]> git.saurik.com Git - apple/xnu.git/blob - bsd/netinet6/ip6_input.c
266547020ffb7c39857b752369126a1a50f3b496
[apple/xnu.git] / bsd / netinet6 / ip6_input.c
1 /*
2 * Copyright (c) 2003-2014 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
29 /*
30 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
31 * All rights reserved.
32 *
33 * Redistribution and use in source and binary forms, with or without
34 * modification, are permitted provided that the following conditions
35 * are met:
36 * 1. Redistributions of source code must retain the above copyright
37 * notice, this list of conditions and the following disclaimer.
38 * 2. Redistributions in binary form must reproduce the above copyright
39 * notice, this list of conditions and the following disclaimer in the
40 * documentation and/or other materials provided with the distribution.
41 * 3. Neither the name of the project nor the names of its contributors
42 * may be used to endorse or promote products derived from this software
43 * without specific prior written permission.
44 *
45 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
46 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
47 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
48 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
49 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
50 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
51 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
52 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
53 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
54 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
55 * SUCH DAMAGE.
56 */
57
58 /*
59 * Copyright (c) 1982, 1986, 1988, 1993
60 * The Regents of the University of California. All rights reserved.
61 *
62 * Redistribution and use in source and binary forms, with or without
63 * modification, are permitted provided that the following conditions
64 * are met:
65 * 1. Redistributions of source code must retain the above copyright
66 * notice, this list of conditions and the following disclaimer.
67 * 2. Redistributions in binary form must reproduce the above copyright
68 * notice, this list of conditions and the following disclaimer in the
69 * documentation and/or other materials provided with the distribution.
70 * 3. All advertising materials mentioning features or use of this software
71 * must display the following acknowledgement:
72 * This product includes software developed by the University of
73 * California, Berkeley and its contributors.
74 * 4. Neither the name of the University nor the names of its contributors
75 * may be used to endorse or promote products derived from this software
76 * without specific prior written permission.
77 *
78 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
79 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
80 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
81 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
82 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
83 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
84 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
85 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
86 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
87 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
88 * SUCH DAMAGE.
89 *
90 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
91 */
92
93 #include <sys/param.h>
94 #include <sys/systm.h>
95 #include <sys/malloc.h>
96 #include <sys/mbuf.h>
97 #include <sys/domain.h>
98 #include <sys/protosw.h>
99 #include <sys/socket.h>
100 #include <sys/socketvar.h>
101 #include <sys/errno.h>
102 #include <sys/time.h>
103 #include <sys/kernel.h>
104 #include <sys/syslog.h>
105 #include <sys/sysctl.h>
106 #include <sys/proc.h>
107 #include <sys/kauth.h>
108 #include <sys/mcache.h>
109
110 #include <mach/mach_time.h>
111 #include <mach/sdt.h>
112 #include <pexpert/pexpert.h>
113 #include <dev/random/randomdev.h>
114
115 #include <net/if.h>
116 #include <net/if_var.h>
117 #include <net/if_types.h>
118 #include <net/if_dl.h>
119 #include <net/route.h>
120 #include <net/kpi_protocol.h>
121 #include <net/ntstat.h>
122 #include <net/init.h>
123 #include <net/net_osdep.h>
124
125 #include <netinet/in.h>
126 #include <netinet/in_systm.h>
127 #if INET
128 #include <netinet/ip.h>
129 #include <netinet/ip_icmp.h>
130 #endif /* INET */
131 #include <netinet/kpi_ipfilter_var.h>
132 #include <netinet/ip6.h>
133 #include <netinet6/in6_var.h>
134 #include <netinet6/ip6_var.h>
135 #include <netinet/in_pcb.h>
136 #include <netinet/icmp6.h>
137 #include <netinet6/in6_ifattach.h>
138 #include <netinet6/nd6.h>
139 #include <netinet6/scope6_var.h>
140 #include <netinet6/ip6protosw.h>
141
142 #if IPSEC
143 #include <netinet6/ipsec.h>
144 #include <netinet6/ipsec6.h>
145 extern int ipsec_bypass;
146 #endif /* IPSEC */
147
148 #if IPFW2
149 #include <netinet6/ip6_fw.h>
150 #endif /* IPFW2 */
151
152 #if DUMMYNET
153 #include <netinet/ip_fw.h>
154 #include <netinet/ip_dummynet.h>
155 #endif /* DUMMYNET */
156
157 /* we need it for NLOOP. */
158 #include "loop.h"
159
160 #if PF
161 #include <net/pfvar.h>
162 #endif /* PF */
163
164 struct ip6protosw *ip6_protox[IPPROTO_MAX];
165
166 static lck_grp_attr_t *in6_ifaddr_rwlock_grp_attr;
167 static lck_grp_t *in6_ifaddr_rwlock_grp;
168 static lck_attr_t *in6_ifaddr_rwlock_attr;
169 decl_lck_rw_data(, in6_ifaddr_rwlock);
170
171 /* Protected by in6_ifaddr_rwlock */
172 struct in6_ifaddr *in6_ifaddrs = NULL;
173
174 #define IN6_IFSTAT_REQUIRE_ALIGNED_64(f) \
175 _CASSERT(!(offsetof(struct in6_ifstat, f) % sizeof (uint64_t)))
176
177 #define ICMP6_IFSTAT_REQUIRE_ALIGNED_64(f) \
178 _CASSERT(!(offsetof(struct icmp6_ifstat, f) % sizeof (uint64_t)))
179
180 #if IPFW2
181 /* firewall hooks */
182 ip6_fw_chk_t *ip6_fw_chk_ptr;
183 ip6_fw_ctl_t *ip6_fw_ctl_ptr;
184 int ip6_fw_enable = 1;
185 #endif /* IPFW2 */
186
187 struct ip6stat ip6stat;
188
189 decl_lck_mtx_data(, proxy6_lock);
190 decl_lck_mtx_data(static, dad6_mutex_data);
191 decl_lck_mtx_data(static, nd6_mutex_data);
192 decl_lck_mtx_data(static, prefix6_mutex_data);
193 lck_mtx_t *dad6_mutex = &dad6_mutex_data;
194 lck_mtx_t *nd6_mutex = &nd6_mutex_data;
195 lck_mtx_t *prefix6_mutex = &prefix6_mutex_data;
196 #ifdef ENABLE_ADDRSEL
197 decl_lck_mtx_data(static, addrsel_mutex_data);
198 lck_mtx_t *addrsel_mutex = &addrsel_mutex_data;
199 #endif
200 static lck_attr_t *ip6_mutex_attr;
201 static lck_grp_t *ip6_mutex_grp;
202 static lck_grp_attr_t *ip6_mutex_grp_attr;
203
204 extern int loopattach_done;
205 extern void addrsel_policy_init(void);
206
207 static void ip6_init_delayed(void);
208 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *);
209
210 #if NSTF
211 extern void stfattach(void);
212 #endif /* NSTF */
213
214 SYSCTL_DECL(_net_inet6_ip6);
215
216 int ip6_doscopedroute = 1;
217 SYSCTL_INT(_net_inet6_ip6, OID_AUTO, scopedroute,
218 CTLFLAG_RD | CTLFLAG_LOCKED, &ip6_doscopedroute, 0,
219 "Enable IPv6 scoped routing");
220
221 static uint32_t ip6_adj_clear_hwcksum = 0;
222 SYSCTL_UINT(_net_inet6_ip6, OID_AUTO, adj_clear_hwcksum,
223 CTLFLAG_RW | CTLFLAG_LOCKED, &ip6_adj_clear_hwcksum, 0,
224 "Invalidate hwcksum info when adjusting length");
225
226 /*
227 * On platforms which require strict alignment (currently for anything but
228 * i386 or x86_64), check if the IP header pointer is 32-bit aligned; if not,
229 * copy the contents of the mbuf chain into a new chain, and free the original
230 * one. Create some head room in the first mbuf of the new chain, in case
231 * it's needed later on.
232 *
233 * RFC 2460 says that IPv6 headers are 64-bit aligned, but network interfaces
234 * mostly align to 32-bit boundaries. Care should be taken never to use 64-bit
235 * load/store operations on the fields in IPv6 headers.
