]> git.proxmox.com Git - qemu.git/blob - slirp/ip_input.c
Merge remote-tracking branch 'sstabellini/saverestore-8' into staging
[qemu.git] / 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 <osdep.h>
43 #include "ip_icmp.h"
44
45 static struct ip *ip_reass(Slirp *slirp, struct ip *ip, struct ipq *fp);
46 static void ip_freef(Slirp *slirp, struct ipq *fp);
47 static void ip_enq(register struct ipasfrag *p,
48 register struct ipasfrag *prev);
49 static void ip_deq(register struct ipasfrag *p);
50
51 /*
52 * IP initialization: fill in IP protocol switch table.
53 * All protocols not implemented in kernel go to raw IP protocol handler.
54 */
55 void
56 ip_init(Slirp *slirp)
57 {
58 slirp->ipq.ip_link.next = slirp->ipq.ip_link.prev = &slirp->ipq.ip_link;
59 udp_init(slirp);
60 tcp_init(slirp);
61 icmp_init(slirp);
62 }
63
64 void ip_cleanup(Slirp *slirp)
65 {
66 udp_cleanup(slirp);
67 tcp_cleanup(slirp);
68 icmp_cleanup(slirp);
69 }
70
71 /*
72 * Ip input routine. Checksum and byte swap header. If fragmented
73 * try to reassemble. Process options. Pass to next level.
74 */
75 void
76 ip_input(struct mbuf *m)
77 {
78 Slirp *slirp = m->slirp;
79 register struct ip *ip;
80 int hlen;
81
82 DEBUG_CALL("ip_input");
83 DEBUG_ARG("m = %lx", (long)m);
84 DEBUG_ARG("m_len = %d", m->m_len);
85
86 if (m->m_len < sizeof (struct ip)) {
87 return;
88 }
89
90 ip = mtod(m, struct ip *);
91
92 if (ip->ip_v != IPVERSION) {
93 goto bad;
94 }
95
96 hlen = ip->ip_hl << 2;
97 if (hlen<sizeof(struct ip ) || hlen>m->m_len) {/* min header length */
98 goto bad; /* or packet too short */
99 }
100
101 /* keep ip header intact for ICMP reply
102 * ip->ip_sum = cksum(m, hlen);
103 * if (ip->ip_sum) {
104 */
105 if(cksum(m,hlen)) {
106 goto bad;
107 }
108
109 /*
110 * Convert fields to host representation.
111 */
112 NTOHS(ip->ip_len);
113 if (ip->ip_len < hlen) {
114 goto bad;
115 }
116 NTOHS(ip->ip_id);
117 NTOHS(ip->ip_off);
118
119 /*
120 * Check that the amount of data in the buffers
121 * is as at least much as the IP header would have us expect.
122 * Trim mbufs if longer than we expect.
123 * Drop packet if shorter than we expect.
124 */
125 if (m->m_len < ip->ip_len) {
126 goto bad;
127 }
128
129 /* Should drop packet if mbuf too long? hmmm... */
130 if (m->m_len > ip->ip_len)
131 m_adj(m, ip->ip_len - m->m_len);
132
133 /* check ip_ttl for a correct ICMP reply */
134 if(ip->ip_ttl==0) {
135 icmp_error(m, ICMP_TIMXCEED,ICMP_TIMXCEED_INTRANS, 0,"ttl");
136 goto bad;
137 }
138
139 /*
140 * If offset or IP_MF are set, must reassemble.
141 * Otherwise, nothing need be done.
142 * (We could look in the reassembly queue to see
143 * if the packet was previously fragmented,
144 * but it's not worth the time; just let them time out.)
145 *
146 * XXX This should fail, don't fragment yet
147 */
148 if (ip->ip_off &~ IP_DF) {
149 register struct ipq *fp;
150 struct qlink *l;
151 /*
152 * Look for queue of fragments
153 * of this datagram.
