]> git.saurik.com Git - apple/xnu.git/blame - bsd/netinet6/ip6_input.c
xnu-2782.40.9.tar.gz
[apple/xnu.git] / bsd / netinet6 / ip6_input.c
CommitLineData
2d21ac55 1/*
04b8595b 2 * Copyright (c) 2003-2015 Apple Inc. All rights reserved.
2d21ac55
A
3 *
4 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
39236c6e 5 *
2d21ac55
A
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.
39236c6e 14 *
2d21ac55
A
15 * Please obtain a copy of the License at
16 * http://www.opensource.apple.com/apsl/ and read it before using this file.
39236c6e 17 *
2d21ac55
A
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.
39236c6e 25 *
2d21ac55
A
26 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
27 */
1c79356b
A
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 */
9bccf70c 92
1c79356b
A
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>
6d2010ae 105#include <sys/sysctl.h>
1c79356b 106#include <sys/proc.h>
91447636 107#include <sys/kauth.h>
6d2010ae 108#include <sys/mcache.h>
6d2010ae 109
39236c6e
A
110#include <mach/mach_time.h>
111#include <mach/sdt.h>
6d2010ae 112#include <pexpert/pexpert.h>
39236c6e 113#include <dev/random/randomdev.h>
1c79356b
A
114
115#include <net/if.h>
9bccf70c 116#include <net/if_var.h>
1c79356b
A
117#include <net/if_types.h>
118#include <net/if_dl.h>
119#include <net/route.h>
2d21ac55 120#include <net/kpi_protocol.h>
6d2010ae 121#include <net/ntstat.h>
316670eb 122#include <net/init.h>
39236c6e 123#include <net/net_osdep.h>
1c79356b
A
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>
39236c6e
A
130#endif /* INET */
131#include <netinet/kpi_ipfilter_var.h>
1c79356b
A
132#include <netinet/ip6.h>
133#include <netinet6/in6_var.h>
134#include <netinet6/ip6_var.h>
1c79356b 135#include <netinet/in_pcb.h>
1c79356b
A
136#include <netinet/icmp6.h>
137#include <netinet6/in6_ifattach.h>
138#include <netinet6/nd6.h>
6d2010ae 139#include <netinet6/scope6_var.h>
39236c6e 140#include <netinet6/ip6protosw.h>
1c79356b 141
9bccf70c
A
142#if IPSEC
143#include <netinet6/ipsec.h>
9bccf70c 144#include <netinet6/ipsec6.h>
9bccf70c 145extern int ipsec_bypass;
39236c6e 146#endif /* IPSEC */
1c79356b 147
39236c6e 148#if IPFW2
1c79356b 149#include <netinet6/ip6_fw.h>
39236c6e 150#endif /* IPFW2 */
1c79356b 151
316670eb
A
152#if DUMMYNET
153#include <netinet/ip_fw.h>
154#include <netinet/ip_dummynet.h>
155#endif /* DUMMYNET */
156
1c79356b 157/* we need it for NLOOP. */
1c79356b 158#include "loop.h"
1c79356b 159
b0d623f7
A
160#if PF
161#include <net/pfvar.h>
162#endif /* PF */
163
39236c6e 164struct ip6protosw *ip6_protox[IPPROTO_MAX];
6d2010ae
A
165
166static lck_grp_attr_t *in6_ifaddr_rwlock_grp_attr;
167static lck_grp_t *in6_ifaddr_rwlock_grp;
168static lck_attr_t *in6_ifaddr_rwlock_attr;
169decl_lck_rw_data(, in6_ifaddr_rwlock);
170
171/* Protected by in6_ifaddr_rwlock */
172struct in6_ifaddr *in6_ifaddrs = NULL;
9bccf70c 173
39236c6e 174#define IN6_IFSTAT_REQUIRE_ALIGNED_64(f) \
6d2010ae 175 _CASSERT(!(offsetof(struct in6_ifstat, f) % sizeof (uint64_t)))
316670eb 176
39236c6e 177#define ICMP6_IFSTAT_REQUIRE_ALIGNED_64(f) \
6d2010ae 178 _CASSERT(!(offsetof(struct icmp6_ifstat, f) % sizeof (uint64_t)))
9bccf70c 179
7e4a7d39 180#if IPFW2
1c79356b
A
181/* firewall hooks */
182ip6_fw_chk_t *ip6_fw_chk_ptr;
183ip6_fw_ctl_t *ip6_fw_ctl_ptr;
9bccf70c 184int ip6_fw_enable = 1;
39236c6e 185#endif /* IPFW2 */
1c79356b
A
186
187struct ip6stat ip6stat;
188
316670eb 189decl_lck_mtx_data(, proxy6_lock);
39236c6e
A
190decl_lck_mtx_data(static, dad6_mutex_data);
191decl_lck_mtx_data(static, nd6_mutex_data);
192decl_lck_mtx_data(static, prefix6_mutex_data);
316670eb
A
193lck_mtx_t *dad6_mutex = &dad6_mutex_data;
194lck_mtx_t *nd6_mutex = &nd6_mutex_data;
195lck_mtx_t *prefix6_mutex = &prefix6_mutex_data;
6d2010ae 196#ifdef ENABLE_ADDRSEL
39236c6e 197decl_lck_mtx_data(static, addrsel_mutex_data);
316670eb 198lck_mtx_t *addrsel_mutex = &addrsel_mutex_data;
6d2010ae 199#endif
39236c6e
A
200static lck_attr_t *ip6_mutex_attr;
201static lck_grp_t *ip6_mutex_grp;
202static lck_grp_attr_t *ip6_mutex_grp_attr;
203
91447636 204extern int loopattach_done;
6d2010ae 205extern void addrsel_policy_init(void);
9bccf70c 206
316670eb 207static void ip6_init_delayed(void);
91447636 208static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *);
1c79356b 209
39236c6e
A
210#if NSTF
211extern void stfattach(void);
212#endif /* NSTF */
1c79356b 213
6d2010ae
A
214SYSCTL_DECL(_net_inet6_ip6);
215
39236c6e
A
216int ip6_doscopedroute = 1;
217SYSCTL_INT(_net_inet6_ip6, OID_AUTO, scopedroute,
218 CTLFLAG_RD | CTLFLAG_LOCKED, &ip6_doscopedroute, 0,
219 "Enable IPv6 scoped routing");
2d21ac55 220
39236c6e
A
221static uint32_t ip6_adj_clear_hwcksum = 0;
222SYSCTL_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");
316670eb
A
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) && \
39236c6e
A
246 (_m)->m_pkthdr.pkt_hdr != NULL) \
247 (_m)->m_pkthdr.pkt_hdr = NULL; \
316670eb
A
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; \
39236c6e 253 _action; \
316670eb
A
254 } else { \
255 VERIFY(_n != (_m)); \
256 (_m) = _n; \
257 } \
258 } \
259} while (0)
260#endif /* !__i386__ && !__x86_64__ */
261
91447636 262static void
39236c6e 263ip6_proto_input(protocol_family_t protocol, mbuf_t packet)
91447636 264{
39236c6e 265#pragma unused(protocol)
91447636
A
266 ip6_input(packet);
267}
268
1c79356b
A
269/*
270 * IP6 initialization: fill in IP6 protocol switch table.
271 * All protocols not implemented in kernel go to raw IP6 protocol handler.
