Annotation of src/sys/netinet/ip_input.c, Revision 1.5
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
1.5 ! mycroft 34: * $Id: ip_input.c,v 1.4 1993/12/06 04:59:37 hpeyerl Exp $
1.1 cgd 35: */
36:
1.5 ! mycroft 37: #include <sys/param.h>
! 38: #include <sys/systm.h>
! 39: #include <sys/malloc.h>
! 40: #include <sys/mbuf.h>
! 41: #include <sys/domain.h>
! 42: #include <sys/protosw.h>
! 43: #include <sys/socket.h>
! 44: #include <sys/errno.h>
! 45: #include <sys/time.h>
! 46: #include <sys/kernel.h>
1.1 cgd 47:
1.5 ! mycroft 48: #include <net/if.h>
! 49: #include <net/route.h>
1.1 cgd 50:
1.5 ! mycroft 51: #include <netinet/in.h>
! 52: #include <netinet/in_systm.h>
! 53: #include <netinet/ip.h>
! 54: #include <netinet/in_pcb.h>
! 55: #include <netinet/in_var.h>
! 56: #include <netinet/ip_var.h>
! 57: #include <netinet/ip_icmp.h>
1.1 cgd 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: }
1.4 hpeyerl 270: #ifdef MULTICAST
271: if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) {
272: struct in_multi *inm;
273: #ifdef MROUTING
274: extern struct socket *ip_mrouter;
275:
276: if (ip_mrouter) {
277: /*
278: * If we are acting as a multicast router, all
279: * incoming multicast packets are passed to the
280: * kernel-level multicast forwarding function.
281: * The packet is returned (relatively) intact; if
282: * ip_mforward() returns a non-zero value, the packet
283: * must be discarded, else it may be accepted below.
284: *
285: * (The IP ident field is put in the same byte order
286: * as expected when ip_mforward() is called from
287: * ip_output().)
288: */
289: ip->ip_id = htons(ip->ip_id);
290: if (ip_mforward(ip, m->m_pkthdr.rcvif, m) != 0) {
291: m_freem(m);
292: goto next;
293: }
294: ip->ip_id = ntohs(ip->ip_id);
295:
296: /*
297: * The process-level routing demon needs to receive
298: * all multicast IGMP packets, whether or not this
299: * host belongs to their destination groups.
300: */
301: if (ip->ip_p == IPPROTO_IGMP)
302: goto ours;
303: }
304: #endif
305: /*
306: * See if we belong to the destination multicast group on the
307: * arrival interface.
308: */
309: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
310: if (inm == NULL) {
311: m_freem(m);
312: goto next;
313: }
314: goto ours;
315: }
316: #endif
1.1 cgd 317: if (ip->ip_dst.s_addr == (u_long)INADDR_BROADCAST)
318: goto ours;
319: if (ip->ip_dst.s_addr == INADDR_ANY)
320: goto ours;
321:
322: /*
323: * Not for us; forward if possible and desirable.
324: */
325: if (ipforwarding == 0) {
326: ipstat.ips_cantforward++;
327: m_freem(m);
328: } else
329: ip_forward(m, 0);
330: goto next;
331:
332: ours:
333: /*
334: * If offset or IP_MF are set, must reassemble.
335: * Otherwise, nothing need be done.
336: * (We could look in the reassembly queue to see
337: * if the packet was previously fragmented,
338: * but it's not worth the time; just let them time out.)
339: */
340: if (ip->ip_off &~ IP_DF) {
341: if (m->m_flags & M_EXT) { /* XXX */
342: if ((m = m_pullup(m, sizeof (struct ip))) == 0) {
343: ipstat.ips_toosmall++;
344: goto next;
345: }
346: ip = mtod(m, struct ip *);
347: }
348: /*
349: * Look for queue of fragments
350: * of this datagram.
