Commit | Line | Data |
---|---|---|
15637ed4 RG |
1 | /* |
2 | * Copyright (c) 1982, 1986, 1988 Regents of the University of California. | |
3 | * 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. 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. | |
20 | * | |
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. | |
32 | * | |
38e82238 | 33 | * from: @(#)ip_input.c 7.19 (Berkeley) 5/25/91 |
d2462e50 | 34 | * $Id: ip_input.c,v 1.3 1993/11/07 17:47:58 wollman Exp $ |
15637ed4 RG |
35 | */ |
36 | ||
37 | #include "param.h" | |
38 | #include "systm.h" | |
39 | #include "malloc.h" | |
40 | #include "mbuf.h" | |
41 | #include "domain.h" | |
42 | #include "protosw.h" | |
43 | #include "socket.h" | |
44 | #include "errno.h" | |
45 | #include "time.h" | |
46 | #include "kernel.h" | |
47 | ||
48 | #include "../net/if.h" | |
49 | #include "../net/route.h" | |
50 | ||
51 | #include "in.h" | |
52 | #include "in_systm.h" | |
53 | #include "ip.h" | |
54 | #include "in_pcb.h" | |
55 | #include "in_var.h" | |
56 | #include "ip_var.h" | |
57 | #include "ip_icmp.h" | |
58 | ||
59 | #ifndef IPFORWARDING | |
60 | #ifdef GATEWAY | |
61 | #define IPFORWARDING 1 /* forward IP packets not for us */ | |
d2462e50 | 62 | #else /* not GATEWAY */ |
15637ed4 | 63 | #define IPFORWARDING 0 /* don't forward IP packets not for us */ |
d2462e50 GW |
64 | #endif /* not GATEWAY */ |
65 | #endif /* not IPFORWARDING */ | |
66 | ||
67 | /* | |
68 | * NB: RFC 1122, ``Requirements for Internet Hosts: Communication Layers'', | |
69 | * absolutely forbids hosts (which are not acting as gateways) from sending | |
70 | * ICMP redirects. | |
71 | */ | |
15637ed4 | 72 | #ifndef IPSENDREDIRECTS |
d2462e50 | 73 | #ifdef GATEWAY |
15637ed4 | 74 | #define IPSENDREDIRECTS 1 |
d2462e50 GW |
75 | #else /* not GATEWAY */ |
76 | #define IPSENDREDIRECTS 0 | |
77 | #endif /* not GATEWAY */ | |
78 | #endif /* not IPSENDREDIRECTS */ | |
79 | ||
15637ed4 RG |
80 | int ipforwarding = IPFORWARDING; |
81 | int ipsendredirects = IPSENDREDIRECTS; | |
82 | #ifdef DIAGNOSTIC | |
83 | int ipprintfs = 0; | |
84 | #endif | |
85 | ||
86 | extern struct domain inetdomain; | |
87 | extern struct protosw inetsw[]; | |
88 | u_char ip_protox[IPPROTO_MAX]; | |
89 | int ipqmaxlen = IFQ_MAXLEN; | |
90 | struct in_ifaddr *in_ifaddr; /* first inet address */ | |
8ace4366 GW |
91 | struct ipstat ipstat; |
92 | struct ipq ipq; | |
93 | u_short ip_id; | |
15637ed4 RG |
94 | |
95 | /* | |
96 | * We need to save the IP options in case a protocol wants to respond | |
97 | * to an incoming packet over the same route if the packet got here | |
98 | * using IP source routing. This allows connection establishment and | |
99 | * maintenance when the remote end is on a network that is not known | |
100 | * to us. | |
101 | */ | |
102 | int ip_nhops = 0; | |
103 | static struct ip_srcrt { | |
104 | struct in_addr dst; /* final destination */ | |
105 | char nop; /* one NOP to align */ | |
106 | char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ | |
107 | struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; | |
108 | } ip_srcrt; | |
109 | ||
110 | #ifdef GATEWAY | |
111 | extern int if_index; | |
112 | u_long *ip_ifmatrix; | |
113 | #endif | |
114 | ||
115 | /* | |
116 | * IP initialization: fill in IP protocol switch table. | |
117 | * All protocols not implemented in kernel go to raw IP protocol handler. | |
118 | */ | |
119 | ip_init() | |
120 | { | |
121 | register struct protosw *pr; | |
122 | register int i; | |
123 | ||
124 | pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); | |
125 | if (pr == 0) | |
126 | panic("ip_init"); | |
127 | for (i = 0; i < IPPROTO_MAX; i++) | |
128 | ip_protox[i] = pr - inetsw; | |
129 | for (pr = inetdomain.dom_protosw; | |
130 | pr < inetdomain.dom_protoswNPROTOSW; pr++) | |
131 | if (pr->pr_domain->dom_family == PF_INET && | |
132 | pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) | |
133 | ip_protox[pr->pr_protocol] = pr - inetsw; | |
134 | ipq.next = ipq.prev = &ipq; | |
135 | ip_id = time.tv_sec & 0xffff; | |
136 | ipintrq.ifq_maxlen = ipqmaxlen; | |
137 | #ifdef GATEWAY | |
138 | i = (if_index + 1) * (if_index + 1) * sizeof (u_long); | |
139 | if ((ip_ifmatrix = (u_long *) malloc(i, M_RTABLE, M_WAITOK)) == 0) | |
140 | panic("no memory for ip_ifmatrix"); | |