272 */
273void
39236c6e 274ip6_init(struct ip6protosw *pp, struct domain *dp)
1c79356b 275{
39236c6e
A
276 static int ip6_initialized = 0;
277 struct protosw *pr;
1c79356b 278 struct timeval tv;
39236c6e
A
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);
1c79356b 287
39236c6e
A
288 if (ip6_initialized)
289 return;
290 ip6_initialized = 1;
99c3a104 291
6d2010ae
A
292 PE_parse_boot_argn("net.inet6.ip6.scopedroute", &ip6_doscopedroute,
293 sizeof (ip6_doscopedroute));
294
39236c6e
A
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. */
1c79356b 303 for (i = 0; i < IPPROTO_MAX; i++)
39236c6e
A
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;
1c79356b
A
317 }
318 }
319
91447636 320 ip6_mutex_grp_attr = lck_grp_attr_alloc_init();
91447636
A
321
322 ip6_mutex_grp = lck_grp_alloc_init("ip6", ip6_mutex_grp_attr);
323 ip6_mutex_attr = lck_attr_alloc_init();
91447636 324
316670eb
A
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);
39236c6e 328 scope6_init(ip6_mutex_grp, ip6_mutex_attr);
91447636 329
6d2010ae 330#ifdef ENABLE_ADDRSEL
316670eb 331 lck_mtx_init(addrsel_mutex, ip6_mutex_grp, ip6_mutex_attr);
6d2010ae
A
332#endif
333
316670eb 334 lck_mtx_init(&proxy6_lock, ip6_mutex_grp, ip6_mutex_attr);
6d2010ae
A
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);
39236c6e 362 IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_mcast);
6d2010ae
A
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
39236c6e
A
400 getmicrotime(&tv);
401 ip6_desync_factor =
402 (RandomULong() ^ tv.tv_usec) % MAX_TEMP_DESYNC_FACTOR;
403
b0d623f7 404 in6_ifaddr_init();
6d2010ae 405 ip6_moptions_init();
1c79356b
A
406 nd6_init();
407 frag6_init();
39236c6e 408 icmp6_init(NULL, dp);
6d2010ae 409 addrsel_policy_init();
91447636 410
316670eb
A
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
39236c6e
A
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);
1c79356b
A
426}
427
428static void
316670eb 429ip6_init_delayed(void)
1c79356b 430{
39236c6e 431 (void) in6_ifattach_prelim(lo_ifp);
1c79356b 432
9bccf70c 433 /* timer for regeneranation of temporary addresses randomize ID */
39236c6e
A
434 timeout(in6_tmpaddrtimer, NULL,
435 (ip6_temp_preferred_lifetime - ip6_desync_factor -
436 ip6_temp_regen_advance) * hz);
9bccf70c 437
9bccf70c
A
438#if NSTF
439 stfattach();
39236c6e 440#endif /* NSTF */
6d2010ae 441}
1c79356b
A
442
443void
6d2010ae 444ip6_input(struct mbuf *m)
1c79356b
A
445{
446 struct ip6_hdr *ip6;
39236c6e 447 int off = sizeof (struct ip6_hdr), nest;
1c79356b
A
448 u_int32_t plen;
449 u_int32_t rtalert = ~0;
450 int nxt = 0, ours = 0;
39236c6e
A
451 struct ifnet *inifp, *deliverifp = NULL;
452 ipfilter_t inject_ipfref = NULL;
91447636 453 int seen;
6d2010ae 454 struct in6_ifaddr *ia6 = NULL;
6d2010ae 455 struct sockaddr_in6 *dst6;
316670eb 456#if DUMMYNET
39236c6e 457 struct m_tag *tag;
316670eb 458#endif /* DUMMYNET */
39236c6e
A
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
6d2010ae 467
39236c6e
A
468 /* zero out {rin6, args} */
469 bzero(&ip6ibz, sizeof (ip6ibz));
6d2010ae 470
39236c6e
A
471 /*
472 * Check if the packet we received is valid after interface filter
6d2010ae
A
473 * processing
474 */
475 MBUF_INPUT_CHECK(m, m->m_pkthdr.rcvif);
39236c6e
A
476 inifp = m->m_pkthdr.rcvif;
477 VERIFY(inifp != NULL);
1c79356b 478
316670eb 479 /* Perform IP header alignment fixup, if needed */
39236c6e
A
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);
316670eb
A
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);
39236c6e 503
316670eb 504 args.fwa_pf_rule = dn_tag->dn_pf_rule;
39236c6e 505
316670eb
A
506 m_tag_delete(m, tag);
507 }
39236c6e 508
316670eb
A
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
91447636 516 /*
39236c6e 517 * No need to proccess packet twice if we've already seen it.
91447636
A
518 */
519 inject_ipfref = ipf_get_inject_filter(m);
39236c6e 520 if (inject_ipfref != NULL) {
91447636
A
521 ip6 = mtod(m, struct ip6_hdr *);
522 nxt = ip6->ip6_nxt;
523 seen = 0;
524 goto injectit;
39236c6e 525 } else {
91447636 526 seen = 1;
1c79356b 527 }
9bccf70c 528
1c79356b 529 /*
55e303ae 530 * mbuf statistics
1c79356b
A
531 */
532 if (m->m_flags & M_EXT) {
39236c6e 533 if (m->m_next != NULL)
1c79356b
A
534 ip6stat.ip6s_mext2m++;
535 else
536 ip6stat.ip6s_mext1++;
537 } else {
39236c6e
A
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 {
1c79356b 546 ip6stat.ip6s_m2m[0]++;
39236c6e
A
547 }
548 } else {
1c79356b 549 ip6stat.ip6s_m1++;
39236c6e 550 }
9bccf70c 551#undef M2MMAX
1c79356b
A
552 }
553
316670eb 554 /*
39236c6e 555 * Drop the packet if IPv6 operation is disabled on the interface.
316670eb 556 */
39236c6e 557 if (inifp->if_eflags & IFEF_IPV6_DISABLED)
6d2010ae 558 goto bad;
6d2010ae 559
39236c6e 560 in6_ifstat_inc_na(inifp, ifs6_in_receive);
1c79356b
A
561 ip6stat.ip6s_total++;
562
9bccf70c
A
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 */
39236c6e 568 if (m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
9bccf70c
A
569 struct mbuf *n;
570
2d21ac55 571 MGETHDR(n, M_DONTWAIT, MT_HEADER); /* MAC-OK */
9bccf70c
A
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 }
6d2010ae
A
581 if (n == NULL)
582 goto bad;
9bccf70c
A
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 }
39236c6e 589 IP6_EXTHDR_CHECK(m, 0, sizeof (struct ip6_hdr), { goto done; });
1c79356b 590
39236c6e
A
591 if (m->m_len < sizeof (struct ip6_hdr)) {
592 if ((m = m_pullup(m, sizeof (struct ip6_hdr))) == 0) {
1c79356b
A
593 ip6stat.ip6s_toosmall++;
594 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
6d2010ae 595 goto done;
1c79356b
A
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++;
39236c6e 603 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
1c79356b
A
604 goto bad;
605 }
606
607 ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
1c79356b 608 /*
9bccf70c 609 * Check against address spoofing/corruption.
1c79356b 610 */
04b8595b
A
611 if (!(m->m_pkthdr.pkt_flags & PKTF_LOOP) &&
612 IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src)) {
613 ip6stat.ip6s_badscope++;
614 in6_ifstat_inc(inifp, ifs6_in_addrerr);
615 goto bad;
616 }
1c79356b
A
617 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
618 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
9bccf70c
A
619 /*
620 * XXX: "badscope" is not very suitable for a multicast source.
621 */
622 ip6stat.ip6s_badscope++;
39236c6e 623 in6_ifstat_inc(inifp, ifs6_in_addrerr);
9bccf70c
A
624 goto bad;
625 }
6d2010ae 626 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) &&
39236c6e 627 !(m->m_pkthdr.pkt_flags & PKTF_LOOP)) {
6d2010ae
A
628 /*
629 * In this case, the packet should come from the loopback
630 * interface. However, we cannot just check the if_flags,
631 * because ip6_mloopback() passes the "actual" interface
632 * as the outgoing/incoming interface.
633 */
1c79356b 634 ip6stat.ip6s_badscope++;
39236c6e 635 in6_ifstat_inc(inifp, ifs6_in_addrerr);
1c79356b
A
636 goto bad;
637 }
55e303ae 638
1c79356b 639 /*
9bccf70c
A
640 * The following check is not documented in specs. A malicious
641 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
642 * and bypass security checks (act as if it was from 127.0.0.1 by using
6d2010ae 643 * IPv6 src ::ffff:127.0.0.1). Be cautious.
9bccf70c
A
644 *
645 * This check chokes if we are in an SIIT cloud. As none of BSDs
646 * support IPv4-less kernel compilation, we cannot support SIIT
647 * environment at all. So, it makes more sense for us to reject any
648 * malicious packets for non-SIIT environment, than try to do a
6d2010ae 649 * partial support for SIIT environment.
1c79356b 650 */
9bccf70c
A
651 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
652 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
653 ip6stat.ip6s_badscope++;
39236c6e 654 in6_ifstat_inc(inifp, ifs6_in_addrerr);
9bccf70c
A
655 goto bad;
656 }
1c79356b
A
657#if 0
658 /*
659 * Reject packets with IPv4 compatible addresses (auto tunnel).
660 *
661 * The code forbids auto tunnel relay case in RFC1933 (the check is
662 * stronger than RFC1933). We may want to re-enable it if mech-xx
663 * is revised to forbid relaying case.
664 */
665 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
666 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
667 ip6stat.ip6s_badscope++;
39236c6e 668 in6_ifstat_inc(inifp, ifs6_in_addrerr);
1c79356b
A
669 goto bad;
670 }
671#endif
04b8595b
A
672#if IPFW2
673 /*
674 * Check with the firewall...
675 */
676 if (ip6_fw_enable && ip6_fw_chk_ptr) {
677 u_short port = 0;
678 /* If ipfw says divert, we have to just drop packet */
679 /* use port as a dummy argument */
680 if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) {
681 m_freem(m);
682 m = NULL;
683 }
684 if (!m)
685 goto done;
686 }
687#endif /* IPFW2 */
9bccf70c 688
6d2010ae
A
689 /*
690 * Naively assume we can attribute inbound data to the route we would
691 * use to send to this destination. Asymetric routing breaks this
692 * assumption, but it still allows us to account for traffic from
693 * a remote node in the routing table.
694 * this has a very significant performance impact so we bypass
695 * if nstat_collect is disabled. We may also bypass if the
696 * protocol is tcp in the future because tcp will have a route that
697 * we can use to attribute the data to. That does mean we would not
698 * account for forwarded tcp traffic.
