Fix typo in comment, by Andreas Faerber.
[qemu/dscho.git] / slirp / ip_input.c
blob4f5bfd9a564373b7f23717cefbf0a619daa01d25
1 /*
2 * Copyright (c) 1982, 1986, 1988, 1993
3 * The Regents of the University of California. All rights reserved.
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. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University 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.
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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.
33 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
34 * ip_input.c,v 1.11 1994/11/16 10:17:08 jkh Exp
38 * Changes and additions relating to SLiRP are
39 * Copyright (c) 1995 Danny Gasparovski.
41 * Please read the file COPYRIGHT for the
42 * terms and conditions of the copyright.
45 #include <slirp.h>
46 #include "ip_icmp.h"
48 int ip_defttl;
49 struct ipstat ipstat;
50 struct ipq ipq;
53 * IP initialization: fill in IP protocol switch table.
54 * All protocols not implemented in kernel go to raw IP protocol handler.
56 void
57 ip_init()
59 ipq.next = ipq.prev = (ipqp_32)&ipq;
60 ip_id = tt.tv_sec & 0xffff;
61 udp_init();
62 tcp_init();
63 ip_defttl = IPDEFTTL;
67 * Ip input routine. Checksum and byte swap header. If fragmented
68 * try to reassemble. Process options. Pass to next level.
70 void
71 ip_input(m)
72 struct mbuf *m;
74 register struct ip *ip;
75 int hlen;
77 DEBUG_CALL("ip_input");
78 DEBUG_ARG("m = %lx", (long)m);
79 DEBUG_ARG("m_len = %d", m->m_len);
81 ipstat.ips_total++;
83 if (m->m_len < sizeof (struct ip)) {
84 ipstat.ips_toosmall++;
85 return;
88 ip = mtod(m, struct ip *);
90 if (ip->ip_v != IPVERSION) {
91 ipstat.ips_badvers++;
92 goto bad;
95 hlen = ip->ip_hl << 2;
96 if (hlen<sizeof(struct ip ) || hlen>m->m_len) {/* min header length */
97 ipstat.ips_badhlen++; /* or packet too short */
98 goto bad;
101 /* keep ip header intact for ICMP reply
102 * ip->ip_sum = cksum(m, hlen);
103 * if (ip->ip_sum) {
105 if(cksum(m,hlen)) {
106 ipstat.ips_badsum++;
107 goto bad;
111 * Convert fields to host representation.
113 NTOHS(ip->ip_len);
114 if (ip->ip_len < hlen) {
115 ipstat.ips_badlen++;
116 goto bad;
118 NTOHS(ip->ip_id);
119 NTOHS(ip->ip_off);
122 * Check that the amount of data in the buffers
123 * is as at least much as the IP header would have us expect.
124 * Trim mbufs if longer than we expect.
125 * Drop packet if shorter than we expect.
127 if (m->m_len < ip->ip_len) {
128 ipstat.ips_tooshort++;
129 goto bad;
131 /* Should drop packet if mbuf too long? hmmm... */
132 if (m->m_len > ip->ip_len)
133 m_adj(m, ip->ip_len - m->m_len);
135 /* check ip_ttl for a correct ICMP reply */
136 if(ip->ip_ttl==0 || ip->ip_ttl==1) {
137 icmp_error(m, ICMP_TIMXCEED,ICMP_TIMXCEED_INTRANS, 0,"ttl");
138 goto bad;
142 * Process options and, if not destined for us,
143 * ship it on. ip_dooptions returns 1 when an
144 * error was detected (causing an icmp message
145 * to be sent and the original packet to be freed).
147 /* We do no IP options */
148 /* if (hlen > sizeof (struct ip) && ip_dooptions(m))
149 * goto next;
152 * If offset or IP_MF are set, must reassemble.
153 * Otherwise, nothing need be done.
154 * (We could look in the reassembly queue to see
155 * if the packet was previously fragmented,
156 * but it's not worth the time; just let them time out.)
