1 | #include <machine/rtems-bsd-kernel-space.h> |
---|
2 | |
---|
3 | /* $FreeBSD$ */ |
---|
4 | /* $KAME: if_stf.c,v 1.73 2001/12/03 11:08:30 keiichi Exp $ */ |
---|
5 | |
---|
6 | /*- |
---|
7 | * Copyright (C) 2000 WIDE Project. |
---|
8 | * All rights reserved. |
---|
9 | * |
---|
10 | * Redistribution and use in source and binary forms, with or without |
---|
11 | * modification, are permitted provided that the following conditions |
---|
12 | * are met: |
---|
13 | * 1. Redistributions of source code must retain the above copyright |
---|
14 | * notice, this list of conditions and the following disclaimer. |
---|
15 | * 2. Redistributions in binary form must reproduce the above copyright |
---|
16 | * notice, this list of conditions and the following disclaimer in the |
---|
17 | * documentation and/or other materials provided with the distribution. |
---|
18 | * 3. Neither the name of the project nor the names of its contributors |
---|
19 | * may be used to endorse or promote products derived from this software |
---|
20 | * without specific prior written permission. |
---|
21 | * |
---|
22 | * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND |
---|
23 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
---|
24 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
---|
25 | * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE |
---|
26 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
---|
27 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
---|
28 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
---|
29 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
---|
30 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
---|
31 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
---|
32 | * SUCH DAMAGE. |
---|
33 | */ |
---|
34 | |
---|
35 | /* |
---|
36 | * 6to4 interface, based on RFC3056. |
---|
37 | * |
---|
38 | * 6to4 interface is NOT capable of link-layer (I mean, IPv4) multicasting. |
---|
39 | * There is no address mapping defined from IPv6 multicast address to IPv4 |
---|
40 | * address. Therefore, we do not have IFF_MULTICAST on the interface. |
---|
41 | * |
---|
42 | * Due to the lack of address mapping for link-local addresses, we cannot |
---|
43 | * throw packets toward link-local addresses (fe80::x). Also, we cannot throw |
---|
44 | * packets to link-local multicast addresses (ff02::x). |
---|
45 | * |
---|
46 | * Here are interesting symptoms due to the lack of link-local address: |
---|
47 | * |
---|
48 | * Unicast routing exchange: |
---|
49 | * - RIPng: Impossible. Uses link-local multicast packet toward ff02::9, |
---|
50 | * and link-local addresses as nexthop. |
---|
51 | * - OSPFv6: Impossible. OSPFv6 assumes that there's link-local address |
---|
52 | * assigned to the link, and makes use of them. Also, HELLO packets use |
---|
53 | * link-local multicast addresses (ff02::5 and ff02::6). |
---|
54 | * - BGP4+: Maybe. You can only use global address as nexthop, and global |
---|
55 | * address as TCP endpoint address. |
---|
56 | * |
---|
57 | * Multicast routing protocols: |
---|
58 | * - PIM: Hello packet cannot be used to discover adjacent PIM routers. |
---|
59 | * Adjacent PIM routers must be configured manually (is it really spec-wise |
---|
60 | * correct thing to do?). |
---|
61 | * |
---|
62 | * ICMPv6: |
---|
63 | * - Redirects cannot be used due to the lack of link-local address. |
---|
64 | * |
---|
65 | * stf interface does not have, and will not need, a link-local address. |
---|
66 | * It seems to have no real benefit and does not help the above symptoms much. |
---|
