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 1.15 2006/09/05 00:55:48 dillon 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 #include <sys/thread2.h> 95 MALLOC_DEFINE(M_NETADDR, "Export Host", "Export host address structure"); 96 97 static void encap_add (struct encaptab *); 98 static int mask_match (const struct encaptab *, const struct sockaddr *, 99 const struct sockaddr *); 100 static void encap_fillarg (struct mbuf *, const struct encaptab *); 101 102 #ifndef LIST_HEAD_INITIALIZER 103 /* rely upon BSS initialization */ 104 LIST_HEAD(, encaptab) encaptab; 105 #else 106 LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(&encaptab); 107 #endif 108 109 int (*ipip_input)(struct mbuf **, int *, int); /* hook for mrouting */ 110 111 void 112 encap_init(void) 113 { 114 static int initialized = 0; 115 116 if (initialized) 117 return; 118 initialized++; 119 #if 0 120 /* 121 * we cannot use LIST_INIT() here, since drivers may want to call 122 * encap_attach(), on driver attach. encap_init() will be called 123 * on AF_INET{,6} initialization, which happens after driver 124 * initialization - using LIST_INIT() here can nuke encap_attach() 125 * from drivers. 126 */ 127 LIST_INIT(&encaptab); 128 #endif 129 } 130 131 #ifdef INET 132 int 133 encap4_input(struct mbuf **mp, int *offp, int proto) 134 { 135 struct mbuf *m = *mp; 136 int off = *offp; 137 struct ip *ip; 138 struct sockaddr_in s, d; 139 const struct protosw *psw; 140 struct encaptab *ep, *match; 141 int prio, matchprio; 142 143 ip = mtod(m, struct ip *); 144 *mp = NULL; 145 146 bzero(&s, sizeof s); 147 s.sin_family = AF_INET; 148 s.sin_len = sizeof(struct sockaddr_in); 149 s.sin_addr = ip->ip_src; 150 bzero(&d, sizeof d); 151 d.sin_family = AF_INET; 152 d.sin_len = sizeof(struct sockaddr_in); 153 d.sin_addr = ip->ip_dst; 154 155 match = NULL; 156 matchprio = 0; 157 for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) { 158 if (ep->af != AF_INET) 159 continue; 160 if (ep->proto >= 0 && ep->proto != proto) 161 continue; 162 if (ep->func) 163 prio = (*ep->func)(m, off, proto, ep->arg); 164 else { 165 /* 166 * it's inbound traffic, we need to match in reverse 167 * order 168 */ 169 prio = mask_match(ep, 170 (struct sockaddr *)&d, 171 (struct sockaddr *)&s); 172 } 173 174 /* 175 * We prioritize the matches by using bit length of the 176 * matches. mask_match() and user-supplied matching function 177 * should return the bit length of the matches (for example, 178 * if both src/dst are matched for IPv4, 64 should be returned). 179 * 0 or negative return value means "it did not match". 180 * 181 * The question is, since we have two "mask" portion, we 182 * cannot really define total order between entries. 183 * For example, which of these should be preferred? 184 * mask_match() returns 48 (32 + 16) for both of them. 185 * src=3ffe::/16, dst=3ffe:501::/32 186 * src=3ffe:501::/32, dst=3ffe::/16 187 * 188 * We need to loop through all the possible candidates 189 * to get the best match - the search takes O(n) for 190 * n attachments (i.e. interfaces). 191 */ 192 if (prio <= 0) 193 continue; 194 if (prio > matchprio) { 195 matchprio = prio; 196 match = ep; 197 } 198 } 199 200 if (match) { 201 /* found a match, "match" has the best one */ 202 psw = match->psw; 203 if (psw && psw->pr_input) { 204 encap_fillarg(m, match); 205 *mp = m; 206 (*psw->pr_input)(mp, offp, proto); 207 } else { 208 m_freem(m); 209 } 210 return(IPPROTO_DONE); 211 } 212 213 /* for backward compatibility */ 214 if (proto == IPPROTO_IPV4 && ipip_input) { 215 *mp = m; 216 ipip_input(mp, offp, proto); 217 return(IPPROTO_DONE); 218 } 219 220 /* last resort: inject to raw socket */ 221 *mp = m; 222 rip_input(mp, offp, proto); 223 return(IPPROTO_DONE); 224 } 225 #endif 226 227 #ifdef INET6 228 int 229 encap6_input(struct mbuf **mp, int *offp, int proto) 230 { 231 struct mbuf *m = *mp; 232 struct ip6_hdr *ip6; 233 struct sockaddr_in6 s, d; 234 const struct protosw *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 = 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(struct encaptab *ep) 295 { 296 297 LIST_INSERT_HEAD(&encaptab, ep, chain); 298 } 299 300 /* 301 * sp (src ptr) is always my side, and dp (dst ptr) is always remote side. 302 * length of mask (sm and dm) is assumed to be same as sp/dp. 303 * Return value will be necessary as input (cookie) for encap_detach(). 