Home | History | Annotate | Line # | Download | only in netinet
raw_ip.c revision 1.44
      1 /*	$NetBSD: raw_ip.c,v 1.44 1999/07/05 07:24:38 darrenr 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. All advertising materials mentioning features or use of this software
     45  *    must display the following acknowledgement:
     46  *	This product includes software developed by the University of
     47  *	California, Berkeley and its contributors.
     48  * 4. Neither the name of the University nor the names of its contributors
     49  *    may be used to endorse or promote products derived from this software
     50  *    without specific prior written permission.
     51  *
     52  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     53  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     54  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     55  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     56  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     57  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     58  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     59  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     60  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     61  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     62  * SUCH DAMAGE.
     63  *
     64  *	@(#)raw_ip.c	8.7 (Berkeley) 5/15/95
     65  */
     66 
     67 #include "opt_mrouting.h"
     68 
     69 #include <sys/param.h>
     70 #include <sys/malloc.h>
     71 #include <sys/mbuf.h>
     72 #include <sys/socket.h>
     73 #include <sys/protosw.h>
     74 #include <sys/socketvar.h>
     75 #include <sys/errno.h>
     76 #include <sys/systm.h>
     77 #include <sys/proc.h>
     78 
     79 #include <net/if.h>
     80 #include <net/route.h>
     81 
     82 #include <netinet/in.h>
     83 #include <netinet/in_systm.h>
     84 #include <netinet/ip.h>
     85 #include <netinet/ip_var.h>
     86 #include <netinet/ip_mroute.h>
     87 #include <netinet/ip_icmp.h>
     88 #include <netinet/in_pcb.h>
     89 #include <netinet/in_var.h>
     90 
     91 #include <machine/stdarg.h>
     92 
     93 #ifdef IPSEC
     94 #include <netinet6/ipsec.h>
     95 #endif /*IPSEC*/
     96 
     97 extern u_char ip_protox[];
     98 extern struct  protosw inetsw[];
     99 struct inpcbtable rawcbtable;
    100 
    101 int	 rip_bind __P((struct inpcb *, struct mbuf *));
    102 int	 rip_connect __P((struct inpcb *, struct mbuf *));
    103 void	 rip_disconnect __P((struct inpcb *));
    104 
    105 /*
    106  * Nominal space allocated to a raw ip socket.
    107  */
    108 #define	RIPSNDQ		8192
    109 #define	RIPRCVQ		8192
    110 
    111 /*
    112  * Raw interface to IP protocol.
    113  */
    114 
    115 /*
    116  * Initialize raw connection block q.
    117  */
    118 void
    119 rip_init()
    120 {
    121 
    122 	in_pcbinit(&rawcbtable, 1, 1);
    123 }
    124 
    125 static struct	sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET };
    126 
    127 /*
    128  * Setup generic address and protocol structures
    129  * for raw_input routine, then pass them along with
    130  * mbuf chain.
    131  */
    132 void
    133 #if __STDC__
    134 rip_input(struct mbuf *m, ...)
    135 #else
    136 rip_input(m, va_alist)
    137 	struct mbuf *m;
    138 	va_dcl
    139 #endif
    140 {
    141 	int off, proto;
    142 	register struct ip *ip = mtod(m, struct ip *);
    143 	register struct inpcb *inp;
    144 	struct inpcb *last = 0;
    145 	struct mbuf *opts = 0;
    146 	struct sockaddr_in ripsrc;
    147 	va_list ap;
    148 
    149 	va_start(ap, m);
    150 	off = va_arg(ap, int);
    151 	proto = va_arg(ap, int);
    152 	va_end(ap);
    153 
    154 	ripsrc.sin_family = AF_INET;
    155 	ripsrc.sin_len = sizeof(struct sockaddr_in);
    156 	ripsrc.sin_addr = ip->ip_src;
    157 	ripsrc.sin_port = 0;
    158 	bzero((caddr_t)ripsrc.sin_zero, sizeof(ripsrc.sin_zero));
    159 
    160 	/*
    161 	 * XXX Compatibility: programs using raw IP expect ip_len
    162 	 * XXX to have the header length subtracted.