699 */
700 if (nstat_collect) {
701 struct rtentry *rte =
39236c6e 702 ifnet_cached_rtlookup_inet6(inifp, &ip6->ip6_src);
6d2010ae
A
703 if (rte != NULL) {
704 nstat_route_rx(rte, 1, m->m_pkthdr.len, 0);
705 rtfree(rte);
706 }
707 }
708
39236c6e
A
709 /* for consistency */
710 m->m_pkthdr.pkt_proto = ip6->ip6_nxt;
711
316670eb
A
712#if DUMMYNET
713check_with_pf:
39236c6e 714#endif /* DUMMYNET */
b0d623f7
A
715#if PF
716 /* Invoke inbound packet filter */
6d2010ae
A
717 if (PF_IS_ENABLED) {
718 int error;
316670eb 719#if DUMMYNET
39236c6e
A
720 error = pf_af_hook(inifp, NULL, &m, AF_INET6, TRUE, &args);
721#else /* !DUMMYNET */
722 error = pf_af_hook(inifp, NULL, &m, AF_INET6, TRUE, NULL);
723#endif /* !DUMMYNET */
316670eb 724 if (error != 0 || m == NULL) {
6d2010ae 725 if (m != NULL) {
39236c6e
A
726 panic("%s: unexpected packet %p\n",
727 __func__, m);
6d2010ae
A
728 /* NOTREACHED */
729 }
730 /* Already freed by callee */
731 goto done;
b0d623f7 732 }
6d2010ae 733 ip6 = mtod(m, struct ip6_hdr *);
b0d623f7 734 }
b0d623f7
A
735#endif /* PF */
736
9bccf70c 737 /* drop packets if interface ID portion is already filled */
39236c6e
A
738 if (!(inifp->if_flags & IFF_LOOPBACK) &&
739 !(m->m_pkthdr.pkt_flags & PKTF_LOOP)) {
9bccf70c
A
740 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) &&
741 ip6->ip6_src.s6_addr16[1]) {
742 ip6stat.ip6s_badscope++;
743 goto bad;
744 }
fe8ab488 745 if (IN6_IS_SCOPE_EMBED(&ip6->ip6_dst) &&
9bccf70c 746 ip6->ip6_dst.s6_addr16[1]) {
1c79356b 747 ip6stat.ip6s_badscope++;
1c79356b
A
748 goto bad;
749 }
750 }
751
39236c6e
A
752 if (m->m_pkthdr.pkt_flags & PKTF_IFAINFO) {
753 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
754 ip6->ip6_src.s6_addr16[1] =
755 htons(m->m_pkthdr.src_ifindex);
fe8ab488 756 if (IN6_IS_SCOPE_EMBED(&ip6->ip6_dst))
39236c6e
A
757 ip6->ip6_dst.s6_addr16[1] =
758 htons(m->m_pkthdr.dst_ifindex);
759 } else {
760 if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
761 ip6->ip6_src.s6_addr16[1] = htons(inifp->if_index);
fe8ab488 762 if (IN6_IS_SCOPE_EMBED(&ip6->ip6_dst))
39236c6e
A
763 ip6->ip6_dst.s6_addr16[1] = htons(inifp->if_index);
764 }
9bccf70c 765
1c79356b
A
766 /*
767 * Multicast check
768 */
769 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
6d2010ae 770 struct in6_multi *in6m = NULL;
1c79356b 771
39236c6e 772 in6_ifstat_inc_na(inifp, ifs6_in_mcast);
1c79356b
A
773 /*
774 * See if we belong to the destination multicast group on the
775 * arrival interface.
776 */
6d2010ae 777 in6_multihead_lock_shared();
39236c6e 778 IN6_LOOKUP_MULTI(&ip6->ip6_dst, inifp, in6m);
6d2010ae
A
779 in6_multihead_lock_done();
780 if (in6m != NULL) {
781 IN6M_REMREF(in6m);
1c79356b 782 ours = 1;
fe8ab488 783 } else if (!nd6_prproxy) {
1c79356b
A
784 ip6stat.ip6s_notmember++;
785 ip6stat.ip6s_cantforward++;
39236c6e 786 in6_ifstat_inc(inifp, ifs6_in_discard);
1c79356b
A
787 goto bad;
788 }
39236c6e
A
789 deliverifp = inifp;
790 VERIFY(ia6 == NULL);
1c79356b
A
791 goto hbhcheck;
792 }
793
794 /*
39236c6e
A
795 * Unicast check
796 *
797 * Fast path: see if the target is ourselves.
1c79356b 798 */
39236c6e
A
799 lck_rw_lock_shared(&in6_ifaddr_rwlock);
800 for (ia6 = in6_ifaddrs; ia6 != NULL; ia6 = ia6->ia_next) {
801 /*
802 * No reference is held on the address, as we just need
803 * to test for a few things while holding the RW lock.
804 */
805 if (IN6_ARE_ADDR_EQUAL(&ia6->ia_addr.sin6_addr, &ip6->ip6_dst))
806 break;
807 }
808
809 if (ia6 != NULL) {
810 /*
811 * For performance, test without acquiring the address lock;
812 * a lot of things in the address are set once and never
813 * changed (e.g. ia_ifp.)
814 */
815 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
816 /* this address is ready */
817 ours = 1;
818 deliverifp = ia6->ia_ifp;
819 /*
820 * record dst address information into mbuf.
821 */
822 (void) ip6_setdstifaddr_info(m, 0, ia6);
823 lck_rw_done(&in6_ifaddr_rwlock);
824 goto hbhcheck;
825 }
826 lck_rw_done(&in6_ifaddr_rwlock);
827 ia6 = NULL;
828 /* address is not ready, so discard the packet. */
829 nd6log((LOG_INFO, "%s: packet to an unready address %s->%s\n",
830 __func__, ip6_sprintf(&ip6->ip6_src),
831 ip6_sprintf(&ip6->ip6_dst)));
832 goto bad;
833 }
834 lck_rw_done(&in6_ifaddr_rwlock);
835
836 /*
837 * Slow path: route lookup.
838 */
839 dst6 = SIN6(&rin6.ro_dst);
840 dst6->sin6_len = sizeof (struct sockaddr_in6);
6d2010ae
A
841 dst6->sin6_family = AF_INET6;
842 dst6->sin6_addr = ip6->ip6_dst;
1c79356b 843
39236c6e 844 rtalloc_scoped_ign((struct route *)&rin6,
6d2010ae 845 RTF_PRCLONING, IFSCOPE_NONE);
39236c6e
A
846 if (rin6.ro_rt != NULL)
847 RT_LOCK_SPIN(rin6.ro_rt);
1c79356b 848
39236c6e 849#define rt6_key(r) (SIN6((r)->rt_nodes->rn_key))
1c79356b
A
850
851 /*
852 * Accept the packet if the forwarding interface to the destination
853 * according to the routing table is the loopback interface,
854 * unless the associated route has a gateway.
855 * Note that this approach causes to accept a packet if there is a
856 * route to the loopback interface for the destination of the packet.
857 * But we think it's even useful in some situations, e.g. when using
858 * a special daemon which wants to intercept the packet.
9bccf70c
A
859 *
860 * XXX: some OSes automatically make a cloned route for the destination
861 * of an outgoing packet. If the outgoing interface of the packet
862 * is a loopback one, the kernel would consider the packet to be
863 * accepted, even if we have no such address assinged on the interface.
864 * We check the cloned flag of the route entry to reject such cases,
865 * assuming that route entries for our own addresses are not made by
866 * cloning (it should be true because in6_addloop explicitly installs
867 * the host route). However, we might have to do an explicit check
868 * while it would be less efficient. Or, should we rather install a
869 * reject route for such a case?
1c79356b 870 */
39236c6e
A
871 if (rin6.ro_rt != NULL &&
872 (rin6.ro_rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
9bccf70c 873#if RTF_WASCLONED
39236c6e 874 !(rin6.ro_rt->rt_flags & RTF_WASCLONED) &&
1c79356b 875#endif
39236c6e
A
876 rin6.ro_rt->rt_ifp->if_type == IFT_LOOP) {
877 ia6 = (struct in6_ifaddr *)rin6.ro_rt->rt_ifa;
9bccf70c 878 /*
39236c6e 879 * Packets to a tentative, duplicated, or somehow invalid
9bccf70c 880 * address must not be accepted.
39236c6e
A
881 *
882 * For performance, test without acquiring the address lock;
883 * a lot of things in the address are set once and never
884 * changed (e.g. ia_ifp.)
9bccf70c 885 */
1c79356b 886 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
9bccf70c 887 /* this address is ready */
1c79356b
A
888 ours = 1;
889 deliverifp = ia6->ia_ifp; /* correct? */
39236c6e
A
890 /*
891 * record dst address information into mbuf.
