2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the project nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * $KAME: ip6_input.c,v 1.259 2002/01/21 04:58:09 jinmei Exp $
35 * Copyright (c) 1982, 1986, 1988, 1993
36 * The Regents of the University of California. All rights reserved.
38 * Redistribution and use in source and binary forms, with or without
39 * modification, are permitted provided that the following conditions
41 * 1. Redistributions of source code must retain the above copyright
42 * notice, this list of conditions and the following disclaimer.
43 * 2. Redistributions in binary form must reproduce the above copyright
44 * notice, this list of conditions and the following disclaimer in the
45 * documentation and/or other materials provided with the distribution.
46 * 3. Neither the name of the University nor the names of its contributors
47 * may be used to endorse or promote products derived from this software
48 * without specific prior written permission.
50 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
51 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
52 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
53 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
54 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
55 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
56 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
57 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
58 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
59 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
62 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
65 #include <sys/cdefs.h>
66 __FBSDID("$FreeBSD$");
69 #include "opt_inet6.h"
70 #include "opt_ipsec.h"
71 #include "opt_route.h"
75 #include <sys/param.h>
76 #include <sys/systm.h>
77 #include <sys/hhook.h>
78 #include <sys/malloc.h>
81 #include <sys/domain.h>
82 #include <sys/protosw.h>
84 #include <sys/socket.h>
85 #include <sys/socketvar.h>
86 #include <sys/errno.h>
88 #include <sys/kernel.h>
90 #include <sys/rmlock.h>
91 #include <sys/syslog.h>
92 #include <sys/sysctl.h>
93 #include <sys/eventhandler.h>
96 #include <net/if_var.h>
97 #include <net/if_types.h>
98 #include <net/if_private.h>
99 #include <net/if_dl.h>
100 #include <net/route.h>
101 #include <net/netisr.h>
102 #include <net/rss_config.h>
103 #include <net/pfil.h>
104 #include <net/vnet.h>
106 #include <netinet/in.h>
107 #include <netinet/in_kdtrace.h>
108 #include <netinet/ip_var.h>
109 #include <netinet/in_systm.h>
110 #include <net/if_llatbl.h>
112 #include <netinet/ip.h>
113 #include <netinet/ip_icmp.h>
115 #include <netinet/ip6.h>
116 #include <netinet6/in6_var.h>
117 #include <netinet6/ip6_var.h>
118 #include <netinet/ip_encap.h>
119 #include <netinet/in_pcb.h>
120 #include <netinet/icmp6.h>
121 #include <netinet6/scope6_var.h>
122 #include <netinet6/in6_ifattach.h>
123 #include <netinet6/mld6_var.h>
124 #include <netinet6/nd6.h>
125 #include <netinet6/in6_rss.h>
127 #include <netinet/sctp_pcb.h>
128 #include <netinet6/sctp6_var.h>
131 #include <netipsec/ipsec_support.h>
133 ip6proto_input_t *ip6_protox[IPPROTO_MAX] = {
134 [0 ... IPPROTO_MAX - 1] = rip6_input };
135 ip6proto_ctlinput_t *ip6_ctlprotox[IPPROTO_MAX] = {
136 [0 ... IPPROTO_MAX - 1] = rip6_ctlinput };
138 VNET_DEFINE(struct in6_ifaddrhead, in6_ifaddrhead);
139 VNET_DEFINE(struct in6_ifaddrlisthead *, in6_ifaddrhashtbl);
140 VNET_DEFINE(u_long, in6_ifaddrhmask);
142 static struct netisr_handler ip6_nh = {
144 .nh_handler = ip6_input,
145 .nh_proto = NETISR_IPV6,
147 .nh_m2cpuid = rss_soft_m2cpuid_v6,
148 .nh_policy = NETISR_POLICY_CPU,
149 .nh_dispatch = NETISR_DISPATCH_HYBRID,
151 .nh_policy = NETISR_POLICY_FLOW,
156 sysctl_netinet6_intr_queue_maxlen(SYSCTL_HANDLER_ARGS)
160 netisr_getqlimit(&ip6_nh, &qlimit);
161 error = sysctl_handle_int(oidp, &qlimit, 0, req);
162 if (error || !req->newptr)
166 return (netisr_setqlimit(&ip6_nh, qlimit));
168 SYSCTL_DECL(_net_inet6_ip6);
169 SYSCTL_PROC(_net_inet6_ip6, IPV6CTL_INTRQMAXLEN, intr_queue_maxlen,
170 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE,
171 0, 0, sysctl_netinet6_intr_queue_maxlen, "I",
172 "Maximum size of the IPv6 input queue");
174 VNET_DEFINE_STATIC(bool, ip6_sav) = true;
175 #define V_ip6_sav VNET(ip6_sav)
176 SYSCTL_BOOL(_net_inet6_ip6, OID_AUTO, source_address_validation,
177 CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip6_sav), true,
178 "Drop incoming packets with source address that is a local address");
181 static struct netisr_handler ip6_direct_nh = {
182 .nh_name = "ip6_direct",
183 .nh_handler = ip6_direct_input,
184 .nh_proto = NETISR_IPV6_DIRECT,
185 .nh_m2cpuid = rss_soft_m2cpuid_v6,
186 .nh_policy = NETISR_POLICY_CPU,
187 .nh_dispatch = NETISR_DISPATCH_HYBRID,
191 sysctl_netinet6_intr_direct_queue_maxlen(SYSCTL_HANDLER_ARGS)
195 netisr_getqlimit(&ip6_direct_nh, &qlimit);
196 error = sysctl_handle_int(oidp, &qlimit, 0, req);
197 if (error || !req->newptr)
201 return (netisr_setqlimit(&ip6_direct_nh, qlimit));
203 SYSCTL_PROC(_net_inet6_ip6, IPV6CTL_INTRDQMAXLEN, intr_direct_queue_maxlen,
204 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE,
205 0, 0, sysctl_netinet6_intr_direct_queue_maxlen, "I",
206 "Maximum size of the IPv6 direct input queue");
210 VNET_DEFINE(pfil_head_t, inet6_pfil_head);
212 VNET_PCPUSTAT_DEFINE(struct ip6stat, ip6stat);
213 VNET_PCPUSTAT_SYSINIT(ip6stat);
215 VNET_PCPUSTAT_SYSUNINIT(ip6stat);
218 struct rmlock in6_ifaddr_lock;
219 RM_SYSINIT(in6_ifaddr_lock, &in6_ifaddr_lock, "in6_ifaddr_lock");
221 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *);
224 * IP6 initialization: fill in IP6 protocol switch table.
