npf_inet.c revision 1.47 1 1.47 maxv /* $NetBSD: npf_inet.c,v 1.47 2018/03/23 08:28:54 maxv Exp $ */
2 1.1 rmind
3 1.1 rmind /*-
4 1.29 rmind * Copyright (c) 2009-2014 The NetBSD Foundation, Inc.
5 1.1 rmind * All rights reserved.
6 1.1 rmind *
7 1.1 rmind * This material is based upon work partially supported by The
8 1.1 rmind * NetBSD Foundation under a contract with Mindaugas Rasiukevicius.
9 1.1 rmind *
10 1.1 rmind * Redistribution and use in source and binary forms, with or without
11 1.1 rmind * modification, are permitted provided that the following conditions
12 1.1 rmind * are met:
13 1.1 rmind * 1. Redistributions of source code must retain the above copyright
14 1.1 rmind * notice, this list of conditions and the following disclaimer.
15 1.1 rmind * 2. Redistributions in binary form must reproduce the above copyright
16 1.1 rmind * notice, this list of conditions and the following disclaimer in the
17 1.1 rmind * documentation and/or other materials provided with the distribution.
18 1.1 rmind *
19 1.1 rmind * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
20 1.1 rmind * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
21 1.1 rmind * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
22 1.1 rmind * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
23 1.1 rmind * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24 1.1 rmind * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25 1.1 rmind * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26 1.1 rmind * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27 1.1 rmind * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28 1.1 rmind * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29 1.1 rmind * POSSIBILITY OF SUCH DAMAGE.
30 1.1 rmind */
31 1.1 rmind
32 1.1 rmind /*
33 1.22 rmind * Various protocol related helper routines.
34 1.12 rmind *
35 1.12 rmind * This layer manipulates npf_cache_t structure i.e. caches requested headers
36 1.12 rmind * and stores which information was cached in the information bit field.
37 1.12 rmind * It is also responsibility of this layer to update or invalidate the cache
38 1.12 rmind * on rewrites (e.g. by translation routines).
39 1.1 rmind */
40 1.1 rmind
41 1.36 christos #ifdef _KERNEL
42 1.1 rmind #include <sys/cdefs.h>
43 1.47 maxv __KERNEL_RCSID(0, "$NetBSD: npf_inet.c,v 1.47 2018/03/23 08:28:54 maxv Exp $");
44 1.1 rmind
45 1.1 rmind #include <sys/param.h>
46 1.11 rmind #include <sys/types.h>
47 1.1 rmind
48 1.4 rmind #include <net/pfil.h>
49 1.4 rmind #include <net/if.h>
50 1.4 rmind #include <net/ethertypes.h>
51 1.4 rmind #include <net/if_ether.h>
52 1.4 rmind
53 1.1 rmind #include <netinet/in_systm.h>
54 1.1 rmind #include <netinet/in.h>
55 1.33 mlelstv #include <netinet6/in6_var.h>
56 1.1 rmind #include <netinet/ip.h>
57 1.4 rmind #include <netinet/ip6.h>
58 1.1 rmind #include <netinet/tcp.h>
59 1.1 rmind #include <netinet/udp.h>
60 1.1 rmind #include <netinet/ip_icmp.h>
61 1.36 christos #endif
62 1.1 rmind
63 1.1 rmind #include "npf_impl.h"
64 1.1 rmind
65 1.1 rmind /*
66 1.27 rmind * npf_fixup{16,32}_cksum: incremental update of the Internet checksum.
67 1.1 rmind */
68 1.1 rmind
69 1.1 rmind uint16_t
70 1.1 rmind npf_fixup16_cksum(uint16_t cksum, uint16_t odatum, uint16_t ndatum)
71 1.1 rmind {
72 1.1 rmind uint32_t sum;
73 1.1 rmind
74 1.1 rmind /*
75 1.1 rmind * RFC 1624:
76 1.1 rmind * HC' = ~(~HC + ~m + m')
77 1.27 rmind *
78 1.27 rmind * Note: 1's complement sum is endian-independent (RFC 1071, page 2).
79 1.1 rmind */
80 1.27 rmind sum = ~cksum & 0xffff;
81 1.27 rmind sum += (~odatum & 0xffff) + ndatum;
82 1.1 rmind sum = (sum >> 16) + (sum & 0xffff);
83 1.1 rmind sum += (sum >> 16);
84 1.1 rmind
85 1.27 rmind return ~sum & 0xffff;
86 1.1 rmind }
87 1.1 rmind
88 1.1 rmind uint16_t
89 1.1 rmind npf_fixup32_cksum(uint16_t cksum, uint32_t odatum, uint32_t ndatum)
90 1.1 rmind {
91 1.27 rmind uint32_t sum;
92 1.27 rmind
93 1.27 rmind /*
94 1.27 rmind * Checksum 32-bit datum as as two 16-bit. Note, the first
95 1.27 rmind * 32->16 bit reduction is not necessary.
96 1.27 rmind */
97 1.27 rmind sum = ~cksum & 0xffff;
98 1.27 rmind sum += (~odatum & 0xffff) + (ndatum & 0xffff);
99 1.1 rmind
100 1.27 rmind sum += (~odatum >> 16) + (ndatum >> 16);
101 1.27 rmind sum = (sum >> 16) + (sum & 0xffff);
102 1.27 rmind sum += (sum >> 16);
103 1.27 rmind return ~sum & 0xffff;
104 1.1 rmind }
105 1.1 rmind
106 1.1 rmind /*
107 1.4 rmind * npf_addr_cksum: calculate checksum of the address, either IPv4 or IPv6.
