npf_inet.c revision 1.46 1 /* $NetBSD: npf_inet.c,v 1.46 2018/03/22 09:04:25 maxv Exp $ */
2
3 /*-
4 * Copyright (c) 2009-2014 The NetBSD Foundation, Inc.
5 * All rights reserved.
6 *
7 * This material is based upon work partially supported by The
8 * NetBSD Foundation under a contract with Mindaugas Rasiukevicius.
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 *
19 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
20 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
21 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
22 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
23 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29 * POSSIBILITY OF SUCH DAMAGE.
30 */
31
32 /*
33 * Various protocol related helper routines.
34 *
35 * This layer manipulates npf_cache_t structure i.e. caches requested headers
36 * and stores which information was cached in the information bit field.
37 * It is also responsibility of this layer to update or invalidate the cache
38 * on rewrites (e.g. by translation routines).
39 */
40
41 #ifdef _KERNEL
42 #include <sys/cdefs.h>
43 __KERNEL_RCSID(0, "$NetBSD: npf_inet.c,v 1.46 2018/03/22 09:04:25 maxv Exp $");
44
45 #include <sys/param.h>
46 #include <sys/types.h>
47
48 #include <net/pfil.h>
49 #include <net/if.h>
50 #include <net/ethertypes.h>
51 #include <net/if_ether.h>
52
53 #include <netinet/in_systm.h>
54 #include <netinet/in.h>
55 #include <netinet6/in6_var.h>
56 #include <netinet/ip.h>
57 #include <netinet/ip6.h>
58 #include <netinet/tcp.h>
59 #include <netinet/udp.h>
60 #include <netinet/ip_icmp.h>
61 #endif
62
63 #include "npf_impl.h"
64
65 /*
66 * npf_fixup{16,32}_cksum: incremental update of the Internet checksum.
67 */
68
69 uint16_t
70 npf_fixup16_cksum(uint16_t cksum, uint16_t odatum, uint16_t ndatum)
71 {
72 uint32_t sum;
73
74 /*
75 * RFC 1624:
76 * HC' = ~(~HC + ~m + m')
77 *
78 * Note: 1's complement sum is endian-independent (RFC 1071, page 2).
79 */
80 sum = ~cksum & 0xffff;
81 sum += (~odatum & 0xffff) + ndatum;
82 sum = (sum >> 16) + (sum & 0xffff);
83 sum += (sum >> 16);
84
85 return ~sum & 0xffff;
86 }
87
88 uint16_t
89 npf_fixup32_cksum(uint16_t cksum, uint32_t odatum, uint32_t ndatum)
90 {
91 uint32_t sum;
92
93 /*
94 * Checksum 32-bit datum as as two 16-bit. Note, the first
95 * 32->16 bit reduction is not necessary.
96 */
97 sum = ~cksum & 0xffff;
98 sum += (~odatum & 0xffff) + (ndatum & 0xffff);
99
100 sum += (~odatum >> 16) + (ndatum >> 16);
101 sum = (sum >> 16) + (sum & 0xffff);
102 sum += (sum >> 16);
103 return ~sum & 0xffff;
104 }
105
106 /*
107 * npf_addr_cksum: calculate checksum of the address, either IPv4 or IPv6.
108 */
109 uint16_t
110 npf_addr_cksum(uint16_t cksum, int sz, const npf_addr_t *oaddr,
111 const npf_addr_t *naddr)
112 {
113 const uint32_t *oip32 = (const uint32_t *)oaddr;
114 const uint32_t *nip32 = (const uint32_t *)naddr;
115
116 KASSERT(sz % sizeof(uint32_t) == 0);
117 do {
118 cksum = npf_fixup32_cksum(cksum, *oip32++, *nip32++);
119 sz -= sizeof(uint32_t);
120 } while (sz);
121
122 return cksum;
123 }
124
125 /*
126 * npf_addr_sum: provide IP addresses as a XORed 32-bit integer.
127 * Note: used for hash function.
