bsd/netinet/ip_encap.c

   1 /*
   2  * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
   3  *
   4  * @APPLE_LICENSE_HEADER_START@
   5  *
   6  * The contents of this file constitute Original Code as defined in and
   7  * are subject to the Apple Public Source License Version 1.1 (the
   8  * "License").  You may not use this file except in compliance with the
   9  * License.  Please obtain a copy of the License at
  10  * http://www.apple.com/publicsource and read it before using this file.
  11  *
  12  * This Original Code and all software distributed under the License are
  13  * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  14  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  15  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  16  * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT.  Please see the
  17  * License for the specific language governing rights and limitations
  18  * under the License.
  19  *
  20  * @APPLE_LICENSE_HEADER_END@
  21  */
  22 /*      $FreeBSD: src/sys/netinet/ip_encap.c,v 1.1.2.2 2001/07/03 11:01:46 ume Exp $    */
  23 /*      $KAME: ip_encap.c,v 1.41 2001/03/15 08:35:08 itojun Exp $       */
  24
  25 /*
  26  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
  27  * All rights reserved.
  28  *
  29  * Redistribution and use in source and binary forms, with or without
  30  * modification, are permitted provided that the following conditions
  31  * are met:
  32  * 1. Redistributions of source code must retain the above copyright
  33  *    notice, this list of conditions and the following disclaimer.
  34  * 2. Redistributions in binary form must reproduce the above copyright
  35  *    notice, this list of conditions and the following disclaimer in the
  36  *    documentation and/or other materials provided with the distribution.
  37  * 3. Neither the name of the project nor the names of its contributors
  38  *    may be used to endorse or promote products derived from this software
  39  *    without specific prior written permission.
  40  *
  41  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
  42  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  43  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  44  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
  45  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  46  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  47  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  48  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  49  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  50  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  51  * SUCH DAMAGE.
  52  */
  53 /*
  54  * My grandfather said that there's a devil inside tunnelling technology...
  55  *
  56  * We have surprisingly many protocols that want packets with IP protocol
  57  * #4 or #41.  Here's a list of protocols that want protocol #41:
  58  *      RFC1933 configured tunnel
  59  *      RFC1933 automatic tunnel
  60  *      RFC2401 IPsec tunnel
  61  *      RFC2473 IPv6 generic packet tunnelling
  62  *      RFC2529 6over4 tunnel
  63  *      mobile-ip6 (uses RFC2473)
  64  *      6to4 tunnel
  65  * Here's a list of protocol that want protocol #4:
  66  *      RFC1853 IPv4-in-IPv4 tunnelling
  67  *      RFC2003 IPv4 encapsulation within IPv4
  68  *      RFC2344 reverse tunnelling for mobile-ip4
  69  *      RFC2401 IPsec tunnel
  70  * Well, what can I say.  They impose different en/decapsulation mechanism
  71  * from each other, so they need separate protocol handler.  The only one
  72  * we can easily determine by protocol # is IPsec, which always has
  73  * AH/ESP/IPComp header right after outer IP header.
  74  *
  75  * So, clearly good old protosw does not work for protocol #4 and #41.
  76  * The code will let you match protocol via src/dst address pair.
  77  */
  78 /* XXX is M_NETADDR correct? */
  79
  80 #include <sys/param.h>
  81 #include <sys/systm.h>
  82 #include <sys/socket.h>
  83 #include <sys/sockio.h>
  84 #include <sys/mbuf.h>
  85 #include <sys/errno.h>
  86 #include <sys/protosw.h>
  87 #include <sys/queue.h>
  88
  89 #include <net/if.h>
  90 #include <net/route.h>
  91
  92 #include <netinet/in.h>
  93 #include <netinet/in_systm.h>
  94 #include <netinet/ip.h>
  95 #include <netinet/ip_var.h>
  96 #include <netinet/ip_encap.h>
  97 #if MROUTING
  98 #include <netinet/ip_mroute.h>
  99 #endif /* MROUTING */
 100
 101 #if INET6
 102 #include <netinet/ip6.h>
 103 #include <netinet6/ip6_var.h>
 104 #include <netinet6/ip6protosw.h>
 105 #endif
 106
 107
 108 #include <net/net_osdep.h>
 109
 110 #ifndef __APPLE__
 111 #include <sys/kernel.h>
 112 #include <sys/malloc.h>
 113 MALLOC_DEFINE(M_NETADDR, "Export Host", "Export host address structure");
 114 #endif
 115
 116 static void encap_add __P((struct encaptab *));
 117 static int mask_match __P((const struct encaptab *, const struct sockaddr *,
 118                 const struct sockaddr *));
 119 static void encap_fillarg __P((struct mbuf *, const struct encaptab *));
 120
 121 #ifndef LIST_HEAD_INITIALIZER
 122 /* rely upon BSS initialization */
 123 LIST_HEAD(, encaptab) encaptab;
 124 #else
 125 LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(&encaptab);
 126 #endif
 127
 128 void
 129 encap_init()
 130 {
 131         static int initialized = 0;
 132
 133         if (initialized)
 134                 return;
 135         initialized++;
 136 #if 0
 137         /*
 138          * we cannot use LIST_INIT() here, since drivers may want to call
 139          * encap_attach(), on driver attach.  encap_init() will be called
 140          * on AF_INET{,6} initialization, which happens after driver
 141          * initialization - using LIST_INIT() here can nuke encap_attach()
 142          * from drivers.