108 1.4 rmind */
109 1.4 rmind uint16_t
110 1.19 rmind npf_addr_cksum(uint16_t cksum, int sz, const npf_addr_t *oaddr,
111 1.19 rmind const npf_addr_t *naddr)
112 1.4 rmind {
113 1.19 rmind const uint32_t *oip32 = (const uint32_t *)oaddr;
114 1.19 rmind const uint32_t *nip32 = (const uint32_t *)naddr;
115 1.4 rmind
116 1.4 rmind KASSERT(sz % sizeof(uint32_t) == 0);
117 1.4 rmind do {
118 1.4 rmind cksum = npf_fixup32_cksum(cksum, *oip32++, *nip32++);
119 1.4 rmind sz -= sizeof(uint32_t);
120 1.4 rmind } while (sz);
121 1.4 rmind
122 1.4 rmind return cksum;
123 1.4 rmind }
124 1.4 rmind
125 1.4 rmind /*
126 1.26 rmind * npf_addr_sum: provide IP addresses as a XORed 32-bit integer.
127 1.4 rmind * Note: used for hash function.
128 1.1 rmind */
129 1.4 rmind uint32_t
130 1.26 rmind npf_addr_mix(const int sz, const npf_addr_t *a1, const npf_addr_t *a2)
131 1.1 rmind {
132 1.4 rmind uint32_t mix = 0;
133 1.1 rmind
134 1.5 rmind KASSERT(sz > 0 && a1 != NULL && a2 != NULL);
135 1.5 rmind
136 1.26 rmind for (int i = 0; i < (sz >> 2); i++) {
137 1.36 christos mix ^= a1->word32[i];
138 1.36 christos mix ^= a2->word32[i];
139 1.4 rmind }
140 1.4 rmind return mix;
141 1.4 rmind }
142 1.1 rmind
143 1.13 rmind /*
144 1.13 rmind * npf_addr_mask: apply the mask to a given address and store the result.
145 1.13 rmind */
146 1.13 rmind void
147 1.13 rmind npf_addr_mask(const npf_addr_t *addr, const npf_netmask_t mask,
148 1.13 rmind const int alen, npf_addr_t *out)
149 1.12 rmind {
150 1.13 rmind const int nwords = alen >> 2;
151 1.12 rmind uint_fast8_t length = mask;
152 1.12 rmind
153 1.12 rmind /* Note: maximum length is 32 for IPv4 and 128 for IPv6. */
154 1.12 rmind KASSERT(length <= NPF_MAX_NETMASK);
155 1.12 rmind
156 1.13 rmind for (int i = 0; i < nwords; i++) {
157 1.13 rmind uint32_t wordmask;
158 1.13 rmind
159 1.12 rmind if (length >= 32) {
160 1.13 rmind wordmask = htonl(0xffffffff);
161 1.12 rmind length -= 32;
162 1.13 rmind } else if (length) {
163 1.13 rmind wordmask = htonl(0xffffffff << (32 - length));
164 1.13 rmind length = 0;
165 1.12 rmind } else {
166 1.13 rmind wordmask = 0;
167 1.12 rmind }
168 1.36 christos out->word32[i] = addr->word32[i] & wordmask;
169 1.12 rmind }
170 1.12 rmind }
171 1.12 rmind
172 1.12 rmind /*
173 1.12 rmind * npf_addr_cmp: compare two addresses, either IPv4 or IPv6.
174 1.12 rmind *
175 1.13 rmind * => Return 0 if equal and negative/positive if less/greater accordingly.
176 1.12 rmind * => Ignore the mask, if NPF_NO_NETMASK is specified.
177 1.12 rmind */
178 1.12 rmind int
179 1.12 rmind npf_addr_cmp(const npf_addr_t *addr1, const npf_netmask_t mask1,
180 1.13 rmind const npf_addr_t *addr2, const npf_netmask_t mask2, const int alen)
181 1.12 rmind {
182 1.13 rmind npf_addr_t realaddr1, realaddr2;
183 1.12 rmind
184 1.12 rmind if (mask1 != NPF_NO_NETMASK) {
185 1.13 rmind npf_addr_mask(addr1, mask1, alen, &realaddr1);
186 1.13 rmind addr1 = &realaddr1;
187 1.12 rmind }
188 1.12 rmind if (mask2 != NPF_NO_NETMASK) {
189 1.13 rmind npf_addr_mask(addr2, mask2, alen, &realaddr2);
190 1.13 rmind addr2 = &realaddr2;
191 1.12 rmind }
192 1.13 rmind return memcmp(addr1, addr2, alen);
193 1.12 rmind }
194 1.12 rmind
195 1.4 rmind /*
196 1.4 rmind * npf_tcpsaw: helper to fetch SEQ, ACK, WIN and return TCP data length.
197 1.12 rmind *
198 1.12 rmind * => Returns all values in host byte-order.
