1 /* $FreeBSD: src/sys/netinet/ip_encap.c,v 1.1.2.5 2003/01/23 21:06:45 sam Exp $ */ 2 /* $DragonFly: src/sys/netinet/ip_encap.c,v 1.10 2005/01/06 09:14:13 hsu Exp $ */ 3 /* $KAME: ip_encap.c,v 1.41 2001/03/15 08:35:08 itojun Exp $ */ 4 5 /* 6 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 7 * All rights reserved. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted provided that the following conditions 11 * are met: 12 * 1. Redistributions of source code must retain the above copyright 13 * notice, this list of conditions and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 3. Neither the name of the project nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 */ 33 /* 34 * My grandfather said that there's a devil inside tunnelling technology... 35 * 36 * We have surprisingly many protocols that want packets with IP protocol 37 * #4 or #41. Here's a list of protocols that want protocol #41: 38 * RFC1933 configured tunnel 39 * RFC1933 automatic tunnel 40 * RFC2401 IPsec tunnel 41 * RFC2473 IPv6 generic packet tunnelling 42 * RFC2529 6over4 tunnel 43 * mobile-ip6 (uses RFC2473) 44 * RFC3056 6to4 tunnel 45 * isatap tunnel 46 * Here's a list of protocol that want protocol #4: 47 * RFC1853 IPv4-in-IPv4 tunnelling 48 * RFC2003 IPv4 encapsulation within IPv4 49 * RFC2344 reverse tunnelling for mobile-ip4 50 * RFC2401 IPsec tunnel 51 * Well, what can I say. They impose different en/decapsulation mechanism 52 * from each other, so they need separate protocol handler. The only one 53 * we can easily determine by protocol # is IPsec, which always has 54 * AH/ESP/IPComp header right after outer IP header. 55 * 56 * So, clearly good old protosw does not work for protocol #4 and #41. 57 * The code will let you match protocol via src/dst address pair. 58 */ 59 /* XXX is M_NETADDR correct? */ 60 61 #include "opt_inet.h" 62 #include "opt_inet6.h" 63 64 #include <sys/param.h> 65 #include <sys/systm.h> 66 #include <sys/socket.h> 67 #include <sys/sockio.h> 68 #include <sys/mbuf.h> 69 #include <sys/errno.h> 70 #include <sys/protosw.h> 71 #include <sys/queue.h> 72 73 #include <net/if.h> 74 #include <net/route.h> 75 76 #include <netinet/in.h> 77 #include <netinet/in_systm.h> 78 #include <netinet/ip.h> 79 #include <netinet/ip_var.h> 80 #include <netinet/ip_encap.h> 81 82 #ifdef INET6 83 #include <netinet/ip6.h> 84 #include <netinet6/ip6_var.h> 85 #include <netinet6/ip6protosw.h> 86 #endif 87 88 #include <machine/stdarg.h> 89 90 #include <net/net_osdep.h> 91 92 #include <sys/kernel.h> 93 #include <sys/malloc.h> 94 MALLOC_DEFINE(M_NETADDR, "Export Host", "Export host address structure"); 95 96 static void encap_add (struct encaptab *); 97 static int mask_match (const struct encaptab *, const struct sockaddr *, 98 const struct sockaddr *); 99 static void encap_fillarg (struct mbuf *, const struct encaptab *); 100 101 #ifndef LIST_HEAD_INITIALIZER 102 /* rely upon BSS initialization */ 103 LIST_HEAD(, encaptab) encaptab; 104 #else 105 LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(&encaptab); 106 #endif 107 108 void (*ipip_input)(struct mbuf *, int, int); /* hook for mrouting */ 109 110 void 111 encap_init() 112 { 113 static int initialized = 0; 114 115 if (initialized) 116 return; 117 initialized++; 118 #if 0 119 /* 120 * we cannot use LIST_INIT() here, since drivers may want to call 121 * encap_attach(), on driver attach. encap_init() will be called 122 * on AF_INET{,6} initialization, which happens after driver 123 * initialization - using LIST_INIT() here can nuke encap_attach() 124 * from drivers. 