    163 	 */
    164 	ip->ip_len -= ip->ip_hl << 2;
    165 
    166 	for (inp = rawcbtable.inpt_queue.cqh_first;
    167 	    inp != (struct inpcb *)&rawcbtable.inpt_queue;
    168 	    inp = inp->inp_queue.cqe_next) {
    169 		if (inp->inp_ip.ip_p && inp->inp_ip.ip_p != proto)
    170 			continue;
    171 		if (!in_nullhost(inp->inp_laddr) &&
    172 		    !in_hosteq(inp->inp_laddr, ip->ip_dst))
    173 			continue;
    174 		if (!in_nullhost(inp->inp_faddr) &&
    175 		    !in_hosteq(inp->inp_faddr, ip->ip_src))
    176 			continue;
    177 		if (last) {
    178 			struct mbuf *n;
    179 			if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) {
    180 				if (last->inp_flags & INP_CONTROLOPTS ||
    181 				    last->inp_socket->so_options & SO_TIMESTAMP)
    182 					ip_savecontrol(last, &opts, ip, n);
    183 				if (sbappendaddr(&last->inp_socket->so_rcv,
    184 				    sintosa(&ripsrc), n, opts) == 0) {
    185 					/* should notify about lost packet */
    186 					m_freem(n);
    187 					if (opts)
    188 						m_freem(opts);
    189 				} else
    190 					sorwakeup(last->inp_socket);
    191 				opts = NULL;
    192 			}
    193 		}
    194 		last = inp;
    195 	}
    196 	if (last) {
    197 		if (last->inp_flags & INP_CONTROLOPTS ||
    198 		    last->inp_socket->so_options & SO_TIMESTAMP)
    199 			ip_savecontrol(last, &opts, ip, m);
    200 		if (sbappendaddr(&last->inp_socket->so_rcv,
    201 		    sintosa(&ripsrc), m, opts) == 0) {
    202 			m_freem(m);
    203 			if (opts)
    204 				m_freem(opts);
    205 		} else
    206 			sorwakeup(last->inp_socket);
    207 	} else {
    208 		if (inetsw[ip_protox[ip->ip_p]].pr_input == rip_input) {
    209 			icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PROTOCOL,0,0);
    210 			ipstat.ips_noproto++;
    211 			ipstat.ips_delivered--;
    212 		} else
    213 			m_freem(m);
    214 	}
    215 	return;
    216 }
    217 
    218 /*
    219  * Generate IP header and pass packet to ip_output.
    220  * Tack on options user may have setup with control call.
    221  */
    222 int
    223 #if __STDC__
    224 rip_output(struct mbuf *m, ...)
    225 #else
    226 rip_output(m, va_alist)
    227 	struct mbuf *m;
    228 	va_dcl
    229 #endif
    230 {
    231 	register struct inpcb *inp;
    232 	register struct ip *ip;
    233 	struct mbuf *opts;
    234 	int flags;
    235 	va_list ap;
    236 
    237 	va_start(ap, m);
    238 	inp = va_arg(ap, struct inpcb *);
    239 	va_end(ap);
    240 
    241 	flags =
    242 	    (inp->inp_socket->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST
    243 	    | IP_RETURNMTU;
    244 
    245 	/*
    246 	 * If the user handed us a complete IP packet, use it.
    247 	 * Otherwise, allocate an mbuf for a header and fill it in.
    248 	 */
    249 	if ((inp->inp_flags & INP_HDRINCL) == 0) {
    250 		if ((m->m_pkthdr.len + sizeof(struct ip)) > IP_MAXPACKET) {
    251 			m_freem(m);
    252 			return (EMSGSIZE);
    253 		}
    254 		M_PREPEND(m, sizeof(struct ip), M_WAIT);
    255 		ip = mtod(m, struct ip *);
    256 		ip->ip_tos = 0;
    257 		ip->ip_off = 0;
    258 		ip->ip_p = inp->inp_ip.ip_p;
    259 		ip->ip_len = m->m_pkthdr.len;
    260 		ip->ip_src = inp->inp_laddr;
    261 		ip->ip_dst = inp->inp_faddr;
    262 		ip->ip_ttl = MAXTTL;
    263 		opts = inp->inp_options;
    264 	} else {
    265 		if (m->m_pkthdr.len > IP_MAXPACKET) {
    266 			m_freem(m);
    267 			return (EMSGSIZE);
    268 		}
    269 		ip = mtod(m, struct ip *);
    270 		if (m->m_pkthdr.len != ip->ip_len) {
    271 			m_freem(m);
    272 			return (EINVAL);
    273 		}
    274 		if (ip->ip_id == 0)
    275 			ip->ip_id = htons(ip_id++);
    276 		opts = NULL;
    277 		/* XXX prevent ip_output from overwriting header fields */
    278 		flags |= IP_RAWOUTPUT;
    279 		ipstat.ips_rawout++;
    280 	}
    281 #ifdef IPSEC
    282 	m->m_pkthdr.rcvif = (struct ifnet *)inp->inp_socket;	/*XXX*/
    283 #endif /*IPSEC*/
    284 	return (ip_output(m, opts, &inp->inp_route, flags, inp->inp_moptions, &inp->inp_errormtu));
    285 }
    286 
    287 /*
    288  * Raw IP socket option processing.
