Annotation of src/sys/netinet/ip_input.c, Revision 1.24
1.24 ! mycroft 1: /* $NetBSD: ip_input.c,v 1.23 1995/06/12 06:46:36 mycroft Exp $ */
1.14 cgd 2:
1.1 cgd 3: /*
1.13 mycroft 4: * Copyright (c) 1982, 1986, 1988, 1993
5: * The Regents of the University of California. All rights reserved.
1.1 cgd 6: *
7: * Redistribution and use in source and binary forms, with or without
8: * modification, are permitted provided that the following conditions
9: * are met:
10: * 1. Redistributions of source code must retain the above copyright
11: * notice, this list of conditions and the following disclaimer.
12: * 2. Redistributions in binary form must reproduce the above copyright
13: * notice, this list of conditions and the following disclaimer in the
14: * documentation and/or other materials provided with the distribution.
15: * 3. All advertising materials mentioning features or use of this software
16: * must display the following acknowledgement:
17: * This product includes software developed by the University of
18: * California, Berkeley and its contributors.
19: * 4. Neither the name of the University nor the names of its contributors
20: * may be used to endorse or promote products derived from this software
21: * without specific prior written permission.
22: *
23: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33: * SUCH DAMAGE.
34: *
1.14 cgd 35: * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
1.1 cgd 36: */
37:
1.5 mycroft 38: #include <sys/param.h>
39: #include <sys/systm.h>
40: #include <sys/malloc.h>
41: #include <sys/mbuf.h>
42: #include <sys/domain.h>
43: #include <sys/protosw.h>
44: #include <sys/socket.h>
45: #include <sys/errno.h>
46: #include <sys/time.h>
47: #include <sys/kernel.h>
1.1 cgd 48:
1.5 mycroft 49: #include <net/if.h>
50: #include <net/route.h>
1.1 cgd 51:
1.5 mycroft 52: #include <netinet/in.h>
53: #include <netinet/in_systm.h>
54: #include <netinet/ip.h>
55: #include <netinet/in_pcb.h>
56: #include <netinet/in_var.h>
57: #include <netinet/ip_var.h>
58: #include <netinet/ip_icmp.h>
1.1 cgd 59:
60: #ifndef IPFORWARDING
61: #ifdef GATEWAY
62: #define IPFORWARDING 1 /* forward IP packets not for us */
63: #else /* GATEWAY */
64: #define IPFORWARDING 0 /* don't forward IP packets not for us */
65: #endif /* GATEWAY */
66: #endif /* IPFORWARDING */
67: #ifndef IPSENDREDIRECTS
68: #define IPSENDREDIRECTS 1
69: #endif
70: int ipforwarding = IPFORWARDING;
71: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 72: int ip_defttl = IPDEFTTL;
1.1 cgd 73: #ifdef DIAGNOSTIC
74: int ipprintfs = 0;
75: #endif
76:
77: extern struct domain inetdomain;
78: extern struct protosw inetsw[];
79: u_char ip_protox[IPPROTO_MAX];
80: int ipqmaxlen = IFQ_MAXLEN;
1.22 mycroft 81: struct in_ifaddrhead in_ifaddr;
1.13 mycroft 82: struct ifqueue ipintrq;
1.1 cgd 83:
84: /*
85: * We need to save the IP options in case a protocol wants to respond
86: * to an incoming packet over the same route if the packet got here
87: * using IP source routing. This allows connection establishment and
88: * maintenance when the remote end is on a network that is not known
89: * to us.
90: */
91: int ip_nhops = 0;
92: static struct ip_srcrt {
93: struct in_addr dst; /* final destination */
94: char nop; /* one NOP to align */
95: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
96: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
97: } ip_srcrt;
98:
1.13 mycroft 99: static void save_rte __P((u_char *, struct in_addr));
1.1 cgd 100: /*
101: * IP initialization: fill in IP protocol switch table.
102: * All protocols not implemented in kernel go to raw IP protocol handler.
103: */
1.8 mycroft 104: void
1.1 cgd 105: ip_init()
106: {
107: register struct protosw *pr;
108: register int i;
109:
110: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
111: if (pr == 0)
112: panic("ip_init");
113: for (i = 0; i < IPPROTO_MAX; i++)
114: ip_protox[i] = pr - inetsw;
115: for (pr = inetdomain.dom_protosw;
116: pr < inetdomain.dom_protoswNPROTOSW; pr++)
117: if (pr->pr_domain->dom_family == PF_INET &&
118: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
119: ip_protox[pr->pr_protocol] = pr - inetsw;
120: ipq.next = ipq.prev = &ipq;
121: ip_id = time.tv_sec & 0xffff;
122: ipintrq.ifq_maxlen = ipqmaxlen;
1.22 mycroft 123: TAILQ_INIT(&in_ifaddr);
1.1 cgd 124: }
125:
126: struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
127: struct route ipforward_rt;
128:
129: /*
130: * Ip input routine. Checksum and byte swap header. If fragmented
131: * try to reassemble. Process options. Pass to next level.
