1 /*-
2 * Copyright (c) 2016 Yandex LLC
3 * Copyright (c) 2016 Andrey V. Elsukov <ae@FreeBSD.org>
4 * All rights reserved.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 *
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 *
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
19 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
21 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
22 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
23 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
24 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
25 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26 */
27
28 #include <sys/param.h>
29 #include <sys/systm.h>
30 #include <sys/counter.h>
31 #include <sys/eventhandler.h>
32 #include <sys/errno.h>
33 #include <sys/kernel.h>
34 #include <sys/lock.h>
35 #include <sys/malloc.h>
36 #include <sys/mbuf.h>
37 #include <sys/module.h>
38 #include <sys/rmlock.h>
39 #include <sys/rwlock.h>
40 #include <sys/socket.h>
41 #include <sys/queue.h>
42 #include <sys/syslog.h>
43 #include <sys/sysctl.h>
44
45 #include <net/if.h>
46 #include <net/if_var.h>
47 #include <net/if_private.h>
48 #include <net/netisr.h>
49 #include <net/pfil.h>
50 #include <net/vnet.h>
51
52 #include <netinet/in.h>
53 #include <netinet/ip_var.h>
54 #include <netinet/ip_fw.h>
55 #include <netinet/ip6.h>
56 #include <netinet/icmp6.h>
57 #include <netinet6/in6_var.h>
58 #include <netinet6/ip6_var.h>
59
60 #include <netpfil/ipfw/ip_fw_private.h>
61 #include <netpfil/ipfw/nptv6/nptv6.h>
62
63 VNET_DEFINE_STATIC(uint16_t, nptv6_eid) = 0;
64 #define V_nptv6_eid VNET(nptv6_eid)
65 #define IPFW_TLV_NPTV6_NAME IPFW_TLV_EACTION_NAME(V_nptv6_eid)
66
67 static eventhandler_tag nptv6_ifaddr_event;
68
69 static struct nptv6_cfg *nptv6_alloc_config(const char *name, uint8_t set);
70 static void nptv6_free_config(struct nptv6_cfg *cfg);
71 static struct nptv6_cfg *nptv6_find(struct namedobj_instance *ni,
72 const char *name, uint8_t set);
73 static int nptv6_rewrite_internal(struct nptv6_cfg *cfg, struct mbuf **mp,
74 int offset);
75 static int nptv6_rewrite_external(struct nptv6_cfg *cfg, struct mbuf **mp,
76 int offset);
77
78 #define NPTV6_LOOKUP(chain, cmd) \
79 (struct nptv6_cfg *)SRV_OBJECT((chain), (cmd)->arg1)
80
81 #ifndef IN6_MASK_ADDR
82 #define IN6_MASK_ADDR(a, m) do { \
83 (a)->s6_addr32[0] &= (m)->s6_addr32[0]; \
84 (a)->s6_addr32[1] &= (m)->s6_addr32[1]; \
85 (a)->s6_addr32[2] &= (m)->s6_addr32[2]; \
86 (a)->s6_addr32[3] &= (m)->s6_addr32[3]; \
87 } while (0)
88 #endif
89 #ifndef IN6_ARE_MASKED_ADDR_EQUAL
90 #define IN6_ARE_MASKED_ADDR_EQUAL(d, a, m) ( \
91 (((d)->s6_addr32[0] ^ (a)->s6_addr32[0]) & (m)->s6_addr32[0]) == 0 && \
92 (((d)->s6_addr32[1] ^ (a)->s6_addr32[1]) & (m)->s6_addr32[1]) == 0 && \
93 (((d)->s6_addr32[2] ^ (a)->s6_addr32[2]) & (m)->s6_addr32[2]) == 0 && \
94 (((d)->s6_addr32[3] ^ (a)->s6_addr32[3]) & (m)->s6_addr32[3]) == 0 )
95 #endif
96
97 #if 0
98 #define NPTV6_DEBUG(fmt, ...) do { \
99 printf("%s: " fmt "\n", __func__, ## __VA_ARGS__); \
100 } while (0)
101 #define NPTV6_IPDEBUG(fmt, ...) do { \
102 char _s[INET6_ADDRSTRLEN], _d[INET6_ADDRSTRLEN]; \
103 printf("%s: " fmt "\n", __func__, ## __VA_ARGS__); \
104 } while (0)
105 #else
106 #define NPTV6_DEBUG(fmt, ...)
107 #define NPTV6_IPDEBUG(fmt, ...)
108 #endif
109
110 static int
nptv6_getlasthdr(struct nptv6_cfg * cfg,struct mbuf * m,int * offset)111 nptv6_getlasthdr(struct nptv6_cfg *cfg, struct mbuf *m, int *offset)
112 {
113 struct ip6_hdr *ip6;
114 struct ip6_hbh *hbh;
115 int proto, hlen;
116
117 hlen = (offset == NULL) ? 0: *offset;
118 if (m->m_len < hlen)
119 return (-1);
120 ip6 = mtodo(m, hlen);
121 hlen += sizeof(*ip6);
122 proto = ip6->ip6_nxt;
123 while (proto == IPPROTO_HOPOPTS || proto == IPPROTO_ROUTING ||
124 proto == IPPROTO_DSTOPTS) {
125 hbh = mtodo(m, hlen);
126 if (m->m_len < hlen)
127 return (-1);
128 proto = hbh->ip6h_nxt;
129 hlen += (hbh->ip6h_len + 1) << 3;
130 }
131 if (offset != NULL)
132 *offset = hlen;
133 return (proto);
134 }
135
136 static int
nptv6_translate_icmpv6(struct nptv6_cfg * cfg,struct mbuf ** mp,int offset)137 nptv6_translate_icmpv6(struct nptv6_cfg *cfg, struct mbuf **mp, int offset)
138 {
139 struct icmp6_hdr *icmp6;
140 struct ip6_hdr *ip6;
141 struct mbuf *m;
142
143 m = *mp;
144 if (offset > m->m_len)
145 return (-1);
146 icmp6 = mtodo(m, offset);
147 NPTV6_DEBUG("ICMPv6 type %d", icmp6->icmp6_type);
148 switch (icmp6->icmp6_type) {
149 case ICMP6_DST_UNREACH:
150 case ICMP6_PACKET_TOO_BIG:
151 case ICMP6_TIME_EXCEEDED:
152 case ICMP6_PARAM_PROB:
153 break;
154 case ICMP6_ECHO_REQUEST:
155 case ICMP6_ECHO_REPLY:
156 /* nothing to translate */
157 return (0);
158 default:
159 /*
160 * XXX: We can add some checks to not translate NDP and MLD
161 * messages. Currently user must explicitly allow these message
162 * types, otherwise packets will be dropped.
