1/*
2 * Copyright (c) 2000-2021 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 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#include <netinet/ip6.h>
107#include <netinet6/ip6_var.h>
108#include <netinet6/ip6protosw.h>
109
110#include <net/net_osdep.h>
111#include <net/sockaddr_utils.h>
112
113#ifndef __APPLE__
114#include <sys/kernel.h>
115#include <sys/malloc.h>
116MALLOC_DEFINE(M_NETADDR, "Export Host", "Export host address structure");
117#endif
118
119static void encap_add_locked(struct encaptab *);
120static int mask_match(const struct encaptab *, const struct sockaddr *,
121 const struct sockaddr *);
122static void encap_fillarg(struct mbuf *, void *arg);
123
124LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(&encaptab);
125
126static LCK_GRP_DECLARE(encaptab_lock_grp, "encaptab lock");
127static LCK_RW_DECLARE(encaptab_lock, &encaptab_lock_grp);
128
129#if INET
130void
131encap4_input(struct mbuf *m, int off)
132{
133 int proto;
134 struct ip *ip;
135 struct sockaddr_in s, d;
136 const struct protosw *psw;
137 struct encaptab *ep, *match;
138 int prio, matchprio;
139 void *match_arg = NULL;
140
141#ifndef __APPLE__
142 va_start(ap, m);
143 off = va_arg(ap, int);
144 proto = va_arg(ap, int);
145 va_end(ap);
146#endif
147
148 /* Expect 32-bit aligned data pointer on strict-align platforms */
149 MBUF_STRICT_DATA_ALIGNMENT_CHECK_32(m);
150
151 ip = mtod(m, struct ip *);
152#ifdef __APPLE__
153 proto = ip->ip_p;
154#endif
155
156 SOCKADDR_ZERO(&s, sizeof(s));
157 s.sin_family = AF_INET;
158 s.sin_len = sizeof(struct sockaddr_in);
159 s.sin_addr = ip->ip_src;
160 SOCKADDR_ZERO(&d, sizeof(d));
161 d.sin_family = AF_INET;
162 d.sin_len = sizeof(struct sockaddr_in);
163 d.sin_addr = ip->ip_dst;
164
165 match = NULL;
166 matchprio = 0;
167
168 lck_rw_lock_shared(lck: &encaptab_lock);
169 for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
170 if (ep->af != AF_INET) {
171 continue;
172 }
173 if (ep->proto >= 0 && ep->proto != proto) {
174 continue;
175 }
176 if (ep->func) {
177 prio = (*ep->func)(m, off, proto, ep->arg);
178 } else {
179 /*
180 * it's inbound traffic, we need to match in reverse
181 * order
182 */
183 prio = mask_match(ep, SA(&d), SA(&s));
184 }
185
186 /*
187 * We prioritize the matches by using bit length of the
188 * matches. mask_match() and user-supplied matching function
189 * should return the bit length of the matches (for example,
190 * if both src/dst are matched for IPv4, 64 should be returned).
191 * 0 or negative return value means "it did not match".
192 *
193 * The question is, since we have two "mask" portion, we
194 * cannot really define total order between entries.
195 * For example, which of these should be preferred?
196 * mask_match() returns 48 (32 + 16) for both of them.
197 * src=3ffe::/16, dst=3ffe:501::/32
198 * src=3ffe:501::/32, dst=3ffe::/16
199 *
200 * We need to loop through all the possible candidates
201 * to get the best match - the search takes O(n) for
202 * n attachments (i.e. interfaces).