125 */ 126 LIST_INIT(&encaptab); 127 #endif 128 } 129 130 #ifdef INET 131 void 132 encap4_input(struct mbuf *m, ...) 133 { 134 int off, proto; 135 struct ip *ip; 136 struct sockaddr_in s, d; 137 const struct protosw *psw; 138 struct encaptab *ep, *match; 139 int prio, matchprio; 140 __va_list ap; 141 142 __va_start(ap, m); 143 off = __va_arg(ap, int); 144 proto = __va_arg(ap, int); 145 __va_end(ap); 146 147 ip = mtod(m, struct ip *); 148 149 bzero(&s, sizeof s); 150 s.sin_family = AF_INET; 151 s.sin_len = sizeof(struct sockaddr_in); 152 s.sin_addr = ip->ip_src; 153 bzero(&d, sizeof d); 154 d.sin_family = AF_INET; 155 d.sin_len = sizeof(struct sockaddr_in); 156 d.sin_addr = ip->ip_dst; 157 158 match = NULL; 159 matchprio = 0; 160 for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) { 161 if (ep->af != AF_INET) 162 continue; 163 if (ep->proto >= 0 && ep->proto != proto) 164 continue; 165 if (ep->func) 166 prio = (*ep->func)(m, off, proto, ep->arg); 167 else { 168 /* 169 * it's inbound traffic, we need to match in reverse 170 * order 171 */ 172 prio = mask_match(ep, (struct sockaddr *)&d, 173 (struct sockaddr *)&s); 174 } 175 176 /* 177 * We prioritize the matches by using bit length of the 178 * matches. mask_match() and user-supplied matching function 179 * should return the bit length of the matches (for example, 180 * if both src/dst are matched for IPv4, 64 should be returned). 181 * 0 or negative return value means "it did not match". 182 * 183 * The question is, since we have two "mask" portion, we 184 * cannot really define total order between entries. 185 * For example, which of these should be preferred? 186 * mask_match() returns 48 (32 + 16) for both of them. 187 * src=3ffe::/16, dst=3ffe:501::/32 188 * src=3ffe:501::/32, dst=3ffe::/16 189 * 190 * We need to loop through all the possible candidates 191 * to get the best match - the search takes O(n) for 192 * n attachments (i.e. interfaces). 193 */ 194 if (prio <= 0) 195 continue; 196 if (prio > matchprio) { 197 matchprio = prio; 198 match = ep; 199 } 200 } 201 202 if (match) { 203 /* found a match, "match" has the best one */ 204 psw = match->psw; 205 if (psw && psw->pr_input) { 206 encap_fillarg(m, match); 207 (*psw->pr_input)(m, off, proto); 208 } else 209 m_freem(m); 210 return; 211 } 212 213 /* for backward compatibility */ 214 if (proto == IPPROTO_IPV4 && ipip_input) { 215 ipip_input(m, off, proto); 216 return; 217 } 218 219 /* last resort: inject to raw socket */ 220 rip_input(m, off, proto); 221 } 222 #endif 223 224 #ifdef INET6 225 int 226 encap6_input(mp, offp, proto) 227 struct mbuf **mp; 228 int *offp; 229 int proto; 230 { 231 struct mbuf *m = *mp; 232 struct ip6_hdr *ip6; 233 struct sockaddr_in6 s, d; 234 const struct ip6protosw *psw; 235 struct encaptab *ep, *match; 236 int prio, matchprio; 237 238 ip6 = mtod(m, struct ip6_hdr *); 239 240 bzero(&s, sizeof s); 241 s.sin6_family = AF_INET6; 242 s.sin6_len = sizeof(struct sockaddr_in6); 243 s.sin6_addr = ip6->ip6_src; 244 bzero(&d, sizeof d); 245 d.sin6_family = AF_INET6; 246 d.sin6_len = sizeof(struct sockaddr_in6); 247 d.sin6_addr = ip6->ip6_dst; 248 249 match = NULL; 250 matchprio = 0; 251 for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) { 252 if (ep->af != AF_INET6) 253 continue; 254 if (ep->proto >= 0 && ep->proto != proto) 255 continue; 256 if (ep->func) 257 prio = (*ep->func)(m, *offp, proto, ep->arg); 258 else { 259 /* 260 * it's inbound traffic, we need to match in reverse 261 * order 262 */ 263 prio = mask_match(ep, (struct sockaddr *)&d, 264 (struct sockaddr *)&s); 265 } 266 267 /* see encap4_input() for issues here */ 268 if (prio <= 0) 269 continue; 270 if (prio > matchprio) { 271 matchprio = prio; 272 match = ep; 273 } 274 } 275 276 if (match) { 277 /* found a match */ 278 psw = (const struct ip6protosw *)match->psw; 279 if (psw && psw->pr_input) { 280 encap_fillarg(m, match); 281 return (*psw->pr_input)(mp, offp, proto); 282 } else { 283 m_freem(m); 284 return IPPROTO_DONE; 285 } 286 } 287 288 /* last resort: inject to raw socket */ 289 return rip6_input(mp, offp, proto); 290 } 291 #endif 292 293 static void 294 encap_add(ep) 295 struct encaptab *ep; 296 { 297 298 LIST_INSERT_HEAD(&encaptab, ep, chain); 299 } 300 301 /* 302 * sp (src ptr) is always my side, and dp (dst ptr) is always remote side. 303 * length of mask (sm and dm) is assumed to be same as sp/dp. 304 * Return value will be necessary as input (cookie) for encap_detach(). 305 */ 306 const struct encaptab * 307 encap_attach(af, proto, sp, sm, dp, dm, psw, arg) 308 int af; 309 int proto; 310 const struct sockaddr *sp, *sm; 311 const struct sockaddr *dp, *dm; 312 const struct protosw *psw; 313 void *arg; 314 { 315 struct encaptab *ep; 316 int error; 317 int s; 318 319 s = splnet(); 320 /* sanity check on args */ 321 if (sp->sa_len > sizeof ep->src || dp->sa_len > sizeof ep->dst) { 322 error = EINVAL; 323 goto fail; 324 } 325 if (sp->sa_len != dp->sa_len) { 326 error = EINVAL; 327 goto fail; 328 } 329 if (af != sp->sa_family || af != dp->sa_family) { 330 error = EINVAL; 331 goto fail; 332 } 333 334 /* check if anyone have already attached with exactly same config */ 335 for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) { 336 if (ep->af != af) 337 continue; 338 if (ep->proto != proto) 339 continue; 340 if (ep->src.ss_len != sp->sa_len || 341 bcmp(&ep->src, sp, sp->sa_len) != 0 || 342 bcmp(&ep->srcmask, sm, sp->sa_len) != 0) 343 continue; 344 if (ep->dst.ss_len != dp->sa_len || 345 bcmp(&ep->dst, dp, dp->sa_len) != 0 || 346 bcmp(&ep->dstmask, dm, dp->sa_len) != 0) 347 continue; 348 349 error = EEXIST; 350 goto fail; 351 } 352 353 ep = malloc(sizeof *ep, M_NETADDR, M_INTWAIT | M_ZERO | M_NULLOK); 354 if (ep == NULL) { 355 error = ENOBUFS; 356 goto fail; 357 } 358 359 ep->af = af; 360 ep->proto = proto; 361 bcopy(sp, &ep->src, sp->sa_len); 362 bcopy(sm, &ep->srcmask, sp->sa_len); 363 bcopy(dp, &ep->dst, dp->sa_len); 364 