141 | #endif | |
142 | } | |
143 | ||
144 | struct ip *ip_reass(); | |
145 | struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; | |
146 | struct route ipforward_rt; | |
147 | ||
148 | /* | |
149 | * Ip input routine. Checksum and byte swap header. If fragmented | |
150 | * try to reassemble. Process options. Pass to next level. | |
151 | */ | |
152 | ipintr() | |
153 | { | |
154 | register struct ip *ip; | |
155 | register struct mbuf *m; | |
156 | register struct ipq *fp; | |
157 | register struct in_ifaddr *ia; | |
158 | int hlen, s; | |
159 | ||
160 | next: | |
161 | /* | |
162 | * Get next datagram off input queue and get IP header | |
163 | * in first mbuf. | |
164 | */ | |
165 | s = splimp(); | |
166 | IF_DEQUEUE(&ipintrq, m); | |
167 | splx(s); | |
168 | if (m == 0) | |
169 | return; | |
170 | #ifdef DIAGNOSTIC | |
171 | if ((m->m_flags & M_PKTHDR) == 0) | |
172 | panic("ipintr no HDR"); | |
173 | #endif | |
174 | /* | |
175 | * If no IP addresses have been set yet but the interfaces | |
176 | * are receiving, can't do anything with incoming packets yet. | |
177 | */ | |
178 | if (in_ifaddr == NULL) | |
179 | goto bad; | |
180 | ipstat.ips_total++; | |
181 | if (m->m_len < sizeof (struct ip) && | |
182 | (m = m_pullup(m, sizeof (struct ip))) == 0) { | |
183 | ipstat.ips_toosmall++; | |
184 | goto next; | |
185 | } | |
186 | ip = mtod(m, struct ip *); | |
187 | hlen = ip->ip_hl << 2; | |
188 | if (hlen < sizeof(struct ip)) { /* minimum header length */ | |
189 | ipstat.ips_badhlen++; | |
190 | goto bad; | |
191 | } | |
192 | if (hlen > m->m_len) { | |
193 | if ((m = m_pullup(m, hlen)) == 0) { | |
194 | ipstat.ips_badhlen++; | |
195 | goto next; | |
196 | } | |
197 | ip = mtod(m, struct ip *); | |
198 | } | |
199 | if (ip->ip_sum = in_cksum(m, hlen)) { | |
200 | ipstat.ips_badsum++; | |
201 | goto bad; | |
202 | } | |
203 | ||
204 | /* | |
205 | * Convert fields to host representation. | |
206 | */ | |
207 | NTOHS(ip->ip_len); | |
208 | if (ip->ip_len < hlen) { | |
209 | ipstat.ips_badlen++; | |
210 | goto bad; | |
211 | } | |
212 | NTOHS(ip->ip_id); | |
213 | NTOHS(ip->ip_off); | |
214 | ||
215 | /* | |
216 | * Check that the amount of data in the buffers | |
217 | * is as at least much as the IP header would have us expect. | |
218 | * Trim mbufs if longer than we expect. | |
219 | * Drop packet if shorter than we expect. | |
220 | */ | |
221 | if (m->m_pkthdr.len < ip->ip_len) { | |
222 | ipstat.ips_tooshort++; | |
223 | goto bad; | |
224 | } | |
225 | if (m->m_pkthdr.len > ip->ip_len) { | |
226 | if (m->m_len == m->m_pkthdr.len) { | |
227 | m->m_len = ip->ip_len; | |
228 | m->m_pkthdr.len = ip->ip_len; | |
229 | } else | |
230 | m_adj(m, ip->ip_len - m->m_pkthdr.len); | |
231 | } | |
232 | ||
233 | /* | |
234 | * Process options and, if not destined for us, | |
235 | * ship it on. ip_dooptions returns 1 when an | |
236 | * error was detected (causing an icmp message | |
237 | * to be sent and the original packet to be freed). | |
238 | */ | |
239 | ip_nhops = 0; /* for source routed packets */ | |
240 | if (hlen > sizeof (struct ip) && ip_dooptions(m)) | |
241 | goto next; | |
242 | ||
243 | /* | |
244 | * Check our list of addresses, to see if the packet is for us. | |
245 | */ | |
246 | for (ia = in_ifaddr; ia; ia = ia->ia_next) { | |
247 | #define satosin(sa) ((struct sockaddr_in *)(sa)) | |
248 | ||
249 | if (IA_SIN(ia)->sin_addr.s_addr == ip->ip_dst.s_addr) | |
250 | goto ours; | |
251 | if ( | |
252 | #ifdef DIRECTED_BROADCAST | |
253 | ia->ia_ifp == m->m_pkthdr.rcvif && | |
254 | #endif | |
255 | (ia->ia_ifp->if_flags & IFF_BROADCAST)) { | |
256 | u_long t; | |
257 | ||
258 | if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr == | |
259 | ip->ip_dst.s_addr) | |
260 | goto ours; | |
261 | if (ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr) | |
262 | goto ours; | |
263 | /* | |
264 | * Look for all-0's host part (old broadcast addr), | |
265 | * either for subnet or net. | |
266 | */ | |
267 | t = ntohl(ip->ip_dst.s_addr); | |
268 | if (t == ia->ia_subnet) | |
269 | goto ours; | |
270 | if (t == ia->ia_net) | |
271 | goto ours; | |
272 | } | |
273 | } | |
274 | if (ip->ip_dst.s_addr == (u_long)INADDR_BROADCAST) | |
275 | goto ours; | |
276 | if (ip->ip_dst.s_addr == INADDR_ANY) | |
