xref: /openbsd-src/sys/netinet/raw_ip.c (revision 78fec973f57e9fc9edd564490c79661460ad807b)
1 /*	$OpenBSD: raw_ip.c,v 1.128 2022/05/15 09:12:20 dlg Exp $	*/
2 /*	$NetBSD: raw_ip.c,v 1.25 1996/02/18 18:58:33 christos Exp $	*/
3 
4 /*
5  * Copyright (c) 1982, 1986, 1988, 1993
6  *	The Regents of the University of California.  All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the name of the University nor the names of its contributors
17  *    may be used to endorse or promote products derived from this software
18  *    without specific prior written permission.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30  * SUCH DAMAGE.
31  *
32  *	@(#)COPYRIGHT	1.1 (NRL) 17 January 1995
33  *
34  * NRL grants permission for redistribution and use in source and binary
35  * forms, with or without modification, of the software and documentation
36  * created at NRL provided that the following conditions are met:
37  *
38  * 1. Redistributions of source code must retain the above copyright
39  *    notice, this list of conditions and the following disclaimer.
40  * 2. Redistributions in binary form must reproduce the above copyright
41  *    notice, this list of conditions and the following disclaimer in the
42  *    documentation and/or other materials provided with the distribution.
43  * 3. All advertising materials mentioning features or use of this software
44  *    must display the following acknowledgements:
45  *	This product includes software developed by the University of
46  *	California, Berkeley and its contributors.
47  *	This product includes software developed at the Information
48  *	Technology Division, US Naval Research Laboratory.
49  * 4. Neither the name of the NRL nor the names of its contributors
50  *    may be used to endorse or promote products derived from this software
51  *    without specific prior written permission.
52  *
53  * THE SOFTWARE PROVIDED BY NRL IS PROVIDED BY NRL AND CONTRIBUTORS ``AS
54  * IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
55  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
56  * PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL NRL OR
57  * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
58  * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
59  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
60  * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
61  * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
62  * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
63  * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
64  *
65  * The views and conclusions contained in the software and documentation
66  * are those of the authors and should not be interpreted as representing
67  * official policies, either expressed or implied, of the US Naval
68  * Research Laboratory (NRL).
69  */
70 
71 #include <sys/param.h>
72 #include <sys/systm.h>
73 #include <sys/mbuf.h>
74 #include <sys/socket.h>
75 #include <sys/protosw.h>
76 #include <sys/socketvar.h>
77 
78 #include <net/if.h>
79 #include <net/if_var.h>
80 #include <net/route.h>
81 
82 #include <netinet/in.h>
83 #include <netinet/ip.h>
84 #include <netinet/ip_mroute.h>
85 #include <netinet/ip_var.h>
86 #include <netinet/in_pcb.h>
87 #include <netinet/in_var.h>
88 #include <netinet/ip_icmp.h>
89 
90 #include <net/pfvar.h>
91 
92 #include "pf.h"
93 
94 struct inpcbtable rawcbtable;
95 
96 /*
97  * Nominal space allocated to a raw ip socket.
98  */
99 #define	RIPSNDQ		8192
100 #define	RIPRCVQ		8192
101 
102 /*
103  * Raw interface to IP protocol.
104  */
105 
106 /*
107  * Initialize raw connection block q.
