xref: /xnu-11215/bsd/netinet6/ip6_input.c (revision 8d741a5d)
1 /*
2  * Copyright (c) 2003-2024 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 #include <net/net_perf.h>
125 #include <net/if_ports_used.h>
126 #include <net/droptap.h>
127 
128 #include <netinet/in.h>
129 #include <netinet/in_systm.h>
130 #if INET
131 #include <netinet/ip.h>
132 #include <netinet/ip_icmp.h>
133 #endif /* INET */
134 #include <netinet/kpi_ipfilter_var.h>
135 #include <netinet/ip6.h>
136 #include <netinet/udp.h>
137 #include <netinet6/in6_var.h>
138 #include <netinet6/ip6_var.h>
139 #include <netinet/in_pcb.h>
140 #include <netinet/icmp6.h>
141 #include <netinet6/in6_ifattach.h>
142 #include <netinet6/nd6.h>
143 #include <netinet6/scope6_var.h>
144 #include <netinet6/ip6protosw.h>
145 
146 #if IPSEC
147 #include <netinet6/ipsec.h>
148 #include <netinet6/ipsec6.h>
149 extern int ipsec_bypass;
150 #endif /* IPSEC */
151 
152 #if DUMMYNET
153 #include <netinet/ip_dummynet.h>
154 #endif /* DUMMYNET */
155 
156 /* we need it for NLOOP. */
157 #include "loop.h"
158 
159 #if PF
160 #include <net/pfvar.h>
161 #endif /* PF */
162 
163 #include <os/log.h>
164 
165 struct ip6protosw *ip6_protox[IPPROTO_MAX];
166 
167 static LCK_GRP_DECLARE(in6_ifaddr_rwlock_grp, "in6_ifaddr_rwlock");
168 LCK_RW_DECLARE(in6_ifaddr_rwlock, &in6_ifaddr_rwlock_grp);
169 
170 /* Protected by in6_ifaddr_rwlock */
171 struct in6_ifaddrhead in6_ifaddrhead;
172 uint32_t in6_ifaddrlist_genid = 0;
173 
174 #define IN6ADDR_NHASH    61
175 u_int32_t in6addr_hashp = 0;                  /* next largest prime */
176 u_int32_t in6addr_nhash = 0;                  /* hash table size */
177 struct in6_ifaddrhashhead *__counted_by(in6addr_nhash) in6_ifaddrhashtbl = 0;
178 
179 #define IN6_IFSTAT_REQUIRE_ALIGNED_64(f)        \
180 	_CASSERT(!(offsetof(struct in6_ifstat, f) % sizeof (uint64_t)))
181 
182 #define ICMP6_IFSTAT_REQUIRE_ALIGNED_64(f)      \
183 	_CASSERT(!(offsetof(struct icmp6_ifstat, f) % sizeof (uint64_t)))
184 
185 struct ip6stat ip6stat;
186 
187 LCK_ATTR_DECLARE(ip6_mutex_attr, 0, 0);
188 LCK_GRP_DECLARE(ip6_mutex_grp, "ip6");
189 
190 LCK_MTX_DECLARE_ATTR(proxy6_lock, &ip6_mutex_grp, &ip6_mutex_attr);
191 LCK_MTX_DECLARE_ATTR(nd6_mutex_data, &ip6_mutex_grp, &ip6_mutex_attr);
192 
193 extern int loopattach_done;
194 extern void addrsel_policy_init(void);
195 
196 static int sysctl_reset_ip6_input_stats SYSCTL_HANDLER_ARGS;
197 static int sysctl_ip6_input_measure_bins SYSCTL_HANDLER_ARGS;
198 static int sysctl_ip6_input_getperf SYSCTL_HANDLER_ARGS;
199 static void ip6_init_delayed(void);
200 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *);
201 
202 static void in6_ifaddrhashtbl_init(void);
203 
204 static struct m_tag *m_tag_kalloc_inet6(u_int32_t id, u_int16_t type, uint16_t len, int wait);
205 static void m_tag_kfree_inet6(struct m_tag *tag);
206 
207 #if NSTF
208 extern void stfattach(void);
209 #endif /* NSTF */
210 
211 SYSCTL_DECL(_net_inet6_ip6);
212 
213 static uint32_t ip6_adj_clear_hwcksum = 0;
214 SYSCTL_UINT(_net_inet6_ip6, OID_AUTO, adj_clear_hwcksum,
215     CTLFLAG_RW | CTLFLAG_LOCKED, &ip6_adj_clear_hwcksum, 0,
216     "Invalidate hwcksum info when adjusting length");
217 
218 static uint32_t ip6_adj_partial_sum = 1;
219 SYSCTL_UINT(_net_inet6_ip6, OID_AUTO, adj_partial_sum,
220     CTLFLAG_RW | CTLFLAG_LOCKED, &ip6_adj_partial_sum, 0,
221     "Perform partial sum adjustment of trailing bytes at IP layer");
222 
223 static int ip6_input_measure = 0;
224 SYSCTL_PROC(_net_inet6_ip6, OID_AUTO, input_perf,
225     CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
226     &ip6_input_measure, 0, sysctl_reset_ip6_input_stats, "I", "Do time measurement");
227 
228 static uint64_t ip6_input_measure_bins = 0;
229 SYSCTL_PROC(_net_inet6_ip6, OID_AUTO, input_perf_bins,
230     CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED, &ip6_input_measure_bins, 0,
231     sysctl_ip6_input_measure_bins, "I",
232     "bins for chaining performance data histogram");
233 
234 static net_perf_t net_perf;
235 SYSCTL_PROC(_net_inet6_ip6, OID_AUTO, input_perf_data,
236     CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED,
237     0, 0, sysctl_ip6_input_getperf, "S,net_perf",
238     "IP6 input performance data (struct net_perf, net/net_perf.h)");
239 
240 /*
241  * ip6_checkinterface controls the receive side of the models for multihoming
242  * that are discussed in RFC 1122.
243  *
244  * sysctl_ip6_checkinterface values are:
245  *  IP6_CHECKINTERFACE_WEAK_ES:
246  *	This corresponds to the Weak End-System model where incoming packets from
247  *	any interface are accepted provided the destination address of the incoming packet
248  *	is assigned to some interface.
249  *
250  *  IP6_CHECKINTERFACE_HYBRID_ES:
251  *	The Hybrid End-System model use the Strong End-System for tunnel interfaces
252  *	(ipsec and utun) and the weak End-System model for other interfaces families.
253  *	This prevents a rogue middle box to probe for signs of TCP connections
254  *	that use the tunnel interface.
255  *
256  *  IP6_CHECKINTERFACE_STRONG_ES:
257  *	The Strong model model requires the packet arrived on an interface that
258  *	is assigned the destination address of the packet.
259  *
260  * Since the routing table and transmit implementation do not implement the Strong ES model,
261  * setting this to a value different from IP6_CHECKINTERFACE_WEAK_ES may lead to unexpected results.
262  *
263  * When forwarding is enabled, the system reverts to the Weak ES model as a router
264  * is expected by design to receive packets from several interfaces to the same address.
265  */
266 #define IP6_CHECKINTERFACE_WEAK_ES       0
267 #define IP6_CHECKINTERFACE_HYBRID_ES     1
268 #define IP6_CHECKINTERFACE_STRONG_ES     2
269 
270 static int ip6_checkinterface = IP6_CHECKINTERFACE_HYBRID_ES;
271 
272 static int sysctl_ip6_checkinterface SYSCTL_HANDLER_ARGS;
273 SYSCTL_PROC(_net_inet6_ip6, OID_AUTO, check_interface,
274     CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
275     0, 0, sysctl_ip6_checkinterface, "I", "Verify packet arrives on correct interface");
276 
277 #if (DEBUG || DEVELOPMENT)
278 #define IP6_CHECK_IFDEBUG 1
279 #else
280 #define IP6_CHECK_IFDEBUG 0
281 #endif /* (DEBUG || DEVELOPMENT) */
282 static int ip6_checkinterface_debug = IP6_CHECK_IFDEBUG;
283 SYSCTL_INT(_net_inet6_ip6, OID_AUTO, checkinterface_debug, CTLFLAG_RW | CTLFLAG_LOCKED,
284     &ip6_checkinterface_debug, IP6_CHECK_IFDEBUG, "");
285 
286 typedef enum ip6_check_if_result {
287 	IP6_CHECK_IF_NONE = 0,
288 	IP6_CHECK_IF_OURS = 1,
289 	IP6_CHECK_IF_DROP = 2,
290 	IP6_CHECK_IF_FORWARD = 3
291 } ip6_check_if_result_t;
292 
293 static ip6_check_if_result_t ip6_input_check_interface(struct mbuf *, struct ip6_hdr *, struct ifnet *, struct route_in6 *rin6, struct ifnet **);
294 
295 /*
296  * On platforms which require strict alignment (currently for anything but
297  * i386 or x86_64 or arm64), check if the IP header pointer is 32-bit aligned; if not,
298  * copy the contents of the mbuf chain into a new chain, and free the original
299  * one.  Create some head room in the first mbuf of the new chain, in case
300  * it's needed later on.
301  *
302  * RFC 2460 says that IPv6 headers are 64-bit aligned, but network interfaces
303  * mostly align to 32-bit boundaries.  Care should be taken never to use 64-bit
304  * load/store operations on the fields in IPv6 headers.
305  */
306 #if defined(__i386__) || defined(__x86_64__) || defined(__arm64__)
307 #define IP6_HDR_ALIGNMENT_FIXUP(_m, _ifp, _action) do { } while (0)
308 #else /* !__i386__ && !__x86_64__ && !__arm64__ */
309 #define IP6_HDR_ALIGNMENT_FIXUP(_m, _ifp, _action) do {             \
310 	if (!IP6_HDR_ALIGNED_P(mtod(_m, caddr_t))) {                    \
311 	        mbuf_ref_t _n;                                          \
312 	        ifnet_ref_t __ifp = (_ifp);                             \
313 	        os_atomic_inc(&(__ifp)->if_alignerrs, relaxed);         \
314 	        if (((_m)->m_flags & M_PKTHDR) &&                       \
315 	            (_m)->m_pkthdr.pkt_hdr != NULL)                     \
316 	                (_m)->m_pkthdr.pkt_hdr = NULL;                  \
317 	        _n = m_defrag_offset(_m, max_linkhdr, M_NOWAIT);        \
318 	        if (_n == NULL) {                                       \
319 	                ip6stat.ip6s_toosmall++;                        \
320 	                m_drop(_m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP_TOO_SMALL, NULL, 0); \
321 	                (_m) = NULL;                                    \
322 	                _action;                                        \
323 	        } else {                                                \
324 	                VERIFY(_n != (_m));                             \
325 	                (_m) = _n;                                      \
326 	        }                                                       \
327 	}                                                               \
328 } while (0)
329 #endif /* !__i386__ && !__x86_64___ && !__arm64__ */
330 
331 static void
ip6_proto_input(protocol_family_t protocol,mbuf_t packet)332 ip6_proto_input(protocol_family_t protocol, mbuf_t packet)
333 {
334 #pragma unused(protocol)
335 #if INET
336 	struct timeval start_tv;
337 	if (ip6_input_measure) {
338 		net_perf_start_time(&net_perf, &start_tv);
339 	}
340 #endif /* INET */
341 	ip6_input(packet);
342 #if INET
343 	if (ip6_input_measure) {
344 		net_perf_measure_time(&net_perf, &start_tv, 1);
345 		net_perf_histogram(&net_perf, 1);
346 	}
347 #endif /* INET */
348 }
349 
350 /*
351  * IP6 initialization: fill in IP6 protocol switch table.
352  * All protocols not implemented in kernel go to raw IP6 protocol handler.