128 */
129 uint32_t
130 npf_addr_mix(const int sz, const npf_addr_t *a1, const npf_addr_t *a2)
131 {
132 uint32_t mix = 0;
133
134 KASSERT(sz > 0 && a1 != NULL && a2 != NULL);
135
136 for (int i = 0; i < (sz >> 2); i++) {
137 mix ^= a1->word32[i];
138 mix ^= a2->word32[i];
139 }
140 return mix;
141 }
142
143 /*
144 * npf_addr_mask: apply the mask to a given address and store the result.
145 */
146 void
147 npf_addr_mask(const npf_addr_t *addr, const npf_netmask_t mask,
148 const int alen, npf_addr_t *out)
149 {
150 const int nwords = alen >> 2;
151 uint_fast8_t length = mask;
152
153 /* Note: maximum length is 32 for IPv4 and 128 for IPv6. */
154 KASSERT(length <= NPF_MAX_NETMASK);
155
156 for (int i = 0; i < nwords; i++) {
157 uint32_t wordmask;
158
159 if (length >= 32) {
160 wordmask = htonl(0xffffffff);
161 length -= 32;
162 } else if (length) {
163 wordmask = htonl(0xffffffff << (32 - length));
164 length = 0;
165 } else {
166 wordmask = 0;
167 }
168 out->word32[i] = addr->word32[i] & wordmask;
169 }
170 }
171
172 /*
173 * npf_addr_cmp: compare two addresses, either IPv4 or IPv6.
174 *
175 * => Return 0 if equal and negative/positive if less/greater accordingly.
176 * => Ignore the mask, if NPF_NO_NETMASK is specified.
177 */
178 int
179 npf_addr_cmp(const npf_addr_t *addr1, const npf_netmask_t mask1,
180 const npf_addr_t *addr2, const npf_netmask_t mask2, const int alen)
181 {
182 npf_addr_t realaddr1, realaddr2;
183
184 if (mask1 != NPF_NO_NETMASK) {
185 npf_addr_mask(addr1, mask1, alen, &realaddr1);
186 addr1 = &realaddr1;
187 }
188 if (mask2 != NPF_NO_NETMASK) {
189 npf_addr_mask(addr2, mask2, alen, &realaddr2);
190 addr2 = &realaddr2;
191 }
192 return memcmp(addr1, addr2, alen);
193 }
194
195 /*
196 * npf_tcpsaw: helper to fetch SEQ, ACK, WIN and return TCP data length.
197 *
198 * => Returns all values in host byte-order.
199 */
200 int
201 npf_tcpsaw(const npf_cache_t *npc, tcp_seq *seq, tcp_seq *ack, uint32_t *win)
202 {
203 const struct tcphdr *th = npc->npc_l4.tcp;
204 u_int thlen;
205
206 KASSERT(npf_iscached(npc, NPC_TCP));
207
208 *seq = ntohl(th->th_seq);
209 *ack = ntohl(th->th_ack);
210 *win = (uint32_t)ntohs(th->th_win);
211 thlen = th->th_off << 2;
212
213 if (npf_iscached(npc, NPC_IP4)) {
214 const struct ip *ip = npc->npc_ip.v4;
215 return ntohs(ip->ip_len) - npc->npc_hlen - thlen;
216 } else if (npf_iscached(npc, NPC_IP6)) {
217 const struct ip6_hdr *ip6 = npc->npc_ip.v6;
218 return ntohs(ip6->ip6_plen) -
219 (npc->npc_hlen - sizeof(*ip6)) - thlen;
220 }
221 return 0;
222 }
223
224 /*
225 * npf_fetch_tcpopts: parse and return TCP options.