199 1.4 rmind */
200 1.4 rmind int
201 1.12 rmind npf_tcpsaw(const npf_cache_t *npc, tcp_seq *seq, tcp_seq *ack, uint32_t *win)
202 1.4 rmind {
203 1.19 rmind const struct tcphdr *th = npc->npc_l4.tcp;
204 1.8 rmind u_int thlen;
205 1.1 rmind
206 1.7 zoltan KASSERT(npf_iscached(npc, NPC_TCP));
207 1.1 rmind
208 1.4 rmind *seq = ntohl(th->th_seq);
209 1.4 rmind *ack = ntohl(th->th_ack);
210 1.4 rmind *win = (uint32_t)ntohs(th->th_win);
211 1.8 rmind thlen = th->th_off << 2;
212 1.1 rmind
213 1.7 zoltan if (npf_iscached(npc, NPC_IP4)) {
214 1.19 rmind const struct ip *ip = npc->npc_ip.v4;
215 1.21 rmind return ntohs(ip->ip_len) - npc->npc_hlen - thlen;
216 1.12 rmind } else if (npf_iscached(npc, NPC_IP6)) {
217 1.19 rmind const struct ip6_hdr *ip6 = npc->npc_ip.v6;
218 1.42 maxv return ntohs(ip6->ip6_plen) -
219 1.42 maxv (npc->npc_hlen - sizeof(*ip6)) - thlen;
220 1.7 zoltan }
221 1.7 zoltan return 0;
222 1.1 rmind }
223 1.1 rmind
224 1.1 rmind /*
225 1.4 rmind * npf_fetch_tcpopts: parse and return TCP options.
226 1.1 rmind */
227 1.1 rmind bool
228 1.32 rmind npf_fetch_tcpopts(npf_cache_t *npc, uint16_t *mss, int *wscale)
229 1.1 rmind {
230 1.32 rmind nbuf_t *nbuf = npc->npc_nbuf;
231 1.19 rmind const struct tcphdr *th = npc->npc_l4.tcp;
232 1.4 rmind int topts_len, step;
233 1.39 maxv uint8_t *nptr;
234 1.4 rmind uint8_t val;
235 1.19 rmind bool ok;
236 1.4 rmind
237 1.7 zoltan KASSERT(npf_iscached(npc, NPC_IP46));
238 1.7 zoltan KASSERT(npf_iscached(npc, NPC_TCP));
239 1.10 rmind
240 1.4 rmind /* Determine if there are any TCP options, get their length. */
241 1.4 rmind topts_len = (th->th_off << 2) - sizeof(struct tcphdr);
242 1.4 rmind if (topts_len <= 0) {
243 1.4 rmind /* No options. */
244 1.1 rmind return false;
245 1.4 rmind }
246 1.4 rmind KASSERT(topts_len <= MAX_TCPOPTLEN);
247 1.1 rmind
248 1.4 rmind /* First step: IP and TCP header up to options. */
249 1.21 rmind step = npc->npc_hlen + sizeof(struct tcphdr);
250 1.19 rmind nbuf_reset(nbuf);
251 1.4 rmind next:
252 1.19 rmind if ((nptr = nbuf_advance(nbuf, step, 1)) == NULL) {
253 1.19 rmind ok = false;
254 1.19 rmind goto done;
255 1.4 rmind }
256 1.39 maxv val = *nptr;
257 1.12 rmind
258 1.4 rmind switch (val) {
259 1.4 rmind case TCPOPT_EOL:
260 1.4 rmind /* Done. */
261 1.19 rmind ok = true;
262 1.19 rmind goto done;
263 1.4 rmind case TCPOPT_NOP:
264 1.4 rmind topts_len--;
265 1.4 rmind step = 1;
266 1.4 rmind break;
267 1.4 rmind case TCPOPT_MAXSEG:
268 1.39 maxv if ((nptr = nbuf_ensure_contig(nbuf, TCPOLEN_MAXSEG)) == NULL) {
269 1.19 rmind ok = false;
270 1.19 rmind goto done;
271 1.4 rmind }
272 1.4 rmind if (mss) {
273 1.19 rmind if (*mss) {
274 1.39 maxv memcpy(nptr + 2, mss, sizeof(uint16_t));
275 1.19 rmind } else {
276 1.39 maxv memcpy(mss, nptr + 2, sizeof(uint16_t));
277 1.19 rmind }
278 1.4 rmind }
279 1.4 rmind topts_len -= TCPOLEN_MAXSEG;
280 1.39 maxv step = TCPOLEN_MAXSEG;
281 1.4 rmind break;
282 1.4 rmind case TCPOPT_WINDOW:
283 1.10 rmind /* TCP Window Scaling (RFC 1323). */
284 1.39 maxv if ((nptr = nbuf_ensure_contig(nbuf, TCPOLEN_WINDOW)) == NULL) {
285 1.19 rmind ok = false;
286 1.19 rmind goto done;
287 1.4 rmind }
288 1.39 maxv val = *(nptr + 2);
289 1.4 rmind *wscale = (val > TCP_MAX_WINSHIFT) ? TCP_MAX_WINSHIFT : val;
290 1.4 rmind topts_len -= TCPOLEN_WINDOW;
291 1.39 maxv step = TCPOLEN_WINDOW;
292 1.4 rmind break;
293 1.4 rmind default:
294 1.39 maxv if ((nptr = nbuf_ensure_contig(nbuf, 2)) == NULL) {
295 1.19 rmind ok = false;
296 1.19 rmind goto done;
297 1.4 rmind }
298 1.39 maxv val = *(nptr + 1);
299 1.16 rmind if (val < 2 || val > topts_len) {
300 1.19 rmind ok = false;
301 1.19 rmind goto done;
302 1.4 rmind }
303 1.4 rmind topts_len -= val;
304 1.39 maxv step = val;
305 1.4 rmind }
306 1.12 rmind
307 1.6 rmind /* Any options left? */
308 1.4 rmind if (__predict_true(topts_len > 0)) {
309 1.4 rmind goto next;
310 1.4 rmind }
311 1.19 rmind ok = true;
312 1.19 rmind done:
313 1.19 rmind if (nbuf_flag_p(nbuf, NBUF_DATAREF_RESET)) {
314 1.32 rmind npf_recache(npc);
315 1.19 rmind }
316 1.19 rmind return ok;
317 1.1 rmind }
318 1.1 rmind
319 1.19 rmind static int
320 1.19 rmind npf_cache_ip(npf_cache_t *npc, nbuf_t *nbuf)
321 1.1 rmind {
322 1.19 rmind const void *nptr = nbuf_dataptr(nbuf);
323 1.19 rmind const uint8_t ver = *(const uint8_t *)nptr;
324 1.19 rmind int flags = 0;
325 1.12 rmind
326 1.43 maxv /*
327 1.43 maxv * We intentionally don't read the L4 payload after IPPROTO_AH.