353  */
354 void
ip6_init(struct ip6protosw * pp,struct domain * dp)355 ip6_init(struct ip6protosw *pp, struct domain *dp)
356 {
357 	static int ip6_initialized = 0;
358 	struct protosw *__single pr;
359 	struct timeval tv;
360 	int i;
361 	domain_unguard_t __single unguard;
362 
363 	domain_proto_mtx_lock_assert_held();
364 	VERIFY((pp->pr_flags & (PR_INITIALIZED | PR_ATTACHED)) == PR_ATTACHED);
365 
366 	_CASSERT((sizeof(struct ip6_hdr) +
367 	    sizeof(struct icmp6_hdr)) <= _MHLEN);
368 
369 	if (ip6_initialized) {
370 		return;
371 	}
372 	ip6_initialized = 1;
373 
374 	eventhandler_lists_ctxt_init(&in6_evhdlr_ctxt);
375 	(void)EVENTHANDLER_REGISTER(&in6_evhdlr_ctxt, in6_event,
376 	    &in6_eventhdlr_callback, eventhandler_entry_dummy_arg,
377 	    EVENTHANDLER_PRI_ANY);
378 
379 	eventhandler_lists_ctxt_init(&in6_clat46_evhdlr_ctxt);
380 	(void)EVENTHANDLER_REGISTER(&in6_clat46_evhdlr_ctxt, in6_clat46_event,
381 	    &in6_clat46_eventhdlr_callback, eventhandler_entry_dummy_arg,
382 	    EVENTHANDLER_PRI_ANY);
383 
384 	for (i = 0; i < IN6_EVENT_MAX; i++) {
385 		VERIFY(in6_event2kev_array[i].in6_event_code == i);
386 	}
387 
388 	pr = pffindproto_locked(PF_INET6, IPPROTO_RAW, SOCK_RAW);
389 	if (pr == NULL) {
390 		panic("%s: Unable to find [PF_INET6,IPPROTO_RAW,SOCK_RAW]",
391 		    __func__);
392 		/* NOTREACHED */
393 	}
394 
395 	/* Initialize the entire ip6_protox[] array to IPPROTO_RAW. */
396 	for (i = 0; i < IPPROTO_MAX; i++) {
397 		ip6_protox[i] = (struct ip6protosw *)pr;
398 	}
399 	/*
400 	 * Cycle through IP protocols and put them into the appropriate place
401 	 * in ip6_protox[], skipping protocols IPPROTO_{IP,RAW}.
402 	 */
403 	VERIFY(dp == inet6domain && dp->dom_family == PF_INET6);
404 	TAILQ_FOREACH(pr, &dp->dom_protosw, pr_entry) {
405 		VERIFY(pr->pr_domain == dp);
406 		if (pr->pr_protocol != 0 && pr->pr_protocol != IPPROTO_RAW) {
407 			/* Be careful to only index valid IP protocols. */
408 			if (pr->pr_protocol < IPPROTO_MAX) {
409 				ip6_protox[pr->pr_protocol] =
410 				    (struct ip6protosw *)pr;
411 			}
412 		}
413 	}
414 
415 	TAILQ_INIT(&in6_ifaddrhead);
416 	in6_ifaddrhashtbl_init();
417 
418 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_receive);
419 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_hdrerr);
420 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_toobig);
421 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_noroute);
422 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_addrerr);
423 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_protounknown);
424 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_truncated);
425 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_discard);
426 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_deliver);
427 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_forward);
428 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_request);
429 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_discard);
430 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_fragok);
431 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_fragfail);
432 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_fragcreat);
433 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_reass_reqd);
434 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_reass_ok);
435 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_reass_fail);
436 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_mcast);
437 	IN6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_mcast);
438 
439 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_msg);
440 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_error);
441 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_dstunreach);
442 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_adminprohib);
443 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_timeexceed);
444 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_paramprob);
445 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_pkttoobig);
446 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_echo);
447 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_echoreply);
448 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_routersolicit);
449 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_routeradvert);
450 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_neighborsolicit);
451 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_neighboradvert);
452 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_redirect);
453 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_mldquery);
454 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_mldreport);
455 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_in_mlddone);
456 
457 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_msg);
458 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_error);
459 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_dstunreach);
460 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_adminprohib);
461 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_timeexceed);
462 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_paramprob);
463 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_pkttoobig);
464 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_echo);
465 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_echoreply);
466 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_routersolicit);
467 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_routeradvert);
468 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_neighborsolicit);
469 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_neighboradvert);
470 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_redirect);
471 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_mldquery);
472 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_mldreport);
473 	ICMP6_IFSTAT_REQUIRE_ALIGNED_64(ifs6_out_mlddone);
474 
475 	getmicrotime(&tv);
476 	ip6_desync_factor =
477 	    (RandomULong() ^ tv.tv_usec) % MAX_TEMP_DESYNC_FACTOR;
478 
479 	PE_parse_boot_argn("in6_embedded_scope", &in6_embedded_scope, sizeof(in6_embedded_scope));
480 	PE_parse_boot_argn("ip6_checkinterface", &i, sizeof(i));
481 	switch (i) {
482 	case IP6_CHECKINTERFACE_WEAK_ES:
483 	case IP6_CHECKINTERFACE_HYBRID_ES:
484 	case IP6_CHECKINTERFACE_STRONG_ES:
485 		ip6_checkinterface = i;
486 		break;
487 	default:
488 		break;
489 	}
490 
491 	in6_ifaddr_init();
492 	ip6_moptions_init();
493 	nd6_init();
494 	frag6_init();
495 	icmp6_init(NULL, dp);
496 	addrsel_policy_init();
497 
498 	/*
499 	 * P2P interfaces often route the local address to the loopback
500 	 * interface. At this point, lo0 hasn't been initialized yet, which
501 	 * means that we need to delay the IPv6 configuration of lo0.
502 	 */
503 	net_init_add(ip6_init_delayed);
504 
505 	unguard = domain_unguard_deploy();
506 	i = proto_register_input(PF_INET6, ip6_proto_input, NULL, 0);
507 	if (i != 0) {
508 		panic("%s: failed to register PF_INET6 protocol: %d",
509 		    __func__, i);
510 		/* NOTREACHED */
511 	}
512 	domain_unguard_release(unguard);
513 }
514 
515 static void
ip6_init_delayed(void)516 ip6_init_delayed(void)
517 {
518 	(void) in6_ifattach_prelim(lo_ifp);
519 
520 	/* timer for regeneranation of temporary addresses randomize ID */
521 	timeout(in6_tmpaddrtimer, NULL,
522 	    (ip6_temp_preferred_lifetime - ip6_desync_factor -
523 	    ip6_temp_regen_advance) * hz);
524 
525 #if NSTF
526 	stfattach();
527 #endif /* NSTF */
528 }
529 
530 static void
ip6_input_adjust(struct mbuf * m,struct ip6_hdr * ip6,uint32_t plen,struct ifnet * inifp)531 ip6_input_adjust(struct mbuf *m, struct ip6_hdr *ip6, uint32_t plen,
532     struct ifnet *inifp)
533 {
534 	boolean_t adjust = TRUE;
535 	uint32_t tot_len = sizeof(*ip6) + plen;
536 
537 	ASSERT(m_pktlen(m) > tot_len);
538 
539 	/*
540 	 * Invalidate hardware checksum info if ip6_adj_clear_hwcksum
541 	 * is set; useful to handle buggy drivers.  Note that this
542 	 * should not be enabled by default, as we may get here due
543 	 * to link-layer padding.
544 	 */
545 	if (ip6_adj_clear_hwcksum &&
546 	    (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) &&
547 	    !(inifp->if_flags & IFF_LOOPBACK) &&
548 	    !(m->m_pkthdr.pkt_flags & PKTF_LOOP)) {
549 		m->m_pkthdr.csum_flags &= ~CSUM_DATA_VALID;
550 		m->m_pkthdr.csum_data = 0;
551 		ip6stat.ip6s_adj_hwcsum_clr++;
552 	}
553 
554 	/*
555 	 * If partial checksum information is available, subtract
556 	 * out the partial sum of postpended extraneous bytes, and
557 	 * update the checksum metadata accordingly.  By doing it
558 	 * here, the upper layer transport only needs to adjust any
559 	 * prepended extraneous bytes (else it will do both.)
560 	 */
561 	if (ip6_adj_partial_sum &&
562 	    (m->m_pkthdr.csum_flags & (CSUM_DATA_VALID | CSUM_PARTIAL)) ==
563 	    (CSUM_DATA_VALID | CSUM_PARTIAL)) {
564 		m->m_pkthdr.csum_rx_val = m_adj_sum16(m,
565 		    m->m_pkthdr.csum_rx_start, m->m_pkthdr.csum_rx_start,
566 		    (tot_len - m->m_pkthdr.csum_rx_start),
567 		    m->m_pkthdr.csum_rx_val);
568 	} else if ((m->m_pkthdr.csum_flags &
569 	    (CSUM_DATA_VALID | CSUM_PARTIAL)) ==
570 	    (CSUM_DATA_VALID | CSUM_PARTIAL)) {
571 		/*
572 		 * If packet has partial checksum info and we decided not
573 		 * to subtract the partial sum of postpended extraneous
574 		 * bytes here (not the default case), leave that work to
575 		 * be handled by the other layers.  For now, only TCP, UDP
576 		 * layers are capable of dealing with this.  For all other
577 		 * protocols (including fragments), trim and ditch the
578 		 * partial sum as those layers might not implement partial
579 		 * checksumming (or adjustment) at all.
580 		 */
581 		if (ip6->ip6_nxt == IPPROTO_TCP ||
582 		    ip6->ip6_nxt == IPPROTO_UDP) {
583 			adjust = FALSE;
584 		} else {
585 			m->m_pkthdr.csum_flags &= ~CSUM_DATA_VALID;
586 			m->m_pkthdr.csum_data = 0;
587 			ip6stat.ip6s_adj_hwcsum_clr++;
588 		}
589 	}
590 
591 	if (adjust) {
592 		ip6stat.ip6s_adj++;
593 		if (m->m_len == m->m_pkthdr.len) {
594 			m->m_len = tot_len;
595 			m->m_pkthdr.len = tot_len;
596 		} else {
597 			m_adj(m, tot_len - m->m_pkthdr.len);
598 		}
599 	}
600 }
601 static ip6_check_if_result_t
ip6_input_check_interface(struct mbuf * m,struct ip6_hdr * ip6,struct ifnet * inifp,struct route_in6 * rin6,struct ifnet ** deliverifp)602 ip6_input_check_interface(struct mbuf *m, struct ip6_hdr *ip6, struct ifnet *inifp, struct route_in6 *rin6, struct ifnet **deliverifp)
603 {
604 	struct in6_ifaddr *__single ia6 = NULL;
605 	struct in6_addr tmp_dst = ip6->ip6_dst; /* copy to avoid unaligned access */
606 	struct in6_ifaddr *__single best_ia6 = NULL;
607 	uint32_t dst_ifscope = IFSCOPE_NONE;
608 	ip6_check_if_result_t result = IP6_CHECK_IF_NONE;
609 
610 	*deliverifp = NULL;
611 
612 	if (m->m_pkthdr.pkt_flags & PKTF_IFAINFO) {
613 		dst_ifscope = m->m_pkthdr.dst_ifindex;
614 	} else {
615 		dst_ifscope = inifp->if_index;
616 	}
617 	/*
618 	 * Check for exact addresses in the hash bucket.
619 	 */
620 	lck_rw_lock_shared(&in6_ifaddr_rwlock);
621 	TAILQ_FOREACH(ia6, IN6ADDR_HASH(&tmp_dst), ia6_hash) {
622 		/*
623 		 * TODO: should we accept loopback
624 		 */
625 		if (in6_are_addr_equal_scoped(&ia6->ia_addr.sin6_addr, &tmp_dst, ia6->ia_ifp->if_index, dst_ifscope)) {
626 			if ((ia6->ia6_flags & (IN6_IFF_NOTREADY | IN6_IFF_CLAT46))) {
627 				continue;
628 			}
629 			best_ia6 = ia6;
630 			if (ia6->ia_ifp == inifp) {
631 				/*
632 				 * TODO: should we also accept locally originated packets
633 				 * or from loopback ???
634 				 */
635 				break;
636 			}
637 			/*
638 			 * Continue the loop in case there's a exact match with another
639 			 * interface
640 			 */
641 		}
642 	}
643 	if (best_ia6 != NULL) {
644 		if (best_ia6->ia_ifp != inifp && ip6_forwarding == 0 &&
645 		    ((ip6_checkinterface == IP6_CHECKINTERFACE_HYBRID_ES &&
646 		    (best_ia6->ia_ifp->if_family == IFNET_FAMILY_IPSEC ||
647 		    best_ia6->ia_ifp->if_family == IFNET_FAMILY_UTUN)) ||
648 		    ip6_checkinterface == IP6_CHECKINTERFACE_STRONG_ES)) {
649 			/*
650 			 * Drop when interface address check is strict and forwarding
651 			 * is disabled
652 			 */
653 			result = IP6_CHECK_IF_DROP;
654 		} else {
655 			result = IP6_CHECK_IF_OURS;
656 			*deliverifp = best_ia6->ia_ifp;
657 			ip6_setdstifaddr_info(m, 0, best_ia6);
658 			ip6_setsrcifaddr_info(m, best_ia6->ia_ifp->if_index, NULL);
659 		}
660 	}
661 	lck_rw_done(&in6_ifaddr_rwlock);
662 
663 	if (result == IP6_CHECK_IF_NONE) {
664 		/*
665 		 * Slow path: route lookup.