 143          */
 144         LIST_INIT(&encaptab);
 145 #endif
 146 }
 147
 148 #if INET
 149 void
 150 encap4_input(m, off)
 151         struct mbuf *m;
 152         int off;
 153 {
 154         int proto;
 155         struct ip *ip;
 156         struct sockaddr_in s, d;
 157         const struct protosw *psw;
 158         struct encaptab *ep, *match;
 159         int prio, matchprio;
 160
 161 #ifndef __APPLE__
 162         va_start(ap, m);
 163         off = va_arg(ap, int);
 164         proto = va_arg(ap, int);
 165         va_end(ap);
 166 #endif
 167
 168         ip = mtod(m, struct ip *);
 169 #ifdef __APPLE__
 170         proto = ip->ip_p;
 171 #endif
 172
 173         bzero(&s, sizeof(s));
 174         s.sin_family = AF_INET;
 175         s.sin_len = sizeof(struct sockaddr_in);
 176         s.sin_addr = ip->ip_src;
 177         bzero(&d, sizeof(d));
 178         d.sin_family = AF_INET;
 179         d.sin_len = sizeof(struct sockaddr_in);
 180         d.sin_addr = ip->ip_dst;
 181
 182         match = NULL;
 183         matchprio = 0;
 184         for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
 185                 if (ep->af != AF_INET)
 186                         continue;
 187                 if (ep->proto >= 0 && ep->proto != proto)
 188                         continue;
 189                 if (ep->func)
 190                         prio = (*ep->func)(m, off, proto, ep->arg);
 191                 else {
 192                         /*
 193                          * it's inbound traffic, we need to match in reverse
 194                          * order
 195                          */
 196                         prio = mask_match(ep, (struct sockaddr *)&d,
 197                             (struct sockaddr *)&s);
 198                 }
 199
 200                 /*
 201                  * We prioritize the matches by using bit length of the
 202                  * matches.  mask_match() and user-supplied matching function
 203                  * should return the bit length of the matches (for example,
 204                  * if both src/dst are matched for IPv4, 64 should be returned).
 205                  * 0 or negative return value means "it did not match".
 206                  *
 207                  * The question is, since we have two "mask" portion, we
 208                  * cannot really define total order between entries.
 209                  * For example, which of these should be preferred?
 210                  * mask_match() returns 48 (32 + 16) for both of them.
 211                  *      src=3ffe::/16, dst=3ffe:501::/32
 212                  *      src=3ffe:501::/32, dst=3ffe::/16
 213                  *
 214                  * We need to loop through all the possible candidates
 215                  * to get the best match - the search takes O(n) for
 216                  * n attachments (i.e. interfaces).
 217                  */
 218                 if (prio <= 0)
 219                         continue;
 220                 if (prio > matchprio) {
 221                         matchprio = prio;
 222                         match = ep;
 223                 }
 224         }
 225
 226         if (match) {
 227                 /* found a match, "match" has the best one */
 228                 psw = (const struct protosw *)match->psw;
 229                 if (psw && psw->pr_input) {
 230                         encap_fillarg(m, match);
 231                         (*psw->pr_input)(m, off);
 232                 } else
 233                         m_freem(m);
 234                 return;
 235         }
 236
 237         /* for backward compatibility */
 238 # if MROUTING
 239 #  define COMPATFUNC    ipip_input
 240 # endif /*MROUTING*/
 241
 242 #if COMPATFUNC
 243         if (proto == IPPROTO_IPV4) {
 244                 COMPATFUNC(m, off);
 245                 return;
 246         }
 247 #endif
 248
 249         /* last resort: inject to raw socket */
 250         rip_input(m, off);
 251 }
 252 #endif
 253
 254 #if INET6
 255 int
 256 encap6_input(mp, offp)
 257         struct mbuf **mp;
 258         int *offp;
 259 {
 260         struct mbuf *m = *mp;
 261         struct ip6_hdr *ip6;
 262         struct sockaddr_in6 s, d;
 263         const struct ip6protosw *psw;
 264         struct encaptab *ep, *match;
 265         int prio, matchprio;
 266         int proto;
 267
 268         ip6 = mtod(m, struct ip6_hdr *);
 269         proto = ip6->ip6_nxt;
 270
 271         bzero(&s, sizeof(s));
 272         s.sin6_family = AF_INET6;
 273         s.sin6_len = sizeof(struct sockaddr_in6);
 274         s.sin6_addr = ip6->ip6_src;
 275         bzero(&d, sizeof(d));
 276         d.sin6_family = AF_INET6;
 277         d.sin6_len = sizeof(struct sockaddr_in6);