225 * All protocols not implemented in kernel go to raw IP6 protocol handler.
228 ip6_vnet_init(void *arg __unused)
230 struct pfil_head_args args;
232 TUNABLE_INT_FETCH("net.inet6.ip6.auto_linklocal",
233 &V_ip6_auto_linklocal);
234 TUNABLE_INT_FETCH("net.inet6.ip6.accept_rtadv", &V_ip6_accept_rtadv);
235 TUNABLE_INT_FETCH("net.inet6.ip6.no_radr", &V_ip6_no_radr);
237 CK_STAILQ_INIT(&V_in6_ifaddrhead);
238 V_in6_ifaddrhashtbl = hashinit(IN6ADDR_NHASH, M_IFADDR,
241 /* Initialize packet filter hooks. */
242 args.pa_version = PFIL_VERSION;
243 args.pa_flags = PFIL_IN | PFIL_OUT;
244 args.pa_type = PFIL_TYPE_IP6;
245 args.pa_headname = PFIL_INET6_NAME;
246 V_inet6_pfil_head = pfil_head_register(&args);
248 if (hhook_head_register(HHOOK_TYPE_IPSEC_IN, AF_INET6,
249 &V_ipsec_hhh_in[HHOOK_IPSEC_INET6],
250 HHOOK_WAITOK | HHOOK_HEADISINVNET) != 0)
251 printf("%s: WARNING: unable to register input helper hook\n",
253 if (hhook_head_register(HHOOK_TYPE_IPSEC_OUT, AF_INET6,
254 &V_ipsec_hhh_out[HHOOK_IPSEC_INET6],
255 HHOOK_WAITOK | HHOOK_HEADISINVNET) != 0)
256 printf("%s: WARNING: unable to register output helper hook\n",
260 addrsel_policy_init();
264 V_ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
266 /* Skip global initialization stuff for non-default instances. */
268 netisr_register_vnet(&ip6_nh);
270 netisr_register_vnet(&ip6_direct_nh);
274 VNET_SYSINIT(ip6_vnet_init, SI_SUB_PROTO_DOMAIN, SI_ORDER_FOURTH,
275 ip6_vnet_init, NULL);
278 ip6_init(void *arg __unused)
282 * Register statically those protocols that are unlikely to ever go
285 IP6PROTO_REGISTER(IPPROTO_ICMPV6, icmp6_input, rip6_ctlinput);
286 IP6PROTO_REGISTER(IPPROTO_DSTOPTS, dest6_input, NULL);
287 IP6PROTO_REGISTER(IPPROTO_ROUTING, route6_input, NULL);
288 IP6PROTO_REGISTER(IPPROTO_FRAGMENT, frag6_input, NULL);
289 IP6PROTO_REGISTER(IPPROTO_IPV4, encap6_input, NULL);
290 IP6PROTO_REGISTER(IPPROTO_IPV6, encap6_input, NULL);
291 IP6PROTO_REGISTER(IPPROTO_ETHERIP, encap6_input, NULL);
292 IP6PROTO_REGISTER(IPPROTO_GRE, encap6_input, NULL);
293 IP6PROTO_REGISTER(IPPROTO_PIM, encap6_input, NULL);
294 #ifdef SCTP /* XXX: has a loadable & static version */
295 IP6PROTO_REGISTER(IPPROTO_SCTP, sctp6_input, sctp6_ctlinput);
298 EVENTHANDLER_REGISTER(vm_lowmem, frag6_drain, NULL, LOWMEM_PRI_DEFAULT);
299 EVENTHANDLER_REGISTER(mbuf_lowmem, frag6_drain, NULL,
302 netisr_register(&ip6_nh);
304 netisr_register(&ip6_direct_nh);
307 SYSINIT(ip6_init, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, ip6_init, NULL);
310 ip6proto_register(uint8_t proto, ip6proto_input_t input,
311 ip6proto_ctlinput_t ctl)
316 if (ip6_protox[proto] == rip6_input) {
317 ip6_protox[proto] = input;
318 ip6_ctlprotox[proto] = ctl;
325 ip6proto_unregister(uint8_t proto)
330 if (ip6_protox[proto] != rip6_input) {
331 ip6_protox[proto] = rip6_input;
332 ip6_ctlprotox[proto] = rip6_ctlinput;
340 ip6_destroy(void *unused __unused)
342 struct ifaddr *ifa, *nifa;
347 netisr_unregister_vnet(&ip6_direct_nh);
349 netisr_unregister_vnet(&ip6_nh);
351 pfil_head_unregister(V_inet6_pfil_head);
352 error = hhook_head_deregister(V_ipsec_hhh_in[HHOOK_IPSEC_INET6]);
354 printf("%s: WARNING: unable to deregister input helper hook "
355 "type HHOOK_TYPE_IPSEC_IN, id HHOOK_IPSEC_INET6: "
356 "error %d returned\n", __func__, error);
358 error = hhook_head_deregister(V_ipsec_hhh_out[HHOOK_IPSEC_INET6]);
360 printf("%s: WARNING: unable to deregister output helper hook "
361 "type HHOOK_TYPE_IPSEC_OUT, id HHOOK_IPSEC_INET6: "
362 "error %d returned\n", __func__, error);
365 /* Cleanup addresses. */
367 CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) {
368 /* Cannot lock here - lock recursion. */
369 /* IF_ADDR_LOCK(ifp); */
370 CK_STAILQ_FOREACH_SAFE(ifa, &ifp->if_addrhead, ifa_link, nifa) {
371 if (ifa->ifa_addr->sa_family != AF_INET6)
375 /* IF_ADDR_UNLOCK(ifp); */
376 in6_ifdetach_destroy(ifp);
377 mld_domifdetach(ifp);
