2/* $KAME: ip6_input.c,v 1.259 2002/01/21 04:58:09 jinmei Exp $ */
3
4/*
5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the project nor the names of its contributors
17 * may be used to endorse or promote products derived from this software
18 * without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 */
32
33/*
34 * Copyright (c) 1982, 1986, 1988, 1993
35 * The Regents of the University of California. All rights reserved.
36 *
37 * Redistribution and use in source and binary forms, with or without
38 * modification, are permitted provided that the following conditions
39 * are met:
40 * 1. Redistributions of source code must retain the above copyright
41 * notice, this list of conditions and the following disclaimer.
42 * 2. Redistributions in binary form must reproduce the above copyright
43 * notice, this list of conditions and the following disclaimer in the
44 * documentation and/or other materials provided with the distribution.
45 * 3. All advertising materials mentioning features or use of this software
46 * must display the following acknowledgement:
47 * This product includes software developed by the University of
48 * California, Berkeley and its contributors.
49 * 4. Neither the name of the University nor the names of its contributors
50 * may be used to endorse or promote products derived from this software
51 * without specific prior written permission.
52 *
53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
56 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
63 * SUCH DAMAGE.
64 *
65 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
66 */
67
68#include "opt_ip6fw.h"
69#include "opt_inet.h"
70#include "opt_inet6.h"
71#include "opt_ipsec.h"
72#include "opt_pfil_hooks.h"
73#include "opt_random_ip_id.h"
74
75#include <sys/param.h>
76#include <sys/systm.h>
77#include <sys/malloc.h>
78#include <sys/mbuf.h>
79#include <sys/proc.h>
80#include <sys/domain.h>
81#include <sys/protosw.h>
82#include <sys/socket.h>
83#include <sys/socketvar.h>
84#include <sys/errno.h>
85#include <sys/time.h>
86#include <sys/kernel.h>
87#include <sys/syslog.h>
88
89#include <net/if.h>
90#include <net/if_types.h>
91#include <net/if_dl.h>
92#include <net/route.h>
93#include <net/netisr.h>
94#ifdef PFIL_HOOKS
95#include <net/pfil.h>
96#endif
97
98#include <netinet/in.h>
99#include <netinet/in_systm.h>
100#ifdef INET
101#include <netinet/ip.h>
102#include <netinet/ip_icmp.h>
103#endif /* INET */
104#include <netinet/ip6.h>
105#include <netinet6/in6_var.h>
106#include <netinet6/ip6_var.h>
107#include <netinet/in_pcb.h>
108#include <netinet/icmp6.h>
109#include <netinet6/scope6_var.h>
110#include <netinet6/in6_ifattach.h>
111#include <netinet6/nd6.h>
112#include <netinet6/in6_prefix.h>
113
114#ifdef IPSEC
115#include <netinet6/ipsec.h>
116#ifdef INET6
117#include <netinet6/ipsec6.h>
118#endif
119#endif
120
121#ifdef FAST_IPSEC
122#include <netipsec/ipsec.h>
123#include <netipsec/ipsec6.h>
124#define IPSEC
125#endif /* FAST_IPSEC */
126
127#include <netinet6/ip6_fw.h>
128
129#include <netinet6/ip6protosw.h>
130
131#include <net/net_osdep.h>
132
133extern struct domain inet6domain;
134
135u_char ip6_protox[IPPROTO_MAX];
136static struct ifqueue ip6intrq;
137static int ip6qmaxlen = IFQ_MAXLEN;
138struct in6_ifaddr *in6_ifaddr;
139
140extern struct callout in6_tmpaddrtimer_ch;
141
142int ip6_forward_srcrt; /* XXX */
143int ip6_sourcecheck; /* XXX */
144int ip6_sourcecheck_interval; /* XXX */
145
146int ip6_ours_check_algorithm;
147
148#ifdef PFIL_HOOKS
149struct pfil_head inet6_pfil_hook;
150#endif
151
152/* firewall hooks */
153ip6_fw_chk_t *ip6_fw_chk_ptr;
154ip6_fw_ctl_t *ip6_fw_ctl_ptr;
155int ip6_fw_enable = 1;
156
157struct ip6stat ip6stat;
158
159static void ip6_init2 __P((void *));
160static struct ip6aux *ip6_setdstifaddr __P((struct mbuf *, struct in6_ifaddr *));
161static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *));
162#ifdef PULLDOWN_TEST
163static struct mbuf *ip6_pullexthdr __P((struct mbuf *, size_t, int));
164#endif
165
166/*
167 * IP6 initialization: fill in IP6 protocol switch table.
168 * All protocols not implemented in kernel go to raw IP6 protocol handler.
169 */
170void
171ip6_init()
172{
173 struct ip6protosw *pr;
174 int i;
175
176#ifdef DIAGNOSTIC
177 if (sizeof(struct protosw) != sizeof(struct ip6protosw))
178 panic("sizeof(protosw) != sizeof(ip6protosw)");
179#endif
180 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
181 if (pr == 0)
182 panic("ip6_init");
183 for (i = 0; i < IPPROTO_MAX; i++)
184 ip6_protox[i] = pr - inet6sw;
185 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
186 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
187 if (pr->pr_domain->dom_family == PF_INET6 &&
188 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
189 ip6_protox[pr->pr_protocol] = pr - inet6sw;
190#ifdef PFIL_HOOKS
191 inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
192 inet6_pfil_hook.ph_af = AF_INET6;
193 if ((i = pfil_head_register(&inet6_pfil_hook)) != 0)
194 printf("%s: WARNING: unable to register pfil hook, "
195 "error %d\n", __func__, i);
196#endif /* PFIL_HOOKS */
197 ip6intrq.ifq_maxlen = ip6qmaxlen;
198 mtx_init(&ip6intrq.ifq_mtx, "ip6_inq", NULL, MTX_DEF);
199 netisr_register(NETISR_IPV6, ip6_input, &ip6intrq, 0);
200 scope6_init();
201 addrsel_policy_init();
202 nd6_init();
203 frag6_init();
204#ifndef RANDOM_IP_ID
205 ip6_flow_seq = arc4random();
206#endif
207 ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
208}
209
210static void
211ip6_init2(dummy)
212 void *dummy;
213{
214
215 /* nd6_timer_init */
216 callout_init(&nd6_timer_ch, 0);
217 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
218
219 /* router renumbering prefix list maintenance */
220 callout_init(&in6_rr_timer_ch, 0);
221 callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL);
222
223 /* timer for regeneranation of temporary addresses randomize ID */
224 callout_init(&in6_tmpaddrtimer_ch, 0);
225 callout_reset(&in6_tmpaddrtimer_ch,
226 (ip6_temp_preferred_lifetime - ip6_desync_factor -
227 ip6_temp_regen_advance) * hz,
228 in6_tmpaddrtimer, NULL);
229}
230
231/* cheat */
232/* This must be after route_init(), which is now SI_ORDER_THIRD */
233SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL);
234
235extern struct route_in6 ip6_forward_rt;
236
237void
238ip6_input(m)
239 struct mbuf *m;
240{
241 struct ip6_hdr *ip6;
242 int off = sizeof(struct ip6_hdr), nest;
243 u_int32_t plen;
244 u_int32_t rtalert = ~0;
245 int nxt, ours = 0;
246 struct ifnet *deliverifp = NULL;
247 struct sockaddr_in6 sa6;
248 u_int32_t srczone, dstzone;
249#ifdef PFIL_HOOKS
250 struct in6_addr odst;
251#endif
252 int srcrt = 0;
253
254 GIANT_REQUIRED; /* XXX for now */
255#ifdef IPSEC
256 /*
257 * should the inner packet be considered authentic?
258 * see comment in ah4_input().
259 */
260 if (m) {
261 m->m_flags &= ~M_AUTHIPHDR;
262 m->m_flags &= ~M_AUTHIPDGM;
263 }
264#endif
265
266 /*
267 * make sure we don't have onion peering information into m_tag.
268 */
269 ip6_delaux(m);
270
271 /*
272 * mbuf statistics
273 */
274 if (m->m_flags & M_EXT) {
275 if (m->m_next)
276 ip6stat.ip6s_mext2m++;
277 else
278 ip6stat.ip6s_mext1++;
279 } else {
280#define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0]))
281 if (m->m_next) {
282 if (m->m_flags & M_LOOP) {
283 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */
284 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
285 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
286 else
287 ip6stat.ip6s_m2m[0]++;
288 } else
289 ip6stat.ip6s_m1++;
290#undef M2MMAX
291 }
292
293 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
294 ip6stat.ip6s_total++;
295
296#ifndef PULLDOWN_TEST
297 /*
298 * L2 bridge code and some other code can return mbuf chain
299 * that does not conform to KAME requirement. too bad.
300 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
301 */
302 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
303 struct mbuf *n;
304
305 MGETHDR(n, M_DONTWAIT, MT_HEADER);
306 if (n)
307 M_MOVE_PKTHDR(n, m);
308 if (n && n->m_pkthdr.len > MHLEN) {
309 MCLGET(n, M_DONTWAIT);
310 if ((n->m_flags & M_EXT) == 0) {
311 m_freem(n);
312 n = NULL;
313 }
314 }
315 if (n == NULL) {
316 m_freem(m);
317 return; /* ENOBUFS */
318 }
319
320 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
321 n->m_len = n->m_pkthdr.len;
322 m_freem(m);
323 m = n;
324 }
325 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */);
326#endif
327
328 if (m->m_len < sizeof(struct ip6_hdr)) {
329 struct ifnet *inifp;
330 inifp = m->m_pkthdr.rcvif;
331 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
332 ip6stat.ip6s_toosmall++;
333 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
334 return;
335 }
336 }
337
338 ip6 = mtod(m, struct ip6_hdr *);
339
340 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
341 ip6stat.ip6s_badvers++;
342 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
343 goto bad;
344 }
345
| 2/* $KAME: ip6_input.c,v 1.259 2002/01/21 04:58:09 jinmei Exp $ */
3
4/*
5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the project nor the names of its contributors
17 * may be used to endorse or promote products derived from this software
18 * without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 */
32
33/*
34 * Copyright (c) 1982, 1986, 1988, 1993
35 * The Regents of the University of California. All rights reserved.
36 *
37 * Redistribution and use in source and binary forms, with or without
38 * modification, are permitted provided that the following conditions
39 * are met:
40 * 1. Redistributions of source code must retain the above copyright
41 * notice, this list of conditions and the following disclaimer.
42 * 2. Redistributions in binary form must reproduce the above copyright
43 * notice, this list of conditions and the following disclaimer in the
44 * documentation and/or other materials provided with the distribution.
45 * 3. All advertising materials mentioning features or use of this software
46 * must display the following acknowledgement:
47 * This product includes software developed by the University of
48 * California, Berkeley and its contributors.
49 * 4. Neither the name of the University nor the names of its contributors
50 * may be used to endorse or promote products derived from this software
51 * without specific prior written permission.
52 *
53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
56 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
63 * SUCH DAMAGE.
