bsd/netinet/ip_encap.c

   1 /*
   2  * Copyright (c) 2000-2011 Apple Inc. All rights reserved.
   3  *
   4  * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
   5  *
   6  * This file contains Original Code and/or Modifications of Original Code
   7  * as defined in and that are subject to the Apple Public Source License
   8  * Version 2.0 (the 'License'). You may not use this file except in
   9  * compliance with the License. The rights granted to you under the License
  10  * may not be used to create, or enable the creation or redistribution of,
  11  * unlawful or unlicensed copies of an Apple operating system, or to
  12  * circumvent, violate, or enable the circumvention or violation of, any
  13  * terms of an Apple operating system software license agreement.
  14  *
  15  * Please obtain a copy of the License at
  16  * http://www.opensource.apple.com/apsl/ and read it before using this file.
  17  *
  18  * The Original Code and all software distributed under the License are
  19  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  20  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  21  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  22  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  23  * Please see the License for the specific language governing rights and
  24  * limitations under the License.
  25  *
  26  * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
  27  */
  28 /*      $FreeBSD: src/sys/netinet/ip_encap.c,v 1.1.2.2 2001/07/03 11:01:46 ume Exp $    */
  29 /*      $KAME: ip_encap.c,v 1.41 2001/03/15 08:35:08 itojun Exp $       */
  30
  31 /*
  32  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
  33  * All rights reserved.
  34  *
  35  * Redistribution and use in source and binary forms, with or without
  36  * modification, are permitted provided that the following conditions
  37  * are met:
  38  * 1. Redistributions of source code must retain the above copyright
  39  *    notice, this list of conditions and the following disclaimer.
  40  * 2. Redistributions in binary form must reproduce the above copyright
  41  *    notice, this list of conditions and the following disclaimer in the
  42  *    documentation and/or other materials provided with the distribution.
  43  * 3. Neither the name of the project nor the names of its contributors
  44  *    may be used to endorse or promote products derived from this software
  45  *    without specific prior written permission.
  46  *
  47  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
  48  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  49  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  50  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
  51  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  52  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  53  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  54  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  55  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  56  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  57  * SUCH DAMAGE.
  58  */
  59 /*
  60  * My grandfather said that there's a devil inside tunnelling technology...
  61  *
  62  * We have surprisingly many protocols that want packets with IP protocol
  63  * #4 or #41.  Here's a list of protocols that want protocol #41:
  64  *      RFC1933 configured tunnel
  65  *      RFC1933 automatic tunnel
  66  *      RFC2401 IPsec tunnel
  67  *      RFC2473 IPv6 generic packet tunnelling
  68  *      RFC2529 6over4 tunnel
  69  *      mobile-ip6 (uses RFC2473)
  70  *      6to4 tunnel
  71  * Here's a list of protocol that want protocol #4:
  72  *      RFC1853 IPv4-in-IPv4 tunnelling
  73  *      RFC2003 IPv4 encapsulation within IPv4
  74  *      RFC2344 reverse tunnelling for mobile-ip4
  75  *      RFC2401 IPsec tunnel
  76  * Well, what can I say.  They impose different en/decapsulation mechanism
  77  * from each other, so they need separate protocol handler.  The only one
  78  * we can easily determine by protocol # is IPsec, which always has
  79  * AH/ESP/IPComp header right after outer IP header.
  80  *
  81  * So, clearly good old protosw does not work for protocol #4 and #41.
  82  * The code will let you match protocol via src/dst address pair.