163 */
164 return (-1);
165 }
166 offset += sizeof(*icmp6);
167 if (offset + sizeof(*ip6) > m->m_pkthdr.len)
168 return (-1);
169 if (offset + sizeof(*ip6) > m->m_len)
170 *mp = m = m_pullup(m, offset + sizeof(*ip6));
171 if (m == NULL)
172 return (-1);
173 ip6 = mtodo(m, offset);
174 NPTV6_IPDEBUG("offset %d, %s -> %s %d", offset,
175 inet_ntop(AF_INET6, &ip6->ip6_src, _s, sizeof(_s)),
176 inet_ntop(AF_INET6, &ip6->ip6_dst, _d, sizeof(_d)),
177 ip6->ip6_nxt);
178 if (IN6_ARE_MASKED_ADDR_EQUAL(&ip6->ip6_src,
179 &cfg->external, &cfg->mask))
180 return (nptv6_rewrite_external(cfg, mp, offset));
181 else if (IN6_ARE_MASKED_ADDR_EQUAL(&ip6->ip6_dst,
182 &cfg->internal, &cfg->mask))
183 return (nptv6_rewrite_internal(cfg, mp, offset));
184 /*
185 * Addresses in the inner IPv6 header doesn't matched to
186 * our prefixes.
187 */
188 return (-1);
189 }
190
191 static int
nptv6_search_index(struct nptv6_cfg * cfg,struct in6_addr * a)192 nptv6_search_index(struct nptv6_cfg *cfg, struct in6_addr *a)
193 {
194 int idx;
195
196 if (cfg->flags & NPTV6_48PLEN)
197 return (3);
198
199 /* Search suitable word index for adjustment */
200 for (idx = 4; idx < 8; idx++)
201 if (a->s6_addr16[idx] != 0xffff)
202 break;
203 /*
204 * RFC 6296 p3.7: If an NPTv6 Translator discovers a datagram with
205 * an IID of all-zeros while performing address mapping, that
206 * datagram MUST be dropped, and an ICMPv6 Parameter Problem error
207 * SHOULD be generated.
208 */
209 if (idx == 8 ||
210 (a->s6_addr32[2] == 0 && a->s6_addr32[3] == 0))
211 return (-1);
212 return (idx);
213 }
214
215 static void
nptv6_copy_addr(struct in6_addr * src,struct in6_addr * dst,struct in6_addr * mask)216 nptv6_copy_addr(struct in6_addr *src, struct in6_addr *dst,
217 struct in6_addr *mask)
218 {
219 int i;
220
221 for (i = 0; i < 8 && mask->s6_addr8[i] != 0; i++) {
222 dst->s6_addr8[i] &= ~mask->s6_addr8[i];
223 dst->s6_addr8[i] |= src->s6_addr8[i] & mask->s6_addr8[i];
224 }
225 }
226
227 static int
nptv6_rewrite_internal(struct nptv6_cfg * cfg,struct mbuf ** mp,int offset)228 nptv6_rewrite_internal(struct nptv6_cfg *cfg, struct mbuf **mp, int offset)
229 {
230 struct in6_addr *addr;
231 struct ip6_hdr *ip6;
232 int idx, proto;
233 uint16_t adj;
234
235 ip6 = mtodo(*mp, offset);
236 NPTV6_IPDEBUG("offset %d, %s -> %s %d", offset,
237 inet_ntop(AF_INET6, &ip6->ip6_src, _s, sizeof(_s)),
238 inet_ntop(AF_INET6, &ip6->ip6_dst, _d, sizeof(_d)),
239 ip6->ip6_nxt);
240 if (offset == 0)
241 addr = &ip6->ip6_src;
242 else {
243 /*
244 * When we rewriting inner IPv6 header, we need to rewrite
245 * destination address back to external prefix. The datagram in
246 * the ICMPv6 payload should looks like it was send from
247 * external prefix.
248 */
249 addr = &ip6->ip6_dst;
250 }
251 idx = nptv6_search_index(cfg, addr);
252 if (idx < 0) {
253 /*
254 * Do not send ICMPv6 error when offset isn't zero.
255 * This means we are rewriting inner IPv6 header in the
256 * ICMPv6 error message.
257 */
258 if (offset == 0) {
259 icmp6_error2(*mp, ICMP6_DST_UNREACH,
260 ICMP6_DST_UNREACH_ADDR, 0, (*mp)->m_pkthdr.rcvif);
261 *mp = NULL;
262 }
263 return (IP_FW_DENY);
264 }
265 adj = addr->s6_addr16[idx];
266 nptv6_copy_addr(&cfg->external, addr, &cfg->mask);
267 adj = cksum_add(adj, cfg->adjustment);
268 if (adj == 0xffff)
269 adj = 0;
270 addr->s6_addr16[idx] = adj;
271 if (offset == 0) {
272 /*
273 * We may need to translate addresses in the inner IPv6
274 * header for ICMPv6 error messages.