158 * XXX This should fail, don't fragment yet
160 if (ip->ip_off &~ IP_DF) {
161 register struct ipq *fp;
163 * Look for queue of fragments
164 * of this datagram.
166 for (fp = (struct ipq *) ipq.next; fp != &ipq;
167 fp = (struct ipq *) fp->next)
168 if (ip->ip_id == fp->ipq_id &&
169 ip->ip_src.s_addr == fp->ipq_src.s_addr &&
170 ip->ip_dst.s_addr == fp->ipq_dst.s_addr &&
171 ip->ip_p == fp->ipq_p)
172 goto found;
173 fp = 0;
174 found:
177 * Adjust ip_len to not reflect header,
178 * set ip_mff if more fragments are expected,
179 * convert offset of this to bytes.
181 ip->ip_len -= hlen;
182 if (ip->ip_off & IP_MF)
183 ((struct ipasfrag *)ip)->ipf_mff |= 1;
184 else
185 ((struct ipasfrag *)ip)->ipf_mff &= ~1;
187 ip->ip_off <<= 3;
190 * If datagram marked as having more fragments
191 * or if this is not the first fragment,
192 * attempt reassembly; if it succeeds, proceed.
194 if (((struct ipasfrag *)ip)->ipf_mff & 1 || ip->ip_off) {
195 ipstat.ips_fragments++;
196 ip = ip_reass((struct ipasfrag *)ip, fp);
197 if (ip == 0)
198 return;
199 ipstat.ips_reassembled++;
200 m = dtom(ip);
201 } else
202 if (fp)
203 ip_freef(fp);
205 } else
206 ip->ip_len -= hlen;
209 * Switch out to protocol's input routine.
211 ipstat.ips_delivered++;
212 switch (ip->ip_p) {
213 case IPPROTO_TCP:
214 tcp_input(m, hlen, (struct socket *)NULL);
215 break;
216 case IPPROTO_UDP:
217 udp_input(m, hlen);
218 break;
219 case IPPROTO_ICMP:
220 icmp_input(m, hlen);
221 break;
222 default:
223 ipstat.ips_noproto++;
224 m_free(m);
226 return;
227 bad:
228 m_freem(m);
229 return;
233 * Take incoming datagram fragment and try to
234 * reassemble it into whole datagram. If a chain for
235 * reassembly of this datagram already exists, then it
236 * is given as fp; otherwise have to make a chain.
238 struct ip *
239 ip_reass(ip, fp)
240 register struct ipasfrag *ip;
241 register struct ipq *fp;
243 register struct mbuf *m = dtom(ip);
244 register struct ipasfrag *q;
245 int hlen = ip->ip_hl << 2;
246 int i, next;
248 DEBUG_CALL("ip_reass");
249 DEBUG_ARG("ip = %lx", (long)ip);
250 DEBUG_ARG("fp = %lx", (long)fp);
251 DEBUG_ARG("m = %lx", (long)m);
254 * Presence of header sizes in mbufs
255 * would confuse code below.
256 * Fragment m_data is concatenated.
258 m->m_data += hlen;
259 m->m_len -= hlen;
262 * If first fragment to arrive, create a reassembly queue.
264 if (fp == 0) {
265 struct mbuf *t;
266 if ((t = m_get()) == NULL) goto dropfrag;
267 fp = mtod(t, struct ipq *);
268 insque_32(fp, &ipq);
269 fp->ipq_ttl = IPFRAGTTL;
270 fp->ipq_p = ip->ip_p;
271 fp->ipq_id = ip->ip_id;
272 fp->ipq_next = fp->ipq_prev = (ipasfragp_32)fp;
273 fp->ipq_src = ((struct ip *)ip)->ip_src;
274 fp->ipq_dst = ((struct ip *)ip)->ip_dst;
275 q = (struct ipasfrag *)fp;
276 goto insert;
280 * Find a segment which begins after this one does.
282 for (q = (struct ipasfrag *)fp->ipq_next; q != (struct ipasfrag *)fp;
283 q = (struct ipasfrag *)q->ipf_next)
284 if (q->ip_off > ip->ip_off)
285 break;
288 * If there is a preceding segment, it may provide some of
289 * our data already. If so, drop the data from the incoming
290 * segment. If it provides all of our data, drop us.