236 */
237 #if defined(__i386__) || defined(__x86_64__)
238 #define IP6_HDR_ALIGNMENT_FIXUP(_m, _ifp, _action) do { } while (0)
239 #else /* !__i386__ && !__x86_64__ */
240 #define IP6_HDR_ALIGNMENT_FIXUP(_m, _ifp, _action) do { \
241 if (!IP6_HDR_ALIGNED_P(mtod(_m, caddr_t))) { \
242 struct mbuf *_n; \
243 struct ifnet *__ifp = (_ifp); \
244 atomic_add_64(&(__ifp)->if_alignerrs, 1); \
245 if (((_m)->m_flags & M_PKTHDR) && \
246 (_m)->m_pkthdr.pkt_hdr != NULL) \
247 (_m)->m_pkthdr.pkt_hdr = NULL; \
248 _n = m_defrag_offset(_m, max_linkhdr, M_NOWAIT); \
249 if (_n == NULL) { \
250 ip6stat.ip6s_toosmall++; \
251 m_freem(_m); \
252 (_m) = NULL; \
253 _action; \
254 } else { \
255 VERIFY(_n != (_m)); \
256 (_m) = _n; \
257 } \
258 } \
259 } while (0)
260 #endif /* !__i386__ && !__x86_64__ */
261
262 static void
263 ip6_proto_input(protocol_family_t protocol, mbuf_t packet)
264 {
265 #pragma unused(protocol)
266 ip6_input(packet);
267 }
268
269 /*
270 * IP6 initialization: fill in IP6 protocol switch table.
271 * All protocols not implemented in kernel go to raw IP6 protocol handler.
272 */
273 void
274 ip6_init(struct ip6protosw *pp, struct domain *dp)
275 {
276 static int ip6_initialized = 0;
277 struct protosw *pr;
278 struct timeval tv;
279 int i;
280 domain_unguard_t unguard;
281
282 domain_proto_mtx_lock_assert_held();
283 VERIFY((pp->pr_flags & (PR_INITIALIZED|PR_ATTACHED)) == PR_ATTACHED);
284
285 _CASSERT((sizeof (struct ip6_hdr) +
286 sizeof (struct icmp6_hdr)) <= _MHLEN);
287
288 if (ip6_initialized)
289 return;
290 ip6_initialized = 1;
291
292 PE_parse_boot_argn("net.inet6.ip6.scopedroute", &ip6_doscopedroute,
293 sizeof (ip6_doscopedroute));
294
295 pr = pffindproto_locked(PF_INET6, IPPROTO_RAW, SOCK_RAW);
296 if (pr == NULL) {
297 panic("%s: Unable to find [PF_INET6,IPPROTO_RAW,SOCK_RAW]\n",
298 __func__);
299 /* NOTREACHED */
300 }
301
302 /* Initialize the entire ip6_protox[] array to IPPROTO_RAW. */
303 for (i = 0; i < IPPROTO_MAX; i++)
304 ip6_protox[i] = (struct ip6protosw *)pr;
305 /*
306 * Cycle through IP protocols and put them into the appropriate place
307 * in ip6_protox[], skipping protocols IPPROTO_{IP,RAW}.
308 */
309 VERIFY(dp == inet6domain && dp->dom_family == PF_INET6);
310 TAILQ_FOREACH(pr, &dp->dom_protosw, pr_entry) {
311 VERIFY(pr->pr_domain == dp);
312 if (pr->pr_protocol != 0 && pr->pr_protocol != IPPROTO_RAW) {
313 /* Be careful to only index valid IP protocols. */
314 if (pr->pr_protocol < IPPROTO_MAX)
315 ip6_protox[pr->pr_protocol] =
316 (struct ip6protosw *)pr;
317 }
318 }
319
320 ip6_mutex_grp_attr = lck_grp_attr_alloc_init();
321
322 ip6_mutex_grp = lck_grp_alloc_init("ip6", ip6_mutex_grp_attr);
323 ip6_mutex_attr = lck_attr_alloc_init();
324
325 lck_mtx_init(dad6_mutex, ip6_mutex_grp, ip6_mutex_attr);
326 lck_mtx_init(nd6_mutex, ip6_mutex_grp, ip6_mutex_attr);
327 lck_mtx_init(prefix6_mutex, ip6_mutex_grp, ip6_mutex_attr);
328 scope6_init(ip6_mutex_grp, ip6_mutex_attr);
329
330 #ifdef ENABLE_ADDRSEL
331 lck_mtx_init(addrsel_mutex, ip6_mutex_grp, ip6_mutex_attr);
332 #endif
333
334 lck_mtx_init(&proxy6_lock, ip6_mutex_grp, ip6_mutex_attr);
335
336 in6_ifaddr_rwlock_grp_attr = lck_grp_attr_alloc_init();
337 in6_ifaddr_rwlock_grp = lck_grp_alloc_init("in6_ifaddr_rwlock",
338 in6_ifaddr_rwlock_grp_attr);
339 in6_ifaddr_rwlock_attr = lck_attr_alloc_init();
340 lck_rw_init(&in6_ifaddr_rwlock, in6_ifaddr_rwlock_grp,
341 in6_ifaddr_rwlock_attr);
342
343 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_receive);
344 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_hdrerr);
345 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_toobig);
346 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_noroute);
347 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_addrerr);
348 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_protounknown);
349 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_truncated);
350 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_discard);
351 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_deliver);
352 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_forward);
353 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_request);
354 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_discard);
355 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_fragok);
356 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_fragfail);
357 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_fragcreat);
358 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_reass_reqd);
359 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_reass_ok);
360 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_reass_fail);
361 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_mcast);
362 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_mcast);
363
364 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_msg);
365 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_error);
366 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_dstunreach);
367 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_adminprohib);
368 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_timeexceed);
369 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_paramprob);
370 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_pkttoobig);
371 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_echo);
372 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_echoreply);
373 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_routersolicit);
374 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_routeradvert);
375 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_neighborsolicit);
376 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_neighboradvert);
377 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_redirect);
378 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_mldquery);
379 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_mldreport);
380 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_mlddone);
381
382 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_msg);
383 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_error);
384 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_dstunreach);
385 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_adminprohib);
386 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_timeexceed);
387 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_paramprob);
388 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_pkttoobig);
389 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_echo);
390 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_echoreply);
391 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_routersolicit);
392 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_routeradvert);
393 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_neighborsolicit);
394 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_neighboradvert);
395 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_redirect);
396 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_mldquery);
397 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_mldreport);
398 ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_mlddone);
399
400 getmicrotime(&tv);
401 ip6_desync_factor =
402 (RandomULong() ^ tv.tv_usec) % MAX_TEMP_DESYNC_FACTOR;
403
404 in6_ifaddr_init();
405 ip6_moptions_init();
406 nd6_init();
407 frag6_init();
408 icmp6_init(NULL, dp);
409 addrsel_policy_init();
410
411 /*
412 * P2P interfaces often route the local address to the loopback
413 * interface. At this point, lo0 hasn't been initialized yet, which
414 * means that we need to delay the IPv6 configuration of lo0.
415 */
416 net_init_add(ip6_init_delayed);
417
418 unguard = domain_unguard_deploy();
419 i = proto_register_input(PF_INET6, ip6_proto_input, NULL, 0);
420 if (i != 0) {
421 panic("%s: failed to register PF_INET6 protocol: %d\n",
422 __func__, i);
423 /* NOTREACHED */
424 }
425 domain_unguard_release(unguard);
426 }
427
428 static void
429 ip6_init_delayed(void)
430 {
431 (void) in6_ifattach_prelim(lo_ifp);
432
433 /* timer for regeneranation of temporary addresses randomize ID */
434 timeout(in6_tmpaddrtimer, NULL,
435 (ip6_temp_preferred_lifetime - ip6_desync_factor -
436 ip6_temp_regen_advance) * hz);
437
438 #if NSTF
439 stfattach();
440 #endif /* NSTF */
441 }
442
443 void
444 ip6_input(struct mbuf *m)
445 {
446 struct ip6_hdr *ip6;
447 int off = sizeof (struct ip6_hdr), nest;
448 u_int32_t plen;
449 u_int32_t rtalert = ~0;
450 int nxt = 0, ours = 0;
451 struct ifnet *inifp, *deliverifp = NULL;
452 ipfilter_t inject_ipfref = NULL;
453 int seen;
454 struct in6_ifaddr *ia6 = NULL;
455 struct sockaddr_in6 *dst6;
456 #if DUMMYNET
457 struct m_tag *tag;
458 #endif /* DUMMYNET */
459 struct {
460 struct route_in6 rin6;
461 #if DUMMYNET
462 struct ip_fw_args args;
463 #endif /* DUMMYNET */
464 } ip6ibz;
465 #define rin6 ip6ibz.rin6
466 #define args ip6ibz.args
467
468 /* zero out {rin6, args} */
469 bzero(&ip6ibz, sizeof (ip6ibz));
470
471 /*
472 * Check if the packet we received is valid after interface filter
473 * processing
474 */
475 MBUF_INPUT_CHECK(m, m->m_pkthdr.rcvif);
476 inifp = m->m_pkthdr.rcvif;
477 VERIFY(inifp != NULL);
478
479 /* Perform IP header alignment fixup, if needed */
480 IP6_HDR_ALIGNMENT_FIXUP(m, inifp, return);
481
482 m->m_pkthdr.pkt_flags &= ~PKTF_FORWARDED;
483 #if IPSEC
484 /*
485 * should the inner packet be considered authentic?
