bsd/netinet/ip_encap.c

   1 /*
   2  * Copyright (c) 2000,2007 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/errno.h>
  92 #include <sys/protosw.h>
  93 #include <sys/queue.h>
  94
  95 #include <net/if.h>
  96 #include <net/route.h>
  97
  98 #include <netinet/in.h>
  99 #include <netinet/in_systm.h>
 100 #include <netinet/ip.h>
 101 #include <netinet/ip_var.h>
 102 #include <netinet/ip_encap.h>
 103 #if MROUTING
 104 #include <netinet/ip_mroute.h>
 105 #endif /* MROUTING */
 106
 107 #if INET6
 108 #include <netinet/ip6.h>
 109 #include <netinet6/ip6_var.h>
 110 #include <netinet6/ip6protosw.h>
 111 #endif
 112
 113
 114 #include <net/net_osdep.h>
 115
 116 #ifndef __APPLE__
 117 #include <sys/kernel.h>
 118 #include <sys/malloc.h>
 119 MALLOC_DEFINE(M_NETADDR, "Export Host", "Export host address structure");
 120 #endif
 121
 122 static void encap_add(struct encaptab *);
 123 static int mask_match(const struct encaptab *, const struct sockaddr *,
 124                 const struct sockaddr *);
 125 static void encap_fillarg(struct mbuf *, const struct encaptab *);
 126
 127 #ifndef LIST_HEAD_INITIALIZER
 128 /* rely upon BSS initialization */
 129 LIST_HEAD(, encaptab) encaptab;
 130 #else
 131 LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(&encaptab);
 132 #endif
 133
 134 void
 135 encap_init()
 136 {
 137         static int initialized = 0;
 138
 139         if (initialized)
 140                 return;
 141         initialized++;
 142 #if 0
 143         /*
 144          * we cannot use LIST_INIT() here, since drivers may want to call
 145          * encap_attach(), on driver attach.  encap_init() will be called
 146          * on AF_INET{,6} initialization, which happens after driver
 147          * initialization - using LIST_INIT() here can nuke encap_attach()
 148          * from drivers.
 149          */
 150         LIST_INIT(&encaptab);
 151 #endif
 152 }
 153
 154 #if INET
 155 void
 156 encap4_input(m, off)
 157         struct mbuf *m;
 158         int off;
 159 {
 160         int proto;
 161         struct ip *ip;
 162         struct sockaddr_in s, d;
 163         const struct protosw *psw;
 164         struct encaptab *ep, *match;
 165         int prio, matchprio;
 166
 167 #ifndef __APPLE__
 168         va_start(ap, m);
 169         off = va_arg(ap, int);
 170         proto = va_arg(ap, int);
 171         va_end(ap);
 172 #endif
 173
 174         ip = mtod(m, struct ip *);
 175 #ifdef __APPLE__
 176         proto = ip->ip_p;
 177 #endif
 178
 179         bzero(&s, sizeof(s));
 180         s.sin_family = AF_INET;
 181         s.sin_len = sizeof(struct sockaddr_in);
 182         s.sin_addr = ip->ip_src;
 183         bzero(&d, sizeof(d));
 184         d.sin_family = AF_INET;
 185         d.sin_len = sizeof(struct sockaddr_in);
 186         d.sin_addr = ip->ip_dst;
 187
 188         match = NULL;
 189         matchprio = 0;
 190         for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
 191                 if (ep->af != AF_INET)
 192                         continue;
 193                 if (ep->proto >= 0 && ep->proto != proto)
 194                         continue;
 195                 if (ep->func)
 196                         prio = (*ep->func)(m, off, proto, ep->arg);
 197                 else {
 198                         /*
 199                          * it's inbound traffic, we need to match in reverse
 200                          * order
 201                          */
 202                         prio = mask_match(ep, (struct sockaddr *)&d,
 203                             (struct sockaddr *)&s);
 204                 }
 205
 206                 /*
 207                  * We prioritize the matches by using bit length of the
 208                  * matches.  mask_match() and user-supplied matching function
 209                  * should return the bit length of the matches (for example,
 210                  * if both src/dst are matched for IPv4, 64 should be returned).
