pcie_aer: support configurable AER capa version
[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 "qemu/osdep.h"
42 #include "slirp.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 if (!slirp->in_enabled) {
83 goto bad;
84 }
85
86 DEBUG_CALL("ip_input");
87 DEBUG_ARG("m = %p", m);
88 DEBUG_ARG("m_len = %d", m->m_len);
89
90 if (m->m_len < sizeof (struct ip)) {
91 goto bad;
92 }
93
94 ip = mtod(m, struct ip *);
95
96 if (ip->ip_v != IPVERSION) {
97 goto bad;
98 }
99
100 hlen = ip->ip_hl << 2;
101 if (hlen<sizeof(struct ip ) || hlen>m->m_len) {/* min header length */
102 goto bad; /* or packet too short */
103 }
104
105 /* keep ip header intact for ICMP reply
106 * ip->ip_sum = cksum(m, hlen);
107 * if (ip->ip_sum) {
108 */
109 if(cksum(m,hlen)) {
110 goto bad;
111 }
112
113 /*
114 * Convert fields to host representation.
115 */
116 NTOHS(ip->ip_len);
117 if (ip->ip_len < hlen) {
118 goto bad;
119 }
120 NTOHS(ip->ip_id);
121 NTOHS(ip->ip_off);
122
123 /*
124 * Check that the amount of data in the buffers
125 * is as at least much as the IP header would have us expect.
126 * Trim mbufs if longer than we expect.
127 * Drop packet if shorter than we expect.
128 */
129 if (m->m_len < ip->ip_len) {
130 goto bad;
131 }
132
133 /* Should drop packet if mbuf too long? hmmm... */
134 if (m->m_len > ip->ip_len)
135 m_adj(m, ip->ip_len - m->m_len);
136
137 /* check ip_ttl for a correct ICMP reply */
138 if (ip->ip_ttl == 0) {
139 icmp_send_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, 0, "ttl");
140 goto bad;
141 }
142
143 /*
144 * If offset or IP_MF are set, must reassemble.
145 * Otherwise, nothing need be done.
146 * (We could look in the reassembly queue to see
147 * if the packet was previously fragmented,
148 * but it's not worth the time; just let them time out.)
149 *
150 * XXX This should fail, don't fragment yet
151 */
152 if (ip->ip_off &~ IP_DF) {
153 register struct ipq *fp;
154 struct qlink *l;
155 /*
156 * Look for queue of fragments
157 * of this datagram.
158 */
159 for (l = slirp->ipq.ip_link.next; l != &slirp->ipq.ip_link;
160 l = l->next) {
161 fp = container_of(l, struct ipq, ip_link);
162 if (ip->ip_id == fp->ipq_id &&
163 ip->ip_src.s_addr == fp->ipq_src.s_addr &&
164 ip->ip_dst.s_addr == fp->ipq_dst.s_addr &&
165 ip->ip_p == fp->ipq_p)
166 goto found;
167 }
168 fp = NULL;
169 found:
170
171 /*
172 * Adjust ip_len to not reflect header,
173 * set ip_mff if more fragments are expected,
174 * convert offset of this to bytes.
175 */
176 ip->ip_len -= hlen;
177 if (ip->ip_off & IP_MF)
178 ip->ip_tos |= 1;
179 else
180 ip->ip_tos &= ~1;
181
182 ip->ip_off <<= 3;
183
184 /*
185 * If datagram marked as having more fragments
186 * or if this is not the first fragment,
187 * attempt reassembly; if it succeeds, proceed.
188 */
189 if (ip->ip_tos & 1 || ip->ip_off) {
190 ip = ip_reass(slirp, ip, fp);
191 if (ip == NULL)
192 return;
193 m = dtom(slirp, ip);
194 } else
195 if (fp)
196 ip_freef(slirp, fp);
197
198 } else
199 ip->ip_len -= hlen;
200
201 /*
202 * Switch out to protocol's input routine.
203 */
204 switch (ip->ip_p) {
205 case IPPROTO_TCP:
206 tcp_input(m, hlen, (struct socket *)NULL, AF_INET);
207 break;
208 case IPPROTO_UDP:
209 udp_input(m, hlen);
210 break;
211 case IPPROTO_ICMP:
212 icmp_input(m, hlen);
213 break;
214 default:
215 m_free(m);
216 }
217 return;
218 bad:
219 m_free(m);
220 }
221
222 #define iptofrag(P) ((struct ipasfrag *)(((char*)(P)) - sizeof(struct qlink)))
223 #define fragtoip(P) ((struct ip*)(((char*)(P)) + sizeof(struct qlink)))
224 /*
225 * Take incoming datagram fragment and try to
226 * reassemble it into whole datagram. If a chain for
227 * reassembly of this datagram already exists, then it
228 * is given as fp; otherwise have to make a chain.