486 * see comment in ah4_input().
487 */
488 m->m_flags &= ~M_AUTHIPHDR;
489 m->m_flags &= ~M_AUTHIPDGM;
490 #endif /* IPSEC */
491
492 /*
493 * make sure we don't have onion peering information into m_aux.
494 */
495 ip6_delaux(m);
496
497 #if DUMMYNET
498 if ((tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID,
499 KERNEL_TAG_TYPE_DUMMYNET, NULL)) != NULL) {
500 struct dn_pkt_tag *dn_tag;
501
502 dn_tag = (struct dn_pkt_tag *)(tag+1);
503
504 args.fwa_pf_rule = dn_tag->dn_pf_rule;
505
506 m_tag_delete(m, tag);
507 }
508
509 if (args.fwa_pf_rule) {
510 ip6 = mtod(m, struct ip6_hdr *); /* In case PF got disabled */
511
512 goto check_with_pf;
513 }
514 #endif /* DUMMYNET */
515
516 /*
517 * No need to proccess packet twice if we've already seen it.
518 */
519 inject_ipfref = ipf_get_inject_filter(m);
520 if (inject_ipfref != NULL) {
521 ip6 = mtod(m, struct ip6_hdr *);
522 nxt = ip6->ip6_nxt;
523 seen = 0;
524 goto injectit;
525 } else {
526 seen = 1;
527 }
528
529 /*
530 * mbuf statistics
531 */
532 if (m->m_flags & M_EXT) {
533 if (m->m_next != NULL)
534 ip6stat.ip6s_mext2m++;
535 else
536 ip6stat.ip6s_mext1++;
537 } else {
538 #define M2MMAX (sizeof (ip6stat.ip6s_m2m) / sizeof (ip6stat.ip6s_m2m[0]))
539 if (m->m_next != NULL) {
540 if (m->m_pkthdr.pkt_flags & PKTF_LOOP) {
541 /* XXX */
542 ip6stat.ip6s_m2m[ifnet_index(lo_ifp)]++;
543 } else if (inifp->if_index < M2MMAX) {
544 ip6stat.ip6s_m2m[inifp->if_index]++;
545 } else {
546 ip6stat.ip6s_m2m[0]++;
547 }
548 } else {
549 ip6stat.ip6s_m1++;
550 }
551 #undef M2MMAX
552 }
553
554 /*
555 * Drop the packet if IPv6 operation is disabled on the interface.
556 */
557 if (inifp->if_eflags & IFEF_IPV6_DISABLED)
558 goto bad;
559
560 in6_ifstat_inc_na(inifp, ifs6_in_receive);
561 ip6stat.ip6s_total++;
562
563 /*
564 * L2 bridge code and some other code can return mbuf chain
565 * that does not conform to KAME requirement. too bad.
566 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
567 */
568 if (m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
569 struct mbuf *n;
570
571 MGETHDR(n, M_DONTWAIT, MT_HEADER); /* MAC-OK */
572 if (n)
573 M_COPY_PKTHDR(n, m);
574 if (n && m->m_pkthdr.len > MHLEN) {
575 MCLGET(n, M_DONTWAIT);
576 if ((n->m_flags & M_EXT) == 0) {
577 m_freem(n);
578 n = NULL;
579 }
580 }
581 if (n == NULL)
582 goto bad;
583
584 m_copydata(m, 0, m->m_pkthdr.len, mtod(n, caddr_t));
585 n->m_len = m->m_pkthdr.len;
586 m_freem(m);
587 m = n;
588 }
589 IP6_EXTHDR_CHECK(m, 0, sizeof (struct ip6_hdr), { goto done; });
590
591 if (m->m_len < sizeof (struct ip6_hdr)) {
592 if ((m = m_pullup(m, sizeof (struct ip6_hdr))) == 0) {
593 ip6stat.ip6s_toosmall++;
594 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
595 goto done;
596 }
597 }
598
599 ip6 = mtod(m, struct ip6_hdr *);
600
601 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
602 ip6stat.ip6s_badvers++;
603 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
604 goto bad;
605 }
606
607 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
608
609 #if IPFW2
610 /*
611 * Check with the firewall...
612 */
613 if (ip6_fw_enable && ip6_fw_chk_ptr) {
614 u_short port = 0;
615 /* If ipfw says divert, we have to just drop packet */
616 /* use port as a dummy argument */
617 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) {
618 m_freem(m);
619 m = NULL;
620 }
621 if (!m)
622 goto done;
623 }
624 #endif /* IPFW2 */
625
626 /*
627 * Check against address spoofing/corruption.
628 */
629 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
630 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
631 /*
632 * XXX: "badscope" is not very suitable for a multicast source.
633 */
634 ip6stat.ip6s_badscope++;
635 in6_ifstat_inc(inifp, ifs6_in_addrerr);
636 goto bad;
637 }
638 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) &&
639 !(m->m_pkthdr.pkt_flags & PKTF_LOOP)) {
640 /*
641 * In this case, the packet should come from the loopback
642 * interface. However, we cannot just check the if_flags,
643 * because ip6_mloopback() passes the "actual" interface
644 * as the outgoing/incoming interface.
645 */
646 ip6stat.ip6s_badscope++;
647 in6_ifstat_inc(inifp, ifs6_in_addrerr);
648 goto bad;
649 }
650
651 /*
652 * The following check is not documented in specs. A malicious
653 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
654 * and bypass security checks (act as if it was from 127.0.0.1 by using
655 * IPv6 src ::ffff:127.0.0.1). Be cautious.
656 *
657 * This check chokes if we are in an SIIT cloud. As none of BSDs
658 * support IPv4-less kernel compilation, we cannot support SIIT
659 * environment at all. So, it makes more sense for us to reject any
660 * malicious packets for non-SIIT environment, than try to do a
661 * partial support for SIIT environment.
662 */
663 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
664 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
665 ip6stat.ip6s_badscope++;
666 in6_ifstat_inc(inifp, ifs6_in_addrerr);
667 goto bad;
668 }
669 #if 0
670 /*
671 * Reject packets with IPv4 compatible addresses (auto tunnel).
672 *
673 * The code forbids auto tunnel relay case in RFC1933 (the check is
674 * stronger than RFC1933). We may want to re-enable it if mech-xx
675 * is revised to forbid relaying case.
676 */
677 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
678 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
679 ip6stat.ip6s_badscope++;
680 in6_ifstat_inc(inifp, ifs6_in_addrerr);
681 goto bad;
682 }
683 #endif
684
685 /*
686 * Naively assume we can attribute inbound data to the route we would
687 * use to send to this destination. Asymetric routing breaks this
688 * assumption, but it still allows us to account for traffic from
689 * a remote node in the routing table.
690 * this has a very significant performance impact so we bypass
691 * if nstat_collect is disabled. We may also bypass if the
692 * protocol is tcp in the future because tcp will have a route that
693 * we can use to attribute the data to. That does mean we would not
694 * account for forwarded tcp traffic.
695 */
696 if (nstat_collect) {
697 struct rtentry *rte =
698 ifnet_cached_rtlookup_inet6(inifp, &ip6->ip6_src);
699 if (rte != NULL) {
700 nstat_route_rx(rte, 1, m->m_pkthdr.len, 0);
701 rtfree(rte);
702 }
703 }
704
705 /* for consistency */
706 m->m_pkthdr.pkt_proto = ip6->ip6_nxt;
707
708 #if DUMMYNET
709 check_with_pf:
710 #endif /* DUMMYNET */
711 #if PF
712 /* Invoke inbound packet filter */
713 if (PF_IS_ENABLED) {
714 int error;
715 #if DUMMYNET
716 error = pf_af_hook(inifp, NULL, &m, AF_INET6, TRUE, &args);
717 #else /* !DUMMYNET */
718 error = pf_af_hook(inifp, NULL, &m, AF_INET6, TRUE, NULL);
719 #endif /* !DUMMYNET */
720 if (error != 0 || m == NULL) {
721 if (m != NULL) {
722 panic("%s: unexpected packet %p\n",
723 __func__, m);
724 /* NOTREACHED */
725 }
726 /* Already freed by callee */
727 goto done;
728 }
729 ip6 = mtod(m, struct ip6_hdr *);
730 }
731 #endif /* PF */
732
733 /* drop packets if interface ID portion is already filled */
734 if (!(inifp->if_flags & IFF_LOOPBACK) &&
735 !(m->m_pkthdr.pkt_flags & PKTF_LOOP)) {
736 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) &&
737 ip6->ip6_src.s6_addr16[1]) {
738 ip6stat.ip6s_badscope++;
739 goto bad;
740 }
741 if (IN6_IS_SCOPE_EMBED(&ip6->ip6_dst) &&
742 ip6->ip6_dst.s6_addr16[1]) {
743 ip6stat.ip6s_badscope++;
744 goto bad;
745 }
746 }
747
748 if (m->m_pkthdr.pkt_flags & PKTF_IFAINFO) {
749 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
750 ip6->ip6_src.s6_addr16[1] =
751 htons(m->m_pkthdr.src_ifindex);
752 if (IN6_IS_SCOPE_EMBED(&ip6->ip6_dst))
753 ip6->ip6_dst.s6_addr16[1] =
754 htons(m->m_pkthdr.dst_ifindex);
755 } else {
756 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
757 ip6->ip6_src.s6_addr16[1] = htons(inifp->if_index);
758 if (IN6_IS_SCOPE_EMBED(&ip6->ip6_dst))
759 ip6->ip6_dst.s6_addr16[1] = htons(inifp->if_index);
760 }
761
762 /*
763 * Multicast check
764 */
765 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
766 struct in6_multi *in6m = NULL;
767
768 in6_ifstat_inc_na(inifp, ifs6_in_mcast);
769 /*
770 * See if we belong to the destination multicast group on the
771 * arrival interface.
