sys/netinet/ip_encap.c

   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 void     (*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 void
 133 encap4_input(struct mbuf *m, ...)
 134 {
 135         int off, proto;
 136         struct ip *ip;
 137         struct sockaddr_in s, d;
 138         const struct protosw *psw;
 139         struct encaptab *ep, *match;
 140         int prio, matchprio;
 141         __va_list ap;
 142
 143         __va_start(ap, m);
 144         off = __va_arg(ap, int);
 145         proto = __va_arg(ap, int);
 146         __va_end(ap);
 147
 148         ip = mtod(m, struct ip *);
 149
 150         bzero(&s, sizeof s);
 151         s.sin_family = AF_INET;
 152         s.sin_len = sizeof(struct sockaddr_in);
 153         s.sin_addr = ip->ip_src;
 154         bzero(&d, sizeof d);
 155         d.sin_family = AF_INET;
 156         d.sin_len = sizeof(struct sockaddr_in);
 157         d.sin_addr = ip->ip_dst;
 158
 159         match = NULL;
 160         matchprio = 0;
 161         for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
 162                 if (ep->af != AF_INET)
 163                         continue;
 164                 if (ep->proto >= 0 && ep->proto != proto)
 165                         continue;
 166                 if (ep->func)
 167                         prio = (*ep->func)(m, off, proto, ep->arg);
 168                 else {
 169                         /*
 170                          * it's inbound traffic, we need to match in reverse
 171                          * order
 172                          */
 173                         prio = mask_match(ep, (struct sockaddr *)&d,
 174                             (struct sockaddr *)&s);
 175                 }
 176
 177                 /*
 178                  * We prioritize the matches by using bit length of the
 179                  * matches.  mask_match() and user-supplied matching function
 180                  * should return the bit length of the matches (for example,
 181                  * if both src/dst are matched for IPv4, 64 should be returned).
 182                  * 0 or negative return value means "it did not match".
 183                  *
 184                  * The question is, since we have two "mask" portion, we
 185                  * cannot really define total order between entries.
 186                  * For example, which of these should be preferred?
 187                  * mask_match() returns 48 (32 + 16) for both of them.
 188                  *      src=3ffe::/16, dst=3ffe:501::/32
 189                  *      src=3ffe:501::/32, dst=3ffe::/16
 190                  *
 191                  * We need to loop through all the possible candidates
 192                  * to get the best match - the search takes O(n) for
 193                  * n attachments (i.e. interfaces).
 194                  */
 195                 if (prio <= 0)
 196                         continue;
 197                 if (prio > matchprio) {
 198                         matchprio = prio;
 199                         match = ep;
 200                 }
 201         }
 202
 203         if (match) {
 204                 /* found a match, "match" has the best one */
 205                 psw = match->psw;
 206                 if (psw && psw->pr_input) {
 207                         encap_fillarg(m, match);
 208                         (*psw->pr_input)(m, off, proto);
 209                 } else
 210                         m_freem(m);
 211                 return;
 212         }
 213
 214         /* for backward compatibility */
 215         if (proto == IPPROTO_IPV4 && ipip_input) {
 216                 ipip_input(m, off, proto);
 217                 return;
 218         }
 219
 220         /* last resort: inject to raw socket */
 221         rip_input(m, off, proto);
 222 }
 223 #endif
 224
 225 #ifdef INET6
 226 int
 227 encap6_input(struct mbuf **mp, int *offp, int proto)
 228 {
 229         struct mbuf *m = *mp;
 230         struct ip6_hdr *ip6;
 231         struct sockaddr_in6 s, d;
 232         const struct ip6protosw *psw;
 233         struct encaptab *ep, *match;
 234         int prio, matchprio;
 235
 236         ip6 = mtod(m, struct ip6_hdr *);
 237
 238         bzero(&s, sizeof s);
 239         s.sin6_family = AF_INET6;
 240         s.sin6_len = sizeof(struct sockaddr_in6);