328 1.43 maxv */
329 1.43 maxv
330 1.4 rmind switch (ver >> 4) {
331 1.12 rmind case IPVERSION: {
332 1.19 rmind struct ip *ip;
333 1.12 rmind
334 1.19 rmind ip = nbuf_ensure_contig(nbuf, sizeof(struct ip));
335 1.19 rmind if (ip == NULL) {
336 1.38 maxv return NPC_FMTERR;
337 1.4 rmind }
338 1.12 rmind
339 1.46 maxv /* Retrieve the complete header. */
340 1.10 rmind if ((u_int)(ip->ip_hl << 2) < sizeof(struct ip)) {
341 1.38 maxv return NPC_FMTERR;
342 1.4 rmind }
343 1.46 maxv ip = nbuf_ensure_contig(nbuf, (u_int)(ip->ip_hl << 2));
344 1.46 maxv if (ip == NULL) {
345 1.46 maxv return NPC_FMTERR;
346 1.46 maxv }
347 1.46 maxv
348 1.4 rmind if (ip->ip_off & ~htons(IP_DF | IP_RF)) {
349 1.4 rmind /* Note fragmentation. */
350 1.19 rmind flags |= NPC_IPFRAG;
351 1.4 rmind }
352 1.12 rmind
353 1.4 rmind /* Cache: layer 3 - IPv4. */
354 1.14 rmind npc->npc_alen = sizeof(struct in_addr);
355 1.28 rmind npc->npc_ips[NPF_SRC] = (npf_addr_t *)&ip->ip_src;
356 1.28 rmind npc->npc_ips[NPF_DST] = (npf_addr_t *)&ip->ip_dst;
357 1.7 zoltan npc->npc_hlen = ip->ip_hl << 2;
358 1.19 rmind npc->npc_proto = ip->ip_p;
359 1.19 rmind
360 1.19 rmind npc->npc_ip.v4 = ip;
361 1.19 rmind flags |= NPC_IP4;
362 1.4 rmind break;
363 1.12 rmind }
364 1.4 rmind
365 1.12 rmind case (IPV6_VERSION >> 4): {
366 1.19 rmind struct ip6_hdr *ip6;
367 1.19 rmind struct ip6_ext *ip6e;
368 1.37 christos struct ip6_frag *ip6f;
369 1.19 rmind size_t off, hlen;
370 1.38 maxv int frag_present;
371 1.19 rmind
372 1.19 rmind ip6 = nbuf_ensure_contig(nbuf, sizeof(struct ip6_hdr));
373 1.19 rmind if (ip6 == NULL) {
374 1.38 maxv return NPC_FMTERR;
375 1.7 zoltan }
376 1.19 rmind
377 1.44 maxv /*
378 1.44 maxv * XXX: We don't handle IPv6 Jumbograms.
379 1.44 maxv */
380 1.44 maxv
381 1.19 rmind /* Set initial next-protocol value. */
382 1.19 rmind hlen = sizeof(struct ip6_hdr);
383 1.19 rmind npc->npc_proto = ip6->ip6_nxt;
384 1.13 rmind npc->npc_hlen = hlen;
385 1.7 zoltan
386 1.38 maxv frag_present = 0;
387 1.38 maxv
388 1.12 rmind /*
389 1.19 rmind * Advance by the length of the current header.
390 1.12 rmind */
391 1.19 rmind off = nbuf_offset(nbuf);
392 1.38 maxv while ((ip6e = nbuf_advance(nbuf, hlen, sizeof(*ip6e))) != NULL) {
393 1.13 rmind /*
394 1.13 rmind * Determine whether we are going to continue.