772 */
773 in6_multihead_lock_shared();
774 IN6_LOOKUP_MULTI(&ip6->ip6_dst, inifp, in6m);
775 in6_multihead_lock_done();
776 if (in6m != NULL) {
777 IN6M_REMREF(in6m);
778 ours = 1;
779 } else if (!nd6_prproxy) {
780 ip6stat.ip6s_notmember++;
781 ip6stat.ip6s_cantforward++;
782 in6_ifstat_inc(inifp, ifs6_in_discard);
783 goto bad;
784 }
785 deliverifp = inifp;
786 VERIFY(ia6 == NULL);
787 goto hbhcheck;
788 }
789
790 /*
791 * Unicast check
792 *
793 * Fast path: see if the target is ourselves.
794 */
795 lck_rw_lock_shared(&in6_ifaddr_rwlock);
796 for (ia6 = in6_ifaddrs; ia6 != NULL; ia6 = ia6->ia_next) {
797 /*
798 * No reference is held on the address, as we just need
799 * to test for a few things while holding the RW lock.
800 */
801 if (IN6_ARE_ADDR_EQUAL(&ia6->ia_addr.sin6_addr, &ip6->ip6_dst))
802 break;
803 }
804
805 if (ia6 != NULL) {
806 /*
807 * For performance, test without acquiring the address lock;
808 * a lot of things in the address are set once and never
809 * changed (e.g. ia_ifp.)
810 */
811 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
812 /* this address is ready */
813 ours = 1;
814 deliverifp = ia6->ia_ifp;
815 /*
816 * record dst address information into mbuf.
817 */
818 (void) ip6_setdstifaddr_info(m, 0, ia6);
819 lck_rw_done(&in6_ifaddr_rwlock);
820 goto hbhcheck;
821 }
822 lck_rw_done(&in6_ifaddr_rwlock);
823 ia6 = NULL;
824 /* address is not ready, so discard the packet. */
825 nd6log((LOG_INFO, "%s: packet to an unready address %s->%s\n",
826 __func__, ip6_sprintf(&ip6->ip6_src),
827 ip6_sprintf(&ip6->ip6_dst)));
828 goto bad;
829 }
830 lck_rw_done(&in6_ifaddr_rwlock);
831
832 /*
833 * Slow path: route lookup.
834 */
835 dst6 = SIN6(&rin6.ro_dst);
836 dst6->sin6_len = sizeof (struct sockaddr_in6);
837 dst6->sin6_family = AF_INET6;
838 dst6->sin6_addr = ip6->ip6_dst;
839
840 rtalloc_scoped_ign((struct route *)&rin6,
841 RTF_PRCLONING, IFSCOPE_NONE);
842 if (rin6.ro_rt != NULL)
843 RT_LOCK_SPIN(rin6.ro_rt);
844
845 #define rt6_key(r) (SIN6((r)->rt_nodes->rn_key))
846
847 /*
848 * Accept the packet if the forwarding interface to the destination
849 * according to the routing table is the loopback interface,
850 * unless the associated route has a gateway.
851 * Note that this approach causes to accept a packet if there is a
852 * route to the loopback interface for the destination of the packet.
853 * But we think it's even useful in some situations, e.g. when using
854 * a special daemon which wants to intercept the packet.
855 *
856 * XXX: some OSes automatically make a cloned route for the destination
857 * of an outgoing packet. If the outgoing interface of the packet
858 * is a loopback one, the kernel would consider the packet to be
859 * accepted, even if we have no such address assinged on the interface.
860 * We check the cloned flag of the route entry to reject such cases,
861 * assuming that route entries for our own addresses are not made by
862 * cloning (it should be true because in6_addloop explicitly installs
863 * the host route). However, we might have to do an explicit check
864 * while it would be less efficient. Or, should we rather install a
865 * reject route for such a case?
866 */
867 if (rin6.ro_rt != NULL &&
868 (rin6.ro_rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
869 #if RTF_WASCLONED
870 !(rin6.ro_rt->rt_flags & RTF_WASCLONED) &&
871 #endif
872 rin6.ro_rt->rt_ifp->if_type == IFT_LOOP) {
873 ia6 = (struct in6_ifaddr *)rin6.ro_rt->rt_ifa;
874 /*
875 * Packets to a tentative, duplicated, or somehow invalid
876 * address must not be accepted.
877 *
878 * For performance, test without acquiring the address lock;
879 * a lot of things in the address are set once and never
880 * changed (e.g. ia_ifp.)
881 */
882 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
883 /* this address is ready */
884 ours = 1;
885 deliverifp = ia6->ia_ifp; /* correct? */
886 /*
887 * record dst address information into mbuf.
888 */
889 (void) ip6_setdstifaddr_info(m, 0, ia6);
890 RT_UNLOCK(rin6.ro_rt);
891 goto hbhcheck;
892 }
893 RT_UNLOCK(rin6.ro_rt);
894 ia6 = NULL;
895 /* address is not ready, so discard the packet. */
896 nd6log((LOG_INFO, "%s: packet to an unready address %s->%s\n",
897 __func__, ip6_sprintf(&ip6->ip6_src),
898 ip6_sprintf(&ip6->ip6_dst)));
899 goto bad;
900 }
901
902 if (rin6.ro_rt != NULL)
903 RT_UNLOCK(rin6.ro_rt);
904
905 /*
906 * Now there is no reason to process the packet if it's not our own
907 * and we're not a router.
908 */
909 if (!ip6_forwarding) {
910 ip6stat.ip6s_cantforward++;
911 in6_ifstat_inc(inifp, ifs6_in_discard);
912 goto bad;
913 }
914
915 hbhcheck:
916 /*
917 * record dst address information into mbuf, if we don't have one yet.
918 * note that we are unable to record it, if the address is not listed
919 * as our interface address (e.g. multicast addresses, etc.)
920 */
921 if (deliverifp != NULL && ia6 == NULL) {
922 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
923 if (ia6 != NULL) {
924 (void) ip6_setdstifaddr_info(m, 0, ia6);
925 IFA_REMREF(&ia6->ia_ifa);
926 }
927 }
928
929 /*
930 * Process Hop-by-Hop options header if it's contained.
931 * m may be modified in ip6_hopopts_input().
932 * If a JumboPayload option is included, plen will also be modified.
933 */
934 plen = (u_int32_t)ntohs(ip6->ip6_plen);
935 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
936 struct ip6_hbh *hbh;
937
938 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
939 #if 0 /* touches NULL pointer */
940 in6_ifstat_inc(inifp, ifs6_in_discard);
941 #endif
942 goto done; /* m have already been freed */
943 }
944
945 /* adjust pointer */
946 ip6 = mtod(m, struct ip6_hdr *);
947
948 /*
949 * if the payload length field is 0 and the next header field
950 * indicates Hop-by-Hop Options header, then a Jumbo Payload
951 * option MUST be included.
952 */
953 if (ip6->ip6_plen == 0 && plen == 0) {
954 /*
955 * Note that if a valid jumbo payload option is
956 * contained, ip6_hopopts_input() must set a valid
957 * (non-zero) payload length to the variable plen.
958 */
959 ip6stat.ip6s_badoptions++;
960 in6_ifstat_inc(inifp, ifs6_in_discard);
961 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
962 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER,
963 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
964 goto done;
965 }
966 /* ip6_hopopts_input() ensures that mbuf is contiguous */
967 hbh = (struct ip6_hbh *)(ip6 + 1);
968 nxt = hbh->ip6h_nxt;
969
970 /*
971 * If we are acting as a router and the packet contains a
972 * router alert option, see if we know the option value.
973 * Currently, we only support the option value for MLD, in which
974 * case we should pass the packet to the multicast routing
975 * daemon.