108  */
109 void
110 rip_init(void)
111 {
112 	in_pcbinit(&rawcbtable, 1);
113 }
114 
115 struct mbuf	*rip_chkhdr(struct mbuf *, struct mbuf *);
116 
117 int
118 rip_input(struct mbuf **mp, int *offp, int proto, int af)
119 {
120 	struct mbuf *m = *mp;
121 	struct ip *ip = mtod(m, struct ip *);
122 	struct inpcb *inp;
123 	SIMPLEQ_HEAD(, inpcb) inpcblist;
124 	struct in_addr *key;
125 	struct counters_ref ref;
126 	uint64_t *counters;
127 	struct sockaddr_in ripsrc;
128 
129 	KASSERT(af == AF_INET);
130 
131 	memset(&ripsrc, 0, sizeof(ripsrc));
132 	ripsrc.sin_family = AF_INET;
133 	ripsrc.sin_len = sizeof(ripsrc);
134 	ripsrc.sin_addr = ip->ip_src;
135 
136 	key = &ip->ip_dst;
137 #if NPF > 0
138 	if (m->m_pkthdr.pf.flags & PF_TAG_DIVERTED) {
139 		struct pf_divert *divert;
140 
141 		divert = pf_find_divert(m);
142 		KASSERT(divert != NULL);
143 		switch (divert->type) {
144 		case PF_DIVERT_TO:
145 			key = &divert->addr.v4;
146 			break;
147 		case PF_DIVERT_REPLY:
148 			break;
149 		default:
150 			panic("%s: unknown divert type %d, mbuf %p, divert %p",
151 			    __func__, divert->type, m, divert);
152 		}
153 	}
154 #endif
155 	NET_ASSERT_WLOCKED();
156 	SIMPLEQ_INIT(&inpcblist);
157 	mtx_enter(&rawcbtable.inpt_mtx);
158 	TAILQ_FOREACH(inp, &rawcbtable.inpt_queue, inp_queue) {
159 		if (inp->inp_socket->so_state & SS_CANTRCVMORE)
160 			continue;
161 #ifdef INET6
162 		if (inp->inp_flags & INP_IPV6)
163 			continue;
164 #endif
165 		if (rtable_l2(inp->inp_rtableid) !=
166 		    rtable_l2(m->m_pkthdr.ph_rtableid))
167 			continue;
168 
169 		if (inp->inp_ip.ip_p && inp->inp_ip.ip_p != ip->ip_p)
170 			continue;
171 		if (inp->inp_laddr.s_addr &&
172 		    inp->inp_laddr.s_addr != key->s_addr)
173 			continue;
174 		if (inp->inp_faddr.s_addr &&
175 		    inp->inp_faddr.s_addr != ip->ip_src.s_addr)
176 			continue;
177 
178 		in_pcbref(inp);
179 		SIMPLEQ_INSERT_TAIL(&inpcblist, inp, inp_notify);
180 	}
181 	mtx_leave(&rawcbtable.inpt_mtx);
182 
183 	if (SIMPLEQ_EMPTY(&inpcblist)) {
184 		if (ip->ip_p != IPPROTO_ICMP)
185 			icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PROTOCOL,
186 			    0, 0);
187 		else
188 			m_freem(m);
189 
190 		counters = counters_enter(&ref, ipcounters);
191 		counters[ips_noproto]++;
192 		counters[ips_delivered]--;
193 		counters_leave(&ref, ipcounters);
194 	}
195 
196 	while ((inp = SIMPLEQ_FIRST(&inpcblist)) != NULL) {
197 		struct mbuf *n, *opts = NULL;
198 
199 		SIMPLEQ_REMOVE_HEAD(&inpcblist, inp_notify);
200 		if (SIMPLEQ_EMPTY(&inpcblist))
201 			n = m;
202 		else
203 			n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
204 		if (n != NULL) {
205 			if (inp->inp_flags & INP_CONTROLOPTS ||
206 			    inp->inp_socket->so_options & SO_TIMESTAMP)
207 				ip_savecontrol(inp, &opts, ip, n);
208 			if (sbappendaddr(inp->inp_socket,
209 			    &inp->inp_socket->so_rcv,
210 			    sintosa(&ripsrc), n, opts) == 0) {
211 				/* should notify about lost packet */
212 				m_freem(n);
213 				m_freem(opts);
214 			} else
215 				sorwakeup(inp->inp_socket);
216 		}
217 		in_pcbunref(inp);
218 	}
219 	return IPPROTO_DONE;
220 }
221 
222 /*
223  * Generate IP header and pass packet to ip_output.
224  * Tack on options user may have setup with control call.
225  */
226 int
227 rip_output(struct mbuf *m, struct socket *so, struct sockaddr *dstaddr,
228     struct mbuf *control)
229 {
230 	struct sockaddr_in *dst = satosin(dstaddr);
231 	struct ip *ip;
232 	struct inpcb *inp;
233 	int flags, error;
234 
235 	inp = sotoinpcb(so);
236 	flags = IP_ALLOWBROADCAST;
237 
238 	/*
239 	 * If the user handed us a complete IP packet, use it.
240 	 * Otherwise, allocate an mbuf for a header and fill it in.