351: */
352: for (fp = ipq.next; fp != &ipq; fp = fp->next)
353: if (ip->ip_id == fp->ipq_id &&
354: ip->ip_src.s_addr == fp->ipq_src.s_addr &&
355: ip->ip_dst.s_addr == fp->ipq_dst.s_addr &&
356: ip->ip_p == fp->ipq_p)
357: goto found;
358: fp = 0;
359: found:
360:
361: /*
362: * Adjust ip_len to not reflect header,
363: * set ip_mff if more fragments are expected,
364: * convert offset of this to bytes.
365: */
366: ip->ip_len -= hlen;
367: ((struct ipasfrag *)ip)->ipf_mff = 0;
368: if (ip->ip_off & IP_MF)
369: ((struct ipasfrag *)ip)->ipf_mff = 1;
370: ip->ip_off <<= 3;
371:
372: /*
373: * If datagram marked as having more fragments
374: * or if this is not the first fragment,
375: * attempt reassembly; if it succeeds, proceed.
376: */
377: if (((struct ipasfrag *)ip)->ipf_mff || ip->ip_off) {
378: ipstat.ips_fragments++;
379: ip = ip_reass((struct ipasfrag *)ip, fp);
380: if (ip == 0)
381: goto next;
382: else
383: ipstat.ips_reassembled++;
384: m = dtom(ip);
385: } else
386: if (fp)
387: ip_freef(fp);
388: } else
389: ip->ip_len -= hlen;
390:
391: /*
392: * Switch out to protocol's input routine.
393: */
394: ipstat.ips_delivered++;
395: (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
396: goto next;
397: bad:
398: m_freem(m);
399: goto next;
400: }
401:
402: /*
403: * Take incoming datagram fragment and try to
404: * reassemble it into whole datagram. If a chain for
405: * reassembly of this datagram already exists, then it
406: * is given as fp; otherwise have to make a chain.
407: */
408: struct ip *
409: ip_reass(ip, fp)
410: register struct ipasfrag *ip;
411: register struct ipq *fp;
412: {
413: register struct mbuf *m = dtom(ip);
414: register struct ipasfrag *q;
415: struct mbuf *t;
416: int hlen = ip->ip_hl << 2;
417: int i, next;
418:
419: /*
420: * Presence of header sizes in mbufs
421: * would confuse code below.
422: */
423: m->m_data += hlen;
424: m->m_len -= hlen;
425:
426: /*
427: * If first fragment to arrive, create a reassembly queue.
428: */
429: if (fp == 0) {
430: if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL)
431: goto dropfrag;
432: fp = mtod(t, struct ipq *);
433: insque(fp, &ipq);
434: fp->ipq_ttl = IPFRAGTTL;
435: fp->ipq_p = ip->ip_p;
436: fp->ipq_id = ip->ip_id;
437: fp->ipq_next = fp->ipq_prev = (struct ipasfrag *)fp;
438: fp->ipq_src = ((struct ip *)ip)->ip_src;
439: fp->ipq_dst = ((struct ip *)ip)->ip_dst;
440: q = (struct ipasfrag *)fp;
441: goto insert;
442: }
443:
444: /*
445: * Find a segment which begins after this one does.
446: */
447: for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next)
448: if (q->ip_off > ip->ip_off)
449: break;
450:
451: /*
452: * If there is a preceding segment, it may provide some of
453: * our data already. If so, drop the data from the incoming
454: * segment. If it provides all of our data, drop us.
455: */
456: if (q->ipf_prev != (struct ipasfrag *)fp) {
457: i = q->ipf_prev->ip_off + q->ipf_prev->ip_len - ip->ip_off;
458: if (i > 0) {
459: if (i >= ip->ip_len)
460: goto dropfrag;
461: m_adj(dtom(ip), i);
462: ip->ip_off += i;
463: ip->ip_len -= i;
464: }
465: }
466:
467: /*
468: * While we overlap succeeding segments trim them or,
469: * if they are completely covered, dequeue them.
470: */
471: while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) {
472: i = (ip->ip_off + ip->ip_len) - q->ip_off;
473: if (i < q->ip_len) {
474: q->ip_len -= i;
475: q->ip_off += i;
476: m_adj(dtom(q), i);
477: break;
478: }
479: q = q->ipf_next;
480: m_freem(dtom(q->ipf_prev));
481: ip_deq(q->ipf_prev);
482: }
483:
484: insert:
485: /*
486: * Stick new segment in its place;
487: * check for complete reassembly.