 211                  * 0 or negative return value means "it did not match".
 212                  *
 213                  * The question is, since we have two "mask" portion, we
 214                  * cannot really define total order between entries.
 215                  * For example, which of these should be preferred?
 216                  * mask_match() returns 48 (32 + 16) for both of them.
 217                  *      src=3ffe::/16, dst=3ffe:501::/32
 218                  *      src=3ffe:501::/32, dst=3ffe::/16
 219                  *
 220                  * We need to loop through all the possible candidates
 221                  * to get the best match - the search takes O(n) for
 222                  * n attachments (i.e. interfaces).
 223                  */
 224                 if (prio <= 0)
 225                         continue;
 226                 if (prio > matchprio) {
 227                         matchprio = prio;
 228                         match = ep;
 229                 }
 230         }
 231
 232         if (match) {
 233                 /* found a match, "match" has the best one */
 234                 psw = (const struct protosw *)match->psw;
 235                 if (psw && psw->pr_input) {
 236                         encap_fillarg(m, match);
 237                         (*psw->pr_input)(m, off);
 238                 } else
 239                         m_freem(m);
 240                 return;
 241         }
 242
 243         /* for backward compatibility */
 244 # if MROUTING
 245 #  define COMPATFUNC    ipip_input
 246 # endif /*MROUTING*/
 247
 248 #if COMPATFUNC
 249         if (proto == IPPROTO_IPV4) {
 250                 COMPATFUNC(m, off);
 251                 return;
 252         }
 253 #endif
 254
 255         /* last resort: inject to raw socket */
 256         rip_input(m, off);
 257 }
 258 #endif
 259
 260 #if INET6
 261 int
 262 encap6_input(struct mbuf **mp, int *offp, int proto)
 263 {
 264         struct mbuf *m = *mp;
 265         struct ip6_hdr *ip6;
 266         struct sockaddr_in6 s, d;
 267         const struct ip6protosw *psw;
 268         struct encaptab *ep, *match;
 269         int prio, matchprio;
 270
 271         ip6 = mtod(m, struct ip6_hdr *);
 272
 273         bzero(&s, sizeof(s));
 274         s.sin6_family = AF_INET6;
 275         s.sin6_len = sizeof(struct sockaddr_in6);
 276         s.sin6_addr = ip6->ip6_src;
 277         bzero(&d, sizeof(d));
 278         d.sin6_family = AF_INET6;
 279         d.sin6_len = sizeof(struct sockaddr_in6);
 280         d.sin6_addr = ip6->ip6_dst;
 281
 282         match = NULL;
 283         matchprio = 0;
 284         for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
 285                 if (ep->af != AF_INET6)
 286                         continue;
 287                 if (ep->proto >= 0 && ep->proto != proto)
 288                         continue;
 289                 if (ep->func)
 290                         prio = (*ep->func)(m, *offp, proto, ep->arg);
 291                 else {
 292                         /*
 293                          * it's inbound traffic, we need to match in reverse
 294                          * order
 295                          */
 296                         prio = mask_match(ep, (struct sockaddr *)&d,
 297                             (struct sockaddr *)&s);
 298                 }
 299
 300                 /* see encap4_input() for issues here */
 301                 if (prio <= 0)
 302                         continue;
 303                 if (prio > matchprio) {
 304                         matchprio = prio;
 305                         match = ep;
 306                 }
 307         }
 308
 309         if (match) {
 310                 /* found a match */
 311                 psw = (const struct ip6protosw *)match->psw;
 312                 if (psw && psw->pr_input) {
 313                         encap_fillarg(m, match);
 314                         return (*psw->pr_input)(mp, offp, proto);
 315                 } else {
 316                         m_freem(m);
 317                         return IPPROTO_DONE;
 318                 }
 319         }
 320
 321         /* last resort: inject to raw socket */
 322         return rip6_input(mp, offp, proto);
 323 }
 324 #endif
 325
 326 static void
 327 encap_add(ep)
 328         struct encaptab *ep;
 329 {
 330
 331         LIST_INSERT_HEAD(&encaptab, ep, chain);
 332 }
 333
 334 /*
 335  * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
 336  * length of mask (sm and dm) is assumed to be same as sp/dp.
