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);
211 VNET_DEFINE(pfil_head_t, inet6_local_pfil_head);
213 VNET_PCPUSTAT_DEFINE(struct ip6stat, ip6stat);
214 VNET_PCPUSTAT_SYSINIT(ip6stat);
216 VNET_PCPUSTAT_SYSUNINIT(ip6stat);
219 struct rmlock in6_ifaddr_lock;
220 RM_SYSINIT(in6_ifaddr_lock, &in6_ifaddr_lock, "in6_ifaddr_lock");
222 static int ip6_hopopts_input(u_int32_t *, u_int32_t *, struct mbuf **, int *);
225 * IP6 initialization: fill in IP6 protocol switch table.
226 * All protocols not implemented in kernel go to raw IP6 protocol handler.
229 ip6_vnet_init(void *arg __unused)
231 struct pfil_head_args args;
233 TUNABLE_INT_FETCH("net.inet6.ip6.auto_linklocal",
234 &V_ip6_auto_linklocal);
235 TUNABLE_INT_FETCH("net.inet6.ip6.accept_rtadv", &V_ip6_accept_rtadv);
236 TUNABLE_INT_FETCH("net.inet6.ip6.no_radr", &V_ip6_no_radr);
238 CK_STAILQ_INIT(&V_in6_ifaddrhead);
239 V_in6_ifaddrhashtbl = hashinit(IN6ADDR_NHASH, M_IFADDR,
242 /* Initialize packet filter hooks. */
243 args.pa_version = PFIL_VERSION;
244 args.pa_flags = PFIL_IN | PFIL_OUT;
245 args.pa_type = PFIL_TYPE_IP6;
246 args.pa_headname = PFIL_INET6_NAME;
247 V_inet6_pfil_head = pfil_head_register(&args);
249 args.pa_flags = PFIL_OUT;
250 args.pa_headname = PFIL_INET6_LOCAL_NAME;
251 V_inet6_local_pfil_head = pfil_head_register(&args);
253 if (hhook_head_register(HHOOK_TYPE_IPSEC_IN, AF_INET6,
254 &V_ipsec_hhh_in[HHOOK_IPSEC_INET6],
255 HHOOK_WAITOK | HHOOK_HEADISINVNET) != 0)
256 printf("%s: WARNING: unable to register input helper hook\n",
258 if (hhook_head_register(HHOOK_TYPE_IPSEC_OUT, AF_INET6,
259 &V_ipsec_hhh_out[HHOOK_IPSEC_INET6],
260 HHOOK_WAITOK | HHOOK_HEADISINVNET) != 0)
261 printf("%s: WARNING: unable to register output helper hook\n",
265 addrsel_policy_init();
269 V_ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
271 /* Skip global initialization stuff for non-default instances. */
273 netisr_register_vnet(&ip6_nh);
275 netisr_register_vnet(&ip6_direct_nh);
279 VNET_SYSINIT(ip6_vnet_init, SI_SUB_PROTO_DOMAIN, SI_ORDER_FOURTH,
280 ip6_vnet_init, NULL);
283 ip6_init(void *arg __unused)
287 * Register statically those protocols that are unlikely to ever go
290 IP6PROTO_REGISTER(IPPROTO_ICMPV6, icmp6_input, rip6_ctlinput);
291 IP6PROTO_REGISTER(IPPROTO_DSTOPTS, dest6_input, NULL);
292 IP6PROTO_REGISTER(IPPROTO_ROUTING, route6_input, NULL);
293 IP6PROTO_REGISTER(IPPROTO_FRAGMENT, frag6_input, NULL);
294 IP6PROTO_REGISTER(IPPROTO_IPV4, encap6_input, NULL);
295 IP6PROTO_REGISTER(IPPROTO_IPV6, encap6_input, NULL);
296 IP6PROTO_REGISTER(IPPROTO_ETHERIP, encap6_input, NULL);
297 IP6PROTO_REGISTER(IPPROTO_GRE, encap6_input, NULL);
298 IP6PROTO_REGISTER(IPPROTO_PIM, encap6_input, NULL);
299 #ifdef SCTP /* XXX: has a loadable & static version */
300 IP6PROTO_REGISTER(IPPROTO_SCTP, sctp6_input, sctp6_ctlinput);
303 EVENTHANDLER_REGISTER(vm_lowmem, frag6_drain, NULL, LOWMEM_PRI_DEFAULT);
304 EVENTHANDLER_REGISTER(mbuf_lowmem, frag6_drain, NULL,
307 netisr_register(&ip6_nh);
309 netisr_register(&ip6_direct_nh);
312 SYSINIT(ip6_init, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, ip6_init, NULL);
315 ip6proto_register(uint8_t proto, ip6proto_input_t input,
316 ip6proto_ctlinput_t ctl)
321 if (ip6_protox[proto] == rip6_input) {
322 ip6_protox[proto] = input;
323 ip6_ctlprotox[proto] = ctl;
330 ip6proto_unregister(uint8_t proto)
335 if (ip6_protox[proto] != rip6_input) {
336 ip6_protox[proto] = rip6_input;
337 ip6_ctlprotox[proto] = rip6_ctlinput;
345 ip6_destroy(void *unused __unused)
347 struct ifaddr *ifa, *nifa;
352 netisr_unregister_vnet(&ip6_direct_nh);
354 netisr_unregister_vnet(&ip6_nh);
356 pfil_head_unregister(V_inet6_pfil_head);
357 error = hhook_head_deregister(V_ipsec_hhh_in[HHOOK_IPSEC_INET6]);
359 printf("%s: WARNING: unable to deregister input helper hook "
360 "type HHOOK_TYPE_IPSEC_IN, id HHOOK_IPSEC_INET6: "