488: */
489: ip_enq(ip, q->ipf_prev);
490: next = 0;
491: for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) {
492: if (q->ip_off != next)
493: return (0);
494: next += q->ip_len;
495: }
496: if (q->ipf_prev->ipf_mff)
497: return (0);
498:
499: /*
500: * Reassembly is complete; concatenate fragments.
501: */
502: q = fp->ipq_next;
503: m = dtom(q);
504: t = m->m_next;
505: m->m_next = 0;
506: m_cat(m, t);
507: q = q->ipf_next;
508: while (q != (struct ipasfrag *)fp) {
509: t = dtom(q);
510: q = q->ipf_next;
511: m_cat(m, t);
512: }
513:
514: /*
515: * Create header for new ip packet by
516: * modifying header of first packet;
517: * dequeue and discard fragment reassembly header.
518: * Make header visible.
519: */
520: ip = fp->ipq_next;
521: ip->ip_len = next;
522: ((struct ip *)ip)->ip_src = fp->ipq_src;
523: ((struct ip *)ip)->ip_dst = fp->ipq_dst;
524: remque(fp);
525: (void) m_free(dtom(fp));
526: m = dtom(ip);
527: m->m_len += (ip->ip_hl << 2);
528: m->m_data -= (ip->ip_hl << 2);
529: /* some debugging cruft by sklower, below, will go away soon */
530: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
531: register int plen = 0;
532: for (t = m; m; m = m->m_next)
533: plen += m->m_len;
534: t->m_pkthdr.len = plen;
535: }
536: return ((struct ip *)ip);
537:
538: dropfrag:
539: ipstat.ips_fragdropped++;
540: m_freem(m);
541: return (0);
542: }
543:
544: /*
545: * Free a fragment reassembly header and all
546: * associated datagrams.
547: */
548: ip_freef(fp)
549: struct ipq *fp;
550: {
551: register struct ipasfrag *q, *p;
552:
553: for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) {
554: p = q->ipf_next;
555: ip_deq(q);
556: m_freem(dtom(q));
557: }
558: remque(fp);
559: (void) m_free(dtom(fp));
560: }
561:
562: /*
563: * Put an ip fragment on a reassembly chain.
564: * Like insque, but pointers in middle of structure.
565: */
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: */
579: ip_deq(p)
580: register struct ipasfrag *p;
581: {
582:
583: p->ipf_prev->ipf_next = p->ipf_next;
584: p->ipf_next->ipf_prev = p->ipf_prev;
585: }
586:
587: /*
588: * IP timer processing;
589: * if a timer expires on a reassembly
590: * queue, discard it.
591: */
592: ip_slowtimo()
593: {
594: register struct ipq *fp;
595: int s = splnet();
596:
597: fp = ipq.next;
598: if (fp == 0) {
599: splx(s);
600: return;
601: }
602: while (fp != &ipq) {
603: --fp->ipq_ttl;
604: fp = fp->next;
605: if (fp->prev->ipq_ttl == 0) {
606: ipstat.ips_fragtimeout++;
607: ip_freef(fp->prev);
608: }
609: }
610: splx(s);
611: }
612:
613: /*
614: * Drain off all datagram fragments.
615: */
616: ip_drain()
617: {
618:
619: while (ipq.next != &ipq) {
620: ipstat.ips_fragdropped++;
621: ip_freef(ipq.next);
622: }
623: }
624:
625: extern struct in_ifaddr *ifptoia();
626: struct in_ifaddr *ip_rtaddr();
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: */
635: ip_dooptions(m)
636: struct mbuf *m;
637: {
638: register struct ip *ip = mtod(m, struct ip *);
639: register u_char *cp;
640: register struct ip_timestamp *ipt;
641: register struct in_ifaddr *ia;
642: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
643: struct in_addr *sin;
644: n_time ntime;
645:
646: cp = (u_char *)(ip + 1);
647: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
648: for (; cnt > 0; cnt -= optlen, cp += optlen) {
649: opt = cp[IPOPT_OPTVAL];
650: if (opt == IPOPT_EOL)
651: break;
652: if (opt == IPOPT_NOP)
653: optlen = 1;
654: else {
655: optlen = cp[IPOPT_OLEN];
656: if (optlen <= 0 || optlen > cnt) {
657: code = &cp[IPOPT_OLEN] - (u_char *)ip;
658: goto bad;
659: }
660: }
661: switch (opt) {
662:
663: default:
664: break;
665:
666: /*
667: * Source routing with record.