395 1.13 rmind */
396 1.19 rmind switch (npc->npc_proto) {
397 1.13 rmind case IPPROTO_HOPOPTS:
398 1.7 zoltan case IPPROTO_DSTOPTS:
399 1.7 zoltan case IPPROTO_ROUTING:
400 1.19 rmind hlen = (ip6e->ip6e_len + 1) << 3;
401 1.7 zoltan break;
402 1.7 zoltan case IPPROTO_FRAGMENT:
403 1.38 maxv if (frag_present++)
404 1.38 maxv return NPC_FMTERR;
405 1.37 christos ip6f = nbuf_ensure_contig(nbuf, sizeof(*ip6f));
406 1.37 christos if (ip6f == NULL)
407 1.38 maxv return NPC_FMTERR;
408 1.38 maxv
409 1.41 maxv /* RFC6946: Skip dummy fragments. */
410 1.41 maxv if (!ntohs(ip6f->ip6f_offlg & IP6F_OFF_MASK) &&
411 1.41 maxv !(ip6f->ip6f_offlg & IP6F_MORE_FRAG)) {
412 1.41 maxv hlen = sizeof(struct ip6_frag);
413 1.41 maxv break;
414 1.41 maxv }
415 1.41 maxv
416 1.40 maxv hlen = 0;
417 1.40 maxv flags |= NPC_IPFRAG;
418 1.37 christos
419 1.7 zoltan break;
420 1.7 zoltan default:
421 1.13 rmind hlen = 0;
422 1.13 rmind break;
423 1.13 rmind }
424 1.13 rmind
425 1.13 rmind if (!hlen) {
426 1.7 zoltan break;
427 1.7 zoltan }
428 1.19 rmind npc->npc_proto = ip6e->ip6e_nxt;
429 1.13 rmind npc->npc_hlen += hlen;
430 1.13 rmind }
431 1.7 zoltan
432 1.46 maxv if (ip6e == NULL) {
433 1.46 maxv return NPC_FMTERR;
434 1.46 maxv }
435 1.46 maxv
436 1.23 rmind /*
437 1.23 rmind * Re-fetch the header pointers (nbufs might have been
438 1.23 rmind * reallocated). Restore the original offset (if any).
439 1.23 rmind */
440 1.19 rmind nbuf_reset(nbuf);
441 1.23 rmind ip6 = nbuf_dataptr(nbuf);
442 1.19 rmind if (off) {
443 1.19 rmind nbuf_advance(nbuf, off, 0);
444 1.19 rmind }
445 1.19 rmind
446 1.12 rmind /* Cache: layer 3 - IPv6. */
447 1.14 rmind npc->npc_alen = sizeof(struct in6_addr);
448 1.28 rmind npc->npc_ips[NPF_SRC] = (npf_addr_t *)&ip6->ip6_src;
449 1.44 maxv npc->npc_ips[NPF_DST] = (npf_addr_t *)&ip6->ip6_dst;
450 1.19 rmind
451 1.19 rmind npc->npc_ip.v6 = ip6;
452 1.19 rmind flags |= NPC_IP6;
453 1.7 zoltan break;
454 1.12 rmind }
455 1.4 rmind default:
456 1.19 rmind break;
457 1.4 rmind }
458 1.19 rmind return flags;
459 1.1 rmind }
460 1.1 rmind
461 1.1 rmind /*
462 1.4 rmind * npf_cache_all: general routine to cache all relevant IP (v4 or v6)
463 1.12 rmind * and TCP, UDP or ICMP headers.
464 1.19 rmind *
465 1.19 rmind * => nbuf offset shall be set accordingly.
466 1.1 rmind */
467 1.10 rmind int
468 1.32 rmind npf_cache_all(npf_cache_t *npc)
469 1.1 rmind {
470 1.32 rmind nbuf_t *nbuf = npc->npc_nbuf;
471 1.19 rmind int flags, l4flags;
472 1.19 rmind u_int hlen;
473 1.19 rmind
474 1.19 rmind /*
475 1.19 rmind * This routine is a main point where the references are cached,
476 1.19 rmind * therefore clear the flag as we reset.
477 1.19 rmind */
478 1.19 rmind again:
479 1.19 rmind nbuf_unset_flag(nbuf, NBUF_DATAREF_RESET);
480 1.1 rmind
481 1.19 rmind /*
482 1.19 rmind * First, cache the L3 header (IPv4 or IPv6). If IP packet is
483 1.19 rmind * fragmented, then we cannot look into L4.
484 1.19 rmind */
485 1.19 rmind flags = npf_cache_ip(npc, nbuf);
486 1.38 maxv if ((flags & NPC_IP46) == 0 || (flags & NPC_IPFRAG) != 0 ||
487 1.38 maxv (flags & NPC_FMTERR) != 0) {
488 1.47 maxv goto out;
489 1.1 rmind }
490 1.19 rmind hlen = npc->npc_hlen;
491 1.19 rmind
492 1.45 maxv /*
493 1.45 maxv * Note: we guarantee that the potential "Query Id" field of the
494 1.45 maxv * ICMPv4/ICMPv6 packets is in the nbuf. This field is used in the
495 1.45 maxv * ICMP ALG.