67 | * Even if we assign link-locals to interface, we cannot really |
---|
68 | * use link-local unicast/multicast on top of 6to4 cloud (since there's no |
---|
69 | * encapsulation defined for link-local address), and the above analysis does |
---|
70 | * not change. RFC3056 does not mandate the assignment of link-local address |
---|
71 | * either. |
---|
72 | * |
---|
73 | * 6to4 interface has security issues. Refer to |
---|
74 | * http://playground.iijlab.net/i-d/draft-itojun-ipv6-transition-abuse-00.txt |
---|
75 | * for details. The code tries to filter out some of malicious packets. |
---|
76 | * Note that there is no way to be 100% secure. |
---|
77 | */ |
---|
78 | |
---|
79 | #include <sys/param.h> |
---|
80 | #include <sys/systm.h> |
---|
81 | #include <sys/socket.h> |
---|
82 | #include <sys/sockio.h> |
---|
83 | #include <sys/mbuf.h> |
---|
84 | #include <sys/errno.h> |
---|
85 | #include <sys/kernel.h> |
---|
86 | #include <sys/lock.h> |
---|
87 | #include <sys/module.h> |
---|
88 | #include <sys/protosw.h> |
---|
89 | #include <sys/proc.h> |
---|
90 | #include <sys/queue.h> |
---|
91 | #include <sys/rmlock.h> |
---|
92 | #include <sys/sysctl.h> |
---|
93 | #include <machine/cpu.h> |
---|
94 | |
---|
95 | #include <sys/malloc.h> |
---|
96 | |
---|
97 | #include <net/if.h> |
---|
98 | #include <net/if_var.h> |
---|
99 | #include <net/if_clone.h> |
---|
100 | #include <net/route.h> |
---|
101 | #include <net/netisr.h> |
---|
102 | #include <net/if_types.h> |
---|
103 | #include <net/vnet.h> |
---|
104 | |
---|
105 | #include <netinet/in.h> |
---|
106 | #include <netinet/in_fib.h> |
---|
107 | #include <netinet/in_systm.h> |
---|
108 | #include <netinet/ip.h> |
---|
109 | #include <netinet/ip_var.h> |
---|
110 | #include <netinet/in_var.h> |
---|
111 | |
---|
112 | #include <netinet/ip6.h> |
---|
113 | #include <netinet6/ip6_var.h> |
---|
114 | #include <netinet6/in6_var.h> |
---|
115 | #include <netinet/ip_ecn.h> |
---|
116 | |
---|
117 | #include <netinet/ip_encap.h> |
---|
118 | |
---|
119 | #include <machine/stdarg.h> |
---|
120 | |
---|
121 | #include <net/bpf.h> |
---|
122 | |
---|
123 | #include <security/mac/mac_framework.h> |
---|
124 | |
---|
125 | SYSCTL_DECL(_net_link); |
---|
126 | static SYSCTL_NODE(_net_link, IFT_STF, stf, CTLFLAG_RW, 0, "6to4 Interface"); |
---|
127 | |
---|
128 | static int stf_permit_rfc1918 = 0; |
---|
129 | SYSCTL_INT(_net_link_stf, OID_AUTO, permit_rfc1918, CTLFLAG_RWTUN, |
---|
130 | &stf_permit_rfc1918, 0, "Permit the use of private IPv4 addresses"); |
---|
131 | |
---|
132 | #define STFUNIT 0 |
---|
133 | |
---|
134 | #define IN6_IS_ADDR_6TO4(x) (ntohs((x)->s6_addr16[0]) == 0x2002) |
---|
135 | |
---|
136 | /* |
---|
137 | * XXX: Return a pointer with 16-bit aligned. Don't cast it to |
---|
138 | * struct in_addr *; use bcopy() instead. |
---|
139 | */ |
---|
140 | #define GET_V4(x) (&(x)->s6_addr16[1]) |
---|
141 | |
---|
142 | struct stf_softc { |
---|
143 | struct ifnet *sc_ifp; |
---|
144 | struct mtx sc_ro_mtx; |
---|
145 | u_int sc_fibnum; |
---|
146 | const struct encaptab *encap_cookie; |
---|
147 | }; |
---|
148 | #define STF2IFP(sc) ((sc)->sc_ifp) |
---|
149 | |
---|
150 | static const char stfname[] = "stf"; |
---|
151 | |
---|
152 | /* |
---|
153 | * Note that mutable fields in the softc are not currently locked. |
---|
154 | * We do lock sc_ro in stf_output though. |
---|
155 | */ |
---|
156 | static MALLOC_DEFINE(M_STF, stfname, "6to4 Tunnel Interface"); |
---|
157 | static const int ip_stf_ttl = 40; |
---|
158 | |
---|
159 | extern struct domain inetdomain; |
---|
160 | static int in_stf_input(struct mbuf **, int *, int); |