892 */
893 (void) ip6_setdstifaddr_info(m, 0, ia6);
894 RT_UNLOCK(rin6.ro_rt);
1c79356b 895 goto hbhcheck;
1c79356b 896 }
39236c6e
A
897 RT_UNLOCK(rin6.ro_rt);
898 ia6 = NULL;
6d2010ae 899 /* address is not ready, so discard the packet. */
39236c6e
A
900 nd6log((LOG_INFO, "%s: packet to an unready address %s->%s\n",
901 __func__, ip6_sprintf(&ip6->ip6_src),
6d2010ae 902 ip6_sprintf(&ip6->ip6_dst)));
6d2010ae 903 goto bad;
1c79356b
A
904 }
905
39236c6e
A
906 if (rin6.ro_rt != NULL)
907 RT_UNLOCK(rin6.ro_rt);
1c79356b 908
1c79356b
A
909 /*
910 * Now there is no reason to process the packet if it's not our own
911 * and we're not a router.
912 */
913 if (!ip6_forwarding) {
914 ip6stat.ip6s_cantforward++;
39236c6e 915 in6_ifstat_inc(inifp, ifs6_in_discard);
1c79356b
A
916 goto bad;
917 }
918
39236c6e 919hbhcheck:
9bccf70c 920 /*
39236c6e 921 * record dst address information into mbuf, if we don't have one yet.
9bccf70c 922 * note that we are unable to record it, if the address is not listed
316670eb 923 * as our interface address (e.g. multicast addresses, etc.)
9bccf70c 924 */
39236c6e 925 if (deliverifp != NULL && ia6 == NULL) {
9bccf70c 926 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
39236c6e
A
927 if (ia6 != NULL) {
928 (void) ip6_setdstifaddr_info(m, 0, ia6);
6d2010ae 929 IFA_REMREF(&ia6->ia_ifa);
9bccf70c
A
930 }
931 }
932
1c79356b
A
933 /*
934 * Process Hop-by-Hop options header if it's contained.
935 * m may be modified in ip6_hopopts_input().
936 * If a JumboPayload option is included, plen will also be modified.
937 */
938 plen = (u_int32_t)ntohs(ip6->ip6_plen);
939 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
940 struct ip6_hbh *hbh;
941
942 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
39236c6e
A
943#if 0 /* touches NULL pointer */
944 in6_ifstat_inc(inifp, ifs6_in_discard);
1c79356b 945#endif
6d2010ae 946 goto done; /* m have already been freed */
1c79356b 947 }
9bccf70c 948
1c79356b
A
949 /* adjust pointer */
950 ip6 = mtod(m, struct ip6_hdr *);
9bccf70c
A
951
952 /*
55e303ae 953 * if the payload length field is 0 and the next header field
9bccf70c
A
954 * indicates Hop-by-Hop Options header, then a Jumbo Payload
955 * option MUST be included.
956 */
957 if (ip6->ip6_plen == 0 && plen == 0) {
958 /*
959 * Note that if a valid jumbo payload option is
6d2010ae
A
960 * contained, ip6_hopopts_input() must set a valid
961 * (non-zero) payload length to the variable plen.
9bccf70c
A
962 */
963 ip6stat.ip6s_badoptions++;
39236c6e
A
964 in6_ifstat_inc(inifp, ifs6_in_discard);
965 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
966 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER,
967 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
6d2010ae 968 goto done;
9bccf70c 969 }
1c79356b
A
970 /* ip6_hopopts_input() ensures that mbuf is contiguous */
971 hbh = (struct ip6_hbh *)(ip6 + 1);
1c79356b
A
972 nxt = hbh->ip6h_nxt;
973
974 /*
6d2010ae
A
975 * If we are acting as a router and the packet contains a
976 * router alert option, see if we know the option value.
977 * Currently, we only support the option value for MLD, in which
978 * case we should pass the packet to the multicast routing
979 * daemon.
1c79356b 980 */
6d2010ae
A
981 if (rtalert != ~0 && ip6_forwarding) {
982 switch (rtalert) {
983 case IP6OPT_RTALERT_MLD:
984 ours = 1;
985 break;
986 default:
987 /*
988 * RFC2711 requires unrecognized values must be
989 * silently ignored.
990 */
991 break;
992 }
993 }
1c79356b
A
994 } else
995 nxt = ip6->ip6_nxt;
996
997 /*
998 * Check that the amount of data in the buffers
999 * is as at least much as the IPv6 header would have us expect.
1000 * Trim mbufs if longer than we expect.
1001 * Drop packet if shorter than we expect.
1002 */
39236c6e 1003 if (m->m_pkthdr.len - sizeof (struct ip6_hdr) < plen) {
1c79356b 1004 ip6stat.ip6s_tooshort++;
39236c6e 1005 in6_ifstat_inc(inifp, ifs6_in_truncated);
1c79356b
A
1006 goto bad;
1007 }
39236c6e
A
1008 if (m->m_pkthdr.len > sizeof (struct ip6_hdr) + plen) {
1009 /*
1010 * Invalidate hardware checksum info if ip6_adj_clear_hwcksum
1011 * is set; useful to handle buggy drivers. Note that this
1012 * should not be enabled by default, as we may get here due
1013 * to link-layer padding.
1014 */
1015 if (ip6_adj_clear_hwcksum &&
1016 (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) &&
1017 !(inifp->if_flags & IFF_LOOPBACK) &&
1018 !(m->m_pkthdr.pkt_flags & PKTF_LOOP)) {
1019 m->m_pkthdr.csum_flags &= ~CSUM_DATA_VALID;
1020 m->m_pkthdr.csum_data = 0;
1021 ip6stat.ip6s_adj_hwcsum_clr++;
1022 }
1023
1024 ip6stat.ip6s_adj++;
1c79356b 1025 if (m->m_len == m->m_pkthdr.len) {
39236c6e
A
1026 m->m_len = sizeof (struct ip6_hdr) + plen;
1027 m->m_pkthdr.len = sizeof (struct ip6_hdr) + plen;
1028 } else {
1029 m_adj(m, sizeof (struct ip6_hdr) + plen -
1030 m->m_pkthdr.len);
1031 }
1c79356b
A
1032 }
1033
1034 /*
1035 * Forward if desirable.
1036 */
1037 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
316670eb
A
1038 if (!ours && nd6_prproxy) {
1039 /*
1040 * If this isn't for us, this might be a Neighbor
1041 * Solicitation (dst is solicited-node multicast)
1042 * against an address in one of the proxied prefixes;
1043 * if so, claim the packet and let icmp6_input()
1044 * handle the rest.
1045 */
1046 ours = nd6_prproxy_isours(m, ip6, NULL, IFSCOPE_NONE);
1047 VERIFY(!ours ||
39236c6e 1048 (m->m_pkthdr.pkt_flags & PKTF_PROXY_DST));
316670eb 1049 }
6d2010ae
A
1050 if (!ours)
1051 goto bad;
1c79356b 1052 } else if (!ours) {
316670eb
A
1053 /*
1054 * The unicast forwarding function might return the packet
1055 * if we are proxying prefix(es), and if the packet is an
1056 * ICMPv6 packet that has failed the zone checks, but is
1057 * targetted towards a proxied address (this is optimized by
1058 * way of RTF_PROXY test.) If so, claim the packet as ours
1059 * and let icmp6_input() handle the rest. The packet's hop
1060 * limit value is kept intact (it's not decremented). This
1061 * is for supporting Neighbor Unreachability Detection between
1062 * proxied nodes on different links (src is link-local, dst
1063 * is target address.)
1064 */
39236c6e 1065 if ((m = ip6_forward(m, &rin6, 0)) == NULL)
316670eb 1066 goto done;
39236c6e
A
1067 VERIFY(rin6.ro_rt != NULL);
1068 VERIFY(m->m_pkthdr.pkt_flags & PKTF_PROXY_DST);
1069 deliverifp = rin6.ro_rt->rt_ifp;
316670eb
A
1070 ours = 1;
1071 }
1c79356b
A
1072
1073 ip6 = mtod(m, struct ip6_hdr *);
1074
1075 /*
1076 * Malicious party may be able to use IPv4 mapped addr to confuse
1077 * tcp/udp stack and bypass security checks (act as if it was from
1078 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
1079 *
1080 * For SIIT end node behavior, you may want to disable the check.
1081 * However, you will become vulnerable to attacks using IPv4 mapped
1082 * source.
1083 */
1084 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
1085 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
1086 ip6stat.ip6s_badscope++;
39236c6e 1087 in6_ifstat_inc(inifp, ifs6_in_addrerr);
1c79356b
A
1088 goto bad;
1089 }
1090
1091 /*
1092 * Tell launch routine the next header
1093 */
1c79356b 1094 ip6stat.ip6s_delivered++;
39236c6e 1095 in6_ifstat_inc_na(deliverifp, ifs6_in_deliver);
91447636 1096
91447636 1097injectit:
1c79356b
A
1098 nest = 0;
1099
316670eb
A
1100 /*
1101 * Perform IP header alignment fixup again, if needed. Note that
1102 * we do it once for the outermost protocol, and we assume each
1103 * protocol handler wouldn't mess with the alignment afterwards.
