slirp/ip_input.c

   1 /*
   2  * Copyright (c) 1982, 1986, 1988, 1993
   3  *      The Regents of the University of California.  All rights reserved.
   4  *
   5  * Redistribution and use in source and binary forms, with or without
   6  * modification, are permitted provided that the following conditions
   7  * are met:
   8  * 1. Redistributions of source code must retain the above copyright
   9  *    notice, this list of conditions and the following disclaimer.
  10  * 2. Redistributions in binary form must reproduce the above copyright
  11  *    notice, this list of conditions and the following disclaimer in the
  12  *    documentation and/or other materials provided with the distribution.
  13  * 3. Neither the name of the University nor the names of its contributors
  14  *    may be used to endorse or promote products derived from this software
  15  *    without specific prior written permission.
  16  *
  17  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  27  * SUCH DAMAGE.
  28  *
  29  *      @(#)ip_input.c  8.2 (Berkeley) 1/4/94
  30  * ip_input.c,v 1.11 1994/11/16 10:17:08 jkh Exp
  31  */
  32
  33 /*
  34  * Changes and additions relating to SLiRP are
  35  * Copyright (c) 1995 Danny Gasparovski.
  36  *
  37  * Please read the file COPYRIGHT for the
  38  * terms and conditions of the copyright.
  39  */
  40
  41 #include "slirp.h"
  42 #include "ip_icmp.h"
  43
  44 static struct ip *ip_reass(Slirp *slirp, struct ip *ip, struct ipq *fp);
  45 static void ip_freef(Slirp *slirp, struct ipq *fp);
  46 static void ip_enq(register struct ipasfrag *p,
  47                    register struct ipasfrag *prev);
  48 static void ip_deq(register struct ipasfrag *p);
  49
  50 /*
  51  * IP initialization: fill in IP protocol switch table.
  52  * All protocols not implemented in kernel go to raw IP protocol handler.
  53  */
  54 void
  55 ip_init(Slirp *slirp)
  56 {
  57     slirp->ipq.ip_link.next = slirp->ipq.ip_link.prev = &slirp->ipq.ip_link;
  58     udp_init(slirp);
  59     tcp_init(slirp);
  60     icmp_init(slirp);
  61 }
  62
  63 void ip_cleanup(Slirp *slirp)
  64 {
  65     udp_cleanup(slirp);
  66     tcp_cleanup(slirp);
  67     icmp_cleanup(slirp);
  68 }
  69
  70 /*
  71  * Ip input routine.  Checksum and byte swap header.  If fragmented
  72  * try to reassemble.  Process options.  Pass to next level.
  73  */
  74 void
  75 ip_input(struct mbuf *m)
  76 {
  77         Slirp *slirp = m->slirp;
  78         register struct ip *ip;
  79         int hlen;
  80
  81         if (!slirp->in_enabled) {
  82                 goto bad;
  83         }
  84
  85         DEBUG_CALL("ip_input");
  86         DEBUG_ARG("m = %p", m);
  87         DEBUG_ARG("m_len = %d", m->m_len);
  88
  89         if (m->m_len < sizeof (struct ip)) {
  90                 goto bad;
  91         }
  92
  93         ip = mtod(m, struct ip *);
  94
  95         if (ip->ip_v != IPVERSION) {
  96                 goto bad;
  97         }
  98
  99         hlen = ip->ip_hl << 2;
 100         if (hlen<sizeof(struct ip ) || hlen>m->m_len) {/* min header length */
 101           goto bad;                                  /* or packet too short */
 102         }
 103
 104         /* keep ip header intact for ICMP reply
 105          * ip->ip_sum = cksum(m, hlen);
 106          * if (ip->ip_sum) {
 107          */
 108         if(cksum(m,hlen)) {
 109           goto bad;
 110         }
 111
 112         /*
 113          * Convert fields to host representation.
 114          */
 115         NTOHS(ip->ip_len);
 116         if (ip->ip_len < hlen) {
 117                 goto bad;
 118         }
 119         NTOHS(ip->ip_id);
 120         NTOHS(ip->ip_off);
 121
 122         /*
 123          * Check that the amount of data in the buffers
 124          * is as at least much as the IP header would have us expect.
 125          * Trim mbufs if longer than we expect.