64 *
65 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
66 */
67
68#include "opt_ip6fw.h"
69#include "opt_inet.h"
70#include "opt_inet6.h"
71#include "opt_ipsec.h"
72#include "opt_pfil_hooks.h"
73#include "opt_random_ip_id.h"
74
75#include <sys/param.h>
76#include <sys/systm.h>
77#include <sys/malloc.h>
78#include <sys/mbuf.h>
79#include <sys/proc.h>
80#include <sys/domain.h>
81#include <sys/protosw.h>
82#include <sys/socket.h>
83#include <sys/socketvar.h>
84#include <sys/errno.h>
85#include <sys/time.h>
86#include <sys/kernel.h>
87#include <sys/syslog.h>
88
89#include <net/if.h>
90#include <net/if_types.h>
91#include <net/if_dl.h>
92#include <net/route.h>
93#include <net/netisr.h>
94#ifdef PFIL_HOOKS
95#include <net/pfil.h>
96#endif
97
98#include <netinet/in.h>
99#include <netinet/in_systm.h>
100#ifdef INET
101#include <netinet/ip.h>
102#include <netinet/ip_icmp.h>
103#endif /* INET */
104#include <netinet/ip6.h>
105#include <netinet6/in6_var.h>
106#include <netinet6/ip6_var.h>
107#include <netinet/in_pcb.h>
108#include <netinet/icmp6.h>
109#include <netinet6/scope6_var.h>
110#include <netinet6/in6_ifattach.h>
111#include <netinet6/nd6.h>
112#include <netinet6/in6_prefix.h>
113
114#ifdef IPSEC
115#include <netinet6/ipsec.h>
116#ifdef INET6
117#include <netinet6/ipsec6.h>
118#endif
119#endif
120
121#ifdef FAST_IPSEC
122#include <netipsec/ipsec.h>
123#include <netipsec/ipsec6.h>
124#define IPSEC
125#endif /* FAST_IPSEC */
126
127#include <netinet6/ip6_fw.h>
128
129#include <netinet6/ip6protosw.h>
130
131#include <net/net_osdep.h>
132
133extern struct domain inet6domain;
134
135u_char ip6_protox[IPPROTO_MAX];
136static struct ifqueue ip6intrq;
137static int ip6qmaxlen = IFQ_MAXLEN;
138struct in6_ifaddr *in6_ifaddr;
139
140extern struct callout in6_tmpaddrtimer_ch;
141
142int ip6_forward_srcrt; /* XXX */
143int ip6_sourcecheck; /* XXX */
144int ip6_sourcecheck_interval; /* XXX */
145
146int ip6_ours_check_algorithm;
147
148#ifdef PFIL_HOOKS
149struct pfil_head inet6_pfil_hook;
150#endif
151
152/* firewall hooks */
153ip6_fw_chk_t *ip6_fw_chk_ptr;
154ip6_fw_ctl_t *ip6_fw_ctl_ptr;
155int ip6_fw_enable = 1;
156
157struct ip6stat ip6stat;
158
159static void ip6_init2 __P((void *));
160static struct ip6aux *ip6_setdstifaddr __P((struct mbuf *, struct in6_ifaddr *));
161static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *));
162#ifdef PULLDOWN_TEST
163static struct mbuf *ip6_pullexthdr __P((struct mbuf *, size_t, int));
164#endif
165
166/*
167 * IP6 initialization: fill in IP6 protocol switch table.
168 * All protocols not implemented in kernel go to raw IP6 protocol handler.
169 */
170void
171ip6_init()
172{
173 struct ip6protosw *pr;
174 int i;
175
176#ifdef DIAGNOSTIC
177 if (sizeof(struct protosw) != sizeof(struct ip6protosw))
178 panic("sizeof(protosw) != sizeof(ip6protosw)");
179#endif
180 pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
181 if (pr == 0)
182 panic("ip6_init");
183 for (i = 0; i < IPPROTO_MAX; i++)
184 ip6_protox[i] = pr - inet6sw;
185 for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
186 pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
187 if (pr->pr_domain->dom_family == PF_INET6 &&
188 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
189 ip6_protox[pr->pr_protocol] = pr - inet6sw;
190#ifdef PFIL_HOOKS
191 inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
192 inet6_pfil_hook.ph_af = AF_INET6;
193 if ((i = pfil_head_register(&inet6_pfil_hook)) != 0)
194 printf("%s: WARNING: unable to register pfil hook, "
195 "error %d\n", __func__, i);
196#endif /* PFIL_HOOKS */
197 ip6intrq.ifq_maxlen = ip6qmaxlen;
198 mtx_init(&ip6intrq.ifq_mtx, "ip6_inq", NULL, MTX_DEF);
199 netisr_register(NETISR_IPV6, ip6_input, &ip6intrq, 0);
200 scope6_init();
201 addrsel_policy_init();
202 nd6_init();
203 frag6_init();
204#ifndef RANDOM_IP_ID
205 ip6_flow_seq = arc4random();
206#endif
207 ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
208}
209
210static void
211ip6_init2(dummy)
212 void *dummy;
213{
214
215 /* nd6_timer_init */
216 callout_init(&nd6_timer_ch, 0);
217 callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
218
219 /* router renumbering prefix list maintenance */
220 callout_init(&in6_rr_timer_ch, 0);
221 callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL);
222
223 /* timer for regeneranation of temporary addresses randomize ID */
224 callout_init(&in6_tmpaddrtimer_ch, 0);
225 callout_reset(&in6_tmpaddrtimer_ch,
226 (ip6_temp_preferred_lifetime - ip6_desync_factor -
227 ip6_temp_regen_advance) * hz,
228 in6_tmpaddrtimer, NULL);
229}
230
231/* cheat */
232/* This must be after route_init(), which is now SI_ORDER_THIRD */
233SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL);
234
235extern struct route_in6 ip6_forward_rt;
236
237void
238ip6_input(m)
239 struct mbuf *m;
240{
241 struct ip6_hdr *ip6;
242 int off = sizeof(struct ip6_hdr), nest;
243 u_int32_t plen;
244 u_int32_t rtalert = ~0;
245 int nxt, ours = 0;
246 struct ifnet *deliverifp = NULL;
247 struct sockaddr_in6 sa6;
248 u_int32_t srczone, dstzone;
249#ifdef PFIL_HOOKS
250 struct in6_addr odst;
251#endif
252 int srcrt = 0;
253
254 GIANT_REQUIRED; /* XXX for now */
255#ifdef IPSEC
256 /*
257 * should the inner packet be considered authentic?
258 * see comment in ah4_input().
259 */
260 if (m) {
261 m->m_flags &= ~M_AUTHIPHDR;
262 m->m_flags &= ~M_AUTHIPDGM;
263 }
264#endif
265
266 /*
267 * make sure we don't have onion peering information into m_tag.
268 */
269 ip6_delaux(m);
270
271 /*
272 * mbuf statistics
273 */
274 if (m->m_flags & M_EXT) {
275 if (m->m_next)
276 ip6stat.ip6s_mext2m++;
277 else
278 ip6stat.ip6s_mext1++;
279 } else {
280#define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0]))
281 if (m->m_next) {
282 if (m->m_flags & M_LOOP) {
283 ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */
284 } else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
285 ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
286 else
287 ip6stat.ip6s_m2m[0]++;
288 } else
289 ip6stat.ip6s_m1++;
290#undef M2MMAX
291 }
292
293 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
294 ip6stat.ip6s_total++;
295
296#ifndef PULLDOWN_TEST
297 /*
298 * L2 bridge code and some other code can return mbuf chain
299 * that does not conform to KAME requirement. too bad.
300 * XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
301 */
302 if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
303 struct mbuf *n;
304
305 MGETHDR(n, M_DONTWAIT, MT_HEADER);
306 if (n)
307 M_MOVE_PKTHDR(n, m);
308 if (n && n->m_pkthdr.len > MHLEN) {
309 MCLGET(n, M_DONTWAIT);
310 if ((n->m_flags & M_EXT) == 0) {
311 m_freem(n);
312 n = NULL;
313 }
314 }
315 if (n == NULL) {
316 m_freem(m);
317 return; /* ENOBUFS */
318 }
319
320 m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
321 n->m_len = n->m_pkthdr.len;
322 m_freem(m);
323 m = n;
324 }
325 IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */);
326#endif
327
328 if (m->m_len < sizeof(struct ip6_hdr)) {
329 struct ifnet *inifp;
330 inifp = m->m_pkthdr.rcvif;
331 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
332 ip6stat.ip6s_toosmall++;
333 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
334 return;
335 }
336 }
337
338 ip6 = mtod(m, struct ip6_hdr *);
339
340 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
341 ip6stat.ip6s_badvers++;
342 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
343 goto bad;
344 }
345
|
460 * construct source and destination address structures with
461 * disambiguating their scope zones (if there is ambiguity).
462 * XXX: sin6_family and sin6_len will NOT be referred to, but we fill
463 * in these fields just in case.
464 */
465 if (in6_addr2zoneid(m->m_pkthdr.rcvif, &ip6->ip6_src, &srczone) ||
466 in6_addr2zoneid(m->m_pkthdr.rcvif, &ip6->ip6_dst, &dstzone)) {
467 /*
468 * Note that these generic checks cover cases that src or
469 * dst are the loopback address and the receiving interface
470 * is not loopback.
471 */
472 ip6stat.ip6s_badscope++;
473 goto bad;
474 }
475
476 bzero(&sa6, sizeof(sa6));
477 sa6.sin6_family = AF_INET6;
478 sa6.sin6_len = sizeof(struct sockaddr_in6);
479
480 sa6.sin6_addr = ip6->ip6_src;
481 sa6.sin6_scope_id = srczone;
482 if (in6_embedscope(&ip6->ip6_src, &sa6, NULL, NULL)) {
483 /* XXX: should not happen */
484 ip6stat.ip6s_badscope++;
485 goto bad;
486 }
487
488 sa6.sin6_addr = ip6->ip6_dst;
489 sa6.sin6_scope_id = dstzone;
490 if (in6_embedscope(&ip6->ip6_dst, &sa6, NULL, NULL)) {
491 /* XXX: should not happen */
492 ip6stat.ip6s_badscope++;
493 goto bad;
494 }
495
496 /* XXX: ff01::%ifN awareness is not merged, yet. */
497 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_src))
498 ip6->ip6_src.s6_addr16[1] = 0;
499 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst))
500 ip6->ip6_dst.s6_addr16[1] = 0;
501
502 /*
503 * Multicast check
504 */
505 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
506 struct in6_multi *in6m = 0;
507
508 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
509 /*
510 * See if we belong to the destination multicast group on the
511 * arrival interface.
512 */
513 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
514 if (in6m)
515 ours = 1;
516 else if (!ip6_mrouter) {
517 ip6stat.ip6s_notmember++;
518 ip6stat.ip6s_cantforward++;
519 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
520 goto bad;
521 }
522 deliverifp = m->m_pkthdr.rcvif;
523 goto hbhcheck;
524 }
525
526 /*
527 * Unicast check
528 */
529 if (ip6_forward_rt.ro_rt != NULL &&
530 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
531 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
532 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
533 ip6stat.ip6s_forward_cachehit++;
534 else {
535 struct sockaddr_in6 *dst6;
536
537 if (ip6_forward_rt.ro_rt) {
538 /* route is down or destination is different */
539 ip6stat.ip6s_forward_cachemiss++;
540 RTFREE(ip6_forward_rt.ro_rt);
541 ip6_forward_rt.ro_rt = 0;
542 }
543
544 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
545 dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
546 dst6->sin6_len = sizeof(struct sockaddr_in6);
547 dst6->sin6_family = AF_INET6;
548 dst6->sin6_addr = ip6->ip6_dst;
549
550 rtalloc((struct route *)&ip6_forward_rt);
551 }
552
553#define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
554
555 /*
556 * Accept the packet if the forwarding interface to the destination
557 * according to the routing table is the loopback interface,
558 * unless the associated route has a gateway.
