bsd/netinet/ip_encap.c

   1 /*
   2  * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
   3  *
   4  * @APPLE_LICENSE_HEADER_START@
   5  *
   6  * Copyright (c) 1999-2003 Apple Computer, Inc.  All Rights Reserved.
   7  *
   8  * This file contains Original Code and/or Modifications of Original Code
   9  * as defined in and that are subject to the Apple Public Source License
  10  * Version 2.0 (the 'License'). You may not use this file except in
  11  * compliance with the License. Please obtain a copy of the License at
  12  * http://www.opensource.apple.com/apsl/ and read it before using this
  13  * file.
  14  *
  15  * The Original Code and all software distributed under the License are
  16  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  17  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  18  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  19  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  20  * Please see the License for the specific language governing rights and
  21  * limitations under the License.
  22  *
  23  * @APPLE_LICENSE_HEADER_END@
  24  */
  25 /*      $FreeBSD: src/sys/netinet/ip_encap.c,v 1.1.2.2 2001/07/03 11:01:46 ume Exp $    */
  26 /*      $KAME: ip_encap.c,v 1.41 2001/03/15 08:35:08 itojun Exp $       */
  27
  28 /*
  29  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
  30  * All rights reserved.
  31  *
  32  * Redistribution and use in source and binary forms, with or without
  33  * modification, are permitted provided that the following conditions
  34  * are met:
  35  * 1. Redistributions of source code must retain the above copyright
  36  *    notice, this list of conditions and the following disclaimer.
  37  * 2. Redistributions in binary form must reproduce the above copyright
  38  *    notice, this list of conditions and the following disclaimer in the
  39  *    documentation and/or other materials provided with the distribution.
  40  * 3. Neither the name of the project nor the names of its contributors
  41  *    may be used to endorse or promote products derived from this software
  42  *    without specific prior written permission.
  43  *
  44  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
  45  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  46  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  47  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
  48  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  49  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  50  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  51  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  52  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  53  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  54  * SUCH DAMAGE.
  55  */
  56 /*
  57  * My grandfather said that there's a devil inside tunnelling technology...
  58  *
  59  * We have surprisingly many protocols that want packets with IP protocol
  60  * #4 or #41.  Here's a list of protocols that want protocol #41:
  61  *      RFC1933 configured tunnel
  62  *      RFC1933 automatic tunnel
  63  *      RFC2401 IPsec tunnel
  64  *      RFC2473 IPv6 generic packet tunnelling
  65  *      RFC2529 6over4 tunnel
  66  *      mobile-ip6 (uses RFC2473)
  67  *      6to4 tunnel
  68  * Here's a list of protocol that want protocol #4:
  69  *      RFC1853 IPv4-in-IPv4 tunnelling
  70  *      RFC2003 IPv4 encapsulation within IPv4
  71  *      RFC2344 reverse tunnelling for mobile-ip4
  72  *      RFC2401 IPsec tunnel
  73  * Well, what can I say.  They impose different en/decapsulation mechanism
  74  * from each other, so they need separate protocol handler.  The only one
  75  * we can easily determine by protocol # is IPsec, which always has
  76  * AH/ESP/IPComp header right after outer IP header.
  77  *
  78  * So, clearly good old protosw does not work for protocol #4 and #41.
  79  * The code will let you match protocol via src/dst address pair.