132: */
1.8 mycroft 133: void
1.1 cgd 134: ipintr()
135: {
136: register struct ip *ip;
137: register struct mbuf *m;
138: register struct ipq *fp;
139: register struct in_ifaddr *ia;
140: int hlen, s;
141:
142: next:
143: /*
144: * Get next datagram off input queue and get IP header
145: * in first mbuf.
146: */
147: s = splimp();
148: IF_DEQUEUE(&ipintrq, m);
149: splx(s);
1.13 mycroft 150: if (m == 0)
1.1 cgd 151: return;
152: #ifdef DIAGNOSTIC
153: if ((m->m_flags & M_PKTHDR) == 0)
154: panic("ipintr no HDR");
155: #endif
156: /*
157: * If no IP addresses have been set yet but the interfaces
158: * are receiving, can't do anything with incoming packets yet.
159: */
1.22 mycroft 160: if (in_ifaddr.tqh_first == 0)
1.1 cgd 161: goto bad;
162: ipstat.ips_total++;
163: if (m->m_len < sizeof (struct ip) &&
164: (m = m_pullup(m, sizeof (struct ip))) == 0) {
165: ipstat.ips_toosmall++;
166: goto next;
167: }
168: ip = mtod(m, struct ip *);
1.13 mycroft 169: if (ip->ip_v != IPVERSION) {
170: ipstat.ips_badvers++;
171: goto bad;
172: }
1.1 cgd 173: hlen = ip->ip_hl << 2;
174: if (hlen < sizeof(struct ip)) { /* minimum header length */
175: ipstat.ips_badhlen++;
176: goto bad;
177: }
178: if (hlen > m->m_len) {
179: if ((m = m_pullup(m, hlen)) == 0) {
180: ipstat.ips_badhlen++;
181: goto next;
182: }
183: ip = mtod(m, struct ip *);
184: }
185: if (ip->ip_sum = in_cksum(m, hlen)) {
186: ipstat.ips_badsum++;
187: goto bad;
188: }
189:
190: /*
191: * Convert fields to host representation.
192: */
193: NTOHS(ip->ip_len);
194: if (ip->ip_len < hlen) {
195: ipstat.ips_badlen++;
196: goto bad;
197: }
198: NTOHS(ip->ip_id);
199: NTOHS(ip->ip_off);
200:
201: /*
202: * Check that the amount of data in the buffers
203: * is as at least much as the IP header would have us expect.
204: * Trim mbufs if longer than we expect.
205: * Drop packet if shorter than we expect.
206: */
207: if (m->m_pkthdr.len < ip->ip_len) {
208: ipstat.ips_tooshort++;
209: goto bad;
210: }
211: if (m->m_pkthdr.len > ip->ip_len) {
212: if (m->m_len == m->m_pkthdr.len) {
213: m->m_len = ip->ip_len;
214: m->m_pkthdr.len = ip->ip_len;
215: } else
216: m_adj(m, ip->ip_len - m->m_pkthdr.len);
217: }
218:
219: /*
220: * Process options and, if not destined for us,
221: * ship it on. ip_dooptions returns 1 when an
222: * error was detected (causing an icmp message
223: * to be sent and the original packet to be freed).
224: */
225: ip_nhops = 0; /* for source routed packets */
226: if (hlen > sizeof (struct ip) && ip_dooptions(m))
227: goto next;
228:
229: /*
230: * Check our list of addresses, to see if the packet is for us.
231: */
1.22 mycroft 232: for (ia = in_ifaddr.tqh_first; ia; ia = ia->ia_list.tqe_next) {
1.20 mycroft 233: if (ip->ip_dst.s_addr == ia->ia_addr.sin_addr.s_addr)
1.1 cgd 234: goto ours;
235: if (
236: #ifdef DIRECTED_BROADCAST
237: ia->ia_ifp == m->m_pkthdr.rcvif &&
238: #endif
239: (ia->ia_ifp->if_flags & IFF_BROADCAST)) {
1.20 mycroft 240: if (ip->ip_dst.s_addr == ia->ia_broadaddr.sin_addr.s_addr ||
241: ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr ||
242: /*
243: * Look for all-0's host part (old broadcast addr),
244: * either for subnet or net.
245: */
246: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 247: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 248: goto ours;
249: }
250: }
1.18 mycroft 251: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 252: struct in_multi *inm;
253: #ifdef MROUTING
254: extern struct socket *ip_mrouter;
1.10 brezak 255:
256: if (m->m_flags & M_EXT) {
257: if ((m = m_pullup(m, hlen)) == 0) {
258: ipstat.ips_toosmall++;
259: goto next;
260: }
261: ip = mtod(m, struct ip *);
262: }
1.4 hpeyerl 263:
264: if (ip_mrouter) {
265: /*
266: * If we are acting as a multicast router, all
267: * incoming multicast packets are passed to the
268: * kernel-level multicast forwarding function.