559 * Note that this approach causes to accept a packet if there is a
560 * route to the loopback interface for the destination of the packet.
561 * But we think it's even useful in some situations, e.g. when using
562 * a special daemon which wants to intercept the packet.
563 *
564 * XXX: some OSes automatically make a cloned route for the destination
565 * of an outgoing packet. If the outgoing interface of the packet
566 * is a loopback one, the kernel would consider the packet to be
567 * accepted, even if we have no such address assinged on the interface.
568 * We check the cloned flag of the route entry to reject such cases,
569 * assuming that route entries for our own addresses are not made by
570 * cloning (it should be true because in6_addloop explicitly installs
571 * the host route). However, we might have to do an explicit check
572 * while it would be less efficient. Or, should we rather install a
573 * reject route for such a case?
574 */
575 if (ip6_forward_rt.ro_rt &&
576 (ip6_forward_rt.ro_rt->rt_flags &
577 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
578#ifdef RTF_WASCLONED
579 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) &&
580#endif
581#ifdef RTF_CLONED
582 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) &&
583#endif
584#if 0
585 /*
586 * The check below is redundant since the comparison of
587 * the destination and the key of the rtentry has
588 * already done through looking up the routing table.
589 */
590 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
591 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr)
592#endif
593 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
594 struct in6_ifaddr *ia6 =
595 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
596
597 /*
598 * record address information into m_tag.
599 */
600 (void)ip6_setdstifaddr(m, ia6);
601
602 /*
603 * packets to a tentative, duplicated, or somehow invalid
604 * address must not be accepted.
605 */
606 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
607 /* this address is ready */
608 ours = 1;
609 deliverifp = ia6->ia_ifp; /* correct? */
610 /* Count the packet in the ip address stats */
611 ia6->ia_ifa.if_ipackets++;
612 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
613 goto hbhcheck;
614 } else {
615 /* address is not ready, so discard the packet. */
616 nd6log((LOG_INFO,
617 "ip6_input: packet to an unready address %s->%s\n",
618 ip6_sprintf(&ip6->ip6_src),
619 ip6_sprintf(&ip6->ip6_dst)));
620
621 goto bad;
622 }
623 }
624
625 /*
626 * FAITH (Firewall Aided Internet Translator)
627 */
628 if (ip6_keepfaith) {
629 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
630 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
631 /* XXX do we need more sanity checks? */
632 ours = 1;
633 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */
634 goto hbhcheck;
635 }
636 }
637
638 /*
639 * Now there is no reason to process the packet if it's not our own
640 * and we're not a router.
641 */
642 if (!ip6_forwarding) {
643 ip6stat.ip6s_cantforward++;
644 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
645 goto bad;
646 }
647
648 hbhcheck:
649 /*
650 * record address information into m_tag, if we don't have one yet.
651 * note that we are unable to record it, if the address is not listed
652 * as our interface address (e.g. multicast addresses, addresses
653 * within FAITH prefixes and such).
654 */
655 if (deliverifp && !ip6_getdstifaddr(m)) {
656 struct in6_ifaddr *ia6;
657
658 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
659 if (ia6) {
660 if (!ip6_setdstifaddr(m, ia6)) {
661 /*
662 * XXX maybe we should drop the packet here,
663 * as we could not provide enough information
664 * to the upper layers.
665 */
666 }
667 }
668 }
669
670 /*
671 * Process Hop-by-Hop options header if it's contained.
672 * m may be modified in ip6_hopopts_input().
673 * If a JumboPayload option is included, plen will also be modified.
674 */
675 plen = (u_int32_t)ntohs(ip6->ip6_plen);
676 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
677 struct ip6_hbh *hbh;
678
679 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
680#if 0 /*touches NULL pointer*/
681 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
682#endif
683 return; /* m have already been freed */
684 }
685
686 /* adjust pointer */
687 ip6 = mtod(m, struct ip6_hdr *);
688
689 /*
690 * if the payload length field is 0 and the next header field
691 * indicates Hop-by-Hop Options header, then a Jumbo Payload
692 * option MUST be included.
693 */
694 if (ip6->ip6_plen == 0 && plen == 0) {
695 /*
696 * Note that if a valid jumbo payload option is
697 * contained, ip6_hopopts_input() must set a valid
698 * (non-zero) payload length to the variable plen.
699 */
700 ip6stat.ip6s_badoptions++;
701 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
702 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
703 icmp6_error(m, ICMP6_PARAM_PROB,
704 ICMP6_PARAMPROB_HEADER,
705 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
706 return;
707 }
708#ifndef PULLDOWN_TEST
709 /* ip6_hopopts_input() ensures that mbuf is contiguous */
710 hbh = (struct ip6_hbh *)(ip6 + 1);
711#else
712 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
713 sizeof(struct ip6_hbh));
714 if (hbh == NULL) {
715 ip6stat.ip6s_tooshort++;
716 return;
717 }
718#endif
719 nxt = hbh->ip6h_nxt;
720
721 /*
722 * accept the packet if a router alert option is included
723 * and we act as an IPv6 router.
724 */
725 if (rtalert != ~0 && ip6_forwarding)
726 ours = 1;
727 } else
728 nxt = ip6->ip6_nxt;
729
730 /*
731 * Check that the amount of data in the buffers
732 * is as at least much as the IPv6 header would have us expect.
733 * Trim mbufs if longer than we expect.
734 * Drop packet if shorter than we expect.
735 */
736 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
737 ip6stat.ip6s_tooshort++;
738 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
739 goto bad;
740 }
741 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
742 if (m->m_len == m->m_pkthdr.len) {
743 m->m_len = sizeof(struct ip6_hdr) + plen;
744 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
745 } else
746 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
747 }
748
749 /*
750 * Forward if desirable.
751 */
752 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
753 /*
754 * If we are acting as a multicast router, all
755 * incoming multicast packets are passed to the
756 * kernel-level multicast forwarding function.
757 * The packet is returned (relatively) intact; if
758 * ip6_mforward() returns a non-zero value, the packet
759 * must be discarded, else it may be accepted below.
760 */
761 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
762 ip6stat.ip6s_cantforward++;
763 m_freem(m);
764 return;
765 }
766 if (!ours) {
767 m_freem(m);
768 return;
769 }
770 } else if (!ours) {
771 ip6_forward(m, srcrt);
772 return;
773 }
774
775 ip6 = mtod(m, struct ip6_hdr *);
776
777 /*
778 * Malicious party may be able to use IPv4 mapped addr to confuse
779 * tcp/udp stack and bypass security checks (act as if it was from
780 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
781 *
782 * For SIIT end node behavior, you may want to disable the check.
783 * However, you will become vulnerable to attacks using IPv4 mapped
784 * source.
785 */
786 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
787 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
788 ip6stat.ip6s_badscope++;
789 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
790 goto bad;
791 }
792
793 /*
794 * Tell launch routine the next header
795 */
796 ip6stat.ip6s_delivered++;
797 in6_ifstat_inc(deliverifp, ifs6_in_deliver);
798 nest = 0;
799
800 while (nxt != IPPROTO_DONE) {
801 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
802 ip6stat.ip6s_toomanyhdr++;
803 goto bad;
804 }
805
806 /*
807 * protection against faulty packet - there should be
808 * more sanity checks in header chain processing.
809 */
810 if (m->m_pkthdr.len < off) {
811 ip6stat.ip6s_tooshort++;
812 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
813 goto bad;
814 }
815
816#ifdef IPSEC
817 /*
818 * enforce IPsec policy checking if we are seeing last header.
819 * note that we do not visit this with protocols with pcb layer
820 * code - like udp/tcp/raw ip.
821 */
822 if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
823 ipsec6_in_reject(m, NULL)) {
824 ipsec6stat.in_polvio++;
825 goto bad;
826 }
827#endif
828 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
829 }
830 return;
831 bad:
832 m_freem(m);
833}
834
835/*
836 * set/grab in6_ifaddr correspond to IPv6 destination address.
837 * XXX backward compatibility wrapper
838 */
839static struct ip6aux *
840ip6_setdstifaddr(m, ia6)
841 struct mbuf *m;
842 struct in6_ifaddr *ia6;
843{
844 struct ip6aux *ip6a;
845
846 ip6a = ip6_addaux(m);
847 if (ip6a)
848 ip6a->ip6a_dstia6 = ia6;
849 return ip6a; /* NULL if failed to set */
850}
851
852struct in6_ifaddr *
853ip6_getdstifaddr(m)
854 struct mbuf *m;
855{
856 struct ip6aux *ip6a;
857
858 ip6a = ip6_findaux(m);
859 if (ip6a)
860 return ip6a->ip6a_dstia6;
861 else
862 return NULL;
863}
864
865/*
866 * Hop-by-Hop options header processing. If a valid jumbo payload option is
867 * included, the real payload length will be stored in plenp.
868 */
869static int
870ip6_hopopts_input(plenp, rtalertp, mp, offp)
871 u_int32_t *plenp;
872 u_int32_t *rtalertp; /* XXX: should be stored more smart way */
873 struct mbuf **mp;
874 int *offp;
875{
876 struct mbuf *m = *mp;
877 int off = *offp, hbhlen;
878 struct ip6_hbh *hbh;
879 u_int8_t *opt;
880
881 /* validation of the length of the header */
882#ifndef PULLDOWN_TEST
883 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
884 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
885 hbhlen = (hbh->ip6h_len + 1) << 3;
886
887 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
888 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
889#else
890 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
891 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
892 if (hbh == NULL) {
893 ip6stat.ip6s_tooshort++;
894 return -1;
895 }
896 hbhlen = (hbh->ip6h_len + 1) << 3;
897 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
898 hbhlen);
899 if (hbh == NULL) {
900 ip6stat.ip6s_tooshort++;
901 return -1;
902 }
903#endif
904 off += hbhlen;
905 hbhlen -= sizeof(struct ip6_hbh);
906 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
907
908 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
909 hbhlen, rtalertp, plenp) < 0)
910 return (-1);
911
912 *offp = off;
913 *mp = m;
914 return (0);
915}
916
917/*
918 * Search header for all Hop-by-hop options and process each option.
919 * This function is separate from ip6_hopopts_input() in order to
920 * handle a case where the sending node itself process its hop-by-hop
921 * options header. In such a case, the function is called from ip6_output().
922 *
923 * The function assumes that hbh header is located right after the IPv6 header
924 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
925 * opthead + hbhlen is located in continuous memory region.
926 */
927int
928ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
929 struct mbuf *m;
930 u_int8_t *opthead;
931 int hbhlen;
932 u_int32_t *rtalertp;
933 u_int32_t *plenp;
934{
935 struct ip6_hdr *ip6;
936 int optlen = 0;
937 u_int8_t *opt = opthead;
938 u_int16_t rtalert_val;
939 u_int32_t jumboplen;
940 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
941
942 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
943 switch (*opt) {
944 case IP6OPT_PAD1:
945 optlen = 1;
946 break;
947 case IP6OPT_PADN:
948 if (hbhlen < IP6OPT_MINLEN) {
949 ip6stat.ip6s_toosmall++;
950 goto bad;
951 }
952 optlen = *(opt + 1) + 2;
953 break;
954 case IP6OPT_ROUTER_ALERT:
955 /* XXX may need check for alignment */
956 if (hbhlen < IP6OPT_RTALERT_LEN) {
957 ip6stat.ip6s_toosmall++;
958 goto bad;
959 }
960 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
961 /* XXX stat */
962 icmp6_error(m, ICMP6_PARAM_PROB,
963 ICMP6_PARAMPROB_HEADER,
964 erroff + opt + 1 - opthead);
965 return (-1);
966 }
967 optlen = IP6OPT_RTALERT_LEN;
968 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
969 *rtalertp = ntohs(rtalert_val);
970 break;
971 case IP6OPT_JUMBO:
972 /* XXX may need check for alignment */
973 if (hbhlen < IP6OPT_JUMBO_LEN) {
974 ip6stat.ip6s_toosmall++;
975 goto bad;
976 }
977 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
978 /* XXX stat */
979 icmp6_error(m, ICMP6_PARAM_PROB,
980 ICMP6_PARAMPROB_HEADER,
981 erroff + opt + 1 - opthead);
982 return (-1);
983 }
984 optlen = IP6OPT_JUMBO_LEN;
985
986 /*
987 * IPv6 packets that have non 0 payload length
988 * must not contain a jumbo payload option.