275 */
276 proto = nptv6_getlasthdr(cfg, *mp, &offset);
277 if (proto < 0 || (proto == IPPROTO_ICMPV6 &&
278 nptv6_translate_icmpv6(cfg, mp, offset) != 0))
279 return (IP_FW_DENY);
280 NPTV6STAT_INC(cfg, in2ex);
281 }
282 return (0);
283 }
284
285 static int
nptv6_rewrite_external(struct nptv6_cfg * cfg,struct mbuf ** mp,int offset)286 nptv6_rewrite_external(struct nptv6_cfg *cfg, struct mbuf **mp, int offset)
287 {
288 struct in6_addr *addr;
289 struct ip6_hdr *ip6;
290 int idx, proto;
291 uint16_t adj;
292
293 ip6 = mtodo(*mp, offset);
294 NPTV6_IPDEBUG("offset %d, %s -> %s %d", offset,
295 inet_ntop(AF_INET6, &ip6->ip6_src, _s, sizeof(_s)),
296 inet_ntop(AF_INET6, &ip6->ip6_dst, _d, sizeof(_d)),
297 ip6->ip6_nxt);
298 if (offset == 0)
299 addr = &ip6->ip6_dst;
300 else {
301 /*
302 * When we rewriting inner IPv6 header, we need to rewrite
303 * source address back to internal prefix. The datagram in
304 * the ICMPv6 payload should looks like it was send from
305 * internal prefix.
306 */
307 addr = &ip6->ip6_src;
308 }
309 idx = nptv6_search_index(cfg, addr);
310 if (idx < 0) {
311 /*
312 * Do not send ICMPv6 error when offset isn't zero.
313 * This means we are rewriting inner IPv6 header in the
314 * ICMPv6 error message.
315 */
316 if (offset == 0) {
317 icmp6_error2(*mp, ICMP6_DST_UNREACH,
318 ICMP6_DST_UNREACH_ADDR, 0, (*mp)->m_pkthdr.rcvif);
319 *mp = NULL;
320 }
321 return (IP_FW_DENY);
322 }
323 adj = addr->s6_addr16[idx];
324 nptv6_copy_addr(&cfg->internal, addr, &cfg->mask);
325 adj = cksum_add(adj, ~cfg->adjustment);
326 if (adj == 0xffff)
327 adj = 0;
328 addr->s6_addr16[idx] = adj;
329 if (offset == 0) {
330 /*
331 * We may need to translate addresses in the inner IPv6
332 * header for ICMPv6 error messages.
333 */
334 proto = nptv6_getlasthdr(cfg, *mp, &offset);
335 if (proto < 0 || (proto == IPPROTO_ICMPV6 &&
336 nptv6_translate_icmpv6(cfg, mp, offset) != 0))
337 return (IP_FW_DENY);
338 NPTV6STAT_INC(cfg, ex2in);
339 }
340 return (0);
341 }
342
343 /*
344 * ipfw external action handler.
345 */
346 static int
ipfw_nptv6(struct ip_fw_chain * chain,struct ip_fw_args * args,ipfw_insn * cmd,int * done)347 ipfw_nptv6(struct ip_fw_chain *chain, struct ip_fw_args *args,
348 ipfw_insn *cmd, int *done)
349 {
350 struct ip6_hdr *ip6;
351 struct nptv6_cfg *cfg;
352 ipfw_insn *icmd;
353 int ret;
354
355 *done = 0; /* try next rule if not matched */
356 ret = IP_FW_DENY;
357 icmd = cmd + 1;
358 if (cmd->opcode != O_EXTERNAL_ACTION ||
359 cmd->arg1 != V_nptv6_eid ||
360 icmd->opcode != O_EXTERNAL_INSTANCE ||
361 (cfg = NPTV6_LOOKUP(chain, icmd)) == NULL ||
362 (cfg->flags & NPTV6_READY) == 0)
363 return (ret);
364 /*
365 * We need act as router, so when forwarding is disabled -
366 * do nothing.
367 */
368 if (V_ip6_forwarding == 0 || args->f_id.addr_type != 6)
369 return (ret);
370 /*
371 * NOTE: we expect ipfw_chk() did m_pullup() up to upper level
372 * protocol's headers. Also we skip some checks, that ip6_input(),
373 * ip6_forward(), ip6_fastfwd() and ipfw_chk() already did.
374 */
375 ip6 = mtod(args->m, struct ip6_hdr *);
376 NPTV6_IPDEBUG("eid %u, oid %u, %s -> %s %d",
377 cmd->arg1, icmd->arg1,
378 inet_ntop(AF_INET6, &ip6->ip6_src, _s, sizeof(_s)),
379 inet_ntop(AF_INET6, &ip6->ip6_dst, _d, sizeof(_d)),
380 ip6->ip6_nxt);
381 if (IN6_ARE_MASKED_ADDR_EQUAL(&ip6->ip6_src,
382 &cfg->internal, &cfg->mask)) {
383 /*
384 * XXX: Do not translate packets when both src and dst
385 * are from internal prefix.
386 */
387 if (IN6_ARE_MASKED_ADDR_EQUAL(&ip6->ip6_dst,
388 &cfg->internal, &cfg->mask))
389 return (ret);
390 ret = nptv6_rewrite_internal(cfg, &args->m, 0);
391 } else if (IN6_ARE_MASKED_ADDR_EQUAL(&ip6->ip6_dst,
392 &cfg->external, &cfg->mask))
393 ret = nptv6_rewrite_external(cfg, &args->m, 0);
394 else
395 return (ret);
396 /*
397 * If address wasn't rewrited - free mbuf and terminate the search.