226 */
227 bool
228 npf_fetch_tcpopts(npf_cache_t *npc, uint16_t *mss, int *wscale)
229 {
230 nbuf_t *nbuf = npc->npc_nbuf;
231 const struct tcphdr *th = npc->npc_l4.tcp;
232 int topts_len, step;
233 uint8_t *nptr;
234 uint8_t val;
235 bool ok;
236
237 KASSERT(npf_iscached(npc, NPC_IP46));
238 KASSERT(npf_iscached(npc, NPC_TCP));
239
240 /* Determine if there are any TCP options, get their length. */
241 topts_len = (th->th_off << 2) - sizeof(struct tcphdr);
242 if (topts_len <= 0) {
243 /* No options. */
244 return false;
245 }
246 KASSERT(topts_len <= MAX_TCPOPTLEN);
247
248 /* First step: IP and TCP header up to options. */
249 step = npc->npc_hlen + sizeof(struct tcphdr);
250 nbuf_reset(nbuf);
251 next:
252 if ((nptr = nbuf_advance(nbuf, step, 1)) == NULL) {
253 ok = false;
254 goto done;
255 }
256 val = *nptr;
257
258 switch (val) {
259 case TCPOPT_EOL:
260 /* Done. */
261 ok = true;
262 goto done;
263 case TCPOPT_NOP:
264 topts_len--;
265 step = 1;
266 break;
267 case TCPOPT_MAXSEG:
268 if ((nptr = nbuf_ensure_contig(nbuf, TCPOLEN_MAXSEG)) == NULL) {
269 ok = false;
270 goto done;
271 }
272 if (mss) {
273 if (*mss) {
274 memcpy(nptr + 2, mss, sizeof(uint16_t));
275 } else {
276 memcpy(mss, nptr + 2, sizeof(uint16_t));
277 }
278 }
279 topts_len -= TCPOLEN_MAXSEG;
280 step = TCPOLEN_MAXSEG;
281 break;
282 case TCPOPT_WINDOW:
283 /* TCP Window Scaling (RFC 1323). */
284 if ((nptr = nbuf_ensure_contig(nbuf, TCPOLEN_WINDOW)) == NULL) {
285 ok = false;
286 goto done;
287 }
288 val = *(nptr + 2);
289 *wscale = (val > TCP_MAX_WINSHIFT) ? TCP_MAX_WINSHIFT : val;
290 topts_len -= TCPOLEN_WINDOW;
291 step = TCPOLEN_WINDOW;
292 break;
293 default:
294 if ((nptr = nbuf_ensure_contig(nbuf, 2)) == NULL) {
295 ok = false;
296 goto done;
297 }
298 val = *(nptr + 1);
299 if (val < 2 || val > topts_len) {
300 ok = false;
301 goto done;
302 }
303 topts_len -= val;
304 step = val;
305 }
306
307 /* Any options left? */
308 if (__predict_true(topts_len > 0)) {
309 goto next;
310 }
311 ok = true;
312 done:
313 if (nbuf_flag_p(nbuf, NBUF_DATAREF_RESET)) {
314 npf_recache(npc);
315 }
316 return ok;
317 }
318
319 static int
320 npf_cache_ip(npf_cache_t *npc, nbuf_t *nbuf)
321 {
322 const void *nptr = nbuf_dataptr(nbuf);
323 const uint8_t ver = *(const uint8_t *)nptr;
324 int flags = 0;
325
326 /*
327 * We intentionally don't read the L4 payload after IPPROTO_AH.