203 */
204 if (prio <= 0) {
205 continue;
206 }
207 if (prio > matchprio) {
208 matchprio = prio;
209 match = ep;
210 psw = (const struct protosw *)match->psw;
211 match_arg = ep->arg;
212 }
213 }
214 lck_rw_unlock_shared(lck: &encaptab_lock);
215
216 if (match) {
217 /* found a match, "match" has the best one */
218 if (psw && psw->pr_input) {
219 encap_fillarg(m, arg: match_arg);
220 (*psw->pr_input)(m, off);
221 } else {
222 m_freem(m);
223 }
224 return;
225 }
226
227 /* last resort: inject to raw socket */
228 rip_input(m, off);
229}
230#endif
231
232int
233encap6_input(struct mbuf **mp, int *offp, int proto)
234{
235 struct mbuf *m = *mp;
236 struct ip6_hdr *ip6;
237 struct sockaddr_in6 s, d;
238 const struct ip6protosw *psw;
239 struct encaptab *ep, *match;
240 int prio, matchprio;
241 void *match_arg = NULL;
242
243 /* Expect 32-bit aligned data pointer on strict-align platforms */
244 MBUF_STRICT_DATA_ALIGNMENT_CHECK_32(m);
245
246 ip6 = mtod(m, struct ip6_hdr *);
247 SOCKADDR_ZERO(&s, sizeof(s));
248 s.sin6_family = AF_INET6;
249 s.sin6_len = sizeof(struct sockaddr_in6);
250 s.sin6_addr = ip6->ip6_src;
251 SOCKADDR_ZERO(&d, sizeof(d));
252 d.sin6_family = AF_INET6;
253 d.sin6_len = sizeof(struct sockaddr_in6);
254 d.sin6_addr = ip6->ip6_dst;
255
256 match = NULL;
257 matchprio = 0;
258
259 lck_rw_lock_shared(lck: &encaptab_lock);
260 for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
261 if (ep->af != AF_INET6) {
262 continue;
263 }
264 if (ep->proto >= 0 && ep->proto != proto) {
265 continue;
266 }
267 if (ep->func) {
268 prio = (*ep->func)(m, *offp, proto, ep->arg);
269 } else {
270 /*
271 * it's inbound traffic, we need to match in reverse
272 * order
273 */
274 prio = mask_match(ep, SA(&d), SA(&s));
275 }
276
277 /* see encap4_input() for issues here */
278 if (prio <= 0) {
279 continue;
280 }
281 if (prio > matchprio) {
282 matchprio = prio;
283 match = ep;
284 psw = (const struct ip6protosw *)match->psw;
285 match_arg = ep->arg;
286 }
287 }
288 lck_rw_unlock_shared(lck: &encaptab_lock);
289
290 if (match) {
291 /* found a match */
292 if (psw && psw->pr_input) {
293 encap_fillarg(m, arg: match_arg);
294 return (*psw->pr_input)(mp, offp, proto);
295 } else {
296 m_freem(m);
297 return IPPROTO_DONE;
298 }
299 }
300
301 /* last resort: inject to raw socket */
302 return rip6_input(mp, offp, proto);
303}
304
305static void
306encap_add_locked(struct encaptab *ep)
307{
308 LCK_RW_ASSERT(&encaptab_lock, LCK_RW_ASSERT_EXCLUSIVE);
309 LIST_INSERT_HEAD(&encaptab, ep, chain);
310}
311
312/*
313 * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
314 * length of mask (sm and dm) is assumed to be same as sp/dp.
315 * Return value will be necessary as input (cookie) for encap_detach().
