Home | History | Annotate | Line # | Download | only in netinet
      1 /*	$NetBSD: raw_ip.c,v 1.187 2025/06/20 15:15:35 roy 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 1.187 2025/06/20 15:15:35 roy 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 	inpcb_init(&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 		m_freem(opts);
    153 	} else {
    154 		sorwakeup(last->inp_socket);
    155 	}
    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, int off, int proto)
    165 {
    166 	struct ip *ip = mtod(m, struct ip *);
    167 	struct inpcb *inp;
    168 	struct inpcb *last = NULL;
    169 	struct mbuf *n;
    170 	struct sockaddr_in ripsrc;
    171 	int hlen;
    172 
    173 	sockaddr_in_init(&ripsrc, &ip->ip_src, 0);
    174 
    175 	/*
    176 	 * XXX Compatibility: programs using raw IP expect ip_len
    177 	 * XXX to have the header length subtracted, and in host order.
    178 	 * XXX ip_off is also expected to be host order.
    179 	 */
    180 	hlen = ip->ip_hl << 2;
    181 	ip->ip_len = ntohs(ip->ip_len) - hlen;
    182 	NTOHS(ip->ip_off);
    183 
    184 	TAILQ_FOREACH(inp, &rawcbtable.inpt_queue, inp_queue) {
    185 		if (inp->inp_af != AF_INET)
    186 			continue;
    187 		if (in4p_ip(inp).ip_p && in4p_ip(inp).ip_p != proto)
    188 			continue;
    189 		if (!in_nullhost(in4p_laddr(inp)) &&
    190 		    !in_hosteq(in4p_laddr(inp), ip->ip_dst))
    191 			continue;
    192 		if (!in_nullhost(in4p_faddr(inp)) &&
    193 		    !in_hosteq(in4p_faddr(inp), ip->ip_src))
    194 			continue;
    195 
    196 		if (last == NULL) {
    197 			;
    198 		}
    199 #if defined(IPSEC)
    200 		else if (ipsec_used && ipsec_in_reject(m, last)) {
    201 			/* do not inject data into pcb */
    202 		}
    203 #endif
    204 		else if ((n = m_copypacket(m, M_DONTWAIT)) != NULL) {
    205 			rip_sbappendaddr(last, ip, sintosa(&ripsrc), hlen, n);
    206 		}
    207 
    208 		last = inp;
    209 	}
    210 
    211 #if defined(IPSEC)
    212 	if (ipsec_used && last != NULL && ipsec_in_reject(m, last)) {
    213 		m_freem(m);
    214 		IP_STATDEC(IP_STAT_DELIVERED);
    215 		/* do not inject data into pcb */
    216 	} else
    217 #endif
    218 	if (last != NULL) {
    219 		rip_sbappendaddr(last, ip, sintosa(&ripsrc), hlen, m);
    220 	} else if (inetsw[ip_protox[ip->ip_p]].pr_input == rip_input) {
    221 		net_stat_ref_t ips;
    222 
    223 		icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PROTOCOL,
    224 		    0, 0);
    225 		ips = IP_STAT_GETREF();
    226 		_NET_STATINC_REF(ips, IP_STAT_NOPROTO);
    227 		_NET_STATDEC_REF(ips, IP_STAT_DELIVERED);
    228 		IP_STAT_PUTREF();
    229 	} else {
    230 		m_freem(m);
    231 	}
    232 
    233 	return;
    234 }
    235 
    236 int
    237 rip_pcbnotify(struct inpcbtable *table,
    238     struct in_addr faddr, struct in_addr laddr, int proto, int errno,
    239     void (*notify)(struct inpcb *, int))
    240 {
    241 	struct inpcb *inp;
    242 	int nmatch;
    243 
    244 	nmatch = 0;
    245 	TAILQ_FOREACH(inp, &table->inpt_queue, inp_queue) {
    246 		if (inp->inp_af != AF_INET)
    247 			continue;
    248 		if (in4p_ip(inp).ip_p && in4p_ip(inp).ip_p != proto)
    249 			continue;
    250 		if (in_hosteq(in4p_faddr(inp), faddr) &&
    251 		    in_hosteq(in4p_laddr(inp), laddr)) {