976 */
977 if (rtalert != ~0 && ip6_forwarding) {
978 switch (rtalert) {
979 case IP6OPT_RTALERT_MLD:
980 ours = 1;
981 break;
982 default:
983 /*
984 * RFC2711 requires unrecognized values must be
985 * silently ignored.
986 */
987 break;
988 }
989 }
990 } else
991 nxt = ip6->ip6_nxt;
992
993 /*
994 * Check that the amount of data in the buffers
995 * is as at least much as the IPv6 header would have us expect.
996 * Trim mbufs if longer than we expect.
997 * Drop packet if shorter than we expect.
998 */
999 if (m->m_pkthdr.len - sizeof (struct ip6_hdr) < plen) {
1000 ip6stat.ip6s_tooshort++;
1001 in6_ifstat_inc(inifp, ifs6_in_truncated);
1002 goto bad;
1003 }
1004 if (m->m_pkthdr.len > sizeof (struct ip6_hdr) + plen) {
1005 /*
1006 * Invalidate hardware checksum info if ip6_adj_clear_hwcksum
1007 * is set; useful to handle buggy drivers. Note that this
1008 * should not be enabled by default, as we may get here due
1009 * to link-layer padding.
1010 */
1011 if (ip6_adj_clear_hwcksum &&
1012 (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) &&
1013 !(inifp->if_flags & IFF_LOOPBACK) &&
1014 !(m->m_pkthdr.pkt_flags & PKTF_LOOP)) {
1015 m->m_pkthdr.csum_flags &= ~CSUM_DATA_VALID;
1016 m->m_pkthdr.csum_data = 0;
1017 ip6stat.ip6s_adj_hwcsum_clr++;
1018 }
1019
1020 ip6stat.ip6s_adj++;
1021 if (m->m_len == m->m_pkthdr.len) {
1022 m->m_len = sizeof (struct ip6_hdr) + plen;
1023 m->m_pkthdr.len = sizeof (struct ip6_hdr) + plen;
1024 } else {
1025 m_adj(m, sizeof (struct ip6_hdr) + plen -
1026 m->m_pkthdr.len);
1027 }
1028 }
1029
1030 /*
1031 * Forward if desirable.
1032 */
1033 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
1034 if (!ours && nd6_prproxy) {
1035 /*
1036 * If this isn't for us, this might be a Neighbor
1037 * Solicitation (dst is solicited-node multicast)
1038 * against an address in one of the proxied prefixes;
1039 * if so, claim the packet and let icmp6_input()
1040 * handle the rest.
1041 */
1042 ours = nd6_prproxy_isours(m, ip6, NULL, IFSCOPE_NONE);
1043 VERIFY(!ours ||
1044 (m->m_pkthdr.pkt_flags & PKTF_PROXY_DST));
1045 }
1046 if (!ours)
1047 goto bad;
1048 } else if (!ours) {
1049 /*
1050 * The unicast forwarding function might return the packet
1051 * if we are proxying prefix(es), and if the packet is an
1052 * ICMPv6 packet that has failed the zone checks, but is
1053 * targetted towards a proxied address (this is optimized by
1054 * way of RTF_PROXY test.) If so, claim the packet as ours
1055 * and let icmp6_input() handle the rest. The packet's hop
1056 * limit value is kept intact (it's not decremented). This
1057 * is for supporting Neighbor Unreachability Detection between
1058 * proxied nodes on different links (src is link-local, dst
1059 * is target address.)
1060 */
1061 if ((m = ip6_forward(m, &rin6, 0)) == NULL)
1062 goto done;
1063 VERIFY(rin6.ro_rt != NULL);
1064 VERIFY(m->m_pkthdr.pkt_flags & PKTF_PROXY_DST);
1065 deliverifp = rin6.ro_rt->rt_ifp;
1066 ours = 1;
1067 }
1068
1069 ip6 = mtod(m, struct ip6_hdr *);
1070
1071 /*
1072 * Malicious party may be able to use IPv4 mapped addr to confuse
1073 * tcp/udp stack and bypass security checks (act as if it was from
1074 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
1075 *
1076 * For SIIT end node behavior, you may want to disable the check.
1077 * However, you will become vulnerable to attacks using IPv4 mapped
1078 * source.
1079 */
1080 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
1081 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
1082 ip6stat.ip6s_badscope++;
1083 in6_ifstat_inc(inifp, ifs6_in_addrerr);
1084 goto bad;
1085 }
1086
1087 /*
1088 * Tell launch routine the next header
1089 */
1090 ip6stat.ip6s_delivered++;
1091 in6_ifstat_inc_na(deliverifp, ifs6_in_deliver);
1092
1093 injectit:
1094 nest = 0;
1095
1096 /*
1097 * Perform IP header alignment fixup again, if needed. Note that
1098 * we do it once for the outermost protocol, and we assume each
1099 * protocol handler wouldn't mess with the alignment afterwards.
1100 */
1101 IP6_HDR_ALIGNMENT_FIXUP(m, inifp, return);
1102
1103 while (nxt != IPPROTO_DONE) {
1104 struct ipfilter *filter;
1105 int (*pr_input)(struct mbuf **, int *, int);
1106
1107 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
1108 ip6stat.ip6s_toomanyhdr++;
1109 goto bad;
1110 }
1111
1112 /*
1113 * protection against faulty packet - there should be
1114 * more sanity checks in header chain processing.
1115 */
1116 if (m->m_pkthdr.len < off) {
1117 ip6stat.ip6s_tooshort++;
1118 in6_ifstat_inc(inifp, ifs6_in_truncated);
1119 goto bad;
1120 }
1121
1122
1123 #if IPSEC
1124 /*
1125 * enforce IPsec policy checking if we are seeing last header.
1126 * note that we do not visit this with protocols with pcb layer
1127 * code - like udp/tcp/raw ip.
1128 */
1129 if ((ipsec_bypass == 0) &&
1130 (ip6_protox[nxt]->pr_flags & PR_LASTHDR) != 0) {
1131 if (ipsec6_in_reject(m, NULL)) {
1132 IPSEC_STAT_INCREMENT(ipsec6stat.in_polvio);
1133 goto bad;
1134 }
1135 }
1136 #endif /* IPSEC */
1137
1138 /*
1139 * Call IP filter
1140 */
1141 if (!TAILQ_EMPTY(&ipv6_filters)) {
1142 ipf_ref();
1143 TAILQ_FOREACH(filter, &ipv6_filters, ipf_link) {
1144 if (seen == 0) {
1145 if ((struct ipfilter *)inject_ipfref ==
1146 filter)
1147 seen = 1;
1148 } else if (filter->ipf_filter.ipf_input) {
1149 errno_t result;
1150
1151 result = filter->ipf_filter.ipf_input(
1152 filter->ipf_filter.cookie,
1153 (mbuf_t *)&m, off, nxt);
1154 if (result == EJUSTRETURN) {
1155 ipf_unref();
1156 goto done;
1157 }
1158 if (result != 0) {
1159 ipf_unref();
1160 goto bad;
1161 }
1162 }
1163 }
1164 ipf_unref();
1165 }
1166
1167 DTRACE_IP6(receive, struct mbuf *, m, struct inpcb *, NULL,
1168 struct ip6_hdr *, ip6, struct ifnet *, inifp,
1169 struct ip *, NULL, struct ip6_hdr *, ip6);
1170
1171 if ((pr_input = ip6_protox[nxt]->pr_input) == NULL) {
1172 m_freem(m);
1173 m = NULL;
1174 nxt = IPPROTO_DONE;
1175 } else if (!(ip6_protox[nxt]->pr_flags & PR_PROTOLOCK)) {
1176 lck_mtx_lock(inet6_domain_mutex);
1177 nxt = pr_input(&m, &off, nxt);
1178 lck_mtx_unlock(inet6_domain_mutex);
1179 } else {
1180 nxt = pr_input(&m, &off, nxt);
1181 }
1182 }
1183 done:
1184 ROUTE_RELEASE(&rin6);
1185 return;
1186 bad:
1187 m_freem(m);
1188 goto done;
1189 }
1190
1191 void
1192 ip6_setsrcifaddr_info(struct mbuf *m, uint32_t src_idx, struct in6_ifaddr *ia6)
1193 {
1194 VERIFY(m->m_flags & M_PKTHDR);
1195
1196 /*
1197 * If the source ifaddr is specified, pick up the information
1198 * from there; otherwise just grab the passed-in ifindex as the
1199 * caller may not have the ifaddr available.