496 1.45 maxv */
497 1.19 rmind switch (npc->npc_proto) {
498 1.1 rmind case IPPROTO_TCP:
499 1.19 rmind /* Cache: layer 4 - TCP. */
500 1.19 rmind npc->npc_l4.tcp = nbuf_advance(nbuf, hlen,
501 1.19 rmind sizeof(struct tcphdr));
502 1.19 rmind l4flags = NPC_LAYER4 | NPC_TCP;
503 1.10 rmind break;
504 1.1 rmind case IPPROTO_UDP:
505 1.19 rmind /* Cache: layer 4 - UDP. */
506 1.19 rmind npc->npc_l4.udp = nbuf_advance(nbuf, hlen,
507 1.19 rmind sizeof(struct udphdr));
508 1.19 rmind l4flags = NPC_LAYER4 | NPC_UDP;
509 1.10 rmind break;
510 1.1 rmind case IPPROTO_ICMP:
511 1.19 rmind /* Cache: layer 4 - ICMPv4. */
512 1.19 rmind npc->npc_l4.icmp = nbuf_advance(nbuf, hlen,
513 1.45 maxv ICMP_MINLEN);
514 1.19 rmind l4flags = NPC_LAYER4 | NPC_ICMP;
515 1.19 rmind break;
516 1.15 spz case IPPROTO_ICMPV6:
517 1.19 rmind /* Cache: layer 4 - ICMPv6. */
518 1.19 rmind npc->npc_l4.icmp6 = nbuf_advance(nbuf, hlen,
519 1.45 maxv sizeof(struct icmp6_hdr));
520 1.19 rmind l4flags = NPC_LAYER4 | NPC_ICMP;
521 1.19 rmind break;
522 1.19 rmind default:
523 1.19 rmind l4flags = 0;
524 1.10 rmind break;
525 1.1 rmind }
526 1.19 rmind
527 1.47 maxv /* Error out if nbuf_advance failed. */
528 1.47 maxv if (l4flags && npc->npc_l4.hdr == NULL) {
529 1.47 maxv goto err;
530 1.47 maxv }
531 1.47 maxv
532 1.19 rmind if (nbuf_flag_p(nbuf, NBUF_DATAREF_RESET)) {
533 1.19 rmind goto again;
534 1.19 rmind }
535 1.19 rmind
536 1.47 maxv flags |= l4flags;
537 1.47 maxv npc->npc_info |= flags;
538 1.47 maxv return flags;
539 1.47 maxv
540 1.47 maxv err:
541 1.47 maxv flags = NPC_FMTERR;
542 1.47 maxv out:
543 1.47 maxv nbuf_unset_flag(nbuf, NBUF_DATAREF_RESET);
544 1.19 rmind npc->npc_info |= flags;
545 1.19 rmind return flags;
546 1.19 rmind }
547 1.19 rmind
548 1.19 rmind void
549 1.32 rmind npf_recache(npf_cache_t *npc)
550 1.19 rmind {
551 1.32 rmind nbuf_t *nbuf = npc->npc_nbuf;
552 1.24 martin const int mflags __diagused = npc->npc_info & (NPC_IP46 | NPC_LAYER4);
553 1.25 mrg int flags __diagused;
554 1.19 rmind
555 1.19 rmind nbuf_reset(nbuf);
556 1.19 rmind npc->npc_info = 0;
557 1.32 rmind flags = npf_cache_all(npc);
558 1.32 rmind
559 1.19 rmind KASSERT((flags & mflags) == mflags);
560 1.19 rmind KASSERT(nbuf_flag_p(nbuf, NBUF_DATAREF_RESET) == 0);
561 1.1 rmind }
562 1.1 rmind
563 1.1 rmind /*
564 1.19 rmind * npf_rwrip: rewrite required IP address.
565 1.4 rmind */
566 1.4 rmind bool
567 1.28 rmind npf_rwrip(const npf_cache_t *npc, u_int which, const npf_addr_t *addr)
568 1.4 rmind {
569 1.4 rmind KASSERT(npf_iscached(npc, NPC_IP46));
570 1.28 rmind KASSERT(which == NPF_SRC || which == NPF_DST);
571 1.4 rmind
572 1.28 rmind memcpy(npc->npc_ips[which], addr, npc->npc_alen);
573 1.4 rmind return true;
574 1.4 rmind }
575 1.4 rmind
576 1.4 rmind /*
577 1.19 rmind * npf_rwrport: rewrite required TCP/UDP port.
578 1.1 rmind */
579 1.1 rmind bool
580 1.28 rmind npf_rwrport(const npf_cache_t *npc, u_int which, const in_port_t port)
581 1.1 rmind {
582 1.21 rmind const int proto = npc->npc_proto;
583 1.4 rmind in_port_t *oport;
584 1.1 rmind
585 1.4 rmind KASSERT(npf_iscached(npc, NPC_TCP) || npf_iscached(npc, NPC_UDP));
586 1.1 rmind KASSERT(proto == IPPROTO_TCP || proto == IPPROTO_UDP);
587 1.28 rmind KASSERT(which == NPF_SRC || which == NPF_DST);
588 1.1 rmind
589 1.19 rmind /* Get the offset and store the port in it. */
590 1.4 rmind if (proto == IPPROTO_TCP) {
591 1.19 rmind struct tcphdr *th = npc->npc_l4.tcp;
592 1.28 rmind oport = (which == NPF_SRC) ? &th->th_sport : &th->th_dport;
593 1.1 rmind } else {
594 1.19 rmind struct udphdr *uh = npc->npc_l4.udp;
595 1.28 rmind oport = (which == NPF_SRC) ? &uh->uh_sport : &uh->uh_dport;
596 1.1 rmind }
597 1.19 rmind memcpy(oport, &port, sizeof(in_port_t));
598 1.1 rmind return true;
599 1.1 rmind }
600 1.1 rmind
601 1.1 rmind /*
602 1.19 rmind * npf_rwrcksum: rewrite IPv4 and/or TCP/UDP checksum.