666 		 */
667 		struct sockaddr_in6 *__single dst6;
668 
669 		dst6 = SIN6(&rin6->ro_dst);
670 		dst6->sin6_len = sizeof(struct sockaddr_in6);
671 		dst6->sin6_family = AF_INET6;
672 		dst6->sin6_addr = ip6->ip6_dst;
673 		if (!in6_embedded_scope && IN6_IS_SCOPE_EMBED(&ip6->ip6_dst)) {
674 			dst6->sin6_scope_id = dst_ifscope;
675 		}
676 		rtalloc_scoped_ign((struct route *)rin6,
677 		    RTF_PRCLONING, IFSCOPE_NONE);
678 		if (rin6->ro_rt != NULL) {
679 			RT_LOCK_SPIN(rin6->ro_rt);
680 		}
681 
682 #define rt6_key(r) (SIN6(rn_get_key((r)->rt_nodes)))
683 
684 		/*
685 		 * Accept the packet if the forwarding interface to the destination
686 		 * according to the routing table is the loopback interface,
687 		 * unless the associated route has a gateway.
688 		 * Note that this approach causes to accept a packet if there is a
689 		 * route to the loopback interface for the destination of the packet.
690 		 * But we think it's even useful in some situations, e.g. when using
691 		 * a special daemon which wants to intercept the packet.
692 		 *
693 		 * XXX: some OSes automatically make a cloned route for the destination
694 		 * of an outgoing packet.  If the outgoing interface of the packet
695 		 * is a loopback one, the kernel would consider the packet to be
696 		 * accepted, even if we have no such address assinged on the interface.
697 		 * We check the cloned flag of the route entry to reject such cases,
698 		 * assuming that route entries for our own addresses are not made by
699 		 * cloning (it should be true because in6_addloop explicitly installs
700 		 * the host route).  However, we might have to do an explicit check
701 		 * while it would be less efficient.  Or, should we rather install a
702 		 * reject route for such a case?
703 		 */
704 		if (rin6->ro_rt != NULL &&
705 		    (rin6->ro_rt->rt_flags & (RTF_HOST | RTF_GATEWAY)) == RTF_HOST &&
706 #if RTF_WASCLONED
707 		    !(rin6->ro_rt->rt_flags & RTF_WASCLONED) &&
708 #endif
709 		    rin6->ro_rt->rt_ifp->if_type == IFT_LOOP) {
710 			ia6 = ifatoia6(rin6->ro_rt->rt_ifa);
711 			/*
712 			 * Packets to a tentative, duplicated, or somehow invalid
713 			 * address must not be accepted.
714 			 *
715 			 * For performance, test without acquiring the address lock;
716 			 * a lot of things in the address are set once and never
717 			 * changed (e.g. ia_ifp.)
718 			 */
719 			if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
720 				/* this address is ready */
721 				result = IP6_CHECK_IF_OURS;
722 				*deliverifp = ia6->ia_ifp;       /* correct? */
723 				/*
724 				 * record dst address information into mbuf.
725 				 */
726 				(void) ip6_setdstifaddr_info(m, 0, ia6);
727 				(void) ip6_setsrcifaddr_info(m, ia6->ia_ifp->if_index, NULL);
728 			}
729 		}
730 
731 		if (rin6->ro_rt != NULL) {
732 			RT_UNLOCK(rin6->ro_rt);
733 		}
734 	}
735 
736 	if (result == IP6_CHECK_IF_NONE) {
737 		if (ip6_forwarding == 0) {
738 			result = IP6_CHECK_IF_DROP;
739 		} else {
740 			result = IP6_CHECK_IF_FORWARD;
741 			ip6_setdstifaddr_info(m, inifp->if_index, NULL);
742 			ip6_setsrcifaddr_info(m, inifp->if_index, NULL);
743 		}
744 	}
745 
746 	if (result == IP6_CHECK_IF_OURS && *deliverifp != inifp) {
747 		ASSERT(*deliverifp != NULL);
748 		ip6stat.ip6s_rcv_if_weak_match++;
749 
750 		/*  Logging is too noisy when forwarding is enabled */
751 		if (ip6_checkinterface_debug != IP6_CHECKINTERFACE_WEAK_ES && ip6_forwarding != 0) {
752 			char src_str[MAX_IPv6_STR_LEN];
753 			char dst_str[MAX_IPv6_STR_LEN];
754 
755 			inet_ntop(AF_INET6, &ip6->ip6_src, src_str, sizeof(src_str));
756 			inet_ntop(AF_INET6, &ip6->ip6_dst, dst_str, sizeof(dst_str));
757 			os_log_info(OS_LOG_DEFAULT,
758 			    "%s: weak ES interface match to %s for packet from %s to %s proto %u received via %s",
759 			    __func__, (*deliverifp)->if_xname, src_str, dst_str, ip6->ip6_nxt, inifp->if_xname);
760 		}
761 	} else if (result == IP6_CHECK_IF_DROP) {
762 		ip6stat.ip6s_rcv_if_no_match++;
763 		if (ip6_checkinterface_debug > 0) {
764 			char src_str[MAX_IPv6_STR_LEN];
765 			char dst_str[MAX_IPv6_STR_LEN];
766 
767 			inet_ntop(AF_INET6, &ip6->ip6_src, src_str, sizeof(src_str));
768 			inet_ntop(AF_INET6, &ip6->ip6_dst, dst_str, sizeof(dst_str));
769 			os_log(OS_LOG_DEFAULT,
770 			    "%s: no interface match for packet from %s to %s proto %u received via %s",
771 			    __func__, src_str, dst_str, ip6->ip6_nxt, inifp->if_xname);
772 		}
773 	}
774 
775 	return result;
776 }
777 
778 void
ip6_input(struct mbuf * m)779 ip6_input(struct mbuf *m)
780 {
781 	struct ip6_hdr *ip6;
782 	int off = sizeof(struct ip6_hdr), nest;
783 	u_int32_t plen;
784 	u_int32_t rtalert = ~0;
785 	int nxt = 0, ours = 0;
786 	ifnet_ref_t inifp, deliverifp = NULL;
787 	ipfilter_t __single inject_ipfref = NULL;
788 	int seen = 1;
789 #if DUMMYNET
790 	struct m_tag *__single tag;
791 	struct ip_fw_args args = {};
792 #endif /* DUMMYNET */
793 	struct route_in6 rin6 = {};
794 
795 	/*
796 	 * Check if the packet we received is valid after interface filter
797 	 * processing
798 	 */
799 	MBUF_INPUT_CHECK(m, m->m_pkthdr.rcvif);
800 	inifp = m->m_pkthdr.rcvif;
801 	VERIFY(inifp != NULL);
802 
803 	/* Perform IP header alignment fixup, if needed */
804 	IP6_HDR_ALIGNMENT_FIXUP(m, inifp, return );
805 
806 	m->m_pkthdr.pkt_flags &= ~PKTF_FORWARDED;
807 #if IPSEC
808 	/*
809 	 * should the inner packet be considered authentic?
810 	 * see comment in ah4_input().
811 	 */
812 	m->m_flags &= ~M_AUTHIPHDR;
813 	m->m_flags &= ~M_AUTHIPDGM;
814 #endif /* IPSEC */
815 
816 	/*
817 	 * make sure we don't have onion peering information into m_aux.
818 	 */
819 	ip6_delaux(m);
820 
821 #if DUMMYNET
822 	if ((tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID,
823 	    KERNEL_TAG_TYPE_DUMMYNET)) != NULL) {
824 		struct dn_pkt_tag *__single dn_tag;
825 
826 		dn_tag = (struct dn_pkt_tag *)(tag->m_tag_data);
827 
828 		args.fwa_pf_rule = dn_tag->dn_pf_rule;
829 
830 		m_tag_delete(m, tag);
831 	}
832 
833 	if (args.fwa_pf_rule) {
834 		ip6 = mtod(m, struct ip6_hdr *); /* In case PF got disabled */
835 
836 		goto check_with_pf;
837 	}
838 #endif /* DUMMYNET */
839 
840 	/*
841 	 * No need to process packet twice if we've already seen it.
842 	 */
843 	inject_ipfref = ipf_get_inject_filter(m);
844 	if (inject_ipfref != NULL) {
845 		ip6 = mtod(m, struct ip6_hdr *);
846 		nxt = ip6->ip6_nxt;
847 		seen = 0;
848 		goto injectit;
849 	} else {
850 		seen = 1;
851 	}
852 
853 	if (__improbable(m->m_pkthdr.pkt_flags & PKTF_WAKE_PKT)) {
854 		if_ports_used_match_mbuf(inifp, PF_INET6, m);
855 	}
856 
857 	/*
858 	 * mbuf statistics
859 	 */
860 	if (m->m_flags & M_EXT) {
861 		if (m->m_next != NULL) {
862 			ip6stat.ip6s_mext2m++;
863 		} else {
864 			ip6stat.ip6s_mext1++;
865 		}
866 	} else {
867 #define M2MMAX  (sizeof (ip6stat.ip6s_m2m) / sizeof (ip6stat.ip6s_m2m[0]))
868 		if (m->m_next != NULL) {
869 			if (m->m_pkthdr.pkt_flags & PKTF_LOOP) {
870 				/* XXX */
871 				ip6stat.ip6s_m2m[ifnet_index(lo_ifp)]++;
872 			} else if (inifp->if_index < M2MMAX) {
873 				ip6stat.ip6s_m2m[inifp->if_index]++;
874 			} else {
875 				ip6stat.ip6s_m2m[0]++;
876 			}
877 		} else {
878 			ip6stat.ip6s_m1++;
879 		}
880 #undef M2MMAX
881 	}
882 
883 	/*
884 	 * Drop the packet if IPv6 operation is disabled on the interface.
885 	 */
886 	if (inifp->if_eflags & IFEF_IPV6_DISABLED) {
887 		m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP6_IF_IPV6_DISABLED, NULL, 0);
888 		goto bad;
889 	}
890 
891 	in6_ifstat_inc_na(inifp, ifs6_in_receive);
892 	ip6stat.ip6s_total++;
893 
894 	/*
895 	 * L2 bridge code and some other code can return mbuf chain
896 	 * that does not conform to KAME requirement.  too bad.
897 	 * XXX: fails to join if interface MTU > MCLBYTES.  jumbogram?
898 	 */
899 	if (m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
900 		mbuf_ref_t n;
901 
902 		MGETHDR(n, M_DONTWAIT, MT_HEADER);      /* MAC-OK */
903 		if (n) {
904 			M_COPY_PKTHDR(n, m);
905 		}
906 		if (n && m->m_pkthdr.len > MHLEN) {
907 			MCLGET(n, M_DONTWAIT);
908 			if ((n->m_flags & M_EXT) == 0) {
909 				m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP_TOO_SMALL, NULL, 0);
910 				n = NULL;
911 			}
912 		}
913 		if (n == NULL) {
914 			goto bad;
915 		}
916 
917 		m_copydata(m, 0, m->m_pkthdr.len, mtod(n, caddr_t));
918 		n->m_len = m->m_pkthdr.len;
919 		m_freem(m);
920 		m = n;
921 	}
922 	IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), { goto done; });
923 
924 	if (m->m_len < sizeof(struct ip6_hdr)) {
925 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) {
926 			ip6stat.ip6s_toosmall++;
927 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
928 			goto done;
929 		}
930 	}
931 
932 	ip6 = mtod(m, struct ip6_hdr *);
933 
934 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
935 		ip6stat.ip6s_badvers++;
936 		in6_ifstat_inc(inifp, ifs6_in_hdrerr);
937 		m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP_BAD_VERSION, NULL, 0);
938 		goto bad;
939 	}
940 
941 	ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
942 
943 	/*
944 	 * Check against address spoofing/corruption.
945 	 */
946 	if (!(m->m_pkthdr.pkt_flags & PKTF_LOOP) &&
947 	    IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src)) {
948 		ip6stat.ip6s_badscope++;
949 		in6_ifstat_inc(inifp, ifs6_in_addrerr);
950 		m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP6_BAD_SCOPE, NULL, 0);
951 		goto bad;
952 	}
953 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
954 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
955 		/*
956 		 * XXX: "badscope" is not very suitable for a multicast source.
957 		 */
958 		ip6stat.ip6s_badscope++;
959 		in6_ifstat_inc(inifp, ifs6_in_addrerr);
960 		m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP6_BAD_SCOPE, NULL, 0);
961 		goto bad;
962 	}
963 	if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) &&
964 	    !(m->m_pkthdr.pkt_flags & PKTF_LOOP)) {
965 		/*
966 		 * In this case, the packet should come from the loopback
967 		 * interface.  However, we cannot just check the if_flags,
968 		 * because ip6_mloopback() passes the "actual" interface
969 		 * as the outgoing/incoming interface.