292 if (q->ipf_prev != (ipasfragp_32)fp) {
293 i = ((struct ipasfrag *)(q->ipf_prev))->ip_off +
294 ((struct ipasfrag *)(q->ipf_prev))->ip_len - ip->ip_off;
295 if (i > 0) {
296 if (i >= ip->ip_len)
297 goto dropfrag;
298 m_adj(dtom(ip), i);
299 ip->ip_off += i;
300 ip->ip_len -= i;
305 * While we overlap succeeding segments trim them or,
306 * if they are completely covered, dequeue them.
308 while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) {
309 i = (ip->ip_off + ip->ip_len) - q->ip_off;
310 if (i < q->ip_len) {
311 q->ip_len -= i;
312 q->ip_off += i;
313 m_adj(dtom(q), i);
314 break;
316 q = (struct ipasfrag *) q->ipf_next;
317 m_freem(dtom((struct ipasfrag *) q->ipf_prev));
318 ip_deq((struct ipasfrag *) q->ipf_prev);
321 insert:
323 * Stick new segment in its place;
324 * check for complete reassembly.
326 ip_enq(ip, (struct ipasfrag *) q->ipf_prev);
327 next = 0;
328 for (q = (struct ipasfrag *) fp->ipq_next; q != (struct ipasfrag *)fp;
329 q = (struct ipasfrag *) q->ipf_next) {
330 if (q->ip_off != next)
331 return (0);
332 next += q->ip_len;
334 if (((struct ipasfrag *)(q->ipf_prev))->ipf_mff & 1)
335 return (0);
338 * Reassembly is complete; concatenate fragments.
340 q = (struct ipasfrag *) fp->ipq_next;
341 m = dtom(q);
343 q = (struct ipasfrag *) q->ipf_next;
344 while (q != (struct ipasfrag *)fp) {
345 struct mbuf *t;
346 t = dtom(q);
347 q = (struct ipasfrag *) q->ipf_next;
348 m_cat(m, t);
352 * Create header for new ip packet by
353 * modifying header of first packet;
354 * dequeue and discard fragment reassembly header.
355 * Make header visible.
357 ip = (struct ipasfrag *) fp->ipq_next;
360 * If the fragments concatenated to an mbuf that's
361 * bigger than the total size of the fragment, then and
362 * m_ext buffer was alloced. But fp->ipq_next points to
363 * the old buffer (in the mbuf), so we must point ip
364 * into the new buffer.
366 if (m->m_flags & M_EXT) {
367 int delta;
368 delta = (char *)ip - m->m_dat;
369 ip = (struct ipasfrag *)(m->m_ext + delta);
372 /* DEBUG_ARG("ip = %lx", (long)ip);
373 * ip=(struct ipasfrag *)m->m_data; */
375 ip->ip_len = next;
376 ip->ipf_mff &= ~1;
377 ((struct ip *)ip)->ip_src = fp->ipq_src;
378 ((struct ip *)ip)->ip_dst = fp->ipq_dst;
379 remque_32(fp);
380 (void) m_free(dtom(fp));
381 m = dtom(ip);
382 m->m_len += (ip->ip_hl << 2);
383 m->m_data -= (ip->ip_hl << 2);
385 return ((struct ip *)ip);
387 dropfrag:
388 ipstat.ips_fragdropped++;
389 m_freem(m);
390 return (0);
394 * Free a fragment reassembly header and all
395 * associated datagrams.
397 void
398 ip_freef(fp)
399 struct ipq *fp;
401 register struct ipasfrag *q, *p;
403 for (q = (struct ipasfrag *) fp->ipq_next; q != (struct ipasfrag *)fp;
404 q = p) {
405 p = (struct ipasfrag *) q->ipf_next;
406 ip_deq(q);
407 m_freem(dtom(q));
409 remque_32(fp);
410 (void) m_free(dtom(fp));
414 * Put an ip fragment on a reassembly chain.