398 */
399 if (ret != 0) {
400 if (args->m != NULL) {
401 m_freem(args->m);
402 args->m = NULL; /* mark mbuf as consumed */
403 }
404 NPTV6STAT_INC(cfg, dropped);
405 *done = 1;
406 } else {
407 /* Terminate the search if one_pass is set */
408 *done = V_fw_one_pass;
409 /* Update args->f_id when one_pass is off */
410 if (*done == 0) {
411 ip6 = mtod(args->m, struct ip6_hdr *);
412 args->f_id.src_ip6 = ip6->ip6_src;
413 args->f_id.dst_ip6 = ip6->ip6_dst;
414 }
415 }
416 return (ret);
417 }
418
419 static struct nptv6_cfg *
nptv6_alloc_config(const char * name,uint8_t set)420 nptv6_alloc_config(const char *name, uint8_t set)
421 {
422 struct nptv6_cfg *cfg;
423
424 cfg = malloc(sizeof(struct nptv6_cfg), M_IPFW, M_WAITOK | M_ZERO);
425 COUNTER_ARRAY_ALLOC(cfg->stats, NPTV6STATS, M_WAITOK);
426 cfg->no.name = cfg->name;
427 cfg->no.etlv = IPFW_TLV_NPTV6_NAME;
428 cfg->no.set = set;
429 strlcpy(cfg->name, name, sizeof(cfg->name));
430 return (cfg);
431 }
432
433 static void
nptv6_free_config(struct nptv6_cfg * cfg)434 nptv6_free_config(struct nptv6_cfg *cfg)
435 {
436
437 COUNTER_ARRAY_FREE(cfg->stats, NPTV6STATS);
438 free(cfg, M_IPFW);
439 }
440
441 static void
nptv6_export_config(struct ip_fw_chain * ch,struct nptv6_cfg * cfg,ipfw_nptv6_cfg * uc)442 nptv6_export_config(struct ip_fw_chain *ch, struct nptv6_cfg *cfg,
443 ipfw_nptv6_cfg *uc)
444 {
445
446 uc->internal = cfg->internal;
447 if (cfg->flags & NPTV6_DYNAMIC_PREFIX)
448 memcpy(uc->if_name, cfg->if_name, IF_NAMESIZE);
449 else
450 uc->external = cfg->external;
451 uc->plen = cfg->plen;
452 uc->flags = cfg->flags & NPTV6_FLAGSMASK;
453 uc->set = cfg->no.set;
454 strlcpy(uc->name, cfg->no.name, sizeof(uc->name));
455 }
456
457 struct nptv6_dump_arg {
458 struct ip_fw_chain *ch;
459 struct sockopt_data *sd;
460 };
461
462 static int
export_config_cb(struct namedobj_instance * ni,struct named_object * no,void * arg)463 export_config_cb(struct namedobj_instance *ni, struct named_object *no,
464 void *arg)
465 {
466 struct nptv6_dump_arg *da = (struct nptv6_dump_arg *)arg;
467 ipfw_nptv6_cfg *uc;
468
469 uc = (ipfw_nptv6_cfg *)ipfw_get_sopt_space(da->sd, sizeof(*uc));
470 nptv6_export_config(da->ch, (struct nptv6_cfg *)no, uc);
471 return (0);
472 }
473
474 static struct nptv6_cfg *
nptv6_find(struct namedobj_instance * ni,const char * name,uint8_t set)475 nptv6_find(struct namedobj_instance *ni, const char *name, uint8_t set)
476 {
477 struct nptv6_cfg *cfg;
478
479 cfg = (struct nptv6_cfg *)ipfw_objhash_lookup_name_type(ni, set,
480 IPFW_TLV_NPTV6_NAME, name);
481
482 return (cfg);
483 }
484
485 static void
nptv6_calculate_adjustment(struct nptv6_cfg * cfg)486 nptv6_calculate_adjustment(struct nptv6_cfg *cfg)
487 {
488 uint16_t i, e;
489 uint16_t *p;
490
491 /* Calculate checksum of internal prefix */
492 for (i = 0, p = (uint16_t *)&cfg->internal;
493 p < (uint16_t *)(&cfg->internal + 1); p++)
494 i = cksum_add(i, *p);
495
496 /* Calculate checksum of external prefix */
497 for (e = 0, p = (uint16_t *)&cfg->external;
498 p < (uint16_t *)(&cfg->external + 1); p++)
499 e = cksum_add(e, *p);
500
501 /* Adjustment value for Int->Ext direction */
502 cfg->adjustment = cksum_add(~e, i);
503 }
504
505 static int
nptv6_check_prefix(const struct in6_addr * addr)506 nptv6_check_prefix(const struct in6_addr *addr)
507 {
508
509 if (IN6_IS_ADDR_MULTICAST(addr) ||
510 IN6_IS_ADDR_LINKLOCAL(addr) ||
511 IN6_IS_ADDR_LOOPBACK(addr) ||
512 IN6_IS_ADDR_UNSPECIFIED(addr))
513 return (EINVAL);
514 return (0);
515 }
516
517 static void
nptv6_set_external(struct nptv6_cfg * cfg,struct in6_addr * addr)518 nptv6_set_external(struct nptv6_cfg *cfg, struct in6_addr *addr)
519 {
520
521 cfg->external = *addr;
522 IN6_MASK_ADDR(&cfg->external, &cfg->mask);
523 nptv6_calculate_adjustment(cfg);
524 cfg->flags |= NPTV6_READY;
525 }
526
527 /*
528 * Try to determine what prefix to use as external for
529 * configured interface name.