361 "error %d returned\n", __func__, error);
363 error = hhook_head_deregister(V_ipsec_hhh_out[HHOOK_IPSEC_INET6]);
365 printf("%s: WARNING: unable to deregister output helper hook "
366 "type HHOOK_TYPE_IPSEC_OUT, id HHOOK_IPSEC_INET6: "
367 "error %d returned\n", __func__, error);
370 /* Cleanup addresses. */
372 CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) {
373 /* Cannot lock here - lock recursion. */
374 /* IF_ADDR_LOCK(ifp); */
375 CK_STAILQ_FOREACH_SAFE(ifa, &ifp->if_addrhead, ifa_link, nifa) {
376 if (ifa->ifa_addr->sa_family != AF_INET6)
380 /* IF_ADDR_UNLOCK(ifp); */
381 in6_ifdetach_destroy(ifp);
382 mld_domifdetach(ifp);
386 /* Make sure any routes are gone as well. */
387 rib_flush_routes_family(AF_INET6);
391 in6_ifattach_destroy();
393 hashdestroy(V_in6_ifaddrhashtbl, M_IFADDR, V_in6_ifaddrhmask);
396 VNET_SYSUNINIT(inet6, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, ip6_destroy, NULL);
400 ip6_input_hbh(struct mbuf **mp, uint32_t *plen, uint32_t *rtalert, int *off,
407 if (ip6_hopopts_input(plen, rtalert, mp, off)) {
408 #if 0 /*touches NULL pointer*/
409 in6_ifstat_inc((*mp)->m_pkthdr.rcvif, ifs6_in_discard);
411 goto out; /* m have already been freed */
416 ip6 = mtod(m, struct ip6_hdr *);
419 * if the payload length field is 0 and the next header field
420 * indicates Hop-by-Hop Options header, then a Jumbo Payload
421 * option MUST be included.
423 if (ip6->ip6_plen == 0 && *plen == 0) {
425 * Note that if a valid jumbo payload option is
426 * contained, ip6_hopopts_input() must set a valid
427 * (non-zero) payload length to the variable plen.
429 IP6STAT_INC(ip6s_badoptions);
430 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
431 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
432 icmp6_error(m, ICMP6_PARAM_PROB,
433 ICMP6_PARAMPROB_HEADER,
434 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
437 /* ip6_hopopts_input() ensures that mbuf is contiguous */
438 hbh = (struct ip6_hbh *)(ip6 + 1);
439 *nxt = hbh->ip6h_nxt;
442 * If we are acting as a router and the packet contains a
443 * router alert option, see if we know the option value.
444 * Currently, we only support the option value for MLD, in which
445 * case we should pass the packet to the multicast routing
448 if (*rtalert != ~0) {
450 case IP6OPT_RTALERT_MLD:
451 if (V_ip6_forwarding)
456 * RFC2711 requires unrecognized values must be
471 * IPv6 direct input routine.
473 * This is called when reinjecting completed fragments where
474 * all of the previous checking and book-keeping has been done.
477 ip6_direct_input(struct mbuf *m)
482 struct ip6_direct_ctx *ip6dc;
484 mtag = m_tag_locate(m, MTAG_ABI_IPV6, IPV6_TAG_DIRECT, NULL);
485 KASSERT(mtag != NULL, ("Reinjected packet w/o direct ctx tag!"));
487 ip6dc = (struct ip6_direct_ctx *)(mtag + 1);
488 nxt = ip6dc->ip6dc_nxt;
489 off = ip6dc->ip6dc_off;
493 m_tag_delete(m, mtag);
495 while (nxt != IPPROTO_DONE) {
496 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) {
497 IP6STAT_INC(ip6s_toomanyhdr);
502 * protection against faulty packet - there should be
503 * more sanity checks in header chain processing.
505 if (m->m_pkthdr.len < off) {
506 IP6STAT_INC(ip6s_tooshort);
507 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
511 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
512 if (IPSEC_ENABLED(ipv6)) {
513 if (IPSEC_INPUT(ipv6, m, off, nxt) != 0)
518 nxt = ip6_protox[nxt](&m, &off, nxt);
527 ip6_input(struct mbuf *m)
529 struct in6_addr odst;
531 struct in6_ifaddr *ia;
534 u_int32_t rtalert = ~0;
535 int off = sizeof(struct ip6_hdr), nest;
540 * Drop the packet if IPv6 operation is disabled on the interface.
542 rcvif = m->m_pkthdr.rcvif;
543 if ((ND_IFINFO(rcvif)->flags & ND6_IFF_IFDISABLED))
546 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
548 * should the inner packet be considered authentic?
549 * see comment in ah4_input().