970 		 */
971 		ip6stat.ip6s_badscope++;
972 		in6_ifstat_inc(inifp, ifs6_in_addrerr);
973 		m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP6_BAD_SCOPE, NULL, 0);
974 		goto bad;
975 	}
976 
977 	/*
978 	 * The following check is not documented in specs.  A malicious
979 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
980 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
981 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
982 	 *
983 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
984 	 * support IPv4-less kernel compilation, we cannot support SIIT
985 	 * environment at all.  So, it makes more sense for us to reject any
986 	 * malicious packets for non-SIIT environment, than try to do a
987 	 * partial support for SIIT environment.
988 	 */
989 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
990 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
991 		ip6stat.ip6s_badscope++;
992 		in6_ifstat_inc(inifp, ifs6_in_addrerr);
993 		m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP6_BAD_SCOPE, NULL, 0);
994 		goto bad;
995 	}
996 
997 	if (((ntohl(ip6->ip6_flow & IPV6_FLOW_ECN_MASK) >> 20) & IPTOS_ECN_ECT1) == IPTOS_ECN_ECT1) {
998 		m->m_pkthdr.pkt_ext_flags |= PKTF_EXT_L4S;
999 	}
1000 
1001 #if 0
1002 	/*
1003 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
1004 	 *
1005 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
1006 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
1007 	 * is revised to forbid relaying case.
1008 	 */
1009 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
1010 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
1011 		ip6stat.ip6s_badscope++;
1012 		in6_ifstat_inc(inifp, ifs6_in_addrerr);
1013 		m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP6_BAD_SCOPE, NULL, 0);
1014 		goto bad;
1015 	}
1016 #endif
1017 
1018 	/*
1019 	 * Naively assume we can attribute inbound data to the route we would
1020 	 * use to send to this destination. Asymetric routing breaks this
1021 	 * assumption, but it still allows us to account for traffic from
1022 	 * a remote node in the routing table.
1023 	 * this has a very significant performance impact so we bypass
1024 	 * if nstat_collect is disabled. We may also bypass if the
1025 	 * protocol is tcp in the future because tcp will have a route that
1026 	 * we can use to attribute the data to. That does mean we would not
1027 	 * account for forwarded tcp traffic.
1028 	 */
1029 	if (nstat_collect) {
1030 		rtentry_ref_t rte =
1031 		    ifnet_cached_rtlookup_inet6(inifp, &ip6->ip6_src);
1032 		if (rte != NULL) {
1033 			nstat_route_rx(rte, 1, m->m_pkthdr.len, 0);
1034 			rtfree(rte);
1035 		}
1036 	}
1037 
1038 #if DUMMYNET
1039 check_with_pf:
1040 #endif /* DUMMYNET */
1041 #if PF
1042 	/* Invoke inbound packet filter */
1043 	if (PF_IS_ENABLED) {
1044 		int error;
1045 #if DUMMYNET
1046 		error = pf_af_hook(inifp, NULL, &m, AF_INET6, TRUE, &args);
1047 #else /* !DUMMYNET */
1048 		error = pf_af_hook(inifp, NULL, &m, AF_INET6, TRUE, NULL);
1049 #endif /* !DUMMYNET */
1050 		if (error != 0 || m == NULL) {
1051 			if (m != NULL) {
1052 				panic("%s: unexpected packet %p",
1053 				    __func__, m);
1054 				/* NOTREACHED */
1055 			}
1056 			/* Already freed by callee */
1057 			goto done;
1058 		}
1059 		ip6 = mtod(m, struct ip6_hdr *);
1060 	}
1061 #endif /* PF */
1062 
1063 	/* drop packets if interface ID portion is already filled */
1064 	if (!(inifp->if_flags & IFF_LOOPBACK) &&
1065 	    !(m->m_pkthdr.pkt_flags & PKTF_LOOP)) {
1066 		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) &&
1067 		    ip6->ip6_src.s6_addr16[1]) {
1068 			ip6stat.ip6s_badscope++;
1069 			m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP6_BAD_SCOPE, NULL, 0);
1070 			goto bad;
1071 		}
1072 		if (IN6_IS_SCOPE_EMBED(&ip6->ip6_dst) &&
1073 		    ip6->ip6_dst.s6_addr16[1]) {
1074 			ip6stat.ip6s_badscope++;
1075 			m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP6_BAD_SCOPE, NULL, 0);
1076 			goto bad;
1077 		}
1078 	}
1079 
1080 	if ((m->m_pkthdr.pkt_flags & PKTF_IFAINFO) != 0 && in6_embedded_scope) {
1081 		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) {
1082 			ip6->ip6_src.s6_addr16[1] =
1083 			    htons(m->m_pkthdr.src_ifindex);
1084 		}
1085 		if (IN6_IS_SCOPE_EMBED(&ip6->ip6_dst)) {
1086 			ip6->ip6_dst.s6_addr16[1] =
1087 			    htons(m->m_pkthdr.dst_ifindex);
1088 		}
1089 	} else if (in6_embedded_scope) {
1090 		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src)) {
1091 			ip6->ip6_src.s6_addr16[1] = htons(inifp->if_index);
1092 		}
1093 		if (IN6_IS_SCOPE_EMBED(&ip6->ip6_dst)) {
1094 			ip6->ip6_dst.s6_addr16[1] = htons(inifp->if_index);
1095 		}
1096 	}
1097 
1098 	/*
1099 	 * Multicast check
1100 	 */
1101 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
1102 		struct in6_multi *__single in6m = NULL;
1103 
1104 		in6_ifstat_inc_na(inifp, ifs6_in_mcast);
1105 		/*
1106 		 * See if we belong to the destination multicast group on the
1107 		 * arrival interface.
1108 		 */
1109 		in6_multihead_lock_shared();
1110 		IN6_LOOKUP_MULTI(&ip6->ip6_dst, inifp, in6m);
1111 		in6_multihead_lock_done();
1112 		if (in6m != NULL) {
1113 			IN6M_REMREF(in6m);
1114 			ours = 1;
1115 		} else if (!nd6_prproxy) {
1116 			ip6stat.ip6s_notmember++;
1117 			ip6stat.ip6s_cantforward++;
1118 			in6_ifstat_inc(inifp, ifs6_in_discard);
1119 			m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP_UNKNOWN_MULTICAST_GROUP, NULL, 0);
1120 			goto bad;
1121 		}
1122 		deliverifp = inifp;
1123 		/*
1124 		 * record dst address information into mbuf, if we don't have one yet.
1125 		 * note that we are unable to record it, if the address is not listed
1126 		 * as our interface address (e.g. multicast addresses, etc.)
1127 		 */
1128 		if (deliverifp != NULL) {
1129 			struct in6_ifaddr *__single ia6 = NULL;
1130 
1131 			ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
1132 			if (ia6 != NULL) {
1133 				(void) ip6_setdstifaddr_info(m, 0, ia6);
1134 				(void) ip6_setsrcifaddr_info(m, ia6->ia_ifp->if_index, NULL);
1135 				ifa_remref(&ia6->ia_ifa);
1136 			} else {
1137 				(void) ip6_setdstifaddr_info(m, inifp->if_index, NULL);
1138 				(void) ip6_setsrcifaddr_info(m, inifp->if_index, NULL);
1139 			}
1140 		}
1141 		goto hbhcheck;
1142 	} else {
1143 		/*
1144 		 * Unicast check
1145 		 */
1146 		ip6_check_if_result_t check_if_result = IP6_CHECK_IF_NONE;
1147 		check_if_result = ip6_input_check_interface(m, ip6, inifp, &rin6, &deliverifp);
1148 		ASSERT(check_if_result != IP6_CHECK_IF_NONE);
1149 		if (check_if_result == IP6_CHECK_IF_OURS) {
1150 			ours = 1;
1151 			goto hbhcheck;
1152 		} else if (check_if_result == IP6_CHECK_IF_DROP) {
1153 			m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP_RCV_IF_NO_MATCH, NULL, 0);
1154 			goto bad;
1155 		}
1156 	}
1157 
1158 	/*
1159 	 * Now there is no reason to process the packet if it's not our own
1160 	 * and we're not a router.
1161 	 */
1162 	if (!ip6_forwarding) {
1163 		ip6stat.ip6s_cantforward++;
1164 		in6_ifstat_inc(inifp, ifs6_in_discard);
1165 		/*
1166 		 * Raise a kernel event if the packet received on cellular
1167 		 * interface is not intended for local host.
1168 		 * For now limit it to ICMPv6 packets.
1169 		 */
1170 		if (inifp->if_type == IFT_CELLULAR &&
1171 		    ip6->ip6_nxt == IPPROTO_ICMPV6) {
1172 			in6_ifstat_inc(inifp, ifs6_cantfoward_icmp6);
1173 		}
1174 		m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP_CANNOT_FORWARD, NULL, 0);
1175 		goto bad;
1176 	}
1177 
1178 hbhcheck:
1179 	/*
1180 	 * Process Hop-by-Hop options header if it's contained.
1181 	 * m may be modified in ip6_hopopts_input().
1182 	 * If a JumboPayload option is included, plen will also be modified.
1183 	 */
1184 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
1185 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1186 		struct ip6_hbh *__single hbh;
1187 
1188 		/*
1189 		 * Mark the packet to imply that HBH option has been checked.
1190 		 * This can only be true is the packet came in unfragmented
1191 		 * or if the option is in the first fragment
1192 		 */
1193 		m->m_pkthdr.pkt_flags |= PKTF_HBH_CHKED;
1194 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
1195 #if 0   /* touches NULL pointer */
1196 			in6_ifstat_inc(inifp, ifs6_in_discard);
1197 #endif
1198 			goto done;      /* m have already been freed */
1199 		}
1200 
1201 		/* adjust pointer */
1202 		ip6 = mtod(m, struct ip6_hdr *);
1203 
1204 		/*
1205 		 * if the payload length field is 0 and the next header field
1206 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
1207 		 * option MUST be included.
1208 		 */
1209 		if (ip6->ip6_plen == 0 && plen == 0) {
1210 			/*
1211 			 * Note that if a valid jumbo payload option is
1212 			 * contained, ip6_hopopts_input() must set a valid
1213 			 * (non-zero) payload length to the variable plen.
1214 			 */
1215 			ip6stat.ip6s_badoptions++;
1216 			in6_ifstat_inc(inifp, ifs6_in_discard);
1217 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
1218 			icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER,
1219 			    (int)((caddr_t)&ip6->ip6_plen - (caddr_t)ip6));
1220 			goto done;
1221 		}
1222 		/* ip6_hopopts_input() ensures that mbuf is contiguous */
1223 		hbh = (struct ip6_hbh *)(ip6 + 1);
1224 		nxt = hbh->ip6h_nxt;
1225 
1226 		/*
1227 		 * If we are acting as a router and the packet contains a
1228 		 * router alert option, see if we know the option value.
1229 		 * Currently, we only support the option value for MLD, in which
1230 		 * case we should pass the packet to the multicast routing
1231 		 * daemon.
1232 		 */
1233 		if (rtalert != ~0 && ip6_forwarding) {
1234 			switch (rtalert) {
1235 			case IP6OPT_RTALERT_MLD:
1236 				ours = 1;
1237 				break;
1238 			default:
1239 				/*
1240 				 * RFC2711 requires unrecognized values must be
1241 				 * silently ignored.
1242 				 */
1243 				break;
1244 			}
1245 		}
1246 	} else {
1247 		nxt = ip6->ip6_nxt;
1248 	}
1249 
1250 	/*
1251 	 * Check that the amount of data in the buffers
1252 	 * is as at least much as the IPv6 header would have us expect.
1253 	 * Trim mbufs if longer than we expect.
1254 	 * Drop packet if shorter than we expect.
1255 	 */
1256 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
1257 		ip6stat.ip6s_tooshort++;
1258 		in6_ifstat_inc(inifp, ifs6_in_truncated);
1259 		m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP_TOO_SHORT, NULL, 0);
1260 		goto bad;
1261 	}
1262 
1263 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
1264 		ip6_input_adjust(m, ip6, plen, inifp);
1265 	}
1266 
1267 	/*
1268 	 * Forward if desirable.
1269 	 */
1270 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
1271 		if (!ours && nd6_prproxy) {
1272 			/*
1273 			 * If this isn't for us, this might be a Neighbor
1274 			 * Solicitation (dst is solicited-node multicast)
1275 			 * against an address in one of the proxied prefixes;
1276 			 * if so, claim the packet and let icmp6_input()
1277 			 * handle the rest.