  83  */
  84 /* XXX is M_NETADDR correct? */
  85
  86 #include <sys/param.h>
  87 #include <sys/systm.h>
  88 #include <sys/socket.h>
  89 #include <sys/sockio.h>
  90 #include <sys/mbuf.h>
  91 #include <sys/mcache.h>
  92 #include <sys/errno.h>
  93 #include <sys/protosw.h>
  94 #include <sys/queue.h>
  95
  96 #include <net/if.h>
  97 #include <net/route.h>
  98
  99 #include <netinet/in.h>
 100 #include <netinet/in_systm.h>
 101 #include <netinet/ip.h>
 102 #include <netinet/ip_var.h>
 103 #include <netinet/ip_encap.h>
 104 #if MROUTING
 105 #include <netinet/ip_mroute.h>
 106 #endif /* MROUTING */
 107
 108 #if INET6
 109 #include <netinet/ip6.h>
 110 #include <netinet6/ip6_var.h>
 111 #include <netinet6/ip6protosw.h>
 112 #endif
 113
 114
 115 #include <net/net_osdep.h>
 116
 117 #ifndef __APPLE__
 118 #include <sys/kernel.h>
 119 #include <sys/malloc.h>
 120 MALLOC_DEFINE(M_NETADDR, "Export Host", "Export host address structure");
 121 #endif
 122
 123 static void encap_add(struct encaptab *);
 124 static int mask_match(const struct encaptab *, const struct sockaddr *,
 125                 const struct sockaddr *);
 126 static void encap_fillarg(struct mbuf *, const struct encaptab *);
 127
 128 #ifndef LIST_HEAD_INITIALIZER
 129 /* rely upon BSS initialization */
 130 LIST_HEAD(, encaptab) encaptab;
 131 #else
 132 LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(&encaptab);
 133 #endif
 134
 135 void
 136 encap_init()
 137 {
 138         static int initialized = 0;
 139
 140         if (initialized)
 141                 return;
 142         initialized++;
 143 #if 0
 144         /*
 145          * we cannot use LIST_INIT() here, since drivers may want to call
 146          * encap_attach(), on driver attach.  encap_init() will be called
 147          * on AF_INET{,6} initialization, which happens after driver
 148          * initialization - using LIST_INIT() here can nuke encap_attach()
 149          * from drivers.
 150          */
 151         LIST_INIT(&encaptab);
 152 #endif
 153 }
 154
 155 #if INET
 156 void
 157 encap4_input(m, off)
 158         struct mbuf *m;
 159         int off;
 160 {
 161         int proto;
 162         struct ip *ip;
 163         struct sockaddr_in s, d;
 164         const struct protosw *psw;
 165         struct encaptab *ep, *match;
 166         int prio, matchprio;
 167
 168 #ifndef __APPLE__
 169         va_start(ap, m);
 170         off = va_arg(ap, int);
 171         proto = va_arg(ap, int);
 172         va_end(ap);
 173 #endif
 174
 175         /* Expect 32-bit aligned data pointer on strict-align platforms */
 176         MBUF_STRICT_DATA_ALIGNMENT_CHECK_32(m);
 177
 178         ip = mtod(m, struct ip *);
 179 #ifdef __APPLE__
 180         proto = ip->ip_p;
 181 #endif
 182
 183         bzero(&s, sizeof(s));
 184         s.sin_family = AF_INET;
 185         s.sin_len = sizeof(struct sockaddr_in);
 186         s.sin_addr = ip->ip_src;
 187         bzero(&d, sizeof(d));
 188         d.sin_family = AF_INET;
 189         d.sin_len = sizeof(struct sockaddr_in);
 190         d.sin_addr = ip->ip_dst;
 191
 192         match = NULL;
 193         matchprio = 0;
 194         for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
 195                 if (ep->af != AF_INET)
 196                         continue;
 197                 if (ep->proto >= 0 && ep->proto != proto)
 198                         continue;
 199                 if (ep->func)
 200                         prio = (*ep->func)(m, off, proto, ep->arg);
 201                 else {
 202                         /*
 203                          * it's inbound traffic, we need to match in reverse
 204                          * order
 205                          */
 206                         prio = mask_match(ep, (struct sockaddr *)&d,
 207                             (struct sockaddr *)&s);
 208                 }
 209
 210                 /*
 211                  * We prioritize the matches by using bit length of the
 212                  * matches.  mask_match() and user-supplied matching function
 213                  * should return the bit length of the matches (for example,
 214                  * if both src/dst are matched for IPv4, 64 should be returned).