1200 */
1201 if (ia6 != NULL) {
1202 m->m_pkthdr.pkt_flags |= PKTF_IFAINFO;
1203 m->m_pkthdr.src_ifindex = ia6->ia_ifp->if_index;
1204
1205 /* See IN6_IFF comments in in6_var.h */
1206 m->m_pkthdr.src_iff = (ia6->ia6_flags & 0xffff);
1207 } else {
1208 m->m_pkthdr.src_iff = 0;
1209 m->m_pkthdr.src_ifindex = src_idx;
1210 if (src_idx != 0)
1211 m->m_pkthdr.pkt_flags |= PKTF_IFAINFO;
1212 }
1213 }
1214
1215 void
1216 ip6_setdstifaddr_info(struct mbuf *m, uint32_t dst_idx, struct in6_ifaddr *ia6)
1217 {
1218 VERIFY(m->m_flags & M_PKTHDR);
1219
1220 /*
1221 * If the destination ifaddr is specified, pick up the information
1222 * from there; otherwise just grab the passed-in ifindex as the
1223 * caller may not have the ifaddr available.
1224 */
1225 if (ia6 != NULL) {
1226 m->m_pkthdr.pkt_flags |= PKTF_IFAINFO;
1227 m->m_pkthdr.dst_ifindex = ia6->ia_ifp->if_index;
1228
1229 /* See IN6_IFF comments in in6_var.h */
1230 m->m_pkthdr.dst_iff = (ia6->ia6_flags & 0xffff);
1231 } else {
1232 m->m_pkthdr.dst_iff = 0;
1233 m->m_pkthdr.dst_ifindex = dst_idx;
1234 if (dst_idx != 0)
1235 m->m_pkthdr.pkt_flags |= PKTF_IFAINFO;
1236 }
1237 }
1238
1239 int
1240 ip6_getsrcifaddr_info(struct mbuf *m, uint32_t *src_idx, uint32_t *ia6f)
1241 {
1242 VERIFY(m->m_flags & M_PKTHDR);
1243
1244 if (!(m->m_pkthdr.pkt_flags & PKTF_IFAINFO))
1245 return (-1);
1246
1247 if (src_idx != NULL)
1248 *src_idx = m->m_pkthdr.src_ifindex;
1249
1250 if (ia6f != NULL)
1251 *ia6f = m->m_pkthdr.src_iff;
1252
1253 return (0);
1254 }
1255
1256 int
1257 ip6_getdstifaddr_info(struct mbuf *m, uint32_t *dst_idx, uint32_t *ia6f)
1258 {
1259 VERIFY(m->m_flags & M_PKTHDR);
1260
1261 if (!(m->m_pkthdr.pkt_flags & PKTF_IFAINFO))
1262 return (-1);
1263
1264 if (dst_idx != NULL)
1265 *dst_idx = m->m_pkthdr.dst_ifindex;
1266
1267 if (ia6f != NULL)
1268 *ia6f = m->m_pkthdr.dst_iff;
1269
1270 return (0);
1271 }
1272
1273 /*
1274 * Hop-by-Hop options header processing. If a valid jumbo payload option is
1275 * included, the real payload length will be stored in plenp.
1276 */
1277 static int
1278 ip6_hopopts_input(uint32_t *plenp, uint32_t *rtalertp, struct mbuf **mp,
1279 int *offp)
1280 {
1281 struct mbuf *m = *mp;
1282 int off = *offp, hbhlen;
1283 struct ip6_hbh *hbh;
1284 u_int8_t *opt;
1285
1286 /* validation of the length of the header */
1287 IP6_EXTHDR_CHECK(m, off, sizeof (*hbh), return (-1));
1288 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
1289 hbhlen = (hbh->ip6h_len + 1) << 3;
1290
1291 IP6_EXTHDR_CHECK(m, off, hbhlen, return (-1));
1292 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
1293 off += hbhlen;
1294 hbhlen -= sizeof (struct ip6_hbh);
1295 opt = (u_int8_t *)hbh + sizeof (struct ip6_hbh);
1296
1297 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof (struct ip6_hbh),
1298 hbhlen, rtalertp, plenp) < 0)
1299 return (-1);
1300
1301 *offp = off;
1302 *mp = m;
1303 return (0);
1304 }
1305
1306 /*
1307 * Search header for all Hop-by-hop options and process each option.
1308 * This function is separate from ip6_hopopts_input() in order to
1309 * handle a case where the sending node itself process its hop-by-hop
1310 * options header. In such a case, the function is called from ip6_output().
1311 *
1312 * The function assumes that hbh header is located right after the IPv6 header
1313 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
1314 * opthead + hbhlen is located in continuous memory region.
1315 */
1316 int
1317 ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
1318 struct mbuf *m;
1319 u_int8_t *opthead;
1320 int hbhlen;
1321 u_int32_t *rtalertp;
1322 u_int32_t *plenp;
1323 {
1324 struct ip6_hdr *ip6;
1325 int optlen = 0;
1326 u_int8_t *opt = opthead;
1327 u_int16_t rtalert_val;
1328 u_int32_t jumboplen;
1329 const int erroff = sizeof (struct ip6_hdr) + sizeof (struct ip6_hbh);
1330
1331 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
1332 switch (*opt) {
1333 case IP6OPT_PAD1:
1334 optlen = 1;
1335 break;
1336 case IP6OPT_PADN:
1337 if (hbhlen < IP6OPT_MINLEN) {
1338 ip6stat.ip6s_toosmall++;
1339 goto bad;
1340 }
1341 optlen = *(opt + 1) + 2;
1342 break;
1343 case IP6OPT_ROUTER_ALERT:
1344 /* XXX may need check for alignment */
1345 if (hbhlen < IP6OPT_RTALERT_LEN) {
1346 ip6stat.ip6s_toosmall++;
1347 goto bad;
1348 }
1349 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
1350 /* XXX stat */
1351 icmp6_error(m, ICMP6_PARAM_PROB,
1352 ICMP6_PARAMPROB_HEADER,
1353 erroff + opt + 1 - opthead);
1354 return (-1);
1355 }
1356 optlen = IP6OPT_RTALERT_LEN;
1357 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
1358 *rtalertp = ntohs(rtalert_val);
1359 break;
1360 case IP6OPT_JUMBO:
1361 /* XXX may need check for alignment */
1362 if (hbhlen < IP6OPT_JUMBO_LEN) {
1363 ip6stat.ip6s_toosmall++;
1364 goto bad;
1365 }
1366 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
1367 /* XXX stat */
1368 icmp6_error(m, ICMP6_PARAM_PROB,
1369 ICMP6_PARAMPROB_HEADER,
1370 erroff + opt + 1 - opthead);
1371 return (-1);
1372 }
1373 optlen = IP6OPT_JUMBO_LEN;
1374
1375 /*
1376 * IPv6 packets that have non 0 payload length
1377 * must not contain a jumbo payload option.
1378 */
1379 ip6 = mtod(m, struct ip6_hdr *);
1380 if (ip6->ip6_plen) {
1381 ip6stat.ip6s_badoptions++;
1382 icmp6_error(m, ICMP6_PARAM_PROB,
1383 ICMP6_PARAMPROB_HEADER,
1384 erroff + opt - opthead);
1385 return (-1);
1386 }
1387
1388 /*
1389 * We may see jumbolen in unaligned location, so
1390 * we'd need to perform bcopy().
1391 */
1392 bcopy(opt + 2, &jumboplen, sizeof (jumboplen));
1393 jumboplen = (u_int32_t)htonl(jumboplen);
1394
1395 #if 1
1396 /*
1397 * if there are multiple jumbo payload options,
1398 * *plenp will be non-zero and the packet will be
1399 * rejected.
1400 * the behavior may need some debate in ipngwg -
1401 * multiple options does not make sense, however,
1402 * there's no explicit mention in specification.
1403 */
1404 if (*plenp != 0) {
1405 ip6stat.ip6s_badoptions++;
1406 icmp6_error(m, ICMP6_PARAM_PROB,
1407 ICMP6_PARAMPROB_HEADER,
1408 erroff + opt + 2 - opthead);
1409 return (-1);
1410 }
1411 #endif
1412
1413 /*
1414 * jumbo payload length must be larger than 65535.
1415 */
1416 if (jumboplen <= IPV6_MAXPACKET) {
1417 ip6stat.ip6s_badoptions++;
1418 icmp6_error(m, ICMP6_PARAM_PROB,
1419 ICMP6_PARAMPROB_HEADER,
1420 erroff + opt + 2 - opthead);
1421 return (-1);
1422 }
1423 *plenp = jumboplen;
1424
1425 break;
1426 default: /* unknown option */
1427 if (hbhlen < IP6OPT_MINLEN) {
1428 ip6stat.ip6s_toosmall++;
1429 goto bad;
1430 }
1431 optlen = ip6_unknown_opt(opt, m,
1432 erroff + opt - opthead);
1433 if (optlen == -1) {
1434 return (-1);
1435 }
1436 optlen += 2;
1437 break;
1438 }
1439 }
1440
1441 return (0);
1442
1443 bad:
1444 m_freem(m);
1445 return (-1);
1446 }
1447
1448 /*
1449 * Unknown option processing.