 126          * Drop packet if shorter than we expect.
 127          */
 128         if (m->m_len < ip->ip_len) {
 129                 goto bad;
 130         }
 131
 132         /* Should drop packet if mbuf too long? hmmm... */
 133         if (m->m_len > ip->ip_len)
 134            m_adj(m, ip->ip_len - m->m_len);
 135
 136         /* check ip_ttl for a correct ICMP reply */
 137         if (ip->ip_ttl == 0) {
 138             icmp_send_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, 0, "ttl");
 139             goto bad;
 140         }
 141
 142         /*
 143          * If offset or IP_MF are set, must reassemble.
 144          * Otherwise, nothing need be done.
 145          * (We could look in the reassembly queue to see
 146          * if the packet was previously fragmented,
 147          * but it's not worth the time; just let them time out.)
 148          *
 149          * XXX This should fail, don't fragment yet
 150          */
 151         if (ip->ip_off &~ IP_DF) {
 152           register struct ipq *fp;
 153       struct qlink *l;
 154                 /*
 155                  * Look for queue of fragments
 156                  * of this datagram.
 157                  */
 158                 for (l = slirp->ipq.ip_link.next; l != &slirp->ipq.ip_link;
 159                      l = l->next) {
 160             fp = container_of(l, struct ipq, ip_link);
 161             if (ip->ip_id == fp->ipq_id &&
 162                     ip->ip_src.s_addr == fp->ipq_src.s_addr &&
 163                     ip->ip_dst.s_addr == fp->ipq_dst.s_addr &&
 164                     ip->ip_p == fp->ipq_p)
 165                     goto found;
 166         }
 167         fp = NULL;
 168         found:
 169
 170                 /*
 171                  * Adjust ip_len to not reflect header,
 172                  * set ip_mff if more fragments are expected,
 173                  * convert offset of this to bytes.
 174                  */
 175                 ip->ip_len -= hlen;
 176                 if (ip->ip_off & IP_MF)
 177                   ip->ip_tos |= 1;
 178                 else
 179                   ip->ip_tos &= ~1;
 180
 181                 ip->ip_off <<= 3;
 182
 183                 /*
 184                  * If datagram marked as having more fragments
 185                  * or if this is not the first fragment,
 186                  * attempt reassembly; if it succeeds, proceed.
 187                  */
 188                 if (ip->ip_tos & 1 || ip->ip_off) {
 189                         ip = ip_reass(slirp, ip, fp);
 190                         if (ip == NULL)
 191                                 return;
 192                         m = dtom(slirp, ip);
 193                 } else
 194                         if (fp)
 195                            ip_freef(slirp, fp);
 196
 197         } else
 198                 ip->ip_len -= hlen;
 199
 200         /*
 201          * Switch out to protocol's input routine.
 202          */
 203         switch (ip->ip_p) {
 204          case IPPROTO_TCP:
 205                 tcp_input(m, hlen, (struct socket *)NULL, AF_INET);
 206                 break;
 207          case IPPROTO_UDP:
 208                 udp_input(m, hlen);
 209                 break;
 210          case IPPROTO_ICMP:
 211                 icmp_input(m, hlen);
 212                 break;
 213          default:
 214                 m_free(m);
 215         }
 216         return;
 217 bad:
 218         m_free(m);
 219 }
 220
 221 #define iptofrag(P) ((struct ipasfrag *)(((char*)(P)) - sizeof(struct qlink)))
 222 #define fragtoip(P) ((struct ip*)(((char*)(P)) + sizeof(struct qlink)))
 223 /*
 224  * Take incoming datagram fragment and try to
 225  * reassemble it into whole datagram.  If a chain for
 226  * reassembly of this datagram already exists, then it
 227  * is given as fp; otherwise have to make a chain.
 228  */
 229 static struct ip *
 230 ip_reass(Slirp *slirp, struct ip *ip, struct ipq *fp)
 231 {
 232         register struct mbuf *m = dtom(slirp, ip);
 233         register struct ipasfrag *q;
 234         int hlen = ip->ip_hl << 2;
 235         int i, next;
 236
 237         DEBUG_CALL("ip_reass");
 238         DEBUG_ARG("ip = %p", ip);
 239         DEBUG_ARG("fp = %p", fp);
 240         DEBUG_ARG("m = %p", m);
 241
 242         /*
 243          * Presence of header sizes in mbufs
 244          * would confuse code below.