241 	 */
242 	if ((inp->inp_flags & INP_HDRINCL) == 0) {
243 		if ((m->m_pkthdr.len + sizeof(struct ip)) > IP_MAXPACKET) {
244 			m_freem(m);
245 			return (EMSGSIZE);
246 		}
247 		M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
248 		if (!m)
249 			return (ENOBUFS);
250 		ip = mtod(m, struct ip *);
251 		ip->ip_tos = inp->inp_ip.ip_tos;
252 		ip->ip_off = htons(0);
253 		ip->ip_p = inp->inp_ip.ip_p;
254 		ip->ip_len = htons(m->m_pkthdr.len);
255 		ip->ip_src.s_addr = INADDR_ANY;
256 		ip->ip_dst = dst->sin_addr;
257 		ip->ip_ttl = inp->inp_ip.ip_ttl ? inp->inp_ip.ip_ttl : MAXTTL;
258 	} else {
259 		if (m->m_pkthdr.len > IP_MAXPACKET) {
260 			m_freem(m);
261 			return (EMSGSIZE);
262 		}
263 
264 		m = rip_chkhdr(m, inp->inp_options);
265 		if (m == NULL)
266 			return (EINVAL);
267 
268 		ip = mtod(m, struct ip *);
269 		if (ip->ip_id == 0)
270 			ip->ip_id = htons(ip_randomid());
271 		dst->sin_addr = ip->ip_dst;
272 
273 		/* XXX prevent ip_output from overwriting header fields */
274 		flags |= IP_RAWOUTPUT;
275 		ipstat_inc(ips_rawout);
276 	}
277 
278 	if (ip->ip_src.s_addr == INADDR_ANY) {
279 		error = in_pcbselsrc(&ip->ip_src, dst, inp);
280 		if (error != 0)
281 			return (error);
282 	}
283 
284 #ifdef INET6
285 	/*
286 	 * A thought:  Even though raw IP shouldn't be able to set IPv6
287 	 *             multicast options, if it does, the last parameter to
288 	 *             ip_output should be guarded against v6/v4 problems.
289 	 */
290 #endif
291 	/* force routing table */
292 	m->m_pkthdr.ph_rtableid = inp->inp_rtableid;
293 
294 #if NPF > 0
295 	if (inp->inp_socket->so_state & SS_ISCONNECTED &&
296 	    ip->ip_p != IPPROTO_ICMP)
297 		pf_mbuf_link_inpcb(m, inp);
298 #endif
299 
300 	error = ip_output(m, inp->inp_options, &inp->inp_route, flags,
301 	    inp->inp_moptions, inp, 0);
302 	return (error);
303 }
304 
305 struct mbuf *
306 rip_chkhdr(struct mbuf *m, struct mbuf *options)
307 {
308 	struct ip *ip;
309 	int hlen, opt, optlen, cnt;
310 	u_char *cp;
311 
312 	if (m->m_pkthdr.len < sizeof(struct ip)) {
313 		m_freem(m);
314 		return NULL;
315 	}
316 
317 	m = m_pullup(m, sizeof (struct ip));
318 	if (m == NULL)
319 		return NULL;
320 
321 	ip = mtod(m, struct ip *);
322 	hlen = ip->ip_hl << 2;
323 
324 	/* Don't allow packet length sizes that will crash. */
325 	if (hlen < sizeof (struct ip) ||
326 	    ntohs(ip->ip_len) < hlen ||
327 	    ntohs(ip->ip_len) != m->m_pkthdr.len) {
328 		m_freem(m);
329 		return NULL;
330 	}
331 	m = m_pullup(m, hlen);
332 	if (m == NULL)
333 		return NULL;
334 
335 	ip = mtod(m, struct ip *);
336 
337 	if (ip->ip_v != IPVERSION) {
338 		m_freem(m);
339 		return NULL;
340 	}
341 
342 	/*
343 	 * Don't allow both user specified and setsockopt options.
344 	 * If options are present verify them.