    289  */
    290 int
    291 rip_ctloutput(op, so, level, optname, m)
    292 	int op;
    293 	struct socket *so;
    294 	int level, optname;
    295 	struct mbuf **m;
    296 {
    297 	register struct inpcb *inp = sotoinpcb(so);
    298 	int error = 0;
    299 
    300 	if (level != IPPROTO_IP) {
    301 		error = ENOPROTOOPT;
    302 		if (op == PRCO_SETOPT && *m != 0)
    303 			(void) m_free(*m);
    304 	} else switch (op) {
    305 
    306 	case PRCO_SETOPT:
    307 		switch (optname) {
    308 		case IP_HDRINCL:
    309 			if (*m == 0 || (*m)->m_len < sizeof (int))
    310 				error = EINVAL;
    311 			else {
    312 				if (*mtod(*m, int *))
    313 					inp->inp_flags |= INP_HDRINCL;
    314 				else
    315 					inp->inp_flags &= ~INP_HDRINCL;
    316 			}
    317 			if (*m != 0)
    318 				(void) m_free(*m);
    319 			break;
    320 
    321 #ifdef MROUTING
    322 		case MRT_INIT:
    323 		case MRT_DONE:
    324 		case MRT_ADD_VIF:
    325 		case MRT_DEL_VIF:
    326 		case MRT_ADD_MFC:
    327 		case MRT_DEL_MFC:
    328 		case MRT_ASSERT:
    329 			error = ip_mrouter_set(so, optname, m);
    330 			break;
    331 #endif
    332 
    333 		default:
    334 			error = ip_ctloutput(op, so, level, optname, m);
    335 			break;
    336 		}
    337 		break;
    338 
    339 	case PRCO_GETOPT:
    340 		switch (optname) {
    341 		case IP_HDRINCL:
    342 			*m = m_get(M_WAIT, M_SOOPTS);
    343 			(*m)->m_len = sizeof (int);
    344 			*mtod(*m, int *) = inp->inp_flags & INP_HDRINCL ? 1 : 0;
    345 			break;
    346 
    347 #ifdef MROUTING
    348 		case MRT_VERSION:
    349 		case MRT_ASSERT:
    350 			error = ip_mrouter_get(so, optname, m);
    351 			break;
    352 #endif
    353 
    354 		default:
    355 			error = ip_ctloutput(op, so, level, optname, m);
    356 			break;
    357 		}
    358 		break;
    359 	}
    360 	return (error);
    361 }
    362 
    363 int
    364 rip_bind(inp, nam)
    365 	struct inpcb *inp;
    366 	struct mbuf *nam;
    367 {
    368 	struct sockaddr_in *addr = mtod(nam, struct sockaddr_in *);
    369 
    370 	if (nam->m_len != sizeof(*addr))
    371 		return (EINVAL);
    372 	if (ifnet.tqh_first == 0)
    373 		return (EADDRNOTAVAIL);
    374 	if (addr->sin_family != AF_INET &&
    375 	    addr->sin_family != AF_IMPLINK)
    376 		return (EAFNOSUPPORT);
    377 	if (!in_nullhost(addr->sin_addr) &&
    378 	    ifa_ifwithaddr(sintosa(addr)) == 0)
    379 		return (EADDRNOTAVAIL);
    380 	inp->inp_laddr = addr->sin_addr;
    381 	return (0);
    382 }
    383 
    384 int
    385 rip_connect(inp, nam)
    386 	struct inpcb *inp;
    387 	struct mbuf *nam;
    388 {
    389 	struct sockaddr_in *addr = mtod(nam, struct sockaddr_in *);
    390 
    391 	if (nam->m_len != sizeof(*addr))
    392 		return (EINVAL);
    393 	if (ifnet.tqh_first == 0)
    394 		return (EADDRNOTAVAIL);
    395 	if (addr->sin_family != AF_INET &&
    396 	    addr->sin_family != AF_IMPLINK)
    397 		return (EAFNOSUPPORT);
    398 	inp->inp_faddr = addr->sin_addr;
    399 	return (0);
    400 }
    401 
    402 void
    403 rip_disconnect(inp)
    404 	struct inpcb *inp;
    405 {
    406 
    407 	inp->inp_faddr = zeroin_addr;
    408 }
    409 
    410 u_long	rip_sendspace = RIPSNDQ;
    411 u_long	rip_recvspace = RIPRCVQ;
    412 
    413 /*ARGSUSED*/
    414 int
    415 rip_usrreq(so, req, m, nam, control, p)
    416 	register struct socket *so;
    417 	int req;
    418 	struct mbuf *m, *nam, *control;
    419 	struct proc *p;
    420 {
    421 	register struct inpcb *inp;
    422 	int s;
    423 	register int error = 0;
    424 #ifdef MROUTING
    425 	extern struct socket *ip_mrouter;
    426 #endif
    427 