    252 			(*notify)(inp, errno);
    253 			nmatch++;
    254 		}
    255 	}
    256 
    257 	return nmatch;
    258 }
    259 
    260 void *
    261 rip_ctlinput(int cmd, const struct sockaddr *sa, void *v)
    262 {
    263 	struct ip *ip = v;
    264 	void (*notify)(struct inpcb *, int) = inpcb_rtchange;
    265 	int errno;
    266 
    267 	if (sa->sa_family != AF_INET ||
    268 	    sa->sa_len != sizeof(struct sockaddr_in))
    269 		return NULL;
    270 	if ((unsigned)cmd >= PRC_NCMDS)
    271 		return NULL;
    272 	errno = inetctlerrmap[cmd];
    273 	if (PRC_IS_REDIRECT(cmd))
    274 		notify = inpcb_rtchange, ip = 0;
    275 	else if (cmd == PRC_HOSTDEAD)
    276 		ip = 0;
    277 	else if (errno == 0)
    278 		return NULL;
    279 	if (ip) {
    280 		rip_pcbnotify(&rawcbtable, satocsin(sa)->sin_addr,
    281 		    ip->ip_src, ip->ip_p, errno, notify);
    282 
    283 		/* XXX mapped address case */
    284 	} else
    285 		inpcb_notifyall(&rawcbtable, satocsin(sa)->sin_addr, errno,
    286 		    notify);
    287 	return NULL;
    288 }
    289 
    290 /*
    291  * Generate IP header and pass packet to ip_output.
    292  * Tack on options user may have setup with control call.
    293  */
    294 int
    295 rip_output(struct mbuf *m, struct inpcb *inp, struct mbuf *control,
    296     struct lwp *l)
    297 {
    298 	struct ip *ip;
    299 	struct mbuf *opts;
    300 	struct ip_pktopts pktopts;
    301 	kauth_cred_t cred;
    302 	int error, flags;
    303 
    304 	flags = (inp->inp_socket->so_options & SO_DONTROUTE) |
    305 	    IP_ALLOWBROADCAST | IP_RETURNMTU;
    306 
    307 	if (l == NULL)
    308 		cred = NULL;
    309 	else
    310 		cred = l->l_cred;
    311 
    312 	/* Setup IP outgoing packet options */
    313 	memset(&pktopts, 0, sizeof(pktopts));
    314 	error = ip_setpktopts(control, &pktopts, &flags, inp, cred);
    315 	m_freem(control);
    316 	if (error != 0)
    317 		goto release;
    318 
    319 	/*
    320 	 * If the user handed us a complete IP packet, use it.
    321 	 * Otherwise, allocate an mbuf for a header and fill it in.
    322 	 */
    323 	if ((inp->inp_flags & INP_HDRINCL) == 0) {
    324 		if ((m->m_pkthdr.len + sizeof(struct ip)) > IP_MAXPACKET) {
    325 			error = EMSGSIZE;
    326 			goto release;
    327 		}
    328 		M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
    329 		if (!m) {
    330 			error = ENOBUFS;
    331 			goto release;
    332 		}
    333 		ip = mtod(m, struct ip *);
    334 		ip->ip_tos = in4p_ip(inp).ip_tos;
    335 		ip->ip_off = htons(0);
    336 		ip->ip_p = in4p_ip(inp).ip_p;
    337 		ip->ip_len = htons(m->m_pkthdr.len);
    338 		ip->ip_src = pktopts.ippo_laddr.sin_addr;
    339 		ip->ip_dst = in4p_faddr(inp);
    340 		ip->ip_ttl = in4p_ip(inp).ip_ttl ? in4p_ip(inp).ip_ttl : MAXTTL;
    341 		opts = inp->inp_options;
    342 	} else {
    343 		if (m->m_pkthdr.len > IP_MAXPACKET) {
    344 			error = EMSGSIZE;
    345 			goto release;
    346 		}
    347 		if (m->m_pkthdr.len < sizeof(struct ip)) {
    348 			error = EINVAL;
    349 			goto release;
    350 		}
    351 		ip = mtod(m, struct ip *);
    352 
    353 		/*
    354 		 * If the mbuf is read-only, we need to allocate
    355 		 * a new mbuf for the header, since we need to
    356 		 * modify the header.