328 */
329
330 switch (ver >> 4) {
331 case IPVERSION: {
332 struct ip *ip;
333
334 ip = nbuf_ensure_contig(nbuf, sizeof(struct ip));
335 if (ip == NULL) {
336 return NPC_FMTERR;
337 }
338
339 /* Retrieve the complete header. */
340 if ((u_int)(ip->ip_hl << 2) < sizeof(struct ip)) {
341 return NPC_FMTERR;
342 }
343 ip = nbuf_ensure_contig(nbuf, (u_int)(ip->ip_hl << 2));
344 if (ip == NULL) {
345 return NPC_FMTERR;
346 }
347
348 if (ip->ip_off & ~htons(IP_DF | IP_RF)) {
349 /* Note fragmentation. */
350 flags |= NPC_IPFRAG;
351 }
352
353 /* Cache: layer 3 - IPv4. */
354 npc->npc_alen = sizeof(struct in_addr);
355 npc->npc_ips[NPF_SRC] = (npf_addr_t *)&ip->ip_src;
356 npc->npc_ips[NPF_DST] = (npf_addr_t *)&ip->ip_dst;
357 npc->npc_hlen = ip->ip_hl << 2;
358 npc->npc_proto = ip->ip_p;
359
360 npc->npc_ip.v4 = ip;
361 flags |= NPC_IP4;
362 break;
363 }
364
365 case (IPV6_VERSION >> 4): {
366 struct ip6_hdr *ip6;
367 struct ip6_ext *ip6e;
368 struct ip6_frag *ip6f;
369 size_t off, hlen;
370 int frag_present;
371
372 ip6 = nbuf_ensure_contig(nbuf, sizeof(struct ip6_hdr));
373 if (ip6 == NULL) {
374 return NPC_FMTERR;
375 }
376
377 /*
378 * XXX: We don't handle IPv6 Jumbograms.
379 */
380
381 /* Set initial next-protocol value. */
382 hlen = sizeof(struct ip6_hdr);
383 npc->npc_proto = ip6->ip6_nxt;
384 npc->npc_hlen = hlen;
385
386 frag_present = 0;
387
388 /*
389 * Advance by the length of the current header.
390 */
391 off = nbuf_offset(nbuf);
392 while ((ip6e = nbuf_advance(nbuf, hlen, sizeof(*ip6e))) != NULL) {
393 /*
394 * Determine whether we are going to continue.
395 */
396 switch (npc->npc_proto) {
397 case IPPROTO_HOPOPTS:
398 case IPPROTO_DSTOPTS:
399 case IPPROTO_ROUTING:
400 hlen = (ip6e->ip6e_len + 1) << 3;
401 break;
402 case IPPROTO_FRAGMENT:
403 if (frag_present++)
404 return NPC_FMTERR;
405 ip6f = nbuf_ensure_contig(nbuf, sizeof(*ip6f));
406 if (ip6f == NULL)
407 return NPC_FMTERR;
408
409 /* RFC6946: Skip dummy fragments. */
410 if (!ntohs(ip6f->ip6f_offlg & IP6F_OFF_MASK) &&
411 !(ip6f->ip6f_offlg & IP6F_MORE_FRAG)) {
412 hlen = sizeof(struct ip6_frag);
413 break;
414 }
415
416 hlen = 0;
417 flags |= NPC_IPFRAG;
418
419 break;
420 default:
421 hlen = 0;
422 break;
423 }
424
425 if (!hlen) {
426 break;
427 }
428 npc->npc_proto = ip6e->ip6e_nxt;
429 npc->npc_hlen += hlen;
430 }
431
432 if (ip6e == NULL) {
433 return NPC_FMTERR;
434 }
435
436 /*
437 * Re-fetch the header pointers (nbufs might have been
438 * reallocated). Restore the original offset (if any).
439 */
440 nbuf_reset(nbuf);
441 ip6 = nbuf_dataptr(nbuf);
442 if (off) {
443 nbuf_advance(nbuf, off, 0);
444 }
445
446 /* Cache: layer 3 - IPv6. */
447 npc->npc_alen = sizeof(struct in6_addr);
448 npc->npc_ips[NPF_SRC] = (npf_addr_t *)&ip6->ip6_src;
449 npc->npc_ips[NPF_DST] = (npf_addr_t *)&ip6->ip6_dst;
450
451 npc->npc_ip.v6 = ip6;
452 flags |= NPC_IP6;
453 break;
454 }
455 default:
456 break;
457 }
458 return flags;
459 }
460
461 /*
462 * npf_cache_all: general routine to cache all relevant IP (v4 or v6)
463 * and TCP, UDP or ICMP headers.