530 */
531 static void
nptv6_find_prefix(struct ip_fw_chain * ch,struct nptv6_cfg * cfg,struct ifnet * ifp)532 nptv6_find_prefix(struct ip_fw_chain *ch, struct nptv6_cfg *cfg,
533 struct ifnet *ifp)
534 {
535 struct epoch_tracker et;
536 struct ifaddr *ifa;
537 struct in6_ifaddr *ia;
538
539 MPASS(cfg->flags & NPTV6_DYNAMIC_PREFIX);
540 IPFW_UH_WLOCK_ASSERT(ch);
541
542 if (ifp == NULL) {
543 ifp = ifunit_ref(cfg->if_name);
544 if (ifp == NULL)
545 return;
546 }
547 NET_EPOCH_ENTER(et);
548 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
549 if (ifa->ifa_addr->sa_family != AF_INET6)
550 continue;
551 ia = (struct in6_ifaddr *)ifa;
552 if (nptv6_check_prefix(&ia->ia_addr.sin6_addr) ||
553 IN6_ARE_MASKED_ADDR_EQUAL(&ia->ia_addr.sin6_addr,
554 &cfg->internal, &cfg->mask))
555 continue;
556 /* Suitable address is found. */
557 nptv6_set_external(cfg, &ia->ia_addr.sin6_addr);
558 break;
559 }
560 NET_EPOCH_EXIT(et);
561 if_rele(ifp);
562 }
563
564 struct ifaddr_event_args {
565 struct ifnet *ifp;
566 const struct in6_addr *addr;
567 int event;
568 };
569
570 static int
ifaddr_cb(struct namedobj_instance * ni,struct named_object * no,void * arg)571 ifaddr_cb(struct namedobj_instance *ni, struct named_object *no,
572 void *arg)
573 {
574 struct ifaddr_event_args *args;
575 struct ip_fw_chain *ch;
576 struct nptv6_cfg *cfg;
577
578 ch = &V_layer3_chain;
579 cfg = (struct nptv6_cfg *)SRV_OBJECT(ch, no->kidx);
580 if ((cfg->flags & NPTV6_DYNAMIC_PREFIX) == 0)
581 return (0);
582
583 args = arg;
584 /* If interface name doesn't match, ignore */
585 if (strncmp(args->ifp->if_xname, cfg->if_name, IF_NAMESIZE))
586 return (0);
587 if (args->ifp->if_flags & IFF_DYING) { /* XXX: is it possible? */
588 cfg->flags &= ~NPTV6_READY;
589 return (0);
590 }
591 if (args->event == IFADDR_EVENT_DEL) {
592 /* If instance is not ready, ignore */
593 if ((cfg->flags & NPTV6_READY) == 0)
594 return (0);
595 /* If address does not match the external prefix, ignore */
596 if (IN6_ARE_MASKED_ADDR_EQUAL(&cfg->external, args->addr,
597 &cfg->mask) != 0)
598 return (0);
599 /* Otherwise clear READY flag */
600 cfg->flags &= ~NPTV6_READY;
601 } else {/* IFADDR_EVENT_ADD */
602 /* If instance is already ready, ignore */
603 if (cfg->flags & NPTV6_READY)
604 return (0);
605 /* If address is not suitable for prefix, ignore */
606 if (nptv6_check_prefix(args->addr) ||
607 IN6_ARE_MASKED_ADDR_EQUAL(args->addr, &cfg->internal,
608 &cfg->mask))
609 return (0);
610 /* FALLTHROUGH */
611 }
612 MPASS(!(cfg->flags & NPTV6_READY));
613 /* Try to determine the prefix */
614 if_ref(args->ifp);
615 nptv6_find_prefix(ch, cfg, args->ifp);
616 return (0);
617 }
618
619 static void
nptv6_ifaddrevent_handler(void * arg __unused,struct ifnet * ifp,struct ifaddr * ifa,int event)620 nptv6_ifaddrevent_handler(void *arg __unused, struct ifnet *ifp,
621 struct ifaddr *ifa, int event)
622 {
623 struct ifaddr_event_args args;
624 struct ip_fw_chain *ch;
625
626 if (ifa->ifa_addr->sa_family != AF_INET6)
627 return;
628
629 args.ifp = ifp;
630 args.addr = &((struct sockaddr_in6 *)ifa->ifa_addr)->sin6_addr;
631 args.event = event;
632
633 ch = &V_layer3_chain;
634 IPFW_UH_WLOCK(ch);
635 ipfw_objhash_foreach_type(CHAIN_TO_SRV(ch), ifaddr_cb, &args,
636 IPFW_TLV_NPTV6_NAME);
637 IPFW_UH_WUNLOCK(ch);
638 }
639
640 /*
641 * Creates new NPTv6 instance.