269: * The packet is returned (relatively) intact; if
270: * ip_mforward() returns a non-zero value, the packet
271: * must be discarded, else it may be accepted below.
272: *
273: * (The IP ident field is put in the same byte order
274: * as expected when ip_mforward() is called from
275: * ip_output().)
276: */
277: ip->ip_id = htons(ip->ip_id);
1.13 mycroft 278: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
279: ipstat.ips_cantforward++;
1.4 hpeyerl 280: m_freem(m);
281: goto next;
282: }
283: ip->ip_id = ntohs(ip->ip_id);
284:
285: /*
286: * The process-level routing demon needs to receive
287: * all multicast IGMP packets, whether or not this
288: * host belongs to their destination groups.
289: */
290: if (ip->ip_p == IPPROTO_IGMP)
291: goto ours;
1.13 mycroft 292: ipstat.ips_forward++;
1.4 hpeyerl 293: }
294: #endif
295: /*
296: * See if we belong to the destination multicast group on the
297: * arrival interface.
298: */
299: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
300: if (inm == NULL) {
1.13 mycroft 301: ipstat.ips_cantforward++;
1.4 hpeyerl 302: m_freem(m);
303: goto next;
304: }
305: goto ours;
306: }
1.19 mycroft 307: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
308: ip->ip_dst.s_addr == INADDR_ANY)
1.1 cgd 309: goto ours;
310:
311: /*
312: * Not for us; forward if possible and desirable.
313: */
314: if (ipforwarding == 0) {
315: ipstat.ips_cantforward++;
316: m_freem(m);
317: } else
318: ip_forward(m, 0);
319: goto next;
320:
321: ours:
322: /*
323: * If offset or IP_MF are set, must reassemble.
324: * Otherwise, nothing need be done.
325: * (We could look in the reassembly queue to see
326: * if the packet was previously fragmented,
327: * but it's not worth the time; just let them time out.)
328: */
329: if (ip->ip_off &~ IP_DF) {
330: if (m->m_flags & M_EXT) { /* XXX */
331: if ((m = m_pullup(m, sizeof (struct ip))) == 0) {
332: ipstat.ips_toosmall++;
333: goto next;
334: }
335: ip = mtod(m, struct ip *);
336: }
337: /*
338: * Look for queue of fragments
339: * of this datagram.
340: */
341: for (fp = ipq.next; fp != &ipq; fp = fp->next)
342: if (ip->ip_id == fp->ipq_id &&
343: ip->ip_src.s_addr == fp->ipq_src.s_addr &&
344: ip->ip_dst.s_addr == fp->ipq_dst.s_addr &&
345: ip->ip_p == fp->ipq_p)
346: goto found;
347: fp = 0;
348: found:
349:
350: /*
351: * Adjust ip_len to not reflect header,
352: * set ip_mff if more fragments are expected,
353: * convert offset of this to bytes.
354: */
355: ip->ip_len -= hlen;
1.13 mycroft 356: ((struct ipasfrag *)ip)->ipf_mff &= ~1;
1.16 cgd 357: if (ip->ip_off & IP_MF) {
358: /*
359: * Make sure that fragments have a data length
360: * that's a non-zero multiple of 8 bytes.
361: */
1.17 cgd 362: if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
1.16 cgd 363: ipstat.ips_badfrags++;
364: goto bad;
365: }
1.13 mycroft 366: ((struct ipasfrag *)ip)->ipf_mff |= 1;
1.16 cgd 367: }
1.1 cgd 368: ip->ip_off <<= 3;
369:
370: /*
371: * If datagram marked as having more fragments
372: * or if this is not the first fragment,
373: * attempt reassembly; if it succeeds, proceed.
374: */
1.13 mycroft 375: if (((struct ipasfrag *)ip)->ipf_mff & 1 || ip->ip_off) {
1.1 cgd 376: ipstat.ips_fragments++;
377: ip = ip_reass((struct ipasfrag *)ip, fp);
378: if (ip == 0)
379: goto next;
1.13 mycroft 380: ipstat.ips_reassembled++;
1.1 cgd 381: m = dtom(ip);
382: } else
383: if (fp)
384: ip_freef(fp);
385: } else
386: ip->ip_len -= hlen;
387:
388: /*
389: * Switch out to protocol's input routine.
390: */
391: ipstat.ips_delivered++;
392: (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
393: goto next;
394: bad:
395: m_freem(m);
396: goto next;
397: }
398:
399: /*
400: * Take incoming datagram fragment and try to
401: * reassemble it into whole datagram. If a chain for
402: * reassembly of this datagram already exists, then it
403: * is given as fp; otherwise have to make a chain.