464 *
465 * => nbuf offset shall be set accordingly.
466 */
467 int
468 npf_cache_all(npf_cache_t *npc)
469 {
470 nbuf_t *nbuf = npc->npc_nbuf;
471 int flags, l4flags;
472 u_int hlen;
473
474 /*
475 * This routine is a main point where the references are cached,
476 * therefore clear the flag as we reset.
477 */
478 again:
479 nbuf_unset_flag(nbuf, NBUF_DATAREF_RESET);
480
481 /*
482 * First, cache the L3 header (IPv4 or IPv6). If IP packet is
483 * fragmented, then we cannot look into L4.
484 */
485 flags = npf_cache_ip(npc, nbuf);
486 if ((flags & NPC_IP46) == 0 || (flags & NPC_IPFRAG) != 0 ||
487 (flags & NPC_FMTERR) != 0) {
488 nbuf_unset_flag(nbuf, NBUF_DATAREF_RESET);
489 npc->npc_info |= flags;
490 return flags;
491 }
492 hlen = npc->npc_hlen;
493
494 /*
495 * Note: we guarantee that the potential "Query Id" field of the
496 * ICMPv4/ICMPv6 packets is in the nbuf. This field is used in the
497 * ICMP ALG.
498 */
499 switch (npc->npc_proto) {
500 case IPPROTO_TCP:
501 /* Cache: layer 4 - TCP. */
502 npc->npc_l4.tcp = nbuf_advance(nbuf, hlen,
503 sizeof(struct tcphdr));
504 l4flags = NPC_LAYER4 | NPC_TCP;
505 break;
506 case IPPROTO_UDP:
507 /* Cache: layer 4 - UDP. */
508 npc->npc_l4.udp = nbuf_advance(nbuf, hlen,
509 sizeof(struct udphdr));
510 l4flags = NPC_LAYER4 | NPC_UDP;
511 break;
512 case IPPROTO_ICMP:
513 /* Cache: layer 4 - ICMPv4. */
514 npc->npc_l4.icmp = nbuf_advance(nbuf, hlen,
515 ICMP_MINLEN);
516 l4flags = NPC_LAYER4 | NPC_ICMP;
517 break;
518 case IPPROTO_ICMPV6:
519 /* Cache: layer 4 - ICMPv6. */
520 npc->npc_l4.icmp6 = nbuf_advance(nbuf, hlen,
521 sizeof(struct icmp6_hdr));
522 l4flags = NPC_LAYER4 | NPC_ICMP;
523 break;
524 default:
525 l4flags = 0;
526 break;
527 }
528
529 if (nbuf_flag_p(nbuf, NBUF_DATAREF_RESET)) {
530 goto again;
531 }
532
533 /* Add the L4 flags if nbuf_advance() succeeded. */
534 if (l4flags && npc->npc_l4.hdr) {
535 flags |= l4flags;
536 }
537 npc->npc_info |= flags;
538 return flags;
539 }
540
541 void
542 npf_recache(npf_cache_t *npc)
543 {
544 nbuf_t *nbuf = npc->npc_nbuf;
545 const int mflags __diagused = npc->npc_info & (NPC_IP46 | NPC_LAYER4);
546 int flags __diagused;
547
548 nbuf_reset(nbuf);
549 npc->npc_info = 0;
550 flags = npf_cache_all(npc);
551
552 KASSERT((flags & mflags) == mflags);
553 KASSERT(nbuf_flag_p(nbuf, NBUF_DATAREF_RESET) == 0);
554 }
555
556 /*
557 * npf_rwrip: rewrite required IP address.
558 */
559 bool
560 npf_rwrip(const npf_cache_t *npc, u_int which, const npf_addr_t *addr)
561 {
562 KASSERT(npf_iscached(npc, NPC_IP46));
563 KASSERT(which == NPF_SRC || which == NPF_DST);
564
565 memcpy(npc->npc_ips[which], addr, npc->npc_alen);
566 return true;
567 }
568
569 /*
570 * npf_rwrport: rewrite required TCP/UDP port.