 337  * Return value will be necessary as input (cookie) for encap_detach().
 338  */
 339 const struct encaptab *
 340 encap_attach(af, proto, sp, sm, dp, dm, psw, arg)
 341         int af;
 342         int proto;
 343         const struct sockaddr *sp, *sm;
 344         const struct sockaddr *dp, *dm;
 345         const struct protosw *psw;
 346         void *arg;
 347 {
 348         struct encaptab *ep;
 349         int error;
 350
 351         /* sanity check on args */
 352         if (sp->sa_len > sizeof(ep->src) || dp->sa_len > sizeof(ep->dst)) {
 353                 error = EINVAL;
 354                 goto fail;
 355         }
 356         if (sp->sa_len != dp->sa_len) {
 357                 error = EINVAL;
 358                 goto fail;
 359         }
 360         if (af != sp->sa_family || af != dp->sa_family) {
 361                 error = EINVAL;
 362                 goto fail;
 363         }
 364
 365         /* check if anyone have already attached with exactly same config */
 366         for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
 367                 if (ep->af != af)
 368                         continue;
 369                 if (ep->proto != proto)
 370                         continue;
 371                 if (ep->src.ss_len != sp->sa_len ||
 372                     bcmp(&ep->src, sp, sp->sa_len) != 0 ||
 373                     bcmp(&ep->srcmask, sm, sp->sa_len) != 0)
 374                         continue;
 375                 if (ep->dst.ss_len != dp->sa_len ||
 376                     bcmp(&ep->dst, dp, dp->sa_len) != 0 ||
 377                     bcmp(&ep->dstmask, dm, dp->sa_len) != 0)
 378                         continue;
 379
 380                 error = EEXIST;
 381                 goto fail;
 382         }
 383
 384         ep = _MALLOC(sizeof(*ep), M_NETADDR, M_WAITOK); /*XXX*/
 385         if (ep == NULL) {
 386                 error = ENOBUFS;
 387                 goto fail;
 388         }
 389         bzero(ep, sizeof(*ep));
 390
 391         ep->af = af;
 392         ep->proto = proto;
 393         bcopy(sp, &ep->src, sp->sa_len);
 394         bcopy(sm, &ep->srcmask, sp->sa_len);
 395         bcopy(dp, &ep->dst, dp->sa_len);
 396         bcopy(dm, &ep->dstmask, dp->sa_len);
 397         ep->psw = psw;
 398         ep->arg = arg;
 399
 400         encap_add(ep);
 401
 402         error = 0;
 403         return ep;
 404
 405 fail:
 406         return NULL;
 407 }
 408
 409 const struct encaptab *
 410 encap_attach_func(af, proto, func, psw, arg)
 411         int af;
 412         int proto;
 413         int (*func)(const struct mbuf *, int, int, void *);
 414         const struct protosw *psw;
 415         void *arg;
 416 {
 417         struct encaptab *ep;
 418         int error;
 419
 420         /* sanity check on args */
 421         if (!func) {
 422                 error = EINVAL;
 423                 goto fail;
 424         }
 425
 426         ep = _MALLOC(sizeof(*ep), M_NETADDR, M_WAITOK); /*XXX*/
 427         if (ep == NULL) {
 428                 error = ENOBUFS;
 429                 goto fail;
 430         }
 431         bzero(ep, sizeof(*ep));
 432
 433         ep->af = af;
 434         ep->proto = proto;
 435         ep->func = func;
 436         ep->psw = psw;
 437         ep->arg = arg;
 438
 439         encap_add(ep);
 440
 441         error = 0;
 442         return ep;
 443
 444 fail:
 445         return NULL;
 446 }
 447
 448 int