---|
161 | static struct protosw in_stf_protosw = { |
---|
162 | .pr_type = SOCK_RAW, |
---|
163 | .pr_domain = &inetdomain, |
---|
164 | .pr_protocol = IPPROTO_IPV6, |
---|
165 | .pr_flags = PR_ATOMIC|PR_ADDR, |
---|
166 | .pr_input = in_stf_input, |
---|
167 | .pr_output = rip_output, |
---|
168 | .pr_ctloutput = rip_ctloutput, |
---|
169 | .pr_usrreqs = &rip_usrreqs |
---|
170 | }; |
---|
171 | |
---|
172 | static char *stfnames[] = {"stf0", "stf", "6to4", NULL}; |
---|
173 | |
---|
174 | static int stfmodevent(module_t, int, void *); |
---|
175 | static int stf_encapcheck(const struct mbuf *, int, int, void *); |
---|
176 | static int stf_getsrcifa6(struct ifnet *, struct in6_addr *, struct in6_addr *); |
---|
177 | static int stf_output(struct ifnet *, struct mbuf *, const struct sockaddr *, |
---|
178 | struct route *); |
---|
179 | static int isrfc1918addr(struct in_addr *); |
---|
180 | static int stf_checkaddr4(struct stf_softc *, struct in_addr *, |
---|
181 | struct ifnet *); |
---|
182 | static int stf_checkaddr6(struct stf_softc *, struct in6_addr *, |
---|
183 | struct ifnet *); |
---|
184 | static int stf_ioctl(struct ifnet *, u_long, caddr_t); |
---|
185 | |
---|
186 | static int stf_clone_match(struct if_clone *, const char *); |
---|
187 | static int stf_clone_create(struct if_clone *, char *, size_t, caddr_t); |
---|
188 | static int stf_clone_destroy(struct if_clone *, struct ifnet *); |
---|
189 | static struct if_clone *stf_cloner; |
---|
190 | |
---|
191 | static int |
---|
192 | stf_clone_match(struct if_clone *ifc, const char *name) |
---|
193 | { |
---|
194 | int i; |
---|
195 | |
---|
196 | for(i = 0; stfnames[i] != NULL; i++) { |
---|
197 | if (strcmp(stfnames[i], name) == 0) |
---|
198 | return (1); |
---|
199 | } |
---|
200 | |
---|
201 | return (0); |
---|
202 | } |
---|
203 | |
---|
204 | static int |
---|
205 | stf_clone_create(struct if_clone *ifc, char *name, size_t len, caddr_t params) |
---|
206 | { |
---|
207 | char *dp; |
---|
208 | int err, unit, wildcard; |
---|
209 | struct stf_softc *sc; |
---|
210 | struct ifnet *ifp; |
---|
211 | |
---|
212 | err = ifc_name2unit(name, &unit); |
---|
213 | if (err != 0) |
---|
214 | return (err); |
---|
215 | wildcard = (unit < 0); |
---|
216 | |
---|
217 | /* |
---|
218 | * We can only have one unit, but since unit allocation is |
---|
219 | * already locked, we use it to keep from allocating extra |
---|
220 | * interfaces. |
---|
221 | */ |
---|
222 | unit = STFUNIT; |
---|
223 | err = ifc_alloc_unit(ifc, &unit); |
---|
224 | if (err != 0) |
---|
225 | return (err); |
---|
226 | |
---|
227 | sc = malloc(sizeof(struct stf_softc), M_STF, M_WAITOK | M_ZERO); |
---|
228 | ifp = STF2IFP(sc) = if_alloc(IFT_STF); |
---|
229 | if (ifp == NULL) { |
---|
230 | free(sc, M_STF); |
---|
231 | ifc_free_unit(ifc, unit); |
---|
232 | return (ENOSPC); |
---|
233 | } |
---|
234 | ifp->if_softc = sc; |
---|
235 | #ifndef __rtems__ |
---|
236 | sc->sc_fibnum = curthread->td_proc->p_fibnum; |
---|
237 | #else /* __rtems__ */ |
---|
238 | sc->sc_fibnum = BSD_DEFAULT_FIB; |
---|
239 | #endif /* __rtems__ */ |
---|
240 | |
---|
241 | /* |
---|
242 | * Set the name manually rather then using if_initname because |
---|
243 | * we don't conform to the default naming convention for interfaces. |
---|
244 | * In the wildcard case, we need to update the name. |
---|
245 | */ |
---|
246 | if (wildcard) { |
---|
247 | for (dp = name; *dp != '\0'; dp++); |
---|
248 | if (snprintf(dp, len - (dp-name), "%d", unit) > |
---|
249 | len - (dp-name) - 1) { |
---|
250 | /* |
---|