603 1.1 rmind */
604 1.1 rmind bool
605 1.28 rmind npf_rwrcksum(const npf_cache_t *npc, u_int which,
606 1.19 rmind const npf_addr_t *addr, const in_port_t port)
607 1.1 rmind {
608 1.28 rmind const npf_addr_t *oaddr = npc->npc_ips[which];
609 1.21 rmind const int proto = npc->npc_proto;
610 1.19 rmind const int alen = npc->npc_alen;
611 1.18 rmind uint16_t *ocksum;
612 1.18 rmind in_port_t oport;
613 1.18 rmind
614 1.19 rmind KASSERT(npf_iscached(npc, NPC_LAYER4));
615 1.28 rmind KASSERT(which == NPF_SRC || which == NPF_DST);
616 1.18 rmind
617 1.4 rmind if (npf_iscached(npc, NPC_IP4)) {
618 1.19 rmind struct ip *ip = npc->npc_ip.v4;
619 1.19 rmind uint16_t ipsum = ip->ip_sum;
620 1.4 rmind
621 1.19 rmind /* Recalculate IPv4 checksum and rewrite. */
622 1.19 rmind ip->ip_sum = npf_addr_cksum(ipsum, alen, oaddr, addr);
623 1.4 rmind } else {
624 1.4 rmind /* No checksum for IPv6. */
625 1.4 rmind KASSERT(npf_iscached(npc, NPC_IP6));
626 1.4 rmind }
627 1.4 rmind
628 1.18 rmind /* Nothing else to do for ICMP. */
629 1.30 rmind if (proto == IPPROTO_ICMP || proto == IPPROTO_ICMPV6) {
630 1.4 rmind return true;
631 1.4 rmind }
632 1.7 zoltan KASSERT(npf_iscached(npc, NPC_TCP) || npf_iscached(npc, NPC_UDP));
633 1.4 rmind
634 1.18 rmind /*
635 1.18 rmind * Calculate TCP/UDP checksum:
636 1.18 rmind * - Skip if UDP and the current checksum is zero.
637 1.18 rmind * - Fixup the IP address change.
638 1.18 rmind * - Fixup the port change, if required (non-zero).
639 1.18 rmind */
640 1.4 rmind if (proto == IPPROTO_TCP) {
641 1.19 rmind struct tcphdr *th = npc->npc_l4.tcp;
642 1.4 rmind
643 1.18 rmind ocksum = &th->th_sum;
644 1.28 rmind oport = (which == NPF_SRC) ? th->th_sport : th->th_dport;
645 1.4 rmind } else {
646 1.19 rmind struct udphdr *uh = npc->npc_l4.udp;
647 1.4 rmind
648 1.4 rmind KASSERT(proto == IPPROTO_UDP);
649 1.18 rmind ocksum = &uh->uh_sum;
650 1.18 rmind if (*ocksum == 0) {
651 1.4 rmind /* No need to update. */
652 1.4 rmind return true;
653 1.4 rmind }
654 1.28 rmind oport = (which == NPF_SRC) ? uh->uh_sport : uh->uh_dport;
655 1.18 rmind }
656 1.18 rmind
657 1.19 rmind uint16_t cksum = npf_addr_cksum(*ocksum, alen, oaddr, addr);
658 1.18 rmind if (port) {
659 1.18 rmind cksum = npf_fixup16_cksum(cksum, oport, port);
660 1.4 rmind }
661 1.1 rmind
662 1.19 rmind /* Rewrite TCP/UDP checksum. */
663 1.19 rmind memcpy(ocksum, &cksum, sizeof(uint16_t));
664 1.4 rmind return true;
665 1.4 rmind }
666 1.4 rmind
667 1.29 rmind /*
668 1.30 rmind * npf_napt_rwr: perform address and/or port translation.
669 1.30 rmind */
670 1.30 rmind int
671 1.30 rmind npf_napt_rwr(const npf_cache_t *npc, u_int which,
672 1.30 rmind const npf_addr_t *addr, const in_addr_t port)
673 1.30 rmind {
674 1.30 rmind const unsigned proto = npc->npc_proto;
675 1.30 rmind
676 1.30 rmind /*
677 1.30 rmind * Rewrite IP and/or TCP/UDP checksums first, since we need the
678 1.30 rmind * current (old) address/port for the calculations. Then perform
679 1.30 rmind * the address translation i.e. rewrite source or destination.