989 */
990 ip6 = mtod(m, struct ip6_hdr *);
991 if (ip6->ip6_plen) {
992 ip6stat.ip6s_badoptions++;
993 icmp6_error(m, ICMP6_PARAM_PROB,
994 ICMP6_PARAMPROB_HEADER,
995 erroff + opt - opthead);
996 return (-1);
997 }
998
999 /*
1000 * We may see jumbolen in unaligned location, so
1001 * we'd need to perform bcopy().
1002 */
1003 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1004 jumboplen = (u_int32_t)htonl(jumboplen);
1005
1006#if 1
1007 /*
1008 * if there are multiple jumbo payload options,
1009 * *plenp will be non-zero and the packet will be
1010 * rejected.
1011 * the behavior may need some debate in ipngwg -
1012 * multiple options does not make sense, however,
1013 * there's no explicit mention in specification.
1014 */
1015 if (*plenp != 0) {
1016 ip6stat.ip6s_badoptions++;
1017 icmp6_error(m, ICMP6_PARAM_PROB,
1018 ICMP6_PARAMPROB_HEADER,
1019 erroff + opt + 2 - opthead);
1020 return (-1);
1021 }
1022#endif
1023
1024 /*
1025 * jumbo payload length must be larger than 65535.
1026 */
1027 if (jumboplen <= IPV6_MAXPACKET) {
1028 ip6stat.ip6s_badoptions++;
1029 icmp6_error(m, ICMP6_PARAM_PROB,
1030 ICMP6_PARAMPROB_HEADER,
1031 erroff + opt + 2 - opthead);
1032 return (-1);
1033 }
1034 *plenp = jumboplen;
1035
1036 break;
1037 default: /* unknown option */
1038 if (hbhlen < IP6OPT_MINLEN) {
1039 ip6stat.ip6s_toosmall++;
1040 goto bad;
1041 }
1042 optlen = ip6_unknown_opt(opt, m,
1043 erroff + opt - opthead);
1044 if (optlen == -1)
1045 return (-1);
1046 optlen += 2;
1047 break;
1048 }
1049 }
1050
1051 return (0);
1052
1053 bad:
1054 m_freem(m);
1055 return (-1);
1056}
1057
1058/*
1059 * Unknown option processing.
1060 * The third argument `off' is the offset from the IPv6 header to the option,
1061 * which is necessary if the IPv6 header the and option header and IPv6 header
1062 * is not continuous in order to return an ICMPv6 error.
1063 */
1064int
1065ip6_unknown_opt(optp, m, off)
1066 u_int8_t *optp;
1067 struct mbuf *m;
1068 int off;
1069{
1070 struct ip6_hdr *ip6;
1071
1072 switch (IP6OPT_TYPE(*optp)) {
1073 case IP6OPT_TYPE_SKIP: /* ignore the option */
1074 return ((int)*(optp + 1));
1075 case IP6OPT_TYPE_DISCARD: /* silently discard */
1076 m_freem(m);
1077 return (-1);
1078 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1079 ip6stat.ip6s_badoptions++;
1080 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1081 return (-1);
1082 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1083 ip6stat.ip6s_badoptions++;
1084 ip6 = mtod(m, struct ip6_hdr *);
1085 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1086 (m->m_flags & (M_BCAST|M_MCAST)))
1087 m_freem(m);
1088 else
1089 icmp6_error(m, ICMP6_PARAM_PROB,
1090 ICMP6_PARAMPROB_OPTION, off);
1091 return (-1);
1092 }
1093
1094 m_freem(m); /* XXX: NOTREACHED */
1095 return (-1);
1096}
1097
1098/*
1099 * Create the "control" list for this pcb.
1100 * The function will not modify mbuf chain at all.
1101 *
1102 * with KAME mbuf chain restriction:
1103 * The routine will be called from upper layer handlers like tcp6_input().
1104 * Thus the routine assumes that the caller (tcp6_input) have already
1105 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1106 * very first mbuf on the mbuf chain.
1107 */
1108void
1109ip6_savecontrol(in6p, m, mp)
1110 struct inpcb *in6p;
1111 struct mbuf *m, **mp;
1112{
1113#define IS2292(x, y) ((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1114 struct thread *td = curthread; /* XXX */
1115 int privileged = 0;
1116 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1117
1118 if (td && !suser(td))
1119 privileged++;
1120
1121#ifdef SO_TIMESTAMP
1122 if ((in6p->in6p_socket->so_options & SO_TIMESTAMP) != 0) {
1123 struct timeval tv;
1124
1125 microtime(&tv);
1126 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1127 SCM_TIMESTAMP, SOL_SOCKET);
1128 if (*mp)
1129 mp = &(*mp)->m_next;
1130 }
1131#endif
1132
1133 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
1134 return;
1135
1136 /* RFC 2292 sec. 5 */
1137 if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1138 struct in6_pktinfo pi6;
1139
1140 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1141 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1142 pi6.ipi6_ifindex =
1143 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1144
1145 *mp = sbcreatecontrol((caddr_t) &pi6,
1146 sizeof(struct in6_pktinfo),
1147 IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1148 if (*mp)
1149 mp = &(*mp)->m_next;
1150 }
1151
1152 if ((in6p->in6p_flags & IN6P_HOPLIMIT) != 0) {
1153 int hlim = ip6->ip6_hlim & 0xff;
1154
1155 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int),
1156 IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
1157 if (*mp)
1158 mp = &(*mp)->m_next;
1159 }
1160
1161 if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
1162 u_int32_t flowinfo;
1163 int tclass;
1164
1165 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1166 flowinfo >>= 20;
1167
1168 tclass = flowinfo & 0xff;
1169 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(tclass),
1170 IPV6_TCLASS, IPPROTO_IPV6);
1171 if (*mp)
1172 mp = &(*mp)->m_next;
1173 }
1174
1175 /*
1176 * IPV6_HOPOPTS socket option. We require super-user privilege
1177 * for the option, but it might be too strict, since there might
1178 * be some hop-by-hop options which can be returned to normal user.
1179 * See RFC 2292 section 6.
1180 */
1181 if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
1182#ifdef DIAGNOSTIC
1183 if (!privileged)
1184 panic("IN6P_HOPOPTS is set for unprivileged socket");
1185#endif
1186 /*
1187 * Check if a hop-by-hop options header is contatined in the
1188 * received packet, and if so, store the options as ancillary
1189 * data. Note that a hop-by-hop options header must be
1190 * just after the IPv6 header, which is assured through the
1191 * IPv6 input processing.
1192 */
1193 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1194 struct ip6_hbh *hbh;
1195 int hbhlen = 0;
1196#ifdef PULLDOWN_TEST
1197 struct mbuf *ext;
1198#endif
1199
1200#ifndef PULLDOWN_TEST
1201 hbh = (struct ip6_hbh *)(ip6 + 1);
1202 hbhlen = (hbh->ip6h_len + 1) << 3;
1203#else
1204 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1205 ip6->ip6_nxt);
1206 if (ext == NULL) {
1207 ip6stat.ip6s_tooshort++;
1208 return;
1209 }
1210 hbh = mtod(ext, struct ip6_hbh *);
1211 hbhlen = (hbh->ip6h_len + 1) << 3;
1212 if (hbhlen != ext->m_len) {
1213 m_freem(ext);
1214 ip6stat.ip6s_tooshort++;
1215 return;
1216 }
1217#endif
1218
1219 /*
1220 * XXX: We copy the whole header even if a
1221 * jumbo payload option is included, the option which
1222 * is to be removed before returning according to
1223 * RFC2292.
1224 * Note: this constraint is removed in 2292bis.
1225 */
1226 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1227 IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1228 IPPROTO_IPV6);
1229 if (*mp)
1230 mp = &(*mp)->m_next;
1231#ifdef PULLDOWN_TEST
1232 m_freem(ext);
1233#endif
1234 }
1235 }
1236
1237 if ((in6p->in6p_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
1238 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1239
1240 /*
1241 * Search for destination options headers or routing
1242 * header(s) through the header chain, and stores each
1243 * header as ancillary data.
1244 * Note that the order of the headers remains in
1245 * the chain of ancillary data.
1246 */
1247 while (1) { /* is explicit loop prevention necessary? */
1248 struct ip6_ext *ip6e = NULL;
1249 int elen;
1250#ifdef PULLDOWN_TEST
1251 struct mbuf *ext = NULL;
1252#endif
1253
1254 /*
1255 * if it is not an extension header, don't try to
1256 * pull it from the chain.
1257 */
1258 switch (nxt) {
1259 case IPPROTO_DSTOPTS:
1260 case IPPROTO_ROUTING:
1261 case IPPROTO_HOPOPTS:
1262 case IPPROTO_AH: /* is it possible? */
1263 break;
1264 default:
1265 goto loopend;
1266 }
1267
1268#ifndef PULLDOWN_TEST
1269 if (off + sizeof(*ip6e) > m->m_len)
1270 goto loopend;
1271 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1272 if (nxt == IPPROTO_AH)
1273 elen = (ip6e->ip6e_len + 2) << 2;
1274 else
1275 elen = (ip6e->ip6e_len + 1) << 3;
1276 if (off + elen > m->m_len)
1277 goto loopend;
1278#else
1279 ext = ip6_pullexthdr(m, off, nxt);
1280 if (ext == NULL) {
1281 ip6stat.ip6s_tooshort++;
1282 return;
1283 }
1284 ip6e = mtod(ext, struct ip6_ext *);
1285 if (nxt == IPPROTO_AH)
1286 elen = (ip6e->ip6e_len + 2) << 2;
1287 else
1288 elen = (ip6e->ip6e_len + 1) << 3;
1289 if (elen != ext->m_len) {
1290 m_freem(ext);
1291 ip6stat.ip6s_tooshort++;
1292 return;
1293 }
1294#endif
1295
1296 switch (nxt) {
1297 case IPPROTO_DSTOPTS:
1298 if (!(in6p->in6p_flags & IN6P_DSTOPTS))
1299 break;
1300
1301 /*
1302 * We also require super-user privilege for
1303 * the option. See comments on IN6_HOPOPTS.
1304 */
1305 if (!privileged)
1306 break;
1307
1308 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1309 IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1310 IPPROTO_IPV6);
1311 if (*mp)
1312 mp = &(*mp)->m_next;
1313 break;
1314 case IPPROTO_ROUTING:
1315 if (!in6p->in6p_flags & IN6P_RTHDR)
1316 break;
1317
1318 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1319 IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
1320 IPPROTO_IPV6);
1321 if (*mp)
1322 mp = &(*mp)->m_next;
1323 break;
1324 case IPPROTO_HOPOPTS:
1325 case IPPROTO_AH: /* is it possible? */
1326 break;
1327
1328 default:
1329 /*
1330 * other cases have been filtered in the above.