bcopy(dm, &ep->dstmask, dp->sa_len); 365 ep->psw = psw; 366 ep->arg = arg; 367 368 encap_add(ep); 369 370 error = 0; 371 splx(s); 372 return ep; 373 374 fail: 375 splx(s); 376 return NULL; 377 } 378 379 const struct encaptab * 380 encap_attach_func(af, proto, func, psw, arg) 381 int af; 382 int proto; 383 int (*func) (const struct mbuf *, int, int, void *); 384 const struct protosw *psw; 385 void *arg; 386 { 387 struct encaptab *ep; 388 int error; 389 int s; 390 391 s = splnet(); 392 /* sanity check on args */ 393 if (!func) { 394 error = EINVAL; 395 goto fail; 396 } 397 398 ep = malloc(sizeof *ep, M_NETADDR, M_INTWAIT | M_ZERO | M_NULLOK); 399 if (ep == NULL) { 400 error = ENOBUFS; 401 goto fail; 402 } 403 404 ep->af = af; 405 ep->proto = proto; 406 ep->func = func; 407 ep->psw = psw; 408 ep->arg = arg; 409 410 encap_add(ep); 411 412 error = 0; 413 splx(s); 414 return ep; 415 416 fail: 417 splx(s); 418 return NULL; 419 } 420 421 int 422 encap_detach(cookie) 423 const struct encaptab *cookie; 424 { 425 const struct encaptab *ep = cookie; 426 struct encaptab *p; 427 428 for (p = LIST_FIRST(&encaptab); p; p = LIST_NEXT(p, chain)) { 429 if (p == ep) { 430 LIST_REMOVE(p, chain); 431 free(p, M_NETADDR); /*XXX*/ 432 return 0; 433 } 434 } 435 436 return EINVAL; 437 } 438 439 static int 440 mask_match(ep, sp, dp) 441 const struct encaptab *ep; 442 const struct sockaddr *sp; 443 const struct sockaddr *dp; 444 { 445 struct sockaddr_storage s; 446 struct sockaddr_storage d; 447 int i; 448 const u_int8_t *p, *q; 449 u_int8_t *r; 450 int matchlen; 451 452 if (sp->sa_len > sizeof s || dp->sa_len > sizeof d) 453 return 0; 454 if (sp->sa_family != ep->af || dp->sa_family != ep->af) 455 return 0; 456 if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len) 457 return 0; 458 459 matchlen = 0; 460 461 p = (const u_int8_t *)sp; 462 q = (const u_int8_t *)&ep->srcmask; 463 r = (u_int8_t *)&s; 464 for (i = 0 ; i < sp->sa_len; i++) { 465 r[i] = p[i] & q[i]; 466 /* XXX estimate */ 467 matchlen += (q[i] ? 8 : 0); 468 } 469 470 p = (const u_int8_t *)dp; 471 q = (const u_int8_t *)&ep->dstmask; 472 r = (u_int8_t *)&d; 473 for (i = 0 ; i < dp->sa_len; i++) { 474 r[i] = p[i] & q[i]; 475 /* XXX rough estimate */ 476 matchlen += (q[i] ? 8 : 0); 477 } 478 479 /* need to overwrite len/family portion as we don't compare them */ 480 s.ss_len = sp->sa_len; 481 s.ss_family = sp->sa_family; 482 d.ss_len = dp->sa_len; 483 d.ss_family = dp->sa_family; 484 485 if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 && 486 bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) { 487 return matchlen; 488 } else 489 return 0; 490 } 491 492 static void 493 encap_fillarg(m, ep) 494 struct mbuf *m; 495 const struct encaptab *ep; 496 { 497 struct m_tag *tag; 498 499 tag = m_tag_get(PACKET_TAG_ENCAP, sizeof(void *), MB_DONTWAIT); 500 if (tag != NULL) { 501 *(void **)(tag + 1) = ep->arg; 502 m_tag_prepend(m, tag); 503 } 504 } 505 506 void * 507 encap_getarg(m) 508 struct mbuf *m; 509 { 510 void *p = NULL; 511 struct m_tag *tag; 512 513 tag = m_tag_find(m, PACKET_TAG_ENCAP, NULL); 514 if (tag != NULL) { 515 p = (void *)(tag + 1); 516 m_tag_delete(m, tag); 517 } 518 return p; 519 } 520