229 */
230 static struct ip *
231 ip_reass(Slirp *slirp, struct ip *ip, struct ipq *fp)
232 {
233 register struct mbuf *m = dtom(slirp, ip);
234 register struct ipasfrag *q;
235 int hlen = ip->ip_hl << 2;
236 int i, next;
237
238 DEBUG_CALL("ip_reass");
239 DEBUG_ARG("ip = %p", ip);
240 DEBUG_ARG("fp = %p", fp);
241 DEBUG_ARG("m = %p", m);
242
243 /*
244 * Presence of header sizes in mbufs
245 * would confuse code below.
246 * Fragment m_data is concatenated.
247 */
248 m->m_data += hlen;
249 m->m_len -= hlen;
250
251 /*
252 * If first fragment to arrive, create a reassembly queue.
253 */
254 if (fp == NULL) {
255 struct mbuf *t = m_get(slirp);
256
257 if (t == NULL) {
258 goto dropfrag;
259 }
260 fp = mtod(t, struct ipq *);
261 insque(&fp->ip_link, &slirp->ipq.ip_link);
262 fp->ipq_ttl = IPFRAGTTL;
263 fp->ipq_p = ip->ip_p;
264 fp->ipq_id = ip->ip_id;
265 fp->frag_link.next = fp->frag_link.prev = &fp->frag_link;
266 fp->ipq_src = ip->ip_src;
267 fp->ipq_dst = ip->ip_dst;
268 q = (struct ipasfrag *)fp;
269 goto insert;
270 }
271
272 /*
273 * Find a segment which begins after this one does.
274 */
275 for (q = fp->frag_link.next; q != (struct ipasfrag *)&fp->frag_link;
276 q = q->ipf_next)
277 if (q->ipf_off > ip->ip_off)
278 break;
279
280 /*
281 * If there is a preceding segment, it may provide some of
282 * our data already. If so, drop the data from the incoming
283 * segment. If it provides all of our data, drop us.
284 */
285 if (q->ipf_prev != &fp->frag_link) {
286 struct ipasfrag *pq = q->ipf_prev;
287 i = pq->ipf_off + pq->ipf_len - ip->ip_off;
288 if (i > 0) {
289 if (i >= ip->ip_len)
290 goto dropfrag;
291 m_adj(dtom(slirp, ip), i);
292 ip->ip_off += i;
293 ip->ip_len -= i;
294 }
295 }
296
297 /*
298 * While we overlap succeeding segments trim them or,
299 * if they are completely covered, dequeue them.
300 */
301 while (q != (struct ipasfrag*)&fp->frag_link &&
302 ip->ip_off + ip->ip_len > q->ipf_off) {
303 i = (ip->ip_off + ip->ip_len) - q->ipf_off;
304 if (i < q->ipf_len) {
305 q->ipf_len -= i;
306 q->ipf_off += i;
307 m_adj(dtom(slirp, q), i);
308 break;
309 }
310 q = q->ipf_next;
311 m_free(dtom(slirp, q->ipf_prev));
312 ip_deq(q->ipf_prev);
313 }
314
315 insert:
316 /*
317 * Stick new segment in its place;
318 * check for complete reassembly.
319 */
320 ip_enq(iptofrag(ip), q->ipf_prev);
321 next = 0;
322 for (q = fp->frag_link.next; q != (struct ipasfrag*)&fp->frag_link;
323 q = q->ipf_next) {
324 if (q->ipf_off != next)
325 return NULL;
326 next += q->ipf_len;
327 }
328 if (((struct ipasfrag *)(q->ipf_prev))->ipf_tos & 1)
329 return NULL;
330
331 /*
332 * Reassembly is complete; concatenate fragments.
333 */
334 q = fp->frag_link.next;
335 m = dtom(slirp, q);
336
337 q = (struct ipasfrag *) q->ipf_next;
338 while (q != (struct ipasfrag*)&fp->frag_link) {
339 struct mbuf *t = dtom(slirp, q);
340 q = (struct ipasfrag *) q->ipf_next;
341 m_cat(m, t);
342 }
343
344 /*
345 * Create header for new ip packet by
346 * modifying header of first packet;
347 * dequeue and discard fragment reassembly header.
348 * Make header visible.
349 */
350 q = fp->frag_link.next;
351
352 /*
353 * If the fragments concatenated to an mbuf that's
354 * bigger than the total size of the fragment, then and
355 * m_ext buffer was alloced. But fp->ipq_next points to
356 * the old buffer (in the mbuf), so we must point ip
357 * into the new buffer.