1331 * none will visit this case. here we supply
1332 * the code just in case (nxt overwritten or
1333 * other cases).
1334 */
1335#ifdef PULLDOWN_TEST
1336 m_freem(ext);
1337#endif
1338 goto loopend;
1339
1340 }
1341
1342 /* proceed with the next header. */
1343 off += elen;
1344 nxt = ip6e->ip6e_nxt;
1345 ip6e = NULL;
1346#ifdef PULLDOWN_TEST
1347 m_freem(ext);
1348 ext = NULL;
1349#endif
1350 }
1351 loopend:
1352 ;
1353 }
1354
1355#undef IS2292
1356}
1357
1358void
1359ip6_notify_pmtu(in6p, dst, mtu)
1360 struct inpcb *in6p;
1361 struct sockaddr_in6 *dst;
1362 u_int32_t *mtu;
1363{
1364 struct socket *so;
1365 struct mbuf *m_mtu;
1366 struct ip6_mtuinfo mtuctl;
1367
1368 so = in6p->inp_socket;
1369
1370 if (mtu == NULL)
1371 return;
1372
1373#ifdef DIAGNOSTIC
1374 if (so == NULL) /* I believe this is impossible */
1375 panic("ip6_notify_pmtu: socket is NULL");
1376#endif
1377
1378 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */
1379 mtuctl.ip6m_mtu = *mtu;
1380 mtuctl.ip6m_addr = *dst;
1381 in6_recoverscope(&mtuctl.ip6m_addr, &mtuctl.ip6m_addr.sin6_addr, NULL);
1382
1383 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl),
1384 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1385 return;
1386
1387 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu)
1388 == 0) {
1389 m_freem(m_mtu);
1390 /* XXX: should count statistics */
1391 } else
1392 sorwakeup(so);
1393
1394 return;
1395}
1396
1397#ifdef PULLDOWN_TEST
1398/*
1399 * pull single extension header from mbuf chain. returns single mbuf that
1400 * contains the result, or NULL on error.
1401 */
1402static struct mbuf *
1403ip6_pullexthdr(m, off, nxt)
1404 struct mbuf *m;
1405 size_t off;
1406 int nxt;
1407{
1408 struct ip6_ext ip6e;
1409 size_t elen;
1410 struct mbuf *n;
1411
1412#ifdef DIAGNOSTIC
1413 switch (nxt) {
1414 case IPPROTO_DSTOPTS:
1415 case IPPROTO_ROUTING:
1416 case IPPROTO_HOPOPTS:
1417 case IPPROTO_AH: /* is it possible? */
1418 break;
1419 default:
1420 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1421 }
1422#endif
1423
1424 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1425 if (nxt == IPPROTO_AH)
1426 elen = (ip6e.ip6e_len + 2) << 2;
1427 else
1428 elen = (ip6e.ip6e_len + 1) << 3;
1429
1430 MGET(n, M_DONTWAIT, MT_DATA);
1431 if (n && elen >= MLEN) {
1432 MCLGET(n, M_DONTWAIT);
1433 if ((n->m_flags & M_EXT) == 0) {
1434 m_free(n);
1435 n = NULL;
1436 }
1437 }
1438 if (!n)
1439 return NULL;
1440
1441 n->m_len = 0;
1442 if (elen >= M_TRAILINGSPACE(n)) {
1443 m_free(n);
1444 return NULL;
1445 }
1446
1447 m_copydata(m, off, elen, mtod(n, caddr_t));
1448 n->m_len = elen;
1449 return n;
1450}
1451#endif
1452
1453/*
1454 * Get pointer to the previous header followed by the header
1455 * currently processed.
1456 * XXX: This function supposes that
1457 * M includes all headers,
1458 * the next header field and the header length field of each header
1459 * are valid, and
1460 * the sum of each header length equals to OFF.
1461 * Because of these assumptions, this function must be called very
1462 * carefully. Moreover, it will not be used in the near future when
1463 * we develop `neater' mechanism to process extension headers.
1464 */
1465char *
1466ip6_get_prevhdr(m, off)
1467 struct mbuf *m;
1468 int off;
1469{
1470 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1471
1472 if (off == sizeof(struct ip6_hdr))
1473 return (&ip6->ip6_nxt);
1474 else {
1475 int len, nxt;
1476 struct ip6_ext *ip6e = NULL;
1477
1478 nxt = ip6->ip6_nxt;
1479 len = sizeof(struct ip6_hdr);
1480 while (len < off) {
1481 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1482
1483 switch (nxt) {
1484 case IPPROTO_FRAGMENT:
1485 len += sizeof(struct ip6_frag);
1486 break;
1487 case IPPROTO_AH:
1488 len += (ip6e->ip6e_len + 2) << 2;
1489 break;
1490 default:
1491 len += (ip6e->ip6e_len + 1) << 3;
1492 break;
1493 }
1494 nxt = ip6e->ip6e_nxt;
1495 }
1496 if (ip6e)
1497 return (&ip6e->ip6e_nxt);
1498 else
1499 return NULL;
1500 }
1501}
1502
1503/*
1504 * get next header offset. m will be retained.
1505 */
1506int
1507ip6_nexthdr(m, off, proto, nxtp)
1508 struct mbuf *m;
1509 int off;
1510 int proto;
1511 int *nxtp;
1512{
1513 struct ip6_hdr ip6;
1514 struct ip6_ext ip6e;
1515 struct ip6_frag fh;
1516
1517 /* just in case */
1518 if (m == NULL)
1519 panic("ip6_nexthdr: m == NULL");
1520 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1521 return -1;
1522
1523 switch (proto) {
1524 case IPPROTO_IPV6:
1525 if (m->m_pkthdr.len < off + sizeof(ip6))
1526 return -1;
1527 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1528 if (nxtp)
1529 *nxtp = ip6.ip6_nxt;
1530 off += sizeof(ip6);
1531 return off;
1532
1533 case IPPROTO_FRAGMENT:
1534 /*
1535 * terminate parsing if it is not the first fragment,
1536 * it does not make sense to parse through it.
1537 */
1538 if (m->m_pkthdr.len < off + sizeof(fh))
1539 return -1;
1540 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1541 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1542 if (fh.ip6f_offlg & IP6F_OFF_MASK)
1543 return -1;
1544 if (nxtp)
1545 *nxtp = fh.ip6f_nxt;
1546 off += sizeof(struct ip6_frag);
1547 return off;
1548
1549 case IPPROTO_AH:
1550 if (m->m_pkthdr.len < off + sizeof(ip6e))
1551 return -1;
1552 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1553 if (nxtp)
1554 *nxtp = ip6e.ip6e_nxt;
1555 off += (ip6e.ip6e_len + 2) << 2;
1556 return off;
1557
1558 case IPPROTO_HOPOPTS:
1559 case IPPROTO_ROUTING:
1560 case IPPROTO_DSTOPTS:
1561 if (m->m_pkthdr.len < off + sizeof(ip6e))
1562 return -1;
1563 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1564 if (nxtp)
1565 *nxtp = ip6e.ip6e_nxt;
1566 off += (ip6e.ip6e_len + 1) << 3;
1567 return off;
1568
1569 case IPPROTO_NONE:
1570 case IPPROTO_ESP:
1571 case IPPROTO_IPCOMP:
1572 /* give up */
1573 return -1;
1574
1575 default:
1576 return -1;
1577 }
1578
1579 return -1;
1580}
1581
1582/*
1583 * get offset for the last header in the chain. m will be kept untainted.
1584 */
1585int
1586ip6_lasthdr(m, off, proto, nxtp)
1587 struct mbuf *m;
1588 int off;
1589 int proto;
1590 int *nxtp;
1591{
1592 int newoff;
1593 int nxt;
1594
1595 if (!nxtp) {
1596 nxt = -1;
1597 nxtp = &nxt;
1598 }
1599 while (1) {
1600 newoff = ip6_nexthdr(m, off, proto, nxtp);
1601 if (newoff < 0)
1602 return off;
1603 else if (newoff < off)
1604 return -1; /* invalid */
1605 else if (newoff == off)
1606 return newoff;
1607
1608 off = newoff;
1609 proto = *nxtp;
1610 }
1611}
1612
1613struct ip6aux *
1614ip6_addaux(m)
1615 struct mbuf *m;
1616{
1617 struct m_tag *mtag;
1618
1619 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1620 if (!mtag) {
1621 mtag = m_tag_get(PACKET_TAG_IPV6_INPUT, sizeof(struct ip6aux),
1622 M_NOWAIT);
1623 if (mtag) {
1624 m_tag_prepend(m, mtag);
1625 bzero(mtag + 1, sizeof(struct ip6aux));
1626 }
1627 }
1628 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1629}
1630
1631struct ip6aux *
1632ip6_findaux(m)
1633 struct mbuf *m;
1634{
1635 struct m_tag *mtag;
1636
1637 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1638 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1639}
1640
1641void
1642ip6_delaux(m)
1643 struct mbuf *m;
1644{
1645 struct m_tag *mtag;
1646
1647 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1648 if (mtag)
1649 m_tag_delete(m, mtag);
1650}
1651
1652/*
1653 * System control for IP6
1654 */
1655
1656u_char inet6ctlerrmap[PRC_NCMDS] = {
1657 0, 0, 0, 0,
1658 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1659 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1660 EMSGSIZE, EHOSTUNREACH, 0, 0,
1661 0, 0, 0, 0,
1662 ENOPROTOOPT
1663};
| 460 * construct source and destination address structures with
461 * disambiguating their scope zones (if there is ambiguity).
462 * XXX: sin6_family and sin6_len will NOT be referred to, but we fill
463 * in these fields just in case.
464 */
465 if (in6_addr2zoneid(m->m_pkthdr.rcvif, &ip6->ip6_src, &srczone) ||
466 in6_addr2zoneid(m->m_pkthdr.rcvif, &ip6->ip6_dst, &dstzone)) {
467 /*
468 * Note that these generic checks cover cases that src or
469 * dst are the loopback address and the receiving interface
470 * is not loopback.
471 */
472 ip6stat.ip6s_badscope++;
473 goto bad;
474 }
475
476 bzero(&sa6, sizeof(sa6));
477 sa6.sin6_family = AF_INET6;
478 sa6.sin6_len = sizeof(struct sockaddr_in6);
479
480 sa6.sin6_addr = ip6->ip6_src;
481 sa6.sin6_scope_id = srczone;
482 if (in6_embedscope(&ip6->ip6_src, &sa6, NULL, NULL)) {
483 /* XXX: should not happen */
484 ip6stat.ip6s_badscope++;
485 goto bad;
486 }
487
488 sa6.sin6_addr = ip6->ip6_dst;
489 sa6.sin6_scope_id = dstzone;
490 if (in6_embedscope(&ip6->ip6_dst, &sa6, NULL, NULL)) {
491 /* XXX: should not happen */
492 ip6stat.ip6s_badscope++;
493 goto bad;
494 }
495
496 /* XXX: ff01::%ifN awareness is not merged, yet. */
497 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_src))
498 ip6->ip6_src.s6_addr16[1] = 0;
499 if (IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst))
500 ip6->ip6_dst.s6_addr16[1] = 0;
501
502 /*
503 * Multicast check
504 */
505 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
506 struct in6_multi *in6m = 0;
507
508 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
509 /*
510 * See if we belong to the destination multicast group on the
511 * arrival interface.