550 * NB: m cannot be NULL when passed to the input routine
553 m->m_flags &= ~M_AUTHIPHDR;
554 m->m_flags &= ~M_AUTHIPDGM;
558 if (m->m_flags & M_FASTFWD_OURS) {
560 * Firewall changed destination to local.
562 ip6 = mtod(m, struct ip6_hdr *);
569 if (m->m_flags & M_EXT) {
571 IP6STAT_INC(ip6s_mext2m);
573 IP6STAT_INC(ip6s_mext1);
576 struct ifnet *ifp = (m->m_flags & M_LOOP) ? V_loif : rcvif;
577 int ifindex = ifp->if_index;
578 if (ifindex >= IP6S_M2MMAX)
580 IP6STAT_INC(ip6s_m2m[ifindex]);
582 IP6STAT_INC(ip6s_m1);
585 in6_ifstat_inc(rcvif, ifs6_in_receive);
586 IP6STAT_INC(ip6s_total);
589 * L2 bridge code and some other code can return mbuf chain
590 * that does not conform to KAME requirement. too bad.
591 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
593 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
596 if (m->m_pkthdr.len > MHLEN)
597 n = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
599 n = m_gethdr(M_NOWAIT, MT_DATA);
604 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
605 n->m_len = n->m_pkthdr.len;
609 if (m->m_len < sizeof(struct ip6_hdr)) {
610 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
611 IP6STAT_INC(ip6s_toosmall);
612 in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
617 ip6 = mtod(m, struct ip6_hdr *);
618 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
619 IP6STAT_INC(ip6s_badvers);
620 in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
624 IP6STAT_INC(ip6s_nxthist[ip6->ip6_nxt]);
625 IP_PROBE(receive, NULL, NULL, ip6, rcvif, NULL, ip6);
628 * Check against address spoofing/corruption.
630 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
631 IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
633 * XXX: "badscope" is not very suitable for a multicast source.
635 IP6STAT_INC(ip6s_badscope);
636 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
639 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) &&
640 !(m->m_flags & M_LOOP)) {
642 * In this case, the packet should come from the loopback
643 * interface. However, we cannot just check the if_flags,
644 * because ip6_mloopback() passes the "actual" interface
645 * as the outgoing/incoming interface.
647 IP6STAT_INC(ip6s_badscope);
648 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
651 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
652 IPV6_ADDR_MC_SCOPE(&ip6->ip6_dst) == 0) {
655 * Nodes must not originate a packet to a multicast address
656 * whose scop field contains the reserved value 0; if such
657 * a packet is received, it must be silently dropped.
659 IP6STAT_INC(ip6s_badscope);
660 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
664 * The following check is not documented in specs. A malicious
665 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
666 * and bypass security checks (act as if it was from 127.0.0.1 by using
667 * IPv6 src ::ffff:127.0.0.1). Be cautious.
669 * We have supported IPv6-only kernels for a few years and this issue
670 * has not come up. The world seems to move mostly towards not using
671 * v4mapped on the wire, so it makes sense for us to keep rejecting
674 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
675 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
676 IP6STAT_INC(ip6s_badscope);
677 in6_ifstat_inc(rcvif, ifs6_in_addrerr);
682 * Reject packets with IPv4 compatible addresses (auto tunnel).
684 * The code forbids auto tunnel relay case in RFC1933 (the check is
685 * stronger than RFC1933). We may want to re-enable it if mech-xx
686 * is revised to forbid relaying case.
688 if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
689 IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
690 IP6STAT_INC(ip6s_badscope);
691 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
696 * Try to forward the packet, but if we fail continue.
697 * ip6_tryforward() does not generate redirects, so fall
698 * through to normal processing if redirects are required.
699 * ip6_tryforward() does inbound and outbound packet firewall
700 * processing. If firewall has decided that destination becomes
701 * our local address, it sets M_FASTFWD_OURS flag. In this
702 * case skip another inbound firewall processing and update
705 if (V_ip6_forwarding != 0 && V_ip6_sendredirects == 0
706 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
707 && (!IPSEC_ENABLED(ipv6) ||
708 IPSEC_CAPS(ipv6, m, IPSEC_CAP_OPERABLE) == 0)
711 if ((m = ip6_tryforward(m)) == NULL)
713 if (m->m_flags & M_FASTFWD_OURS) {
714 ip6 = mtod(m, struct ip6_hdr *);
718 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
720 * Bypass packet filtering for packets previously handled by IPsec.
722 if (IPSEC_ENABLED(ipv6) &&
723 IPSEC_CAPS(ipv6, m, IPSEC_CAP_BYPASS_FILTER) != 0)
727 * Run through list of hooks for input packets.
729 * NB: Beware of the destination address changing
730 * (e.g. by NAT rewriting). When this happens,
731 * tell ip6_forward to do the right thing.
734 /* Jump over all PFIL processing if hooks are not active. */
735 if (!PFIL_HOOKED_IN(V_inet6_pfil_head))
739 if (pfil_mbuf_in(V_inet6_pfil_head, &m, m->m_pkthdr.rcvif,
742 ip6 = mtod(m, struct ip6_hdr *);
743 srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
744 if ((m->m_flags & (M_IP6_NEXTHOP | M_FASTFWD_OURS)) == M_IP6_NEXTHOP &&
745 m_tag_find(m, PACKET_TAG_IPFORWARD, NULL) != NULL) {
747 * Directly ship the packet on. This allows forwarding
748 * packets originally destined to us to some other directly
757 * Disambiguate address scope zones (if there is ambiguity).