 245          * Fragment m_data is concatenated.
 246          */
 247         m->m_data += hlen;
 248         m->m_len -= hlen;
 249
 250         /*
 251          * If first fragment to arrive, create a reassembly queue.
 252          */
 253         if (fp == NULL) {
 254           struct mbuf *t = m_get(slirp);
 255
 256           if (t == NULL) {
 257               goto dropfrag;
 258           }
 259           fp = mtod(t, struct ipq *);
 260           insque(&fp->ip_link, &slirp->ipq.ip_link);
 261           fp->ipq_ttl = IPFRAGTTL;
 262           fp->ipq_p = ip->ip_p;
 263           fp->ipq_id = ip->ip_id;
 264           fp->frag_link.next = fp->frag_link.prev = &fp->frag_link;
 265           fp->ipq_src = ip->ip_src;
 266           fp->ipq_dst = ip->ip_dst;
 267           q = (struct ipasfrag *)fp;
 268           goto insert;
 269         }
 270
 271         /*
 272          * Find a segment which begins after this one does.
 273          */
 274         for (q = fp->frag_link.next; q != (struct ipasfrag *)&fp->frag_link;
 275             q = q->ipf_next)
 276                 if (q->ipf_off > ip->ip_off)
 277                         break;
 278
 279         /*
 280          * If there is a preceding segment, it may provide some of
 281          * our data already.  If so, drop the data from the incoming
 282          * segment.  If it provides all of our data, drop us.
 283          */
 284         if (q->ipf_prev != &fp->frag_link) {
 285         struct ipasfrag *pq = q->ipf_prev;
 286                 i = pq->ipf_off + pq->ipf_len - ip->ip_off;
 287                 if (i > 0) {
 288                         if (i >= ip->ip_len)
 289                                 goto dropfrag;
 290                         m_adj(dtom(slirp, ip), i);
 291                         ip->ip_off += i;
 292                         ip->ip_len -= i;
 293                 }
 294         }
 295
 296         /*
 297          * While we overlap succeeding segments trim them or,
 298          * if they are completely covered, dequeue them.
 299          */
 300         while (q != (struct ipasfrag*)&fp->frag_link &&
 301             ip->ip_off + ip->ip_len > q->ipf_off) {
 302                 i = (ip->ip_off + ip->ip_len) - q->ipf_off;
 303                 if (i < q->ipf_len) {
 304                         q->ipf_len -= i;
 305                         q->ipf_off += i;
 306                         m_adj(dtom(slirp, q), i);
 307                         break;
 308                 }
 309                 q = q->ipf_next;
 310                 m_free(dtom(slirp, q->ipf_prev));
 311                 ip_deq(q->ipf_prev);
 312         }
 313
 314 insert:
 315         /*
 316          * Stick new segment in its place;
 317          * check for complete reassembly.
 318          */
 319         ip_enq(iptofrag(ip), q->ipf_prev);
 320         next = 0;
 321         for (q = fp->frag_link.next; q != (struct ipasfrag*)&fp->frag_link;
 322             q = q->ipf_next) {
 323                 if (q->ipf_off != next)
 324                         return NULL;
 325                 next += q->ipf_len;
 326         }
 327         if (((struct ipasfrag *)(q->ipf_prev))->ipf_tos & 1)
 328                 return NULL;
 329
 330         /*
 331          * Reassembly is complete; concatenate fragments.
 332          */
 333     q = fp->frag_link.next;
 334         m = dtom(slirp, q);
 335
 336         q = (struct ipasfrag *) q->ipf_next;
 337         while (q != (struct ipasfrag*)&fp->frag_link) {
 338           struct mbuf *t = dtom(slirp, q);
 339           q = (struct ipasfrag *) q->ipf_next;
 340           m_cat(m, t);
 341         }
 342
 343         /*
 344          * Create header for new ip packet by
 345          * modifying header of first packet;
 346          * dequeue and discard fragment reassembly header.
 347          * Make header visible.