    357 		 */
    358 		if (M_READONLY(m)) {
    359 			int hlen = ip->ip_hl << 2;
    360 
    361 			m = m_copyup(m, hlen, (max_linkhdr + 3) & ~3);
    362 			if (m == NULL) {
    363 				error = ENOMEM;
    364 				goto release;
    365 			}
    366 			ip = mtod(m, struct ip *);
    367 		}
    368 
    369 		/* XXX userland passes ip_len and ip_off in host order */
    370 		if (m->m_pkthdr.len != ip->ip_len) {
    371 			error = EINVAL;
    372 			goto release;
    373 		}
    374 		HTONS(ip->ip_len);
    375 		HTONS(ip->ip_off);
    376 
    377 		if (ip->ip_id != 0 || m->m_pkthdr.len < IP_MINFRAGSIZE)
    378 			flags |= IP_NOIPNEWID;
    379 		opts = NULL;
    380 
    381 		/* Prevent ip_output from overwriting header fields. */
    382 		flags |= IP_RAWOUTPUT;
    383 
    384 		IP_STATINC(IP_STAT_RAWOUT);
    385 	}
    386 
    387 	/*
    388 	 * IP output.  Note: if IP_RETURNMTU flag is set, the MTU size
    389 	 * will be stored in inp_errormtu.
    390 	 */
    391 	return ip_output(m, opts, &inp->inp_route, flags, pktopts.ippo_imo,
    392 	    inp);
    393 
    394  release:
    395 	m_freem(m);
    396 	return error;
    397 }
    398 
    399 /*
    400  * Raw IP socket option processing.
    401  */
    402 int
    403 rip_ctloutput(int op, struct socket *so, struct sockopt *sopt)
    404 {
    405 	struct inpcb *inp = sotoinpcb(so);
    406 	int error = 0;
    407 	int optval;
    408 
    409 	if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_NOHEADER) {
    410 		if (op == PRCO_GETOPT) {
    411 			optval = (inp->inp_flags & INP_NOHEADER) ? 1 : 0;
    412 			error = sockopt_set(sopt, &optval, sizeof(optval));
    413 		} else if (op == PRCO_SETOPT) {
    414 			error = sockopt_getint(sopt, &optval);
    415 			if (error)
    416 				goto out;
    417 			if (optval) {
    418 				inp->inp_flags &= ~INP_HDRINCL;
    419 				inp->inp_flags |= INP_NOHEADER;
    420 			} else
    421 				inp->inp_flags &= ~INP_NOHEADER;
    422 		}
    423 		goto out;
    424 	} else if (sopt->sopt_level != IPPROTO_IP)
    425 		return ip_ctloutput(op, so, sopt);
    426 
    427 	switch (op) {
    428 
    429 	case PRCO_SETOPT:
    430 		switch (sopt->sopt_name) {
    431 		case IP_HDRINCL:
    432 			error = sockopt_getint(sopt, &optval);
    433 			if (error)
    434 				break;
    435 			if (optval)
    436 				inp->inp_flags |= INP_HDRINCL;
    437 			else
    438 				inp->inp_flags &= ~INP_HDRINCL;
    439 			break;
    440 
    441 #ifdef MROUTING
    442 		case MRT_INIT:
    443 		case MRT_DONE:
    444 		case MRT_ADD_VIF:
    445 		case MRT_DEL_VIF:
    446 		case MRT_ADD_MFC:
    447 		case MRT_DEL_MFC:
    448 		case MRT_ASSERT:
    449 		case MRT_API_CONFIG:
    450 		case MRT_ADD_BW_UPCALL:
    451 		case MRT_DEL_BW_UPCALL:
    452 			error = ip_mrouter_set(so, sopt);
    453 			break;
    454 #endif
    455 
    456 		default:
    457 			error = ip_ctloutput(op, so, sopt);
    458 			break;
    459 		}
    460 		break;
    461 
    462 	case PRCO_GETOPT:
    463 		switch (sopt->sopt_name) {
    464 		case IP_HDRINCL:
    465 			optval = inp->inp_flags & INP_HDRINCL;
    466 			error = sockopt_set(sopt, &optval, sizeof(optval));
    467 			break;
    468 
    469 #ifdef MROUTING
    470 		case MRT_VERSION:
    471 		case MRT_ASSERT:
    472 		case MRT_API_SUPPORT:
    473 		case MRT_API_CONFIG:
    474 			error = ip_mrouter_get(so, sopt);
    475 			break;
    476 #endif
    477 
    478 		default:
    479 			error = ip_ctloutput(op, so, sopt);
    480 			break;
    481 		}
    482 		break;
    483 	}
    484  out:
    485 	return error;
    486 }
    487 
    488 int
    489 rip_connect_pcb(struct inpcb *inp, struct sockaddr_in *addr)
    490 {
    491 
    492 	if (IFNET_READER_EMPTY())
    493 		return (EADDRNOTAVAIL);
    494 	if (addr->sin_family != AF_INET)
    495 		return (EAFNOSUPPORT);
    496 	if (addr->sin_len != sizeof(*addr))
    497 		return EINVAL;
    498 	in4p_faddr(inp) = addr->sin_addr;
    499 	return (0);
    500 }
    501 
    502 static void
    503 rip_disconnect1(struct inpcb *inp)
    504 {
    505 
    506 	in4p_faddr(inp) = zeroin_addr;
    507 }
    508 
    509 static int
    510 rip_attach(struct socket *so, int proto)
    511 {
    512 	struct inpcb *inp;
    513 	int error;
    514 
    515 	KASSERT(sotoinpcb(so) == NULL);
    516 	sosetlock(so);
    517 
    518 	if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
    519 		error = soreserve(so, rip_sendspace, rip_recvspace);
    520 		if (error) {
    521 			return error;
    522 		}
    523 	}
    524 
    525 	error = inpcb_create(so, &rawcbtable);
    526 	if (error) {
    527 		return error;
    528 	}
    529 	inp = sotoinpcb(so);
    530 	in4p_ip(inp).ip_p = proto;
    531 	KASSERT(solocked(so));
    532 
    533 	return 0;
    534 }
    535 
    536 static void
    537 rip_detach(struct socket *so)
    538 {
    539 	struct inpcb *inp;
    540 
    541 	KASSERT(solocked(so));
    542 	inp = sotoinpcb(so);
    543 	KASSERT(inp != NULL);
    544 
    545 #ifdef MROUTING
    546 	extern struct socket *ip_mrouter;
    547 	if (so == ip_mrouter) {
    548 		ip_mrouter_done();
    549 	}
    550 #endif
    551 	inpcb_destroy(inp);
    552 }
    553 
    554 static int
    555 rip_accept(struct socket *so, struct sockaddr *nam)
    556 {
    557 	KASSERT(solocked(so));
    558 
    559 	panic("rip_accept");
    560 
    561 	return EOPNOTSUPP;
    562 }
    563 
    564 static int
    565 rip_bind(struct socket *so, struct sockaddr *nam, struct lwp *l)