358 */
359 if (m->m_flags & M_EXT) {
360 int delta = (char *)q - m->m_dat;
361 q = (struct ipasfrag *)(m->m_ext + delta);
362 }
363
364 ip = fragtoip(q);
365 ip->ip_len = next;
366 ip->ip_tos &= ~1;
367 ip->ip_src = fp->ipq_src;
368 ip->ip_dst = fp->ipq_dst;
369 remque(&fp->ip_link);
370 (void) m_free(dtom(slirp, fp));
371 m->m_len += (ip->ip_hl << 2);
372 m->m_data -= (ip->ip_hl << 2);
373
374 return ip;
375
376 dropfrag:
377 m_free(m);
378 return NULL;
379 }
380
381 /*
382 * Free a fragment reassembly header and all
383 * associated datagrams.
384 */
385 static void
386 ip_freef(Slirp *slirp, struct ipq *fp)
387 {
388 register struct ipasfrag *q, *p;
389
390 for (q = fp->frag_link.next; q != (struct ipasfrag*)&fp->frag_link; q = p) {
391 p = q->ipf_next;
392 ip_deq(q);
393 m_free(dtom(slirp, q));
394 }
395 remque(&fp->ip_link);
396 (void) m_free(dtom(slirp, fp));
397 }
398
399 /*
400 * Put an ip fragment on a reassembly chain.
401 * Like insque, but pointers in middle of structure.
402 */
403 static void
404 ip_enq(register struct ipasfrag *p, register struct ipasfrag *prev)
405 {
406 DEBUG_CALL("ip_enq");
407 DEBUG_ARG("prev = %p", prev);
408 p->ipf_prev = prev;
409 p->ipf_next = prev->ipf_next;
410 ((struct ipasfrag *)(prev->ipf_next))->ipf_prev = p;
411 prev->ipf_next = p;
412 }
413
414 /*
415 * To ip_enq as remque is to insque.
416 */
417 static void
418 ip_deq(register struct ipasfrag *p)
419 {
420 ((struct ipasfrag *)(p->ipf_prev))->ipf_next = p->ipf_next;
421 ((struct ipasfrag *)(p->ipf_next))->ipf_prev = p->ipf_prev;
422 }
423
424 /*
425 * IP timer processing;
426 * if a timer expires on a reassembly
427 * queue, discard it.
428 */
429 void
430 ip_slowtimo(Slirp *slirp)
431 {
432 struct qlink *l;
433
434 DEBUG_CALL("ip_slowtimo");
435
436 l = slirp->ipq.ip_link.next;
437
438 if (l == NULL)
439 return;
440
441 while (l != &slirp->ipq.ip_link) {
442 struct ipq *fp = container_of(l, struct ipq, ip_link);
443 l = l->next;
444 if (--fp->ipq_ttl == 0) {
445 ip_freef(slirp, fp);
446 }
447 }
448 }
449
450 /*
451 * Do option processing on a datagram,
452 * possibly discarding it if bad options are encountered,
453 * or forwarding it if source-routed.
454 * Returns 1 if packet has been forwarded/freed,
455 * 0 if the packet should be processed further.
456 */
457
458 #ifdef notdef
459
460 int
461 ip_dooptions(m)
462 struct mbuf *m;
463 {
464 register struct ip *ip = mtod(m, struct ip *);
465 register u_char *cp;
466 register struct ip_timestamp *ipt;
467 register struct in_ifaddr *ia;
468 int opt, optlen, cnt, off, code, type, forward = 0;
469 struct in_addr *sin, dst;
470 typedef uint32_t n_time;
471 n_time ntime;
472
473 dst = ip->ip_dst;
474 cp = (u_char *)(ip + 1);
475 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
476 for (; cnt > 0; cnt -= optlen, cp += optlen) {
477 opt = cp[IPOPT_OPTVAL];
478 if (opt == IPOPT_EOL)
479 break;
480 if (opt == IPOPT_NOP)
481 optlen = 1;
482 else {
483 optlen = cp[IPOPT_OLEN];
484 if (optlen <= 0 || optlen > cnt) {
485 code = &cp[IPOPT_OLEN] - (u_char *)ip;
486 goto bad;
487 }
488 }
489 switch (opt) {
490
491 default:
492 break;
493
494 /*
495 * Source routing with record.
496 * Find interface with current destination address.
497 * If none on this machine then drop if strictly routed,
498 * or do nothing if loosely routed.
499 * Record interface address and bring up next address
500 * component. If strictly routed make sure next
501 * address is on directly accessible net.
502 */
503 case IPOPT_LSRR:
504 case IPOPT_SSRR:
505 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
506 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
507 goto bad;
508 }
509 ipaddr.sin_addr = ip->ip_dst;
510 ia = (struct in_ifaddr *)
511 ifa_ifwithaddr((struct sockaddr *)&ipaddr);
512 if (ia == 0) {
513 if (opt == IPOPT_SSRR) {
514 type = ICMP_UNREACH;
515 code = ICMP_UNREACH_SRCFAIL;
516 goto bad;
517 }
518 /*
519 * Loose routing, and not at next destination
520 * yet; nothing to do except forward.