571 */
572 bool
573 npf_rwrport(const npf_cache_t *npc, u_int which, const in_port_t port)
574 {
575 const int proto = npc->npc_proto;
576 in_port_t *oport;
577
578 KASSERT(npf_iscached(npc, NPC_TCP) || npf_iscached(npc, NPC_UDP));
579 KASSERT(proto == IPPROTO_TCP || proto == IPPROTO_UDP);
580 KASSERT(which == NPF_SRC || which == NPF_DST);
581
582 /* Get the offset and store the port in it. */
583 if (proto == IPPROTO_TCP) {
584 struct tcphdr *th = npc->npc_l4.tcp;
585 oport = (which == NPF_SRC) ? &th->th_sport : &th->th_dport;
586 } else {
587 struct udphdr *uh = npc->npc_l4.udp;
588 oport = (which == NPF_SRC) ? &uh->uh_sport : &uh->uh_dport;
589 }
590 memcpy(oport, &port, sizeof(in_port_t));
591 return true;
592 }
593
594 /*
595 * npf_rwrcksum: rewrite IPv4 and/or TCP/UDP checksum.
596 */
597 bool
598 npf_rwrcksum(const npf_cache_t *npc, u_int which,
599 const npf_addr_t *addr, const in_port_t port)
600 {
601 const npf_addr_t *oaddr = npc->npc_ips[which];
602 const int proto = npc->npc_proto;
603 const int alen = npc->npc_alen;
604 uint16_t *ocksum;
605 in_port_t oport;
606
607 KASSERT(npf_iscached(npc, NPC_LAYER4));
608 KASSERT(which == NPF_SRC || which == NPF_DST);
609
610 if (npf_iscached(npc, NPC_IP4)) {
611 struct ip *ip = npc->npc_ip.v4;
612 uint16_t ipsum = ip->ip_sum;
613
614 /* Recalculate IPv4 checksum and rewrite. */
615 ip->ip_sum = npf_addr_cksum(ipsum, alen, oaddr, addr);
616 } else {
617 /* No checksum for IPv6. */
618 KASSERT(npf_iscached(npc, NPC_IP6));
619 }
620
621 /* Nothing else to do for ICMP. */
622 if (proto == IPPROTO_ICMP || proto == IPPROTO_ICMPV6) {
623 return true;
624 }
625 KASSERT(npf_iscached(npc, NPC_TCP) || npf_iscached(npc, NPC_UDP));
626
627 /*
628 * Calculate TCP/UDP checksum:
629 * - Skip if UDP and the current checksum is zero.
630 * - Fixup the IP address change.
631 * - Fixup the port change, if required (non-zero).
632 */
633 if (proto == IPPROTO_TCP) {
634 struct tcphdr *th = npc->npc_l4.tcp;
635
636 ocksum = &th->th_sum;
637 oport = (which == NPF_SRC) ? th->th_sport : th->th_dport;
638 } else {
639 struct udphdr *uh = npc->npc_l4.udp;
640
641 KASSERT(proto == IPPROTO_UDP);
642 ocksum = &uh->uh_sum;
643 if (*ocksum == 0) {
644 /* No need to update. */
645 return true;
646 }
647 oport = (which == NPF_SRC) ? uh->uh_sport : uh->uh_dport;
648 }
649
650 uint16_t cksum = npf_addr_cksum(*ocksum, alen, oaddr, addr);
651 if (port) {
652 cksum = npf_fixup16_cksum(cksum, oport, port);
653 }
654
655 /* Rewrite TCP/UDP checksum. */
656 memcpy(ocksum, &cksum, sizeof(uint16_t));
657 return true;
658 }
659
660 /*
661 * npf_napt_rwr: perform address and/or port translation.