758 * We first make sure that the original source or destination address
759 * is not in our internal form for scoped addresses. Such addresses
760 * are not necessarily invalid spec-wise, but we cannot accept them due
761 * to the usage conflict.
762 * in6_setscope() then also checks and rejects the cases where src or
763 * dst are the loopback address and the receiving interface
766 if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
767 IP6STAT_INC(ip6s_badscope); /* XXX */
770 if (in6_setscope(&ip6->ip6_src, rcvif, NULL) ||
771 in6_setscope(&ip6->ip6_dst, rcvif, NULL)) {
772 IP6STAT_INC(ip6s_badscope);
775 if (m->m_flags & M_FASTFWD_OURS) {
776 m->m_flags &= ~M_FASTFWD_OURS;
781 * Multicast check. Assume packet is for us to avoid
782 * prematurely taking locks.
784 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
786 in6_ifstat_inc(rcvif, ifs6_in_mcast);
791 * XXX: For now we keep link-local IPv6 addresses with embedded
792 * scope zone id, therefore we use zero zoneid here.
794 ia = in6ifa_ifwithaddr(&ip6->ip6_dst, 0 /* XXX */, false);
796 if (ia->ia6_flags & IN6_IFF_NOTREADY) {
797 char ip6bufs[INET6_ADDRSTRLEN];
798 char ip6bufd[INET6_ADDRSTRLEN];
799 /* address is not ready, so discard the packet. */
801 "ip6_input: packet to an unready address %s->%s\n",
802 ip6_sprintf(ip6bufs, &ip6->ip6_src),
803 ip6_sprintf(ip6bufd, &ip6->ip6_dst)));
806 if (V_ip6_sav && !(m->m_flags & M_LOOP) &&
807 __predict_false(in6_localip_fib(&ip6->ip6_src,
809 IP6STAT_INC(ip6s_badscope); /* XXX */
812 /* Count the packet in the ip address stats */
813 counter_u64_add(ia->ia_ifa.ifa_ipackets, 1);
814 counter_u64_add(ia->ia_ifa.ifa_ibytes, m->m_pkthdr.len);
820 * Now there is no reason to process the packet if it's not our own
821 * and we're not a router.
823 if (!V_ip6_forwarding) {
824 IP6STAT_INC(ip6s_cantforward);
830 * Process Hop-by-Hop options header if it's contained.
831 * m may be modified in ip6_hopopts_input().
832 * If a JumboPayload option is included, plen will also be modified.
834 plen = (u_int32_t)ntohs(ip6->ip6_plen);
835 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
836 if (ip6_input_hbh(&m, &plen, &rtalert, &off, &nxt, &ours) != 0)
842 * Use mbuf flags to propagate Router Alert option to
843 * ICMPv6 layer, as hop-by-hop options have been stripped.
846 m->m_flags |= M_RTALERT_MLD;
849 * Check that the amount of data in the buffers
850 * is as at least much as the IPv6 header would have us expect.
851 * Trim mbufs if longer than we expect.
852 * Drop packet if shorter than we expect.
854 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
855 IP6STAT_INC(ip6s_tooshort);
856 in6_ifstat_inc(rcvif, ifs6_in_truncated);
859 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
860 if (m->m_len == m->m_pkthdr.len) {
861 m->m_len = sizeof(struct ip6_hdr) + plen;
862 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
864 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
868 * Forward if desirable.
871 IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
873 * If we are acting as a multicast router, all
874 * incoming multicast packets are passed to the
875 * kernel-level multicast forwarding function.
876 * The packet is returned (relatively) intact; if
877 * ip6_mforward() returns a non-zero value, the packet
878 * must be discarded, else it may be accepted below.
880 * XXX TODO: Check hlim and multicast scope here to avoid
881 * unnecessarily calling into ip6_mforward().
883 if (ip6_mforward && ip6_mforward(ip6, rcvif, m)) {
884 IP6STAT_INC(ip6s_cantforward);
888 ip6_forward(m, srcrt);
893 * We are going to ship the packet to the local protocol stack. Call the
894 * filter again for this 'output' action, allowing redirect-like rules
895 * to adjust the source address.
897 if (PFIL_HOOKED_OUT(V_inet6_local_pfil_head)) {
898 if (pfil_mbuf_out(V_inet6_local_pfil_head, &m, V_loif, NULL) !=
901 if (m == NULL) /* consumed by filter */
903 ip6 = mtod(m, struct ip6_hdr *);
907 * Tell launch routine the next header
909 IP6STAT_INC(ip6s_delivered);
910 in6_ifstat_inc(rcvif, ifs6_in_deliver);
913 while (nxt != IPPROTO_DONE) {
914 if (V_ip6_hdrnestlimit && (++nest > V_ip6_hdrnestlimit)) {
915 IP6STAT_INC(ip6s_toomanyhdr);
920 * protection against faulty packet - there should be
921 * more sanity checks in header chain processing.
