1/*
2 * Copyright (c) 2000-2012 Apple Inc. All rights reserved.
3 *
4 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
5 *
6 * This file contains Original Code and/or Modifications of Original Code
7 * as defined in and that are subject to the Apple Public Source License
8 * Version 2.0 (the 'License'). You may not use this file except in
9 * compliance with the License. The rights granted to you under the License
10 * may not be used to create, or enable the creation or redistribution of,
11 * unlawful or unlicensed copies of an Apple operating system, or to
12 * circumvent, violate, or enable the circumvention or violation of, any
13 * terms of an Apple operating system software license agreement.
14 *
15 * Please obtain a copy of the License at
16 * http://www.opensource.apple.com/apsl/ and read it before using this file.
17 *
18 * The Original Code and all software distributed under the License are
19 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
20 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
21 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
22 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
23 * Please see the License for the specific language governing rights and
24 * limitations under the License.
25 *
26 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
27 */
28/*	$FreeBSD: src/sys/netinet/ip_encap.c,v 1.1.2.2 2001/07/03 11:01:46 ume Exp $	*/
29/*	$KAME: ip_encap.c,v 1.41 2001/03/15 08:35:08 itojun Exp $	*/
30
31/*
32 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
33 * All rights reserved.
34 *
35 * Redistribution and use in source and binary forms, with or without
36 * modification, are permitted provided that the following conditions
37 * are met:
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. Neither the name of the project nor the names of its contributors
44 *    may be used to endorse or promote products derived from this software
45 *    without specific prior written permission.
46 *
47 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
48 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
49 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
50 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
51 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
52 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
53 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
54 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
55 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
56 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
57 * SUCH DAMAGE.
58 */
59/*
60 * My grandfather said that there's a devil inside tunnelling technology...
61 *
62 * We have surprisingly many protocols that want packets with IP protocol
63 * #4 or #41.  Here's a list of protocols that want protocol #41:
64 *	RFC1933 configured tunnel
65 *	RFC1933 automatic tunnel
66 *	RFC2401 IPsec tunnel
67 *	RFC2473 IPv6 generic packet tunnelling
68 *	RFC2529 6over4 tunnel
69 *	mobile-ip6 (uses RFC2473)
70 *	6to4 tunnel
71 * Here's a list of protocol that want protocol #4:
72 *	RFC1853 IPv4-in-IPv4 tunnelling
73 *	RFC2003 IPv4 encapsulation within IPv4
74 *	RFC2344 reverse tunnelling for mobile-ip4
75 *	RFC2401 IPsec tunnel
76 * Well, what can I say.  They impose different en/decapsulation mechanism
77 * from each other, so they need separate protocol handler.  The only one
78 * we can easily determine by protocol # is IPsec, which always has
79 * AH/ESP/IPComp header right after outer IP header.
80 *
81 * So, clearly good old protosw does not work for protocol #4 and #41.
82 * The code will let you match protocol via src/dst address pair.
83 */
84/* XXX is M_NETADDR correct? */
85
86#include <sys/param.h>
87#include <sys/systm.h>
88#include <sys/socket.h>
89#include <sys/sockio.h>
90#include <sys/mbuf.h>
91#include <sys/mcache.h>
92#include <sys/errno.h>
93#include <sys/domain.h>
94#include <sys/protosw.h>
95#include <sys/queue.h>
96
97#include <net/if.h>
98#include <net/route.h>
99
100#include <netinet/in.h>
101#include <netinet/in_systm.h>
102#include <netinet/ip.h>
103#include <netinet/ip_var.h>
104#include <netinet/ip_encap.h>
105
106#if INET6
107#include <netinet/ip6.h>
108#include <netinet6/ip6_var.h>
109#include <netinet6/ip6protosw.h>
110#endif
111
112
113#include <net/net_osdep.h>
114
115#ifndef __APPLE__
116#include <sys/kernel.h>
117#include <sys/malloc.h>
118MALLOC_DEFINE(M_NETADDR, "Export Host", "Export host address structure");
119#endif
120
121static void encap_init(struct protosw *, struct domain *);
122static void encap_add(struct encaptab *);
123static int mask_match(const struct encaptab *, const struct sockaddr *,
124		const struct sockaddr *);
125static void encap_fillarg(struct mbuf *, const struct encaptab *);
126
127#ifndef LIST_HEAD_INITIALIZER
128/* rely upon BSS initialization */
129LIST_HEAD(, encaptab) encaptab;
130#else
131LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(&encaptab);
132#endif
133
134static void
135encap_init(struct protosw *pp, struct domain *dp)
136{
137#pragma unused(dp)
138	static int encap_initialized = 0;
139
140	VERIFY((pp->pr_flags & (PR_INITIALIZED|PR_ATTACHED)) == PR_ATTACHED);
141
142	/* This gets called by more than one protocols, so initialize once */
143	if (encap_initialized)
144		return;
145	encap_initialized = 1;
146#if 0
147	/*
148	 * we cannot use LIST_INIT() here, since drivers may want to call
149	 * encap_attach(), on driver attach.  encap_init() will be called
150	 * on AF_INET{,6} initialization, which happens after driver
151	 * initialization - using LIST_INIT() here can nuke encap_attach()
152	 * from drivers.