512 */
513 IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
514 if (in6m)
515 ours = 1;
516 else if (!ip6_mrouter) {
517 ip6stat.ip6s_notmember++;
518 ip6stat.ip6s_cantforward++;
519 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
520 goto bad;
521 }
522 deliverifp = m->m_pkthdr.rcvif;
523 goto hbhcheck;
524 }
525
526 /*
527 * Unicast check
528 */
529 if (ip6_forward_rt.ro_rt != NULL &&
530 (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
531 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
532 &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
533 ip6stat.ip6s_forward_cachehit++;
534 else {
535 struct sockaddr_in6 *dst6;
536
537 if (ip6_forward_rt.ro_rt) {
538 /* route is down or destination is different */
539 ip6stat.ip6s_forward_cachemiss++;
540 RTFREE(ip6_forward_rt.ro_rt);
541 ip6_forward_rt.ro_rt = 0;
542 }
543
544 bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
545 dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
546 dst6->sin6_len = sizeof(struct sockaddr_in6);
547 dst6->sin6_family = AF_INET6;
548 dst6->sin6_addr = ip6->ip6_dst;
549
550 rtalloc((struct route *)&ip6_forward_rt);
551 }
552
553#define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
554
555 /*
556 * Accept the packet if the forwarding interface to the destination
557 * according to the routing table is the loopback interface,
558 * unless the associated route has a gateway.
559 * Note that this approach causes to accept a packet if there is a
560 * route to the loopback interface for the destination of the packet.
561 * But we think it's even useful in some situations, e.g. when using
562 * a special daemon which wants to intercept the packet.
563 *
564 * XXX: some OSes automatically make a cloned route for the destination
565 * of an outgoing packet. If the outgoing interface of the packet
566 * is a loopback one, the kernel would consider the packet to be
567 * accepted, even if we have no such address assinged on the interface.
568 * We check the cloned flag of the route entry to reject such cases,
569 * assuming that route entries for our own addresses are not made by
570 * cloning (it should be true because in6_addloop explicitly installs
571 * the host route). However, we might have to do an explicit check
572 * while it would be less efficient. Or, should we rather install a
573 * reject route for such a case?
574 */
575 if (ip6_forward_rt.ro_rt &&
576 (ip6_forward_rt.ro_rt->rt_flags &
577 (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
578#ifdef RTF_WASCLONED
579 !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) &&
580#endif
581#ifdef RTF_CLONED
582 !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) &&
583#endif
584#if 0
585 /*
586 * The check below is redundant since the comparison of
587 * the destination and the key of the rtentry has
588 * already done through looking up the routing table.
589 */
590 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
591 &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr)
592#endif
593 ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
594 struct in6_ifaddr *ia6 =
595 (struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
596
597 /*
598 * record address information into m_tag.
599 */
600 (void)ip6_setdstifaddr(m, ia6);
601
602 /*
603 * packets to a tentative, duplicated, or somehow invalid
604 * address must not be accepted.
605 */
606 if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
607 /* this address is ready */
608 ours = 1;
609 deliverifp = ia6->ia_ifp; /* correct? */
610 /* Count the packet in the ip address stats */
611 ia6->ia_ifa.if_ipackets++;
612 ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
613 goto hbhcheck;
614 } else {
615 /* address is not ready, so discard the packet. */
616 nd6log((LOG_INFO,
617 "ip6_input: packet to an unready address %s->%s\n",
618 ip6_sprintf(&ip6->ip6_src),
619 ip6_sprintf(&ip6->ip6_dst)));
620
621 goto bad;
622 }
623 }
624
625 /*
626 * FAITH (Firewall Aided Internet Translator)
627 */
628 if (ip6_keepfaith) {
629 if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
630 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
631 /* XXX do we need more sanity checks? */
632 ours = 1;
633 deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */
634 goto hbhcheck;
635 }
636 }
637
638 /*
639 * Now there is no reason to process the packet if it's not our own
640 * and we're not a router.
641 */
642 if (!ip6_forwarding) {
643 ip6stat.ip6s_cantforward++;
644 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
645 goto bad;
646 }
647
648 hbhcheck:
649 /*
650 * record address information into m_tag, if we don't have one yet.
651 * note that we are unable to record it, if the address is not listed
652 * as our interface address (e.g. multicast addresses, addresses
653 * within FAITH prefixes and such).
654 */
655 if (deliverifp && !ip6_getdstifaddr(m)) {
656 struct in6_ifaddr *ia6;
657
658 ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
659 if (ia6) {
660 if (!ip6_setdstifaddr(m, ia6)) {
661 /*
662 * XXX maybe we should drop the packet here,
663 * as we could not provide enough information
664 * to the upper layers.
665 */
666 }
667 }
668 }
669
670 /*
671 * Process Hop-by-Hop options header if it's contained.
672 * m may be modified in ip6_hopopts_input().
673 * If a JumboPayload option is included, plen will also be modified.
674 */
675 plen = (u_int32_t)ntohs(ip6->ip6_plen);
676 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
677 struct ip6_hbh *hbh;
678
679 if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
680#if 0 /*touches NULL pointer*/
681 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
682#endif
683 return; /* m have already been freed */
684 }
685
686 /* adjust pointer */
687 ip6 = mtod(m, struct ip6_hdr *);
688
689 /*
690 * if the payload length field is 0 and the next header field
691 * indicates Hop-by-Hop Options header, then a Jumbo Payload
692 * option MUST be included.
693 */
694 if (ip6->ip6_plen == 0 && plen == 0) {
695 /*
696 * Note that if a valid jumbo payload option is
697 * contained, ip6_hopopts_input() must set a valid
698 * (non-zero) payload length to the variable plen.
699 */
700 ip6stat.ip6s_badoptions++;
701 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
702 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
703 icmp6_error(m, ICMP6_PARAM_PROB,
704 ICMP6_PARAMPROB_HEADER,
705 (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
706 return;
707 }
708#ifndef PULLDOWN_TEST
709 /* ip6_hopopts_input() ensures that mbuf is contiguous */
710 hbh = (struct ip6_hbh *)(ip6 + 1);
711#else
712 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
713 sizeof(struct ip6_hbh));
714 if (hbh == NULL) {
715 ip6stat.ip6s_tooshort++;
716 return;
717 }
718#endif
719 nxt = hbh->ip6h_nxt;
720
721 /*
722 * accept the packet if a router alert option is included
723 * and we act as an IPv6 router.
724 */
725 if (rtalert != ~0 && ip6_forwarding)
726 ours = 1;
727 } else
728 nxt = ip6->ip6_nxt;
729
730 /*
731 * Check that the amount of data in the buffers
732 * is as at least much as the IPv6 header would have us expect.
733 * Trim mbufs if longer than we expect.
734 * Drop packet if shorter than we expect.
735 */
736 if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
737 ip6stat.ip6s_tooshort++;
738 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
739 goto bad;
740 }
741 if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
742 if (m->m_len == m->m_pkthdr.len) {
743 m->m_len = sizeof(struct ip6_hdr) + plen;
744 m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
745 } else
746 m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
747 }
748
749 /*
750 * Forward if desirable.
751 */
752 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
753 /*
754 * If we are acting as a multicast router, all
755 * incoming multicast packets are passed to the
756 * kernel-level multicast forwarding function.
757 * The packet is returned (relatively) intact; if
758 * ip6_mforward() returns a non-zero value, the packet
759 * must be discarded, else it may be accepted below.
760 */
761 if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
762 ip6stat.ip6s_cantforward++;
763 m_freem(m);
764 return;
765 }
766 if (!ours) {
767 m_freem(m);
768 return;
769 }
770 } else if (!ours) {
771 ip6_forward(m, srcrt);
772 return;
773 }
774
775 ip6 = mtod(m, struct ip6_hdr *);
776
777 /*
778 * Malicious party may be able to use IPv4 mapped addr to confuse
779 * tcp/udp stack and bypass security checks (act as if it was from
780 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1). Be cautious.
781 *
782 * For SIIT end node behavior, you may want to disable the check.
783 * However, you will become vulnerable to attacks using IPv4 mapped
784 * source.
785 */
786 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
787 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
788 ip6stat.ip6s_badscope++;
789 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
790 goto bad;
791 }
792
793 /*
794 * Tell launch routine the next header
795 */
796 ip6stat.ip6s_delivered++;
797 in6_ifstat_inc(deliverifp, ifs6_in_deliver);
798 nest = 0;
799
800 while (nxt != IPPROTO_DONE) {
801 if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
802 ip6stat.ip6s_toomanyhdr++;
803 goto bad;
804 }
805
806 /*
807 * protection against faulty packet - there should be
808 * more sanity checks in header chain processing.
809 */
810 if (m->m_pkthdr.len < off) {
811 ip6stat.ip6s_tooshort++;
812 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
813 goto bad;
814 }
815
816#ifdef IPSEC
817 /*
818 * enforce IPsec policy checking if we are seeing last header.
819 * note that we do not visit this with protocols with pcb layer
820 * code - like udp/tcp/raw ip.
821 */
822 if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
823 ipsec6_in_reject(m, NULL)) {
824 ipsec6stat.in_polvio++;
825 goto bad;
826 }
827#endif
828 nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
829 }
830 return;
831 bad:
832 m_freem(m);
833}
834
835/*
836 * set/grab in6_ifaddr correspond to IPv6 destination address.
837 * XXX backward compatibility wrapper
838 */
839static struct ip6aux *
840ip6_setdstifaddr(m, ia6)
841 struct mbuf *m;
842 struct in6_ifaddr *ia6;
843{
844 struct ip6aux *ip6a;
845
846 ip6a = ip6_addaux(m);
847 if (ip6a)
848 ip6a->ip6a_dstia6 = ia6;
849 return ip6a; /* NULL if failed to set */
850}
851
852struct in6_ifaddr *
853ip6_getdstifaddr(m)
854 struct mbuf *m;
855{
856 struct ip6aux *ip6a;
857
858 ip6a = ip6_findaux(m);
859 if (ip6a)
860 return ip6a->ip6a_dstia6;
861 else
862 return NULL;
863}
864
865/*
866 * Hop-by-Hop options header processing. If a valid jumbo payload option is
867 * included, the real payload length will be stored in plenp.
868 */
869static int
870ip6_hopopts_input(plenp, rtalertp, mp, offp)
871 u_int32_t *plenp;
872 u_int32_t *rtalertp; /* XXX: should be stored more smart way */
873 struct mbuf **mp;
874 int *offp;
875{
876 struct mbuf *m = *mp;
877 int off = *offp, hbhlen;
878 struct ip6_hbh *hbh;
879 u_int8_t *opt;
880
881 /* validation of the length of the header */
882#ifndef PULLDOWN_TEST
883 IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
884 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
885 hbhlen = (hbh->ip6h_len + 1) << 3;
886
887 IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
888 hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
889#else
890 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
891 sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
892 if (hbh == NULL) {
893 ip6stat.ip6s_tooshort++;
894 return -1;
895 }
896 hbhlen = (hbh->ip6h_len + 1) << 3;
897 IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
898 hbhlen);
899 if (hbh == NULL) {
900 ip6stat.ip6s_tooshort++;
901 return -1;
902 }
903#endif
904 off += hbhlen;
905 hbhlen -= sizeof(struct ip6_hbh);
906 opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
907
908 if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
909 hbhlen, rtalertp, plenp) < 0)
910 return (-1);
911
912 *offp = off;
913 *mp = m;
914 return (0);
915}
916
917/*
918 * Search header for all Hop-by-hop options and process each option.