 241         s.sin6_addr = ip6->ip6_src;
 242         bzero(&d, sizeof d);
 243         d.sin6_family = AF_INET6;
 244         d.sin6_len = sizeof(struct sockaddr_in6);
 245         d.sin6_addr = ip6->ip6_dst;
 246
 247         match = NULL;
 248         matchprio = 0;
 249         for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
 250                 if (ep->af != AF_INET6)
 251                         continue;
 252                 if (ep->proto >= 0 && ep->proto != proto)
 253                         continue;
 254                 if (ep->func)
 255                         prio = (*ep->func)(m, *offp, proto, ep->arg);
 256                 else {
 257                         /*
 258                          * it's inbound traffic, we need to match in reverse
 259                          * order
 260                          */
 261                         prio = mask_match(ep, (struct sockaddr *)&d,
 262                             (struct sockaddr *)&s);
 263                 }
 264
 265                 /* see encap4_input() for issues here */
 266                 if (prio <= 0)
 267                         continue;
 268                 if (prio > matchprio) {
 269                         matchprio = prio;
 270                         match = ep;
 271                 }
 272         }
 273
 274         if (match) {
 275                 /* found a match */
 276                 psw = (const struct ip6protosw *)match->psw;
 277                 if (psw && psw->pr_input) {
 278                         encap_fillarg(m, match);
 279                         return (*psw->pr_input)(mp, offp, proto);
 280                 } else {
 281                         m_freem(m);
 282                         return IPPROTO_DONE;
 283                 }
 284         }
 285
 286         /* last resort: inject to raw socket */
 287         return rip6_input(mp, offp, proto);
 288 }
 289 #endif
 290
 291 static void
 292 encap_add(struct encaptab *ep)
 293 {
 294
 295         LIST_INSERT_HEAD(&encaptab, ep, chain);
 296 }
 297
 298 /*
 299  * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
 300  * length of mask (sm and dm) is assumed to be same as sp/dp.
 301  * Return value will be necessary as input (cookie) for encap_detach().
 302  */
 303 const struct encaptab *
 304 encap_attach(int af, int proto, const struct sockaddr *sp,
 305              const struct sockaddr *sm, const struct sockaddr *dp,
 306              const struct sockaddr *dm, const struct protosw *psw, void *arg)
 307 {
 308         struct encaptab *ep;
 309         int error;
 310
 311         crit_enter();
 312         /* sanity check on args */
 313         if (sp->sa_len > sizeof ep->src || dp->sa_len > sizeof ep->dst) {
 314                 error = EINVAL;
 315                 goto fail;
 316         }
 317         if (sp->sa_len != dp->sa_len) {
 318                 error = EINVAL;
 319                 goto fail;
 320         }
 321         if (af != sp->sa_family || af != dp->sa_family) {
 322                 error = EINVAL;
 323                 goto fail;
 324         }
 325
 326         /* check if anyone have already attached with exactly same config */
 327         for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
 328                 if (ep->af != af)
 329                         continue;
 330                 if (ep->proto != proto)
 331                         continue;
 332                 if (ep->src.ss_len != sp->sa_len ||
 333                     bcmp(&ep->src, sp, sp->sa_len) != 0 ||
 334                     bcmp(&ep->srcmask, sm, sp->sa_len) != 0)
 335                         continue;
 336                 if (ep->dst.ss_len != dp->sa_len ||
 337                     bcmp(&ep->dst, dp, dp->sa_len) != 0 ||
 338                     bcmp(&ep->dstmask, dm, dp->sa_len) != 0)
 339                         continue;
 340
 341                 error = EEXIST;
 342                 goto fail;
 343         }
 344
 345         ep = kmalloc(sizeof *ep, M_NETADDR, M_INTWAIT | M_ZERO | M_NULLOK);
 346         if (ep == NULL) {
 347                 error = ENOBUFS;
 348                 goto fail;
 349         }
 350
 351         ep->af = af;
 352         ep->proto = proto;
 353         bcopy(sp, &ep->src, sp->sa_len);
 354         bcopy(sm, &ep->srcmask, sp->sa_len);