668: * Find interface with current destination address.
669: * If none on this machine then drop if strictly routed,
670: * or do nothing if loosely routed.
671: * Record interface address and bring up next address
672: * component. If strictly routed make sure next
673: * address is on directly accessible net.
674: */
675: case IPOPT_LSRR:
676: case IPOPT_SSRR:
677: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
678: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
679: goto bad;
680: }
681: ipaddr.sin_addr = ip->ip_dst;
682: ia = (struct in_ifaddr *)
683: ifa_ifwithaddr((struct sockaddr *)&ipaddr);
684: if (ia == 0) {
685: if (opt == IPOPT_SSRR) {
686: type = ICMP_UNREACH;
687: code = ICMP_UNREACH_SRCFAIL;
688: goto bad;
689: }
690: /*
691: * Loose routing, and not at next destination
692: * yet; nothing to do except forward.
693: */
694: break;
695: }
696: off--; /* 0 origin */
697: if (off > optlen - sizeof(struct in_addr)) {
698: /*
699: * End of source route. Should be for us.
700: */
701: save_rte(cp, ip->ip_src);
702: break;
703: }
704: /*
705: * locate outgoing interface
706: */
707: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
708: sizeof(ipaddr.sin_addr));
709: if (opt == IPOPT_SSRR) {
710: #define INA struct in_ifaddr *
711: #define SA struct sockaddr *
712: if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0)
713: ia = in_iaonnetof(in_netof(ipaddr.sin_addr));
714: } else
715: ia = ip_rtaddr(ipaddr.sin_addr);
716: if (ia == 0) {
717: type = ICMP_UNREACH;
718: code = ICMP_UNREACH_SRCFAIL;
719: goto bad;
720: }
721: ip->ip_dst = ipaddr.sin_addr;
722: bcopy((caddr_t)&(IA_SIN(ia)->sin_addr),
723: (caddr_t)(cp + off), sizeof(struct in_addr));
724: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
725: forward = 1;
726: break;
727:
728: case IPOPT_RR:
729: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
730: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
731: goto bad;
732: }
733: /*
734: * If no space remains, ignore.
735: */
736: off--; /* 0 origin */
737: if (off > optlen - sizeof(struct in_addr))
738: break;
739: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
740: sizeof(ipaddr.sin_addr));
741: /*
742: * locate outgoing interface; if we're the destination,
743: * use the incoming interface (should be same).
744: */
745: if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
746: (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
747: type = ICMP_UNREACH;
748: code = ICMP_UNREACH_HOST;
749: goto bad;
750: }
751: bcopy((caddr_t)&(IA_SIN(ia)->sin_addr),
752: (caddr_t)(cp + off), sizeof(struct in_addr));
753: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
754: break;
755:
756: case IPOPT_TS:
757: code = cp - (u_char *)ip;
758: ipt = (struct ip_timestamp *)cp;
759: if (ipt->ipt_len < 5)
760: goto bad;
761: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (long)) {
762: if (++ipt->ipt_oflw == 0)
763: goto bad;
764: break;
765: }
766: sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
767: switch (ipt->ipt_flg) {
768:
769: case IPOPT_TS_TSONLY:
770: break;
771:
772: case IPOPT_TS_TSANDADDR:
773: if (ipt->ipt_ptr + sizeof(n_time) +
774: sizeof(struct in_addr) > ipt->ipt_len)
775: goto bad;
776: ia = ifptoia(m->m_pkthdr.rcvif);
777: bcopy((caddr_t)&IA_SIN(ia)->sin_addr,
778: (caddr_t)sin, sizeof(struct in_addr));
779: ipt->ipt_ptr += sizeof(struct in_addr);
780: break;
781:
782: case IPOPT_TS_PRESPEC:
783: if (ipt->ipt_ptr + sizeof(n_time) +
784: sizeof(struct in_addr) > ipt->ipt_len)
785: goto bad;
786: bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
787: sizeof(struct in_addr));
788: if (ifa_ifwithaddr((SA)&ipaddr) == 0)
789: continue;
790: ipt->ipt_ptr += sizeof(struct in_addr);
791: break;
792:
793: default:
794: goto bad;
795: }
796: ntime = iptime();
797: bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
798: sizeof(n_time));
799: ipt->ipt_ptr += sizeof(n_time);
800: }
801: }
802: if (forward) {
803: ip_forward(m, 1);
804: return (1);
805: } else
806: return (0);
807: bad:
808: icmp_error(m, type, code);
809: return (1);
810: }
811:
812: /*
813: * Given address of next destination (final or next hop),
814: * return internet address info of interface to be used to get there.
