VIF:
    451 		case MRT_DEL_VIF:
    452 		case MRT_ADD_MFC:
    453 		case MRT_DEL_MFC:
    454 		case MRT_ASSERT:
    455 		case MRT_API_CONFIG:
    456 		case MRT_ADD_BW_UPCALL:
    457 		case MRT_DEL_BW_UPCALL:
    458 			error = ip_mrouter_set(so, sopt);
    459 			break;
    460 #endif
    461 
    462 		default:
    463 			error = ip_ctloutput(op, so, sopt);
    464 			break;
    465 		}
    466 		break;
    467 
    468 	case PRCO_GETOPT:
    469 		switch (sopt->sopt_name) {
    470 		case IP_HDRINCL:
    471 			optval = inp->inp_flags & INP_HDRINCL;
    472 			error = sockopt_set(sopt, &optval, sizeof(optval));
    473 			break;
    474 
    475 #ifdef MROUTING
    476 		case MRT_VERSION:
    477 		case MRT_ASSERT:
    478 		case MRT_API_SUPPORT:
    479 		case MRT_API_CONFIG:
    480 			error = ip_mrouter_get(so, sopt);
    481 			break;
    482 #endif
    483 
    484 		default:
    485 			error = ip_ctloutput(op, so, sopt);
    486 			break;
    487 		}
    488 		break;
    489 	}
    490  out:
    491 	return error;
    492 }
    493 
    494 int
    495 rip_connect_pcb(struct inpcb *inp, struct sockaddr_in *addr)
    496 {
    497 
    498 	if (IFNET_READER_EMPTY())
    499 		return (EADDRNOTAVAIL);
    500 	if (addr->sin_family != AF_INET)
    501 		return (EAFNOSUPPORT);
    502 	if (addr->sin_len != sizeof(*addr))
    503 		return EINVAL;
    504 	inp->inp_faddr = addr->sin_addr;
    505 	return (0);
    506 }
    507 
    508 static void
    509 rip_disconnect1(struct inpcb *inp)
    510 {
    511 
    512 	inp->inp_faddr = zeroin_addr;
    513 }
    514 
    515 static int
    516 rip_attach(struct socket *so, int proto)
    517 {
    518 	struct inpcb *inp;
    519 	int error;
    520 
    521 	KASSERT(sotoinpcb(so) == NULL);
    522 	sosetlock(so);
    523 
    524 	if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
    525 		error = soreserve(so, rip_sendspace, rip_recvspace);
    526 		if (error) {
    527 			return error;
    528 		}
    529 	}
    530 
    531 	error = in_pcballoc(so, &rawcbtable);
    532 	if (error) {
    533 		return error;
    534 	}
    535 	inp = sotoinpcb(so);
    536 	inp->inp_ip.ip_p = proto;
    537 	KASSERT(solocked(so));
    538 
    539 	return 0;
    540 }
    541 
    542 static void
    543 rip_detach(struct socket *so)
    544 {
    545 	struct inpcb *inp;
    546 
    547 	KASSERT(solocked(so));
    548 	inp = sotoinpcb(so);
    549 	KASSERT(inp != NULL);
    550 
    551 #ifdef MROUTING
    552 	extern struct socket *ip_mrouter;
    553 	if (so == ip_mrouter) {
    554 		ip_mrouter_done();
    555 	}
    556 #endif
    557 	in_pcbdetach(inp);
    558 }
    559 
    560 static int
    561 rip_accept(struct socket *so, struct sockaddr *nam)
    562 {
    563 	KASSERT(solocked(so));
    564 
    565 	panic("rip_accept");