404: */
405: struct ip *
406: ip_reass(ip, fp)
407: register struct ipasfrag *ip;
408: register struct ipq *fp;
409: {
410: register struct mbuf *m = dtom(ip);
411: register struct ipasfrag *q;
412: struct mbuf *t;
413: int hlen = ip->ip_hl << 2;
414: int i, next;
415:
416: /*
417: * Presence of header sizes in mbufs
418: * would confuse code below.
419: */
420: m->m_data += hlen;
421: m->m_len -= hlen;
422:
423: /*
424: * If first fragment to arrive, create a reassembly queue.
425: */
426: if (fp == 0) {
427: if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL)
428: goto dropfrag;
429: fp = mtod(t, struct ipq *);
430: insque(fp, &ipq);
431: fp->ipq_ttl = IPFRAGTTL;
432: fp->ipq_p = ip->ip_p;
433: fp->ipq_id = ip->ip_id;
434: fp->ipq_next = fp->ipq_prev = (struct ipasfrag *)fp;
435: fp->ipq_src = ((struct ip *)ip)->ip_src;
436: fp->ipq_dst = ((struct ip *)ip)->ip_dst;
437: q = (struct ipasfrag *)fp;
438: goto insert;
439: }
440:
441: /*
442: * Find a segment which begins after this one does.
443: */
444: for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next)
445: if (q->ip_off > ip->ip_off)
446: break;
447:
448: /*
449: * If there is a preceding segment, it may provide some of
450: * our data already. If so, drop the data from the incoming
451: * segment. If it provides all of our data, drop us.
452: */
453: if (q->ipf_prev != (struct ipasfrag *)fp) {
454: i = q->ipf_prev->ip_off + q->ipf_prev->ip_len - ip->ip_off;
455: if (i > 0) {
456: if (i >= ip->ip_len)
457: goto dropfrag;
458: m_adj(dtom(ip), i);
459: ip->ip_off += i;
460: ip->ip_len -= i;
461: }
462: }
463:
464: /*
465: * While we overlap succeeding segments trim them or,
466: * if they are completely covered, dequeue them.
467: */
468: while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) {
469: i = (ip->ip_off + ip->ip_len) - q->ip_off;
470: if (i < q->ip_len) {
471: q->ip_len -= i;
472: q->ip_off += i;
473: m_adj(dtom(q), i);
474: break;
475: }
476: q = q->ipf_next;
477: m_freem(dtom(q->ipf_prev));
478: ip_deq(q->ipf_prev);
479: }
480:
481: insert:
482: /*
483: * Stick new segment in its place;
484: * check for complete reassembly.
485: */
486: ip_enq(ip, q->ipf_prev);
487: next = 0;
488: for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) {
489: if (q->ip_off != next)
490: return (0);
491: next += q->ip_len;
492: }
1.13 mycroft 493: if (q->ipf_prev->ipf_mff & 1)
1.1 cgd 494: return (0);
495:
496: /*
497: * Reassembly is complete; concatenate fragments.
498: */
499: q = fp->ipq_next;
500: m = dtom(q);
501: t = m->m_next;
502: m->m_next = 0;
503: m_cat(m, t);
504: q = q->ipf_next;
505: while (q != (struct ipasfrag *)fp) {
506: t = dtom(q);
507: q = q->ipf_next;
508: m_cat(m, t);
509: }
510:
511: /*
512: * Create header for new ip packet by
513: * modifying header of first packet;
514: * dequeue and discard fragment reassembly header.
515: * Make header visible.
516: */
517: ip = fp->ipq_next;
518: ip->ip_len = next;
1.13 mycroft 519: ip->ipf_mff &= ~1;
1.1 cgd 520: ((struct ip *)ip)->ip_src = fp->ipq_src;
521: ((struct ip *)ip)->ip_dst = fp->ipq_dst;
522: remque(fp);
523: (void) m_free(dtom(fp));
524: m = dtom(ip);
525: m->m_len += (ip->ip_hl << 2);
526: m->m_data -= (ip->ip_hl << 2);
527: /* some debugging cruft by sklower, below, will go away soon */
528: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
529: register int plen = 0;
530: for (t = m; m; m = m->m_next)
531: plen += m->m_len;
532: t->m_pkthdr.len = plen;
533: }
534: return ((struct ip *)ip);
535:
536: dropfrag:
537: ipstat.ips_fragdropped++;
538: m_freem(m);
539: return (0);
540: }
541:
542: /*
543: * Free a fragment reassembly header and all
544: * associated datagrams.
545: */
1.8 mycroft 546: void
1.1 cgd 547: ip_freef(fp)
548: struct ipq *fp;
549: {
550: register struct ipasfrag *q, *p;
551:
552: for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) {
553: p = q->ipf_next;
554: ip_deq(q);
555: m_freem(dtom(q));
556: }
557: remque(fp);
558: (void) m_free(dtom(fp));
559: }
560:
561: /*
562: * Put an ip fragment on a reassembly chain.