1104 */
39236c6e 1105 IP6_HDR_ALIGNMENT_FIXUP(m, inifp, return);
316670eb 1106
1c79356b 1107 while (nxt != IPPROTO_DONE) {
91447636 1108 struct ipfilter *filter;
6d2010ae 1109 int (*pr_input)(struct mbuf **, int *, int);
0b4c1975 1110
1c79356b
A
1111 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
1112 ip6stat.ip6s_toomanyhdr++;
6d2010ae 1113 goto bad;
1c79356b
A
1114 }
1115
1116 /*
1117 * protection against faulty packet - there should be
1118 * more sanity checks in header chain processing.
1119 */
9bccf70c 1120 if (m->m_pkthdr.len < off) {
1c79356b 1121 ip6stat.ip6s_tooshort++;
39236c6e 1122 in6_ifstat_inc(inifp, ifs6_in_truncated);
9bccf70c
A
1123 goto bad;
1124 }
6d2010ae 1125
9bccf70c
A
1126
1127#if IPSEC
1128 /*
1129 * enforce IPsec policy checking if we are seeing last header.
1130 * note that we do not visit this with protocols with pcb layer
1131 * code - like udp/tcp/raw ip.
1132 */
316670eb
A
1133 if ((ipsec_bypass == 0) &&
1134 (ip6_protox[nxt]->pr_flags & PR_LASTHDR) != 0) {
91447636 1135 if (ipsec6_in_reject(m, NULL)) {
2d21ac55 1136 IPSEC_STAT_INCREMENT(ipsec6stat.in_polvio);
6d2010ae 1137 goto bad;
39236c6e 1138 }
1c79356b 1139 }
39236c6e 1140#endif /* IPSEC */
9bccf70c 1141
91447636 1142 /*
2d21ac55 1143 * Call IP filter
91447636 1144 */
2d21ac55 1145 if (!TAILQ_EMPTY(&ipv6_filters)) {
91447636
A
1146 ipf_ref();
1147 TAILQ_FOREACH(filter, &ipv6_filters, ipf_link) {
1148 if (seen == 0) {
316670eb
A
1149 if ((struct ipfilter *)inject_ipfref ==
1150 filter)
91447636
A
1151 seen = 1;
1152 } else if (filter->ipf_filter.ipf_input) {
1153 errno_t result;
316670eb 1154
91447636 1155 result = filter->ipf_filter.ipf_input(
39236c6e
A
1156 filter->ipf_filter.cookie,
1157 (mbuf_t *)&m, off, nxt);
91447636
A
1158 if (result == EJUSTRETURN) {
1159 ipf_unref();
6d2010ae 1160 goto done;
91447636
A
1161 }
1162 if (result != 0) {
1163 ipf_unref();
6d2010ae 1164 goto bad;
91447636
A
1165 }
1166 }
1167 }
1168 ipf_unref();
1169 }
0b4c1975 1170
6d2010ae 1171 DTRACE_IP6(receive, struct mbuf *, m, struct inpcb *, NULL,
39236c6e
A
1172 struct ip6_hdr *, ip6, struct ifnet *, inifp,
1173 struct ip *, NULL, struct ip6_hdr *, ip6);
6d2010ae 1174
0b4c1975
A
1175 if ((pr_input = ip6_protox[nxt]->pr_input) == NULL) {
1176 m_freem(m);
1177 m = NULL;
1178 nxt = IPPROTO_DONE;
1179 } else if (!(ip6_protox[nxt]->pr_flags & PR_PROTOLOCK)) {
91447636 1180 lck_mtx_lock(inet6_domain_mutex);
6d2010ae 1181 nxt = pr_input(&m, &off, nxt);
91447636 1182 lck_mtx_unlock(inet6_domain_mutex);
0b4c1975 1183 } else {
6d2010ae 1184 nxt = pr_input(&m, &off, nxt);
91447636 1185 }
1c79356b 1186 }
6d2010ae 1187done:
39236c6e 1188 ROUTE_RELEASE(&rin6);
1c79356b 1189 return;
39236c6e 1190bad:
1c79356b 1191 m_freem(m);
6d2010ae 1192 goto done;
1c79356b
A
1193}
1194
39236c6e
A
1195void
1196ip6_setsrcifaddr_info(struct mbuf *m, uint32_t src_idx, struct in6_ifaddr *ia6)
9bccf70c 1197{
39236c6e
A
1198 VERIFY(m->m_flags & M_PKTHDR);
1199
1200 /*
1201 * If the source ifaddr is specified, pick up the information
1202 * from there; otherwise just grab the passed-in ifindex as the
1203 * caller may not have the ifaddr available.
1204 */
1205 if (ia6 != NULL) {
1206 m->m_pkthdr.pkt_flags |= PKTF_IFAINFO;
1207 m->m_pkthdr.src_ifindex = ia6->ia_ifp->if_index;
1208
1209 /* See IN6_IFF comments in in6_var.h */
1210 m->m_pkthdr.src_iff = (ia6->ia6_flags & 0xffff);
1211 } else {
1212 m->m_pkthdr.src_iff = 0;
1213 m->m_pkthdr.src_ifindex = src_idx;
1214 if (src_idx != 0)
1215 m->m_pkthdr.pkt_flags |= PKTF_IFAINFO;
6d2010ae 1216 }
9bccf70c
A
1217}
1218
39236c6e
A
1219void
1220ip6_setdstifaddr_info(struct mbuf *m, uint32_t dst_idx, struct in6_ifaddr *ia6)
9bccf70c 1221{
39236c6e 1222 VERIFY(m->m_flags & M_PKTHDR);
9bccf70c 1223
39236c6e
A
1224 /*
1225 * If the destination ifaddr is specified, pick up the information
1226 * from there; otherwise just grab the passed-in ifindex as the
1227 * caller may not have the ifaddr available.
1228 */
1229 if (ia6 != NULL) {
1230 m->m_pkthdr.pkt_flags |= PKTF_IFAINFO;
1231 m->m_pkthdr.dst_ifindex = ia6->ia_ifp->if_index;
1232
1233 /* See IN6_IFF comments in in6_var.h */
1234 m->m_pkthdr.dst_iff = (ia6->ia6_flags & 0xffff);
1235 } else {
1236 m->m_pkthdr.dst_iff = 0;
1237 m->m_pkthdr.dst_ifindex = dst_idx;
1238 if (dst_idx != 0)
1239 m->m_pkthdr.pkt_flags |= PKTF_IFAINFO;
6d2010ae 1240 }
39236c6e
A
1241}
1242
1243int
1244ip6_getsrcifaddr_info(struct mbuf *m, uint32_t *src_idx, uint32_t *ia6f)
1245{
1246 VERIFY(m->m_flags & M_PKTHDR);
1247
1248 if (!(m->m_pkthdr.pkt_flags & PKTF_IFAINFO))
1249 return (-1);
1250
1251 if (src_idx != NULL)
1252 *src_idx = m->m_pkthdr.src_ifindex;
1253
1254 if (ia6f != NULL)
1255 *ia6f = m->m_pkthdr.src_iff;
1256
1257 return (0);
1258}
1259
1260int
1261ip6_getdstifaddr_info(struct mbuf *m, uint32_t *dst_idx, uint32_t *ia6f)
1262{
1263 VERIFY(m->m_flags & M_PKTHDR);
1264
1265 if (!(m->m_pkthdr.pkt_flags & PKTF_IFAINFO))
1266 return (-1);
1267
1268 if (dst_idx != NULL)
1269 *dst_idx = m->m_pkthdr.dst_ifindex;
1270
1271 if (ia6f != NULL)
1272 *ia6f = m->m_pkthdr.dst_iff;
1273
1274 return (0);
9bccf70c
A
1275}
1276
1c79356b
A
1277/*
1278 * Hop-by-Hop options header processing. If a valid jumbo payload option is
1279 * included, the real payload length will be stored in plenp.
1280 */
1281static int
6d2010ae
A
1282ip6_hopopts_input(uint32_t *plenp, uint32_t *rtalertp, struct mbuf **mp,
1283 int *offp)
1c79356b 1284{
9bccf70c 1285 struct mbuf *m = *mp;
1c79356b
A
1286 int off = *offp, hbhlen;
1287 struct ip6_hbh *hbh;
1288 u_int8_t *opt;
1289
1290 /* validation of the length of the header */
39236c6e 1291 IP6_EXTHDR_CHECK(m, off, sizeof (*hbh), return (-1));
1c79356b
A
1292 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
1293 hbhlen = (hbh->ip6h_len + 1) << 3;
1294
39236c6e 1295 IP6_EXTHDR_CHECK(m, off, hbhlen, return (-1));
1c79356b 1296 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
1c79356b 1297 off += hbhlen;
39236c6e
A
1298 hbhlen -= sizeof (struct ip6_hbh);
1299 opt = (u_int8_t *)hbh + sizeof (struct ip6_hbh);
1c79356b 1300
39236c6e
A
1301 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof (struct ip6_hbh),
1302 hbhlen, rtalertp, plenp) < 0)
6d2010ae 1303 return (-1);
1c79356b
A
1304
1305 *offp = off;
1306 *mp = m;
6d2010ae 1307 return (0);
1c79356b
A
1308}
1309
1310/*
1311 * Search header for all Hop-by-hop options and process each option.