 355         bcopy(dp, &ep->dst, dp->sa_len);
 356         bcopy(dm, &ep->dstmask, dp->sa_len);
 357         ep->psw = psw;
 358         ep->arg = arg;
 359
 360         encap_add(ep);
 361
 362         error = 0;
 363         crit_exit();
 364         return ep;
 365
 366 fail:
 367         crit_exit();
 368         return NULL;
 369 }
 370
 371 const struct encaptab *
 372 encap_attach_func(int af, int proto,
 373                   int (*func)(const struct mbuf *, int, int, void *),
 374                   const struct protosw *psw, void *arg)
 375 {
 376         struct encaptab *ep;
 377         int error;
 378
 379         crit_enter();
 380         /* sanity check on args */
 381         if (!func) {
 382                 error = EINVAL;
 383                 goto fail;
 384         }
 385
 386         ep = kmalloc(sizeof *ep, M_NETADDR, M_INTWAIT | M_ZERO | M_NULLOK);
 387         if (ep == NULL) {
 388                 error = ENOBUFS;
 389                 goto fail;
 390         }
 391
 392         ep->af = af;
 393         ep->proto = proto;
 394         ep->func = func;
 395         ep->psw = psw;
 396         ep->arg = arg;
 397
 398         encap_add(ep);
 399
 400         error = 0;
 401         crit_exit();
 402         return ep;
 403
 404 fail:
 405         crit_exit();
 406         return NULL;
 407 }
 408
 409 int
 410 encap_detach(const struct encaptab *cookie)
 411 {
 412         const struct encaptab *ep = cookie;
 413         struct encaptab *p;
 414
 415         for (p = LIST_FIRST(&encaptab); p; p = LIST_NEXT(p, chain)) {
 416                 if (p == ep) {
 417                         LIST_REMOVE(p, chain);
 418                         kfree(p, M_NETADDR);    /*XXX*/
 419                         return 0;
 420                 }
 421         }
 422
 423         return EINVAL;
 424 }
 425
 426 static int
 427 mask_match(const struct encaptab *ep, const struct sockaddr *sp,
 428            const struct sockaddr *dp)
 429 {
 430         struct sockaddr_storage s;
 431         struct sockaddr_storage d;
 432         int i;
 433         const u_int8_t *p, *q;
 434         u_int8_t *r;
 435         int matchlen;
 436
 437         if (sp->sa_len > sizeof s || dp->sa_len > sizeof d)
 438                 return 0;
 439         if (sp->sa_family != ep->af || dp->sa_family != ep->af)
 440                 return 0;
 441         if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len)
 442                 return 0;
 443
 444         matchlen = 0;
 445
 446         p = (const u_int8_t *)sp;
 447         q = (const u_int8_t *)&ep->srcmask;
 448         r = (u_int8_t *)&s;
 449         for (i = 0 ; i < sp->sa_len; i++) {
 450                 r[i] = p[i] & q[i];
 451                 /* XXX estimate */
 452                 matchlen += (q[i] ? 8 : 0);
 453         }
 454
 455         p = (const u_int8_t *)dp;
 456         q = (const u_int8_t *)&ep->dstmask;
 457         r = (u_int8_t *)&d;
 458         for (i = 0 ; i < dp->sa_len; i++) {
 459                 r[i] = p[i] & q[i];
 460                 /* XXX rough estimate */
 461                 matchlen += (q[i] ? 8 : 0);
 462         }
 463
 464         /* need to overwrite len/family portion as we don't compare them */
 465         s.ss_len = sp->sa_len;
 466         s.ss_family = sp->sa_family;
 467         d.ss_len = dp->sa_len;
 468         d.ss_family = dp->sa_family;
 469
 470         if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 &&
 471             bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) {
 472                 return matchlen;
 473         } else
 474                 return 0;
 475 }
 476
 477 static void
 478 encap_fillarg(struct mbuf *m, const struct encaptab *ep)
 479 {
 480         struct m_tag *tag;
 481
 482         tag = m_tag_get(PACKET_TAG_ENCAP, sizeof(void *), MB_DONTWAIT);
 483         if (tag != NULL) {
 484                 *(void **)m_tag_data(tag) = ep->arg;
 485                 m_tag_prepend(m, tag);
 486         }
 487 }
 488
 489 void *
 490 encap_getarg(struct mbuf *m)
 491 {
 492         void *p = NULL;
 493         struct m_tag *tag;
 494
 495         tag = m_tag_find(m, PACKET_TAG_ENCAP, NULL);
 496         if (tag != NULL) {
 497                 p = *(void **)m_tag_data(tag);
 498                 m_tag_delete(m, tag);
 499         }
 500         return p;
 501 }