923 if (m->m_pkthdr.len < off) {
924 IP6STAT_INC(ip6s_tooshort);
925 in6_ifstat_inc(rcvif, ifs6_in_truncated);
929 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
930 if (IPSEC_ENABLED(ipv6)) {
931 if (IPSEC_INPUT(ipv6, m, off, nxt) != 0)
936 nxt = ip6_protox[nxt](&m, &off, nxt);
940 in6_ifstat_inc(rcvif, ifs6_in_discard);
946 * Hop-by-Hop options header processing. If a valid jumbo payload option is
947 * included, the real payload length will be stored in plenp.
949 * rtalertp - XXX: should be stored more smart way
952 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
953 struct mbuf **mp, int *offp)
955 struct mbuf *m = *mp;
956 int off = *offp, hbhlen;
959 /* validation of the length of the header */
960 if (m->m_len < off + sizeof(*hbh)) {
961 m = m_pullup(m, off + sizeof(*hbh));
963 IP6STAT_INC(ip6s_exthdrtoolong);
968 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
969 hbhlen = (hbh->ip6h_len + 1) << 3;
971 if (m->m_len < off + hbhlen) {
972 m = m_pullup(m, off + hbhlen);
974 IP6STAT_INC(ip6s_exthdrtoolong);
979 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
981 hbhlen -= sizeof(struct ip6_hbh);
982 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
983 hbhlen, rtalertp, plenp) < 0) {
994 * Search header for all Hop-by-hop options and process each option.
995 * This function is separate from ip6_hopopts_input() in order to
996 * handle a case where the sending node itself process its hop-by-hop
997 * options header. In such a case, the function is called from ip6_output().
999 * The function assumes that hbh header is located right after the IPv6 header
1000 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
1001 * opthead + hbhlen is located in contiguous memory region.
1004 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
1005 u_int32_t *rtalertp, u_int32_t *plenp)
1007 struct ip6_hdr *ip6;
1009 u_int8_t *opt = opthead;
1010 u_int16_t rtalert_val;
1011 u_int32_t jumboplen;
1012 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
1014 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
1020 if (hbhlen < IP6OPT_MINLEN) {
1021 IP6STAT_INC(ip6s_toosmall);
1024 optlen = *(opt + 1) + 2;
1026 case IP6OPT_ROUTER_ALERT:
1027 /* XXX may need check for alignment */
1028 if (hbhlen < IP6OPT_RTALERT_LEN) {
1029 IP6STAT_INC(ip6s_toosmall);
1032 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
1034 icmp6_error(m, ICMP6_PARAM_PROB,
1035 ICMP6_PARAMPROB_HEADER,
1036 erroff + opt + 1 - opthead);
1039 optlen = IP6OPT_RTALERT_LEN;
1040 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
1041 *rtalertp = ntohs(rtalert_val);
1044 /* XXX may need check for alignment */
1045 if (hbhlen < IP6OPT_JUMBO_LEN) {
1046 IP6STAT_INC(ip6s_toosmall);
1049 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
1051 icmp6_error(m, ICMP6_PARAM_PROB,
1052 ICMP6_PARAMPROB_HEADER,
1053 erroff + opt + 1 - opthead);
1056 optlen = IP6OPT_JUMBO_LEN;
1059 * IPv6 packets that have non 0 payload length
1060 * must not contain a jumbo payload option.
1062 ip6 = mtod(m, struct ip6_hdr *);
1063 if (ip6->ip6_plen) {
1064 IP6STAT_INC(ip6s_badoptions);
1065 icmp6_error(m, ICMP6_PARAM_PROB,
1066 ICMP6_PARAMPROB_HEADER,
1067 erroff + opt - opthead);
1072 * We may see jumbolen in unaligned location, so
1073 * we'd need to perform bcopy().
1075 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1076 jumboplen = (u_int32_t)htonl(jumboplen);
1080 * if there are multiple jumbo payload options,
1081 * *plenp will be non-zero and the packet will be
1083 * the behavior may need some debate in ipngwg -
1084 * multiple options does not make sense, however,
1085 * there's no explicit mention in specification.
1088 IP6STAT_INC(ip6s_badoptions);
1089 icmp6_error(m, ICMP6_PARAM_PROB,
1090 ICMP6_PARAMPROB_HEADER,
1091 erroff + opt + 2 - opthead);
1097 * jumbo payload length must be larger than 65535.
1099 if (jumboplen <= IPV6_MAXPACKET) {
1100 IP6STAT_INC(ip6s_badoptions);
1101 icmp6_error(m, ICMP6_PARAM_PROB,
1102 ICMP6_PARAMPROB_HEADER,
1103 erroff + opt + 2 - opthead);
1109 default: /* unknown option */
1110 if (hbhlen < IP6OPT_MINLEN) {
1111 IP6STAT_INC(ip6s_toosmall);
1114 optlen = ip6_unknown_opt(opt, m,
1115 erroff + opt - opthead);
1131 * Unknown option processing.
1132 * The third argument `off' is the offset from the IPv6 header to the option,
1133 * which is necessary if the IPv6 header the and option header and IPv6 header
1134 * is not contiguous in order to return an ICMPv6 error.