  80  */
  81 /* XXX is M_NETADDR correct? */
  82
  83 #include <sys/param.h>
  84 #include <sys/systm.h>
  85 #include <sys/socket.h>
  86 #include <sys/sockio.h>
  87 #include <sys/mbuf.h>
  88 #include <sys/errno.h>
  89 #include <sys/protosw.h>
  90 #include <sys/queue.h>
  91
  92 #include <net/if.h>
  93 #include <net/route.h>
  94
  95 #include <netinet/in.h>
  96 #include <netinet/in_systm.h>
  97 #include <netinet/ip.h>
  98 #include <netinet/ip_var.h>
  99 #include <netinet/ip_encap.h>
 100 #if MROUTING
 101 #include <netinet/ip_mroute.h>
 102 #endif /* MROUTING */
 103
 104 #if INET6
 105 #include <netinet/ip6.h>
 106 #include <netinet6/ip6_var.h>
 107 #include <netinet6/ip6protosw.h>
 108 #endif
 109
 110
 111 #include <net/net_osdep.h>
 112
 113 #ifndef __APPLE__
 114 #include <sys/kernel.h>
 115 #include <sys/malloc.h>
 116 MALLOC_DEFINE(M_NETADDR, "Export Host", "Export host address structure");
 117 #endif
 118
 119 static void encap_add __P((struct encaptab *));
 120 static int mask_match __P((const struct encaptab *, const struct sockaddr *,
 121                 const struct sockaddr *));
 122 static void encap_fillarg __P((struct mbuf *, const struct encaptab *));
 123
 124 #ifndef LIST_HEAD_INITIALIZER
 125 /* rely upon BSS initialization */
 126 LIST_HEAD(, encaptab) encaptab;
 127 #else
 128 LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(&encaptab);
 129 #endif
 130
 131 void
 132 encap_init()
 133 {
 134         static int initialized = 0;
 135
 136         if (initialized)
 137                 return;
 138         initialized++;
 139 #if 0
 140         /*
 141          * we cannot use LIST_INIT() here, since drivers may want to call
 142          * encap_attach(), on driver attach.  encap_init() will be called
 143          * on AF_INET{,6} initialization, which happens after driver
 144          * initialization - using LIST_INIT() here can nuke encap_attach()
 145          * from drivers.
 146          */
 147         LIST_INIT(&encaptab);
 148 #endif
 149 }
 150
 151 #if INET
 152 void
 153 encap4_input(m, off)
 154         struct mbuf *m;
 155         int off;
 156 {
 157         int proto;
 158         struct ip *ip;
 159         struct sockaddr_in s, d;
 160         const struct protosw *psw;
 161         struct encaptab *ep, *match;
 162         int prio, matchprio;
 163
 164 #ifndef __APPLE__
 165         va_start(ap, m);
 166         off = va_arg(ap, int);
 167         proto = va_arg(ap, int);
 168         va_end(ap);
 169 #endif
 170
 171         ip = mtod(m, struct ip *);
 172 #ifdef __APPLE__
 173         proto = ip->ip_p;
 174 #endif
 175
 176         bzero(&s, sizeof(s));
 177         s.sin_family = AF_INET;
 178         s.sin_len = sizeof(struct sockaddr_in);
 179         s.sin_addr = ip->ip_src;
 180         bzero(&d, sizeof(d));
 181         d.sin_family = AF_INET;
 182         d.sin_len = sizeof(struct sockaddr_in);
 183         d.sin_addr = ip->ip_dst;
 184
 185         match = NULL;
 186         matchprio = 0;
 187         for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
 188                 if (ep->af != AF_INET)
 189                         continue;
 190                 if (ep->proto >= 0 && ep->proto != proto)
 191                         continue;
 192                 if (ep->func)
 193                         prio = (*ep->func)(m, off, proto, ep->arg);
 194                 else {
 195                         /*
 196                          * it's inbound traffic, we need to match in reverse
 197                          * order
 198                          */
 199                         prio = mask_match(ep, (struct sockaddr *)&d,
 200                             (struct sockaddr *)&s);
 201                 }
 202
 203                 /*
 204                  * We prioritize the matches by using bit length of the
 205                  * matches.  mask_match() and user-supplied matching function
 206                  * should return the bit length of the matches (for example,
 207                  * if both src/dst are matched for IPv4, 64 should be returned).
 208                  * 0 or negative return value means "it did not match".