381 /* Make sure any routes are gone as well. */
382 rib_flush_routes_family(AF_INET6);
386 in6_ifattach_destroy();
388 hashdestroy(V_in6_ifaddrhashtbl, M_IFADDR, V_in6_ifaddrhmask);
391 VNET_SYSUNINIT(inet6, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, ip6_destroy, NULL);
395 ip6_input_hbh(struct mbuf **mp, uint32_t *plen, uint32_t *rtalert, int *off,
402 if (ip6_hopopts_input(plen, rtalert, mp, off)) {
403 #if 0 /*touches NULL pointer*/
404 in6_ifstat_inc((*mp)->m_pkthdr.rcvif, ifs6_in_discard);
406 goto out; /* m have already been freed */
411 ip6 = mtod(m, struct ip6_hdr *);
414 * if the payload length field is 0 and the next header field
415 * indicates Hop-by-Hop Options header, then a Jumbo Payload
416 * option MUST be included.
418 if (ip6->ip6_plen == 0 && *plen == 0) {
420 * Note that if a valid jumbo payload option is
421 * contained, ip6_hopopts_input() must set a valid
422 * (non-zero) payload length to the variable plen.
424 IP6STAT_INC(ip6s_badoptions);
425 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
426 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
427 icmp6_error(m, ICMP6_PARAM_PROB,
428 ICMP6_PARAMPROB_HEADER,
429 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
432 /* ip6_hopopts_input() ensures that mbuf is contiguous */
433 hbh = (struct ip6_hbh *)(ip6 + 1);
434 *nxt = hbh->ip6h_nxt;
437 * If we are acting as a router and the packet contains a
438 * router alert option, see if we know the option value.
439 * Currently, we only support the option value for MLD, in which
440 * case we should pass the packet to the multicast routing
443 if (*rtalert != ~0) {
445 case IP6OPT_RTALERT_MLD:
446 if (V_ip6_forwarding)
451 * RFC2711 requires unrecognized values must be
466 * IPv6 direct input routine.
468 * This is called when reinjecting completed fragments where
469 * all of the previous checking and book-keeping has been done.
472 ip6_direct_input(struct mbuf *m)
477 struct ip6_direct_ctx *ip6dc;
479 mtag = m_tag_locate(m, MTAG_ABI_IPV6, IPV6_TAG_DIRECT, NULL);
480 KASSERT(mtag != NULL, ("Reinjected packet w/o direct ctx tag!"));
482 ip6dc = (struct ip6_direct_ctx *)(mtag + 1);
483 nxt = ip6dc->ip6dc_nxt;
484 off = ip6dc->ip6dc_off;
488 m_tag_delete(m, mtag);
490 while (nxt != IPPROTO_DONE) {
491 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) {
492 IP6STAT_INC(ip6s_toomanyhdr);
497 * protection against faulty packet - there should be
498 * more sanity checks in header chain processing.
500 if (m->m_pkthdr.len < off) {
501 IP6STAT_INC(ip6s_tooshort);
502 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
506 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
507 if (IPSEC_ENABLED(ipv6)) {
508 if (IPSEC_INPUT(ipv6, m, off, nxt) != 0)
513 nxt = ip6_protox[nxt](&m, &off, nxt);
522 ip6_input(struct mbuf *m)
524 struct in6_addr odst;
526 struct in6_ifaddr *ia;
529 u_int32_t rtalert = ~0;
530 int off = sizeof(struct ip6_hdr), nest;
535 * Drop the packet if IPv6 operation is disabled on the interface.
537 rcvif = m->m_pkthdr.rcvif;
538 if ((ND_IFINFO(rcvif)->flags & ND6_IFF_IFDISABLED))
541 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
543 * should the inner packet be considered authentic?
544 * see comment in ah4_input().
545 * NB: m cannot be NULL when passed to the input routine
548 m->m_flags &= ~M_AUTHIPHDR;
549 m->m_flags &= ~M_AUTHIPDGM;
553 if (m->m_flags & M_FASTFWD_OURS) {
555 * Firewall changed destination to local.
557 ip6 = mtod(m, struct ip6_hdr *);
564 if (m->m_flags & M_EXT) {
566 IP6STAT_INC(ip6s_mext2m);
568 IP6STAT_INC(ip6s_mext1);
571 struct ifnet *ifp = (m->m_flags & M_LOOP) ? V_loif : rcvif;
572 int ifindex = ifp->if_index;
573 if (ifindex >= IP6S_M2MMAX)
575 IP6STAT_INC(ip6s_m2m[ifindex]);
577 IP6STAT_INC(ip6s_m1);
580 in6_ifstat_inc(rcvif, ifs6_in_receive);
581 IP6STAT_INC(ip6s_total);
584 * L2 bridge code and some other code can return mbuf chain
585 * that does not conform to KAME requirement. too bad.