1278 			 */
1279 			ours = nd6_prproxy_isours(m, ip6, NULL, IFSCOPE_NONE);
1280 			VERIFY(!ours ||
1281 			    (m->m_pkthdr.pkt_flags & PKTF_PROXY_DST));
1282 		}
1283 		if (!ours) {
1284 			m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP6_UNPROXIED_NS, NULL, 0);
1285 			goto bad;
1286 		}
1287 	} else if (!ours) {
1288 		/*
1289 		 * The unicast forwarding function might return the packet
1290 		 * if we are proxying prefix(es), and if the packet is an
1291 		 * ICMPv6 packet that has failed the zone checks, but is
1292 		 * targetted towards a proxied address (this is optimized by
1293 		 * way of RTF_PROXY test.)  If so, claim the packet as ours
1294 		 * and let icmp6_input() handle the rest.  The packet's hop
1295 		 * limit value is kept intact (it's not decremented).  This
1296 		 * is for supporting Neighbor Unreachability Detection between
1297 		 * proxied nodes on different links (src is link-local, dst
1298 		 * is target address.)
1299 		 */
1300 		if ((m = ip6_forward(m, &rin6, 0)) == NULL) {
1301 			goto done;
1302 		}
1303 		VERIFY(rin6.ro_rt != NULL);
1304 		VERIFY(m->m_pkthdr.pkt_flags & PKTF_PROXY_DST);
1305 		deliverifp = rin6.ro_rt->rt_ifp;
1306 		ours = 1;
1307 	}
1308 
1309 	ip6 = mtod(m, struct ip6_hdr *);
1310 
1311 	/*
1312 	 * Malicious party may be able to use IPv4 mapped addr to confuse
1313 	 * tcp/udp stack and bypass security checks (act as if it was from
1314 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
1315 	 *
1316 	 * For SIIT end node behavior, you may want to disable the check.
1317 	 * However, you will  become vulnerable to attacks using IPv4 mapped
1318 	 * source.
1319 	 */
1320 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
1321 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
1322 		ip6stat.ip6s_badscope++;
1323 		in6_ifstat_inc(inifp, ifs6_in_addrerr);
1324 		m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP6_BAD_SCOPE, NULL, 0);
1325 		goto bad;
1326 	}
1327 
1328 	/*
1329 	 * Tell launch routine the next header
1330 	 */
1331 	ip6stat.ip6s_delivered++;
1332 	in6_ifstat_inc_na(deliverifp, ifs6_in_deliver);
1333 
1334 injectit:
1335 	nest = 0;
1336 
1337 	/*
1338 	 * Perform IP header alignment fixup again, if needed.  Note that
1339 	 * we do it once for the outermost protocol, and we assume each
1340 	 * protocol handler wouldn't mess with the alignment afterwards.
1341 	 */
1342 	IP6_HDR_ALIGNMENT_FIXUP(m, inifp, return );
1343 
1344 	while (nxt != IPPROTO_DONE) {
1345 		struct ipfilter *__single filter;
1346 		int (*pr_input)(struct mbuf **, int *, int);
1347 
1348 		/*
1349 		 * This would imply either IPPROTO_HOPOPTS was not the first
1350 		 * option or it did not come in the first fragment.
1351 		 */
1352 		if (nxt == IPPROTO_HOPOPTS &&
1353 		    (m->m_pkthdr.pkt_flags & PKTF_HBH_CHKED) == 0) {
1354 			/*
1355 			 * This implies that HBH option was not contained
1356 			 * in the first fragment
1357 			 */
1358 			ip6stat.ip6s_badoptions++;
1359 			m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP6_BAD_OPTION, NULL, 0);
1360 			goto bad;
1361 		}
1362 
1363 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
1364 			ip6stat.ip6s_toomanyhdr++;
1365 			m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP6_TOO_MANY_OPTIONS, NULL, 0);
1366 			goto bad;
1367 		}
1368 
1369 		/*
1370 		 * protection against faulty packet - there should be
1371 		 * more sanity checks in header chain processing.
1372 		 */
1373 		if (m->m_pkthdr.len < off) {
1374 			ip6stat.ip6s_tooshort++;
1375 			in6_ifstat_inc(inifp, ifs6_in_truncated);
1376 			m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP_TOO_SHORT, NULL, 0);
1377 			goto bad;
1378 		}
1379 
1380 #if IPSEC
1381 		/*
1382 		 * enforce IPsec policy checking if we are seeing last header.
1383 		 * note that we do not visit this with protocols with pcb layer
1384 		 * code - like udp/tcp/raw ip.
1385 		 */
1386 		if ((ipsec_bypass == 0) &&
1387 		    (ip6_protox[nxt]->pr_flags & PR_LASTHDR) != 0) {
1388 			if (ipsec6_in_reject(m, NULL)) {
1389 				IPSEC_STAT_INCREMENT(ipsec6stat.in_polvio);
1390 				m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IPSEC_REJECT, NULL, 0);
1391 				goto bad;
1392 			}
1393 		}
1394 #endif /* IPSEC */
1395 
1396 		/*
1397 		 * Call IP filter
1398 		 */
1399 		if (!TAILQ_EMPTY(&ipv6_filters) && !IFNET_IS_INTCOPROC(inifp)) {
1400 			ipf_ref();
1401 			TAILQ_FOREACH(filter, &ipv6_filters, ipf_link) {
1402 				if (seen == 0) {
1403 					if ((struct ipfilter *)inject_ipfref ==
1404 					    filter) {
1405 						seen = 1;
1406 					}
1407 				} else if (filter->ipf_filter.ipf_input) {
1408 					errno_t result;
1409 
1410 					result = filter->ipf_filter.ipf_input(
1411 						filter->ipf_filter.cookie,
1412 						(mbuf_t *)&m, off, (uint8_t)nxt);
1413 					if (result == EJUSTRETURN) {
1414 						ipf_unref();
1415 						goto done;
1416 					}
1417 					if (result != 0) {
1418 						ipf_unref();
1419 						m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP_FILTER_DROP, NULL, 0);
1420 						goto bad;
1421 					}
1422 				}
1423 			}
1424 			ipf_unref();
1425 		}
1426 
1427 		DTRACE_IP6(receive, struct mbuf *, m, struct inpcb *, NULL,
1428 		    struct ip6_hdr *, ip6, struct ifnet *, inifp,
1429 		    struct ip *, NULL, struct ip6_hdr *, ip6);
1430 
1431 		if ((pr_input = ip6_protox[nxt]->pr_input) == NULL) {
1432 			m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP_NO_PROTO, NULL, 0);
1433 			m = NULL;
1434 			nxt = IPPROTO_DONE;
1435 		} else if (!(ip6_protox[nxt]->pr_flags & PR_PROTOLOCK)) {
1436 			lck_mtx_lock(inet6_domain_mutex);
1437 			nxt = pr_input(&m, &off, nxt);
1438 			lck_mtx_unlock(inet6_domain_mutex);
1439 		} else {
1440 			nxt = pr_input(&m, &off, nxt);
1441 		}
1442 	}
1443 done:
1444 	ROUTE_RELEASE(&rin6);
1445 	return;
1446 bad:
1447 	goto done;
1448 }
1449 
1450 void
ip6_setsrcifaddr_info(struct mbuf * m,uint32_t src_idx,struct in6_ifaddr * ia6)1451 ip6_setsrcifaddr_info(struct mbuf *m, uint32_t src_idx, struct in6_ifaddr *ia6)
1452 {
1453 	VERIFY(m->m_flags & M_PKTHDR);
1454 	m->m_pkthdr.pkt_ext_flags &= ~PKTF_EXT_OUTPUT_SCOPE;
1455 	/*
1456 	 * If the source ifaddr is specified, pick up the information
1457 	 * from there; otherwise just grab the passed-in ifindex as the
1458 	 * caller may not have the ifaddr available.
1459 	 */
1460 	if (ia6 != NULL) {
1461 		m->m_pkthdr.pkt_flags |= PKTF_IFAINFO;
1462 		m->m_pkthdr.src_ifindex = ia6->ia_ifp->if_index;
1463 
1464 		/* See IN6_IFF comments in in6_var.h */
1465 		m->m_pkthdr.src_iff = (ia6->ia6_flags & 0xffff);
1466 	} else {
1467 		m->m_pkthdr.src_iff = 0;
1468 		m->m_pkthdr.src_ifindex = (uint16_t)src_idx;
1469 		if (src_idx != 0) {
1470 			m->m_pkthdr.pkt_flags |= PKTF_IFAINFO;
1471 		}
1472 	}
1473 }
1474 
1475 void
ip6_setdstifaddr_info(struct mbuf * m,uint32_t dst_idx,struct in6_ifaddr * ia6)1476 ip6_setdstifaddr_info(struct mbuf *m, uint32_t dst_idx, struct in6_ifaddr *ia6)
1477 {
1478 	VERIFY(m->m_flags & M_PKTHDR);
1479 	m->m_pkthdr.pkt_ext_flags &= ~PKTF_EXT_OUTPUT_SCOPE;
1480 
1481 	/*
1482 	 * If the destination ifaddr is specified, pick up the information
1483 	 * from there; otherwise just grab the passed-in ifindex as the
1484 	 * caller may not have the ifaddr available.
1485 	 */
1486 	if (ia6 != NULL) {
1487 		m->m_pkthdr.pkt_flags |= PKTF_IFAINFO;
1488 		m->m_pkthdr.dst_ifindex = ia6->ia_ifp->if_index;
1489 
1490 		/* See IN6_IFF comments in in6_var.h */
1491 		m->m_pkthdr.dst_iff = (ia6->ia6_flags & 0xffff);
1492 	} else {
1493 		m->m_pkthdr.dst_iff = 0;
1494 		m->m_pkthdr.dst_ifindex = (uint16_t)dst_idx;
1495 		if (dst_idx != 0) {
1496 			m->m_pkthdr.pkt_flags |= PKTF_IFAINFO;
1497 		}
1498 	}
1499 }
1500 
1501 int
ip6_getsrcifaddr_info(struct mbuf * m,uint32_t * src_idx,uint32_t * ia6f)1502 ip6_getsrcifaddr_info(struct mbuf *m, uint32_t *src_idx, uint32_t *ia6f)
1503 {
1504 	VERIFY(m->m_flags & M_PKTHDR);
1505 
1506 	if (!(m->m_pkthdr.pkt_flags & PKTF_IFAINFO)) {
1507 		return -1;
1508 	}
1509 
1510 	if (src_idx != NULL) {
1511 		*src_idx = m->m_pkthdr.src_ifindex;
1512 	}
1513 
1514 	if (ia6f != NULL) {
1515 		*ia6f = m->m_pkthdr.src_iff;
1516 	}
1517 
1518 	return 0;
1519 }
1520 
1521 int
ip6_getdstifaddr_info(struct mbuf * m,uint32_t * dst_idx,uint32_t * ia6f)1522 ip6_getdstifaddr_info(struct mbuf *m, uint32_t *dst_idx, uint32_t *ia6f)
1523 {
1524 	VERIFY(m->m_flags & M_PKTHDR);
1525 
1526 	if (!(m->m_pkthdr.pkt_flags & PKTF_IFAINFO)) {
1527 		return -1;
1528 	}
1529 
1530 	if (dst_idx != NULL) {
1531 		*dst_idx = m->m_pkthdr.dst_ifindex;
1532 	}
1533 
1534 	if (ia6f != NULL) {
1535 		*ia6f = m->m_pkthdr.dst_iff;
1536 	}
1537 
1538 	return 0;
1539 }
1540 
1541 uint32_t
ip6_input_getsrcifscope(struct mbuf * m)1542 ip6_input_getsrcifscope(struct mbuf *m)
1543 {
1544 	VERIFY(m->m_flags & M_PKTHDR);
1545 
1546 	if (m->m_pkthdr.rcvif != NULL) {
1547 		return m->m_pkthdr.rcvif->if_index;
1548 	}
1549 
1550 	uint32_t src_ifscope = IFSCOPE_NONE;
1551 	ip6_getsrcifaddr_info(m, &src_ifscope, NULL);
1552 	return src_ifscope;
1553 }
1554 
1555 uint32_t
ip6_input_getdstifscope(struct mbuf * m)1556 ip6_input_getdstifscope(struct mbuf *m)
1557 {
1558 	VERIFY(m->m_flags & M_PKTHDR);
1559 
1560 	if (m->m_pkthdr.rcvif != NULL) {
1561 		return m->m_pkthdr.rcvif->if_index;
1562 	}
1563 
1564 	uint32_t dst_ifscope = IFSCOPE_NONE;
1565 	ip6_getdstifaddr_info(m, &dst_ifscope, NULL);
1566 	return dst_ifscope;
1567 }
1568 
1569 /*
1570  * Hop-by-Hop options header processing. If a valid jumbo payload option is
1571  * included, the real payload length will be stored in plenp.