 215                  * 0 or negative return value means "it did not match".
 216                  *
 217                  * The question is, since we have two "mask" portion, we
 218                  * cannot really define total order between entries.
 219                  * For example, which of these should be preferred?
 220                  * mask_match() returns 48 (32 + 16) for both of them.
 221                  *      src=3ffe::/16, dst=3ffe:501::/32
 222                  *      src=3ffe:501::/32, dst=3ffe::/16
 223                  *
 224                  * We need to loop through all the possible candidates
 225                  * to get the best match - the search takes O(n) for
 226                  * n attachments (i.e. interfaces).
 227                  */
 228                 if (prio <= 0)
 229                         continue;
 230                 if (prio > matchprio) {
 231                         matchprio = prio;
 232                         match = ep;
 233                 }
 234         }
 235
 236         if (match) {
 237                 /* found a match, "match" has the best one */
 238                 psw = (const struct protosw *)match->psw;
 239                 if (psw && psw->pr_input) {
 240                         encap_fillarg(m, match);
 241                         (*psw->pr_input)(m, off);
 242                 } else
 243                         m_freem(m);
 244                 return;
 245         }
 246
 247         /* for backward compatibility */
 248 # if MROUTING
 249 #  define COMPATFUNC    ipip_input
 250 # endif /*MROUTING*/
 251
 252 #if COMPATFUNC
 253         if (proto == IPPROTO_IPV4) {
 254                 COMPATFUNC(m, off);
 255                 return;
 256         }
 257 #endif
 258
 259         /* last resort: inject to raw socket */
 260         rip_input(m, off);
 261 }
 262 #endif
 263
 264 #if INET6
 265 int
 266 encap6_input(struct mbuf **mp, int *offp, int proto)
 267 {
 268         struct mbuf *m = *mp;
 269         struct ip6_hdr *ip6;
 270         struct sockaddr_in6 s, d;
 271         const struct ip6protosw *psw;
 272         struct encaptab *ep, *match;
 273         int prio, matchprio;
 274
 275         /* Expect 32-bit aligned data pointer on strict-align platforms */
 276         MBUF_STRICT_DATA_ALIGNMENT_CHECK_32(m);
 277
 278         ip6 = mtod(m, struct ip6_hdr *);
 279         bzero(&s, sizeof(s));
 280         s.sin6_family = AF_INET6;
 281         s.sin6_len = sizeof(struct sockaddr_in6);
 282         s.sin6_addr = ip6->ip6_src;
 283         bzero(&d, sizeof(d));
 284         d.sin6_family = AF_INET6;
 285         d.sin6_len = sizeof(struct sockaddr_in6);
 286         d.sin6_addr = ip6->ip6_dst;
 287
 288         match = NULL;
 289         matchprio = 0;
 290         for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
 291                 if (ep->af != AF_INET6)
 292                         continue;
 293                 if (ep->proto >= 0 && ep->proto != proto)
 294                         continue;
 295                 if (ep->func)
 296                         prio = (*ep->func)(m, *offp, proto, ep->arg);
 297                 else {
 298                         /*
 299                          * it's inbound traffic, we need to match in reverse
 300                          * order
 301                          */
 302                         prio = mask_match(ep, (struct sockaddr *)&d,
 303                             (struct sockaddr *)&s);
 304                 }
 305
 306                 /* see encap4_input() for issues here */
 307                 if (prio <= 0)
 308                         continue;
 309                 if (prio > matchprio) {
 310                         matchprio = prio;
 311                         match = ep;
 312                 }
 313         }
 314
 315         if (match) {
 316                 /* found a match */
 317                 psw = (const struct ip6protosw *)match->psw;
 318                 if (psw && psw->pr_input) {
 319                         encap_fillarg(m, match);
 320                         return (*psw->pr_input)(mp, offp, proto);
 321                 } else {
 322                         m_freem(m);
 323                         return IPPROTO_DONE;
 324                 }
 325         }
 326
 327         /* last resort: inject to raw socket */
 328         return rip6_input(mp, offp, proto);
 329 }
 330 #endif
 331
 332 static void
 333 encap_add(ep)
 334         struct encaptab *ep;
 335 {
 336
 337         LIST_INSERT_HEAD(&encaptab, ep, chain);
 338 }
 339
 340 /*
 341  * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
 342  * length of mask (sm and dm) is assumed to be same as sp/dp.