    566 
    567 	return EOPNOTSUPP;
    568 }
    569 
    570 static int
    571 rip_bind(struct socket *so, struct sockaddr *nam, struct lwp *l)
    572 {
    573 	struct inpcb *inp = sotoinpcb(so);
    574 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
    575 	int error = 0;
    576 	int s, ss;
    577 	struct ifaddr *ifa;
    578 
    579 	KASSERT(solocked(so));
    580 	KASSERT(inp != NULL);
    581 	KASSERT(nam != NULL);
    582 
    583 	if (addr->sin_len != sizeof(*addr))
    584 		return EINVAL;
    585 
    586 	s = splsoftnet();
    587 	if (IFNET_READER_EMPTY()) {
    588 		error = EADDRNOTAVAIL;
    589 		goto release;
    590 	}
    591 	if (addr->sin_family != AF_INET) {
    592 		error = EAFNOSUPPORT;
    593 		goto release;
    594 	}
    595 	ss = pserialize_read_enter();
    596 	if ((ifa = ifa_ifwithaddr(sintosa(addr))) == NULL &&
    597 	    !in_nullhost(addr->sin_addr))
    598 	{
    599 		pserialize_read_exit(ss);
    600 		error = EADDRNOTAVAIL;
    601 		goto release;
    602 	}
    603         if (ifa && (ifatoia(ifa))->ia4_flags & IN6_IFF_DUPLICATED) {
    604 		pserialize_read_exit(ss);
    605 		error = EADDRNOTAVAIL;
    606 		goto release;
    607 	}
    608 	pserialize_read_exit(ss);
    609 
    610 	inp->inp_laddr = addr->sin_addr;
    611 
    612 release:
    613 	splx(s);
    614 	return error;
    615 }
    616 
    617 static int
    618 rip_listen(struct socket *so, struct lwp *l)
    619 {
    620 	KASSERT(solocked(so));
    621 
    622 	return EOPNOTSUPP;
    623 }
    624 
    625 static int
    626 rip_connect(struct socket *so, struct sockaddr *nam, struct lwp *l)
    627 {
    628 	struct inpcb *inp = sotoinpcb(so);
    629 	int error = 0;
    630 	int s;
    631 
    632 	KASSERT(solocked(so));
    633 	KASSERT(inp != NULL);
    634 	KASSERT(nam != NULL);
    635 
    636 	s = splsoftnet();
    637 	error = rip_connect_pcb(inp, (struct sockaddr_in *)nam);
    638 	if (! error)
    639 		soisconnected(so);
    640 	splx(s);
    641 
    642 	return error;
    643 }
    644 
    645 static int
    646 rip_connect2(struct socket *so, struct socket *so2)
    647 {
    648 	KASSERT(solocked(so));
    649 
    650 	return EOPNOTSUPP;
    651 }
    652 
    653 static int
    654 rip_disconnect(struct socket *so)
    655 {
    656 	struct inpcb *inp = sotoinpcb(so);
    657 	int s;
    658 
    659 	KASSERT(solocked(so));
    660 	KASSERT(inp != NULL);
    661 
    662 	s = splsoftnet();
    663 	soisdisconnected(so);
    664 	rip_disconnect1(inp);
    665 	splx(s);
    666 
    667 	return 0;
    668 }
    669 
    670 static int
    671 rip_shutdown(struct socket *so)
    672 {
    673 	int s;
    674 
    675 	KASSERT(solocked(so));
    676 
    677 	/*
    678 	 * Mark the connection as being incapable of further input.