1572  */
1573 static int
ip6_hopopts_input(uint32_t * plenp,uint32_t * rtalertp,struct mbuf ** mp,int * offp)1574 ip6_hopopts_input(uint32_t *plenp, uint32_t *rtalertp, struct mbuf **mp,
1575     int *offp)
1576 {
1577 	mbuf_ref_t m = *mp;
1578 	int off = *offp, hbhlen;
1579 	struct ip6_hbh *hbh;
1580 	u_int8_t *opt;
1581 
1582 	/* validation of the length of the header */
1583 	IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), return (-1));
1584 	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
1585 	hbhlen = (hbh->ip6h_len + 1) << 3;
1586 
1587 	IP6_EXTHDR_CHECK(m, off, hbhlen, return (-1));
1588 	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
1589 	off += hbhlen;
1590 	hbhlen -= sizeof(struct ip6_hbh);
1591 	opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
1592 
1593 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
1594 	    hbhlen, rtalertp, plenp) < 0) {
1595 		return -1;
1596 	}
1597 
1598 	*offp = off;
1599 	*mp = m;
1600 	return 0;
1601 }
1602 
1603 /*
1604  * Search header for all Hop-by-hop options and process each option.
1605  * This function is separate from ip6_hopopts_input() in order to
1606  * handle a case where the sending node itself process its hop-by-hop
1607  * options header. In such a case, the function is called from ip6_output().
1608  *
1609  * The function assumes that hbh header is located right after the IPv6 header
1610  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
1611  * opthead + hbhlen is located in continuous memory region.
1612  */
1613 int
ip6_process_hopopts(struct mbuf * m,u_int8_t * __sized_by (inhbhlen)opthead,int inhbhlen,u_int32_t * rtalertp,u_int32_t * plenp)1614 ip6_process_hopopts(struct mbuf *m, u_int8_t *__sized_by(inhbhlen)opthead, int inhbhlen,
1615     u_int32_t *rtalertp, u_int32_t *plenp)
1616 {
1617 	struct ip6_hdr *__single ip6;
1618 	int optlen = 0;
1619 	int hbhlen = inhbhlen;
1620 	u_int8_t *opt = opthead;
1621 	u_int16_t rtalert_val;
1622 	u_int32_t jumboplen;
1623 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
1624 
1625 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
1626 		switch (*opt) {
1627 		case IP6OPT_PAD1:
1628 			optlen = 1;
1629 			break;
1630 		case IP6OPT_PADN:
1631 			if (hbhlen < IP6OPT_MINLEN) {
1632 				ip6stat.ip6s_toosmall++;
1633 				m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP_TOO_SMALL, NULL, 0);
1634 				goto bad;
1635 			}
1636 			optlen = *(opt + 1) + 2;
1637 			break;
1638 		case IP6OPT_ROUTER_ALERT:
1639 			/* XXX may need check for alignment */
1640 			if (hbhlen < IP6OPT_RTALERT_LEN) {
1641 				ip6stat.ip6s_toosmall++;
1642 				goto bad;
1643 			}
1644 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
1645 				/* XXX stat */
1646 				icmp6_error(m, ICMP6_PARAM_PROB,
1647 				    ICMP6_PARAMPROB_HEADER,
1648 				    (int)(erroff + opt + 1 - opthead));
1649 				return -1;
1650 			}
1651 			optlen = IP6OPT_RTALERT_LEN;
1652 			bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
1653 			*rtalertp = ntohs(rtalert_val);
1654 			break;
1655 		case IP6OPT_JUMBO:
1656 			/* XXX may need check for alignment */
1657 			if (hbhlen < IP6OPT_JUMBO_LEN) {
1658 				ip6stat.ip6s_toosmall++;
1659 				m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP_TOO_SMALL, NULL, 0);
1660 				goto bad;
1661 			}
1662 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
1663 				/* XXX stat */
1664 				icmp6_error(m, ICMP6_PARAM_PROB,
1665 				    ICMP6_PARAMPROB_HEADER,
1666 				    (int)(erroff + opt + 1 - opthead));
1667 				return -1;
1668 			}
1669 			optlen = IP6OPT_JUMBO_LEN;
1670 
1671 			/*
1672 			 * IPv6 packets that have non 0 payload length
1673 			 * must not contain a jumbo payload option.
1674 			 */
1675 			ip6 = mtod(m, struct ip6_hdr *);
1676 			if (ip6->ip6_plen) {
1677 				ip6stat.ip6s_badoptions++;
1678 				icmp6_error(m, ICMP6_PARAM_PROB,
1679 				    ICMP6_PARAMPROB_HEADER,
1680 				    (int)(erroff + opt - opthead));
1681 				return -1;
1682 			}
1683 
1684 			/*
1685 			 * We may see jumbolen in unaligned location, so
1686 			 * we'd need to perform bcopy().
1687 			 */
1688 			bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1689 			jumboplen = (u_int32_t)htonl(jumboplen);
1690 
1691 #if 1
1692 			/*
1693 			 * if there are multiple jumbo payload options,
1694 			 * *plenp will be non-zero and the packet will be
1695 			 * rejected.
1696 			 * the behavior may need some debate in ipngwg -
1697 			 * multiple options does not make sense, however,
1698 			 * there's no explicit mention in specification.
1699 			 */
1700 			if (*plenp != 0) {
1701 				ip6stat.ip6s_badoptions++;
1702 				icmp6_error(m, ICMP6_PARAM_PROB,
1703 				    ICMP6_PARAMPROB_HEADER,
1704 				    (int)(erroff + opt + 2 - opthead));
1705 				return -1;
1706 			}
1707 #endif
1708 
1709 			/*
1710 			 * jumbo payload length must be larger than 65535.
1711 			 */
1712 			if (jumboplen <= IPV6_MAXPACKET) {
1713 				ip6stat.ip6s_badoptions++;
1714 				icmp6_error(m, ICMP6_PARAM_PROB,
1715 				    ICMP6_PARAMPROB_HEADER,
1716 				    (int)(erroff + opt + 2 - opthead));
1717 				return -1;
1718 			}
1719 			*plenp = jumboplen;
1720 
1721 			break;
1722 		default:                /* unknown option */
1723 			if (hbhlen < IP6OPT_MINLEN) {
1724 				ip6stat.ip6s_toosmall++;
1725 				m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP_TOO_SMALL, NULL, 0);
1726 				goto bad;
1727 			}
1728 			optlen = ip6_unknown_opt(opt, hbhlen, m,
1729 			    erroff + opt - opthead);
1730 			if (optlen == -1) {
1731 				return -1;
1732 			}
1733 			optlen += 2;
1734 			break;
1735 		}
1736 	}
1737 
1738 	return 0;
1739 
1740 bad:
1741 	return -1;
1742 }
1743 
1744 /*
1745  * Unknown option processing.
1746  * The fourth argument `off' is the offset from the IPv6 header to the option,
1747  * which is necessary if the IPv6 header the and option header and IPv6 header
1748  * is not continuous in order to return an ICMPv6 error.
1749  */
1750 int
ip6_unknown_opt(uint8_t * __counted_by (optplen)optp,size_t optplen,struct mbuf * m,size_t off)1751 ip6_unknown_opt(uint8_t *__counted_by(optplen) optp, size_t optplen, struct mbuf *m, size_t off)
1752 {
1753 #pragma unused(optplen)
1754 
1755 	struct ip6_hdr *__single ip6;
1756 
1757 	switch (IP6OPT_TYPE(*optp)) {
1758 	case IP6OPT_TYPE_SKIP: /* ignore the option */
1759 		return (int)*(optp + 1);
1760 
1761 	case IP6OPT_TYPE_DISCARD:       /* silently discard */
1762 		m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP6_OPT_DISCARD, NULL, 0);
1763 		return -1;
1764 
1765 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1766 		ip6stat.ip6s_badoptions++;
1767 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, (int)off);
1768 		return -1;
1769 
1770 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1771 		ip6stat.ip6s_badoptions++;
1772 		ip6 = mtod(m, struct ip6_hdr *);
1773 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1774 		    (m->m_flags & (M_BCAST | M_MCAST))) {
1775 			m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_IP6_BAD_OPTION, NULL, 0);
1776 		} else {
1777 			icmp6_error(m, ICMP6_PARAM_PROB,
1778 			    ICMP6_PARAMPROB_OPTION, (int)off);
1779 		}
1780 		return -1;
1781 	}
1782 
1783 	m_drop(m, DROPTAP_FLAG_DIR_IN | DROPTAP_FLAG_L2_MISSING, DROP_REASON_UNSPECIFIED, NULL, 0); /* XXX: NOTREACHED */
1784 	return -1;
1785 }
1786 
1787 /*
1788  * Create the "control" list for this pcb.
1789  * These functions will not modify mbuf chain at all.
1790  *
1791  * With KAME mbuf chain restriction:
1792  * The routine will be called from upper layer handlers like tcp6_input().
1793  * Thus the routine assumes that the caller (tcp6_input) have already
1794  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1795  * very first mbuf on the mbuf chain.
1796  *
1797  * ip6_savecontrol_v4 will handle those options that are possible to be
1798  * set on a v4-mapped socket.
1799  * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those
1800  * options and handle the v6-only ones itself.
1801  */
1802 struct mbuf **
ip6_savecontrol_v4(struct inpcb * inp,struct mbuf * m,struct mbuf ** mp,int * v4only)1803 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp,
1804     int *v4only)
1805 {
1806 	struct ip6_hdr *__single ip6 = mtod(m, struct ip6_hdr *);
1807 
1808 	if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) {
1809 		struct timeval tv;
1810 
1811 		getmicrotime(&tv);
1812 		mp = sbcreatecontrol_mbuf((caddr_t)&tv, sizeof(tv),
1813 		    SCM_TIMESTAMP, SOL_SOCKET, mp);
1814 		if (*mp == NULL) {
1815 			return NULL;
1816 		}
1817 	}
1818 	if ((inp->inp_socket->so_options & SO_TIMESTAMP_MONOTONIC) != 0) {
1819 		uint64_t time;
1820 
1821 		time = mach_absolute_time();
1822 		mp = sbcreatecontrol_mbuf((caddr_t)&time, sizeof(time),
1823 		    SCM_TIMESTAMP_MONOTONIC, SOL_SOCKET, mp);
1824 		if (*mp == NULL) {
1825 			return NULL;
1826 		}
1827 	}
1828 	if ((inp->inp_socket->so_options & SO_TIMESTAMP_CONTINUOUS) != 0) {
1829 		uint64_t time;
1830 
1831 		time = mach_continuous_time();
1832 		mp = sbcreatecontrol_mbuf((caddr_t)&time, sizeof(time),
1833 		    SCM_TIMESTAMP_CONTINUOUS, SOL_SOCKET, mp);
1834 		if (*mp == NULL) {
1835 			return NULL;
1836 		}
1837 	}
1838 	if ((inp->inp_socket->so_flags & SOF_RECV_TRAFFIC_CLASS) != 0) {
1839 		int tc = m_get_traffic_class(m);
1840 
1841 		mp = sbcreatecontrol_mbuf((caddr_t)&tc, sizeof(tc),
1842 		    SO_TRAFFIC_CLASS, SOL_SOCKET, mp);
1843 		if (*mp == NULL) {
1844 			return NULL;
1845 		}
1846 	}
1847 
1848 	if ((inp->inp_socket->so_flags & SOF_RECV_WAKE_PKT) &&
1849 	    (m->m_pkthdr.pkt_flags & PKTF_WAKE_PKT)) {
1850 		int flag = 1;
1851 
1852 		mp = sbcreatecontrol_mbuf((caddr_t)&flag, sizeof(flag),
1853 		    SO_RECV_WAKE_PKT, SOL_SOCKET, mp);
1854 		if (*mp == NULL) {
1855 			return NULL;
1856 		}
1857 	}
1858 
1859 #define IS2292(inp, x, y)       (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y))
1860 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1861 		if (v4only != NULL) {
1862 			*v4only = 1;
1863 		}
1864 
1865 		// Send ECN flags for v4-mapped addresses
1866 		if ((inp->inp_flags & IN6P_TCLASS) != 0) {
1867 			struct ip *__single ip_header = mtod(m, struct ip *);
1868 
1869 			int tclass = (int)(ip_header->ip_tos);
1870 			mp = sbcreatecontrol_mbuf((caddr_t)&tclass, sizeof(tclass),
1871 			    IPV6_TCLASS, IPPROTO_IPV6, mp);
1872 			if (*mp == NULL) {
1873 				return NULL;
1874 			}
1875 		}
1876 
1877 		// Send IN6P_PKTINFO for v4-mapped address
1878 		if ((inp->inp_flags & IN6P_PKTINFO) != 0 || SOFLOW_ENABLED(inp->inp_socket)) {
1879 			struct in6_pktinfo pi6 = {
1880 				.ipi6_addr = IN6ADDR_V4MAPPED_INIT,
1881 				.ipi6_ifindex = (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0,
1882 			};
1883 
1884 			struct ip *__single ip_header = mtod(m, struct ip *);
1885 			bcopy(&ip_header->ip_dst, &pi6.ipi6_addr.s6_addr32[3], sizeof(struct in_addr));
1886 
1887 			mp = sbcreatecontrol_mbuf((caddr_t)&pi6,
1888 			    sizeof(struct in6_pktinfo),
1889 			    IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO),
1890 			    IPPROTO_IPV6, mp);
1891 			if (*mp == NULL) {
1892 				return NULL;
1893 			}
1894 		}
1895 		return mp;
1896 	}
1897 
1898 	/* RFC 2292 sec. 5 */
1899 	if ((inp->inp_flags & IN6P_PKTINFO) != 0 || SOFLOW_ENABLED(inp->inp_socket)) {
1900 		struct in6_pktinfo pi6;
1901 
1902 		bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1903 		in6_clearscope(&pi6.ipi6_addr); /* XXX */
1904 		pi6.ipi6_ifindex =
1905 		    (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1906 
1907 		mp = sbcreatecontrol_mbuf((caddr_t)&pi6,
1908 		    sizeof(struct in6_pktinfo),
1909 		    IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO),
1910 		    IPPROTO_IPV6, mp);
1911 		if (*mp == NULL) {
1912 			return NULL;
1913 		}
1914 	}
1915 
1916 	if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) {
1917 		int hlim = ip6->ip6_hlim & 0xff;
1918 
1919 		mp = sbcreatecontrol_mbuf((caddr_t)&hlim, sizeof(int),
1920 		    IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT),
1921 		    IPPROTO_IPV6, mp);
1922 		if (*mp == NULL) {
1923 			return NULL;
1924 		}
1925 	}
1926 
1927 	if (v4only != NULL) {
1928 		*v4only = 0;
1929 	}
1930 	return mp;
1931 }
1932 
1933 int
ip6_savecontrol(struct inpcb * in6p,struct mbuf * m,struct mbuf ** mp)1934 ip6_savecontrol(struct inpcb *in6p, struct mbuf *m, struct mbuf **mp)
1935 {
1936 	struct mbuf **np;
1937 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1938 	int v4only = 0;
1939 
1940 	*mp = NULL;
1941 	np = ip6_savecontrol_v4(in6p, m, mp, &v4only);
1942 	if (np == NULL) {
1943 		goto no_mbufs;
1944 	}
1945 
1946 	mp = np;
1947 	if (v4only) {
1948 		return 0;
1949 	}
1950 
1951 	if ((in6p->inp_flags & IN6P_TCLASS) != 0) {
1952 		u_int32_t flowinfo;
1953 		int tclass;
1954 
1955 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1956 		flowinfo >>= 20;
1957 
1958 		tclass = flowinfo & 0xff;
1959 		mp = sbcreatecontrol_mbuf((caddr_t)&tclass, sizeof(tclass),
1960 		    IPV6_TCLASS, IPPROTO_IPV6, mp);
1961 		if (*mp == NULL) {
1962 			goto no_mbufs;
1963 		}
1964 	}
1965 
1966 	/*
1967 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
1968 	 * privilege for the option (see ip6_ctloutput), but it might be too
1969 	 * strict, since there might be some hop-by-hop options which can be
1970 	 * returned to normal user.