662 */
663 int
664 npf_napt_rwr(const npf_cache_t *npc, u_int which,
665 const npf_addr_t *addr, const in_addr_t port)
666 {
667 const unsigned proto = npc->npc_proto;
668
669 /*
670 * Rewrite IP and/or TCP/UDP checksums first, since we need the
671 * current (old) address/port for the calculations. Then perform
672 * the address translation i.e. rewrite source or destination.
673 */
674 if (!npf_rwrcksum(npc, which, addr, port)) {
675 return EINVAL;
676 }
677 if (!npf_rwrip(npc, which, addr)) {
678 return EINVAL;
679 }
680 if (port == 0) {
681 /* Done. */
682 return 0;
683 }
684
685 switch (proto) {
686 case IPPROTO_TCP:
687 case IPPROTO_UDP:
688 /* Rewrite source/destination port. */
689 if (!npf_rwrport(npc, which, port)) {
690 return EINVAL;
691 }
692 break;
693 case IPPROTO_ICMP:
694 case IPPROTO_ICMPV6:
695 KASSERT(npf_iscached(npc, NPC_ICMP));
696 /* Nothing. */
697 break;
698 default:
699 return ENOTSUP;
700 }
701 return 0;
702 }
703
704 /*
705 * IPv6-to-IPv6 Network Prefix Translation (NPTv6), as per RFC 6296.
706 */
707
708 int
709 npf_npt66_rwr(const npf_cache_t *npc, u_int which, const npf_addr_t *pref,
710 npf_netmask_t len, uint16_t adj)
711 {
712 npf_addr_t *addr = npc->npc_ips[which];
713 unsigned remnant, word, preflen = len >> 4;
714 uint32_t sum;
715
716 KASSERT(which == NPF_SRC || which == NPF_DST);
717
718 if (!npf_iscached(npc, NPC_IP6)) {
719 return EINVAL;
720 }
721 if (len <= 48) {
722 /*
723 * The word to adjust. Cannot translate the 0xffff
724 * subnet if /48 or shorter.
725 */
726 word = 3;
727 if (addr->word16[word] == 0xffff) {
728 return EINVAL;
729 }
730 } else {
731 /*
732 * Also, all 0s or 1s in the host part are disallowed for
733 * longer than /48 prefixes.
734 */
735 if ((addr->word32[2] == 0 && addr->word32[3] == 0) ||
736 (addr->word32[2] == ~0U && addr->word32[3] == ~0U))
737 return EINVAL;
738
739 /* Determine the 16-bit word to adjust. */
740 for (word = 4; word < 8; word++)
741 if (addr->word16[word] != 0xffff)
742 break;
743 }
744
745 /* Rewrite the prefix. */
746 for (unsigned i = 0; i < preflen; i++) {
747 addr->word16[i] = pref->word16[i];
748 }
749
750 /*
751 * If prefix length is within a 16-bit word (not dividable by 16),
752 * then prepare a mask, determine the word and adjust it.
753 */
754 if ((remnant = len - (preflen << 4)) != 0) {
755 const uint16_t wordmask = (1U << remnant) - 1;
756 const unsigned i = preflen;
757
758 addr->word16[i] = (pref->word16[i] & wordmask) |
759 (addr->word16[i] & ~wordmask);
760 }
761
762 /*
763 * Performing 1's complement sum/difference.
764 */
765 sum = addr->word16[word] + adj;
766 while (sum >> 16) {
767 sum = (sum >> 16) + (sum & 0xffff);
768 }
769 if (sum == 0xffff) {
770 /* RFC 1071. */
771 sum = 0x0000;
772 }
773 addr->word16[word] = sum;
774 return 0;
775 }
776
777 #if defined(DDB) || defined(_NPF_TESTING)
778
779 const char *
780 npf_addr_dump(const npf_addr_t *addr, int alen)
781 {
782 if (alen == sizeof(struct in_addr)) {
783 struct in_addr ip;
784 memcpy(&ip, addr, alen);
785 return inet_ntoa(ip);
786 }
787 return "[IPv6]";
788 }
789
790 #endif
791