586 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
588 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
591 if (m->m_pkthdr.len > MHLEN)
592 n = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
594 n = m_gethdr(M_NOWAIT, MT_DATA);
599 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
600 n->m_len = n->m_pkthdr.len;
604 if (m->m_len < sizeof(struct ip6_hdr)) {
605 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
606 IP6STAT_INC(ip6s_toosmall);
607 in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
612 ip6 = mtod(m, struct ip6_hdr *);
613 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
614 IP6STAT_INC(ip6s_badvers);
615 in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
619 IP6STAT_INC(ip6s_nxthist[ip6->ip6_nxt]);
620 IP_PROBE(receive, NULL, NULL, ip6, rcvif, NULL, ip6);
623 * Check against address spoofing/corruption.
625 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
626 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
628 * XXX: "badscope" is not very suitable for a multicast source.
630 IP6STAT_INC(ip6s_badscope);
631 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
634 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) &&
635 !(m->m_flags & M_LOOP)) {
637 * In this case, the packet should come from the loopback
638 * interface. However, we cannot just check the if_flags,
639 * because ip6_mloopback() passes the "actual" interface
640 * as the outgoing/incoming interface.
642 IP6STAT_INC(ip6s_badscope);
643 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
646 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
647 IPV6_ADDR_MC_SCOPE(&ip6->ip6_dst) == 0) {
650 * Nodes must not originate a packet to a multicast address
651 * whose scop field contains the reserved value 0; if such
652 * a packet is received, it must be silently dropped.
654 IP6STAT_INC(ip6s_badscope);
655 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
659 * The following check is not documented in specs. A malicious
660 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
661 * and bypass security checks (act as if it was from 127.0.0.1 by using
662 * IPv6 src ::ffff:127.0.0.1). Be cautious.
664 * We have supported IPv6-only kernels for a few years and this issue
665 * has not come up. The world seems to move mostly towards not using
666 * v4mapped on the wire, so it makes sense for us to keep rejecting
669 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
670 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
671 IP6STAT_INC(ip6s_badscope);
672 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
677 * Reject packets with IPv4 compatible addresses (auto tunnel).
679 * The code forbids auto tunnel relay case in RFC1933 (the check is
680 * stronger than RFC1933). We may want to re-enable it if mech-xx
681 * is revised to forbid relaying case.
683 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
684 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
685 IP6STAT_INC(ip6s_badscope);
686 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
691 * Try to forward the packet, but if we fail continue.
692 * ip6_tryforward() does not generate redirects, so fall
693 * through to normal processing if redirects are required.
694 * ip6_tryforward() does inbound and outbound packet firewall
695 * processing. If firewall has decided that destination becomes
696 * our local address, it sets M_FASTFWD_OURS flag. In this
697 * case skip another inbound firewall processing and update
700 if (V_ip6_forwarding != 0 && V_ip6_sendredirects == 0
701 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
702 && (!IPSEC_ENABLED(ipv6) ||
703 IPSEC_CAPS(ipv6, m, IPSEC_CAP_OPERABLE) == 0)
706 if ((m = ip6_tryforward(m)) == NULL)
708 if (m->m_flags & M_FASTFWD_OURS) {
709 ip6 = mtod(m, struct ip6_hdr *);
713 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
715 * Bypass packet filtering for packets previously handled by IPsec.
717 if (IPSEC_ENABLED(ipv6) &&
718 IPSEC_CAPS(ipv6, m, IPSEC_CAP_BYPASS_FILTER) != 0)
722 * Run through list of hooks for input packets.
724 * NB: Beware of the destination address changing
725 * (e.g. by NAT rewriting). When this happens,
726 * tell ip6_forward to do the right thing.
729 /* Jump over all PFIL processing if hooks are not active. */
730 if (!PFIL_HOOKED_IN(V_inet6_pfil_head))
734 if (pfil_mbuf_in(V_inet6_pfil_head, &m, m->m_pkthdr.rcvif,
737 ip6 = mtod(m, struct ip6_hdr *);
738 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
739 if ((m->m_flags & (M_IP6_NEXTHOP | M_FASTFWD_OURS)) == M_IP6_NEXTHOP &&
740 m_tag_find(m, PACKET_TAG_IPFORWARD, NULL) != NULL) {
742 * Directly ship the packet on. This allows forwarding
743 * packets originally destined to us to some other directly
752 * Disambiguate address scope zones (if there is ambiguity).
753 * We first make sure that the original source or destination address
754 * is not in our internal form for scoped addresses. Such addresses
755 * are not necessarily invalid spec-wise, but we cannot accept them due
756 * to the usage conflict.
757 * in6_setscope() then also checks and rejects the cases where src or
758 * dst are the loopback address and the receiving interface
761 if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
762 IP6STAT_INC(ip6s_badscope); /* XXX */
765 if (in6_setscope(&ip6->ip6_src, rcvif, NULL) ||
766 in6_setscope(&ip6->ip6_dst, rcvif, NULL)) {
767 IP6STAT_INC(ip6s_badscope);
770 if (m->m_flags & M_FASTFWD_OURS) {
771 m->m_flags &= ~M_FASTFWD_OURS;
776 * Multicast check. Assume packet is for us to avoid
777 * prematurely taking locks.
779 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
781 in6_ifstat_inc(rcvif, ifs6_in_mcast);
786 * XXX: For now we keep link-local IPv6 addresses with embedded
787 * scope zone id, therefore we use zero zoneid here.