1312 * This function is separate from ip6_hopopts_input() in order to
1313 * handle a case where the sending node itself process its hop-by-hop
1314 * options header. In such a case, the function is called from ip6_output().
9bccf70c
A
1315 *
1316 * The function assumes that hbh header is located right after the IPv6 header
1317 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
1318 * opthead + hbhlen is located in continuous memory region.
1c79356b
A
1319 */
1320int
1321ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
1322 struct mbuf *m;
1323 u_int8_t *opthead;
1324 int hbhlen;
1325 u_int32_t *rtalertp;
1326 u_int32_t *plenp;
1327{
1328 struct ip6_hdr *ip6;
1329 int optlen = 0;
1330 u_int8_t *opt = opthead;
1331 u_int16_t rtalert_val;
9bccf70c 1332 u_int32_t jumboplen;
39236c6e 1333 const int erroff = sizeof (struct ip6_hdr) + sizeof (struct ip6_hbh);
1c79356b
A
1334
1335 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
9bccf70c
A
1336 switch (*opt) {
1337 case IP6OPT_PAD1:
1338 optlen = 1;
1339 break;
1340 case IP6OPT_PADN:
1341 if (hbhlen < IP6OPT_MINLEN) {
1342 ip6stat.ip6s_toosmall++;
1343 goto bad;
1344 }
1345 optlen = *(opt + 1) + 2;
1346 break;
6d2010ae 1347 case IP6OPT_ROUTER_ALERT:
9bccf70c
A
1348 /* XXX may need check for alignment */
1349 if (hbhlen < IP6OPT_RTALERT_LEN) {
1350 ip6stat.ip6s_toosmall++;
1351 goto bad;
1352 }
1353 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
1354 /* XXX stat */
1355 icmp6_error(m, ICMP6_PARAM_PROB,
1356 ICMP6_PARAMPROB_HEADER,
1357 erroff + opt + 1 - opthead);
39236c6e 1358 return (-1);
9bccf70c
A
1359 }
1360 optlen = IP6OPT_RTALERT_LEN;
1361 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
1362 *rtalertp = ntohs(rtalert_val);
1363 break;
1364 case IP6OPT_JUMBO:
1365 /* XXX may need check for alignment */
1366 if (hbhlen < IP6OPT_JUMBO_LEN) {
1367 ip6stat.ip6s_toosmall++;
1368 goto bad;
1369 }
1370 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
1371 /* XXX stat */
1372 icmp6_error(m, ICMP6_PARAM_PROB,
1373 ICMP6_PARAMPROB_HEADER,
1374 erroff + opt + 1 - opthead);
39236c6e 1375 return (-1);
9bccf70c
A
1376 }
1377 optlen = IP6OPT_JUMBO_LEN;
1378
1379 /*
1380 * IPv6 packets that have non 0 payload length
1381 * must not contain a jumbo payload option.
1382 */
1383 ip6 = mtod(m, struct ip6_hdr *);
1384 if (ip6->ip6_plen) {
1385 ip6stat.ip6s_badoptions++;
1386 icmp6_error(m, ICMP6_PARAM_PROB,
1387 ICMP6_PARAMPROB_HEADER,
1388 erroff + opt - opthead);
39236c6e 1389 return (-1);
9bccf70c
A
1390 }
1391
1392 /*
1393 * We may see jumbolen in unaligned location, so
1394 * we'd need to perform bcopy().
1395 */
39236c6e 1396 bcopy(opt + 2, &jumboplen, sizeof (jumboplen));
9bccf70c
A
1397 jumboplen = (u_int32_t)htonl(jumboplen);
1398
1399#if 1
1400 /*
1401 * if there are multiple jumbo payload options,
1402 * *plenp will be non-zero and the packet will be
1403 * rejected.
1404 * the behavior may need some debate in ipngwg -
1405 * multiple options does not make sense, however,
1406 * there's no explicit mention in specification.
1407 */
1408 if (*plenp != 0) {
1409 ip6stat.ip6s_badoptions++;
1410 icmp6_error(m, ICMP6_PARAM_PROB,
1411 ICMP6_PARAMPROB_HEADER,
1412 erroff + opt + 2 - opthead);
39236c6e 1413 return (-1);
9bccf70c 1414 }
1c79356b 1415#endif
9bccf70c
A
1416
1417 /*
1418 * jumbo payload length must be larger than 65535.
1419 */
1420 if (jumboplen <= IPV6_MAXPACKET) {
1421 ip6stat.ip6s_badoptions++;
1422 icmp6_error(m, ICMP6_PARAM_PROB,
1423 ICMP6_PARAMPROB_HEADER,
1424 erroff + opt + 2 - opthead);
39236c6e 1425 return (-1);
9bccf70c
A
1426 }
1427 *plenp = jumboplen;
1428
1429 break;
1430 default: /* unknown option */
1431 if (hbhlen < IP6OPT_MINLEN) {
1432 ip6stat.ip6s_toosmall++;
1433 goto bad;
1434 }
1435 optlen = ip6_unknown_opt(opt, m,
6d2010ae 1436 erroff + opt - opthead);
91447636 1437 if (optlen == -1) {
39236c6e 1438 return (-1);
91447636 1439 }
9bccf70c
A
1440 optlen += 2;
1441 break;
1c79356b
A
1442 }
1443 }
1444
39236c6e 1445 return (0);
1c79356b 1446
39236c6e 1447bad:
1c79356b 1448 m_freem(m);
39236c6e 1449 return (-1);
1c79356b
A
1450}
1451
1452/*
1453 * Unknown option processing.
1454 * The third argument `off' is the offset from the IPv6 header to the option,
1455 * which is necessary if the IPv6 header the and option header and IPv6 header
1456 * is not continuous in order to return an ICMPv6 error.
1457 */
1458int
6d2010ae 1459ip6_unknown_opt(uint8_t *optp, struct mbuf *m, int off)
1c79356b
A
1460{
1461 struct ip6_hdr *ip6;
1462
9bccf70c
A
1463 switch (IP6OPT_TYPE(*optp)) {
1464 case IP6OPT_TYPE_SKIP: /* ignore the option */
39236c6e
A
1465 return ((int)*(optp + 1));
1466
9bccf70c
A
1467 case IP6OPT_TYPE_DISCARD: /* silently discard */
1468 m_freem(m);
39236c6e
A
1469 return (-1);
1470
9bccf70c
A
1471 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1472 ip6stat.ip6s_badoptions++;
1473 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
39236c6e
A
1474 return (-1);
1475
9bccf70c
A
1476 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1477 ip6stat.ip6s_badoptions++;
1478 ip6 = mtod(m, struct ip6_hdr *);
1479 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
39236c6e 1480 (m->m_flags & (M_BCAST|M_MCAST))) {
9bccf70c 1481 m_freem(m);
39236c6e 1482 } else {
9bccf70c 1483 icmp6_error(m, ICMP6_PARAM_PROB,
39236c6e
A
1484 ICMP6_PARAMPROB_OPTION, off);
1485 }
1486 return (-1);
1c79356b
A
1487 }
1488
1489 m_freem(m); /* XXX: NOTREACHED */
6d2010ae 1490 return (-1);
1c79356b
A
1491}
1492
1493/*
1494 * Create the "control" list for this pcb.
6d2010ae 1495 * These functions will not modify mbuf chain at all.
1c79356b 1496 *
6d2010ae 1497 * With KAME mbuf chain restriction:
1c79356b
A
1498 * The routine will be called from upper layer handlers like tcp6_input().
1499 * Thus the routine assumes that the caller (tcp6_input) have already
1500 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1501 * very first mbuf on the mbuf chain.
6d2010ae
A
1502 *
1503 * ip6_savecontrol_v4 will handle those options that are possible to be
1504 * set on a v4-mapped socket.
1505 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those
1506 * options and handle the v6-only ones itself.