    679 	 */
    680 	s = splsoftnet();
    681 	socantsendmore(so);
    682 	splx(s);
    683 
    684 	return 0;
    685 }
    686 
    687 static int
    688 rip_abort(struct socket *so)
    689 {
    690 	KASSERT(solocked(so));
    691 
    692 	panic("rip_abort");
    693 
    694 	return EOPNOTSUPP;
    695 }
    696 
    697 static int
    698 rip_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp)
    699 {
    700 	return in_control(so, cmd, nam, ifp);
    701 }
    702 
    703 static int
    704 rip_stat(struct socket *so, struct stat *ub)
    705 {
    706 	KASSERT(solocked(so));
    707 
    708 	/* stat: don't bother with a blocksize. */
    709 	return 0;
    710 }
    711 
    712 static int
    713 rip_peeraddr(struct socket *so, struct sockaddr *nam)
    714 {
    715 	int s;
    716 
    717 	KASSERT(solocked(so));
    718 	KASSERT(sotoinpcb(so) != NULL);
    719 	KASSERT(nam != NULL);
    720 
    721 	s = splsoftnet();
    722 	in_setpeeraddr(sotoinpcb(so), (struct sockaddr_in *)nam);
    723 	splx(s);
    724 
    725 	return 0;
    726 }
    727 
    728 static int
    729 rip_sockaddr(struct socket *so, struct sockaddr *nam)
    730 {
    731 	int s;
    732 
    733 	KASSERT(solocked(so));
    734 	KASSERT(sotoinpcb(so) != NULL);
    735 	KASSERT(nam != NULL);
    736 
    737 	s = splsoftnet();
    738 	in_setsockaddr(sotoinpcb(so), (struct sockaddr_in *)nam);
    739 	splx(s);
    740 
    741 	return 0;
    742 }
    743 
    744 static int
    745 rip_rcvd(struct socket *so, int flags, struct lwp *l)
    746 {
    747 	KASSERT(solocked(so));
    748 
    749 	return EOPNOTSUPP;
    750 }
    751 
    752 static int
    753 rip_recvoob(struct socket *so, struct mbuf *m, int flags)
    754 {
    755 	KASSERT(solocked(so));
    756 
    757 	return EOPNOTSUPP;
    758 }
    759 
    760 static int
    761 rip_send(struct socket *so, struct mbuf *m, struct sockaddr *nam,
    762     struct mbuf *control, struct lwp *l)
    763 {
    764 	struct inpcb *inp = sotoinpcb(so);
    765 	int error = 0;
    766 	int s;
    767 
    768 	KASSERT(solocked(so));
    769 	KASSERT(inp != NULL);
    770 	KASSERT(m != NULL);
    771 
    772 	/*
    773 	 * Ship a packet out.  The appropriate raw output
    774 	 * routine handles any massaging necessary.
    775 	 */
    776 	s = splsoftnet();
    777 	if (nam) {
    778 		if ((so->so_state & SS_ISCONNECTED) != 0) {
    779 			error = EISCONN;
    780 			goto die;
    781 		}
    782 		error = rip_connect_pcb(inp, (struct sockaddr_in *)nam);
    783 		if (error)
    784 			goto die;
    785 	} else {
    786 		if ((so->so_state & SS_ISCONNECTED) == 0) {
    787 			error = ENOTCONN;
    788 			goto die;
    789 		}
    790 	}
    791 	error = rip_output(m, inp, control, l);
    792 	m = NULL;
    793 	control = NULL;
    794 	if (nam)
    795 		rip_disconnect1(inp);
    796  die:
    797 	if (m != NULL)
    798 		m_freem(m);
    799 	if (control != NULL)
    800 		m_freem(control);
    801 
    802 	splx(s);
    803 	return error;
    804 }
    805 
    806 static int
    807 rip_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control)
    808 {
    809 	KASSERT(solocked(so));
    810 