789 ia = in6ifa_ifwithaddr(&ip6->ip6_dst, 0 /* XXX */, false);
791 if (ia->ia6_flags & IN6_IFF_NOTREADY) {
792 char ip6bufs[INET6_ADDRSTRLEN];
793 char ip6bufd[INET6_ADDRSTRLEN];
794 /* address is not ready, so discard the packet. */
796 "ip6_input: packet to an unready address %s->%s\n",
797 ip6_sprintf(ip6bufs, &ip6->ip6_src),
798 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
801 if (V_ip6_sav && !(m->m_flags & M_LOOP) &&
802 __predict_false(in6_localip_fib(&ip6->ip6_src,
804 IP6STAT_INC(ip6s_badscope); /* XXX */
807 /* Count the packet in the ip address stats */
808 counter_u64_add(ia->ia_ifa.ifa_ipackets, 1);
809 counter_u64_add(ia->ia_ifa.ifa_ibytes, m->m_pkthdr.len);
815 * Now there is no reason to process the packet if it's not our own
816 * and we're not a router.
818 if (!V_ip6_forwarding) {
819 IP6STAT_INC(ip6s_cantforward);
825 * Process Hop-by-Hop options header if it's contained.
826 * m may be modified in ip6_hopopts_input().
827 * If a JumboPayload option is included, plen will also be modified.
829 plen = (u_int32_t)ntohs(ip6->ip6_plen);
830 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
831 if (ip6_input_hbh(&m, &plen, &rtalert, &off, &nxt, &ours) != 0)
837 * Use mbuf flags to propagate Router Alert option to
838 * ICMPv6 layer, as hop-by-hop options have been stripped.
841 m->m_flags |= M_RTALERT_MLD;
844 * Check that the amount of data in the buffers
845 * is as at least much as the IPv6 header would have us expect.
846 * Trim mbufs if longer than we expect.
847 * Drop packet if shorter than we expect.
849 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
850 IP6STAT_INC(ip6s_tooshort);
851 in6_ifstat_inc(rcvif, ifs6_in_truncated);
854 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
855 if (m->m_len == m->m_pkthdr.len) {
856 m->m_len = sizeof(struct ip6_hdr) + plen;
857 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
859 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
863 * Forward if desirable.
866 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
868 * If we are acting as a multicast router, all
869 * incoming multicast packets are passed to the
870 * kernel-level multicast forwarding function.
871 * The packet is returned (relatively) intact; if
872 * ip6_mforward() returns a non-zero value, the packet
873 * must be discarded, else it may be accepted below.
875 * XXX TODO: Check hlim and multicast scope here to avoid
876 * unnecessarily calling into ip6_mforward().
878 if (ip6_mforward && ip6_mforward(ip6, rcvif, m)) {
879 IP6STAT_INC(ip6s_cantforward);
883 ip6_forward(m, srcrt);
888 * Tell launch routine the next header
890 IP6STAT_INC(ip6s_delivered);
891 in6_ifstat_inc(rcvif, ifs6_in_deliver);
894 while (nxt != IPPROTO_DONE) {
895 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) {
896 IP6STAT_INC(ip6s_toomanyhdr);
901 * protection against faulty packet - there should be
902 * more sanity checks in header chain processing.
904 if (m->m_pkthdr.len < off) {
905 IP6STAT_INC(ip6s_tooshort);
906 in6_ifstat_inc(rcvif, ifs6_in_truncated);
910 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
911 if (IPSEC_ENABLED(ipv6)) {
912 if (IPSEC_INPUT(ipv6, m, off, nxt) != 0)
917 nxt = ip6_protox[nxt](&m, &off, nxt);
921 in6_ifstat_inc(rcvif, ifs6_in_discard);
927 * Hop-by-Hop options header processing. If a valid jumbo payload option is
928 * included, the real payload length will be stored in plenp.
930 * rtalertp - XXX: should be stored more smart way
933 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
934 struct mbuf **mp, int *offp)
936 struct mbuf *m = *mp;
937 int off = *offp, hbhlen;
940 /* validation of the length of the header */
941 if (m->m_len < off + sizeof(*hbh)) {
942 m = m_pullup(m, off + sizeof(*hbh));
944 IP6STAT_INC(ip6s_exthdrtoolong);
949 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
950 hbhlen = (hbh->ip6h_len + 1) << 3;
952 if (m->m_len < off + hbhlen) {
953 m = m_pullup(m, off + hbhlen);
955 IP6STAT_INC(ip6s_exthdrtoolong);
960 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
962 hbhlen -= sizeof(struct ip6_hbh);
963 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
964 hbhlen, rtalertp, plenp) < 0) {
975 * Search header for all Hop-by-hop options and process each option.
976 * This function is separate from ip6_hopopts_input() in order to
977 * handle a case where the sending node itself process its hop-by-hop
978 * options header. In such a case, the function is called from ip6_output().
980 * The function assumes that hbh header is located right after the IPv6 header
981 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
982 * opthead + hbhlen is located in contiguous memory region.
985 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
986 u_int32_t *rtalertp, u_int32_t *plenp)
990 u_int8_t *opt = opthead;
991 u_int16_t rtalert_val;
993 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
995 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
1001 if (hbhlen < IP6OPT_MINLEN) {
1002 IP6STAT_INC(ip6s_toosmall);
1005 optlen = *(opt + 1) + 2;
1007 case IP6OPT_ROUTER_ALERT:
1008 /* XXX may need check for alignment */
1009 if (hbhlen < IP6OPT_RTALERT_LEN) {
1010 IP6STAT_INC(ip6s_toosmall);
1013 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
1015 icmp6_error(m, ICMP6_PARAM_PROB,
1016 ICMP6_PARAMPROB_HEADER,
1017 erroff + opt + 1 - opthead);
1020 optlen = IP6OPT_RTALERT_LEN;
1021 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
1022 *rtalertp = ntohs(rtalert_val);
1025 /* XXX may need check for alignment */
1026 if (hbhlen < IP6OPT_JUMBO_LEN) {
1027 IP6STAT_INC(ip6s_toosmall);
1030 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
1032 icmp6_error(m, ICMP6_PARAM_PROB,
1033 ICMP6_PARAMPROB_HEADER,
1034 erroff + opt + 1 - opthead);
1037 optlen = IP6OPT_JUMBO_LEN;
1040 * IPv6 packets that have non 0 payload length
1041 * must not contain a jumbo payload option.