345 	 */
346 	if (hlen != sizeof(struct ip)) {
347 		if (options) {
348 			m_freem(m);
349 			return NULL;
350 		} else {
351 			cp = (u_char *)(ip + 1);
352 			cnt = hlen - sizeof(struct ip);
353 			for (; cnt > 0; cnt -= optlen, cp += optlen) {
354 				opt = cp[IPOPT_OPTVAL];
355 				if (opt == IPOPT_EOL)
356 					break;
357 				if (opt == IPOPT_NOP)
358 					optlen = 1;
359 				else {
360 					if (cnt < IPOPT_OLEN + sizeof(*cp)) {
361 						m_freem(m);
362 						return NULL;
363 					}
364 					optlen = cp[IPOPT_OLEN];
365 					if (optlen < IPOPT_OLEN + sizeof(*cp) ||
366 					    optlen > cnt) {
367 						m_freem(m);
368 						return NULL;
369 					}
370 				}
371 			}
372 		}
373 	}
374 
375 	return m;
376 }
377 
378 /*
379  * Raw IP socket option processing.
380  */
381 int
382 rip_ctloutput(int op, struct socket *so, int level, int optname,
383     struct mbuf *m)
384 {
385 	struct inpcb *inp = sotoinpcb(so);
386 	int error;
387 
388 	if (level != IPPROTO_IP)
389 		return (EINVAL);
390 
391 	switch (optname) {
392 
393 	case IP_HDRINCL:
394 		error = 0;
395 		if (op == PRCO_SETOPT) {
396 			if (m == NULL || m->m_len < sizeof (int))
397 				error = EINVAL;
398 			else if (*mtod(m, int *))
399 				inp->inp_flags |= INP_HDRINCL;
400 			else
401 				inp->inp_flags &= ~INP_HDRINCL;
402 		} else {
403 			m->m_len = sizeof(int);
404 			*mtod(m, int *) = inp->inp_flags & INP_HDRINCL;
405 		}
406 		return (error);
407 
408 	case MRT_INIT:
409 	case MRT_DONE:
410 	case MRT_ADD_VIF:
411 	case MRT_DEL_VIF:
412 	case MRT_ADD_MFC:
413 	case MRT_DEL_MFC:
414 	case MRT_VERSION:
415 	case MRT_ASSERT:
416 	case MRT_API_SUPPORT:
417 	case MRT_API_CONFIG:
418 #ifdef MROUTING
419 		switch (op) {
420 		case PRCO_SETOPT:
421 			error = ip_mrouter_set(so, optname, m);
422 			break;
423 		case PRCO_GETOPT:
424 			error = ip_mrouter_get(so, optname, m);
425 			break;
426 		default:
427 			error = EINVAL;
428 			break;
429 		}
430 		return (error);
431 #else
432 		return (EOPNOTSUPP);
433 #endif
434 	}
435 	return (ip_ctloutput(op, so, level, optname, m));
436 }
437 
438 u_long	rip_sendspace = RIPSNDQ;
439 u_long	rip_recvspace = RIPRCVQ;
440 
441 /*ARGSUSED*/
442 int
443 rip_usrreq(struct socket *so, int req, struct mbuf *m, struct mbuf *nam,
444     struct mbuf *control, struct proc *p)
445 {
446 	struct inpcb *inp;
447 	int error = 0;
448 
449 	if (req == PRU_CONTROL)
450 		return (in_control(so, (u_long)m, (caddr_t)nam,
451 		    (struct ifnet *)control));
452 
453 	soassertlocked(so);
454 
455 	inp = sotoinpcb(so);
456 	if (inp == NULL) {
457 		error = EINVAL;
458 		goto release;
459 	}
460 
461 	switch (req) {
462 
463 	case PRU_DISCONNECT:
464 		if ((so->so_state & SS_ISCONNECTED) == 0) {
465 			error = ENOTCONN;
466 			break;
467 		}
468 		soisdisconnected(so);
469 		inp->inp_faddr.s_addr = INADDR_ANY;
470 		break;
471 	case PRU_ABORT:
472 		soisdisconnected(so);
473 		if (inp == NULL)
474 			panic("rip_abort");
475 #ifdef MROUTING
476 		if (so == ip_mrouter[inp->inp_rtableid])
477 			ip_mrouter_done(so);
478 #endif
479 		in_pcbdetach(inp);
480 		break;
481 
482 	case PRU_BIND:
483 	    {
484 		struct sockaddr_in *addr;
485 
486 		if ((error = in_nam2sin(nam, &addr)))
487 			break;
488 		if (!((so->so_options & SO_BINDANY) ||
489 		    addr->sin_addr.s_addr == INADDR_ANY ||
490 		    addr->sin_addr.s_addr == INADDR_BROADCAST ||
491 		    in_broadcast(addr->sin_addr, inp->inp_rtableid) ||
492 		    ifa_ifwithaddr(sintosa(addr), inp->inp_rtableid))) {
493 			error = EADDRNOTAVAIL;
494 			break;
495 		}
496 		inp->inp_laddr = addr->sin_addr;
497 		break;
498 	    }
499 	case PRU_CONNECT:
500 	    {
501 		struct sockaddr_in *addr;
502 
503 		if ((error = in_nam2sin(nam, &addr)))
504 			break;
505 		inp->inp_faddr = addr->sin_addr;
506 		soisconnected(so);
507 		break;
508 	    }
509 
510 	case PRU_CONNECT2:
511 		error = EOPNOTSUPP;
512 		break;
513 
514 	/*
515 	 * Mark the connection as being incapable of further input.