    811 	m_freem(m);
    812 	m_freem(control);
    813 
    814 	return EOPNOTSUPP;
    815 }
    816 
    817 static int
    818 rip_purgeif(struct socket *so, struct ifnet *ifp)
    819 {
    820 	int s;
    821 
    822 	s = splsoftnet();
    823 	mutex_enter(softnet_lock);
    824 	in_pcbpurgeif0(&rawcbtable, ifp);
    825 #ifdef NET_MPSAFE
    826 	mutex_exit(softnet_lock);
    827 #endif
    828 	in_purgeif(ifp);
    829 #ifdef NET_MPSAFE
    830 	mutex_enter(softnet_lock);
    831 #endif
    832 	in_pcbpurgeif(&rawcbtable, ifp);
    833 	mutex_exit(softnet_lock);
    834 	splx(s);
    835 
    836 	return 0;
    837 }
    838 
    839 PR_WRAP_USRREQS(rip)
    840 #define	rip_attach	rip_attach_wrapper
    841 #define	rip_detach	rip_detach_wrapper
    842 #define	rip_accept	rip_accept_wrapper
    843 #define	rip_bind	rip_bind_wrapper
    844 #define	rip_listen	rip_listen_wrapper
    845 #define	rip_connect	rip_connect_wrapper
    846 #define	rip_connect2	rip_connect2_wrapper
    847 #define	rip_disconnect	rip_disconnect_wrapper
    848 #define	rip_shutdown	rip_shutdown_wrapper
    849 #define	rip_abort	rip_abort_wrapper
    850 #define	rip_ioctl	rip_ioctl_wrapper
    851 #define	rip_stat	rip_stat_wrapper
    852 #define	rip_peeraddr	rip_peeraddr_wrapper
    853 #define	rip_sockaddr	rip_sockaddr_wrapper
    854 #define	rip_rcvd	rip_rcvd_wrapper
    855 #define	rip_recvoob	rip_recvoob_wrapper
    856 #define	rip_send	rip_send_wrapper
    857 #define	rip_sendoob	rip_sendoob_wrapper
    858 #define	rip_purgeif	rip_purgeif_wrapper
    859 
    860 const struct pr_usrreqs rip_usrreqs = {
    861 	.pr_attach	= rip_attach,
    862 	.pr_detach	= rip_detach,
    863 	.pr_accept	= rip_accept,
    864 	.pr_bind	= rip_bind,
    865 	.pr_listen	= rip_listen,
    866 	.pr_connect	= rip_connect,
    867 	.pr_connect2	= rip_connect2,
    868 	.pr_disconnect	= rip_disconnect,
    869 	.pr_shutdown	= rip_shutdown,
    870 	.pr_abort	= rip_abort,
    871 	.pr_ioctl	= rip_ioctl,
    872 	.pr_stat	= rip_stat,
    873 	.pr_peeraddr	= rip_peeraddr,
    874 	.pr_sockaddr	= rip_sockaddr,
    875 	.pr_rcvd	= rip_rcvd,
    876 	.pr_recvoob	= rip_recvoob,
    877 	.pr_send	= rip_send,
    878 	.pr_sendoob	= rip_sendoob,
    879 	.pr_purgeif	= rip_purgeif,
    880 };
    881 
    882 static void
    883 sysctl_net_inet_raw_setup(struct sysctllog **clog)
    884 {
    885 
    886 	sysctl_createv(clog, 0, NULL, NULL,
    887 		       CTLFLAG_PERMANENT,
    888 		       CTLTYPE_NODE, "inet", NULL,
    889 		       NULL, 0, NULL, 0,
    890 		       CTL_NET, PF_INET, CTL_EOL);
    891 	sysctl_createv(clog, 0, NULL, NULL,
    892 		       CTLFLAG_PERMANENT,
    893 		       CTLTYPE_NODE, "raw",
    894 		       SYSCTL_DESCR("Raw IPv4 settings"),
    895 		       NULL, 0, NULL, 0,
    896 		       CTL_NET, PF_INET, IPPROTO_RAW, CTL_EOL);
    897 
    898 	sysctl_createv(clog, 0, NULL, NULL,
    899 		       CTLFLAG_PERMANENT,
    900 		       CTLTYPE_STRUCT, "pcblist",
    901 		       SYSCTL_DESCR("Raw IPv4 control block list"),
    902 		       sysctl_inpcblist, 0, &rawcbtable, 0,
    903 		       CTL_NET, PF_INET, IPPROTO_RAW,
    904 		       CTL_CREATE, CTL_EOL);
    905 }
    906