1043 ip6 = mtod(m, struct ip6_hdr *);
1044 if (ip6->ip6_plen) {
1045 IP6STAT_INC(ip6s_badoptions);
1046 icmp6_error(m, ICMP6_PARAM_PROB,
1047 ICMP6_PARAMPROB_HEADER,
1048 erroff + opt - opthead);
1053 * We may see jumbolen in unaligned location, so
1054 * we'd need to perform bcopy().
1056 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1057 jumboplen = (u_int32_t)htonl(jumboplen);
1061 * if there are multiple jumbo payload options,
1062 * *plenp will be non-zero and the packet will be
1064 * the behavior may need some debate in ipngwg -
1065 * multiple options does not make sense, however,
1066 * there's no explicit mention in specification.
1069 IP6STAT_INC(ip6s_badoptions);
1070 icmp6_error(m, ICMP6_PARAM_PROB,
1071 ICMP6_PARAMPROB_HEADER,
1072 erroff + opt + 2 - opthead);
1078 * jumbo payload length must be larger than 65535.
1080 if (jumboplen <= IPV6_MAXPACKET) {
1081 IP6STAT_INC(ip6s_badoptions);
1082 icmp6_error(m, ICMP6_PARAM_PROB,
1083 ICMP6_PARAMPROB_HEADER,
1084 erroff + opt + 2 - opthead);
1090 default: /* unknown option */
1091 if (hbhlen < IP6OPT_MINLEN) {
1092 IP6STAT_INC(ip6s_toosmall);
1095 optlen = ip6_unknown_opt(opt, m,
1096 erroff + opt - opthead);
1112 * Unknown option processing.
1113 * The third argument `off' is the offset from the IPv6 header to the option,
1114 * which is necessary if the IPv6 header the and option header and IPv6 header
1115 * is not contiguous in order to return an ICMPv6 error.
1118 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1120 struct ip6_hdr *ip6;
1122 switch (IP6OPT_TYPE(*optp)) {
1123 case IP6OPT_TYPE_SKIP: /* ignore the option */
1124 return ((int)*(optp + 1));
1125 case IP6OPT_TYPE_DISCARD: /* silently discard */
1128 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1129 IP6STAT_INC(ip6s_badoptions);
1130 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1132 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1133 IP6STAT_INC(ip6s_badoptions);
1134 ip6 = mtod(m, struct ip6_hdr *);
1135 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1136 (m->m_flags & (M_BCAST|M_MCAST)))
1139 icmp6_error(m, ICMP6_PARAM_PROB,
1140 ICMP6_PARAMPROB_OPTION, off);
1144 m_freem(m); /* XXX: NOTREACHED */
1149 * Create the "control" list for this pcb.
1150 * These functions will not modify mbuf chain at all.
1152 * The routine will be called from upper layer handlers like tcp6_input().
1153 * Thus the routine assumes that the caller (tcp6_input) have already
1154 * called m_pullup() and all the extension headers are located in the
1155 * very first mbuf on the mbuf chain.
1157 * ip6_savecontrol_v4 will handle those options that are possible to be
1158 * set on a v4-mapped socket.
1159 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those
1160 * options and handle the v6-only ones itself.
1163 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp,
1166 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1169 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) {
1175 struct bintime boottimebin, bt1;
1176 struct timespec ts1;
1180 switch (inp->inp_socket->so_ts_clock) {
1181 case SO_TS_REALTIME_MICRO:
1182 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1184 mbuf_tstmp2timespec(m, &ts1);
1185 timespec2bintime(&ts1, &bt1);
1186 getboottimebin(&boottimebin);
1187 bintime_add(&bt1, &boottimebin);
1188 bintime2timeval(&bt1, &t.tv);
1192 *mp = sbcreatecontrol(&t.tv, sizeof(t.tv),
1193 SCM_TIMESTAMP, SOL_SOCKET, M_NOWAIT);
1195 mp = &(*mp)->m_next;
1201 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1203 mbuf_tstmp2timespec(m, &ts1);
1204 timespec2bintime(&ts1, &t.bt);
1205 getboottimebin(&boottimebin);
1206 bintime_add(&t.bt, &boottimebin);
1210 *mp = sbcreatecontrol(&t.bt, sizeof(t.bt), SCM_BINTIME,
1211 SOL_SOCKET, M_NOWAIT);
1213 mp = &(*mp)->m_next;
1218 case SO_TS_REALTIME:
1219 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1221 mbuf_tstmp2timespec(m, &t.ts);
1222 getboottimebin(&boottimebin);
1223 bintime2timespec(&boottimebin, &ts1);
1224 timespecadd(&t.ts, &ts1, &t.ts);
1228 *mp = sbcreatecontrol(&t.ts, sizeof(t.ts),
1229 SCM_REALTIME, SOL_SOCKET, M_NOWAIT);
1231 mp = &(*mp)->m_next;
1236 case SO_TS_MONOTONIC:
1237 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1239 mbuf_tstmp2timespec(m, &t.ts);
1242 *mp = sbcreatecontrol(&t.ts, sizeof(t.ts),
1243 SCM_MONOTONIC, SOL_SOCKET, M_NOWAIT);
1245 mp = &(*mp)->m_next;
1251 panic("unknown (corrupted) so_ts_clock");
1253 if (stamped && (m->m_flags & (M_PKTHDR | M_TSTMP)) ==
1254 (M_PKTHDR | M_TSTMP)) {
1255 struct sock_timestamp_info sti;
1257 bzero(&sti, sizeof(sti));
1258 sti.st_info_flags = ST_INFO_HW;
1259 if ((m->m_flags & M_TSTMP_HPREC) != 0)
1260 sti.st_info_flags |= ST_INFO_HW_HPREC;