251 | * This can only be a programmer error and |
---|
252 | * there's no straightforward way to recover if |
---|
253 | * it happens. |
---|
254 | */ |
---|
255 | panic("if_clone_create(): interface name too long"); |
---|
256 | } |
---|
257 | } |
---|
258 | strlcpy(ifp->if_xname, name, IFNAMSIZ); |
---|
259 | ifp->if_dname = stfname; |
---|
260 | ifp->if_dunit = IF_DUNIT_NONE; |
---|
261 | |
---|
262 | mtx_init(&(sc)->sc_ro_mtx, "stf ro", NULL, MTX_DEF); |
---|
263 | sc->encap_cookie = encap_attach_func(AF_INET, IPPROTO_IPV6, |
---|
264 | stf_encapcheck, &in_stf_protosw, sc); |
---|
265 | if (sc->encap_cookie == NULL) { |
---|
266 | if_printf(ifp, "attach failed\n"); |
---|
267 | free(sc, M_STF); |
---|
268 | ifc_free_unit(ifc, unit); |
---|
269 | return (ENOMEM); |
---|
270 | } |
---|
271 | |
---|
272 | ifp->if_mtu = IPV6_MMTU; |
---|
273 | ifp->if_ioctl = stf_ioctl; |
---|
274 | ifp->if_output = stf_output; |
---|
275 | ifp->if_snd.ifq_maxlen = ifqmaxlen; |
---|
276 | if_attach(ifp); |
---|
277 | bpfattach(ifp, DLT_NULL, sizeof(u_int32_t)); |
---|
278 | return (0); |
---|
279 | } |
---|
280 | |
---|
281 | static int |
---|
282 | stf_clone_destroy(struct if_clone *ifc, struct ifnet *ifp) |
---|
283 | { |
---|
284 | struct stf_softc *sc = ifp->if_softc; |
---|
285 | int err; |
---|
286 | |
---|
287 | err = encap_detach(sc->encap_cookie); |
---|
288 | KASSERT(err == 0, ("Unexpected error detaching encap_cookie")); |
---|
289 | mtx_destroy(&(sc)->sc_ro_mtx); |
---|
290 | bpfdetach(ifp); |
---|
291 | if_detach(ifp); |
---|
292 | if_free(ifp); |
---|
293 | |
---|
294 | free(sc, M_STF); |
---|
295 | ifc_free_unit(ifc, STFUNIT); |
---|
296 | |
---|
297 | return (0); |
---|
298 | } |
---|
299 | |
---|
300 | static int |
---|
301 | stfmodevent(module_t mod, int type, void *data) |
---|
302 | { |
---|
303 | |
---|
304 | switch (type) { |
---|
305 | case MOD_LOAD: |
---|
306 | stf_cloner = if_clone_advanced(stfname, 0, stf_clone_match, |
---|
307 | stf_clone_create, stf_clone_destroy); |
---|
308 | break; |
---|
309 | case MOD_UNLOAD: |
---|
310 | if_clone_detach(stf_cloner); |
---|
311 | break; |
---|
312 | default: |
---|
313 | return (EOPNOTSUPP); |
---|
314 | } |
---|
315 | |
---|
316 | return (0); |
---|
317 | } |
---|
318 | |
---|
319 | static moduledata_t stf_mod = { |
---|
320 | "if_stf", |
---|
321 | stfmodevent, |
---|
322 | 0 |
---|
323 | }; |
---|
324 | |
---|
325 | DECLARE_MODULE(if_stf, stf_mod, SI_SUB_PSEUDO, SI_ORDER_ANY); |
---|
326 | |
---|
327 | static int |
---|
328 | stf_encapcheck(const struct mbuf *m, int off, int proto, void *arg) |
---|
329 | { |
---|
330 | struct ip ip; |
---|
331 | struct stf_softc *sc; |
---|
332 | struct in_addr a, b, mask; |
---|
333 | struct in6_addr addr6, mask6; |
---|
334 | |
---|
335 | sc = (struct stf_softc *)arg; |
---|
336 | if (sc == NULL) |
---|
337 | return 0; |
---|
338 | |
---|
339 | if ((STF2IFP(sc)->if_flags & IFF_UP) == 0) |
---|
340 | return 0; |
---|
341 | |
---|
342 | /* IFF_LINK0 means "no decapsulation" */ |
---|
343 | if ((STF2IFP(sc)->if_flags & IFF_LINK0) != 0) |
---|
344 | return 0; |
---|
345 | |
---|
346 | if (proto != IPPROTO_IPV6) |
---|
347 | return 0; |
---|
348 | |
---|
349 | m_copydata(m, 0, sizeof(ip), (caddr_t)&ip); |
---|
350 | |
---|
351 | if (ip.ip_v != 4) |
---|
352 | return 0; |
---|
353 | |
---|
354 | if (stf_getsrcifa6(STF2IFP(sc), &addr6, &mask6) != 0) |
---|
355 | return (0); |
---|
356 | |
---|
357 | /* |
---|
358 | * check if IPv4 dst matches the IPv4 address derived from the |
---|
359 | * local 6to4 address. |
---|
360 | * success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:... |