 449 encap_detach(cookie)
 450         const struct encaptab *cookie;
 451 {
 452         const struct encaptab *ep = cookie;
 453         struct encaptab *p;
 454
 455         for (p = LIST_FIRST(&encaptab); p; p = LIST_NEXT(p, chain)) {
 456                 if (p == ep) {
 457                         LIST_REMOVE(p, chain);
 458                         _FREE(p, M_NETADDR);    /*XXX*/
 459                         return 0;
 460                 }
 461         }
 462
 463         return EINVAL;
 464 }
 465
 466 static int
 467 mask_match(ep, sp, dp)
 468         const struct encaptab *ep;
 469         const struct sockaddr *sp;
 470         const struct sockaddr *dp;
 471 {
 472         struct sockaddr_storage s;
 473         struct sockaddr_storage d;
 474         int i;
 475         const u_int8_t *p, *q;
 476         u_int8_t *r;
 477         int matchlen;
 478
 479         if (sp->sa_len > sizeof(s) || dp->sa_len > sizeof(d))
 480                 return 0;
 481         if (sp->sa_family != ep->af || dp->sa_family != ep->af)
 482                 return 0;
 483         if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len)
 484                 return 0;
 485
 486         matchlen = 0;
 487
 488         p = (const u_int8_t *)sp;
 489         q = (const u_int8_t *)&ep->srcmask;
 490         r = (u_int8_t *)&s;
 491         for (i = 0 ; i < sp->sa_len; i++) {
 492                 r[i] = p[i] & q[i];
 493                 /* XXX estimate */
 494                 matchlen += (q[i] ? 8 : 0);
 495         }
 496
 497         p = (const u_int8_t *)dp;
 498         q = (const u_int8_t *)&ep->dstmask;
 499         r = (u_int8_t *)&d;
 500         for (i = 0 ; i < dp->sa_len; i++) {
 501                 r[i] = p[i] & q[i];
 502                 /* XXX rough estimate */
 503                 matchlen += (q[i] ? 8 : 0);
 504         }
 505
 506         /* need to overwrite len/family portion as we don't compare them */
 507         s.ss_len = sp->sa_len;
 508         s.ss_family = sp->sa_family;
 509         d.ss_len = dp->sa_len;
 510         d.ss_family = dp->sa_family;
 511
 512         if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 &&
 513             bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) {
 514                 return matchlen;
 515         } else
 516                 return 0;
 517 }
 518
 519 struct encaptabtag {
 520         void*                   *arg;
 521 };
 522
 523 static void
 524 encap_fillarg(
 525         struct mbuf *m,
 526         const struct encaptab *ep)
 527 {
 528         struct m_tag    *tag;
 529         struct encaptabtag *et;
 530
 531         tag = m_tag_create(KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_ENCAP,
 532                                           sizeof(struct encaptabtag), M_WAITOK, m);
 533
 534         if (tag != NULL) {
 535                 et = (struct encaptabtag*)(tag + 1);
 536                 et->arg = ep->arg;
 537                 m_tag_prepend(m, tag);
 538         }
 539 }
 540
 541 void *
 542 encap_getarg(m)
 543         struct mbuf *m;
 544 {
 545         struct m_tag    *tag;
 546         struct encaptabtag *et;
 547         void *p = NULL;
 548
 549         tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_ENCAP, NULL);
 550         if (tag) {
 551                 et = (struct encaptabtag*)(tag + 1);
 552                 p = et->arg;
 553                 m_tag_delete(m, tag);
 554         }
 555
 556         return p;
 557 }