 343  * Return value will be necessary as input (cookie) for encap_detach().
 344  */
 345 const struct encaptab *
 346 encap_attach(af, proto, sp, sm, dp, dm, psw, arg)
 347         int af;
 348         int proto;
 349         const struct sockaddr *sp, *sm;
 350         const struct sockaddr *dp, *dm;
 351         const struct protosw *psw;
 352         void *arg;
 353 {
 354         struct encaptab *ep;
 355         int error;
 356
 357         /* sanity check on args */
 358         if (sp->sa_len > sizeof(ep->src) || dp->sa_len > sizeof(ep->dst)) {
 359                 error = EINVAL;
 360                 goto fail;
 361         }
 362         if (sp->sa_len != dp->sa_len) {
 363                 error = EINVAL;
 364                 goto fail;
 365         }
 366         if (af != sp->sa_family || af != dp->sa_family) {
 367                 error = EINVAL;
 368                 goto fail;
 369         }
 370
 371         /* check if anyone have already attached with exactly same config */
 372         for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
 373                 if (ep->af != af)
 374                         continue;
 375                 if (ep->proto != proto)
 376                         continue;
 377                 if (ep->src.ss_len != sp->sa_len ||
 378                     bcmp(&ep->src, sp, sp->sa_len) != 0 ||
 379                     bcmp(&ep->srcmask, sm, sp->sa_len) != 0)
 380                         continue;
 381                 if (ep->dst.ss_len != dp->sa_len ||
 382                     bcmp(&ep->dst, dp, dp->sa_len) != 0 ||
 383                     bcmp(&ep->dstmask, dm, dp->sa_len) != 0)
 384                         continue;
 385
 386                 error = EEXIST;
 387                 goto fail;
 388         }
 389
 390         ep = _MALLOC(sizeof(*ep), M_NETADDR, M_WAITOK); /*XXX*/
 391         if (ep == NULL) {
 392                 error = ENOBUFS;
 393                 goto fail;
 394         }
 395         bzero(ep, sizeof(*ep));
 396
 397         ep->af = af;
 398         ep->proto = proto;
 399         bcopy(sp, &ep->src, sp->sa_len);
 400         bcopy(sm, &ep->srcmask, sp->sa_len);
 401         bcopy(dp, &ep->dst, dp->sa_len);
 402         bcopy(dm, &ep->dstmask, dp->sa_len);
 403         ep->psw = psw;
 404         ep->arg = arg;
 405
 406         encap_add(ep);
 407
 408         error = 0;
 409         return ep;
 410
 411 fail:
 412         return NULL;
 413 }
 414
 415 const struct encaptab *
 416 encap_attach_func(af, proto, func, psw, arg)
 417         int af;
 418         int proto;
 419         int (*func)(const struct mbuf *, int, int, void *);
 420         const struct protosw *psw;
 421         void *arg;
 422 {
 423         struct encaptab *ep;
 424         int error;
 425
 426         /* sanity check on args */
 427         if (!func) {
 428                 error = EINVAL;
 429                 goto fail;
 430         }
 431
 432         ep = _MALLOC(sizeof(*ep), M_NETADDR, M_WAITOK); /*XXX*/
 433         if (ep == NULL) {
 434                 error = ENOBUFS;
 435                 goto fail;
 436         }
 437         bzero(ep, sizeof(*ep));
 438
 439         ep->af = af;
 440         ep->proto = proto;
 441         ep->func = func;
 442         ep->psw = psw;
 443         ep->arg = arg;
 444
 445         encap_add(ep);
 446
 447         error = 0;
 448         return ep;
 449
 450 fail:
 451         return NULL;
 452 }
 453
 454 int