1261 *mp = sbcreatecontrol(&sti, sizeof(sti), SCM_TIME_INFO,
1262 SOL_SOCKET, M_NOWAIT);
1264 mp = &(*mp)->m_next;
1269 #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y))
1270 /* RFC 2292 sec. 5 */
1271 if ((inp->inp_flags & IN6P_PKTINFO) != 0) {
1272 struct in6_pktinfo pi6;
1274 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1278 ip = mtod(m, struct ip *);
1279 pi6.ipi6_addr.s6_addr32[0] = 0;
1280 pi6.ipi6_addr.s6_addr32[1] = 0;
1281 pi6.ipi6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP;
1282 pi6.ipi6_addr.s6_addr32[3] = ip->ip_dst.s_addr;
1284 /* We won't hit this code */
1285 bzero(&pi6.ipi6_addr, sizeof(struct in6_addr));
1288 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1289 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1292 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1294 *mp = sbcreatecontrol(&pi6, sizeof(struct in6_pktinfo),
1295 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6,
1298 mp = &(*mp)->m_next;
1301 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) {
1304 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1308 ip = mtod(m, struct ip *);
1311 /* We won't hit this code */
1315 hlim = ip6->ip6_hlim & 0xff;
1317 *mp = sbcreatecontrol(&hlim, sizeof(int),
1318 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT),
1319 IPPROTO_IPV6, M_NOWAIT);
1321 mp = &(*mp)->m_next;
1324 if ((inp->inp_flags & IN6P_TCLASS) != 0) {
1327 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1331 ip = mtod(m, struct ip *);
1332 tclass = ip->ip_tos;
1334 /* We won't hit this code */
1340 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1342 tclass = flowinfo & 0xff;
1344 *mp = sbcreatecontrol(&tclass, sizeof(int), IPV6_TCLASS,
1345 IPPROTO_IPV6, M_NOWAIT);
1347 mp = &(*mp)->m_next;
1350 if (v4only != NULL) {
1351 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1362 ip6_savecontrol(struct inpcb *inp, struct mbuf *m, struct mbuf **mp)
1364 struct ip6_hdr *ip6;
1367 mp = ip6_savecontrol_v4(inp, m, mp, &v4only);
1371 ip6 = mtod(m, struct ip6_hdr *);
1373 * IPV6_HOPOPTS socket option. Recall that we required super-user
1374 * privilege for the option (see ip6_ctloutput), but it might be too
1375 * strict, since there might be some hop-by-hop options which can be
1376 * returned to normal user.
1377 * See also RFC 2292 section 6 (or RFC 3542 section 8).
1379 if ((inp->inp_flags & IN6P_HOPOPTS) != 0) {
1381 * Check if a hop-by-hop options header is contatined in the
1382 * received packet, and if so, store the options as ancillary
1383 * data. Note that a hop-by-hop options header must be
1384 * just after the IPv6 header, which is assured through the
1385 * IPv6 input processing.
1387 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1388 struct ip6_hbh *hbh;
1391 hbh = (struct ip6_hbh *)(ip6 + 1);
1392 hbhlen = (hbh->ip6h_len + 1) << 3;
1395 * XXX: We copy the whole header even if a
1396 * jumbo payload option is included, the option which
1397 * is to be removed before returning according to
1399 * Note: this constraint is removed in RFC3542
1401 *mp = sbcreatecontrol(hbh, hbhlen,
1402 IS2292(inp, IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1403 IPPROTO_IPV6, M_NOWAIT);
1405 mp = &(*mp)->m_next;
1409 if ((inp->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
1410 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1413 * Search for destination options headers or routing
1414 * header(s) through the header chain, and stores each
1415 * header as ancillary data.
1416 * Note that the order of the headers remains in
1417 * the chain of ancillary data.
1419 while (1) { /* is explicit loop prevention necessary? */
1420 struct ip6_ext *ip6e = NULL;
1424 * if it is not an extension header, don't try to
1425 * pull it from the chain.
1428 case IPPROTO_DSTOPTS:
1429 case IPPROTO_ROUTING:
1430 case IPPROTO_HOPOPTS:
1431 case IPPROTO_AH: /* is it possible? */
1437 if (off + sizeof(*ip6e) > m->m_len)
1439 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1440 if (nxt == IPPROTO_AH)
1441 elen = (ip6e->ip6e_len + 2) << 2;
1443 elen = (ip6e->ip6e_len + 1) << 3;
1444 if (off + elen > m->m_len)
1448 case IPPROTO_DSTOPTS:
1449 if (!(inp->inp_flags & IN6P_DSTOPTS))
1452 *mp = sbcreatecontrol(ip6e, elen,
1453 IS2292(inp, IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1454 IPPROTO_IPV6, M_NOWAIT);
1456 mp = &(*mp)->m_next;
1458 case IPPROTO_ROUTING:
1459 if (!(inp->inp_flags & IN6P_RTHDR))
1462 *mp = sbcreatecontrol(ip6e, elen,
1463 IS2292(inp, IPV6_2292RTHDR, IPV6_RTHDR),
1464 IPPROTO_IPV6, M_NOWAIT);
1466 mp = &(*mp)->m_next;
1468 case IPPROTO_HOPOPTS:
1469 case IPPROTO_AH: /* is it possible? */
1474 * other cases have been filtered in the above.