563: * Like insque, but pointers in middle of structure.
564: */
1.8 mycroft 565: void
1.1 cgd 566: ip_enq(p, prev)
567: register struct ipasfrag *p, *prev;
568: {
569:
570: p->ipf_prev = prev;
571: p->ipf_next = prev->ipf_next;
572: prev->ipf_next->ipf_prev = p;
573: prev->ipf_next = p;
574: }
575:
576: /*
577: * To ip_enq as remque is to insque.
578: */
1.8 mycroft 579: void
1.1 cgd 580: ip_deq(p)
581: register struct ipasfrag *p;
582: {
583:
584: p->ipf_prev->ipf_next = p->ipf_next;
585: p->ipf_next->ipf_prev = p->ipf_prev;
586: }
587:
588: /*
589: * IP timer processing;
590: * if a timer expires on a reassembly
591: * queue, discard it.
592: */
1.8 mycroft 593: void
1.1 cgd 594: ip_slowtimo()
595: {
596: register struct ipq *fp;
1.24 ! mycroft 597: int s = splsoftnet();
1.1 cgd 598:
599: fp = ipq.next;
600: if (fp == 0) {
601: splx(s);
602: return;
603: }
604: while (fp != &ipq) {
605: --fp->ipq_ttl;
606: fp = fp->next;
607: if (fp->prev->ipq_ttl == 0) {
608: ipstat.ips_fragtimeout++;
609: ip_freef(fp->prev);
610: }
611: }
612: splx(s);
613: }
614:
615: /*
616: * Drain off all datagram fragments.
617: */
1.8 mycroft 618: void
1.1 cgd 619: ip_drain()
620: {
621:
622: while (ipq.next != &ipq) {
623: ipstat.ips_fragdropped++;
624: ip_freef(ipq.next);
625: }
626: }
627:
628: /*
629: * Do option processing on a datagram,
630: * possibly discarding it if bad options are encountered,
631: * or forwarding it if source-routed.
632: * Returns 1 if packet has been forwarded/freed,
633: * 0 if the packet should be processed further.
634: */
1.8 mycroft 635: int
1.1 cgd 636: ip_dooptions(m)
637: struct mbuf *m;
638: {
639: register struct ip *ip = mtod(m, struct ip *);
640: register u_char *cp;
641: register struct ip_timestamp *ipt;
642: register struct in_ifaddr *ia;
643: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.13 mycroft 644: struct in_addr *sin, dst;
1.1 cgd 645: n_time ntime;
646:
1.13 mycroft 647: dst = ip->ip_dst;
1.1 cgd 648: cp = (u_char *)(ip + 1);
649: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
650: for (; cnt > 0; cnt -= optlen, cp += optlen) {
651: opt = cp[IPOPT_OPTVAL];
652: if (opt == IPOPT_EOL)
653: break;
654: if (opt == IPOPT_NOP)
655: optlen = 1;
656: else {
657: optlen = cp[IPOPT_OLEN];
658: if (optlen <= 0 || optlen > cnt) {
659: code = &cp[IPOPT_OLEN] - (u_char *)ip;
660: goto bad;
661: }
662: }
663: switch (opt) {
664:
665: default:
666: break;
667:
668: /*
669: * Source routing with record.
670: * Find interface with current destination address.
671: * If none on this machine then drop if strictly routed,
672: * or do nothing if loosely routed.
673: * Record interface address and bring up next address
674: * component. If strictly routed make sure next
675: * address is on directly accessible net.
676: */
677: case IPOPT_LSRR:
678: case IPOPT_SSRR:
679: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
680: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
681: goto bad;
682: }
683: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 684: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 685: if (ia == 0) {
686: if (opt == IPOPT_SSRR) {
687: type = ICMP_UNREACH;
688: code = ICMP_UNREACH_SRCFAIL;
689: goto bad;
690: }
691: /*
692: * Loose routing, and not at next destination
693: * yet; nothing to do except forward.
694: */
695: break;
696: }
697: off--; /* 0 origin */
698: if (off > optlen - sizeof(struct in_addr)) {
699: /*
700: * End of source route. Should be for us.
701: */
702: save_rte(cp, ip->ip_src);
703: break;
704: }
705: /*
706: * locate outgoing interface
707: */
708: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
709: sizeof(ipaddr.sin_addr));
710: if (opt == IPOPT_SSRR) {
711: #define INA struct in_ifaddr *
712: #define SA struct sockaddr *
713: if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0)
1.13 mycroft 714: ia = (INA)ifa_ifwithnet((SA)&ipaddr);
1.1 cgd 715: } else
716: ia = ip_rtaddr(ipaddr.sin_addr);
717: if (ia == 0) {
718: type = ICMP_UNREACH;
719: code = ICMP_UNREACH_SRCFAIL;
720: goto bad;
721: }
722: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 723: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 724: (caddr_t)(cp + off), sizeof(struct in_addr));
725: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 726: /*
727: * Let ip_intr's mcast routing check handle mcast pkts
728: */
1.18 mycroft 729: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 730: break;
731:
732: case IPOPT_RR:
733: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
734: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
735: goto bad;
736: }
737: /*
738: * If no space remains, ignore.