919 * This function is separate from ip6_hopopts_input() in order to
920 * handle a case where the sending node itself process its hop-by-hop
921 * options header. In such a case, the function is called from ip6_output().
922 *
923 * The function assumes that hbh header is located right after the IPv6 header
924 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
925 * opthead + hbhlen is located in continuous memory region.
926 */
927int
928ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
929 struct mbuf *m;
930 u_int8_t *opthead;
931 int hbhlen;
932 u_int32_t *rtalertp;
933 u_int32_t *plenp;
934{
935 struct ip6_hdr *ip6;
936 int optlen = 0;
937 u_int8_t *opt = opthead;
938 u_int16_t rtalert_val;
939 u_int32_t jumboplen;
940 const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
941
942 for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
943 switch (*opt) {
944 case IP6OPT_PAD1:
945 optlen = 1;
946 break;
947 case IP6OPT_PADN:
948 if (hbhlen < IP6OPT_MINLEN) {
949 ip6stat.ip6s_toosmall++;
950 goto bad;
951 }
952 optlen = *(opt + 1) + 2;
953 break;
954 case IP6OPT_ROUTER_ALERT:
955 /* XXX may need check for alignment */
956 if (hbhlen < IP6OPT_RTALERT_LEN) {
957 ip6stat.ip6s_toosmall++;
958 goto bad;
959 }
960 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
961 /* XXX stat */
962 icmp6_error(m, ICMP6_PARAM_PROB,
963 ICMP6_PARAMPROB_HEADER,
964 erroff + opt + 1 - opthead);
965 return (-1);
966 }
967 optlen = IP6OPT_RTALERT_LEN;
968 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
969 *rtalertp = ntohs(rtalert_val);
970 break;
971 case IP6OPT_JUMBO:
972 /* XXX may need check for alignment */
973 if (hbhlen < IP6OPT_JUMBO_LEN) {
974 ip6stat.ip6s_toosmall++;
975 goto bad;
976 }
977 if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
978 /* XXX stat */
979 icmp6_error(m, ICMP6_PARAM_PROB,
980 ICMP6_PARAMPROB_HEADER,
981 erroff + opt + 1 - opthead);
982 return (-1);
983 }
984 optlen = IP6OPT_JUMBO_LEN;
985
986 /*
987 * IPv6 packets that have non 0 payload length
988 * must not contain a jumbo payload option.
989 */
990 ip6 = mtod(m, struct ip6_hdr *);
991 if (ip6->ip6_plen) {
992 ip6stat.ip6s_badoptions++;
993 icmp6_error(m, ICMP6_PARAM_PROB,
994 ICMP6_PARAMPROB_HEADER,
995 erroff + opt - opthead);
996 return (-1);
997 }
998
999 /*
1000 * We may see jumbolen in unaligned location, so
1001 * we'd need to perform bcopy().
1002 */
1003 bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
1004 jumboplen = (u_int32_t)htonl(jumboplen);
1005
1006#if 1
1007 /*
1008 * if there are multiple jumbo payload options,
1009 * *plenp will be non-zero and the packet will be
1010 * rejected.
1011 * the behavior may need some debate in ipngwg -
1012 * multiple options does not make sense, however,
1013 * there's no explicit mention in specification.
1014 */
1015 if (*plenp != 0) {
1016 ip6stat.ip6s_badoptions++;
1017 icmp6_error(m, ICMP6_PARAM_PROB,
1018 ICMP6_PARAMPROB_HEADER,
1019 erroff + opt + 2 - opthead);
1020 return (-1);
1021 }
1022#endif
1023
1024 /*
1025 * jumbo payload length must be larger than 65535.
1026 */
1027 if (jumboplen <= IPV6_MAXPACKET) {
1028 ip6stat.ip6s_badoptions++;
1029 icmp6_error(m, ICMP6_PARAM_PROB,
1030 ICMP6_PARAMPROB_HEADER,
1031 erroff + opt + 2 - opthead);
1032 return (-1);
1033 }
1034 *plenp = jumboplen;
1035
1036 break;
1037 default: /* unknown option */
1038 if (hbhlen < IP6OPT_MINLEN) {
1039 ip6stat.ip6s_toosmall++;
1040 goto bad;
1041 }
1042 optlen = ip6_unknown_opt(opt, m,
1043 erroff + opt - opthead);
1044 if (optlen == -1)
1045 return (-1);
1046 optlen += 2;
1047 break;
1048 }
1049 }
1050
1051 return (0);
1052
1053 bad:
1054 m_freem(m);
1055 return (-1);
1056}
1057
1058/*
1059 * Unknown option processing.
1060 * The third argument `off' is the offset from the IPv6 header to the option,
1061 * which is necessary if the IPv6 header the and option header and IPv6 header
1062 * is not continuous in order to return an ICMPv6 error.
1063 */
1064int
1065ip6_unknown_opt(optp, m, off)
1066 u_int8_t *optp;
1067 struct mbuf *m;
1068 int off;
1069{
1070 struct ip6_hdr *ip6;
1071
1072 switch (IP6OPT_TYPE(*optp)) {
1073 case IP6OPT_TYPE_SKIP: /* ignore the option */
1074 return ((int)*(optp + 1));
1075 case IP6OPT_TYPE_DISCARD: /* silently discard */
1076 m_freem(m);
1077 return (-1);
1078 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1079 ip6stat.ip6s_badoptions++;
1080 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1081 return (-1);
1082 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1083 ip6stat.ip6s_badoptions++;
1084 ip6 = mtod(m, struct ip6_hdr *);
1085 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1086 (m->m_flags & (M_BCAST|M_MCAST)))
1087 m_freem(m);
1088 else
1089 icmp6_error(m, ICMP6_PARAM_PROB,
1090 ICMP6_PARAMPROB_OPTION, off);
1091 return (-1);
1092 }
1093
1094 m_freem(m); /* XXX: NOTREACHED */
1095 return (-1);
1096}
1097
1098/*
1099 * Create the "control" list for this pcb.
1100 * The function will not modify mbuf chain at all.
1101 *
1102 * with KAME mbuf chain restriction:
1103 * The routine will be called from upper layer handlers like tcp6_input().
1104 * Thus the routine assumes that the caller (tcp6_input) have already
1105 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1106 * very first mbuf on the mbuf chain.
1107 */
1108void
1109ip6_savecontrol(in6p, m, mp)
1110 struct inpcb *in6p;
1111 struct mbuf *m, **mp;
1112{
1113#define IS2292(x, y) ((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
1114 struct thread *td = curthread; /* XXX */
1115 int privileged = 0;
1116 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1117
1118 if (td && !suser(td))
1119 privileged++;
1120
1121#ifdef SO_TIMESTAMP
1122 if ((in6p->in6p_socket->so_options & SO_TIMESTAMP) != 0) {
1123 struct timeval tv;
1124
1125 microtime(&tv);
1126 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1127 SCM_TIMESTAMP, SOL_SOCKET);
1128 if (*mp)
1129 mp = &(*mp)->m_next;
1130 }
1131#endif
1132
1133 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
1134 return;
1135
1136 /* RFC 2292 sec. 5 */
1137 if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1138 struct in6_pktinfo pi6;
1139
1140 bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1141 in6_clearscope(&pi6.ipi6_addr); /* XXX */
1142 pi6.ipi6_ifindex =
1143 (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1144
1145 *mp = sbcreatecontrol((caddr_t) &pi6,
1146 sizeof(struct in6_pktinfo),
1147 IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
1148 if (*mp)
1149 mp = &(*mp)->m_next;
1150 }
1151
1152 if ((in6p->in6p_flags & IN6P_HOPLIMIT) != 0) {
1153 int hlim = ip6->ip6_hlim & 0xff;
1154
1155 *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int),
1156 IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
1157 if (*mp)
1158 mp = &(*mp)->m_next;
1159 }
1160
1161 if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
1162 u_int32_t flowinfo;
1163 int tclass;
1164
1165 flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
1166 flowinfo >>= 20;
1167
1168 tclass = flowinfo & 0xff;
1169 *mp = sbcreatecontrol((caddr_t) &tclass, sizeof(tclass),
1170 IPV6_TCLASS, IPPROTO_IPV6);
1171 if (*mp)
1172 mp = &(*mp)->m_next;
1173 }
1174
1175 /*
1176 * IPV6_HOPOPTS socket option. We require super-user privilege
1177 * for the option, but it might be too strict, since there might
1178 * be some hop-by-hop options which can be returned to normal user.
1179 * See RFC 2292 section 6.
1180 */
1181 if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
1182#ifdef DIAGNOSTIC
1183 if (!privileged)
1184 panic("IN6P_HOPOPTS is set for unprivileged socket");
1185#endif
1186 /*
1187 * Check if a hop-by-hop options header is contatined in the
1188 * received packet, and if so, store the options as ancillary
1189 * data. Note that a hop-by-hop options header must be
1190 * just after the IPv6 header, which is assured through the
1191 * IPv6 input processing.
1192 */
1193 if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1194 struct ip6_hbh *hbh;
1195 int hbhlen = 0;
1196#ifdef PULLDOWN_TEST
1197 struct mbuf *ext;
1198#endif
1199
1200#ifndef PULLDOWN_TEST
1201 hbh = (struct ip6_hbh *)(ip6 + 1);
1202 hbhlen = (hbh->ip6h_len + 1) << 3;
1203#else
1204 ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1205 ip6->ip6_nxt);
1206 if (ext == NULL) {
1207 ip6stat.ip6s_tooshort++;
1208 return;
1209 }
1210 hbh = mtod(ext, struct ip6_hbh *);
1211 hbhlen = (hbh->ip6h_len + 1) << 3;
1212 if (hbhlen != ext->m_len) {
1213 m_freem(ext);
1214 ip6stat.ip6s_tooshort++;
1215 return;
1216 }
1217#endif
1218
1219 /*
1220 * XXX: We copy the whole header even if a
1221 * jumbo payload option is included, the option which
1222 * is to be removed before returning according to
1223 * RFC2292.
1224 * Note: this constraint is removed in 2292bis.
1225 */
1226 *mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1227 IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
1228 IPPROTO_IPV6);
1229 if (*mp)
1230 mp = &(*mp)->m_next;
1231#ifdef PULLDOWN_TEST
1232 m_freem(ext);
1233#endif
1234 }
1235 }
1236
1237 if ((in6p->in6p_flags & (IN6P_RTHDR | IN6P_DSTOPTS)) != 0) {
1238 int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1239
1240 /*
1241 * Search for destination options headers or routing
1242 * header(s) through the header chain, and stores each
1243 * header as ancillary data.
1244 * Note that the order of the headers remains in
1245 * the chain of ancillary data.
1246 */
1247 while (1) { /* is explicit loop prevention necessary? */
1248 struct ip6_ext *ip6e = NULL;
1249 int elen;
1250#ifdef PULLDOWN_TEST
1251 struct mbuf *ext = NULL;
1252#endif
1253
1254 /*
1255 * if it is not an extension header, don't try to
1256 * pull it from the chain.