1971 	 * See also RFC 2292 section 6 (or RFC 3542 section 8).
1972 	 */
1973 	if ((in6p->inp_flags & IN6P_HOPOPTS) != 0) {
1974 		/*
1975 		 * Check if a hop-by-hop options header is contatined in the
1976 		 * received packet, and if so, store the options as ancillary
1977 		 * data. Note that a hop-by-hop options header must be
1978 		 * just after the IPv6 header, which is assured through the
1979 		 * IPv6 input processing.
1980 		 */
1981 		ip6 = mtod(m, struct ip6_hdr *);
1982 		if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1983 			struct ip6_hbh *hbh;
1984 			int hbhlen = 0;
1985 			hbh = (struct ip6_hbh *)(ip6 + 1);
1986 			hbhlen = (hbh->ip6h_len + 1) << 3;
1987 
1988 			/*
1989 			 * XXX: We copy the whole header even if a
1990 			 * jumbo payload option is included, the option which
1991 			 * is to be removed before returning according to
1992 			 * RFC2292.
1993 			 * Note: this constraint is removed in RFC3542
1994 			 */
1995 			mp = sbcreatecontrol_mbuf((caddr_t)hbh, hbhlen,
1996 			    IS2292(in6p, IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1997 			    IPPROTO_IPV6, mp);
1998 
1999 			if (*mp == NULL) {
2000 				goto no_mbufs;
2001 			}
2002 		}
2003 	}
2004 
2005 	if ((in6p->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
2006 		int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
2007 
2008 		/*
2009 		 * Search for destination options headers or routing
2010 		 * header(s) through the header chain, and stores each
2011 		 * header as ancillary data.
2012 		 * Note that the order of the headers remains in
2013 		 * the chain of ancillary data.
2014 		 */
2015 		while (1) {     /* is explicit loop prevention necessary? */
2016 			struct ip6_ext *ip6e = NULL;
2017 			int elen;
2018 
2019 			/*
2020 			 * if it is not an extension header, don't try to
2021 			 * pull it from the chain.
2022 			 */
2023 			switch (nxt) {
2024 			case IPPROTO_DSTOPTS:
2025 			case IPPROTO_ROUTING:
2026 			case IPPROTO_HOPOPTS:
2027 			case IPPROTO_AH: /* is it possible? */
2028 				break;
2029 			default:
2030 				goto loopend;
2031 			}
2032 
2033 			if (off + sizeof(*ip6e) > m->m_len) {
2034 				goto loopend;
2035 			}
2036 			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
2037 			if (nxt == IPPROTO_AH) {
2038 				elen = (ip6e->ip6e_len + 2) << 2;
2039 			} else {
2040 				elen = (ip6e->ip6e_len + 1) << 3;
2041 			}
2042 			if (off + elen > m->m_len) {
2043 				goto loopend;
2044 			}
2045 
2046 			switch (nxt) {
2047 			case IPPROTO_DSTOPTS:
2048 				if (!(in6p->inp_flags & IN6P_DSTOPTS)) {
2049 					break;
2050 				}
2051 
2052 				mp = sbcreatecontrol_mbuf((caddr_t)ip6e, elen,
2053 				    IS2292(in6p, IPV6_2292DSTOPTS,
2054 				    IPV6_DSTOPTS), IPPROTO_IPV6, mp);
2055 				if (*mp == NULL) {
2056 					goto no_mbufs;
2057 				}
2058 				break;
2059 			case IPPROTO_ROUTING:
2060 				if (!(in6p->inp_flags & IN6P_RTHDR)) {
2061 					break;
2062 				}
2063 
2064 				mp = sbcreatecontrol_mbuf((caddr_t)ip6e, elen,
2065 				    IS2292(in6p, IPV6_2292RTHDR, IPV6_RTHDR),
2066 				    IPPROTO_IPV6, mp);
2067 				if (*mp == NULL) {
2068 					goto no_mbufs;
2069 				}
2070 				break;
2071 			case IPPROTO_HOPOPTS:
2072 			case IPPROTO_AH: /* is it possible? */
2073 				break;
2074 
2075 			default:
2076 				/*
2077 				 * other cases have been filtered in the above.
2078 				 * none will visit this case.  here we supply
2079 				 * the code just in case (nxt overwritten or
2080 				 * other cases).
2081 				 */
2082 				goto loopend;
2083 			}
2084 
2085 			/* proceed with the next header. */
2086 			off += elen;
2087 			nxt = ip6e->ip6e_nxt;
2088 			ip6e = NULL;
2089 		}
2090 loopend:
2091 		;
2092 	}
2093 	return 0;
2094 no_mbufs:
2095 	ip6stat.ip6s_pktdropcntrl++;
2096 	/* XXX increment a stat to show the failure */
2097 	return ENOBUFS;
2098 }
2099 #undef IS2292
2100 
2101 void
ip6_notify_pmtu(struct inpcb * in6p,struct sockaddr_in6 * dst,u_int32_t * mtu)2102 ip6_notify_pmtu(struct inpcb *in6p, struct sockaddr_in6 *dst, u_int32_t *mtu)
2103 {
2104 	struct socket *__single so;
2105 	mbuf_ref_t m_mtu;
2106 	struct ip6_mtuinfo mtuctl;
2107 
2108 	so =  in6p->inp_socket;
2109 
2110 	if ((in6p->inp_flags & IN6P_MTU) == 0) {
2111 		return;
2112 	}
2113 
2114 	if (mtu == NULL) {
2115 		return;
2116 	}
2117 
2118 	if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) && SOCK_CHECK_PROTO(so, IPPROTO_TCP)) {
2119 		return;
2120 	}
2121 
2122 	if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
2123 	    !in6_are_addr_equal_scoped(&in6p->in6p_faddr, &dst->sin6_addr, in6p->inp_fifscope, dst->sin6_scope_id)) {
2124 		return;
2125 	}
2126 
2127 	bzero(&mtuctl, sizeof(mtuctl));         /* zero-clear for safety */
2128 	mtuctl.ip6m_mtu = *mtu;
2129 	mtuctl.ip6m_addr = *dst;
2130 	if (!in6_embedded_scope) {
2131 		mtuctl.ip6m_addr.sin6_scope_id = dst->sin6_scope_id;
2132 	}
2133 	if (sa6_recoverscope(&mtuctl.ip6m_addr, TRUE)) {
2134 		return;
2135 	}
2136 
2137 	if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl),
2138 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) {
2139 		return;
2140 	}
2141 
2142 	if (sbappendaddr(&so->so_rcv, SA(dst), NULL, m_mtu, NULL) == 0) {
2143 		return;
2144 	}
2145 	sorwakeup(so);
2146 }
2147 
2148 /*
2149  * Get pointer to the previous header followed by the header
2150  * currently processed.
2151  * XXX: This function supposes that
2152  *	M includes all headers,
2153  *	the next header field and the header length field of each header
2154  *	are valid, and
2155  *	the sum of each header length equals to OFF.
2156  * Because of these assumptions, this function must be called very
2157  * carefully. Moreover, it will not be used in the near future when
2158  * we develop `neater' mechanism to process extension headers.
2159  */
2160 char *
ip6_get_prevhdr(struct mbuf * m,int off)2161 ip6_get_prevhdr(struct mbuf *m, int off)
2162 {
2163 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
2164 
2165 	if (off == sizeof(struct ip6_hdr)) {
2166 		return (char *)&ip6->ip6_nxt;
2167 	} else {
2168 		int len, nxt;
2169 		struct ip6_ext *ip6e = NULL;
2170 
2171 		nxt = ip6->ip6_nxt;
2172 		len = sizeof(struct ip6_hdr);
2173 		while (len < off) {
2174 			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
2175 
2176 			switch (nxt) {
2177 			case IPPROTO_FRAGMENT:
2178 				len += sizeof(struct ip6_frag);
2179 				break;
2180 			case IPPROTO_AH:
2181 				len += (ip6e->ip6e_len + 2) << 2;
2182 				break;
2183 			default:
2184 				len += (ip6e->ip6e_len + 1) << 3;
2185 				break;
2186 			}
2187 			nxt = ip6e->ip6e_nxt;
2188 		}
2189 		if (ip6e) {
2190 			return (char *)&ip6e->ip6e_nxt;
2191 		} else {
2192 			return NULL;
2193 		}
2194 	}
2195 }
2196 
2197 /*
2198  * get next header offset.  m will be retained.
2199  */
2200 int
ip6_nexthdr(struct mbuf * m,int off,int proto,int * nxtp)2201 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
2202 {
2203 	struct ip6_hdr ip6;
2204 	struct ip6_ext ip6e;
2205 	struct ip6_frag fh;
2206 
2207 	/* just in case */
2208 	VERIFY(m != NULL);
2209 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off) {
2210 		return -1;
2211 	}
2212 
2213 	switch (proto) {
2214 	case IPPROTO_IPV6:
2215 		if (m->m_pkthdr.len < off + sizeof(ip6)) {
2216 			return -1;
2217 		}
2218 		m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
2219 		if (nxtp) {
2220 			*nxtp = ip6.ip6_nxt;
2221 		}
2222 		off += sizeof(ip6);
2223 		return off;
2224 
2225 	case IPPROTO_FRAGMENT:
2226 		/*
2227 		 * terminate parsing if it is not the first fragment,
2228 		 * it does not make sense to parse through it.