415 * Like insque, but pointers in middle of structure.
417 void
418 ip_enq(p, prev)
419 register struct ipasfrag *p, *prev;
421 DEBUG_CALL("ip_enq");
422 DEBUG_ARG("prev = %lx", (long)prev);
423 p->ipf_prev = (ipasfragp_32) prev;
424 p->ipf_next = prev->ipf_next;
425 ((struct ipasfrag *)(prev->ipf_next))->ipf_prev = (ipasfragp_32) p;
426 prev->ipf_next = (ipasfragp_32) p;
430 * To ip_enq as remque is to insque.
432 void
433 ip_deq(p)
434 register struct ipasfrag *p;
436 ((struct ipasfrag *)(p->ipf_prev))->ipf_next = p->ipf_next;
437 ((struct ipasfrag *)(p->ipf_next))->ipf_prev = p->ipf_prev;
441 * IP timer processing;
442 * if a timer expires on a reassembly
443 * queue, discard it.
445 void
446 ip_slowtimo()
448 register struct ipq *fp;
450 DEBUG_CALL("ip_slowtimo");
452 fp = (struct ipq *) ipq.next;
453 if (fp == 0)
454 return;
456 while (fp != &ipq) {
457 --fp->ipq_ttl;
458 fp = (struct ipq *) fp->next;
459 if (((struct ipq *)(fp->prev))->ipq_ttl == 0) {
460 ipstat.ips_fragtimeout++;
461 ip_freef((struct ipq *) fp->prev);
467 * Do option processing on a datagram,
468 * possibly discarding it if bad options are encountered,
469 * or forwarding it if source-routed.
470 * Returns 1 if packet has been forwarded/freed,
471 * 0 if the packet should be processed further.
474 #ifdef notdef
477 ip_dooptions(m)
478 struct mbuf *m;
480 register struct ip *ip = mtod(m, struct ip *);
481 register u_char *cp;
482 register struct ip_timestamp *ipt;
483 register struct in_ifaddr *ia;
484 /* int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; */
485 int opt, optlen, cnt, off, code, type, forward = 0;
486 struct in_addr *sin, dst;
487 typedef u_int32_t n_time;
488 n_time ntime;
490 dst = ip->ip_dst;
491 cp = (u_char *)(ip + 1);
492 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
493 for (; cnt > 0; cnt -= optlen, cp += optlen) {
494 opt = cp[IPOPT_OPTVAL];
495 if (opt == IPOPT_EOL)
496 break;
497 if (opt == IPOPT_NOP)
498 optlen = 1;
499 else {
500 optlen = cp[IPOPT_OLEN];
501 if (optlen <= 0 || optlen > cnt) {
502 code = &cp[IPOPT_OLEN] - (u_char *)ip;
503 goto bad;
506 switch (opt) {
508 default:
509 break;
512 * Source routing with record.
513 * Find interface with current destination address.
514 * If none on this machine then drop if strictly routed,
515 * or do nothing if loosely routed.
516 * Record interface address and bring up next address
517 * component. If strictly routed make sure next
518 * address is on directly accessible net.
520 case IPOPT_LSRR:
521 case IPOPT_SSRR:
522 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
523 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
524 goto bad;
526 ipaddr.sin_addr = ip->ip_dst;
527 ia = (struct in_ifaddr *)
528 ifa_ifwithaddr((struct sockaddr *)&ipaddr);
529 if (ia == 0) {
530 if (opt == IPOPT_SSRR) {
531 type = ICMP_UNREACH;
532 code = ICMP_UNREACH_SRCFAIL;
533 goto bad;
536 * Loose routing, and not at next destination
537 * yet; nothing to do except forward.
539 break;
541 off--; / * 0 origin * /
542 if (off > optlen - sizeof(struct in_addr)) {
544 * End of source route. Should be for us.