153	 */
154	LIST_INIT(&encaptab);
155#endif
156}
157
158void
159encap4_init(struct protosw *pp, struct domain *dp)
160{
161	encap_init(pp, dp);
162}
163
164void
165encap6_init(struct ip6protosw *pp, struct domain *dp)
166{
167	encap_init((struct protosw *)pp, dp);
168}
169
170#if INET
171void
172encap4_input(m, off)
173	struct mbuf *m;
174	int off;
175{
176	int proto;
177	struct ip *ip;
178	struct sockaddr_in s, d;
179	const struct protosw *psw;
180	struct encaptab *ep, *match;
181	int prio, matchprio;
182
183#ifndef __APPLE__
184	va_start(ap, m);
185	off = va_arg(ap, int);
186	proto = va_arg(ap, int);
187	va_end(ap);
188#endif
189
190	/* Expect 32-bit aligned data pointer on strict-align platforms */
191	MBUF_STRICT_DATA_ALIGNMENT_CHECK_32(m);
192
193	ip = mtod(m, struct ip *);
194#ifdef __APPLE__
195	proto = ip->ip_p;
196#endif
197
198	bzero(&s, sizeof(s));
199	s.sin_family = AF_INET;
200	s.sin_len = sizeof(struct sockaddr_in);
201	s.sin_addr = ip->ip_src;
202	bzero(&d, sizeof(d));
203	d.sin_family = AF_INET;
204	d.sin_len = sizeof(struct sockaddr_in);
205	d.sin_addr = ip->ip_dst;
206
207	match = NULL;
208	matchprio = 0;
209	for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
210		if (ep->af != AF_INET)
211			continue;
212		if (ep->proto >= 0 && ep->proto != proto)
213			continue;
214		if (ep->func)
215			prio = (*ep->func)(m, off, proto, ep->arg);
216		else {
217			/*
218			 * it's inbound traffic, we need to match in reverse
219			 * order
220			 */
221			prio = mask_match(ep, (struct sockaddr *)&d,
222			    (struct sockaddr *)&s);
223		}
224
225		/*
226		 * We prioritize the matches by using bit length of the
227		 * matches.  mask_match() and user-supplied matching function
228		 * should return the bit length of the matches (for example,
229		 * if both src/dst are matched for IPv4, 64 should be returned).
230		 * 0 or negative return value means "it did not match".
231		 *
232		 * The question is, since we have two "mask" portion, we
233		 * cannot really define total order between entries.
234		 * For example, which of these should be preferred?
235		 * mask_match() returns 48 (32 + 16) for both of them.
236		 *	src=3ffe::/16, dst=3ffe:501::/32
237		 *	src=3ffe:501::/32, dst=3ffe::/16
238		 *
239		 * We need to loop through all the possible candidates
240		 * to get the best match - the search takes O(n) for
241		 * n attachments (i.e. interfaces).