154 */
155 for (l = slirp->ipq.ip_link.next; l != &slirp->ipq.ip_link;
156 l = l->next) {
157 fp = container_of(l, struct ipq, ip_link);
158 if (ip->ip_id == fp->ipq_id &&
159 ip->ip_src.s_addr == fp->ipq_src.s_addr &&
160 ip->ip_dst.s_addr == fp->ipq_dst.s_addr &&
161 ip->ip_p == fp->ipq_p)
162 goto found;
163 }
164 fp = NULL;
165 found:
166
167 /*
168 * Adjust ip_len to not reflect header,
169 * set ip_mff if more fragments are expected,
170 * convert offset of this to bytes.
171 */
172 ip->ip_len -= hlen;
173 if (ip->ip_off & IP_MF)
174 ip->ip_tos |= 1;
175 else
176 ip->ip_tos &= ~1;
177
178 ip->ip_off <<= 3;
179
180 /*
181 * If datagram marked as having more fragments
182 * or if this is not the first fragment,
183 * attempt reassembly; if it succeeds, proceed.
184 */
185 if (ip->ip_tos & 1 || ip->ip_off) {
186 ip = ip_reass(slirp, ip, fp);
187 if (ip == NULL)
188 return;
189 m = dtom(slirp, ip);
190 } else
191 if (fp)
192 ip_freef(slirp, fp);
193
194 } else
195 ip->ip_len -= hlen;
196
197 /*
198 * Switch out to protocol's input routine.
199 */
200 switch (ip->ip_p) {
201 case IPPROTO_TCP:
202 tcp_input(m, hlen, (struct socket *)NULL);
203 break;
204 case IPPROTO_UDP:
205 udp_input(m, hlen);
206 break;
207 case IPPROTO_ICMP:
208 icmp_input(m, hlen);
209 break;
210 default:
211 m_free(m);
212 }
213 return;
214 bad:
215 m_free(m);
216 return;
217 }
218
219 #define iptofrag(P) ((struct ipasfrag *)(((char*)(P)) - sizeof(struct qlink)))
220 #define fragtoip(P) ((struct ip*)(((char*)(P)) + sizeof(struct qlink)))
221 /*
222 * Take incoming datagram fragment and try to
223 * reassemble it into whole datagram. If a chain for
224 * reassembly of this datagram already exists, then it
225 * is given as fp; otherwise have to make a chain.
226 */
227 static struct ip *
228 ip_reass(Slirp *slirp, struct ip *ip, struct ipq *fp)
229 {
230 register struct mbuf *m = dtom(slirp, ip);
231 register struct ipasfrag *q;
232 int hlen = ip->ip_hl << 2;
233 int i, next;
234
235 DEBUG_CALL("ip_reass");
236 DEBUG_ARG("ip = %lx", (long)ip);
237 DEBUG_ARG("fp = %lx", (long)fp);
238 DEBUG_ARG("m = %lx", (long)m);
239
240 /*
241 * Presence of header sizes in mbufs
242 * would confuse code below.
243 * Fragment m_data is concatenated.
244 */
245 m->m_data += hlen;
246 m->m_len -= hlen;
247
248 /*
249 * If first fragment to arrive, create a reassembly queue.
250 */
251 if (fp == NULL) {
252 struct mbuf *t = m_get(slirp);
253
254 if (t == NULL) {
255 goto dropfrag;
256 }
257 fp = mtod(t, struct ipq *);
258 insque(&fp->ip_link, &slirp->ipq.ip_link);
259 fp->ipq_ttl = IPFRAGTTL;
260 fp->ipq_p = ip->ip_p;
261 fp->ipq_id = ip->ip_id;
262 fp->frag_link.next = fp->frag_link.prev = &fp->frag_link;
263 fp->ipq_src = ip->ip_src;
264 fp->ipq_dst = ip->ip_dst;
265 q = (struct ipasfrag *)fp;
266 goto insert;
267 }
268
269 /*
270 * Find a segment which begins after this one does.
271 */
272 for (q = fp->frag_link.next; q != (struct ipasfrag *)&fp->frag_link;
273 q = q->ipf_next)
274 if (q->ipf_off > ip->ip_off)
275 break;
276
277 /*
278 * If there is a preceding segment, it may provide some of
279 * our data already. If so, drop the data from the incoming
280 * segment. If it provides all of our data, drop us.