2229 		 */
2230 		if (m->m_pkthdr.len < off + sizeof(fh)) {
2231 			return -1;
2232 		}
2233 		m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
2234 		/* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
2235 		if (fh.ip6f_offlg & IP6F_OFF_MASK) {
2236 			return -1;
2237 		}
2238 		if (nxtp) {
2239 			*nxtp = fh.ip6f_nxt;
2240 		}
2241 		off += sizeof(struct ip6_frag);
2242 		return off;
2243 
2244 	case IPPROTO_AH:
2245 		if (m->m_pkthdr.len < off + sizeof(ip6e)) {
2246 			return -1;
2247 		}
2248 		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
2249 		if (nxtp) {
2250 			*nxtp = ip6e.ip6e_nxt;
2251 		}
2252 		off += (ip6e.ip6e_len + 2) << 2;
2253 		return off;
2254 
2255 	case IPPROTO_HOPOPTS:
2256 	case IPPROTO_ROUTING:
2257 	case IPPROTO_DSTOPTS:
2258 		if (m->m_pkthdr.len < off + sizeof(ip6e)) {
2259 			return -1;
2260 		}
2261 		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
2262 		if (nxtp) {
2263 			*nxtp = ip6e.ip6e_nxt;
2264 		}
2265 		off += (ip6e.ip6e_len + 1) << 3;
2266 		return off;
2267 
2268 	case IPPROTO_NONE:
2269 	case IPPROTO_ESP:
2270 	case IPPROTO_IPCOMP:
2271 		/* give up */
2272 		return -1;
2273 
2274 	default:
2275 		return -1;
2276 	}
2277 }
2278 
2279 /*
2280  * get offset for the last header in the chain.  m will be kept untainted.
2281  */
2282 int
ip6_lasthdr(struct mbuf * m,int off,int proto,int * nxtp)2283 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
2284 {
2285 	int newoff;
2286 	int nxt;
2287 
2288 	if (!nxtp) {
2289 		nxt = -1;
2290 		nxtp = &nxt;
2291 	}
2292 	while (1) {
2293 		newoff = ip6_nexthdr(m, off, proto, nxtp);
2294 		if (newoff < 0) {
2295 			return off;
2296 		} else if (newoff < off) {
2297 			return -1;    /* invalid */
2298 		} else if (newoff == off) {
2299 			return newoff;
2300 		}
2301 
2302 		off = newoff;
2303 		proto = *nxtp;
2304 	}
2305 }
2306 
2307 boolean_t
ip6_pkt_has_ulp(struct mbuf * m)2308 ip6_pkt_has_ulp(struct mbuf *m)
2309 {
2310 	int off = 0, nxt = IPPROTO_NONE;
2311 
2312 	off = ip6_lasthdr(m, 0, IPPROTO_IPV6, &nxt);
2313 	if (off < 0 || m->m_pkthdr.len < off) {
2314 		return FALSE;
2315 	}
2316 
2317 	switch (nxt) {
2318 	case IPPROTO_TCP:
2319 		if (off + sizeof(struct tcphdr) > m->m_pkthdr.len) {
2320 			return FALSE;
2321 		}
2322 		break;
2323 	case IPPROTO_UDP:
2324 		if (off + sizeof(struct udphdr) > m->m_pkthdr.len) {
2325 			return FALSE;
2326 		}
2327 		break;
2328 	case IPPROTO_ICMPV6:
2329 		if (off + sizeof(uint32_t) > m->m_pkthdr.len) {
2330 			return FALSE;
2331 		}
2332 		break;
2333 	case IPPROTO_NONE:
2334 		return TRUE;
2335 	case IPPROTO_ESP:
2336 		return TRUE;
2337 	case IPPROTO_IPCOMP:
2338 		return TRUE;
2339 	default:
2340 		return FALSE;
2341 	}
2342 	return TRUE;
2343 }
2344 
2345 struct ip6aux *
ip6_addaux(struct mbuf * m)2346 ip6_addaux(struct mbuf *m)
2347 {
2348 	struct m_tag *__single tag;
2349 
2350 	/* Check if one is already allocated */
2351 	tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID,
2352 	    KERNEL_TAG_TYPE_INET6);
2353 	if (tag == NULL) {
2354 		/* Allocate a tag */
2355 		tag = m_tag_create(KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_INET6,
2356 		    sizeof(struct ip6aux), M_DONTWAIT, m);
2357 
2358 		/* Attach it to the mbuf */
2359 		if (tag) {
2360 			m_tag_prepend(m, tag);
2361 		}
2362 	}
2363 
2364 	return tag ? (struct ip6aux *)(tag->m_tag_data) : NULL;
2365 }
2366 
2367 struct ip6aux *
ip6_findaux(struct mbuf * m)2368 ip6_findaux(struct mbuf *m)
2369 {
2370 	struct m_tag *__single tag;
2371 
2372 	tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID,
2373 	    KERNEL_TAG_TYPE_INET6);
2374 
2375 	return tag != NULL ? (struct ip6aux *)(tag->m_tag_data) : NULL;
2376 }
2377 
2378 void
ip6_delaux(struct mbuf * m)2379 ip6_delaux(struct mbuf *m)
2380 {
2381 	struct m_tag *__single tag;
2382 
2383 	tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID,
2384 	    KERNEL_TAG_TYPE_INET6);
2385 	if (tag != NULL) {
2386 		m_tag_delete(m, tag);
2387 	}
2388 }
2389 
2390 struct inet6_tag_container {
2391 	struct m_tag    inet6_m_tag;
2392 	struct ip6aux   inet6_ip6a;
2393 };
2394 
2395 struct m_tag *
m_tag_kalloc_inet6(u_int32_t id,u_int16_t type,uint16_t len,int wait)2396 m_tag_kalloc_inet6(u_int32_t id, u_int16_t type, uint16_t len, int wait)
2397 {
2398 	struct inet6_tag_container *tag_container;
2399 	struct m_tag *tag = NULL;
2400 
2401 	assert3u(id, ==, KERNEL_MODULE_TAG_ID);
2402 	assert3u(type, ==, KERNEL_TAG_TYPE_INET6);
2403 	assert3u(len, ==, sizeof(struct ip6aux));
2404 
2405 	if (len != sizeof(struct ip6aux)) {
2406 		return NULL;
2407 	}
2408 
2409 	tag_container = kalloc_type(struct inet6_tag_container, wait | M_ZERO);
2410 	if (tag_container != NULL) {
2411 		tag =  &tag_container->inet6_m_tag;
2412 
2413 		assert3p(tag, ==, tag_container);
2414 
2415 		M_TAG_INIT(tag, id, type, len, &tag_container->inet6_ip6a, NULL);
2416 	}
2417 
2418 	return tag;
2419 }
2420 
2421 void
m_tag_kfree_inet6(struct m_tag * tag)2422 m_tag_kfree_inet6(struct m_tag *tag)
2423 {
2424 	struct inet6_tag_container *__single tag_container = (struct inet6_tag_container *)tag;
2425 
2426 	assert3u(tag->m_tag_len, ==, sizeof(struct ip6aux));
2427 
2428 	kfree_type(struct inet6_tag_container, tag_container);
2429 }
2430 
2431 void
ip6_register_m_tag(void)2432 ip6_register_m_tag(void)
2433 {
2434 	int error;
2435 
2436 	error = m_register_internal_tag_type(KERNEL_TAG_TYPE_INET6, sizeof(struct ip6aux),
2437 	    m_tag_kalloc_inet6, m_tag_kfree_inet6);
2438 
2439 	assert3u(error, ==, 0);
2440 }
2441 
2442 /*
2443  * Drain callback
2444  */
2445 void
ip6_drain(void)2446 ip6_drain(void)
2447 {
2448 	frag6_drain();          /* fragments */
2449 	in6_rtqdrain();         /* protocol cloned routes */
2450 	nd6_drain(NULL);        /* cloned routes: ND6 */
2451 }
2452 
2453 /*
2454  * System control for IP6
2455  */
2456 
2457 u_char  inet6ctlerrmap[PRC_NCMDS] = {
2458 	0, 0, 0, 0,
2459 	0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
2460 	EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
2461 	EMSGSIZE, EHOSTUNREACH, 0, 0,
2462 	0, 0, 0, 0,
2463 	ENOPROTOOPT, ECONNREFUSED
2464 };
2465 
2466 static int
2467 sysctl_reset_ip6_input_stats SYSCTL_HANDLER_ARGS
2468 {
2469 #pragma unused(arg1, arg2)
2470 	int error, i;
2471 
2472 	i = ip6_input_measure;
2473 	error = sysctl_handle_int(oidp, &i, 0, req);
2474 	if (error || req->newptr == USER_ADDR_NULL) {
2475 		goto done;
2476 	}
2477 	/* impose bounds */
2478 	if (i < 0 || i > 1) {
2479 		error = EINVAL;
2480 		goto done;
2481 	}
2482 	if (ip6_input_measure != i && i == 1) {
2483 		net_perf_initialize(&net_perf, ip6_input_measure_bins);
2484 	}
2485 	ip6_input_measure = i;
2486 done:
2487 	return error;
2488 }
2489 
2490 static int
2491 sysctl_ip6_input_measure_bins SYSCTL_HANDLER_ARGS
2492 {
2493 #pragma unused(arg1, arg2)
2494 	int error;
2495 	uint64_t i;
2496 
2497 	i = ip6_input_measure_bins;
2498 	error = sysctl_handle_quad(oidp, &i, 0, req);
2499 	if (error || req->newptr == USER_ADDR_NULL) {
2500 		goto done;
2501 	}
2502 	/* validate data */
2503 	if (!net_perf_validate_bins(i)) {
2504 		error = EINVAL;
2505 		goto done;
2506 	}
2507 	ip6_input_measure_bins = i;
2508 done:
2509 	return error;
2510 }
2511 
2512 static int
2513 sysctl_ip6_input_getperf SYSCTL_HANDLER_ARGS
2514 {
2515 #pragma unused(oidp, arg1, arg2)
2516 	if (req->oldptr == USER_ADDR_NULL) {
2517 		req->oldlen = (size_t)sizeof(struct net_perf);
2518 	}
2519 
2520 	return SYSCTL_OUT(req, &net_perf, MIN(sizeof(net_perf), req->oldlen));
2521 }
2522 
2523 
2524 /*
2525  * Initialize IPv6 source address hash table.
2526  */
2527 static void
in6_ifaddrhashtbl_init(void)2528 in6_ifaddrhashtbl_init(void)
2529 {
2530 	int i, k, p;
2531 	uint32_t nhash = 0;
2532 	uint32_t hash_size;
2533 
2534 	if (in6_ifaddrhashtbl != NULL) {
2535 		return;
2536 	}
2537 
2538 	PE_parse_boot_argn("ina6ddr_nhash", &nhash,
2539 	    sizeof(in6addr_nhash));
2540 	if (nhash == 0) {
2541 		nhash = IN6ADDR_NHASH;
2542 	}
2543 
2544 	hash_size = nhash * sizeof(*in6_ifaddrhashtbl);
2545 
2546 	in6_ifaddrhashtbl = zalloc_permanent(
2547 		hash_size,
2548 		ZALIGN_PTR);
2549 	in6addr_nhash = nhash;
2550 
2551 	/*
2552 	 * Generate the next largest prime greater than in6addr_nhash.
2553 	 */
2554 	k = (in6addr_nhash % 2 == 0) ? in6addr_nhash + 1 : in6addr_nhash + 2;
2555 	for (;;) {
2556 		p = 1;
2557 		for (i = 3; i * i <= k; i += 2) {
2558 			if (k % i == 0) {
2559 				p = 0;
2560 			}
2561 		}
2562 		if (p == 1) {
2563 			break;
2564 		}
2565 		k += 2;
2566 	}
2567 	in6addr_hashp = k;
2568 }
2569 
2570 static int
2571 sysctl_ip6_checkinterface SYSCTL_HANDLER_ARGS
2572 {
2573 #pragma unused(arg1, arg2)
2574 	int error, i;
2575 
2576 	i = ip6_checkinterface;
2577 	error = sysctl_handle_int(oidp, &i, 0, req);
2578 	if (error || req->newptr == USER_ADDR_NULL) {
2579 		return error;
2580 	}
2581 
2582 	switch (i) {
2583 	case IP6_CHECKINTERFACE_WEAK_ES:
2584 	case IP6_CHECKINTERFACE_HYBRID_ES:
2585 	case IP6_CHECKINTERFACE_STRONG_ES:
2586 		if (ip6_checkinterface != i) {
2587 			ip6_checkinterface = i;
2588 			os_log(OS_LOG_DEFAULT, "%s: ip6_checkinterface is now %d\n",
2589 			    __func__, ip6_checkinterface);
2590 		}
2591 		break;
2592 	default:
2593 		error = EINVAL;
2594 		break;
2595 	}
2596 	return error;
2597 }
2598