242		 */
243		if (prio <= 0)
244			continue;
245		if (prio > matchprio) {
246			matchprio = prio;
247			match = ep;
248		}
249	}
250
251	if (match) {
252		/* found a match, "match" has the best one */
253		psw = (const struct protosw *)match->psw;
254		if (psw && psw->pr_input) {
255			encap_fillarg(m, match);
256			(*psw->pr_input)(m, off);
257		} else
258			m_freem(m);
259		return;
260	}
261
262	/* last resort: inject to raw socket */
263	rip_input(m, off);
264}
265#endif
266
267#if INET6
268int
269encap6_input(struct mbuf **mp, int *offp, int proto)
270{
271	struct mbuf *m = *mp;
272	struct ip6_hdr *ip6;
273	struct sockaddr_in6 s, d;
274	const struct ip6protosw *psw;
275	struct encaptab *ep, *match;
276	int prio, matchprio;
277
278	/* Expect 32-bit aligned data pointer on strict-align platforms */
279	MBUF_STRICT_DATA_ALIGNMENT_CHECK_32(m);
280
281	ip6 = mtod(m, struct ip6_hdr *);
282	bzero(&s, sizeof(s));
283	s.sin6_family = AF_INET6;
284	s.sin6_len = sizeof(struct sockaddr_in6);
285	s.sin6_addr = ip6->ip6_src;
286	bzero(&d, sizeof(d));
287	d.sin6_family = AF_INET6;
288	d.sin6_len = sizeof(struct sockaddr_in6);
289	d.sin6_addr = ip6->ip6_dst;
290
291	match = NULL;
292	matchprio = 0;
293	for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
294		if (ep->af != AF_INET6)
295			continue;
296		if (ep->proto >= 0 && ep->proto != proto)
297			continue;
298		if (ep->func)
299			prio = (*ep->func)(m, *offp, proto, ep->arg);
300		else {
301			/*
302			 * it's inbound traffic, we need to match in reverse
303			 * order
304			 */
305			prio = mask_match(ep, (struct sockaddr *)&d,
306			    (struct sockaddr *)&s);
307		}
308
309		/* see encap4_input() for issues here */
310		if (prio <= 0)
311			continue;
312		if (prio > matchprio) {
313			matchprio = prio;
314			match = ep;
315		}
316	}
317
318	if (match) {
319		/* found a match */
320		psw = (const struct ip6protosw *)match->psw;
321		if (psw && psw->pr_input) {
322			encap_fillarg(m, match);
323			return (*psw->pr_input)(mp, offp, proto);
324		} else {
325			m_freem(m);
326			return IPPROTO_DONE;
327		}
328	}
329
330	/* last resort: inject to raw socket */
331	return rip6_input(mp, offp, proto);
332}
333#endif
334
335static void
336encap_add(ep)
337	struct encaptab *ep;
338{
339
340	LIST_INSERT_HEAD(&encaptab, ep, chain);
341}
342
343/*
344 * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
345 * length of mask (sm and dm) is assumed to be same as sp/dp.
346 * Return value will be necessary as input (cookie) for encap_detach().
347 */
348const struct encaptab *
349encap_attach(af, proto, sp, sm, dp, dm, psw, arg)
350	int af;
351	int proto;
352	const struct sockaddr *sp, *sm;
353	const struct sockaddr *dp, *dm;
354	const struct protosw *psw;
355	void *arg;
356{
357	struct encaptab *ep;
358	int error;
359
360	/* sanity check on args */
361	if (sp->sa_len > sizeof(ep->src) || dp->sa_len > sizeof(ep->dst)) {
362		error = EINVAL;
363		goto fail;
364	}
365	if (sp->sa_len != dp->sa_len) {
366		error = EINVAL;
367		goto fail;
368	}
369	if (af != sp->sa_family || af != dp->sa_family) {
370		error = EINVAL;
371		goto fail;
372	}
373
374	/* check if anyone have already attached with exactly same config */
375	for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
376		if (ep->af != af)
377			continue;
378		if (ep->proto != proto)
379			continue;
380		if (ep->src.ss_len != sp->sa_len ||
381		    bcmp(&ep->src, sp, sp->sa_len) != 0 ||
382		    bcmp(&ep->srcmask, sm, sp->sa_len) != 0)
383			continue;
384		if (ep->dst.ss_len != dp->sa_len ||
385		    bcmp(&ep->dst, dp, dp->sa_len) != 0 ||
386		    bcmp(&ep->dstmask, dm, dp->sa_len) != 0)
387			continue;
388
389		error = EEXIST;
390		goto fail;
391	}
392