277 | goto ours; | |
278 | ||
279 | /* | |
280 | * Not for us; forward if possible and desirable. | |
281 | */ | |
282 | if (ipforwarding == 0) { | |
283 | ipstat.ips_cantforward++; | |
284 | m_freem(m); | |
285 | } else | |
286 | ip_forward(m, 0); | |
287 | goto next; | |
288 | ||
289 | ours: | |
290 | /* | |
291 | * If offset or IP_MF are set, must reassemble. | |
292 | * Otherwise, nothing need be done. | |
293 | * (We could look in the reassembly queue to see | |
294 | * if the packet was previously fragmented, | |
295 | * but it's not worth the time; just let them time out.) | |
296 | */ | |
297 | if (ip->ip_off &~ IP_DF) { | |
298 | if (m->m_flags & M_EXT) { /* XXX */ | |
299 | if ((m = m_pullup(m, sizeof (struct ip))) == 0) { | |
300 | ipstat.ips_toosmall++; | |
301 | goto next; | |
302 | } | |
303 | ip = mtod(m, struct ip *); | |
304 | } | |
305 | /* | |
306 | * Look for queue of fragments | |
307 | * of this datagram. | |
308 | */ | |
309 | for (fp = ipq.next; fp != &ipq; fp = fp->next) | |
310 | if (ip->ip_id == fp->ipq_id && | |
311 | ip->ip_src.s_addr == fp->ipq_src.s_addr && | |
312 | ip->ip_dst.s_addr == fp->ipq_dst.s_addr && | |
313 | ip->ip_p == fp->ipq_p) | |
314 | goto found; | |
315 | fp = 0; | |
316 | found: | |
317 | ||
318 | /* | |
319 | * Adjust ip_len to not reflect header, | |
320 | * set ip_mff if more fragments are expected, | |
321 | * convert offset of this to bytes. | |
322 | */ | |
323 | ip->ip_len -= hlen; | |
324 | ((struct ipasfrag *)ip)->ipf_mff = 0; | |
325 | if (ip->ip_off & IP_MF) | |
326 | ((struct ipasfrag *)ip)->ipf_mff = 1; | |
327 | ip->ip_off <<= 3; | |
328 | ||
329 | /* | |
330 | * If datagram marked as having more fragments | |
331 | * or if this is not the first fragment, | |
332 | * attempt reassembly; if it succeeds, proceed. | |
333 | */ | |
334 | if (((struct ipasfrag *)ip)->ipf_mff || ip->ip_off) { | |
335 | ipstat.ips_fragments++; | |
336 | ip = ip_reass((struct ipasfrag *)ip, fp); | |
337 | if (ip == 0) | |
338 | goto next; | |
339 | else | |
340 | ipstat.ips_reassembled++; | |
341 | m = dtom(ip); | |
342 | } else | |
343 | if (fp) | |
344 | ip_freef(fp); | |
345 | } else | |
346 | ip->ip_len -= hlen; | |
347 | ||
348 | /* | |
349 | * Switch out to protocol's input routine. | |
350 | */ | |
351 | ipstat.ips_delivered++; | |
352 | (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen); | |
353 | goto next; | |
354 | bad: | |
355 | m_freem(m); | |
356 | goto next; | |
357 | } | |
358 | ||
359 | /* | |
360 | * Take incoming datagram fragment and try to | |
361 | * reassemble it into whole datagram. If a chain for | |
362 | * reassembly of this datagram already exists, then it | |
363 | * is given as fp; otherwise have to make a chain. | |
364 | */ | |
365 | struct ip * | |
366 | ip_reass(ip, fp) | |
367 | register struct ipasfrag *ip; | |
368 | register struct ipq *fp; | |
369 | { | |
370 | register struct mbuf *m = dtom(ip); | |
371 | register struct ipasfrag *q; | |
372 | struct mbuf *t; | |
373 | int hlen = ip->ip_hl << 2; | |
374 | int i, next; | |
375 | ||
376 | /* | |
377 | * Presence of header sizes in mbufs | |
378 | * would confuse code below. | |
379 | */ | |
380 | m->m_data += hlen; | |
381 | m->m_len -= hlen; | |
382 | ||
383 | /* | |
384 | * If first fragment to arrive, create a reassembly queue. | |
385 | */ | |
386 | if (fp == 0) { | |
387 | if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL) | |
388 | goto dropfrag; | |
389 | fp = mtod(t, struct ipq *); | |
390 | insque(fp, &ipq); | |
391 | fp->ipq_ttl = IPFRAGTTL; | |
392 | fp->ipq_p = ip->ip_p; | |
393 | fp->ipq_id = ip->ip_id; | |
394 | fp->ipq_next = fp->ipq_prev = (struct ipasfrag *)fp; | |
395 | fp->ipq_src = ((struct ip *)ip)->ip_src; | |
396 | fp->ipq_dst = ((struct ip *)ip)->ip_dst; | |
397 | q = (struct ipasfrag *)fp; | |
398 | goto insert; | |
399 | } | |
400 | ||
401 | /* | |
402 | * Find a segment which begins after this one does. | |
403 | */ | |
404 | for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) | |
405 | if (q->ip_off > ip->ip_off) | |
406 | break; | |
407 | ||
408 | /* | |
409 | * If there is a preceding segment, it may provide some of | |
410 | * our data already. If so, drop the data from the incoming | |
411 | * segment. If it provides all of our data, drop us. | |
412 | */ | |