304 */ 305 const struct encaptab * 306 encap_attach(int af, int proto, const struct sockaddr *sp, 307 const struct sockaddr *sm, const struct sockaddr *dp, 308 const struct sockaddr *dm, const struct protosw *psw, void *arg) 309 { 310 struct encaptab *ep; 311 int error; 312 313 crit_enter(); 314 /* sanity check on args */ 315 if (sp->sa_len > sizeof ep->src || dp->sa_len > sizeof ep->dst) { 316 error = EINVAL; 317 goto fail; 318 } 319 if (sp->sa_len != dp->sa_len) { 320 error = EINVAL; 321 goto fail; 322 } 323 if (af != sp->sa_family || af != dp->sa_family) { 324 error = EINVAL; 325 goto fail; 326 } 327 328 /* check if anyone have already attached with exactly same config */ 329 for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) { 330 if (ep->af != af) 331 continue; 332 if (ep->proto != proto) 333 continue; 334 if (ep->src.ss_len != sp->sa_len || 335 bcmp(&ep->src, sp, sp->sa_len) != 0 || 336 bcmp(&ep->srcmask, sm, sp->sa_len) != 0) 337 continue; 338 if (ep->dst.ss_len != dp->sa_len || 339 bcmp(&ep->dst, dp, dp->sa_len) != 0 || 340 bcmp(&ep->dstmask, dm, dp->sa_len) != 0) 341 continue; 342 343 error = EEXIST; 344 goto fail; 345 } 346 347 ep = kmalloc(sizeof *ep, M_NETADDR, M_INTWAIT | M_ZERO | M_NULLOK); 348 if (ep == NULL) { 349 error = ENOBUFS; 350 goto fail; 351 } 352 353 ep->af = af; 354 ep->proto = proto; 355 bcopy(sp, &ep->src, sp->sa_len); 356 bcopy(sm, &ep->srcmask, sp->sa_len); 357 bcopy(dp, &ep->dst, dp->sa_len); 358 bcopy(dm, &ep->dstmask, dp->sa_len); 359 ep->psw = psw; 360 ep->arg = arg; 361 362 encap_add(ep); 363 364 error = 0; 365 crit_exit(); 366 return ep; 367 368 fail: 369 crit_exit(); 370 return NULL; 371 } 372 373 const struct encaptab * 374 encap_attach_func(int af, int proto, 375 int (*func)(const struct mbuf *, int, int, void *), 376 const struct protosw *psw, void *arg) 377 { 378 struct encaptab *ep; 379 int error; 380 381 crit_enter(); 382 /* sanity check on args */ 383 if (!func) { 384 error = EINVAL; 385 goto fail; 386 } 387 388 ep = kmalloc(sizeof *ep, M_NETADDR, M_INTWAIT | M_ZERO | M_NULLOK); 389 if (ep == NULL) { 390 error = ENOBUFS; 391 goto fail; 392 } 393 394 ep->af = af; 395 ep->proto = proto; 396 ep->func = func; 397 ep->psw = psw; 398 ep->arg = arg; 399 400 encap_add(ep); 401 402 error = 0; 403 crit_exit(); 404 return ep; 405 406 fail: 407 crit_exit(); 408 return NULL; 409 } 410 411 int 412 encap_detach(const struct encaptab *cookie) 413 { 414 const struct encaptab *ep = cookie; 415 struct encaptab *p; 416 417 for (p = LIST_FIRST(&encaptab); p; p = LIST_NEXT(p, chain)) { 418 if (p == ep) { 419 LIST_REMOVE(p, chain); 420 kfree(p, M_NETADDR); /*XXX*/ 421 return 0; 422 } 423 } 424 425 return EINVAL; 426 } 427 428 static int 429 mask_match(const struct encaptab *ep, const struct sockaddr *sp, 430 const struct sockaddr *dp) 431 { 432 struct sockaddr_storage s; 433 struct sockaddr_storage d; 434 int i; 435 const u_int8_t *p, *q; 436 u_int8_t *r; 437 int matchlen; 438 439 if (sp->sa_len > sizeof s || dp->sa_len > sizeof d) 440 return 0; 441 if (sp->sa_family != ep->af || dp->sa_family != ep->af) 442 return 0; 443 if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len) 444 return 0; 445 446 matchlen = 0; 447 448 p = (const u_int8_t *)sp; 449 q = (const u_int8_t *)&ep->srcmask; 450 r = (u_int8_t *)&s; 451 for (i = 0 ; i < sp->sa_len; i++) { 452 r[i] = p[i] & q[i]; 453 /* XXX estimate */ 454 matchlen += (q[i] ? 8 : 0); 455 } 456 457 p = (const u_int8_t *)dp; 458 q = (const u_int8_t *)&ep->dstmask; 459 r = (u_int8_t *)&d; 460 for (i = 0 ; i < dp->sa_len; i++) { 461 r[i] = p[i] & q[i]; 462 /* XXX rough estimate */ 463 matchlen += (q[i] ? 8 : 0); 464 } 465 466 /* need to overwrite len/family portion as we don't compare them */ 467 s.ss_len = sp->sa_len; 468 s.ss_family = sp->sa_family; 469 d.ss_len = dp->sa_len; 470 d.ss_family = dp->sa_family; 471 472 if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 && 473 bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) { 474 return matchlen; 475 } else 476 return 0; 477 } 478 479 static void 480 encap_fillarg(struct mbuf *m, const struct encaptab *ep) 481 { 482 struct m_tag *tag; 483 484 tag = m_tag_get(PACKET_TAG_ENCAP, sizeof(void *), MB_DONTWAIT); 485 if (tag != NULL) { 486 *(void **)m_tag_data(tag) = ep->arg; 487 m_tag_prepend(m, tag); 488 } 489 } 490 491 void * 492 encap_getarg(struct mbuf *m) 493 { 494 void *p = NULL; 495 struct m_tag *tag; 496 497 tag = m_tag_find(m, PACKET_TAG_ENCAP, NULL); 498 if (tag != NULL) { 499 p = *(void **)m_tag_data(tag); 500 m_tag_delete(m, tag); 501 } 502 return p; 503 } 504