393	ep = _MALLOC(sizeof(*ep), M_NETADDR, M_WAITOK);	/*XXX*/
394	if (ep == NULL) {
395		error = ENOBUFS;
396		goto fail;
397	}
398	bzero(ep, sizeof(*ep));
399
400	ep->af = af;
401	ep->proto = proto;
402	bcopy(sp, &ep->src, sp->sa_len);
403	bcopy(sm, &ep->srcmask, sp->sa_len);
404	bcopy(dp, &ep->dst, dp->sa_len);
405	bcopy(dm, &ep->dstmask, dp->sa_len);
406	ep->psw = psw;
407	ep->arg = arg;
408
409	encap_add(ep);
410
411	error = 0;
412	return ep;
413
414fail:
415	return NULL;
416}
417
418const struct encaptab *
419encap_attach_func(af, proto, func, psw, arg)
420	int af;
421	int proto;
422	int (*func)(const struct mbuf *, int, int, void *);
423	const struct protosw *psw;
424	void *arg;
425{
426	struct encaptab *ep;
427	int error;
428
429	/* sanity check on args */
430	if (!func) {
431		error = EINVAL;
432		goto fail;
433	}
434
435	ep = _MALLOC(sizeof(*ep), M_NETADDR, M_WAITOK);	/*XXX*/
436	if (ep == NULL) {
437		error = ENOBUFS;
438		goto fail;
439	}
440	bzero(ep, sizeof(*ep));
441
442	ep->af = af;
443	ep->proto = proto;
444	ep->func = func;
445	ep->psw = psw;
446	ep->arg = arg;
447
448	encap_add(ep);
449
450	error = 0;
451	return ep;
452
453fail:
454	return NULL;
455}
456
457int
458encap_detach(cookie)
459	const struct encaptab *cookie;
460{
461	const struct encaptab *ep = cookie;
462	struct encaptab *p;
463
464	for (p = LIST_FIRST(&encaptab); p; p = LIST_NEXT(p, chain)) {
465		if (p == ep) {
466			LIST_REMOVE(p, chain);
467			_FREE(p, M_NETADDR);	/*XXX*/
468			return 0;
469		}
470	}
471
472	return EINVAL;
473}
474
475static int
476mask_match(ep, sp, dp)
477	const struct encaptab *ep;
478	const struct sockaddr *sp;
479	const struct sockaddr *dp;
480{
481	struct sockaddr_storage s;
482	struct sockaddr_storage d;
483	int i;
484	const u_int8_t *p, *q;
485	u_int8_t *r;
486	int matchlen;
487
488	if (sp->sa_len > sizeof(s) || dp->sa_len > sizeof(d))
489		return 0;
490	if (sp->sa_family != ep->af || dp->sa_family != ep->af)
491		return 0;
492	if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len)
493		return 0;
494
495	matchlen = 0;
496
497	p = (const u_int8_t *)sp;
498	q = (const u_int8_t *)&ep->srcmask;
499	r = (u_int8_t *)&s;
500	for (i = 0 ; i < sp->sa_len; i++) {
501		r[i] = p[i] & q[i];
502		/* XXX estimate */
503		matchlen += (q[i] ? 8 : 0);
504	}
505
506	p = (const u_int8_t *)dp;
507	q = (const u_int8_t *)&ep->dstmask;
508	r = (u_int8_t *)&d;
509	for (i = 0 ; i < dp->sa_len; i++) {
510		r[i] = p[i] & q[i];
511		/* XXX rough estimate */
512		matchlen += (q[i] ? 8 : 0);
513	}
514
515	/* need to overwrite len/family portion as we don't compare them */
516	s.ss_len = sp->sa_len;
517	s.ss_family = sp->sa_family;
518	d.ss_len = dp->sa_len;
519	d.ss_family = dp->sa_family;
520
521	if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 &&
522	    bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) {
523		return matchlen;
524	} else
525		return 0;
526}
527
528struct encaptabtag {
529	void*			*arg;
530};
531
532static void
533encap_fillarg(
534	struct mbuf *m,
535	const struct encaptab *ep)
536{
537	struct m_tag	*tag;
538	struct encaptabtag *et;
539
540	tag = m_tag_create(KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_ENCAP,
541					  sizeof(struct encaptabtag), M_WAITOK, m);
542
543	if (tag != NULL) {
544		et = (struct encaptabtag*)(tag + 1);
545		et->arg = ep->arg;
546		m_tag_prepend(m, tag);
547	}
548}
549
550void *
551encap_getarg(m)
552	struct mbuf *m;
553{
554	struct m_tag	*tag;
555	struct encaptabtag *et;
556	void *p = NULL;
557
558	tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_ENCAP, NULL);
559	if (tag) {
560		et = (struct encaptabtag*)(tag + 1);
561		p = et->arg;
562		m_tag_delete(m, tag);
563	}
564
565	return p;
566}
567