---|
361 | */ |
---|
362 | if (bcmp(GET_V4(&addr6), &ip.ip_dst, sizeof(ip.ip_dst)) != 0) |
---|
363 | return 0; |
---|
364 | |
---|
365 | /* |
---|
366 | * check if IPv4 src matches the IPv4 address derived from the |
---|
367 | * local 6to4 address masked by prefixmask. |
---|
368 | * success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24 |
---|
369 | * fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24 |
---|
370 | */ |
---|
371 | bzero(&a, sizeof(a)); |
---|
372 | bcopy(GET_V4(&addr6), &a, sizeof(a)); |
---|
373 | bcopy(GET_V4(&mask6), &mask, sizeof(mask)); |
---|
374 | a.s_addr &= mask.s_addr; |
---|
375 | b = ip.ip_src; |
---|
376 | b.s_addr &= mask.s_addr; |
---|
377 | if (a.s_addr != b.s_addr) |
---|
378 | return 0; |
---|
379 | |
---|
380 | /* stf interface makes single side match only */ |
---|
381 | return 32; |
---|
382 | } |
---|
383 | |
---|
384 | static int |
---|
385 | stf_getsrcifa6(struct ifnet *ifp, struct in6_addr *addr, struct in6_addr *mask) |
---|
386 | { |
---|
387 | struct ifaddr *ia; |
---|
388 | struct in_ifaddr *ia4; |
---|
389 | struct in6_ifaddr *ia6; |
---|
390 | struct sockaddr_in6 *sin6; |
---|
391 | struct in_addr in; |
---|
392 | |
---|
393 | if_addr_rlock(ifp); |
---|
394 | TAILQ_FOREACH(ia, &ifp->if_addrhead, ifa_link) { |
---|
395 | if (ia->ifa_addr->sa_family != AF_INET6) |
---|
396 | continue; |
---|
397 | sin6 = (struct sockaddr_in6 *)ia->ifa_addr; |
---|
398 | if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) |
---|
399 | continue; |
---|
400 | |
---|
401 | bcopy(GET_V4(&sin6->sin6_addr), &in, sizeof(in)); |
---|
402 | LIST_FOREACH(ia4, INADDR_HASH(in.s_addr), ia_hash) |
---|
403 | if (ia4->ia_addr.sin_addr.s_addr == in.s_addr) |
---|
404 | break; |
---|
405 | if (ia4 == NULL) |
---|
406 | continue; |
---|
407 | |
---|
408 | ia6 = (struct in6_ifaddr *)ia; |
---|
409 | |
---|
410 | *addr = sin6->sin6_addr; |
---|
411 | *mask = ia6->ia_prefixmask.sin6_addr; |
---|
412 | if_addr_runlock(ifp); |
---|
413 | return (0); |
---|
414 | } |
---|
415 | if_addr_runlock(ifp); |
---|
416 | |
---|
417 | return (ENOENT); |
---|
418 | } |
---|
419 | |
---|
420 | static int |
---|
421 | stf_output(struct ifnet *ifp, struct mbuf *m, const struct sockaddr *dst, |
---|
422 | struct route *ro) |
---|
423 | { |
---|
424 | struct stf_softc *sc; |
---|
425 | const struct sockaddr_in6 *dst6; |
---|
426 | struct in_addr in4; |
---|
427 | const void *ptr; |
---|
428 | u_int8_t tos; |
---|
429 | struct ip *ip; |
---|
430 | struct ip6_hdr *ip6; |
---|
431 | struct in6_addr addr6, mask6; |
---|
432 | int error; |
---|
433 | |
---|
434 | #ifdef MAC |
---|
435 | error = mac_ifnet_check_transmit(ifp, m); |
---|
436 | if (error) { |
---|
437 | m_freem(m); |
---|
438 | return (error); |
---|
439 | } |
---|
440 | #endif |
---|
441 | |
---|
442 | sc = ifp->if_softc; |
---|
443 | dst6 = (const struct sockaddr_in6 *)dst; |
---|
444 | |
---|
445 | /* just in case */ |
---|
446 | if ((ifp->if_flags & IFF_UP) == 0) { |
---|
447 | m_freem(m); |
---|
448 | if_inc_counter(ifp, IFCOUNTER_OERRORS, 1); |
---|
449 | return ENETDOWN; |
---|
450 | } |
---|
451 | |
---|
452 | /* |
---|
453 | * If we don't have an ip4 address that match my inner ip6 address, |
---|
454 | * we shouldn't generate output. Without this check, we'll end up |
---|
455 | * using wrong IPv4 source. |
---|
456 | */ |
---|
457 | if (stf_getsrcifa6(ifp, &addr6, &mask6) != 0) { |
---|
458 | m_freem(m); |
---|
459 | if_inc_counter(ifp, IFCOUNTER_OERRORS, 1); |
---|
460 | return ENETDOWN; |
---|
461 | } |
---|
462 | |
---|
463 | if (m->m_len < sizeof(*ip6)) { |
---|