1137 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
1139 struct ip6_hdr *ip6;
1141 switch (IP6OPT_TYPE(*optp)) {
1142 case IP6OPT_TYPE_SKIP: /* ignore the option */
1143 return ((int)*(optp + 1));
1144 case IP6OPT_TYPE_DISCARD: /* silently discard */
1147 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1148 IP6STAT_INC(ip6s_badoptions);
1149 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1151 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1152 IP6STAT_INC(ip6s_badoptions);
1153 ip6 = mtod(m, struct ip6_hdr *);
1154 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1155 (m->m_flags & (M_BCAST|M_MCAST)))
1158 icmp6_error(m, ICMP6_PARAM_PROB,
1159 ICMP6_PARAMPROB_OPTION, off);
1163 m_freem(m); /* XXX: NOTREACHED */
1168 * Create the "control" list for this pcb.
1169 * These functions will not modify mbuf chain at all.
1171 * The routine will be called from upper layer handlers like tcp6_input().
1172 * Thus the routine assumes that the caller (tcp6_input) have already
1173 * called m_pullup() and all the extension headers are located in the
1174 * very first mbuf on the mbuf chain.
1176 * ip6_savecontrol_v4 will handle those options that are possible to be
1177 * set on a v4-mapped socket.
1178 * ip6_savecontrol will directly call ip6_savecontrol_v4 to handle those
1179 * options and handle the v6-only ones itself.
1182 ip6_savecontrol_v4(struct inpcb *inp, struct mbuf *m, struct mbuf **mp,
1185 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1188 if ((inp->inp_socket->so_options & SO_TIMESTAMP) != 0) {
1194 struct bintime boottimebin, bt1;
1195 struct timespec ts1;
1199 switch (inp->inp_socket->so_ts_clock) {
1200 case SO_TS_REALTIME_MICRO:
1201 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1203 mbuf_tstmp2timespec(m, &ts1);
1204 timespec2bintime(&ts1, &bt1);
1205 getboottimebin(&boottimebin);
1206 bintime_add(&bt1, &boottimebin);
1207 bintime2timeval(&bt1, &t.tv);
1211 *mp = sbcreatecontrol(&t.tv, sizeof(t.tv),
1212 SCM_TIMESTAMP, SOL_SOCKET, M_NOWAIT);
1214 mp = &(*mp)->m_next;
1220 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1222 mbuf_tstmp2timespec(m, &ts1);
1223 timespec2bintime(&ts1, &t.bt);
1224 getboottimebin(&boottimebin);
1225 bintime_add(&t.bt, &boottimebin);
1229 *mp = sbcreatecontrol(&t.bt, sizeof(t.bt), SCM_BINTIME,
1230 SOL_SOCKET, M_NOWAIT);
1232 mp = &(*mp)->m_next;
1237 case SO_TS_REALTIME:
1238 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1240 mbuf_tstmp2timespec(m, &t.ts);
1241 getboottimebin(&boottimebin);
1242 bintime2timespec(&boottimebin, &ts1);
1243 timespecadd(&t.ts, &ts1, &t.ts);
1247 *mp = sbcreatecontrol(&t.ts, sizeof(t.ts),
1248 SCM_REALTIME, SOL_SOCKET, M_NOWAIT);
1250 mp = &(*mp)->m_next;
1255 case SO_TS_MONOTONIC:
1256 if ((m->m_flags & (M_PKTHDR | M_TSTMP)) == (M_PKTHDR |
1258 mbuf_tstmp2timespec(m, &t.ts);
1261 *mp = sbcreatecontrol(&t.ts, sizeof(t.ts),
1262 SCM_MONOTONIC, SOL_SOCKET, M_NOWAIT);
1264 mp = &(*mp)->m_next;
1270 panic("unknown (corrupted) so_ts_clock");
1272 if (stamped && (m->m_flags & (M_PKTHDR | M_TSTMP)) ==
1273 (M_PKTHDR | M_TSTMP)) {
1274 struct sock_timestamp_info sti;
1276 bzero(&sti, sizeof(sti));
1277 sti.st_info_flags = ST_INFO_HW;
1278 if ((m->m_flags & M_TSTMP_HPREC) != 0)
1279 sti.st_info_flags |= ST_INFO_HW_HPREC;
1280 *mp = sbcreatecontrol(&sti, sizeof(sti), SCM_TIME_INFO,
1281 SOL_SOCKET, M_NOWAIT);
1283 mp = &(*mp)->m_next;
1288 #define IS2292(inp, x, y) (((inp)->inp_flags & IN6P_RFC2292) ? (x) : (y))
1289 /* RFC 2292 sec. 5 */
1290 if ((inp->inp_flags & IN6P_PKTINFO) != 0) {
1291 struct in6_pktinfo pi6;
1293 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1297 ip = mtod(m, struct ip *);
1298 pi6.ipi6_addr.s6_addr32[0] = 0;
1299 pi6.ipi6_addr.s6_addr32[1] = 0;
1300 pi6.ipi6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP;
1301 pi6.ipi6_addr.s6_addr32[3] = ip->ip_dst.s_addr;
1303 /* We won't hit this code */
1304 bzero(&pi6.ipi6_addr, sizeof(struct in6_addr));
1307 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1308 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1311 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1313 *mp = sbcreatecontrol(&pi6, sizeof(struct in6_pktinfo),