    428 	if (req == PRU_CONTROL)
    429 		return (in_control(so, (long)m, (caddr_t)nam,
    430 		    (struct ifnet *)control, p));
    431 
    432 	s = splsoftnet();
    433 	inp = sotoinpcb(so);
    434 #ifdef DIAGNOSTIC
    435 	if (req != PRU_SEND && req != PRU_SENDOOB && control)
    436 		panic("rip_usrreq: unexpected control mbuf");
    437 #endif
    438 	if (inp == 0 && req != PRU_ATTACH) {
    439 		error = EINVAL;
    440 		goto release;
    441 	}
    442 
    443 	switch (req) {
    444 
    445 	case PRU_ATTACH:
    446 		if (inp != 0) {
    447 			error = EISCONN;
    448 			break;
    449 		}
    450 		if (p == 0 || (error = suser(p->p_ucred, &p->p_acflag))) {
    451 			error = EACCES;
    452 			break;
    453 		}
    454 		if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
    455 			error = soreserve(so, rip_sendspace, rip_recvspace);
    456 			if (error)
    457 				break;
    458 		}
    459 		error = in_pcballoc(so, &rawcbtable);
    460 		if (error)
    461 			break;
    462 		inp = sotoinpcb(so);
    463 		inp->inp_ip.ip_p = (long)nam;
    464 #ifdef IPSEC
    465 		error = ipsec_init_policy(&inp->inp_sp);
    466 #endif /*IPSEC*/
    467 		break;
    468 
    469 	case PRU_DETACH:
    470 #ifdef MROUTING
    471 		if (so == ip_mrouter)
    472 			ip_mrouter_done();
    473 #endif
    474 		in_pcbdetach(inp);
    475 		break;
    476 
    477 	case PRU_BIND:
    478 		error = rip_bind(inp, nam);
    479 		break;
    480 
    481 	case PRU_LISTEN:
    482 		error = EOPNOTSUPP;
    483 		break;
    484 
    485 	case PRU_CONNECT:
    486 		error = rip_connect(inp, nam);
    487 		if (error)
    488 			break;
    489 		soisconnected(so);
    490 		break;
    491 
    492 	case PRU_CONNECT2:
    493 		error = EOPNOTSUPP;
    494 		break;
    495 
    496 	case PRU_DISCONNECT:
    497 		soisdisconnected(so);
    498 		rip_disconnect(inp);
    499 		break;
    500 
    501 	/*
    502 	 * Mark the connection as being incapable of further input.
    503 	 */
    504 	case PRU_SHUTDOWN:
    505 		socantsendmore(so);
    506 		break;
    507 
    508 	case PRU_RCVD:
    509 		error = EOPNOTSUPP;
    510 		break;
    511 
    512 	/*
    513 	 * Ship a packet out.  The appropriate raw output
    514 	 * routine handles any massaging necessary.
    515 	 */
    516 	case PRU_SEND:
    517 		if (control && control->m_len) {
    518 			m_freem(control);
    519 			m_freem(m);
    520 			error = EINVAL;
    521 			break;
    522 		}
    523 	{
    524 		if (nam) {
    525 			if ((so->so_state & SS_ISCONNECTED) != 0) {
    526 				error = EISCONN;
    527 				goto die;
    528 			}
    529 			error = rip_connect(inp, nam);
    530 			if (error) {
    531 			die:
    532 				m_freem(m);
    533 				break;
    534 			}
    535 		} else {
    536 			if ((so->so_state & SS_ISCONNECTED) == 0) {
    537 				error = ENOTCONN;
    538 				goto die;
    539 			}
    540 		}
    541 		error = rip_output(m, inp);
    542 		if (nam)
    543 			rip_disconnect(inp);
    544 	}
    545 		break;
    546 
    547 	case PRU_SENSE:
    548 		/*
    549 		 * stat: don't bother with a blocksize.
    550 		 */
    551 		splx(s);
    552 		return (0);
    553 
    554 	case PRU_RCVOOB:
    555 		error = EOPNOTSUPP;
    556 		break;
    557 
    558 	case PRU_SENDOOB:
    559 		m_freem(control);
    560 		m_freem(m);
    561 		error = EOPNOTSUPP;
    562 		break;
    563 
    564 	case PRU_SOCKADDR:
    565 		in_setsockaddr(inp, nam);
    566 		break;
    567 
    568 	case PRU_PEERADDR:
    569 		in_setpeeraddr(inp, nam);
    570 		break;
    571 
    572 	default:
    573 		panic("rip_usrreq");
    574 	}
    575 
    576 release:
    577 	splx(s);
    578 	return (error);
    579 }
    580