1c79356b 1507 */
6d2010ae
A
1508struct mbuf **
1509ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp,
1510 int *v4only)
1c79356b 1511{
6d2010ae 1512 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1c79356b 1513
6d2010ae 1514 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) {
1c79356b
A
1515 struct timeval tv;
1516
39236c6e
A
1517 getmicrotime(&tv);
1518 mp = sbcreatecontrol_mbuf((caddr_t)&tv, sizeof (tv),
6d2010ae 1519 SCM_TIMESTAMP, SOL_SOCKET, mp);
39236c6e
A
1520 if (*mp == NULL)
1521 return (NULL);
1c79356b 1522 }
39236c6e
A
1523 if ((inp->inp_socket->so_options & SO_TIMESTAMP_MONOTONIC) != 0) {
1524 uint64_t time;
1c79356b 1525
39236c6e
A
1526 time = mach_absolute_time();
1527 mp = sbcreatecontrol_mbuf((caddr_t)&time, sizeof (time),
1528 SCM_TIMESTAMP_MONOTONIC, SOL_SOCKET, mp);
1529 if (*mp == NULL)
1530 return (NULL);
1531 }
6d2010ae 1532 if ((inp->inp_socket->so_flags & SOF_RECV_TRAFFIC_CLASS) != 0) {
316670eb
A
1533 int tc = m_get_traffic_class(m);
1534
39236c6e
A
1535 mp = sbcreatecontrol_mbuf((caddr_t)&tc, sizeof (tc),
1536 SO_TRAFFIC_CLASS, SOL_SOCKET, mp);
316670eb
A
1537 if (*mp == NULL)
1538 return (NULL);
6d2010ae
A
1539 }
1540
1541 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1542 if (v4only != NULL)
1543 *v4only = 1;
1544 return (mp);
1545 }
1546
39236c6e 1547#define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y))
1c79356b 1548 /* RFC 2292 sec. 5 */
6d2010ae 1549 if ((inp->inp_flags & IN6P_PKTINFO) != 0) {
9bccf70c 1550 struct in6_pktinfo pi6;
6d2010ae 1551
39236c6e 1552 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof (struct in6_addr));
6d2010ae
A
1553 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1554 pi6.ipi6_ifindex =
1555 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1556
39236c6e
A
1557 mp = sbcreatecontrol_mbuf((caddr_t)&pi6,
1558 sizeof (struct in6_pktinfo),
1559 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO),
1560 IPPROTO_IPV6, mp);
1561 if (*mp == NULL)
1562 return (NULL);
1c79356b
A
1563 }
1564
6d2010ae 1565 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) {
9bccf70c 1566 int hlim = ip6->ip6_hlim & 0xff;
6d2010ae 1567
39236c6e 1568 mp = sbcreatecontrol_mbuf((caddr_t)&hlim, sizeof (int),
6d2010ae
A
1569 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT),
1570 IPPROTO_IPV6, mp);
39236c6e
A
1571 if (*mp == NULL)
1572 return (NULL);
1c79356b
A
1573 }
1574
6d2010ae
A
1575 if (v4only != NULL)
1576 *v4only = 0;
1577 return (mp);
1578}
1579
1580int
1581ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp)
1582{
1583 struct mbuf **np;
1584 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1585 int v4only = 0;
1586
1587 *mp = NULL;
1588 np = ip6_savecontrol_v4(in6p, m, mp, &v4only);
1589 if (np == NULL)
1590 goto no_mbufs;
b0d623f7 1591
6d2010ae
A
1592 mp = np;
1593 if (v4only)
39236c6e 1594 return (0);
b0d623f7 1595
6d2010ae
A
1596 if ((in6p->inp_flags & IN6P_TCLASS) != 0) {
1597 u_int32_t flowinfo;
1598 int tclass;
1599
1600 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1601 flowinfo >>= 20;
1602
1603 tclass = flowinfo & 0xff;
39236c6e 1604 mp = sbcreatecontrol_mbuf((caddr_t)&tclass, sizeof (tclass),
6d2010ae 1605 IPV6_TCLASS, IPPROTO_IPV6, mp);
39236c6e 1606 if (*mp == NULL)
6d2010ae
A
1607 goto no_mbufs;
1608 }
b0d623f7 1609
1c79356b 1610 /*
91447636
A
1611 * IPV6_HOPOPTS socket option. Recall that we required super-user
1612 * privilege for the option (see ip6_ctloutput), but it might be too
1613 * strict, since there might be some hop-by-hop options which can be
1614 * returned to normal user.
6d2010ae 1615 * See also RFC 2292 section 6 (or RFC 3542 section 8).
1c79356b 1616 */
6d2010ae 1617 if ((in6p->inp_flags & IN6P_HOPOPTS) != 0) {
1c79356b
A
1618 /*
1619 * Check if a hop-by-hop options header is contatined in the
1620 * received packet, and if so, store the options as ancillary
1621 * data. Note that a hop-by-hop options header must be
6d2010ae
A
1622 * just after the IPv6 header, which is assured through the
1623 * IPv6 input processing.
1c79356b 1624 */
91447636 1625 ip6 = mtod(m, struct ip6_hdr *);
1c79356b 1626 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
9bccf70c
A
1627 struct ip6_hbh *hbh;
1628 int hbhlen = 0;
1c79356b
A
1629 hbh = (struct ip6_hbh *)(ip6 + 1);
1630 hbhlen = (hbh->ip6h_len + 1) << 3;
1c79356b 1631
1c79356b 1632 /*
6d2010ae
A
1633 * XXX: We copy the whole header even if a
1634 * jumbo payload option is included, the option which
1635 * is to be removed before returning according to
1636 * RFC2292.
1637 * Note: this constraint is removed in RFC3542
1c79356b 1638 */
6d2010ae
A
1639 mp = sbcreatecontrol_mbuf((caddr_t)hbh, hbhlen,
1640 IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1641 IPPROTO_IPV6, mp);
1642
6d2010ae
A
1643 if (*mp == NULL) {
1644 goto no_mbufs;
9bccf70c 1645 }
9bccf70c
A
1646 }
1647 }
1648
6d2010ae 1649 if ((in6p->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
39236c6e 1650 int nxt = ip6->ip6_nxt, off = sizeof (struct ip6_hdr);
1c79356b
A
1651
1652 /*
1653 * Search for destination options headers or routing
1654 * header(s) through the header chain, and stores each
1655 * header as ancillary data.
1656 * Note that the order of the headers remains in
1657 * the chain of ancillary data.
1658 */
1659 while (1) { /* is explicit loop prevention necessary? */
1660 struct ip6_ext *ip6e = NULL;
1661 int elen;
1c79356b
A
1662
1663 /*
1664 * if it is not an extension header, don't try to
1665 * pull it from the chain.
1666 */
1667 switch (nxt) {
1668 case IPPROTO_DSTOPTS:
1669 case IPPROTO_ROUTING:
1670 case IPPROTO_HOPOPTS:
1671 case IPPROTO_AH: /* is it possible? */
1672 break;
1673 default:
1674 goto loopend;
1675 }
1676
39236c6e 1677 if (off + sizeof (*ip6e) > m->m_len)
1c79356b
A
1678 goto loopend;
1679 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1680 if (nxt == IPPROTO_AH)
1681 elen = (ip6e->ip6e_len + 2) << 2;
1682 else
1683 elen = (ip6e->ip6e_len + 1) << 3;
1684 if (off + elen > m->m_len)
1685 goto loopend;
1c79356b
A
1686
1687 switch (nxt) {
1688 case IPPROTO_DSTOPTS:
6d2010ae 1689 if (!(in6p->inp_flags & IN6P_DSTOPTS))
1c79356b
A
1690 break;
1691
6d2010ae 1692 mp = sbcreatecontrol_mbuf((caddr_t)ip6e, elen,
39236c6e
A
1693 IS2292(in6p, IPV6_2292DSTOPTS,
1694 IPV6_DSTOPTS), IPPROTO_IPV6, mp);
1695 if (*mp == NULL) {
6d2010ae
A
1696 goto no_mbufs;
1697 }
1c79356b 1698 break;
1c79356b 1699 case IPPROTO_ROUTING:
6d2010ae 1700 if (!in6p->inp_flags & IN6P_RTHDR)
1c79356b
A
1701 break;
1702
6d2010ae
A
1703 mp = sbcreatecontrol_mbuf((caddr_t)ip6e, elen,
1704 IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR),
1705 IPPROTO_IPV6, mp);
1706 if (*mp == NULL) {
6d2010ae
A
1707 goto no_mbufs;
1708 }
1c79356b 1709 break;
1c79356b
A
1710 case IPPROTO_HOPOPTS:
1711 case IPPROTO_AH: /* is it possible? */
1712 break;
1713
1714 default:
1715 /*
6d2010ae 1716 * other cases have been filtered in the above.
1c79356b
A
1717 * none will visit this case. here we supply
1718 * the code just in case (nxt overwritten or
1719 * other cases).
1720 */
1c79356b
A
1721 goto loopend;
1722
1723 }
1724
1725 /* proceed with the next header. */
1726 off += elen;
1727 nxt = ip6e->ip6e_nxt;
1728 ip6e = NULL;
1c79356b 1729 }
39236c6e 1730loopend:
9bccf70c 1731 ;
1c79356b 1732 }
39236c6e 1733 return (0);
6d2010ae
A
1734no_mbufs:
1735 ip6stat.ip6s_pktdropcntrl++;
1736 /* XXX increment a stat to show the failure */
39236c6e 1737 return (ENOBUFS);
6d2010ae
A
1738}
1739#undef IS2292
1740
1741void
1742ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu)
1743{
1744 struct socket *so;
1745 struct mbuf *m_mtu;
1746 struct ip6_mtuinfo mtuctl;
1747
1748 so = in6p->inp_socket;
1c79356b 1749
6d2010ae
A
1750 if (mtu == NULL)
1751 return;
1752
1753#ifdef DIAGNOSTIC
39236c6e 1754 if (so == NULL) { /* I believe this is impossible */
6d2010ae 1755 panic("ip6_notify_pmtu: socket is NULL");
39236c6e
A
1756 /* NOTREACHED */
1757 }
6d2010ae
A
1758#endif
1759
39236c6e 1760 bzero(&mtuctl, sizeof (mtuctl)); /* zero-clear for safety */
6d2010ae
A
1761 mtuctl.ip6m_mtu = *mtu;
1762 mtuctl.ip6m_addr = *dst;
316670eb 1763 if (sa6_recoverscope(&mtuctl.ip6m_addr, TRUE))
6d2010ae
A
1764 return;
1765
39236c6e 1766 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof (mtuctl),
6d2010ae
A
1767 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1768 return;
1769
39236c6e 1770 if (sbappendaddr(&so->so_rcv, SA(dst), NULL, m_mtu, NULL) == 0) {
6d2010ae
A
1771 m_freem(m_mtu);
1772 /* XXX: should count statistics */
39236c6e 1773 } else {
6d2010ae 1774 sorwakeup(so);
39236c6e 1775 }
1c79356b
A
1776}
1777
1c79356b
A
1778/*
1779 * Get pointer to the previous header followed by the header
1780 * currently processed.