281 */
282 if (q->ipf_prev != &fp->frag_link) {
283 struct ipasfrag *pq = q->ipf_prev;
284 i = pq->ipf_off + pq->ipf_len - ip->ip_off;
285 if (i > 0) {
286 if (i >= ip->ip_len)
287 goto dropfrag;
288 m_adj(dtom(slirp, ip), i);
289 ip->ip_off += i;
290 ip->ip_len -= i;
291 }
292 }
293
294 /*
295 * While we overlap succeeding segments trim them or,
296 * if they are completely covered, dequeue them.
297 */
298 while (q != (struct ipasfrag*)&fp->frag_link &&
299 ip->ip_off + ip->ip_len > q->ipf_off) {
300 i = (ip->ip_off + ip->ip_len) - q->ipf_off;
301 if (i < q->ipf_len) {
302 q->ipf_len -= i;
303 q->ipf_off += i;
304 m_adj(dtom(slirp, q), i);
305 break;
306 }
307 q = q->ipf_next;
308 m_free(dtom(slirp, q->ipf_prev));
309 ip_deq(q->ipf_prev);
310 }
311
312 insert:
313 /*
314 * Stick new segment in its place;
315 * check for complete reassembly.
316 */
317 ip_enq(iptofrag(ip), q->ipf_prev);
318 next = 0;
319 for (q = fp->frag_link.next; q != (struct ipasfrag*)&fp->frag_link;
320 q = q->ipf_next) {
321 if (q->ipf_off != next)
322 return NULL;
323 next += q->ipf_len;
324 }
325 if (((struct ipasfrag *)(q->ipf_prev))->ipf_tos & 1)
326 return NULL;
327
328 /*
329 * Reassembly is complete; concatenate fragments.
330 */
331 q = fp->frag_link.next;
332 m = dtom(slirp, q);
333
334 q = (struct ipasfrag *) q->ipf_next;
335 while (q != (struct ipasfrag*)&fp->frag_link) {
336 struct mbuf *t = dtom(slirp, q);
337 q = (struct ipasfrag *) q->ipf_next;
338 m_cat(m, t);
339 }
340
341 /*
342 * Create header for new ip packet by
343 * modifying header of first packet;
344 * dequeue and discard fragment reassembly header.
345 * Make header visible.
346 */
347 q = fp->frag_link.next;
348
349 /*
350 * If the fragments concatenated to an mbuf that's
351 * bigger than the total size of the fragment, then and
352 * m_ext buffer was alloced. But fp->ipq_next points to
353 * the old buffer (in the mbuf), so we must point ip
354 * into the new buffer.
355 */
356 if (m->m_flags & M_EXT) {
357 int delta = (char *)q - m->m_dat;
358 q = (struct ipasfrag *)(m->m_ext + delta);
359 }
360
361 ip = fragtoip(q);
362 ip->ip_len = next;
363 ip->ip_tos &= ~1;
364 ip->ip_src = fp->ipq_src;
365 ip->ip_dst = fp->ipq_dst;
366 remque(&fp->ip_link);
367 (void) m_free(dtom(slirp, fp));
368 m->m_len += (ip->ip_hl << 2);
369 m->m_data -= (ip->ip_hl << 2);
370
371 return ip;
372
373 dropfrag:
374 m_free(m);
375 return NULL;
376 }
377
378 /*
379 * Free a fragment reassembly header and all
380 * associated datagrams.
381 */
382 static void
383 ip_freef(Slirp *slirp, struct ipq *fp)
384 {
385 register struct ipasfrag *q, *p;
386
387 for (q = fp->frag_link.next; q != (struct ipasfrag*)&fp->frag_link; q = p) {
388 p = q->ipf_next;
389 ip_deq(q);
390 m_free(dtom(slirp, q));
391 }
392 remque(&fp->ip_link);
393 (void) m_free(dtom(slirp, fp));
394 }
395
396 /*
397 * Put an ip fragment on a reassembly chain.
398 * Like insque, but pointers in middle of structure.