464 | m = m_pullup(m, sizeof(*ip6)); |
---|
465 | if (!m) { |
---|
466 | if_inc_counter(ifp, IFCOUNTER_OERRORS, 1); |
---|
467 | return ENOBUFS; |
---|
468 | } |
---|
469 | } |
---|
470 | ip6 = mtod(m, struct ip6_hdr *); |
---|
471 | tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff; |
---|
472 | |
---|
473 | /* |
---|
474 | * Pickup the right outer dst addr from the list of candidates. |
---|
475 | * ip6_dst has priority as it may be able to give us shorter IPv4 hops. |
---|
476 | */ |
---|
477 | ptr = NULL; |
---|
478 | if (IN6_IS_ADDR_6TO4(&ip6->ip6_dst)) |
---|
479 | ptr = GET_V4(&ip6->ip6_dst); |
---|
480 | else if (IN6_IS_ADDR_6TO4(&dst6->sin6_addr)) |
---|
481 | ptr = GET_V4(&dst6->sin6_addr); |
---|
482 | else { |
---|
483 | m_freem(m); |
---|
484 | if_inc_counter(ifp, IFCOUNTER_OERRORS, 1); |
---|
485 | return ENETUNREACH; |
---|
486 | } |
---|
487 | bcopy(ptr, &in4, sizeof(in4)); |
---|
488 | |
---|
489 | if (bpf_peers_present(ifp->if_bpf)) { |
---|
490 | /* |
---|
491 | * We need to prepend the address family as |
---|
492 | * a four byte field. Cons up a dummy header |
---|
493 | * to pacify bpf. This is safe because bpf |
---|
494 | * will only read from the mbuf (i.e., it won't |
---|
495 | * try to free it or keep a pointer a to it). |
---|
496 | */ |
---|
497 | u_int af = AF_INET6; |
---|
498 | bpf_mtap2(ifp->if_bpf, &af, sizeof(af), m); |
---|
499 | } |
---|
500 | |
---|
501 | M_PREPEND(m, sizeof(struct ip), M_NOWAIT); |
---|
502 | if (m == NULL) { |
---|
503 | if_inc_counter(ifp, IFCOUNTER_OERRORS, 1); |
---|
504 | return ENOBUFS; |
---|
505 | } |
---|
506 | ip = mtod(m, struct ip *); |
---|
507 | |
---|
508 | bzero(ip, sizeof(*ip)); |
---|
509 | |
---|
510 | bcopy(GET_V4(&addr6), &ip->ip_src, sizeof(ip->ip_src)); |
---|
511 | bcopy(&in4, &ip->ip_dst, sizeof(ip->ip_dst)); |
---|
512 | ip->ip_p = IPPROTO_IPV6; |
---|
513 | ip->ip_ttl = ip_stf_ttl; |
---|
514 | ip->ip_len = htons(m->m_pkthdr.len); |
---|
515 | if (ifp->if_flags & IFF_LINK1) |
---|
516 | ip_ecn_ingress(ECN_ALLOWED, &ip->ip_tos, &tos); |
---|
517 | else |
---|
518 | ip_ecn_ingress(ECN_NOCARE, &ip->ip_tos, &tos); |
---|
519 | |
---|
520 | M_SETFIB(m, sc->sc_fibnum); |
---|
521 | if_inc_counter(ifp, IFCOUNTER_OPACKETS, 1); |
---|
522 | error = ip_output(m, NULL, NULL, 0, NULL, NULL); |
---|
523 | |
---|
524 | return error; |
---|
525 | } |
---|
526 | |
---|
527 | static int |
---|
528 | isrfc1918addr(struct in_addr *in) |
---|
529 | { |
---|
530 | /* |
---|
531 | * returns 1 if private address range: |
---|
532 | * 10.0.0.0/8 172.16.0.0/12 192.168.0.0/16 |
---|
533 | */ |
---|
534 | if (stf_permit_rfc1918 == 0 && ( |
---|
535 | (ntohl(in->s_addr) & 0xff000000) >> 24 == 10 || |
---|
536 | (ntohl(in->s_addr) & 0xfff00000) >> 16 == 172 * 256 + 16 || |
---|
537 | (ntohl(in->s_addr) & 0xffff0000) >> 16 == 192 * 256 + 168)) |
---|
538 | return 1; |
---|
539 | |
---|
540 | return 0; |
---|
541 | } |
---|
542 | |
---|
543 | static int |
---|
544 | stf_checkaddr4(struct stf_softc *sc, struct in_addr *in, struct ifnet *inifp) |
---|
545 | { |
---|
546 | struct rm_priotracker in_ifa_tracker; |
---|
547 | struct in_ifaddr *ia4; |
---|
548 | |
---|
549 | /* |
---|
550 | * reject packets with the following address: |
---|
551 | * 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8 |
---|
552 | */ |
---|
553 | if (IN_MULTICAST(ntohl(in->s_addr))) |
---|
554 | return -1; |
---|
555 | switch ((ntohl(in->s_addr) & 0xff000000) >> 24) { |
---|
556 | case 0: case 127: case 255: |
---|
557 | return -1; |
---|
558 | } |
---|
559 | |
---|
560 | /* |
---|
561 | * reject packets with private address range. |