1257 */
1258 switch (nxt) {
1259 case IPPROTO_DSTOPTS:
1260 case IPPROTO_ROUTING:
1261 case IPPROTO_HOPOPTS:
1262 case IPPROTO_AH: /* is it possible? */
1263 break;
1264 default:
1265 goto loopend;
1266 }
1267
1268#ifndef PULLDOWN_TEST
1269 if (off + sizeof(*ip6e) > m->m_len)
1270 goto loopend;
1271 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1272 if (nxt == IPPROTO_AH)
1273 elen = (ip6e->ip6e_len + 2) << 2;
1274 else
1275 elen = (ip6e->ip6e_len + 1) << 3;
1276 if (off + elen > m->m_len)
1277 goto loopend;
1278#else
1279 ext = ip6_pullexthdr(m, off, nxt);
1280 if (ext == NULL) {
1281 ip6stat.ip6s_tooshort++;
1282 return;
1283 }
1284 ip6e = mtod(ext, struct ip6_ext *);
1285 if (nxt == IPPROTO_AH)
1286 elen = (ip6e->ip6e_len + 2) << 2;
1287 else
1288 elen = (ip6e->ip6e_len + 1) << 3;
1289 if (elen != ext->m_len) {
1290 m_freem(ext);
1291 ip6stat.ip6s_tooshort++;
1292 return;
1293 }
1294#endif
1295
1296 switch (nxt) {
1297 case IPPROTO_DSTOPTS:
1298 if (!(in6p->in6p_flags & IN6P_DSTOPTS))
1299 break;
1300
1301 /*
1302 * We also require super-user privilege for
1303 * the option. See comments on IN6_HOPOPTS.
1304 */
1305 if (!privileged)
1306 break;
1307
1308 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1309 IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
1310 IPPROTO_IPV6);
1311 if (*mp)
1312 mp = &(*mp)->m_next;
1313 break;
1314 case IPPROTO_ROUTING:
1315 if (!in6p->in6p_flags & IN6P_RTHDR)
1316 break;
1317
1318 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1319 IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
1320 IPPROTO_IPV6);
1321 if (*mp)
1322 mp = &(*mp)->m_next;
1323 break;
1324 case IPPROTO_HOPOPTS:
1325 case IPPROTO_AH: /* is it possible? */
1326 break;
1327
1328 default:
1329 /*
1330 * other cases have been filtered in the above.
1331 * none will visit this case. here we supply
1332 * the code just in case (nxt overwritten or
1333 * other cases).
1334 */
1335#ifdef PULLDOWN_TEST
1336 m_freem(ext);
1337#endif
1338 goto loopend;
1339
1340 }
1341
1342 /* proceed with the next header. */
1343 off += elen;
1344 nxt = ip6e->ip6e_nxt;
1345 ip6e = NULL;
1346#ifdef PULLDOWN_TEST
1347 m_freem(ext);
1348 ext = NULL;
1349#endif
1350 }
1351 loopend:
1352 ;
1353 }
1354
1355#undef IS2292
1356}
1357
1358void
1359ip6_notify_pmtu(in6p, dst, mtu)
1360 struct inpcb *in6p;
1361 struct sockaddr_in6 *dst;
1362 u_int32_t *mtu;
1363{
1364 struct socket *so;
1365 struct mbuf *m_mtu;
1366 struct ip6_mtuinfo mtuctl;
1367
1368 so = in6p->inp_socket;
1369
1370 if (mtu == NULL)
1371 return;
1372
1373#ifdef DIAGNOSTIC
1374 if (so == NULL) /* I believe this is impossible */
1375 panic("ip6_notify_pmtu: socket is NULL");
1376#endif
1377
1378 bzero(&mtuctl, sizeof(mtuctl)); /* zero-clear for safety */
1379 mtuctl.ip6m_mtu = *mtu;
1380 mtuctl.ip6m_addr = *dst;
1381 in6_recoverscope(&mtuctl.ip6m_addr, &mtuctl.ip6m_addr.sin6_addr, NULL);
1382
1383 if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl),
1384 IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
1385 return;
1386
1387 if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu)
1388 == 0) {
1389 m_freem(m_mtu);
1390 /* XXX: should count statistics */
1391 } else
1392 sorwakeup(so);
1393
1394 return;
1395}
1396
1397#ifdef PULLDOWN_TEST
1398/*
1399 * pull single extension header from mbuf chain. returns single mbuf that
1400 * contains the result, or NULL on error.
1401 */
1402static struct mbuf *
1403ip6_pullexthdr(m, off, nxt)
1404 struct mbuf *m;
1405 size_t off;
1406 int nxt;
1407{
1408 struct ip6_ext ip6e;
1409 size_t elen;
1410 struct mbuf *n;
1411
1412#ifdef DIAGNOSTIC
1413 switch (nxt) {
1414 case IPPROTO_DSTOPTS:
1415 case IPPROTO_ROUTING:
1416 case IPPROTO_HOPOPTS:
1417 case IPPROTO_AH: /* is it possible? */
1418 break;
1419 default:
1420 printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1421 }
1422#endif
1423
1424 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1425 if (nxt == IPPROTO_AH)
1426 elen = (ip6e.ip6e_len + 2) << 2;
1427 else
1428 elen = (ip6e.ip6e_len + 1) << 3;
1429
1430 MGET(n, M_DONTWAIT, MT_DATA);
1431 if (n && elen >= MLEN) {
1432 MCLGET(n, M_DONTWAIT);
1433 if ((n->m_flags & M_EXT) == 0) {
1434 m_free(n);
1435 n = NULL;
1436 }
1437 }
1438 if (!n)
1439 return NULL;
1440
1441 n->m_len = 0;
1442 if (elen >= M_TRAILINGSPACE(n)) {
1443 m_free(n);
1444 return NULL;
1445 }
1446
1447 m_copydata(m, off, elen, mtod(n, caddr_t));
1448 n->m_len = elen;
1449 return n;
1450}
1451#endif
1452
1453/*
1454 * Get pointer to the previous header followed by the header
1455 * currently processed.
1456 * XXX: This function supposes that
1457 * M includes all headers,
1458 * the next header field and the header length field of each header
1459 * are valid, and
1460 * the sum of each header length equals to OFF.
1461 * Because of these assumptions, this function must be called very
1462 * carefully. Moreover, it will not be used in the near future when
1463 * we develop `neater' mechanism to process extension headers.
1464 */
1465char *
1466ip6_get_prevhdr(m, off)
1467 struct mbuf *m;
1468 int off;
1469{
1470 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1471
1472 if (off == sizeof(struct ip6_hdr))
1473 return (&ip6->ip6_nxt);
1474 else {
1475 int len, nxt;
1476 struct ip6_ext *ip6e = NULL;
1477
1478 nxt = ip6->ip6_nxt;
1479 len = sizeof(struct ip6_hdr);
1480 while (len < off) {
1481 ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1482
1483 switch (nxt) {
1484 case IPPROTO_FRAGMENT:
1485 len += sizeof(struct ip6_frag);
1486 break;
1487 case IPPROTO_AH:
1488 len += (ip6e->ip6e_len + 2) << 2;
1489 break;
1490 default:
1491 len += (ip6e->ip6e_len + 1) << 3;
1492 break;
1493 }
1494 nxt = ip6e->ip6e_nxt;
1495 }
1496 if (ip6e)
1497 return (&ip6e->ip6e_nxt);
1498 else
1499 return NULL;
1500 }
1501}
1502
1503/*
1504 * get next header offset. m will be retained.
1505 */
1506int
1507ip6_nexthdr(m, off, proto, nxtp)
1508 struct mbuf *m;
1509 int off;
1510 int proto;
1511 int *nxtp;
1512{
1513 struct ip6_hdr ip6;
1514 struct ip6_ext ip6e;
1515 struct ip6_frag fh;
1516
1517 /* just in case */
1518 if (m == NULL)
1519 panic("ip6_nexthdr: m == NULL");
1520 if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1521 return -1;
1522
1523 switch (proto) {
1524 case IPPROTO_IPV6:
1525 if (m->m_pkthdr.len < off + sizeof(ip6))
1526 return -1;
1527 m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1528 if (nxtp)
1529 *nxtp = ip6.ip6_nxt;
1530 off += sizeof(ip6);
1531 return off;
1532
1533 case IPPROTO_FRAGMENT:
1534 /*
1535 * terminate parsing if it is not the first fragment,
1536 * it does not make sense to parse through it.
1537 */
1538 if (m->m_pkthdr.len < off + sizeof(fh))
1539 return -1;
1540 m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1541 /* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1542 if (fh.ip6f_offlg & IP6F_OFF_MASK)
1543 return -1;
1544 if (nxtp)
1545 *nxtp = fh.ip6f_nxt;
1546 off += sizeof(struct ip6_frag);
1547 return off;
1548
1549 case IPPROTO_AH:
1550 if (m->m_pkthdr.len < off + sizeof(ip6e))
1551 return -1;
1552 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1553 if (nxtp)
1554 *nxtp = ip6e.ip6e_nxt;
1555 off += (ip6e.ip6e_len + 2) << 2;
1556 return off;
1557
1558 case IPPROTO_HOPOPTS:
1559 case IPPROTO_ROUTING:
1560 case IPPROTO_DSTOPTS:
1561 if (m->m_pkthdr.len < off + sizeof(ip6e))
1562 return -1;
1563 m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1564 if (nxtp)
1565 *nxtp = ip6e.ip6e_nxt;
1566 off += (ip6e.ip6e_len + 1) << 3;
1567 return off;
1568
1569 case IPPROTO_NONE:
1570 case IPPROTO_ESP:
1571 case IPPROTO_IPCOMP:
1572 /* give up */
1573 return -1;
1574
1575 default:
1576 return -1;
1577 }
1578
1579 return -1;
1580}
1581
1582/*
1583 * get offset for the last header in the chain. m will be kept untainted.
1584 */
1585int
1586ip6_lasthdr(m, off, proto, nxtp)
1587 struct mbuf *m;
1588 int off;
1589 int proto;
1590 int *nxtp;
1591{
1592 int newoff;
1593 int nxt;
1594
1595 if (!nxtp) {
1596 nxt = -1;
1597 nxtp = &nxt;
1598 }
1599 while (1) {
1600 newoff = ip6_nexthdr(m, off, proto, nxtp);
1601 if (newoff < 0)
1602 return off;
1603 else if (newoff < off)
1604 return -1; /* invalid */
1605 else if (newoff == off)
1606 return newoff;
1607
1608 off = newoff;
1609 proto = *nxtp;
1610 }
1611}
1612
1613struct ip6aux *
1614ip6_addaux(m)
1615 struct mbuf *m;
1616{
1617 struct m_tag *mtag;
1618
1619 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1620 if (!mtag) {
1621 mtag = m_tag_get(PACKET_TAG_IPV6_INPUT, sizeof(struct ip6aux),
1622 M_NOWAIT);
1623 if (mtag) {
1624 m_tag_prepend(m, mtag);
1625 bzero(mtag + 1, sizeof(struct ip6aux));
1626 }
1627 }
1628 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1629}
1630
1631struct ip6aux *
1632ip6_findaux(m)
1633 struct mbuf *m;
1634{
1635 struct m_tag *mtag;
1636
1637 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1638 return mtag ? (struct ip6aux *)(mtag + 1) : NULL;
1639}
1640
1641void
1642ip6_delaux(m)
1643 struct mbuf *m;
1644{
1645 struct m_tag *mtag;
1646
1647 mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1648 if (mtag)
1649 m_tag_delete(m, mtag);
1650}
1651
1652/*
1653 * System control for IP6
1654 */
1655
1656u_char inet6ctlerrmap[PRC_NCMDS] = {
1657 0, 0, 0, 0,
1658 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1659 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1660 EMSGSIZE, EHOSTUNREACH, 0, 0,
1661 0, 0, 0, 0,
1662 ENOPROTOOPT
1663};
|