 278         d.sin6_addr = ip6->ip6_dst;
 279
 280         match = NULL;
 281         matchprio = 0;
 282         for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
 283                 if (ep->af != AF_INET6)
 284                         continue;
 285                 if (ep->proto >= 0 && ep->proto != proto)
 286                         continue;
 287                 if (ep->func)
 288                         prio = (*ep->func)(m, *offp, proto, ep->arg);
 289                 else {
 290                         /*
 291                          * it's inbound traffic, we need to match in reverse
 292                          * order
 293                          */
 294                         prio = mask_match(ep, (struct sockaddr *)&d,
 295                             (struct sockaddr *)&s);
 296                 }
 297
 298                 /* see encap4_input() for issues here */
 299                 if (prio <= 0)
 300                         continue;
 301                 if (prio > matchprio) {
 302                         matchprio = prio;
 303                         match = ep;
 304                 }
 305         }
 306
 307         if (match) {
 308                 /* found a match */
 309                 psw = (const struct ip6protosw *)match->psw;
 310                 if (psw && psw->pr_input) {
 311                         encap_fillarg(m, match);
 312                         return (*psw->pr_input)(mp, offp);
 313                 } else {
 314                         m_freem(m);
 315                         return IPPROTO_DONE;
 316                 }
 317         }
 318
 319         /* last resort: inject to raw socket */
 320         return rip6_input(mp, offp);
 321 }
 322 #endif
 323
 324 static void
 325 encap_add(ep)
 326         struct encaptab *ep;
 327 {
 328
 329         LIST_INSERT_HEAD(&encaptab, ep, chain);
 330 }
 331
 332 /*
 333  * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
 334  * length of mask (sm and dm) is assumed to be same as sp/dp.
 335  * Return value will be necessary as input (cookie) for encap_detach().
 336  */
 337 const struct encaptab *
 338 encap_attach(af, proto, sp, sm, dp, dm, psw, arg)
 339         int af;
 340         int proto;
 341         const struct sockaddr *sp, *sm;
 342         const struct sockaddr *dp, *dm;
 343         const struct protosw *psw;
 344         void *arg;
 345 {
 346         struct encaptab *ep;
 347         int error;
 348         int s;
 349
 350         s = splnet();
 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         splx(s);
 404         return ep;
 405
 406 fail:
 407         splx(s);
 408         return NULL;
 409 }
 410
 411 const struct encaptab *
 412 encap_attach_func(af, proto, func, psw, arg)
 413         int af;
 414         int proto;
 415         int (*func) __P((const struct mbuf *, int, int, void *));
 416         const struct protosw *psw;
 417         void *arg;
 418 {
 419         struct encaptab *ep;
 420         int error;
 421         int s;
 422
 423         s = splnet();
 424         /* sanity check on args */
 425         if (!func) {
 426                 error = EINVAL;
 427                 goto fail;
 428         }
 429
 430         ep = _MALLOC(sizeof(*ep), M_NETADDR, M_WAITOK); /*XXX*/
 431         if (ep == NULL) {
 432                 error = ENOBUFS;
 433                 goto fail;
 434         }
 435         bzero(ep, sizeof(*ep));
 436
 437         ep->af = af;
 438         ep->proto = proto;
 439         ep->func = func;
 440         ep->psw = psw;
 441         ep->arg = arg;
 442
 443         encap_add(ep);
 444
 445         error = 0;
 446         splx(s);
 447         return ep;
 448
 449 fail:
 450         splx(s);
 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 static void
 526 encap_fillarg(m, ep)
 527         struct mbuf *m;
 528         const struct encaptab *ep;
 529 {
 530 #if 0
 531         m->m_pkthdr.aux = ep->arg;
 532 #else
 533         struct mbuf *n;
 534
 535         n = m_aux_add(m, AF_INET, IPPROTO_IPV4);
 536         if (n) {
 537                 *mtod(n, void **) = ep->arg;
 538                 n->m_len = sizeof(void *);
 539         }
 540 #endif
 541 }
 542
 543 void *
 544 encap_getarg(m)
 545         struct mbuf *m;
 546 {
 547         void *p;
 548 #if 0
 549         p = m->m_pkthdr.aux;
 550         m->m_pkthdr.aux = NULL;
 551         return p;
 552 #else
 553         struct mbuf *n;
 554
 555         p = NULL;
 556         n = m_aux_find(m, AF_INET, IPPROTO_IPV4);
 557         if (n) {
 558                 if (n->m_len == sizeof(void *))
 559                         p = *mtod(n, void **);
 560                 m_aux_delete(m, n);
 561         }
 562         return p;
 563 #endif
 564 }