1450 * The third argument `off' is the offset from the IPv6 header to the option,
1451 * which is necessary if the IPv6 header the and option header and IPv6 header
1452 * is not continuous in order to return an ICMPv6 error.
1453 */
1454 int
1455 ip6_unknown_opt(uint8_t *optp, struct mbuf *m, int off)
1456 {
1457 struct ip6_hdr *ip6;
1458
1459 switch (IP6OPT_TYPE(*optp)) {
1460 case IP6OPT_TYPE_SKIP: /* ignore the option */
1461 return ((int)*(optp + 1));
1462
1463 case IP6OPT_TYPE_DISCARD: /* silently discard */
1464 m_freem(m);
1465 return (-1);
1466
1467 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1468 ip6stat.ip6s_badoptions++;
1469 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1470 return (-1);
1471
1472 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1473 ip6stat.ip6s_badoptions++;
1474 ip6 = mtod(m, struct ip6_hdr *);
1475 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1476 (m->m_flags & (M_BCAST|M_MCAST))) {
1477 m_freem(m);
1478 } else {
1479 icmp6_error(m, ICMP6_PARAM_PROB,
1480 ICMP6_PARAMPROB_OPTION, off);
1481 }
1482 return (-1);
1483 }
1484
1485 m_freem(m); /* XXX: NOTREACHED */
1486 return (-1);
1487 }
1488
1489 /*
1490 * Create the "control" list for this pcb.
1491 * These functions will not modify mbuf chain at all.
1492 *
1493 * With KAME mbuf chain restriction:
1494 * The routine will be called from upper layer handlers like tcp6_input().
1495 * Thus the routine assumes that the caller (tcp6_input) have already
1496 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1497 * very first mbuf on the mbuf chain.
1498 *
1499 * ip6_savecontrol_v4 will handle those options that are possible to be
1500 * set on a v4-mapped socket.
1501 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those
1502 * options and handle the v6-only ones itself.
1503 */
1504 struct mbuf **
1505 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp,
1506 int *v4only)
1507 {
1508 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1509
1510 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) {
1511 struct timeval tv;
1512
1513 getmicrotime(&tv);
1514 mp = sbcreatecontrol_mbuf((caddr_t)&tv, sizeof (tv),
1515 SCM_TIMESTAMP, SOL_SOCKET, mp);
1516 if (*mp == NULL)
1517 return (NULL);
1518 }
1519 if ((inp->inp_socket->so_options & SO_TIMESTAMP_MONOTONIC) != 0) {
1520 uint64_t time;
1521
1522 time = mach_absolute_time();
1523 mp = sbcreatecontrol_mbuf((caddr_t)&time, sizeof (time),
1524 SCM_TIMESTAMP_MONOTONIC, SOL_SOCKET, mp);
1525 if (*mp == NULL)
1526 return (NULL);
1527 }
1528 if ((inp->inp_socket->so_flags & SOF_RECV_TRAFFIC_CLASS) != 0) {
1529 int tc = m_get_traffic_class(m);
1530
1531 mp = sbcreatecontrol_mbuf((caddr_t)&tc, sizeof (tc),
1532 SO_TRAFFIC_CLASS, SOL_SOCKET, mp);
1533 if (*mp == NULL)
1534 return (NULL);
1535 }
1536
1537 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1538 if (v4only != NULL)
1539 *v4only = 1;
1540 return (mp);
1541 }
1542
1543 #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y))
1544 /* RFC 2292 sec. 5 */
1545 if ((inp->inp_flags & IN6P_PKTINFO) != 0) {
1546 struct in6_pktinfo pi6;
1547
1548 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof (struct in6_addr));
1549 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1550 pi6.ipi6_ifindex =
1551 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1552
1553 mp = sbcreatecontrol_mbuf((caddr_t)&pi6,
1554 sizeof (struct in6_pktinfo),
1555 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO),
1556 IPPROTO_IPV6, mp);
1557 if (*mp == NULL)
1558 return (NULL);
1559 }
1560
1561 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) {
1562 int hlim = ip6->ip6_hlim & 0xff;
1563
1564 mp = sbcreatecontrol_mbuf((caddr_t)&hlim, sizeof (int),
1565 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT),
1566 IPPROTO_IPV6, mp);
1567 if (*mp == NULL)
1568 return (NULL);
1569 }
1570
1571 if (v4only != NULL)
1572 *v4only = 0;
1573 return (mp);
1574 }
1575
1576 int
1577 ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp)
1578 {
1579 struct mbuf **np;
1580 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1581 int v4only = 0;
1582
1583 *mp = NULL;
1584 np = ip6_savecontrol_v4(in6p, m, mp, &v4only);
1585 if (np == NULL)
1586 goto no_mbufs;
1587
1588 mp = np;
1589 if (v4only)
1590 return (0);
1591
1592 if ((in6p->inp_flags & IN6P_TCLASS) != 0) {
1593 u_int32_t flowinfo;
1594 int tclass;
1595
1596 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1597 flowinfo >>= 20;
1598
1599 tclass = flowinfo & 0xff;
1600 mp = sbcreatecontrol_mbuf((caddr_t)&tclass, sizeof (tclass),
1601 IPV6_TCLASS, IPPROTO_IPV6, mp);
1602 if (*mp == NULL)
1603 goto no_mbufs;
1604 }
1605
1606 /*
1607 * IPV6_HOPOPTS socket option. Recall that we required super-user
1608 * privilege for the option (see ip6_ctloutput), but it might be too
1609 * strict, since there might be some hop-by-hop options which can be
1610 * returned to normal user.
1611 * See also RFC 2292 section 6 (or RFC 3542 section 8).
1612 */
1613 if ((in6p->inp_flags & IN6P_HOPOPTS) != 0) {
1614 /*
1615 * Check if a hop-by-hop options header is contatined in the
1616 * received packet, and if so, store the options as ancillary
1617 * data. Note that a hop-by-hop options header must be
1618 * just after the IPv6 header, which is assured through the
1619 * IPv6 input processing.
1620 */
1621 ip6 = mtod(m, struct ip6_hdr *);
1622 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1623 struct ip6_hbh *hbh;
1624 int hbhlen = 0;
1625 hbh = (struct ip6_hbh *)(ip6 + 1);
1626 hbhlen = (hbh->ip6h_len + 1) << 3;
1627
1628 /*
1629 * XXX: We copy the whole header even if a
1630 * jumbo payload option is included, the option which
1631 * is to be removed before returning according to
1632 * RFC2292.
1633 * Note: this constraint is removed in RFC3542
1634 */
1635 mp = sbcreatecontrol_mbuf((caddr_t)hbh, hbhlen,
1636 IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1637 IPPROTO_IPV6, mp);
1638
1639 if (*mp == NULL) {
1640 goto no_mbufs;
1641 }
1642 }
1643 }
1644
1645 if ((in6p->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
1646 int nxt = ip6->ip6_nxt, off = sizeof (struct ip6_hdr);
1647
1648 /*
1649 * Search for destination options headers or routing
1650 * header(s) through the header chain, and stores each
1651 * header as ancillary data.
1652 * Note that the order of the headers remains in
1653 * the chain of ancillary data.
1654 */
1655 while (1) { /* is explicit loop prevention necessary? */
1656 struct ip6_ext *ip6e = NULL;
1657 int elen;
1658
1659 /*
1660 * if it is not an extension header, don't try to
1661 * pull it from the chain.
1662 */
1663 switch (nxt) {
1664 case IPPROTO_DSTOPTS:
1665 case IPPROTO_ROUTING:
1666 case IPPROTO_HOPOPTS:
1667 case IPPROTO_AH: /* is it possible? */
1668 break;
1669 default:
1670 goto loopend;
1671 }
1672
1673 if (off + sizeof (*ip6e) > m->m_len)
1674 goto loopend;
1675 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1676 if (nxt == IPPROTO_AH)
1677 elen = (ip6e->ip6e_len + 2) << 2;
1678 else
1679 elen = (ip6e->ip6e_len + 1) << 3;
1680 if (off + elen > m->m_len)
1681 goto loopend;
1682
1683 switch (nxt) {
1684 case IPPROTO_DSTOPTS:
1685 if (!(in6p->inp_flags & IN6P_DSTOPTS))
1686 break;
1687
1688 mp = sbcreatecontrol_mbuf((caddr_t)ip6e, elen,
1689 IS2292(in6p, IPV6_2292DSTOPTS,
1690 IPV6_DSTOPTS), IPPROTO_IPV6, mp);
1691 if (*mp == NULL) {
1692 goto no_mbufs;
1693 }
1694 break;
1695 case IPPROTO_ROUTING:
1696 if (!in6p->inp_flags & IN6P_RTHDR)
1697 break;
1698
1699 mp = sbcreatecontrol_mbuf((caddr_t)ip6e, elen,
1700 IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR),
1701 IPPROTO_IPV6, mp);
1702 if (*mp == NULL) {
1703 goto no_mbufs;
1704 }
1705 break;
1706 case IPPROTO_HOPOPTS:
1707 case IPPROTO_AH: /* is it possible? */
1708 break;
1709
1710 default:
1711 /*
1712 * other cases have been filtered in the above.