316 */
317const struct encaptab *
318encap_attach(int af, int proto, const struct sockaddr *sp,
319 const struct sockaddr *sm, const struct sockaddr *dp,
320 const struct sockaddr *dm, const struct protosw *psw, void *arg)
321{
322 struct encaptab *ep = NULL;
323 struct encaptab *new_ep = NULL;
324 int error;
325
326 /* sanity check on args */
327 if (sp->sa_len > sizeof(new_ep->src) || dp->sa_len > sizeof(new_ep->dst)) {
328 error = EINVAL;
329 goto fail;
330 }
331 if (sp->sa_len != dp->sa_len) {
332 error = EINVAL;
333 goto fail;
334 }
335 if (af != sp->sa_family || af != dp->sa_family) {
336 error = EINVAL;
337 goto fail;
338 }
339
340 new_ep = kalloc_type(struct encaptab, Z_WAITOK | Z_ZERO | Z_NOFAIL);
341
342 /* check if anyone have already attached with exactly same config */
343 lck_rw_lock_exclusive(lck: &encaptab_lock);
344 for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
345 if (ep->af != af) {
346 continue;
347 }
348 if (ep->proto != proto) {
349 continue;
350 }
351 if (ep->src.ss_len != sp->sa_len ||
352 SOCKADDR_CMP(&ep->src, sp, sp->sa_len) != 0 ||
353 SOCKADDR_CMP(&ep->srcmask, sm, sp->sa_len) != 0) {
354 continue;
355 }
356 if (ep->dst.ss_len != dp->sa_len ||
357 SOCKADDR_CMP(&ep->dst, dp, dp->sa_len) != 0 ||
358 SOCKADDR_CMP(&ep->dstmask, dm, dp->sa_len) != 0) {
359 continue;
360 }
361
362 error = EEXIST;
363 goto fail_locked;
364 }
365
366 new_ep->af = af;
367 new_ep->proto = proto;
368 SOCKADDR_COPY(sp, &new_ep->src, sp->sa_len);
369 SOCKADDR_COPY(sm, &new_ep->srcmask, sp->sa_len);
370 SOCKADDR_COPY(dp, &new_ep->dst, dp->sa_len);
371 SOCKADDR_COPY(dm, &new_ep->dstmask, dp->sa_len);
372 new_ep->psw = psw;
373 new_ep->arg = arg;
374
375 encap_add_locked(ep: new_ep);
376 lck_rw_unlock_exclusive(lck: &encaptab_lock);
377
378 error = 0;
379 return new_ep;
380
381fail_locked:
382 lck_rw_unlock_exclusive(lck: &encaptab_lock);
383 if (new_ep != NULL) {
384 kfree_type(struct encaptab, new_ep);
385 }
386fail:
387 return NULL;
388}
389
390const struct encaptab *
391encap_attach_func( int af, int proto,
392 int (*func)(const struct mbuf *, int, int, void *),
393 const struct protosw *psw, void *arg)
394{
395 struct encaptab *ep;
396 int error;
397
398 /* sanity check on args */
399 if (!func) {
400 error = EINVAL;
401 goto fail;
402 }
403
404 ep = kalloc_type(struct encaptab, Z_WAITOK | Z_ZERO | Z_NOFAIL); /* XXX */
405
406 ep->af = af;
407 ep->proto = proto;
408 ep->func = func;
409 ep->psw = psw;
410 ep->arg = arg;
411
412 lck_rw_lock_exclusive(lck: &encaptab_lock);
413 encap_add_locked(ep);
414 lck_rw_unlock_exclusive(lck: &encaptab_lock);
415
416 error = 0;
417 return ep;
418
419fail:
420 return NULL;
421}
422
423int
424encap_detach(const struct encaptab *cookie)
425{
426 const struct encaptab *ep = cookie;
427 struct encaptab *p;
428
429 lck_rw_lock_exclusive(lck: &encaptab_lock);
430 for (p = LIST_FIRST(&encaptab); p; p = LIST_NEXT(p, chain)) {
431 if (p == ep) {
432 LIST_REMOVE(p, chain);
433 lck_rw_unlock_exclusive(lck: &encaptab_lock);
434 kfree_type(struct encaptab, p); /*XXX*/
435 return 0;
436 }
437 }
438 lck_rw_unlock_exclusive(lck: &encaptab_lock);
439
440 return EINVAL;
441}
442
443static int
444mask_match(const struct encaptab *ep, const struct sockaddr *sp,
445 const struct sockaddr *dp)
446{
447 struct sockaddr_storage s;
448 struct sockaddr_storage d;