 209                  *
 210                  * The question is, since we have two "mask" portion, we
 211                  * cannot really define total order between entries.
 212                  * For example, which of these should be preferred?
 213                  * mask_match() returns 48 (32 + 16) for both of them.
 214                  *      src=3ffe::/16, dst=3ffe:501::/32
 215                  *      src=3ffe:501::/32, dst=3ffe::/16
 216                  *
 217                  * We need to loop through all the possible candidates
 218                  * to get the best match - the search takes O(n) for
 219                  * n attachments (i.e. interfaces).
 220                  */
 221                 if (prio <= 0)
 222                         continue;
 223                 if (prio > matchprio) {
 224                         matchprio = prio;
 225                         match = ep;
 226                 }
 227         }
 228
 229         if (match) {
 230                 /* found a match, "match" has the best one */
 231                 psw = (const struct protosw *)match->psw;
 232                 if (psw && psw->pr_input) {
 233                         encap_fillarg(m, match);
 234                         (*psw->pr_input)(m, off);
 235                 } else
 236                         m_freem(m);
 237                 return;
 238         }
 239
 240         /* for backward compatibility */
 241 # if MROUTING
 242 #  define COMPATFUNC    ipip_input
 243 # endif /*MROUTING*/
 244
 245 #if COMPATFUNC
 246         if (proto == IPPROTO_IPV4) {
 247                 COMPATFUNC(m, off);
 248                 return;
 249         }
 250 #endif
 251
 252         /* last resort: inject to raw socket */
 253         rip_input(m, off);
 254 }
 255 #endif
 256
 257 #if INET6
 258 int
 259 encap6_input(mp, offp)
 260         struct mbuf **mp;
 261         int *offp;
 262 {
 263         struct mbuf *m = *mp;
 264         struct ip6_hdr *ip6;
 265         struct sockaddr_in6 s, d;
 266         const struct ip6protosw *psw;
 267         struct encaptab *ep, *match;
 268         int prio, matchprio;
 269         int proto;
 270
 271         ip6 = mtod(m, struct ip6_hdr *);
 272         proto = ip6->ip6_nxt;
 273
 274         bzero(&s, sizeof(s));
 275         s.sin6_family = AF_INET6;
 276         s.sin6_len = sizeof(struct sockaddr_in6);
 277         s.sin6_addr = ip6->ip6_src;
 278         bzero(&d, sizeof(d));
 279         d.sin6_family = AF_INET6;
 280         d.sin6_len = sizeof(struct sockaddr_in6);
 281         d.sin6_addr = ip6->ip6_dst;
 282
 283         match = NULL;
 284         matchprio = 0;
 285         for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
 286                 if (ep->af != AF_INET6)
 287                         continue;
 288                 if (ep->proto >= 0 && ep->proto != proto)
 289                         continue;
 290                 if (ep->func)
 291                         prio = (*ep->func)(m, *offp, proto, ep->arg);
 292                 else {
 293                         /*
 294                          * it's inbound traffic, we need to match in reverse
 295                          * order
 296                          */
 297                         prio = mask_match(ep, (struct sockaddr *)&d,
 298                             (struct sockaddr *)&s);
 299                 }
 300
 301                 /* see encap4_input() for issues here */
 302                 if (prio <= 0)
 303                         continue;
 304                 if (prio > matchprio) {
 305                         matchprio = prio;
 306                         match = ep;
 307                 }
 308         }
 309
 310         if (match) {
 311                 /* found a match */
 312                 psw = (const struct ip6protosw *)match->psw;
 313                 if (psw && psw->pr_input) {
 314                         encap_fillarg(m, match);
 315                         return (*psw->pr_input)(mp, offp);
 316                 } else {
 317                         m_freem(m);
 318                         return IPPROTO_DONE;
 319                 }
 320         }
 321
 322         /* last resort: inject to raw socket */
 323         return rip6_input(mp, offp);
 324 }
 325 #endif
 326
 327 static void
 328 encap_add(ep)
 329         struct encaptab *ep;
 330 {
 331
 332         LIST_INSERT_HEAD(&encaptab, ep, chain);
 333 }
 334
 335 /*
 336  * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
 337  * length of mask (sm and dm) is assumed to be same as sp/dp.