413 | if (q->ipf_prev != (struct ipasfrag *)fp) { | |
414 | i = q->ipf_prev->ip_off + q->ipf_prev->ip_len - ip->ip_off; | |
415 | if (i > 0) { | |
416 | if (i >= ip->ip_len) | |
417 | goto dropfrag; | |
418 | m_adj(dtom(ip), i); | |
419 | ip->ip_off += i; | |
420 | ip->ip_len -= i; | |
421 | } | |
422 | } | |
423 | ||
424 | /* | |
425 | * While we overlap succeeding segments trim them or, | |
426 | * if they are completely covered, dequeue them. | |
427 | */ | |
428 | while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) { | |
429 | i = (ip->ip_off + ip->ip_len) - q->ip_off; | |
430 | if (i < q->ip_len) { | |
431 | q->ip_len -= i; | |
432 | q->ip_off += i; | |
433 | m_adj(dtom(q), i); | |
434 | break; | |
435 | } | |
436 | q = q->ipf_next; | |
437 | m_freem(dtom(q->ipf_prev)); | |
438 | ip_deq(q->ipf_prev); | |
439 | } | |
440 | ||
441 | insert: | |
442 | /* | |
443 | * Stick new segment in its place; | |
444 | * check for complete reassembly. | |
445 | */ | |
446 | ip_enq(ip, q->ipf_prev); | |
447 | next = 0; | |
448 | for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) { | |
449 | if (q->ip_off != next) | |
450 | return (0); | |
451 | next += q->ip_len; | |
452 | } | |
453 | if (q->ipf_prev->ipf_mff) | |
454 | return (0); | |
455 | ||
456 | /* | |
457 | * Reassembly is complete; concatenate fragments. | |
458 | */ | |
459 | q = fp->ipq_next; | |
460 | m = dtom(q); | |
461 | t = m->m_next; | |
462 | m->m_next = 0; | |
463 | m_cat(m, t); | |
464 | q = q->ipf_next; | |
465 | while (q != (struct ipasfrag *)fp) { | |
466 | t = dtom(q); | |
467 | q = q->ipf_next; | |
468 | m_cat(m, t); | |
469 | } | |
470 | ||
471 | /* | |
472 | * Create header for new ip packet by | |
473 | * modifying header of first packet; | |
474 | * dequeue and discard fragment reassembly header. | |
475 | * Make header visible. | |
476 | */ | |
477 | ip = fp->ipq_next; | |
478 | ip->ip_len = next; | |
479 | ((struct ip *)ip)->ip_src = fp->ipq_src; | |
480 | ((struct ip *)ip)->ip_dst = fp->ipq_dst; | |
481 | remque(fp); | |
482 | (void) m_free(dtom(fp)); | |
483 | m = dtom(ip); | |
484 | m->m_len += (ip->ip_hl << 2); | |
485 | m->m_data -= (ip->ip_hl << 2); | |
486 | /* some debugging cruft by sklower, below, will go away soon */ | |
487 | if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ | |
488 | register int plen = 0; | |
489 | for (t = m; m; m = m->m_next) | |
490 | plen += m->m_len; | |
491 | t->m_pkthdr.len = plen; | |
492 | } | |
493 | return ((struct ip *)ip); | |
494 | ||
495 | dropfrag: | |
496 | ipstat.ips_fragdropped++; | |
497 | m_freem(m); | |
498 | return (0); | |
499 | } | |
500 | ||
501 | /* | |
502 | * Free a fragment reassembly header and all | |
503 | * associated datagrams. | |
504 | */ | |
505 | ip_freef(fp) | |
506 | struct ipq *fp; | |
507 | { | |
508 | register struct ipasfrag *q, *p; | |
509 | ||
510 | for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) { | |
511 | p = q->ipf_next; | |
512 | ip_deq(q); | |
513 | m_freem(dtom(q)); | |
514 | } | |
515 | remque(fp); | |
516 | (void) m_free(dtom(fp)); | |
517 | } | |
518 | ||
519 | /* | |
520 | * Put an ip fragment on a reassembly chain. | |
521 | * Like insque, but pointers in middle of structure. | |
522 | */ | |
523 | ip_enq(p, prev) | |
524 | register struct ipasfrag *p, *prev; | |
525 | { | |
526 | ||
527 | p->ipf_prev = prev; | |
528 | p->ipf_next = prev->ipf_next; | |
529 | prev->ipf_next->ipf_prev = p; | |
530 | prev->ipf_next = p; | |
531 | } | |
532 | ||
533 | /* | |
534 | * To ip_enq as remque is to insque. | |
535 | */ | |
536 | ip_deq(p) | |
537 | register struct ipasfrag *p; | |
538 | { | |
539 | ||
540 | p->ipf_prev->ipf_next = p->ipf_next; | |
541 | p->ipf_next->ipf_prev = p->ipf_prev; | |
542 | } | |
543 | ||
544 | /* | |
545 | * IP timer processing; | |
546 | * if a timer expires on a reassembly | |
547 | * queue, discard it. | |
548 | */ | |
549 | ip_slowtimo() | |
550 | { | |
551 | register struct ipq *fp; | |
552 | int s = splnet(); | |
553 | ||
554 | fp = ipq.next; | |
555 | if (fp == 0) { | |
556 | splx(s); | |
557 | return; | |
558 | } | |
559 | while (fp != &ipq) { | |
560 | --fp->ipq_ttl; | |
561 | fp = fp->next; | |
562 | if (fp->prev->ipq_ttl == 0) { | |
563 | ipstat.ips_fragtimeout++; | |
564 | ip_freef(fp->prev); | |
565 | } | |
566 | } | |