---|
562 | * (requirement from RFC3056 section 2 1st paragraph) |
---|
563 | */ |
---|
564 | if (isrfc1918addr(in)) |
---|
565 | return -1; |
---|
566 | |
---|
567 | /* |
---|
568 | * reject packets with broadcast |
---|
569 | */ |
---|
570 | IN_IFADDR_RLOCK(&in_ifa_tracker); |
---|
571 | TAILQ_FOREACH(ia4, &V_in_ifaddrhead, ia_link) { |
---|
572 | if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0) |
---|
573 | continue; |
---|
574 | if (in->s_addr == ia4->ia_broadaddr.sin_addr.s_addr) { |
---|
575 | IN_IFADDR_RUNLOCK(&in_ifa_tracker); |
---|
576 | return -1; |
---|
577 | } |
---|
578 | } |
---|
579 | IN_IFADDR_RUNLOCK(&in_ifa_tracker); |
---|
580 | |
---|
581 | /* |
---|
582 | * perform ingress filter |
---|
583 | */ |
---|
584 | if (sc && (STF2IFP(sc)->if_flags & IFF_LINK2) == 0 && inifp) { |
---|
585 | struct nhop4_basic nh4; |
---|
586 | |
---|
587 | if (fib4_lookup_nh_basic(sc->sc_fibnum, *in, 0, 0, &nh4) != 0) |
---|
588 | return (-1); |
---|
589 | |
---|
590 | if (nh4.nh_ifp != inifp) |
---|
591 | return (-1); |
---|
592 | } |
---|
593 | |
---|
594 | return 0; |
---|
595 | } |
---|
596 | |
---|
597 | static int |
---|
598 | stf_checkaddr6(struct stf_softc *sc, struct in6_addr *in6, struct ifnet *inifp) |
---|
599 | { |
---|
600 | /* |
---|
601 | * check 6to4 addresses |
---|
602 | */ |
---|
603 | if (IN6_IS_ADDR_6TO4(in6)) { |
---|
604 | struct in_addr in4; |
---|
605 | bcopy(GET_V4(in6), &in4, sizeof(in4)); |
---|
606 | return stf_checkaddr4(sc, &in4, inifp); |
---|
607 | } |
---|
608 | |
---|
609 | /* |
---|
610 | * reject anything that look suspicious. the test is implemented |
---|
611 | * in ip6_input too, but we check here as well to |
---|
612 | * (1) reject bad packets earlier, and |
---|
613 | * (2) to be safe against future ip6_input change. |
---|
614 | */ |
---|
615 | if (IN6_IS_ADDR_V4COMPAT(in6) || IN6_IS_ADDR_V4MAPPED(in6)) |
---|
616 | return -1; |
---|
617 | |
---|
618 | return 0; |
---|
619 | } |
---|
620 | |
---|
621 | static int |
---|
622 | in_stf_input(struct mbuf **mp, int *offp, int proto) |
---|
623 | { |
---|
624 | struct stf_softc *sc; |
---|
625 | struct ip *ip; |
---|
626 | struct ip6_hdr *ip6; |
---|
627 | struct mbuf *m; |
---|
628 | u_int8_t otos, itos; |
---|
629 | struct ifnet *ifp; |
---|
630 | int off; |
---|
631 | |
---|
632 | m = *mp; |
---|
633 | off = *offp; |
---|
634 | |
---|
635 | if (proto != IPPROTO_IPV6) { |
---|
636 | m_freem(m); |
---|
637 | return (IPPROTO_DONE); |
---|
638 | } |
---|
639 | |
---|
640 | ip = mtod(m, struct ip *); |
---|
641 | |
---|
642 | sc = (struct stf_softc *)encap_getarg(m); |
---|
643 | |
---|
644 | if (sc == NULL || (STF2IFP(sc)->if_flags & IFF_UP) == 0) { |
---|
645 | m_freem(m); |
---|
646 | return (IPPROTO_DONE); |
---|
647 | } |
---|
648 | |
---|
649 | ifp = STF2IFP(sc); |
---|
650 | |
---|
651 | #ifdef MAC |
---|
652 | mac_ifnet_create_mbuf(ifp, m); |
---|
653 | #endif |
---|
654 | |
---|
655 | /* |
---|
656 | * perform sanity check against outer src/dst. |
---|
657 | * for source, perform ingress filter as well. |
---|
658 | */ |
---|
659 | if (stf_checkaddr4(sc, &ip->ip_dst, NULL) < 0 || |
---|
660 | stf_checkaddr4(sc, &ip->ip_src, m->m_pkthdr.rcvif) < 0) { |
---|
661 | m_freem(m); |
---|
662 | return (IPPROTO_DONE); |
---|
663 | } |
---|
664 | |
---|
665 | otos = ip->ip_tos; |
---|
666 | m_adj(m, off); |
---|
667 | |
---|
668 | if (m->m_len < sizeof(*ip6)) { |
---|
669 | m = m_pullup(m, sizeof(*ip6)); |
---|
670 | if (!m) |
---|
671 | return (IPPROTO_DONE); |
---|
672 | } |
---|
673 | ip6 = mtod(m, struct ip6_hdr *); |
---|
674 | |
---|
675 | /* |