1713 * none will visit this case. here we supply
1714 * the code just in case (nxt overwritten or
1715 * other cases).
1716 */
1717 goto loopend;
1718
1719 }
1720
1721 /* proceed with the next header. */
1722 off += elen;
1723 nxt = ip6e->ip6e_nxt;
1724 ip6e = NULL;
1725 }
1726 loopend:
1727 ;
1728 }
1729 return (0);
1730 no_mbufs:
1731 ip6stat.ip6s_pktdropcntrl++;
1732 /* XXX increment a stat to show the failure */
1733 return (ENOBUFS);
1734 }
1735 #undef IS2292
1736
1737 void
1738 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu)
1739 {
1740 struct socket *so;
1741 struct mbuf *m_mtu;
1742 struct ip6_mtuinfo mtuctl;
1743
1744 so = in6p->inp_socket;
1745
1746 if (mtu == NULL)
1747 return;
1748
1749 #ifdef DIAGNOSTIC
1750 if (so == NULL) { /* I believe this is impossible */
1751 panic("ip6_notify_pmtu: socket is NULL");
1752 /* NOTREACHED */
1753 }
1754 #endif
1755
1756 bzero(&mtuctl, sizeof (mtuctl)); /* zero-clear for safety */
1757 mtuctl.ip6m_mtu = *mtu;
1758 mtuctl.ip6m_addr = *dst;
1759 if (sa6_recoverscope(&mtuctl.ip6m_addr, TRUE))
1760 return;
1761
1762 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof (mtuctl),
1763 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1764 return;
1765
1766 if (sbappendaddr(&so->so_rcv, SA(dst), NULL, m_mtu, NULL) == 0) {
1767 m_freem(m_mtu);
1768 /* XXX: should count statistics */
1769 } else {
1770 sorwakeup(so);
1771 }
1772 }
1773
1774 /*
1775 * Get pointer to the previous header followed by the header
1776 * currently processed.
1777 * XXX: This function supposes that
1778 * M includes all headers,
1779 * the next header field and the header length field of each header
1780 * are valid, and
1781 * the sum of each header length equals to OFF.
1782 * Because of these assumptions, this function must be called very
1783 * carefully. Moreover, it will not be used in the near future when
1784 * we develop `neater' mechanism to process extension headers.
1785 */
1786 char *
1787 ip6_get_prevhdr(m, off)
1788 struct mbuf *m;
1789 int off;
1790 {
1791 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1792
1793 if (off == sizeof (struct ip6_hdr)) {
1794 return ((char *)&ip6->ip6_nxt);
1795 } else {
1796 int len, nxt;
1797 struct ip6_ext *ip6e = NULL;
1798
1799 nxt = ip6->ip6_nxt;
1800 len = sizeof (struct ip6_hdr);
1801 while (len < off) {
1802 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1803
1804 switch (nxt) {
1805 case IPPROTO_FRAGMENT:
1806 len += sizeof (struct ip6_frag);
1807 break;
1808 case IPPROTO_AH:
1809 len += (ip6e->ip6e_len + 2) << 2;
1810 break;
1811 default:
1812 len += (ip6e->ip6e_len + 1) << 3;
1813 break;
1814 }
1815 nxt = ip6e->ip6e_nxt;
1816 }
1817 if (ip6e)
1818 return ((char *)&ip6e->ip6e_nxt);
1819 else
1820 return (NULL);
1821 }
1822 }
1823
1824 /*
1825 * get next header offset. m will be retained.
1826 */
1827 int
1828 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1829 {
1830 struct ip6_hdr ip6;
1831 struct ip6_ext ip6e;
1832 struct ip6_frag fh;
1833
1834 /* just in case */
1835 VERIFY(m != NULL);
1836 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1837 return (-1);
1838
1839 switch (proto) {
1840 case IPPROTO_IPV6:
1841 if (m->m_pkthdr.len < off + sizeof (ip6))
1842 return (-1);
1843 m_copydata(m, off, sizeof (ip6), (caddr_t)&ip6);
1844 if (nxtp)
1845 *nxtp = ip6.ip6_nxt;
1846 off += sizeof (ip6);
1847 return (off);
1848
1849 case IPPROTO_FRAGMENT:
1850 /*
1851 * terminate parsing if it is not the first fragment,
1852 * it does not make sense to parse through it.
1853 */
1854 if (m->m_pkthdr.len < off + sizeof (fh))
1855 return (-1);
1856 m_copydata(m, off, sizeof (fh), (caddr_t)&fh);
1857 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1858 if (fh.ip6f_offlg & IP6F_OFF_MASK)
1859 return (-1);
1860 if (nxtp)
1861 *nxtp = fh.ip6f_nxt;
1862 off += sizeof (struct ip6_frag);
1863 return (off);
1864
1865 case IPPROTO_AH:
1866 if (m->m_pkthdr.len < off + sizeof (ip6e))
1867 return (-1);
1868 m_copydata(m, off, sizeof (ip6e), (caddr_t)&ip6e);
1869 if (nxtp)
1870 *nxtp = ip6e.ip6e_nxt;
1871 off += (ip6e.ip6e_len + 2) << 2;
1872 return (off);
1873
1874 case IPPROTO_HOPOPTS:
1875 case IPPROTO_ROUTING:
1876 case IPPROTO_DSTOPTS:
1877 if (m->m_pkthdr.len < off + sizeof (ip6e))
1878 return (-1);
1879 m_copydata(m, off, sizeof (ip6e), (caddr_t)&ip6e);
1880 if (nxtp)
1881 *nxtp = ip6e.ip6e_nxt;
1882 off += (ip6e.ip6e_len + 1) << 3;
1883 return (off);
1884
1885 case IPPROTO_NONE:
1886 case IPPROTO_ESP:
1887 case IPPROTO_IPCOMP:
1888 /* give up */
1889 return (-1);
1890
1891 default:
1892 return (-1);
1893 }
1894
1895 return (-1);
1896 }
1897
1898 /*
1899 * get offset for the last header in the chain. m will be kept untainted.
1900 */
1901 int
1902 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1903 {
1904 int newoff;
1905 int nxt;
1906
1907 if (!nxtp) {
1908 nxt = -1;
1909 nxtp = &nxt;
1910 }
1911 while (1) {
1912 newoff = ip6_nexthdr(m, off, proto, nxtp);
1913 if (newoff < 0)
1914 return (off);
1915 else if (newoff < off)
1916 return (-1); /* invalid */
1917 else if (newoff == off)
1918 return (newoff);
1919
1920 off = newoff;
1921 proto = *nxtp;
1922 }
1923 }
1924
1925 struct ip6aux *
1926 ip6_addaux(struct mbuf *m)
1927 {
1928 struct m_tag *tag;
1929
1930 /* Check if one is already allocated */
1931 tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID,
1932 KERNEL_TAG_TYPE_INET6, NULL);
1933 if (tag == NULL) {
1934 /* Allocate a tag */
1935 tag = m_tag_create(KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_INET6,
1936 sizeof (struct ip6aux), M_DONTWAIT, m);
1937
1938 /* Attach it to the mbuf */
1939 if (tag) {
1940 m_tag_prepend(m, tag);
1941 }
1942 }
1943
1944 return (tag ? (struct ip6aux *)(tag + 1) : NULL);
1945 }
1946
1947 struct ip6aux *
1948 ip6_findaux(struct mbuf *m)
1949 {
1950 struct m_tag *tag;
1951
1952 tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID,
1953 KERNEL_TAG_TYPE_INET6, NULL);
1954
1955 return (tag ? (struct ip6aux *)(tag + 1) : NULL);
1956 }
1957
1958 void
1959 ip6_delaux(struct mbuf *m)
1960 {
1961 struct m_tag *tag;
1962
1963 tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID,
1964 KERNEL_TAG_TYPE_INET6, NULL);
1965 if (tag) {
1966 m_tag_delete(m, tag);
1967 }
1968 }
1969
1970 /*
1971 * Drain callback
1972 */
1973 void
1974 ip6_drain(void)
1975 {
1976 frag6_drain(); /* fragments */
1977 in6_rtqdrain(); /* protocol cloned routes */
1978 nd6_drain(NULL); /* cloned routes: ND6 */
1979 }
1980
1981 /*
1982 * System control for IP6
1983 */
1984
1985 u_char inet6ctlerrmap[PRC_NCMDS] = {
1986 0, 0, 0, 0,
1987 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1988 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1989 EMSGSIZE, EHOSTUNREACH, 0, 0,
1990 0, 0, 0, 0,
1991 ENOPROTOOPT
1992 };