567 | splx(s); | |
568 | } | |
569 | ||
570 | /* | |
571 | * Drain off all datagram fragments. | |
572 | */ | |
573 | ip_drain() | |
574 | { | |
575 | ||
576 | while (ipq.next != &ipq) { | |
577 | ipstat.ips_fragdropped++; | |
578 | ip_freef(ipq.next); | |
579 | } | |
580 | } | |
581 | ||
582 | extern struct in_ifaddr *ifptoia(); | |
583 | struct in_ifaddr *ip_rtaddr(); | |
584 | ||
585 | /* | |
586 | * Do option processing on a datagram, | |
587 | * possibly discarding it if bad options are encountered, | |
588 | * or forwarding it if source-routed. | |
589 | * Returns 1 if packet has been forwarded/freed, | |
590 | * 0 if the packet should be processed further. | |
591 | */ | |
592 | ip_dooptions(m) | |
593 | struct mbuf *m; | |
594 | { | |
595 | register struct ip *ip = mtod(m, struct ip *); | |
596 | register u_char *cp; | |
597 | register struct ip_timestamp *ipt; | |
598 | register struct in_ifaddr *ia; | |
599 | int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; | |
600 | struct in_addr *sin; | |
601 | n_time ntime; | |
602 | ||
603 | cp = (u_char *)(ip + 1); | |
604 | cnt = (ip->ip_hl << 2) - sizeof (struct ip); | |
605 | for (; cnt > 0; cnt -= optlen, cp += optlen) { | |
606 | opt = cp[IPOPT_OPTVAL]; | |
607 | if (opt == IPOPT_EOL) | |
608 | break; | |
609 | if (opt == IPOPT_NOP) | |
610 | optlen = 1; | |
611 | else { | |
612 | optlen = cp[IPOPT_OLEN]; | |
613 | if (optlen <= 0 || optlen > cnt) { | |
614 | code = &cp[IPOPT_OLEN] - (u_char *)ip; | |
615 | goto bad; | |
616 | } | |
617 | } | |
618 | switch (opt) { | |
619 | ||
620 | default: | |
621 | break; | |
622 | ||
623 | /* | |
624 | * Source routing with record. | |
625 | * Find interface with current destination address. | |
626 | * If none on this machine then drop if strictly routed, | |
627 | * or do nothing if loosely routed. | |
628 | * Record interface address and bring up next address | |
629 | * component. If strictly routed make sure next | |
630 | * address is on directly accessible net. | |
631 | */ | |
632 | case IPOPT_LSRR: | |
633 | case IPOPT_SSRR: | |
634 | if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { | |
635 | code = &cp[IPOPT_OFFSET] - (u_char *)ip; | |
636 | goto bad; | |
637 | } | |
638 | ipaddr.sin_addr = ip->ip_dst; | |
639 | ia = (struct in_ifaddr *) | |
640 | ifa_ifwithaddr((struct sockaddr *)&ipaddr); | |
641 | if (ia == 0) { | |
642 | if (opt == IPOPT_SSRR) { | |
643 | type = ICMP_UNREACH; | |
644 | code = ICMP_UNREACH_SRCFAIL; | |
645 | goto bad; | |
646 | } | |
647 | /* | |
648 | * Loose routing, and not at next destination | |
649 | * yet; nothing to do except forward. | |
650 | */ | |
651 | break; | |
652 | } | |
653 | off--; /* 0 origin */ | |
654 | if (off > optlen - sizeof(struct in_addr)) { | |
655 | /* | |
656 | * End of source route. Should be for us. | |
657 | */ | |
658 | save_rte(cp, ip->ip_src); | |
659 | break; | |
660 | } | |
661 | /* | |
662 | * locate outgoing interface | |
663 | */ | |
664 | bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, | |
665 | sizeof(ipaddr.sin_addr)); | |
666 | if (opt == IPOPT_SSRR) { | |
667 | #define INA struct in_ifaddr * | |
668 | #define SA struct sockaddr * | |
669 | if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0) | |
670 | ia = in_iaonnetof(in_netof(ipaddr.sin_addr)); | |
671 | } else | |
672 | ia = ip_rtaddr(ipaddr.sin_addr); | |
673 | if (ia == 0) { | |
674 | type = ICMP_UNREACH; | |
675 | code = ICMP_UNREACH_SRCFAIL; | |
676 | goto bad; | |
677 | } | |
678 | ip->ip_dst = ipaddr.sin_addr; | |
679 | bcopy((caddr_t)&(IA_SIN(ia)->sin_addr), | |
680 | (caddr_t)(cp + off), sizeof(struct in_addr)); | |
681 | cp[IPOPT_OFFSET] += sizeof(struct in_addr); | |
682 | forward = 1; | |
683 | break; | |
684 | ||
685 | case IPOPT_RR: | |
686 | if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { | |
687 | code = &cp[IPOPT_OFFSET] - (u_char *)ip; | |
688 | goto bad; | |
689 | } | |
690 | /* | |
691 | * If no space remains, ignore. | |
692 | */ | |
693 | off--; /* 0 origin */ | |
694 | if (off > optlen - sizeof(struct in_addr)) | |
695 | break; | |
696 | bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, | |
697 | sizeof(ipaddr.sin_addr)); | |
698 | /* | |
699 | * locate outgoing interface; if we're the destination, | |
700 | * use the incoming interface (should be same). | |