---|
676 | * perform sanity check against inner src/dst. |
---|
677 | * for source, perform ingress filter as well. |
---|
678 | */ |
---|
679 | if (stf_checkaddr6(sc, &ip6->ip6_dst, NULL) < 0 || |
---|
680 | stf_checkaddr6(sc, &ip6->ip6_src, m->m_pkthdr.rcvif) < 0) { |
---|
681 | m_freem(m); |
---|
682 | return (IPPROTO_DONE); |
---|
683 | } |
---|
684 | |
---|
685 | itos = (ntohl(ip6->ip6_flow) >> 20) & 0xff; |
---|
686 | if ((ifp->if_flags & IFF_LINK1) != 0) |
---|
687 | ip_ecn_egress(ECN_ALLOWED, &otos, &itos); |
---|
688 | else |
---|
689 | ip_ecn_egress(ECN_NOCARE, &otos, &itos); |
---|
690 | ip6->ip6_flow &= ~htonl(0xff << 20); |
---|
691 | ip6->ip6_flow |= htonl((u_int32_t)itos << 20); |
---|
692 | |
---|
693 | m->m_pkthdr.rcvif = ifp; |
---|
694 | |
---|
695 | if (bpf_peers_present(ifp->if_bpf)) { |
---|
696 | /* |
---|
697 | * We need to prepend the address family as |
---|
698 | * a four byte field. Cons up a dummy header |
---|
699 | * to pacify bpf. This is safe because bpf |
---|
700 | * will only read from the mbuf (i.e., it won't |
---|
701 | * try to free it or keep a pointer a to it). |
---|
702 | */ |
---|
703 | u_int32_t af = AF_INET6; |
---|
704 | bpf_mtap2(ifp->if_bpf, &af, sizeof(af), m); |
---|
705 | } |
---|
706 | |
---|
707 | /* |
---|
708 | * Put the packet to the network layer input queue according to the |
---|
709 | * specified address family. |
---|
710 | * See net/if_gif.c for possible issues with packet processing |
---|
711 | * reorder due to extra queueing. |
---|
712 | */ |
---|
713 | if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1); |
---|
714 | if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); |
---|
715 | M_SETFIB(m, ifp->if_fib); |
---|
716 | netisr_dispatch(NETISR_IPV6, m); |
---|
717 | return (IPPROTO_DONE); |
---|
718 | } |
---|
719 | |
---|
720 | static int |
---|
721 | stf_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data) |
---|
722 | { |
---|
723 | struct ifaddr *ifa; |
---|
724 | struct ifreq *ifr; |
---|
725 | struct sockaddr_in6 *sin6; |
---|
726 | struct in_addr addr; |
---|
727 | int error, mtu; |
---|
728 | |
---|
729 | error = 0; |
---|
730 | switch (cmd) { |
---|
731 | case SIOCSIFADDR: |
---|
732 | ifa = (struct ifaddr *)data; |
---|
733 | if (ifa == NULL || ifa->ifa_addr->sa_family != AF_INET6) { |
---|
734 | error = EAFNOSUPPORT; |
---|
735 | break; |
---|
736 | } |
---|
737 | sin6 = (struct sockaddr_in6 *)ifa->ifa_addr; |
---|
738 | if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) { |
---|
739 | error = EINVAL; |
---|
740 | break; |
---|
741 | } |
---|
742 | bcopy(GET_V4(&sin6->sin6_addr), &addr, sizeof(addr)); |
---|
743 | if (isrfc1918addr(&addr)) { |
---|
744 | error = EINVAL; |
---|
745 | break; |
---|
746 | } |
---|
747 | |
---|
748 | ifp->if_flags |= IFF_UP; |
---|
749 | break; |
---|
750 | |
---|
751 | case SIOCADDMULTI: |
---|
752 | case SIOCDELMULTI: |
---|
753 | ifr = (struct ifreq *)data; |
---|
754 | if (ifr && ifr->ifr_addr.sa_family == AF_INET6) |
---|
755 | ; |
---|
756 | else |
---|
757 | error = EAFNOSUPPORT; |
---|
758 | break; |
---|
759 | |
---|
760 | case SIOCGIFMTU: |
---|
761 | break; |
---|
762 | |
---|
763 | case SIOCSIFMTU: |
---|
764 | ifr = (struct ifreq *)data; |
---|
765 | mtu = ifr->ifr_mtu; |
---|
766 | /* RFC 4213 3.2 ideal world MTU */ |
---|
767 | if (mtu < IPV6_MINMTU || mtu > IF_MAXMTU - 20) |
---|
768 | return (EINVAL); |
---|
769 | ifp->if_mtu = mtu; |
---|
770 | break; |
---|
771 | |
---|
772 | default: |
---|
773 | error = EINVAL; |
---|
774 | break; |
---|
775 | } |
---|
776 | |
---|
777 | return error; |
---|
778 | } |
---|