739: */
740: off--; /* 0 origin */
741: if (off > optlen - sizeof(struct in_addr))
742: break;
743: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
744: sizeof(ipaddr.sin_addr));
745: /*
746: * locate outgoing interface; if we're the destination,
747: * use the incoming interface (should be same).
748: */
749: if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
750: (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
751: type = ICMP_UNREACH;
752: code = ICMP_UNREACH_HOST;
753: goto bad;
754: }
1.20 mycroft 755: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 756: (caddr_t)(cp + off), sizeof(struct in_addr));
757: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
758: break;
759:
760: case IPOPT_TS:
761: code = cp - (u_char *)ip;
762: ipt = (struct ip_timestamp *)cp;
763: if (ipt->ipt_len < 5)
764: goto bad;
1.15 cgd 765: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.1 cgd 766: if (++ipt->ipt_oflw == 0)
767: goto bad;
768: break;
769: }
770: sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
771: switch (ipt->ipt_flg) {
772:
773: case IPOPT_TS_TSONLY:
774: break;
775:
776: case IPOPT_TS_TSANDADDR:
777: if (ipt->ipt_ptr + sizeof(n_time) +
778: sizeof(struct in_addr) > ipt->ipt_len)
779: goto bad;
1.13 mycroft 780: ipaddr.sin_addr = dst;
781: ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
782: m->m_pkthdr.rcvif);
783: if (ia == 0)
784: continue;
1.20 mycroft 785: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 786: (caddr_t)sin, sizeof(struct in_addr));
787: ipt->ipt_ptr += sizeof(struct in_addr);
788: break;
789:
790: case IPOPT_TS_PRESPEC:
791: if (ipt->ipt_ptr + sizeof(n_time) +
792: sizeof(struct in_addr) > ipt->ipt_len)
793: goto bad;
794: bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
795: sizeof(struct in_addr));
796: if (ifa_ifwithaddr((SA)&ipaddr) == 0)
797: continue;
798: ipt->ipt_ptr += sizeof(struct in_addr);
799: break;
800:
801: default:
802: goto bad;
803: }
804: ntime = iptime();
805: bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
806: sizeof(n_time));
807: ipt->ipt_ptr += sizeof(n_time);
808: }
809: }
810: if (forward) {
811: ip_forward(m, 1);
812: return (1);
1.13 mycroft 813: }
814: return (0);
1.1 cgd 815: bad:
1.13 mycroft 816: ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */
817: icmp_error(m, type, code, 0, 0);
818: ipstat.ips_badoptions++;
1.1 cgd 819: return (1);
820: }
821:
822: /*
823: * Given address of next destination (final or next hop),
824: * return internet address info of interface to be used to get there.
825: */
826: struct in_ifaddr *
827: ip_rtaddr(dst)
828: struct in_addr dst;
829: {
830: register struct sockaddr_in *sin;
831:
1.19 mycroft 832: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 833:
834: if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) {
835: if (ipforward_rt.ro_rt) {
836: RTFREE(ipforward_rt.ro_rt);
837: ipforward_rt.ro_rt = 0;
838: }
839: sin->sin_family = AF_INET;
840: sin->sin_len = sizeof(*sin);
841: sin->sin_addr = dst;
842:
843: rtalloc(&ipforward_rt);
844: }
845: if (ipforward_rt.ro_rt == 0)
846: return ((struct in_ifaddr *)0);
1.19 mycroft 847: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 848: }
849:
850: /*
851: * Save incoming source route for use in replies,
852: * to be picked up later by ip_srcroute if the receiver is interested.
853: */
1.13 mycroft 854: void
1.1 cgd 855: save_rte(option, dst)
856: u_char *option;
857: struct in_addr dst;
858: {
859: unsigned olen;
860:
861: olen = option[IPOPT_OLEN];
862: #ifdef DIAGNOSTIC
863: if (ipprintfs)
864: printf("save_rte: olen %d\n", olen);
865: #endif
866: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
867: return;
868: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
869: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
870: ip_srcrt.dst = dst;
871: }
872:
873: /*
874: * Retrieve incoming source route for use in replies,
875: * in the same form used by setsockopt.
876: * The first hop is placed before the options, will be removed later.