546 save_rte(cp, ip->ip_src);
547 break;
550 * locate outgoing interface
552 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
553 sizeof(ipaddr.sin_addr));
554 if (opt == IPOPT_SSRR) {
555 #define INA struct in_ifaddr *
556 #define SA struct sockaddr *
557 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0)
558 ia = (INA)ifa_ifwithnet((SA)&ipaddr);
559 } else
560 ia = ip_rtaddr(ipaddr.sin_addr);
561 if (ia == 0) {
562 type = ICMP_UNREACH;
563 code = ICMP_UNREACH_SRCFAIL;
564 goto bad;
566 ip->ip_dst = ipaddr.sin_addr;
567 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr),
568 (caddr_t)(cp + off), sizeof(struct in_addr));
569 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
571 * Let ip_intr's mcast routing check handle mcast pkts
573 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr));
574 break;
576 case IPOPT_RR:
577 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
578 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
579 goto bad;
582 * If no space remains, ignore.
584 off--; * 0 origin *
585 if (off > optlen - sizeof(struct in_addr))
586 break;
587 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
588 sizeof(ipaddr.sin_addr));
590 * locate outgoing interface; if we're the destination,
591 * use the incoming interface (should be same).
593 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
594 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
595 type = ICMP_UNREACH;
596 code = ICMP_UNREACH_HOST;
597 goto bad;
599 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr),
600 (caddr_t)(cp + off), sizeof(struct in_addr));
601 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
602 break;
604 case IPOPT_TS:
605 code = cp - (u_char *)ip;
606 ipt = (struct ip_timestamp *)cp;
607 if (ipt->ipt_len < 5)
608 goto bad;
609 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
610 if (++ipt->ipt_oflw == 0)
611 goto bad;
612 break;
614 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
615 switch (ipt->ipt_flg) {
617 case IPOPT_TS_TSONLY:
618 break;
620 case IPOPT_TS_TSANDADDR:
621 if (ipt->ipt_ptr + sizeof(n_time) +
622 sizeof(struct in_addr) > ipt->ipt_len)
623 goto bad;
624 ipaddr.sin_addr = dst;
625 ia = (INA)ifaof_ i f p foraddr((SA)&ipaddr,
626 m->m_pkthdr.rcvif);
627 if (ia == 0)
628 continue;
629 bcopy((caddr_t)&IA_SIN(ia)->sin_addr,
630 (caddr_t)sin, sizeof(struct in_addr));
631 ipt->ipt_ptr += sizeof(struct in_addr);
632 break;
634 case IPOPT_TS_PRESPEC:
635 if (ipt->ipt_ptr + sizeof(n_time) +
636 sizeof(struct in_addr) > ipt->ipt_len)
637 goto bad;
638 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
639 sizeof(struct in_addr));
640 if (ifa_ifwithaddr((SA)&ipaddr) == 0)
641 continue;
642 ipt->ipt_ptr += sizeof(struct in_addr);
643 break;
645 default:
646 goto bad;
648 ntime = iptime();
649 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
650 sizeof(n_time));
651 ipt->ipt_ptr += sizeof(n_time);
654 if (forward) {
655 ip_forward(m, 1);
656 return (1);
660 return (0);
661 bad:
662 /* ip->ip_len -= ip->ip_hl << 2; XXX icmp_error adds in hdr length */
664 /* Not yet */
665 icmp_error(m, type, code, 0, 0);
667 ipstat.ips_badoptions++;
668 return (1);
671 #endif /* notdef */
674 * Strip out IP options, at higher
675 * level protocol in the kernel.
676 * Second argument is buffer to which options
677 * will be moved, and return value is their length.
678 * (XXX) should be deleted; last arg currently ignored.
680 void
681 ip_stripoptions(m, mopt)
682 register struct mbuf *m;
683 struct mbuf *mopt;
685 register int i;
686 struct ip *ip = mtod(m, struct ip *);
687 register caddr_t opts;
688 int olen;
690 olen = (ip->ip_hl<<2) - sizeof (struct ip);
691 opts = (caddr_t)(ip + 1);
692 i = m->m_len - (sizeof (struct ip) + olen);
693 memcpy(opts, opts + olen, (unsigned)i);
694 m->m_len -= olen;
696 ip->ip_hl = sizeof(struct ip) >> 2;