642 * Data layout (v0)(current):
643 * Request: [ ipfw_obj_lheader ipfw_nptv6_cfg ]
644 *
645 * Returns 0 on success
646 */
647 static int
nptv6_create(struct ip_fw_chain * ch,ip_fw3_opheader * op3,struct sockopt_data * sd)648 nptv6_create(struct ip_fw_chain *ch, ip_fw3_opheader *op3,
649 struct sockopt_data *sd)
650 {
651 struct in6_addr mask;
652 ipfw_obj_lheader *olh;
653 ipfw_nptv6_cfg *uc;
654 struct namedobj_instance *ni;
655 struct nptv6_cfg *cfg;
656
657 if (sd->valsize != sizeof(*olh) + sizeof(*uc))
658 return (EINVAL);
659
660 olh = (ipfw_obj_lheader *)sd->kbuf;
661 uc = (ipfw_nptv6_cfg *)(olh + 1);
662 if (ipfw_check_object_name_generic(uc->name) != 0)
663 return (EINVAL);
664 if (uc->plen < 8 || uc->plen > 64 || uc->set >= IPFW_MAX_SETS)
665 return (EINVAL);
666 if (nptv6_check_prefix(&uc->internal))
667 return (EINVAL);
668 in6_prefixlen2mask(&mask, uc->plen);
669 if ((uc->flags & NPTV6_DYNAMIC_PREFIX) == 0 && (
670 nptv6_check_prefix(&uc->external) ||
671 IN6_ARE_MASKED_ADDR_EQUAL(&uc->external, &uc->internal, &mask)))
672 return (EINVAL);
673
674 ni = CHAIN_TO_SRV(ch);
675 IPFW_UH_RLOCK(ch);
676 if (nptv6_find(ni, uc->name, uc->set) != NULL) {
677 IPFW_UH_RUNLOCK(ch);
678 return (EEXIST);
679 }
680 IPFW_UH_RUNLOCK(ch);
681
682 cfg = nptv6_alloc_config(uc->name, uc->set);
683 cfg->plen = uc->plen;
684 cfg->flags = uc->flags & NPTV6_FLAGSMASK;
685 if (cfg->plen <= 48)
686 cfg->flags |= NPTV6_48PLEN;
687 cfg->mask = mask;
688 cfg->internal = uc->internal;
689 IN6_MASK_ADDR(&cfg->internal, &mask);
690 if (cfg->flags & NPTV6_DYNAMIC_PREFIX)
691 memcpy(cfg->if_name, uc->if_name, IF_NAMESIZE);
692 else
693 nptv6_set_external(cfg, &uc->external);
694
695 if ((uc->flags & NPTV6_DYNAMIC_PREFIX) != 0 &&
696 nptv6_ifaddr_event == NULL)
697 nptv6_ifaddr_event = EVENTHANDLER_REGISTER(
698 ifaddr_event_ext, nptv6_ifaddrevent_handler, NULL,
699 EVENTHANDLER_PRI_ANY);
700
701 IPFW_UH_WLOCK(ch);
702 if (ipfw_objhash_alloc_idx(ni, &cfg->no.kidx) != 0) {
703 IPFW_UH_WUNLOCK(ch);
704 nptv6_free_config(cfg);
705 return (ENOSPC);
706 }
707 ipfw_objhash_add(ni, &cfg->no);
708 SRV_OBJECT(ch, cfg->no.kidx) = cfg;
709 if (cfg->flags & NPTV6_DYNAMIC_PREFIX)
710 nptv6_find_prefix(ch, cfg, NULL);
711 IPFW_UH_WUNLOCK(ch);
712
713 return (0);
714 }
715
716 /*
717 * Destroys NPTv6 instance.
718 * Data layout (v0)(current):
719 * Request: [ ipfw_obj_header ]
720 *
721 * Returns 0 on success
722 */
723 static int
nptv6_destroy(struct ip_fw_chain * ch,ip_fw3_opheader * op3,struct sockopt_data * sd)724 nptv6_destroy(struct ip_fw_chain *ch, ip_fw3_opheader *op3,
725 struct sockopt_data *sd)
726 {
727 ipfw_obj_header *oh;
728 struct nptv6_cfg *cfg;
729
730 if (sd->valsize != sizeof(*oh))
731 return (EINVAL);
732
733 oh = (ipfw_obj_header *)sd->kbuf;
734 if (ipfw_check_object_name_generic(oh->ntlv.name) != 0)
735 return (EINVAL);
736
737 IPFW_UH_WLOCK(ch);
738 cfg = nptv6_find(CHAIN_TO_SRV(ch), oh->ntlv.name, oh->ntlv.set);
739 if (cfg == NULL) {
740 IPFW_UH_WUNLOCK(ch);
741 return (ESRCH);
742 }
743 if (cfg->no.refcnt > 0) {
744 IPFW_UH_WUNLOCK(ch);
745 return (EBUSY);
746 }
747
748 ipfw_reset_eaction_instance(ch, V_nptv6_eid, cfg->no.kidx);
749 SRV_OBJECT(ch, cfg->no.kidx) = NULL;
750 ipfw_objhash_del(CHAIN_TO_SRV(ch), &cfg->no);
751 ipfw_objhash_free_idx(CHAIN_TO_SRV(ch), cfg->no.kidx);
752 IPFW_UH_WUNLOCK(ch);
753
754 nptv6_free_config(cfg);
755 return (0);
756 }
757
758 /*
759 * Get or change nptv6 instance config.
760 * Request: [ ipfw_obj_header [ ipfw_nptv6_cfg ] ]
761 */
762 static int
nptv6_config(struct ip_fw_chain * chain,ip_fw3_opheader * op,struct sockopt_data * sd)763 nptv6_config(struct ip_fw_chain *chain, ip_fw3_opheader *op,
764 struct sockopt_data *sd)
765 {
766
767 return (EOPNOTSUPP);
768 }
769
770 /*
771 * Lists all NPTv6 instances currently available in kernel.