877: */
878: struct mbuf *
879: ip_srcroute()
880: {
881: register struct in_addr *p, *q;
882: register struct mbuf *m;
883:
884: if (ip_nhops == 0)
885: return ((struct mbuf *)0);
886: m = m_get(M_DONTWAIT, MT_SOOPTS);
887: if (m == 0)
888: return ((struct mbuf *)0);
889:
1.13 mycroft 890: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 891:
892: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
893: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
894: OPTSIZ;
895: #ifdef DIAGNOSTIC
896: if (ipprintfs)
897: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
898: #endif
899:
900: /*
901: * First save first hop for return route
902: */
903: p = &ip_srcrt.route[ip_nhops - 1];
904: *(mtod(m, struct in_addr *)) = *p--;
905: #ifdef DIAGNOSTIC
906: if (ipprintfs)
907: printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr));
908: #endif
909:
910: /*
911: * Copy option fields and padding (nop) to mbuf.
912: */
913: ip_srcrt.nop = IPOPT_NOP;
914: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
915: bcopy((caddr_t)&ip_srcrt.nop,
916: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
917: q = (struct in_addr *)(mtod(m, caddr_t) +
918: sizeof(struct in_addr) + OPTSIZ);
919: #undef OPTSIZ
920: /*
921: * Record return path as an IP source route,
922: * reversing the path (pointers are now aligned).
923: */
924: while (p >= ip_srcrt.route) {
925: #ifdef DIAGNOSTIC
926: if (ipprintfs)
927: printf(" %lx", ntohl(q->s_addr));
928: #endif
929: *q++ = *p--;
930: }
931: /*
932: * Last hop goes to final destination.
933: */
934: *q = ip_srcrt.dst;
935: #ifdef DIAGNOSTIC
936: if (ipprintfs)
937: printf(" %lx\n", ntohl(q->s_addr));
938: #endif
939: return (m);
940: }
941:
942: /*
943: * Strip out IP options, at higher
944: * level protocol in the kernel.
945: * Second argument is buffer to which options
946: * will be moved, and return value is their length.
947: * XXX should be deleted; last arg currently ignored.
948: */
1.8 mycroft 949: void
1.1 cgd 950: ip_stripoptions(m, mopt)
951: register struct mbuf *m;
952: struct mbuf *mopt;
953: {
954: register int i;
955: struct ip *ip = mtod(m, struct ip *);
956: register caddr_t opts;
957: int olen;
958:
959: olen = (ip->ip_hl<<2) - sizeof (struct ip);
960: opts = (caddr_t)(ip + 1);
961: i = m->m_len - (sizeof (struct ip) + olen);
962: bcopy(opts + olen, opts, (unsigned)i);
963: m->m_len -= olen;
964: if (m->m_flags & M_PKTHDR)
965: m->m_pkthdr.len -= olen;
966: ip->ip_hl = sizeof(struct ip) >> 2;
967: }
968:
1.23 mycroft 969: int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 970: 0, 0, 0, 0,
971: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
972: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
973: EMSGSIZE, EHOSTUNREACH, 0, 0,
974: 0, 0, 0, 0,
975: ENOPROTOOPT
976: };
977:
978: /*
979: * Forward a packet. If some error occurs return the sender
980: * an icmp packet. Note we can't always generate a meaningful
981: * icmp message because icmp doesn't have a large enough repertoire
982: * of codes and types.
983: *
984: * If not forwarding, just drop the packet. This could be confusing
985: * if ipforwarding was zero but some routing protocol was advancing
986: * us as a gateway to somewhere. However, we must let the routing
987: * protocol deal with that.
988: *
989: * The srcrt parameter indicates whether the packet is being forwarded
990: * via a source route.
991: */
1.13 mycroft 992: void
1.1 cgd 993: ip_forward(m, srcrt)
994: struct mbuf *m;
995: int srcrt;
996: {
997: register struct ip *ip = mtod(m, struct ip *);
998: register struct sockaddr_in *sin;
999: register struct rtentry *rt;
1000: int error, type = 0, code;
1001: struct mbuf *mcopy;
1.13 mycroft 1002: n_long dest;
1003: struct ifnet *destifp;
1.1 cgd 1004:
1.13 mycroft 1005: dest = 0;
1.1 cgd 1006: #ifdef DIAGNOSTIC
1007: if (ipprintfs)
1008: printf("forward: src %x dst %x ttl %x\n", ip->ip_src,
1009: ip->ip_dst, ip->ip_ttl);
1010: #endif
1011: if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1012: ipstat.ips_cantforward++;
1013: m_freem(m);
1014: return;
1015: }
1016: HTONS(ip->ip_id);
1017: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1018: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1019: return;
1020: }
1021: ip->ip_ttl -= IPTTLDEC;
1022:
1.19 mycroft 1023: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1024: if ((rt = ipforward_rt.ro_rt) == 0 ||
1025: ip->ip_dst.s_addr != sin->sin_addr.s_addr) {
1026: if (ipforward_rt.ro_rt) {
1027: RTFREE(ipforward_rt.ro_rt);
1028: ipforward_rt.ro_rt = 0;
1029: }
1030: sin->sin_family = AF_INET;
1031: sin->sin_len = sizeof(*sin);
1032: sin->sin_addr = ip->ip_dst;
1033:
1034: rtalloc(&ipforward_rt);
1035: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1036: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1037: return;
1038: }
1039: rt = ipforward_rt.ro_rt;
1040: }
1041:
1042: /*
1043: * Save at most 64 bytes of the packet in case
1044: * we need to generate an ICMP message to the src.