701 | */ | |
702 | if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && | |
703 | (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { | |
704 | type = ICMP_UNREACH; | |
705 | code = ICMP_UNREACH_HOST; | |
706 | goto bad; | |
707 | } | |
708 | bcopy((caddr_t)&(IA_SIN(ia)->sin_addr), | |
709 | (caddr_t)(cp + off), sizeof(struct in_addr)); | |
710 | cp[IPOPT_OFFSET] += sizeof(struct in_addr); | |
711 | break; | |
712 | ||
713 | case IPOPT_TS: | |
714 | code = cp - (u_char *)ip; | |
715 | ipt = (struct ip_timestamp *)cp; | |
716 | if (ipt->ipt_len < 5) | |
717 | goto bad; | |
718 | if (ipt->ipt_ptr > ipt->ipt_len - sizeof (long)) { | |
719 | if (++ipt->ipt_oflw == 0) | |
720 | goto bad; | |
721 | break; | |
722 | } | |
723 | sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); | |
724 | switch (ipt->ipt_flg) { | |
725 | ||
726 | case IPOPT_TS_TSONLY: | |
727 | break; | |
728 | ||
729 | case IPOPT_TS_TSANDADDR: | |
730 | if (ipt->ipt_ptr + sizeof(n_time) + | |
731 | sizeof(struct in_addr) > ipt->ipt_len) | |
732 | goto bad; | |
733 | ia = ifptoia(m->m_pkthdr.rcvif); | |
734 | bcopy((caddr_t)&IA_SIN(ia)->sin_addr, | |
735 | (caddr_t)sin, sizeof(struct in_addr)); | |
736 | ipt->ipt_ptr += sizeof(struct in_addr); | |
737 | break; | |
738 | ||
739 | case IPOPT_TS_PRESPEC: | |
740 | if (ipt->ipt_ptr + sizeof(n_time) + | |
741 | sizeof(struct in_addr) > ipt->ipt_len) | |
742 | goto bad; | |
743 | bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, | |
744 | sizeof(struct in_addr)); | |
745 | if (ifa_ifwithaddr((SA)&ipaddr) == 0) | |
746 | continue; | |
747 | ipt->ipt_ptr += sizeof(struct in_addr); | |
748 | break; | |
749 | ||
750 | default: | |
751 | goto bad; | |
752 | } | |
753 | ntime = iptime(); | |
754 | bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, | |
755 | sizeof(n_time)); | |
756 | ipt->ipt_ptr += sizeof(n_time); | |
757 | } | |
758 | } | |
759 | if (forward) { | |
760 | ip_forward(m, 1); | |
761 | return (1); | |
762 | } else | |
763 | return (0); | |
764 | bad: | |
765 | icmp_error(m, type, code); | |
766 | return (1); | |
767 | } | |
768 | ||
769 | /* | |
770 | * Given address of next destination (final or next hop), | |
771 | * return internet address info of interface to be used to get there. | |
772 | */ | |
773 | struct in_ifaddr * | |
774 | ip_rtaddr(dst) | |
775 | struct in_addr dst; | |
776 | { | |
777 | register struct sockaddr_in *sin; | |
778 | ||
779 | sin = (struct sockaddr_in *) &ipforward_rt.ro_dst; | |
780 | ||
781 | if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) { | |
782 | if (ipforward_rt.ro_rt) { | |
783 | RTFREE(ipforward_rt.ro_rt); | |
784 | ipforward_rt.ro_rt = 0; | |
785 | } | |
786 | sin->sin_family = AF_INET; | |
787 | sin->sin_len = sizeof(*sin); | |
788 | sin->sin_addr = dst; | |
789 | ||
790 | rtalloc(&ipforward_rt); | |
791 | } | |
792 | if (ipforward_rt.ro_rt == 0) | |
793 | return ((struct in_ifaddr *)0); | |
794 | return ((struct in_ifaddr *) ipforward_rt.ro_rt->rt_ifa); | |
795 | } | |
796 | ||
797 | /* | |
798 | * Save incoming source route for use in replies, | |
799 | * to be picked up later by ip_srcroute if the receiver is interested. | |
800 | */ | |
801 | save_rte(option, dst) | |
802 | u_char *option; | |
803 | struct in_addr dst; | |
804 | { | |
805 | unsigned olen; | |
806 | ||
807 | olen = option[IPOPT_OLEN]; | |
808 | #ifdef DIAGNOSTIC | |
809 | if (ipprintfs) | |
810 | printf("save_rte: olen %d\n", olen); | |
811 | #endif | |
812 | if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) | |
813 | return; | |
814 | bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); | |
815 | ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); | |
816 | ip_srcrt.dst = dst; | |
817 | } | |
818 | ||
819 | /* | |
820 | * Retrieve incoming source route for use in replies, | |
821 | * in the same form used by setsockopt. | |
822 | * The first hop is placed before the options, will be removed later. | |
823 | */ | |
824 | struct mbuf * | |
825 | ip_srcroute() | |
826 | { | |
827 | register struct in_addr *p, *q; | |
828 | register struct mbuf *m; | |
829 | ||
830 | if (ip_nhops == 0) | |
831 | return ((struct mbuf *)0); | |
832 | m = m_get(M_DONTWAIT, MT_SOOPTS); | |
833 | if (m == 0) | |
834 | return ((struct mbuf *)0); | |
835 | ||