772 * Data layout (v0)(current):
773 * Request: [ ipfw_obj_lheader ]
774 * Reply: [ ipfw_obj_lheader ipfw_nptv6_cfg x N ]
775 *
776 * Returns 0 on success
777 */
778 static int
nptv6_list(struct ip_fw_chain * ch,ip_fw3_opheader * op3,struct sockopt_data * sd)779 nptv6_list(struct ip_fw_chain *ch, ip_fw3_opheader *op3,
780 struct sockopt_data *sd)
781 {
782 ipfw_obj_lheader *olh;
783 struct nptv6_dump_arg da;
784
785 /* Check minimum header size */
786 if (sd->valsize < sizeof(ipfw_obj_lheader))
787 return (EINVAL);
788
789 olh = (ipfw_obj_lheader *)ipfw_get_sopt_header(sd, sizeof(*olh));
790
791 IPFW_UH_RLOCK(ch);
792 olh->count = ipfw_objhash_count_type(CHAIN_TO_SRV(ch),
793 IPFW_TLV_NPTV6_NAME);
794 olh->objsize = sizeof(ipfw_nptv6_cfg);
795 olh->size = sizeof(*olh) + olh->count * olh->objsize;
796
797 if (sd->valsize < olh->size) {
798 IPFW_UH_RUNLOCK(ch);
799 return (ENOMEM);
800 }
801 memset(&da, 0, sizeof(da));
802 da.ch = ch;
803 da.sd = sd;
804 ipfw_objhash_foreach_type(CHAIN_TO_SRV(ch), export_config_cb,
805 &da, IPFW_TLV_NPTV6_NAME);
806 IPFW_UH_RUNLOCK(ch);
807
808 return (0);
809 }
810
811 #define __COPY_STAT_FIELD(_cfg, _stats, _field) \
812 (_stats)->_field = NPTV6STAT_FETCH(_cfg, _field)
813 static void
export_stats(struct ip_fw_chain * ch,struct nptv6_cfg * cfg,struct ipfw_nptv6_stats * stats)814 export_stats(struct ip_fw_chain *ch, struct nptv6_cfg *cfg,
815 struct ipfw_nptv6_stats *stats)
816 {
817
818 __COPY_STAT_FIELD(cfg, stats, in2ex);
819 __COPY_STAT_FIELD(cfg, stats, ex2in);
820 __COPY_STAT_FIELD(cfg, stats, dropped);
821 }
822
823 /*
824 * Get NPTv6 statistics.
825 * Data layout (v0)(current):
826 * Request: [ ipfw_obj_header ]
827 * Reply: [ ipfw_obj_header ipfw_obj_ctlv [ uint64_t x N ]]
828 *
829 * Returns 0 on success
830 */
831 static int
nptv6_stats(struct ip_fw_chain * ch,ip_fw3_opheader * op,struct sockopt_data * sd)832 nptv6_stats(struct ip_fw_chain *ch, ip_fw3_opheader *op,
833 struct sockopt_data *sd)
834 {
835 struct ipfw_nptv6_stats stats;
836 struct nptv6_cfg *cfg;
837 ipfw_obj_header *oh;
838 ipfw_obj_ctlv *ctlv;
839 size_t sz;
840
841 sz = sizeof(ipfw_obj_header) + sizeof(ipfw_obj_ctlv) + sizeof(stats);
842 if (sd->valsize % sizeof(uint64_t))
843 return (EINVAL);
844 if (sd->valsize < sz)
845 return (ENOMEM);
846 oh = (ipfw_obj_header *)ipfw_get_sopt_header(sd, sz);
847 if (oh == NULL)
848 return (EINVAL);
849 if (ipfw_check_object_name_generic(oh->ntlv.name) != 0 ||
850 oh->ntlv.set >= IPFW_MAX_SETS)
851 return (EINVAL);
852 memset(&stats, 0, sizeof(stats));
853
854 IPFW_UH_RLOCK(ch);
855 cfg = nptv6_find(CHAIN_TO_SRV(ch), oh->ntlv.name, oh->ntlv.set);
856 if (cfg == NULL) {
857 IPFW_UH_RUNLOCK(ch);
858 return (ESRCH);
859 }
860 export_stats(ch, cfg, &stats);
861 IPFW_UH_RUNLOCK(ch);
862
863 ctlv = (ipfw_obj_ctlv *)(oh + 1);
864 memset(ctlv, 0, sizeof(*ctlv));
865 ctlv->head.type = IPFW_TLV_COUNTERS;
866 ctlv->head.length = sz - sizeof(ipfw_obj_header);
867 ctlv->count = sizeof(stats) / sizeof(uint64_t);
868 ctlv->objsize = sizeof(uint64_t);
869 ctlv->version = 1;
870 memcpy(ctlv + 1, &stats, sizeof(stats));
871 return (0);
872 }
873
874 /*
875 * Reset NPTv6 statistics.
876 * Data layout (v0)(current):
877 * Request: [ ipfw_obj_header ]
878 *
879 * Returns 0 on success
880 */
881 static int
nptv6_reset_stats(struct ip_fw_chain * ch,ip_fw3_opheader * op,struct sockopt_data * sd)882 nptv6_reset_stats(struct ip_fw_chain *ch, ip_fw3_opheader *op,
883 struct sockopt_data *sd)
884 {
885 struct nptv6_cfg *cfg;
886 ipfw_obj_header *oh;
887
888 if (sd->valsize != sizeof(*oh))
889 return (EINVAL);
890 oh = (ipfw_obj_header *)sd->kbuf;
891 if (ipfw_check_object_name_generic(oh->ntlv.name) != 0 ||
892 oh->ntlv.set >= IPFW_MAX_SETS)
893 return (EINVAL);
894
895 IPFW_UH_WLOCK(ch);
896 cfg = nptv6_find(CHAIN_TO_SRV(ch), oh->ntlv.name, oh->ntlv.set);
897 if (cfg == NULL) {
898 IPFW_UH_WUNLOCK(ch);
899 return (ESRCH);
900 }
901 COUNTER_ARRAY_ZERO(cfg->stats, NPTV6STATS);
902 IPFW_UH_WUNLOCK(ch);
903 return (0);
904 }
905
906 static struct ipfw_sopt_handler scodes[] = {
907 { IP_FW_NPTV6_CREATE, 0, HDIR_SET, nptv6_create },
908 { IP_FW_NPTV6_DESTROY,0, HDIR_SET, nptv6_destroy },