1045: */
1046: mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64));
1047:
1048: /*
1049: * If forwarding packet using same interface that it came in on,
1050: * perhaps should send a redirect to sender to shortcut a hop.
1051: * Only send redirect if source is sending directly to us,
1052: * and if packet was not source routed (or has any options).
1053: * Also, don't send redirect if forwarding using a default route
1054: * or a route modified by a redirect.
1055: */
1056: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1057: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1058: satosin(rt_key(rt))->sin_addr.s_addr != 0 &&
1059: ipsendredirects && !srcrt) {
1.19 mycroft 1060: if (rt->rt_ifa &&
1061: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1062: ifatoia(rt->rt_ifa)->ia_subnet) {
1.1 cgd 1063: if (rt->rt_flags & RTF_GATEWAY)
1.13 mycroft 1064: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1.1 cgd 1065: else
1.13 mycroft 1066: dest = ip->ip_dst.s_addr;
1067: /* Router requirements says to only send host redirects */
1.1 cgd 1068: type = ICMP_REDIRECT;
1.13 mycroft 1069: code = ICMP_REDIRECT_HOST;
1.1 cgd 1070: #ifdef DIAGNOSTIC
1071: if (ipprintfs)
1.15 cgd 1072: printf("redirect (%d) to %lx\n", code, (u_int32_t)dest);
1.1 cgd 1073: #endif
1074: }
1075: }
1076:
1.13 mycroft 1077: error = ip_output(m, (struct mbuf *)0, &ipforward_rt, IP_FORWARDING
1.9 mycroft 1078: #ifdef DIRECTED_BROADCAST
1.13 mycroft 1079: | IP_ALLOWBROADCAST
1.9 mycroft 1080: #endif
1.13 mycroft 1081: , 0);
1.1 cgd 1082: if (error)
1083: ipstat.ips_cantforward++;
1084: else {
1085: ipstat.ips_forward++;
1086: if (type)
1087: ipstat.ips_redirectsent++;
1088: else {
1089: if (mcopy)
1090: m_freem(mcopy);
1091: return;
1092: }
1093: }
1094: if (mcopy == NULL)
1095: return;
1.13 mycroft 1096: destifp = NULL;
1097:
1.1 cgd 1098: switch (error) {
1099:
1100: case 0: /* forwarded, but need redirect */
1101: /* type, code set above */
1102: break;
1103:
1104: case ENETUNREACH: /* shouldn't happen, checked above */
1105: case EHOSTUNREACH:
1106: case ENETDOWN:
1107: case EHOSTDOWN:
1108: default:
1109: type = ICMP_UNREACH;
1110: code = ICMP_UNREACH_HOST;
1111: break;
1112:
1113: case EMSGSIZE:
1114: type = ICMP_UNREACH;
1115: code = ICMP_UNREACH_NEEDFRAG;
1.13 mycroft 1116: if (ipforward_rt.ro_rt)
1117: destifp = ipforward_rt.ro_rt->rt_ifp;
1.1 cgd 1118: ipstat.ips_cantfrag++;
1119: break;
1120:
1121: case ENOBUFS:
1122: type = ICMP_SOURCEQUENCH;
1123: code = 0;
1124: break;
1125: }
1.13 mycroft 1126: icmp_error(mcopy, type, code, dest, destifp);
1127: }
1128:
1129: int
1130: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1131: int *name;
1132: u_int namelen;
1133: void *oldp;
1134: size_t *oldlenp;
1135: void *newp;
1136: size_t newlen;
1137: {
1138: /* All sysctl names at this level are terminal. */
1139: if (namelen != 1)
1140: return (ENOTDIR);
1141:
1142: switch (name[0]) {
1143: case IPCTL_FORWARDING:
1144: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1145: case IPCTL_SENDREDIRECTS:
1146: return (sysctl_int(oldp, oldlenp, newp, newlen,
1147: &ipsendredirects));
1148: case IPCTL_DEFTTL:
1149: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1150: #ifdef notyet
1151: case IPCTL_DEFMTU:
1152: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1153: #endif
1154: default:
1155: return (EOPNOTSUPP);
1156: }
1157: /* NOTREACHED */
1.1 cgd 1158: }
CVSweb <webmaster@jp.NetBSD.org>