1781 * XXX: This function supposes that
1782 * M includes all headers,
1783 * the next header field and the header length field of each header
1784 * are valid, and
1785 * the sum of each header length equals to OFF.
1786 * Because of these assumptions, this function must be called very
1787 * carefully. Moreover, it will not be used in the near future when
1788 * we develop `neater' mechanism to process extension headers.
1789 */
1790char *
1791ip6_get_prevhdr(m, off)
1792 struct mbuf *m;
1793 int off;
1794{
1795 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1796
39236c6e
A
1797 if (off == sizeof (struct ip6_hdr)) {
1798 return ((char *)&ip6->ip6_nxt);
1799 } else {
1c79356b
A
1800 int len, nxt;
1801 struct ip6_ext *ip6e = NULL;
1802
1803 nxt = ip6->ip6_nxt;
39236c6e 1804 len = sizeof (struct ip6_hdr);
1c79356b
A
1805 while (len < off) {
1806 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1807
9bccf70c 1808 switch (nxt) {
1c79356b 1809 case IPPROTO_FRAGMENT:
39236c6e 1810 len += sizeof (struct ip6_frag);
1c79356b
A
1811 break;
1812 case IPPROTO_AH:
1813 len += (ip6e->ip6e_len + 2) << 2;
1814 break;
1815 default:
1816 len += (ip6e->ip6e_len + 1) << 3;
1817 break;
1818 }
1819 nxt = ip6e->ip6e_nxt;
1820 }
1821 if (ip6e)
39236c6e 1822 return ((char *)&ip6e->ip6e_nxt);
1c79356b 1823 else
39236c6e 1824 return (NULL);
1c79356b
A
1825 }
1826}
1827
1828/*
1829 * get next header offset. m will be retained.
1830 */
1831int
39236c6e 1832ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
1c79356b
A
1833{
1834 struct ip6_hdr ip6;
1835 struct ip6_ext ip6e;
1836 struct ip6_frag fh;
1837
1838 /* just in case */
39236c6e 1839 VERIFY(m != NULL);
1c79356b 1840 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
39236c6e 1841 return (-1);
1c79356b
A
1842
1843 switch (proto) {
1844 case IPPROTO_IPV6:
39236c6e
A
1845 if (m->m_pkthdr.len < off + sizeof (ip6))
1846 return (-1);
1847 m_copydata(m, off, sizeof (ip6), (caddr_t)&ip6);
1c79356b
A
1848 if (nxtp)
1849 *nxtp = ip6.ip6_nxt;
39236c6e
A
1850 off += sizeof (ip6);
1851 return (off);
1c79356b
A
1852
1853 case IPPROTO_FRAGMENT:
1854 /*
1855 * terminate parsing if it is not the first fragment,
1856 * it does not make sense to parse through it.
1857 */
39236c6e
A
1858 if (m->m_pkthdr.len < off + sizeof (fh))
1859 return (-1);
1860 m_copydata(m, off, sizeof (fh), (caddr_t)&fh);
91447636
A
1861 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1862 if (fh.ip6f_offlg & IP6F_OFF_MASK)
39236c6e 1863 return (-1);
1c79356b
A
1864 if (nxtp)
1865 *nxtp = fh.ip6f_nxt;
39236c6e
A
1866 off += sizeof (struct ip6_frag);
1867 return (off);
1c79356b
A
1868
1869 case IPPROTO_AH:
39236c6e
A
1870 if (m->m_pkthdr.len < off + sizeof (ip6e))
1871 return (-1);
1872 m_copydata(m, off, sizeof (ip6e), (caddr_t)&ip6e);
1c79356b
A
1873 if (nxtp)
1874 *nxtp = ip6e.ip6e_nxt;
1875 off += (ip6e.ip6e_len + 2) << 2;
39236c6e 1876 return (off);
1c79356b
A
1877
1878 case IPPROTO_HOPOPTS:
1879 case IPPROTO_ROUTING:
1880 case IPPROTO_DSTOPTS:
39236c6e
A
1881 if (m->m_pkthdr.len < off + sizeof (ip6e))
1882 return (-1);
1883 m_copydata(m, off, sizeof (ip6e), (caddr_t)&ip6e);
1c79356b
A
1884 if (nxtp)
1885 *nxtp = ip6e.ip6e_nxt;
1886 off += (ip6e.ip6e_len + 1) << 3;
39236c6e 1887 return (off);
1c79356b
A
1888
1889 case IPPROTO_NONE:
1890 case IPPROTO_ESP:
1891 case IPPROTO_IPCOMP:
1892 /* give up */
39236c6e 1893 return (-1);
1c79356b
A
1894
1895 default:
39236c6e 1896 return (-1);
1c79356b
A
1897 }
1898
39236c6e 1899 return (-1);
1c79356b
A
1900}
1901
1902/*
1903 * get offset for the last header in the chain. m will be kept untainted.
1904 */
1905int
39236c6e 1906ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
1c79356b
A
1907{
1908 int newoff;
1909 int nxt;
1910
1911 if (!nxtp) {
1912 nxt = -1;
1913 nxtp = &nxt;
1914 }
1915 while (1) {
1916 newoff = ip6_nexthdr(m, off, proto, nxtp);
1917 if (newoff < 0)
39236c6e 1918 return (off);
1c79356b 1919 else if (newoff < off)
39236c6e 1920 return (-1); /* invalid */
1c79356b 1921 else if (newoff == off)
39236c6e 1922 return (newoff);
1c79356b
A
1923
1924 off = newoff;
1925 proto = *nxtp;
1926 }
1927}
1928
2d21ac55 1929struct ip6aux *
316670eb 1930ip6_addaux(struct mbuf *m)
9bccf70c 1931{
2d21ac55 1932 struct m_tag *tag;
316670eb 1933
2d21ac55 1934 /* Check if one is already allocated */
316670eb
A
1935 tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID,
1936 KERNEL_TAG_TYPE_INET6, NULL);
2d21ac55
A
1937 if (tag == NULL) {
1938 /* Allocate a tag */
6d2010ae
A
1939 tag = m_tag_create(KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_INET6,
1940 sizeof (struct ip6aux), M_DONTWAIT, m);
b0d623f7 1941
2d21ac55
A
1942 /* Attach it to the mbuf */
1943 if (tag) {
1944 m_tag_prepend(m, tag);
55e303ae 1945 }
9bccf70c 1946 }
316670eb
A
1947
1948 return (tag ? (struct ip6aux *)(tag + 1) : NULL);
9bccf70c
A
1949}
1950
2d21ac55 1951struct ip6aux *
316670eb 1952ip6_findaux(struct mbuf *m)
9bccf70c 1953{
2d21ac55 1954 struct m_tag *tag;
316670eb
A
1955
1956 tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID,
1957 KERNEL_TAG_TYPE_INET6, NULL);
1958
1959 return (tag ? (struct ip6aux *)(tag + 1) : NULL);
9bccf70c
A
1960}
1961
1962void
316670eb 1963ip6_delaux(struct mbuf *m)
9bccf70c 1964{
2d21ac55 1965 struct m_tag *tag;
9bccf70c 1966
316670eb
A
1967 tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID,
1968 KERNEL_TAG_TYPE_INET6, NULL);
2d21ac55
A
1969 if (tag) {
1970 m_tag_delete(m, tag);
1971 }
9bccf70c
A
1972}
1973
6d2010ae 1974/*
39236c6e 1975 * Drain callback
6d2010ae
A
1976 */
1977void
39236c6e 1978ip6_drain(void)
6d2010ae 1979{
39236c6e
A
1980 frag6_drain(); /* fragments */
1981 in6_rtqdrain(); /* protocol cloned routes */
1982 nd6_drain(NULL); /* cloned routes: ND6 */
6d2010ae
A
1983}
1984
1c79356b
A
1985/*
1986 * System control for IP6
1987 */
1988
1989u_char inet6ctlerrmap[PRC_NCMDS] = {
1990 0, 0, 0, 0,
1991 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1992 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1993 EMSGSIZE, EHOSTUNREACH, 0, 0,
1994 0, 0, 0, 0,
1995 ENOPROTOOPT
1996};