 338  * Return value will be necessary as input (cookie) for encap_detach().
 339  */
 340 const struct encaptab *
 341 encap_attach(af, proto, sp, sm, dp, dm, psw, arg)
 342         int af;
 343         int proto;
 344         const struct sockaddr *sp, *sm;
 345         const struct sockaddr *dp, *dm;
 346         const struct protosw *psw;
 347         void *arg;
 348 {
 349         struct encaptab *ep;
 350         int error;
 351         int s;
 352
 353         s = splnet();
 354         /* sanity check on args */
 355         if (sp->sa_len > sizeof(ep->src) || dp->sa_len > sizeof(ep->dst)) {
 356                 error = EINVAL;
 357                 goto fail;
 358         }
 359         if (sp->sa_len != dp->sa_len) {
 360                 error = EINVAL;
 361                 goto fail;
 362         }
 363         if (af != sp->sa_family || af != dp->sa_family) {
 364                 error = EINVAL;
 365                 goto fail;
 366         }
 367
 368         /* check if anyone have already attached with exactly same config */
 369         for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
 370                 if (ep->af != af)
 371                         continue;
 372                 if (ep->proto != proto)
 373                         continue;
 374                 if (ep->src.ss_len != sp->sa_len ||
 375                     bcmp(&ep->src, sp, sp->sa_len) != 0 ||
 376                     bcmp(&ep->srcmask, sm, sp->sa_len) != 0)
 377                         continue;
 378                 if (ep->dst.ss_len != dp->sa_len ||
 379                     bcmp(&ep->dst, dp, dp->sa_len) != 0 ||
 380                     bcmp(&ep->dstmask, dm, dp->sa_len) != 0)
 381                         continue;
 382
 383                 error = EEXIST;
 384                 goto fail;
 385         }
 386
 387         ep = _MALLOC(sizeof(*ep), M_NETADDR, M_WAITOK); /*XXX*/
 388         if (ep == NULL) {
 389                 error = ENOBUFS;
 390                 goto fail;
 391         }
 392         bzero(ep, sizeof(*ep));
 393
 394         ep->af = af;
 395         ep->proto = proto;
 396         bcopy(sp, &ep->src, sp->sa_len);
 397         bcopy(sm, &ep->srcmask, sp->sa_len);
 398         bcopy(dp, &ep->dst, dp->sa_len);
 399         bcopy(dm, &ep->dstmask, dp->sa_len);
 400         ep->psw = psw;
 401         ep->arg = arg;
 402
 403         encap_add(ep);
 404
 405         error = 0;
 406         splx(s);
 407         return ep;
 408
 409 fail:
 410         splx(s);
 411         return NULL;
 412 }
 413
 414 const struct encaptab *
 415 encap_attach_func(af, proto, func, psw, arg)
 416         int af;
 417         int proto;
 418         int (*func) __P((const struct mbuf *, int, int, void *));
 419         const struct protosw *psw;
 420         void *arg;
 421 {
 422         struct encaptab *ep;
 423         int error;
 424         int s;
 425
 426         s = splnet();
 427         /* sanity check on args */
 428         if (!func) {
 429                 error = EINVAL;
 430                 goto fail;
 431         }
 432
 433         ep = _MALLOC(sizeof(*ep), M_NETADDR, M_WAITOK); /*XXX*/
 434         if (ep == NULL) {
 435                 error = ENOBUFS;
 436                 goto fail;
 437         }
 438         bzero(ep, sizeof(*ep));
 439
 440         ep->af = af;
 441         ep->proto = proto;
 442         ep->func = func;
 443         ep->psw = psw;
 444         ep->arg = arg;
 445
 446         encap_add(ep);