449 int i;
450 const u_int8_t *p, *q;
451 u_int8_t *r;
452 int matchlen;
453
454 if (sp->sa_len > sizeof(s) || dp->sa_len > sizeof(d)) {
455 return 0;
456 }
457 if (sp->sa_family != ep->af || dp->sa_family != ep->af) {
458 return 0;
459 }
460 if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len) {
461 return 0;
462 }
463
464 matchlen = 0;
465
466 p = (const u_int8_t *)sp;
467 q = (const u_int8_t *)&ep->srcmask;
468 r = (u_int8_t *)&s;
469 for (i = 0; i < sp->sa_len; i++) {
470 r[i] = p[i] & q[i];
471 /* XXX estimate */
472 matchlen += (q[i] ? 8 : 0);
473 }
474
475 p = (const u_int8_t *)dp;
476 q = (const u_int8_t *)&ep->dstmask;
477 r = (u_int8_t *)&d;
478 for (i = 0; i < dp->sa_len; i++) {
479 r[i] = p[i] & q[i];
480 /* XXX rough estimate */
481 matchlen += (q[i] ? 8 : 0);
482 }
483
484 /* need to overwrite len/family portion as we don't compare them */
485 s.ss_len = sp->sa_len;
486 s.ss_family = sp->sa_family;
487 d.ss_len = dp->sa_len;
488 d.ss_family = dp->sa_family;
489
490 if (bcmp(s1: &s, s2: &ep->src, n: ep->src.ss_len) == 0 &&
491 bcmp(s1: &d, s2: &ep->dst, n: ep->dst.ss_len) == 0) {
492 return matchlen;
493 } else {
494 return 0;
495 }
496}
497
498struct encaptabtag {
499 void* *arg;
500};
501
502static void
503encap_fillarg(
504 struct mbuf *m,
505 void *arg)
506{
507 struct m_tag *tag;
508 struct encaptabtag *et;
509
510 tag = m_tag_create(KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_ENCAP,
511 sizeof(struct encaptabtag), M_WAITOK, m);
512
513 if (tag != NULL) {
514 et = (struct encaptabtag*)(tag->m_tag_data);
515 et->arg = arg;
516 m_tag_prepend(m, tag);
517 }
518}
519
520void *
521encap_getarg(struct mbuf *m)
522{
523 struct m_tag *tag;
524 struct encaptabtag *et;
525 void *p = NULL;
526
527 tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_ENCAP);
528 if (tag) {
529 et = (struct encaptabtag*)(tag->m_tag_data);
530 p = et->arg;
531 m_tag_delete(m, tag);
532 }
533
534 return p;
535}
536
537struct encaptab_tag_container {
538 struct m_tag encaptab_m_tag;
539 struct encaptabtag encaptab_tag;
540};
541
542static struct m_tag *
543m_tag_kalloc_encap(u_int32_t id, u_int16_t type, uint16_t len, int wait)
544{
545 struct encaptab_tag_container *tag_container;
546 struct m_tag *tag = NULL;
547
548 assert3u(id, ==, KERNEL_MODULE_TAG_ID);
549 assert3u(type, ==, KERNEL_TAG_TYPE_ENCAP);
550 assert3u(len, ==, sizeof(struct encaptabtag));
551
552 if (len != sizeof(struct encaptabtag)) {
553 return NULL;
554 }
555
556 tag_container = kalloc_type(struct encaptab_tag_container, wait | M_ZERO);
557 if (tag_container != NULL) {
558 tag = &tag_container->encaptab_m_tag;
559
560 assert3p(tag, ==, tag_container);
561
562 M_TAG_INIT(tag, id, type, len, &tag_container->encaptab_tag, NULL);
563 }
564
565 return tag;
566}
567
568static void
569m_tag_kfree_encap(struct m_tag *tag)
570{
571 struct encaptab_tag_container *tag_container = (struct encaptab_tag_container *)tag;
572
573 assert3u(tag->m_tag_len, ==, sizeof(struct encaptabtag));
574
575 kfree_type(struct encaptab_tag_container, tag_container);
576}
577
578void
579encap_register_m_tag(void)
580{
581 int error;
582
583 error = m_register_internal_tag_type(type: KERNEL_TAG_TYPE_ENCAP, len: sizeof(struct encaptabtag),
584 alloc_func: m_tag_kalloc_encap, free_func: m_tag_kfree_encap);
585
586 assert3u(error, ==, 0);
587}
588