    566 {
    567 	struct inpcb *inp = sotoinpcb(so);
    568 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
    569 	int error = 0;
    570 	int s, ss;
    571 	struct ifaddr *ifa;
    572 
    573 	KASSERT(solocked(so));
    574 	KASSERT(inp != NULL);
    575 	KASSERT(nam != NULL);
    576 
    577 	if (addr->sin_len != sizeof(*addr))
    578 		return EINVAL;
    579 
    580 	s = splsoftnet();
    581 	if (IFNET_READER_EMPTY()) {
    582 		error = EADDRNOTAVAIL;
    583 		goto release;
    584 	}
    585 	if (addr->sin_family != AF_INET) {
    586 		error = EAFNOSUPPORT;
    587 		goto release;
    588 	}
    589 	ss = pserialize_read_enter();
    590 	if ((ifa = ifa_ifwithaddr(sintosa(addr))) == NULL &&
    591 	    (inp->inp_flags & INP_BINDANY) == 0 &&
    592 	    !in_nullhost(addr->sin_addr))
    593 	{
    594 		pserialize_read_exit(ss);
    595 		error = EADDRNOTAVAIL;
    596 		goto release;
    597 	}
    598         if (ifa && (ifatoia(ifa))->ia4_flags & IN6_IFF_DUPLICATED) {
    599 		pserialize_read_exit(ss);
    600 		error = EADDRNOTAVAIL;
    601 		goto release;
    602 	}
    603 	pserialize_read_exit(ss);
    604 
    605 	in4p_laddr(inp) = addr->sin_addr;
    606 
    607 release:
    608 	splx(s);
    609 	return error;
    610 }
    611 
    612 static int
    613 rip_listen(struct socket *so, struct lwp *l)
    614 {
    615 	KASSERT(solocked(so));
    616 
    617 	return EOPNOTSUPP;
    618 }
    619 
    620 static int
    621 rip_connect(struct socket *so, struct sockaddr *nam, struct lwp *l)
    622 {
    623 	struct inpcb *inp = sotoinpcb(so);
    624 	int error = 0;
    625 	int s;
    626 
    627 	KASSERT(solocked(so));
    628 	KASSERT(inp != NULL);
    629 	KASSERT(nam != NULL);
    630 
    631 	s = splsoftnet();
    632 	error = rip_connect_pcb(inp, (struct sockaddr_in *)nam);
    633 	if (! error)
    634 		soisconnected(so);
    635 	splx(s);
    636 
    637 	return error;
    638 }
    639 
    640 static int
    641 rip_connect2(struct socket *so, struct socket *so2)
    642 {
    643 	KASSERT(solocked(so));
    644 
    645 	return EOPNOTSUPP;
    646 }
    647 
    648 static int
    649 rip_disconnect(struct socket *so)
    650 {
    651 	struct inpcb *inp = sotoinpcb(so);
    652 	int s;
    653 
    654 	KASSERT(solocked(so));
    655 	KASSERT(inp != NULL);
    656 
    657 	s = splsoftnet();
    658 	soisdisconnected(so);
    659 	rip_disconnect1(inp);
    660 	splx(s);
    661 
    662 	return 0;
    663 }
    664 
    665 static int
    666 rip_shutdown(struct socket *so)
    667 {
    668 	int s;
    669 
    670 	KASSERT(solocked(so));
    671 
    672 	/*
    673 	 * Mark the connection as being incapable of further input.