516 	 */
517 	case PRU_SHUTDOWN:
518 		socantsendmore(so);
519 		break;
520 
521 	/*
522 	 * Ship a packet out.  The appropriate raw output
523 	 * routine handles any massaging necessary.
524 	 */
525 	case PRU_SEND:
526 	    {
527 		struct sockaddr_in dst;
528 
529 		memset(&dst, 0, sizeof(dst));
530 		dst.sin_family = AF_INET;
531 		dst.sin_len = sizeof(dst);
532 		if (so->so_state & SS_ISCONNECTED) {
533 			if (nam) {
534 				error = EISCONN;
535 				break;
536 			}
537 			dst.sin_addr = inp->inp_faddr;
538 		} else {
539 			struct sockaddr_in *addr;
540 
541 			if (nam == NULL) {
542 				error = ENOTCONN;
543 				break;
544 			}
545 			if ((error = in_nam2sin(nam, &addr)))
546 				break;
547 			dst.sin_addr = addr->sin_addr;
548 		}
549 #ifdef IPSEC
550 		/* XXX Find an IPsec TDB */
551 #endif
552 		error = rip_output(m, so, sintosa(&dst), NULL);
553 		m = NULL;
554 		break;
555 	    }
556 
557 	case PRU_SENSE:
558 		/*
559 		 * stat: don't bother with a blocksize.
560 		 */
561 		break;
562 
563 	/*
564 	 * Not supported.
565 	 */
566 	case PRU_LISTEN:
567 	case PRU_ACCEPT:
568 	case PRU_SENDOOB:
569 	case PRU_RCVD:
570 	case PRU_RCVOOB:
571 		error = EOPNOTSUPP;
572 		break;
573 
574 	case PRU_SOCKADDR:
575 		in_setsockaddr(inp, nam);
576 		break;
577 
578 	case PRU_PEERADDR:
579 		in_setpeeraddr(inp, nam);
580 		break;
581 
582 	default:
583 		panic("rip_usrreq");
584 	}
585 release:
586 	if (req != PRU_RCVD && req != PRU_RCVOOB && req != PRU_SENSE) {
587 		m_freem(control);
588 		m_freem(m);
589 	}
590 	return (error);
591 }
592 
593 int
594 rip_attach(struct socket *so, int proto)
595 {
596 	struct inpcb *inp;
597 	int error;
598 
599 	if (so->so_pcb)
600 		panic("rip_attach");
601 	if ((so->so_state & SS_PRIV) == 0)
602 		return EACCES;
603 	if (proto < 0 || proto >= IPPROTO_MAX)
604 		return EPROTONOSUPPORT;
605 
606 	if ((error = soreserve(so, rip_sendspace, rip_recvspace)))
607 		return error;
608 	NET_ASSERT_LOCKED();
609 	if ((error = in_pcballoc(so, &rawcbtable)))
610 		return error;
611 	inp = sotoinpcb(so);
612 	inp->inp_ip.ip_p = proto;
613 	return 0;
614 }
615 
616 int
617 rip_detach(struct socket *so)
618 {
619 	struct inpcb *inp = sotoinpcb(so);
620 
621 	soassertlocked(so);
622 
623 	if (inp == NULL)
624 		return (EINVAL);
625 
626 #ifdef MROUTING
627 	if (so == ip_mrouter[inp->inp_rtableid])
628 		ip_mrouter_done(so);
629 #endif
630 	in_pcbdetach(inp);
631 
632 	return (0);
633 }
634