 348          */
 349         q = fp->frag_link.next;
 350
 351         /*
 352          * If the fragments concatenated to an mbuf that's
 353          * bigger than the total size of the fragment, then and
 354          * m_ext buffer was alloced. But fp->ipq_next points to
 355          * the old buffer (in the mbuf), so we must point ip
 356          * into the new buffer.
 357          */
 358         if (m->m_flags & M_EXT) {
 359           int delta = (char *)q - m->m_dat;
 360           q = (struct ipasfrag *)(m->m_ext + delta);
 361         }
 362
 363     ip = fragtoip(q);
 364         ip->ip_len = next;
 365         ip->ip_tos &= ~1;
 366         ip->ip_src = fp->ipq_src;
 367         ip->ip_dst = fp->ipq_dst;
 368         remque(&fp->ip_link);
 369         (void) m_free(dtom(slirp, fp));
 370         m->m_len += (ip->ip_hl << 2);
 371         m->m_data -= (ip->ip_hl << 2);
 372
 373         return ip;
 374
 375 dropfrag:
 376         m_free(m);
 377         return NULL;
 378 }
 379
 380 /*
 381  * Free a fragment reassembly header and all
 382  * associated datagrams.
 383  */
 384 static void
 385 ip_freef(Slirp *slirp, struct ipq *fp)
 386 {
 387         register struct ipasfrag *q, *p;
 388
 389         for (q = fp->frag_link.next; q != (struct ipasfrag*)&fp->frag_link; q = p) {
 390                 p = q->ipf_next;
 391                 ip_deq(q);
 392                 m_free(dtom(slirp, q));
 393         }
 394         remque(&fp->ip_link);
 395         (void) m_free(dtom(slirp, fp));
 396 }
 397
 398 /*
 399  * Put an ip fragment on a reassembly chain.
 400  * Like insque, but pointers in middle of structure.
 401  */
 402 static void
 403 ip_enq(register struct ipasfrag *p, register struct ipasfrag *prev)
 404 {
 405         DEBUG_CALL("ip_enq");
 406         DEBUG_ARG("prev = %p", prev);
 407         p->ipf_prev =  prev;
 408         p->ipf_next = prev->ipf_next;
 409         ((struct ipasfrag *)(prev->ipf_next))->ipf_prev = p;
 410         prev->ipf_next = p;
 411 }
 412
 413 /*
 414  * To ip_enq as remque is to insque.
 415  */
 416 static void
 417 ip_deq(register struct ipasfrag *p)
 418 {
 419         ((struct ipasfrag *)(p->ipf_prev))->ipf_next = p->ipf_next;
 420         ((struct ipasfrag *)(p->ipf_next))->ipf_prev = p->ipf_prev;
 421 }
 422
 423 /*
 424  * IP timer processing;
 425  * if a timer expires on a reassembly
 426  * queue, discard it.
 427  */
 428 void
 429 ip_slowtimo(Slirp *slirp)
 430 {
 431     struct qlink *l;
 432
 433         DEBUG_CALL("ip_slowtimo");
 434
 435     l = slirp->ipq.ip_link.next;
 436
 437         if (l == NULL)
 438            return;
 439
 440     while (l != &slirp->ipq.ip_link) {
 441         struct ipq *fp = container_of(l, struct ipq, ip_link);
 442         l = l->next;
 443                 if (--fp->ipq_ttl == 0) {
 444                         ip_freef(slirp, fp);
 445                 }
 446     }
 447 }
 448
 449 /*
 450  * Strip out IP options, at higher
 451  * level protocol in the kernel.
 452  * Second argument is buffer to which options
 453  * will be moved, and return value is their length.
 454  * (XXX) should be deleted; last arg currently ignored.
 455  */
 456 void
 457 ip_stripoptions(register struct mbuf *m, struct mbuf *mopt)
 458 {
 459         register int i;
 460         struct ip *ip = mtod(m, struct ip *);
 461         register char *opts;
 462         int olen;
 463
 464         olen = (ip->ip_hl<<2) - sizeof (struct ip);
 465         opts = (char *)(ip + 1);
 466         i = m->m_len - (sizeof (struct ip) + olen);
 467         memcpy(opts, opts  + olen, (unsigned)i);
 468         m->m_len -= olen;
 469
 470         ip->ip_hl = sizeof(struct ip) >> 2;
 471 }