    674 	 */
    675 	s = splsoftnet();
    676 	socantsendmore(so);
    677 	splx(s);
    678 
    679 	return 0;
    680 }
    681 
    682 static int
    683 rip_abort(struct socket *so)
    684 {
    685 	KASSERT(solocked(so));
    686 
    687 	panic("rip_abort");
    688 
    689 	return EOPNOTSUPP;
    690 }
    691 
    692 static int
    693 rip_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp)
    694 {
    695 	return in_control(so, cmd, nam, ifp);
    696 }
    697 
    698 static int
    699 rip_stat(struct socket *so, struct stat *ub)
    700 {
    701 	KASSERT(solocked(so));
    702 
    703 	/* stat: don't bother with a blocksize. */
    704 	return 0;
    705 }
    706 
    707 static int
    708 rip_peeraddr(struct socket *so, struct sockaddr *nam)
    709 {
    710 	int s;
    711 
    712 	KASSERT(solocked(so));
    713 	KASSERT(sotoinpcb(so) != NULL);
    714 	KASSERT(nam != NULL);
    715 
    716 	s = splsoftnet();
    717 	inpcb_fetch_peeraddr(sotoinpcb(so), (struct sockaddr_in *)nam);
    718 	splx(s);
    719 
    720 	return 0;
    721 }
    722 
    723 static int
    724 rip_sockaddr(struct socket *so, struct sockaddr *nam)
    725 {
    726 	int s;
    727 
    728 	KASSERT(solocked(so));
    729 	KASSERT(sotoinpcb(so) != NULL);
    730 	KASSERT(nam != NULL);
    731 
    732 	s = splsoftnet();
    733 	inpcb_fetch_sockaddr(sotoinpcb(so), (struct sockaddr_in *)nam);
    734 	splx(s);
    735 
    736 	return 0;
    737 }
    738 
    739 static int
    740 rip_rcvd(struct socket *so, int flags, struct lwp *l)
    741 {
    742 	KASSERT(solocked(so));
    743 
    744 	return EOPNOTSUPP;
    745 }
    746 
    747 static int
    748 rip_recvoob(struct socket *so, struct mbuf *m, int flags)
    749 {
    750 	KASSERT(solocked(so));
    751 
    752 	return EOPNOTSUPP;
    753 }
    754 
    755 static int
    756 rip_send(struct socket *so, struct mbuf *m, struct sockaddr *nam,
    757     struct mbuf *control, struct lwp *l)
    758 {
    759 	struct inpcb *inp = sotoinpcb(so);
    760 	int error = 0;
    761 	int s;
    762 
    763 	KASSERT(solocked(so));
    764 	KASSERT(inp != NULL);
    765 	KASSERT(m != NULL);
    766 
    767 	/*
    768 	 * Ship a packet out.  The appropriate raw output
    769 	 * routine handles any massaging necessary.
    770 	 */
    771 	s = splsoftnet();
    772 	if (nam) {
    773 		if ((so->so_state & SS_ISCONNECTED) != 0) {
    774 			error = EISCONN;
    775 			goto die;
    776 		}
    777 		error = rip_connect_pcb(inp, (struct sockaddr_in *)nam);
    778 		if (error)
    779 			goto die;
    780 	} else {
    781 		if ((so->so_state & SS_ISCONNECTED) == 0) {
    782 			error = ENOTCONN;
    783 			goto die;
    784 		}
    785 	}
    786 	error = rip_output(m, inp, control, l);
    787 	m = NULL;
    788 	control = NULL;
    789 	if (nam)
    790 		rip_disconnect1(inp);
    791  die:
    792 	m_freem(m);
    793 	m_freem(control);
    794 
    795 	splx(s);
    796 	return error;
    797 }
    798 
    799 static int
    800 rip_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control)
    801 {
    802 	KASSERT(solocked(so));
    803 
    804 	m_freem(m);