836 | #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) | |
837 | ||
838 | /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ | |
839 | m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + | |
840 | OPTSIZ; | |
841 | #ifdef DIAGNOSTIC | |
842 | if (ipprintfs) | |
843 | printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); | |
844 | #endif | |
845 | ||
846 | /* | |
847 | * First save first hop for return route | |
848 | */ | |
849 | p = &ip_srcrt.route[ip_nhops - 1]; | |
850 | *(mtod(m, struct in_addr *)) = *p--; | |
851 | #ifdef DIAGNOSTIC | |
852 | if (ipprintfs) | |
853 | printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr)); | |
854 | #endif | |
855 | ||
856 | /* | |
857 | * Copy option fields and padding (nop) to mbuf. | |
858 | */ | |
859 | ip_srcrt.nop = IPOPT_NOP; | |
860 | ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; | |
861 | bcopy((caddr_t)&ip_srcrt.nop, | |
862 | mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); | |
863 | q = (struct in_addr *)(mtod(m, caddr_t) + | |
864 | sizeof(struct in_addr) + OPTSIZ); | |
865 | #undef OPTSIZ | |
866 | /* | |
867 | * Record return path as an IP source route, | |
868 | * reversing the path (pointers are now aligned). | |
869 | */ | |
870 | while (p >= ip_srcrt.route) { | |
871 | #ifdef DIAGNOSTIC | |
872 | if (ipprintfs) | |
873 | printf(" %lx", ntohl(q->s_addr)); | |
874 | #endif | |
875 | *q++ = *p--; | |
876 | } | |
877 | /* | |
878 | * Last hop goes to final destination. | |
879 | */ | |
880 | *q = ip_srcrt.dst; | |
881 | #ifdef DIAGNOSTIC | |
882 | if (ipprintfs) | |
883 | printf(" %lx\n", ntohl(q->s_addr)); | |
884 | #endif | |
885 | return (m); | |
886 | } | |
887 | ||
888 | /* | |
889 | * Strip out IP options, at higher | |
890 | * level protocol in the kernel. | |
891 | * Second argument is buffer to which options | |
892 | * will be moved, and return value is their length. | |
893 | * XXX should be deleted; last arg currently ignored. | |
894 | */ | |
895 | ip_stripoptions(m, mopt) | |
896 | register struct mbuf *m; | |
897 | struct mbuf *mopt; | |
898 | { | |
899 | register int i; | |
900 | struct ip *ip = mtod(m, struct ip *); | |
901 | register caddr_t opts; | |
902 | int olen; | |
903 | ||
904 | olen = (ip->ip_hl<<2) - sizeof (struct ip); | |
905 | opts = (caddr_t)(ip + 1); | |
906 | i = m->m_len - (sizeof (struct ip) + olen); | |
907 | bcopy(opts + olen, opts, (unsigned)i); | |
908 | m->m_len -= olen; | |
909 | if (m->m_flags & M_PKTHDR) | |
910 | m->m_pkthdr.len -= olen; | |
911 | ip->ip_hl = sizeof(struct ip) >> 2; | |
912 | } | |
913 | ||
914 | u_char inetctlerrmap[PRC_NCMDS] = { | |
915 | 0, 0, 0, 0, | |
916 | 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, | |
917 | EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, | |
918 | EMSGSIZE, EHOSTUNREACH, 0, 0, | |
919 | 0, 0, 0, 0, | |
920 | ENOPROTOOPT | |
921 | }; | |
922 | ||
923 | /* | |
924 | * Forward a packet. If some error occurs return the sender | |
925 | * an icmp packet. Note we can't always generate a meaningful | |
926 | * icmp message because icmp doesn't have a large enough repertoire | |
927 | * of codes and types. | |
928 | * | |
929 | * If not forwarding, just drop the packet. This could be confusing | |
930 | * if ipforwarding was zero but some routing protocol was advancing | |
931 | * us as a gateway to somewhere. However, we must let the routing | |
932 | * protocol deal with that. | |
933 | * | |
934 | * The srcrt parameter indicates whether the packet is being forwarded | |
935 | * via a source route. | |
936 | */ | |
937 | ip_forward(m, srcrt) | |
938 | struct mbuf *m; | |
939 | int srcrt; | |
940 | { | |
941 | register struct ip *ip = mtod(m, struct ip *); | |
942 | register struct sockaddr_in *sin; | |
943 | register struct rtentry *rt; | |
944 | int error, type = 0, code; | |
945 | struct mbuf *mcopy; | |
946 | struct in_addr dest; | |
947 | ||
948 | dest.s_addr = 0; | |
949 | #ifdef DIAGNOSTIC | |
950 | if (ipprintfs) | |
951 | printf("forward: src %x dst %x ttl %x\n", ip->ip_src, | |
952 | ip->ip_dst, ip->ip_ttl); | |
953 | #endif | |
954 | if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) { | |
955 | ipstat.ips_cantforward++; | |
956 | m_freem(m); | |
957 | return; | |
958 | } | |
959 | HTONS(ip->ip_id); | |
960 | if (ip->ip_ttl <= IPTTLDEC) { | |