 447
 448         error = 0;
 449         splx(s);
 450         return ep;
 451
 452 fail:
 453         splx(s);
 454         return NULL;
 455 }
 456
 457 int
 458 encap_detach(cookie)
 459         const struct encaptab *cookie;
 460 {
 461         const struct encaptab *ep = cookie;
 462         struct encaptab *p;
 463
 464         for (p = LIST_FIRST(&encaptab); p; p = LIST_NEXT(p, chain)) {
 465                 if (p == ep) {
 466                         LIST_REMOVE(p, chain);
 467                         _FREE(p, M_NETADDR);    /*XXX*/
 468                         return 0;
 469                 }
 470         }
 471
 472         return EINVAL;
 473 }
 474
 475 static int
 476 mask_match(ep, sp, dp)
 477         const struct encaptab *ep;
 478         const struct sockaddr *sp;
 479         const struct sockaddr *dp;
 480 {
 481         struct sockaddr_storage s;
 482         struct sockaddr_storage d;
 483         int i;
 484         const u_int8_t *p, *q;
 485         u_int8_t *r;
 486         int matchlen;
 487
 488         if (sp->sa_len > sizeof(s) || dp->sa_len > sizeof(d))
 489                 return 0;
 490         if (sp->sa_family != ep->af || dp->sa_family != ep->af)
 491                 return 0;
 492         if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len)
 493                 return 0;
 494
 495         matchlen = 0;
 496
 497         p = (const u_int8_t *)sp;
 498         q = (const u_int8_t *)&ep->srcmask;
 499         r = (u_int8_t *)&s;
 500         for (i = 0 ; i < sp->sa_len; i++) {
 501                 r[i] = p[i] & q[i];
 502                 /* XXX estimate */
 503                 matchlen += (q[i] ? 8 : 0);
 504         }
 505
 506         p = (const u_int8_t *)dp;
 507         q = (const u_int8_t *)&ep->dstmask;
 508         r = (u_int8_t *)&d;
 509         for (i = 0 ; i < dp->sa_len; i++) {
 510                 r[i] = p[i] & q[i];
 511                 /* XXX rough estimate */
 512                 matchlen += (q[i] ? 8 : 0);
 513         }
 514
 515         /* need to overwrite len/family portion as we don't compare them */
 516         s.ss_len = sp->sa_len;
 517         s.ss_family = sp->sa_family;
 518         d.ss_len = dp->sa_len;
 519         d.ss_family = dp->sa_family;
 520
 521         if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 &&
 522             bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) {
 523                 return matchlen;
 524         } else
 525                 return 0;
 526 }
 527
 528 static void
 529 encap_fillarg(m, ep)
 530         struct mbuf *m;
 531         const struct encaptab *ep;
 532 {
 533 #if 0
 534         m->m_pkthdr.aux = ep->arg;
 535 #else
 536         struct mbuf *n;
 537
 538         n = m_aux_add(m, AF_INET, IPPROTO_IPV4);
 539         if (n) {
 540                 *mtod(n, void **) = ep->arg;
 541                 n->m_len = sizeof(void *);
 542         }
 543 #endif
 544 }
 545
 546 void *
 547 encap_getarg(m)
 548         struct mbuf *m;
 549 {
 550         void *p;
 551 #if 0
 552         p = m->m_pkthdr.aux;
 553         m->m_pkthdr.aux = NULL;
 554         return p;
 555 #else
 556         struct mbuf *n;
 557
 558         p = NULL;
 559         n = m_aux_find(m, AF_INET, IPPROTO_IPV4);
 560         if (n) {
 561                 if (n->m_len == sizeof(void *))
 562                         p = *mtod(n, void **);
 563                 m_aux_delete(m, n);
 564         }
 565         return p;
 566 #endif
 567 }