    805 	m_freem(control);
    806 
    807 	return EOPNOTSUPP;
    808 }
    809 
    810 static int
    811 rip_purgeif(struct socket *so, struct ifnet *ifp)
    812 {
    813 	int s;
    814 
    815 	s = splsoftnet();
    816 	mutex_enter(softnet_lock);
    817 	inpcb_purgeif0(&rawcbtable, ifp);
    818 #ifdef NET_MPSAFE
    819 	mutex_exit(softnet_lock);
    820 #endif
    821 	in_purgeif(ifp);
    822 #ifdef NET_MPSAFE
    823 	mutex_enter(softnet_lock);
    824 #endif
    825 	inpcb_purgeif(&rawcbtable, ifp);
    826 	mutex_exit(softnet_lock);
    827 	splx(s);
    828 
    829 	return 0;
    830 }
    831 
    832 PR_WRAP_USRREQS(rip)
    833 #define	rip_attach	rip_attach_wrapper
    834 #define	rip_detach	rip_detach_wrapper
    835 #define	rip_accept	rip_accept_wrapper
    836 #define	rip_bind	rip_bind_wrapper
    837 #define	rip_listen	rip_listen_wrapper
    838 #define	rip_connect	rip_connect_wrapper
    839 #define	rip_connect2	rip_connect2_wrapper
    840 #define	rip_disconnect	rip_disconnect_wrapper
    841 #define	rip_shutdown	rip_shutdown_wrapper
    842 #define	rip_abort	rip_abort_wrapper
    843 #define	rip_ioctl	rip_ioctl_wrapper
    844 #define	rip_stat	rip_stat_wrapper
    845 #define	rip_peeraddr	rip_peeraddr_wrapper
    846 #define	rip_sockaddr	rip_sockaddr_wrapper
    847 #define	rip_rcvd	rip_rcvd_wrapper
    848 #define	rip_recvoob	rip_recvoob_wrapper
    849 #define	rip_send	rip_send_wrapper
    850 #define	rip_sendoob	rip_sendoob_wrapper
    851 #define	rip_purgeif	rip_purgeif_wrapper
    852 
    853 const struct pr_usrreqs rip_usrreqs = {
    854 	.pr_attach	= rip_attach,
    855 	.pr_detach	= rip_detach,
    856 	.pr_accept	= rip_accept,
    857 	.pr_bind	= rip_bind,
    858 	.pr_listen	= rip_listen,
    859 	.pr_connect	= rip_connect,
    860 	.pr_connect2	= rip_connect2,
    861 	.pr_disconnect	= rip_disconnect,
    862 	.pr_shutdown	= rip_shutdown,
    863 	.pr_abort	= rip_abort,
    864 	.pr_ioctl	= rip_ioctl,
    865 	.pr_stat	= rip_stat,
    866 	.pr_peeraddr	= rip_peeraddr,
    867 	.pr_sockaddr	= rip_sockaddr,
    868 	.pr_rcvd	= rip_rcvd,
    869 	.pr_recvoob	= rip_recvoob,
    870 	.pr_send	= rip_send,
    871 	.pr_sendoob	= rip_sendoob,
    872 	.pr_purgeif	= rip_purgeif,
    873 };
    874 
    875 static void
    876 sysctl_net_inet_raw_setup(struct sysctllog **clog)
    877 {
    878 
    879 	sysctl_createv(clog, 0, NULL, NULL,
    880 		       CTLFLAG_PERMANENT,
    881 		       CTLTYPE_NODE, "inet", NULL,
    882 		       NULL, 0, NULL, 0,
    883 		       CTL_NET, PF_INET, CTL_EOL);
    884 	sysctl_createv(clog, 0, NULL, NULL,
    885 		       CTLFLAG_PERMANENT,
    886 		       CTLTYPE_NODE, "raw",
    887 		       SYSCTL_DESCR("Raw IPv4 settings"),
    888 		       NULL, 0, NULL, 0,
    889 		       CTL_NET, PF_INET, IPPROTO_RAW, CTL_EOL);
    890 
    891 	sysctl_createv(clog, 0, NULL, NULL,
    892 		       CTLFLAG_PERMANENT,
    893 		       CTLTYPE_STRUCT, "pcblist",
    894 		       SYSCTL_DESCR("Raw IPv4 control block list"),
    895 		       sysctl_inpcblist, 0, &rawcbtable, 0,
    896 		       CTL_NET, PF_INET, IPPROTO_RAW,
    897 		       CTL_CREATE, CTL_EOL);
    898 }
    899