680 1.30 rmind */
681 1.30 rmind if (!npf_rwrcksum(npc, which, addr, port)) {
682 1.30 rmind return EINVAL;
683 1.30 rmind }
684 1.30 rmind if (!npf_rwrip(npc, which, addr)) {
685 1.30 rmind return EINVAL;
686 1.30 rmind }
687 1.30 rmind if (port == 0) {
688 1.30 rmind /* Done. */
689 1.30 rmind return 0;
690 1.30 rmind }
691 1.30 rmind
692 1.30 rmind switch (proto) {
693 1.30 rmind case IPPROTO_TCP:
694 1.30 rmind case IPPROTO_UDP:
695 1.30 rmind /* Rewrite source/destination port. */
696 1.30 rmind if (!npf_rwrport(npc, which, port)) {
697 1.30 rmind return EINVAL;
698 1.30 rmind }
699 1.30 rmind break;
700 1.30 rmind case IPPROTO_ICMP:
701 1.30 rmind case IPPROTO_ICMPV6:
702 1.30 rmind KASSERT(npf_iscached(npc, NPC_ICMP));
703 1.30 rmind /* Nothing. */
704 1.30 rmind break;
705 1.30 rmind default:
706 1.30 rmind return ENOTSUP;
707 1.30 rmind }
708 1.30 rmind return 0;
709 1.30 rmind }
710 1.30 rmind
711 1.30 rmind /*
712 1.29 rmind * IPv6-to-IPv6 Network Prefix Translation (NPTv6), as per RFC 6296.
713 1.29 rmind */
714 1.29 rmind
715 1.29 rmind int
716 1.29 rmind npf_npt66_rwr(const npf_cache_t *npc, u_int which, const npf_addr_t *pref,
717 1.29 rmind npf_netmask_t len, uint16_t adj)
718 1.29 rmind {
719 1.29 rmind npf_addr_t *addr = npc->npc_ips[which];
720 1.29 rmind unsigned remnant, word, preflen = len >> 4;
721 1.29 rmind uint32_t sum;
722 1.29 rmind
723 1.29 rmind KASSERT(which == NPF_SRC || which == NPF_DST);
724 1.29 rmind
725 1.29 rmind if (!npf_iscached(npc, NPC_IP6)) {
726 1.29 rmind return EINVAL;
727 1.29 rmind }
728 1.29 rmind if (len <= 48) {
729 1.29 rmind /*
730 1.29 rmind * The word to adjust. Cannot translate the 0xffff
731 1.29 rmind * subnet if /48 or shorter.
732 1.29 rmind */
733 1.29 rmind word = 3;
734 1.36 christos if (addr->word16[word] == 0xffff) {
735 1.29 rmind return EINVAL;
736 1.29 rmind }
737 1.29 rmind } else {
738 1.29 rmind /*
739 1.29 rmind * Also, all 0s or 1s in the host part are disallowed for
740 1.29 rmind * longer than /48 prefixes.
741 1.29 rmind */
742 1.36 christos if ((addr->word32[2] == 0 && addr->word32[3] == 0) ||
743 1.36 christos (addr->word32[2] == ~0U && addr->word32[3] == ~0U))
744 1.29 rmind return EINVAL;
745 1.29 rmind
746 1.29 rmind /* Determine the 16-bit word to adjust. */
747 1.29 rmind for (word = 4; word < 8; word++)
748 1.36 christos if (addr->word16[word] != 0xffff)
749 1.29 rmind break;
750 1.29 rmind }
751 1.29 rmind
752 1.29 rmind /* Rewrite the prefix. */
753 1.29 rmind for (unsigned i = 0; i < preflen; i++) {
754 1.36 christos addr->word16[i] = pref->word16[i];
755 1.29 rmind }
756 1.29 rmind
757 1.29 rmind /*
758 1.29 rmind * If prefix length is within a 16-bit word (not dividable by 16),
759 1.29 rmind * then prepare a mask, determine the word and adjust it.
760 1.29 rmind */
761 1.29 rmind if ((remnant = len - (preflen << 4)) != 0) {
762 1.29 rmind const uint16_t wordmask = (1U << remnant) - 1;
763 1.29 rmind const unsigned i = preflen;
764 1.29 rmind
765 1.36 christos addr->word16[i] = (pref->word16[i] & wordmask) |
766 1.36 christos (addr->word16[i] & ~wordmask);
767 1.29 rmind }
768 1.29 rmind
769 1.29 rmind /*
770 1.29 rmind * Performing 1's complement sum/difference.
771 1.29 rmind */
772 1.36 christos sum = addr->word16[word] + adj;
773 1.29 rmind while (sum >> 16) {
774 1.29 rmind sum = (sum >> 16) + (sum & 0xffff);
775 1.29 rmind }
776 1.29 rmind if (sum == 0xffff) {
777 1.29 rmind /* RFC 1071. */
778 1.29 rmind sum = 0x0000;
779 1.29 rmind }
780 1.36 christos addr->word16[word] = sum;
781 1.29 rmind return 0;
782 1.29 rmind }
783 1.29 rmind
784 1.13 rmind #if defined(DDB) || defined(_NPF_TESTING)
785 1.13 rmind
786 1.31 rmind const char *
787 1.31 rmind npf_addr_dump(const npf_addr_t *addr, int alen)
788 1.13 rmind {
789 1.31 rmind if (alen == sizeof(struct in_addr)) {
790 1.31 rmind struct in_addr ip;
791 1.31 rmind memcpy(&ip, addr, alen);
792 1.31 rmind return inet_ntoa(ip);
793 1.31 rmind }
794 1.36 christos return "[IPv6]";
795 1.13 rmind }
796 1.13 rmind
797 1.13 rmind #endif
798