1475 * none will visit this case. here we supply
1476 * the code just in case (nxt overwritten or
1482 /* proceed with the next header. */
1484 nxt = ip6e->ip6e_nxt;
1491 if (inp->inp_flags2 & INP_RECVFLOWID) {
1492 uint32_t flowid, flow_type;
1494 flowid = m->m_pkthdr.flowid;
1495 flow_type = M_HASHTYPE_GET(m);
1498 * XXX should handle the failure of one or the
1499 * other - don't populate both?
1501 *mp = sbcreatecontrol(&flowid, sizeof(uint32_t), IPV6_FLOWID,
1502 IPPROTO_IPV6, M_NOWAIT);
1504 mp = &(*mp)->m_next;
1505 *mp = sbcreatecontrol(&flow_type, sizeof(uint32_t),
1506 IPV6_FLOWTYPE, IPPROTO_IPV6, M_NOWAIT);
1508 mp = &(*mp)->m_next;
1512 if (inp->inp_flags2 & INP_RECVRSSBUCKETID) {
1513 uint32_t flowid, flow_type;
1514 uint32_t rss_bucketid;
1516 flowid = m->m_pkthdr.flowid;
1517 flow_type = M_HASHTYPE_GET(m);
1519 if (rss_hash2bucket(flowid, flow_type, &rss_bucketid) == 0) {
1520 *mp = sbcreatecontrol(&rss_bucketid, sizeof(uint32_t),
1521 IPV6_RSSBUCKETID, IPPROTO_IPV6, M_NOWAIT);
1523 mp = &(*mp)->m_next;
1532 ip6_notify_pmtu(struct inpcb *inp, struct sockaddr_in6 *dst, u_int32_t mtu)
1536 struct ip6_mtuinfo mtuctl;
1538 KASSERT(inp != NULL, ("%s: inp == NULL", __func__));
1540 * Notify the error by sending IPV6_PATHMTU ancillary data if
1541 * application wanted to know the MTU value.
1542 * NOTE: we notify disconnected sockets, because some udp
1543 * applications keep sending sockets disconnected.
1544 * NOTE: our implementation doesn't notify connected sockets that has
1545 * foreign address that is different than given destination addresses
1546 * (this is permitted by RFC 3542).
1548 if ((inp->inp_flags & IN6P_MTU) == 0 || (
1549 !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
1550 !IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &dst->sin6_addr)))
1553 mtuctl.ip6m_mtu = mtu;
1554 mtuctl.ip6m_addr = *dst;
1555 if (sa6_recoverscope(&mtuctl.ip6m_addr))
1558 if ((m_mtu = sbcreatecontrol(&mtuctl, sizeof(mtuctl), IPV6_PATHMTU,
1559 IPPROTO_IPV6, M_NOWAIT)) == NULL)
1562 so = inp->inp_socket;
1563 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu)
1567 /* XXX: should count statistics */
1573 * Get pointer to the previous header followed by the header
1574 * currently processed.
1577 ip6_get_prevhdr(const struct mbuf *m, int off)
1579 struct ip6_ext ip6e;
1580 struct ip6_hdr *ip6;
1583 if (off == sizeof(struct ip6_hdr))
1584 return (offsetof(struct ip6_hdr, ip6_nxt));
1585 if (off < sizeof(struct ip6_hdr))
1586 panic("%s: off < sizeof(struct ip6_hdr)", __func__);
1588 ip6 = mtod(m, struct ip6_hdr *);
1590 len = sizeof(struct ip6_hdr);
1593 m_copydata(m, len, sizeof(ip6e), (caddr_t)&ip6e);
1595 case IPPROTO_FRAGMENT:
1596 nlen = sizeof(struct ip6_frag);
1599 nlen = (ip6e.ip6e_len + 2) << 2;
1602 nlen = (ip6e.ip6e_len + 1) << 3;
1605 nxt = ip6e.ip6e_nxt;
1607 return (len - nlen);
1611 * get next header offset. m will be retained.
1614 ip6_nexthdr(const struct mbuf *m, int off, int proto, int *nxtp)
1617 struct ip6_ext ip6e;
1622 panic("ip6_nexthdr: m == NULL");
1623 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1628 if (m->m_pkthdr.len < off + sizeof(ip6))
1630 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1632 *nxtp = ip6.ip6_nxt;
1636 case IPPROTO_FRAGMENT:
1638 * terminate parsing if it is not the first fragment,
1639 * it does not make sense to parse through it.
1641 if (m->m_pkthdr.len < off + sizeof(fh))
1643 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1644 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1645 if (fh.ip6f_offlg & IP6F_OFF_MASK)
1648 *nxtp = fh.ip6f_nxt;
1649 off += sizeof(struct ip6_frag);
1653 if (m->m_pkthdr.len < off + sizeof(ip6e))
1655 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1657 *nxtp = ip6e.ip6e_nxt;
1658 off += (ip6e.ip6e_len + 2) << 2;
1661 case IPPROTO_HOPOPTS:
1662 case IPPROTO_ROUTING:
1663 case IPPROTO_DSTOPTS:
1664 if (m->m_pkthdr.len < off + sizeof(ip6e))
1666 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1668 *nxtp = ip6e.ip6e_nxt;
1669 off += (ip6e.ip6e_len + 1) << 3;
1674 case IPPROTO_IPCOMP:
1686 * get offset for the last header in the chain. m will be kept untainted.
1689 ip6_lasthdr(const struct mbuf *m, int off, int proto, int *nxtp)
1699 newoff = ip6_nexthdr(m, off, proto, nxtp);
1702 else if (newoff < off)
1703 return -1; /* invalid */
1704 else if (newoff == off)