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