909 { IP_FW_NPTV6_CONFIG, 0, HDIR_BOTH, nptv6_config },
910 { IP_FW_NPTV6_LIST, 0, HDIR_GET, nptv6_list },
911 { IP_FW_NPTV6_STATS, 0, HDIR_GET, nptv6_stats },
912 { IP_FW_NPTV6_RESET_STATS,0, HDIR_SET, nptv6_reset_stats },
913 };
914
915 static int
nptv6_classify(ipfw_insn * cmd,uint16_t * puidx,uint8_t * ptype)916 nptv6_classify(ipfw_insn *cmd, uint16_t *puidx, uint8_t *ptype)
917 {
918 ipfw_insn *icmd;
919
920 icmd = cmd - 1;
921 NPTV6_DEBUG("opcode %d, arg1 %d, opcode0 %d, arg1 %d",
922 cmd->opcode, cmd->arg1, icmd->opcode, icmd->arg1);
923 if (icmd->opcode != O_EXTERNAL_ACTION ||
924 icmd->arg1 != V_nptv6_eid)
925 return (1);
926
927 *puidx = cmd->arg1;
928 *ptype = 0;
929 return (0);
930 }
931
932 static void
nptv6_update_arg1(ipfw_insn * cmd,uint16_t idx)933 nptv6_update_arg1(ipfw_insn *cmd, uint16_t idx)
934 {
935
936 cmd->arg1 = idx;
937 NPTV6_DEBUG("opcode %d, arg1 -> %d", cmd->opcode, cmd->arg1);
938 }
939
940 static int
nptv6_findbyname(struct ip_fw_chain * ch,struct tid_info * ti,struct named_object ** pno)941 nptv6_findbyname(struct ip_fw_chain *ch, struct tid_info *ti,
942 struct named_object **pno)
943 {
944 int err;
945
946 err = ipfw_objhash_find_type(CHAIN_TO_SRV(ch), ti,
947 IPFW_TLV_NPTV6_NAME, pno);
948 NPTV6_DEBUG("uidx %u, type %u, err %d", ti->uidx, ti->type, err);
949 return (err);
950 }
951
952 static struct named_object *
nptv6_findbykidx(struct ip_fw_chain * ch,uint16_t idx)953 nptv6_findbykidx(struct ip_fw_chain *ch, uint16_t idx)
954 {
955 struct namedobj_instance *ni;
956 struct named_object *no;
957
958 IPFW_UH_WLOCK_ASSERT(ch);
959 ni = CHAIN_TO_SRV(ch);
960 no = ipfw_objhash_lookup_kidx(ni, idx);
961 KASSERT(no != NULL, ("NPT with index %d not found", idx));
962
963 NPTV6_DEBUG("kidx %u -> %s", idx, no->name);
964 return (no);
965 }
966
967 static int
nptv6_manage_sets(struct ip_fw_chain * ch,uint16_t set,uint8_t new_set,enum ipfw_sets_cmd cmd)968 nptv6_manage_sets(struct ip_fw_chain *ch, uint16_t set, uint8_t new_set,
969 enum ipfw_sets_cmd cmd)
970 {
971
972 return (ipfw_obj_manage_sets(CHAIN_TO_SRV(ch), IPFW_TLV_NPTV6_NAME,
973 set, new_set, cmd));
974 }
975
976 static struct opcode_obj_rewrite opcodes[] = {
977 {
978 .opcode = O_EXTERNAL_INSTANCE,
979 .etlv = IPFW_TLV_EACTION /* just show it isn't table */,
980 .classifier = nptv6_classify,
981 .update = nptv6_update_arg1,
982 .find_byname = nptv6_findbyname,
983 .find_bykidx = nptv6_findbykidx,
984 .manage_sets = nptv6_manage_sets,
985 },
986 };
987
988 static int
destroy_config_cb(struct namedobj_instance * ni,struct named_object * no,void * arg)989 destroy_config_cb(struct namedobj_instance *ni, struct named_object *no,
990 void *arg)
991 {
992 struct nptv6_cfg *cfg;
993 struct ip_fw_chain *ch;
994
995 ch = (struct ip_fw_chain *)arg;
996 IPFW_UH_WLOCK_ASSERT(ch);
997
998 cfg = (struct nptv6_cfg *)SRV_OBJECT(ch, no->kidx);
999 SRV_OBJECT(ch, no->kidx) = NULL;
1000 ipfw_objhash_del(ni, &cfg->no);
1001 ipfw_objhash_free_idx(ni, cfg->no.kidx);
1002 nptv6_free_config(cfg);
1003 return (0);
1004 }
1005
1006 int
nptv6_init(struct ip_fw_chain * ch,int first)1007 nptv6_init(struct ip_fw_chain *ch, int first)
1008 {
1009
1010 V_nptv6_eid = ipfw_add_eaction(ch, ipfw_nptv6, "nptv6");
1011 if (V_nptv6_eid == 0)
1012 return (ENXIO);
1013 IPFW_ADD_SOPT_HANDLER(first, scodes);
1014 IPFW_ADD_OBJ_REWRITER(first, opcodes);
1015 return (0);
1016 }
1017
1018 void
nptv6_uninit(struct ip_fw_chain * ch,int last)1019 nptv6_uninit(struct ip_fw_chain *ch, int last)
1020 {
1021
1022 if (last && nptv6_ifaddr_event != NULL)
1023 EVENTHANDLER_DEREGISTER(ifaddr_event_ext, nptv6_ifaddr_event);
1024 IPFW_DEL_OBJ_REWRITER(last, opcodes);
1025 IPFW_DEL_SOPT_HANDLER(last, scodes);
1026 ipfw_del_eaction(ch, V_nptv6_eid);
1027 /*
1028 * Since we already have deregistered external action,
1029 * our named objects become unaccessible via rules, because
1030 * all rules were truncated by ipfw_del_eaction().
1031 * So, we can unlink and destroy our named objects without holding
1032 * IPFW_WLOCK().
1033 */
1034 IPFW_UH_WLOCK(ch);
1035 ipfw_objhash_foreach_type(CHAIN_TO_SRV(ch), destroy_config_cb, ch,
1036 IPFW_TLV_NPTV6_NAME);
1037 V_nptv6_eid = 0;
1038 IPFW_UH_WUNLOCK(ch);
1039 }
1040