 455 encap_detach(cookie)
 456         const struct encaptab *cookie;
 457 {
 458         const struct encaptab *ep = cookie;
 459         struct encaptab *p;
 460
 461         for (p = LIST_FIRST(&encaptab); p; p = LIST_NEXT(p, chain)) {
 462                 if (p == ep) {
 463                         LIST_REMOVE(p, chain);
 464                         _FREE(p, M_NETADDR);    /*XXX*/
 465                         return 0;
 466                 }
 467         }
 468
 469         return EINVAL;
 470 }
 471
 472 static int
 473 mask_match(ep, sp, dp)
 474         const struct encaptab *ep;
 475         const struct sockaddr *sp;
 476         const struct sockaddr *dp;
 477 {
 478         struct sockaddr_storage s;
 479         struct sockaddr_storage d;
 480         int i;
 481         const u_int8_t *p, *q;
 482         u_int8_t *r;
 483         int matchlen;
 484
 485         if (sp->sa_len > sizeof(s) || dp->sa_len > sizeof(d))
 486                 return 0;
 487         if (sp->sa_family != ep->af || dp->sa_family != ep->af)
 488                 return 0;
 489         if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len)
 490                 return 0;
 491
 492         matchlen = 0;
 493
 494         p = (const u_int8_t *)sp;
 495         q = (const u_int8_t *)&ep->srcmask;
 496         r = (u_int8_t *)&s;
 497         for (i = 0 ; i < sp->sa_len; i++) {
 498                 r[i] = p[i] & q[i];
 499                 /* XXX estimate */
 500                 matchlen += (q[i] ? 8 : 0);
 501         }
 502
 503         p = (const u_int8_t *)dp;
 504         q = (const u_int8_t *)&ep->dstmask;
 505         r = (u_int8_t *)&d;
 506         for (i = 0 ; i < dp->sa_len; i++) {
 507                 r[i] = p[i] & q[i];
 508                 /* XXX rough estimate */
 509                 matchlen += (q[i] ? 8 : 0);
 510         }
 511
 512         /* need to overwrite len/family portion as we don't compare them */
 513         s.ss_len = sp->sa_len;
 514         s.ss_family = sp->sa_family;
 515         d.ss_len = dp->sa_len;
 516         d.ss_family = dp->sa_family;
 517
 518         if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 &&
 519             bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) {
 520                 return matchlen;
 521         } else
 522                 return 0;
 523 }
 524
 525 struct encaptabtag {
 526         void*                   *arg;
 527 };
 528
 529 static void
 530 encap_fillarg(
 531         struct mbuf *m,
 532         const struct encaptab *ep)
 533 {
 534         struct m_tag    *tag;
 535         struct encaptabtag *et;
 536
 537         tag = m_tag_create(KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_ENCAP,
 538                                           sizeof(struct encaptabtag), M_WAITOK, m);
 539
 540         if (tag != NULL) {
 541                 et = (struct encaptabtag*)(tag + 1);
 542                 et->arg = ep->arg;
 543                 m_tag_prepend(m, tag);
 544         }
 545 }
 546
 547 void *
 548 encap_getarg(m)
 549         struct mbuf *m;
 550 {
 551         struct m_tag    *tag;
 552         struct encaptabtag *et;
 553         void *p = NULL;
 554
 555         tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_ENCAP, NULL);
 556         if (tag) {
 557                 et = (struct encaptabtag*)(tag + 1);
 558                 p = et->arg;
 559                 m_tag_delete(m, tag);
 560         }
 561
 562         return p;
 563 }