399 */
400 static void
401 ip_enq(register struct ipasfrag *p, register struct ipasfrag *prev)
402 {
403 DEBUG_CALL("ip_enq");
404 DEBUG_ARG("prev = %lx", (long)prev);
405 p->ipf_prev = prev;
406 p->ipf_next = prev->ipf_next;
407 ((struct ipasfrag *)(prev->ipf_next))->ipf_prev = p;
408 prev->ipf_next = p;
409 }
410
411 /*
412 * To ip_enq as remque is to insque.
413 */
414 static void
415 ip_deq(register struct ipasfrag *p)
416 {
417 ((struct ipasfrag *)(p->ipf_prev))->ipf_next = p->ipf_next;
418 ((struct ipasfrag *)(p->ipf_next))->ipf_prev = p->ipf_prev;
419 }
420
421 /*
422 * IP timer processing;
423 * if a timer expires on a reassembly
424 * queue, discard it.
425 */
426 void
427 ip_slowtimo(Slirp *slirp)
428 {
429 struct qlink *l;
430
431 DEBUG_CALL("ip_slowtimo");
432
433 l = slirp->ipq.ip_link.next;
434
435 if (l == NULL)
436 return;
437
438 while (l != &slirp->ipq.ip_link) {
439 struct ipq *fp = container_of(l, struct ipq, ip_link);
440 l = l->next;
441 if (--fp->ipq_ttl == 0) {
442 ip_freef(slirp, fp);
443 }
444 }
445 }
446
447 /*
448 * Do option processing on a datagram,
449 * possibly discarding it if bad options are encountered,
450 * or forwarding it if source-routed.
451 * Returns 1 if packet has been forwarded/freed,
452 * 0 if the packet should be processed further.
453 */
454
455 #ifdef notdef
456
457 int
458 ip_dooptions(m)
459 struct mbuf *m;
460 {
461 register struct ip *ip = mtod(m, struct ip *);
462 register u_char *cp;
463 register struct ip_timestamp *ipt;
464 register struct in_ifaddr *ia;
465 int opt, optlen, cnt, off, code, type, forward = 0;
466 struct in_addr *sin, dst;
467 typedef uint32_t n_time;
468 n_time ntime;
469
470 dst = ip->ip_dst;
471 cp = (u_char *)(ip + 1);
472 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
473 for (; cnt > 0; cnt -= optlen, cp += optlen) {
474 opt = cp[IPOPT_OPTVAL];
475 if (opt == IPOPT_EOL)
476 break;
477 if (opt == IPOPT_NOP)
478 optlen = 1;
479 else {
480 optlen = cp[IPOPT_OLEN];
481 if (optlen <= 0 || optlen > cnt) {
482 code = &cp[IPOPT_OLEN] - (u_char *)ip;
483 goto bad;
484 }
485 }
486 switch (opt) {
487
488 default:
489 break;
490
491 /*
492 * Source routing with record.
493 * Find interface with current destination address.
494 * If none on this machine then drop if strictly routed,
495 * or do nothing if loosely routed.
496 * Record interface address and bring up next address
497 * component. If strictly routed make sure next
498 * address is on directly accessible net.
499 */
500 case IPOPT_LSRR:
501 case IPOPT_SSRR:
502 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
503 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
504 goto bad;
505 }
506 ipaddr.sin_addr = ip->ip_dst;
507 ia = (struct in_ifaddr *)
508 ifa_ifwithaddr((struct sockaddr *)&ipaddr);
509 if (ia == 0) {
510 if (opt == IPOPT_SSRR) {
511 type = ICMP_UNREACH;
512 code = ICMP_UNREACH_SRCFAIL;
513 goto bad;
514 }
515 /*
516 * Loose routing, and not at next destination
517 * yet; nothing to do except forward.
518 */
519 break;
520 }
521 off--; /* 0 origin */
522 if (off > optlen - sizeof(struct in_addr)) {
523 /*
524 * End of source route. Should be for us.