961 | icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest); | |
962 | return; | |
963 | } | |
964 | ip->ip_ttl -= IPTTLDEC; | |
965 | ||
966 | sin = (struct sockaddr_in *)&ipforward_rt.ro_dst; | |
967 | if ((rt = ipforward_rt.ro_rt) == 0 || | |
968 | ip->ip_dst.s_addr != sin->sin_addr.s_addr) { | |
969 | if (ipforward_rt.ro_rt) { | |
970 | RTFREE(ipforward_rt.ro_rt); | |
971 | ipforward_rt.ro_rt = 0; | |
972 | } | |
973 | sin->sin_family = AF_INET; | |
974 | sin->sin_len = sizeof(*sin); | |
975 | sin->sin_addr = ip->ip_dst; | |
976 | ||
977 | rtalloc(&ipforward_rt); | |
978 | if (ipforward_rt.ro_rt == 0) { | |
979 | icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest); | |
980 | return; | |
981 | } | |
982 | rt = ipforward_rt.ro_rt; | |
983 | } | |
984 | ||
985 | /* | |
986 | * Save at most 64 bytes of the packet in case | |
987 | * we need to generate an ICMP message to the src. | |
988 | */ | |
989 | mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64)); | |
990 | ||
991 | #ifdef GATEWAY | |
992 | ip_ifmatrix[rt->rt_ifp->if_index + | |
993 | if_index * m->m_pkthdr.rcvif->if_index]++; | |
994 | #endif | |
995 | /* | |
996 | * If forwarding packet using same interface that it came in on, | |
997 | * perhaps should send a redirect to sender to shortcut a hop. | |
998 | * Only send redirect if source is sending directly to us, | |
999 | * and if packet was not source routed (or has any options). | |
1000 | * Also, don't send redirect if forwarding using a default route | |
1001 | * or a route modified by a redirect. | |
1002 | */ | |
1003 | #define satosin(sa) ((struct sockaddr_in *)(sa)) | |
1004 | if (rt->rt_ifp == m->m_pkthdr.rcvif && | |
1005 | (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && | |
1006 | satosin(rt_key(rt))->sin_addr.s_addr != 0 && | |
1007 | ipsendredirects && !srcrt) { | |
1008 | struct in_ifaddr *ia; | |
1009 | u_long src = ntohl(ip->ip_src.s_addr); | |
1010 | u_long dst = ntohl(ip->ip_dst.s_addr); | |
1011 | ||
1012 | if ((ia = ifptoia(m->m_pkthdr.rcvif)) && | |
1013 | (src & ia->ia_subnetmask) == ia->ia_subnet) { | |
1014 | if (rt->rt_flags & RTF_GATEWAY) | |
1015 | dest = satosin(rt->rt_gateway)->sin_addr; | |
1016 | else | |
1017 | dest = ip->ip_dst; | |
1018 | /* | |
1019 | * If the destination is reached by a route to host, | |
1020 | * is on a subnet of a local net, or is directly | |
1021 | * on the attached net (!), use host redirect. | |
1022 | * (We may be the correct first hop for other subnets.) | |
1023 | */ | |
1024 | #define RTA(rt) ((struct in_ifaddr *)(rt->rt_ifa)) | |
1025 | type = ICMP_REDIRECT; | |
1026 | if ((rt->rt_flags & RTF_HOST) || | |
1027 | (rt->rt_flags & RTF_GATEWAY) == 0) | |
1028 | code = ICMP_REDIRECT_HOST; | |
1029 | else if (RTA(rt)->ia_subnetmask != RTA(rt)->ia_netmask && | |
1030 | (dst & RTA(rt)->ia_netmask) == RTA(rt)->ia_net) | |
1031 | code = ICMP_REDIRECT_HOST; | |
1032 | else | |
1033 | code = ICMP_REDIRECT_NET; | |
1034 | #ifdef DIAGNOSTIC | |
1035 | if (ipprintfs) | |
1036 | printf("redirect (%d) to %x\n", code, dest.s_addr); | |
1037 | #endif | |
1038 | } | |
1039 | } | |
1040 | ||
1041 | error = ip_output(m, (struct mbuf *)0, &ipforward_rt, IP_FORWARDING); | |
1042 | if (error) | |
1043 | ipstat.ips_cantforward++; | |
1044 | else { | |
1045 | ipstat.ips_forward++; | |
1046 | if (type) | |
1047 | ipstat.ips_redirectsent++; | |
1048 | else { | |
1049 | if (mcopy) | |
1050 | m_freem(mcopy); | |
1051 | return; | |
1052 | } | |
1053 | } | |
1054 | if (mcopy == NULL) | |
1055 | return; | |
1056 | switch (error) { | |
1057 | ||
1058 | case 0: /* forwarded, but need redirect */ | |
1059 | /* type, code set above */ | |
1060 | break; | |
1061 | ||
1062 | case ENETUNREACH: /* shouldn't happen, checked above */ | |
1063 | case EHOSTUNREACH: | |
1064 | case ENETDOWN: | |
1065 | case EHOSTDOWN: | |
1066 | default: | |
1067 | type = ICMP_UNREACH; | |
1068 | code = ICMP_UNREACH_HOST; | |
1069 | break; | |
1070 | ||
1071 | case EMSGSIZE: | |
1072 | type = ICMP_UNREACH; | |
1073 | code = ICMP_UNREACH_NEEDFRAG; | |
1074 | ipstat.ips_cantfrag++; | |
1075 | break; | |
1076 | ||
1077 | case ENOBUFS: | |
1078 | type = ICMP_SOURCEQUENCH; | |
1079 | code = 0; | |
1080 | break; | |
1081 | } | |
1082 | icmp_error(mcopy, type, code, dest); | |
1083 | } |