521 */
522 break;
523 }
524 off--; /* 0 origin */
525 if (off > optlen - sizeof(struct in_addr)) {
526 /*
527 * End of source route. Should be for us.
528 */
529 save_rte(cp, ip->ip_src);
530 break;
531 }
532 /*
533 * locate outgoing interface
534 */
535 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
536 sizeof(ipaddr.sin_addr));
537 if (opt == IPOPT_SSRR) {
538 #define INA struct in_ifaddr *
539 #define SA struct sockaddr *
540 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0)
541 ia = (INA)ifa_ifwithnet((SA)&ipaddr);
542 } else
543 ia = ip_rtaddr(ipaddr.sin_addr);
544 if (ia == 0) {
545 type = ICMP_UNREACH;
546 code = ICMP_UNREACH_SRCFAIL;
547 goto bad;
548 }
549 ip->ip_dst = ipaddr.sin_addr;
550 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr),
551 (caddr_t)(cp + off), sizeof(struct in_addr));
552 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
553 /*
554 * Let ip_intr's mcast routing check handle mcast pkts
555 */
556 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr));
557 break;
558
559 case IPOPT_RR:
560 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
561 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
562 goto bad;
563 }
564 /*
565 * If no space remains, ignore.
566 */
567 off--; /* 0 origin */
568 if (off > optlen - sizeof(struct in_addr))
569 break;
570 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
571 sizeof(ipaddr.sin_addr));
572 /*
573 * locate outgoing interface; if we're the destination,
574 * use the incoming interface (should be same).
575 */
576 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
577 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
578 type = ICMP_UNREACH;
579 code = ICMP_UNREACH_HOST;
580 goto bad;
581 }
582 bcopy((caddr_t)&(IA_SIN(ia)->sin_addr),
583 (caddr_t)(cp + off), sizeof(struct in_addr));
584 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
585 break;
586
587 case IPOPT_TS:
588 code = cp - (u_char *)ip;
589 ipt = (struct ip_timestamp *)cp;
590 if (ipt->ipt_len < 5)
591 goto bad;
592 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
593 if (++ipt->ipt_oflw == 0)
594 goto bad;
595 break;
596 }
597 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
598 switch (ipt->ipt_flg) {
599
600 case IPOPT_TS_TSONLY:
601 break;
602
603 case IPOPT_TS_TSANDADDR:
604 if (ipt->ipt_ptr + sizeof(n_time) +
605 sizeof(struct in_addr) > ipt->ipt_len)
606 goto bad;
607 ipaddr.sin_addr = dst;
608 ia = (INA)ifaof_ i f p foraddr((SA)&ipaddr,
609 m->m_pkthdr.rcvif);
610 if (ia == 0)
611 continue;
612 bcopy((caddr_t)&IA_SIN(ia)->sin_addr,
613 (caddr_t)sin, sizeof(struct in_addr));
614 ipt->ipt_ptr += sizeof(struct in_addr);
615 break;
616
617 case IPOPT_TS_PRESPEC:
618 if (ipt->ipt_ptr + sizeof(n_time) +
619 sizeof(struct in_addr) > ipt->ipt_len)
620 goto bad;
621 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
622 sizeof(struct in_addr));
623 if (ifa_ifwithaddr((SA)&ipaddr) == 0)
624 continue;
625 ipt->ipt_ptr += sizeof(struct in_addr);
626 break;
627
628 default:
629 goto bad;
630 }
631 ntime = iptime();
632 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
633 sizeof(n_time));
634 ipt->ipt_ptr += sizeof(n_time);
635 }
636 }
637 if (forward) {
638 ip_forward(m, 1);
639 return (1);
640 }
641 return (0);
642 bad:
643 icmp_send_error(m, type, code, 0, 0);
644
645 return (1);
646 }
647
648 #endif /* notdef */
649
650 /*
651 * Strip out IP options, at higher
652 * level protocol in the kernel.
653 * Second argument is buffer to which options
654 * will be moved, and return value is their length.
655 * (XXX) should be deleted; last arg currently ignored.
656 */
657 void
658 ip_stripoptions(register struct mbuf *m, struct mbuf *mopt)
659 {
660 register int i;
661 struct ip *ip = mtod(m, struct ip *);
662 register caddr_t opts;
663 int olen;
664
665 olen = (ip->ip_hl<<2) - sizeof (struct ip);
666 opts = (caddr_t)(ip + 1);
667 i = m->m_len - (sizeof (struct ip) + olen);
668 memcpy(opts, opts + olen, (unsigned)i);
669 m->m_len -= olen;
670
671 ip->ip_hl = sizeof(struct ip) >> 2;
672 }