525 */
526 save_rte(cp, ip->ip_src);
527 break;
528 }
529 /*
530 * locate outgoing interface
531 */
532 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
533 sizeof(ipaddr.sin_addr));
534 if (opt == IPOPT_SSRR) {
535 #define INA struct in_ifaddr *
536 #define SA struct sockaddr *
537 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0)
538 ia = (INA)ifa_ifwithnet((SA)&ipaddr);
539 } else
540 ia = ip_rtaddr(ipaddr.sin_addr);
541 if (ia == 0) {
542 type = ICMP_UNREACH;
543 code = ICMP_UNREACH_SRCFAIL;
544 goto bad;
545 }
546 ip->ip_dst = ipaddr.sin_addr;
547 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr),
548 (caddr_t)(cp + off), sizeof(struct in_addr));
549 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
550 /*
551 * Let ip_intr's mcast routing check handle mcast pkts
552 */
553 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr));
554 break;
555
556 case IPOPT_RR:
557 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
558 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
559 goto bad;
560 }
561 /*
562 * If no space remains, ignore.
563 */
564 off--; /* 0 origin */
565 if (off > optlen - sizeof(struct in_addr))
566 break;
567 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
568 sizeof(ipaddr.sin_addr));
569 /*
570 * locate outgoing interface; if we're the destination,
571 * use the incoming interface (should be same).
572 */
573 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
574 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
575 type = ICMP_UNREACH;
576 code = ICMP_UNREACH_HOST;
577 goto bad;
578 }
579 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr),
580 (caddr_t)(cp + off), sizeof(struct in_addr));
581 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
582 break;
583
584 case IPOPT_TS:
585 code = cp - (u_char *)ip;
586 ipt = (struct ip_timestamp *)cp;
587 if (ipt->ipt_len < 5)
588 goto bad;
589 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
590 if (++ipt->ipt_oflw == 0)
591 goto bad;
592 break;
593 }
594 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
595 switch (ipt->ipt_flg) {
596
597 case IPOPT_TS_TSONLY:
598 break;
599
600 case IPOPT_TS_TSANDADDR:
601 if (ipt->ipt_ptr + sizeof(n_time) +
602 sizeof(struct in_addr) > ipt->ipt_len)
603 goto bad;
604 ipaddr.sin_addr = dst;
605 ia = (INA)ifaof_ i f p foraddr((SA)&ipaddr,
606 m->m_pkthdr.rcvif);
607 if (ia == 0)
608 continue;
609 bcopy((caddr_t)&IA_SIN(ia)->sin_addr,
610 (caddr_t)sin, sizeof(struct in_addr));
611 ipt->ipt_ptr += sizeof(struct in_addr);
612 break;
613
614 case IPOPT_TS_PRESPEC:
615 if (ipt->ipt_ptr + sizeof(n_time) +
616 sizeof(struct in_addr) > ipt->ipt_len)
617 goto bad;
618 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
619 sizeof(struct in_addr));
620 if (ifa_ifwithaddr((SA)&ipaddr) == 0)
621 continue;
622 ipt->ipt_ptr += sizeof(struct in_addr);
623 break;
624
625 default:
626 goto bad;
627 }
628 ntime = iptime();
629 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
630 sizeof(n_time));
631 ipt->ipt_ptr += sizeof(n_time);
632 }
633 }
634 if (forward) {
635 ip_forward(m, 1);
636 return (1);
637 }
638 return (0);
639 bad:
640 icmp_error(m, type, code, 0, 0);
641
642 return (1);
643 }
644
645 #endif /* notdef */
646
647 /*
648 * Strip out IP options, at higher
649 * level protocol in the kernel.
650 * Second argument is buffer to which options
651 * will be moved, and return value is their length.
652 * (XXX) should be deleted; last arg currently ignored.
653 */
654 void
655 ip_stripoptions(register struct mbuf *m, struct mbuf *mopt)
656 {
657 register int i;
658 struct ip *ip = mtod(m, struct ip *);
659 register caddr_t opts;
660 int olen;
661
662 olen = (ip->ip_hl<<2) - sizeof (struct ip);
663 opts = (caddr_t)(ip + 1);
664 i = m->m_len - (sizeof (struct ip) + olen);
665 memcpy(opts, opts + olen, (unsigned)i);
666 m->m_len -= olen;
667
668 ip->ip_hl = sizeof(struct ip) >> 2;
669 }