1314 IS2292(inp, IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6,
1317 mp = &(*mp)->m_next;
1320 if ((inp->inp_flags & IN6P_HOPLIMIT) != 0) {
1323 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1327 ip = mtod(m, struct ip *);
1330 /* We won't hit this code */
1334 hlim = ip6->ip6_hlim & 0xff;
1336 *mp = sbcreatecontrol(&hlim, sizeof(int),
1337 IS2292(inp, IPV6_2292HOPLIMIT, IPV6_HOPLIMIT),
1338 IPPROTO_IPV6, M_NOWAIT);
1340 mp = &(*mp)->m_next;
1343 if ((inp->inp_flags & IN6P_TCLASS) != 0) {
1346 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1350 ip = mtod(m, struct ip *);
1351 tclass = ip->ip_tos;
1353 /* We won't hit this code */
1359 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1361 tclass = flowinfo & 0xff;
1363 *mp = sbcreatecontrol(&tclass, sizeof(int), IPV6_TCLASS,
1364 IPPROTO_IPV6, M_NOWAIT);
1366 mp = &(*mp)->m_next;
1369 if (v4only != NULL) {
1370 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1381 ip6_savecontrol(struct inpcb *inp, struct mbuf *m, struct mbuf **mp)
1383 struct ip6_hdr *ip6;
1386 mp = ip6_savecontrol_v4(inp, m, mp, &v4only);
1390 ip6 = mtod(m, struct ip6_hdr *);
1392 * IPV6_HOPOPTS socket option. Recall that we required super-user
1393 * privilege for the option (see ip6_ctloutput), but it might be too
1394 * strict, since there might be some hop-by-hop options which can be
1395 * returned to normal user.
1396 * See also RFC 2292 section 6 (or RFC 3542 section 8).
1398 if ((inp->inp_flags & IN6P_HOPOPTS) != 0) {
1400 * Check if a hop-by-hop options header is contatined in the
1401 * received packet, and if so, store the options as ancillary
1402 * data. Note that a hop-by-hop options header must be
1403 * just after the IPv6 header, which is assured through the
1404 * IPv6 input processing.
1406 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1407 struct ip6_hbh *hbh;
1410 hbh = (struct ip6_hbh *)(ip6 + 1);
1411 hbhlen = (hbh->ip6h_len + 1) << 3;
1414 * XXX: We copy the whole header even if a
1415 * jumbo payload option is included, the option which
1416 * is to be removed before returning according to
1418 * Note: this constraint is removed in RFC3542
1420 *mp = sbcreatecontrol(hbh, hbhlen,
1421 IS2292(inp, IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1422 IPPROTO_IPV6, M_NOWAIT);
1424 mp = &(*mp)->m_next;
1428 if ((inp->inp_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
1429 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1432 * Search for destination options headers or routing
1433 * header(s) through the header chain, and stores each
1434 * header as ancillary data.
1435 * Note that the order of the headers remains in
1436 * the chain of ancillary data.
1438 while (1) { /* is explicit loop prevention necessary? */
1439 struct ip6_ext *ip6e = NULL;
1443 * if it is not an extension header, don't try to
1444 * pull it from the chain.
1447 case IPPROTO_DSTOPTS:
1448 case IPPROTO_ROUTING:
1449 case IPPROTO_HOPOPTS:
1450 case IPPROTO_AH: /* is it possible? */
1456 if (off + sizeof(*ip6e) > m->m_len)
1458 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1459 if (nxt == IPPROTO_AH)
1460 elen = (ip6e->ip6e_len + 2) << 2;
1462 elen = (ip6e->ip6e_len + 1) << 3;
1463 if (off + elen > m->m_len)
1467 case IPPROTO_DSTOPTS:
1468 if (!(inp->inp_flags & IN6P_DSTOPTS))
1471 *mp = sbcreatecontrol(ip6e, elen,
1472 IS2292(inp, IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1473 IPPROTO_IPV6, M_NOWAIT);
1475 mp = &(*mp)->m_next;
1477 case IPPROTO_ROUTING:
1478 if (!(inp->inp_flags & IN6P_RTHDR))
1481 *mp = sbcreatecontrol(ip6e, elen,
1482 IS2292(inp, IPV6_2292RTHDR, IPV6_RTHDR),
1483 IPPROTO_IPV6, M_NOWAIT);
1485 mp = &(*mp)->m_next;
1487 case IPPROTO_HOPOPTS:
1488 case IPPROTO_AH: /* is it possible? */
1493 * other cases have been filtered in the above.
1494 * none will visit this case. here we supply
1495 * the code just in case (nxt overwritten or
1501 /* proceed with the next header. */
1503 nxt = ip6e->ip6e_nxt;
1510 if (inp->inp_flags2 & INP_RECVFLOWID) {
1511 uint32_t flowid, flow_type;
1513 flowid = m->m_pkthdr.flowid;
1514 flow_type = M_HASHTYPE_GET(m);
1517 * XXX should handle the failure of one or the
1518 * other - don't populate both?
1520 *mp = sbcreatecontrol(&flowid, sizeof(uint32_t), IPV6_FLOWID,
1521 IPPROTO_IPV6, M_NOWAIT);
1523 mp = &(*mp)->m_next;
1524 *mp = sbcreatecontrol(&flow_type, sizeof(uint32_t),
1525 IPV6_FLOWTYPE, IPPROTO_IPV6, M_NOWAIT);
1527 mp = &(*mp)->m_next;
1531 if (inp->inp_flags2 & INP_RECVRSSBUCKETID) {
1532 uint32_t flowid, flow_type;
1533 uint32_t rss_bucketid;
1535 flowid = m->m_pkthdr.flowid;
1536 flow_type = M_HASHTYPE_GET(m);
1538 if (rss_hash2bucket(flowid, flow_type, &rss_bucketid) == 0) {
1539 *mp = sbcreatecontrol(&rss_bucketid, sizeof(uint32_t),
1540 IPV6_RSSBUCKETID, IPPROTO_IPV6, M_NOWAIT);
1542 mp = &(*mp)->m_next;
1551 ip6_notify_pmtu(struct inpcb *inp, struct sockaddr_in6 *dst, u_int32_t mtu)
1555 struct ip6_mtuinfo mtuctl;
1557 KASSERT(inp != NULL, ("%s: inp == NULL", __func__));
1559 * Notify the error by sending IPV6_PATHMTU ancillary data if
1560 * application wanted to know the MTU value.
1561 * NOTE: we notify disconnected sockets, because some udp
1562 * applications keep sending sockets disconnected.
1563 * NOTE: our implementation doesn't notify connected sockets that has
1564 * foreign address that is different than given destination addresses
1565 * (this is permitted by RFC 3542).
1567 if ((inp->inp_flags & IN6P_MTU) == 0 || (
1568 !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
1569 !IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &dst->sin6_addr)))
1572 mtuctl.ip6m_mtu = mtu;
1573 mtuctl.ip6m_addr = *dst;
1574 if (sa6_recoverscope(&mtuctl.ip6m_addr))
1577 if ((m_mtu = sbcreatecontrol(&mtuctl, sizeof(mtuctl), IPV6_PATHMTU,
1578 IPPROTO_IPV6, M_NOWAIT)) == NULL)
1581 so = inp->inp_socket;
1582 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu)
1586 /* XXX: should count statistics */
1592 * Get pointer to the previous header followed by the header
1593 * currently processed.
1596 ip6_get_prevhdr(const struct mbuf *m, int off)
1598 struct ip6_ext ip6e;
1599 struct ip6_hdr *ip6;
1602 if (off == sizeof(struct ip6_hdr))
1603 return (offsetof(struct ip6_hdr, ip6_nxt));
1604 if (off < sizeof(struct ip6_hdr))
1605 panic("%s: off < sizeof(struct ip6_hdr)", __func__);
1607 ip6 = mtod(m, struct ip6_hdr *);
1609 len = sizeof(struct ip6_hdr);
1612 m_copydata(m, len, sizeof(ip6e), (caddr_t)&ip6e);
1614 case IPPROTO_FRAGMENT:
1615 nlen = sizeof(struct ip6_frag);
1618 nlen = (ip6e.ip6e_len + 2) << 2;
1621 nlen = (ip6e.ip6e_len + 1) << 3;
1624 nxt = ip6e.ip6e_nxt;
1626 return (len - nlen);
1630 * get next header offset. m will be retained.
1633 ip6_nexthdr(const struct mbuf *m, int off, int proto, int *nxtp)
1636 struct ip6_ext ip6e;
1641 panic("ip6_nexthdr: m == NULL");
1642 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1647 if (m->m_pkthdr.len < off + sizeof(ip6))
1649 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1651 *nxtp = ip6.ip6_nxt;
1655 case IPPROTO_FRAGMENT:
1657 * terminate parsing if it is not the first fragment,
1658 * it does not make sense to parse through it.
1660 if (m->m_pkthdr.len < off + sizeof(fh))
1662 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1663 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1664 if (fh.ip6f_offlg & IP6F_OFF_MASK)
1667 *nxtp = fh.ip6f_nxt;
1668 off += sizeof(struct ip6_frag);
1672 if (m->m_pkthdr.len < off + sizeof(ip6e))
1674 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1676 *nxtp = ip6e.ip6e_nxt;
1677 off += (ip6e.ip6e_len + 2) << 2;
1680 case IPPROTO_HOPOPTS:
1681 case IPPROTO_ROUTING:
1682 case IPPROTO_DSTOPTS:
1683 if (m->m_pkthdr.len < off + sizeof(ip6e))
1685 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1687 *nxtp = ip6e.ip6e_nxt;
1688 off += (ip6e.ip6e_len + 1) << 3;
1693 case IPPROTO_IPCOMP:
1705 * get offset for the last header in the chain. m will be kept untainted.
1708 ip6_lasthdr(const struct mbuf *m, int off, int proto, int *nxtp)
1718 newoff = ip6_nexthdr(m, off, proto, nxtp);
1721 else if (newoff < off)
1722 return -1; /* invalid */
1723 else if (newoff == off)