815: */
816: struct in_ifaddr *
817: ip_rtaddr(dst)
818: struct in_addr dst;
819: {
820: register struct sockaddr_in *sin;
821:
822: sin = (struct sockaddr_in *) &ipforward_rt.ro_dst;
823:
824: if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) {
825: if (ipforward_rt.ro_rt) {
826: RTFREE(ipforward_rt.ro_rt);
827: ipforward_rt.ro_rt = 0;
828: }
829: sin->sin_family = AF_INET;
830: sin->sin_len = sizeof(*sin);
831: sin->sin_addr = dst;
832:
833: rtalloc(&ipforward_rt);
834: }
835: if (ipforward_rt.ro_rt == 0)
836: return ((struct in_ifaddr *)0);
837: return ((struct in_ifaddr *) ipforward_rt.ro_rt->rt_ifa);
838: }
839:
840: /*
841: * Save incoming source route for use in replies,
842: * to be picked up later by ip_srcroute if the receiver is interested.
843: */
844: save_rte(option, dst)
845: u_char *option;
846: struct in_addr dst;
847: {
848: unsigned olen;
849:
850: olen = option[IPOPT_OLEN];
851: #ifdef DIAGNOSTIC
852: if (ipprintfs)
853: printf("save_rte: olen %d\n", olen);
854: #endif
855: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
856: return;
857: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
858: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
859: ip_srcrt.dst = dst;
860: }
861:
862: /*
863: * Retrieve incoming source route for use in replies,
864: * in the same form used by setsockopt.
865: * The first hop is placed before the options, will be removed later.
866: */
867: struct mbuf *
868: ip_srcroute()
869: {
870: register struct in_addr *p, *q;
871: register struct mbuf *m;
872:
873: if (ip_nhops == 0)
874: return ((struct mbuf *)0);
875: m = m_get(M_DONTWAIT, MT_SOOPTS);
876: if (m == 0)
877: return ((struct mbuf *)0);
878:
879: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
880:
881: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
882: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
883: OPTSIZ;
884: #ifdef DIAGNOSTIC
885: if (ipprintfs)
886: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
887: #endif
888:
889: /*
890: * First save first hop for return route
891: */
892: p = &ip_srcrt.route[ip_nhops - 1];
893: *(mtod(m, struct in_addr *)) = *p--;
894: #ifdef DIAGNOSTIC
895: if (ipprintfs)
896: printf(" hops %lx", ntohl(mtod(m, struct in_addr *)->s_addr));
897: #endif
898:
899: /*
900: * Copy option fields and padding (nop) to mbuf.
901: */
902: ip_srcrt.nop = IPOPT_NOP;
903: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
904: bcopy((caddr_t)&ip_srcrt.nop,
905: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
906: q = (struct in_addr *)(mtod(m, caddr_t) +
907: sizeof(struct in_addr) + OPTSIZ);
908: #undef OPTSIZ
909: /*
910: * Record return path as an IP source route,
911: * reversing the path (pointers are now aligned).
912: */
913: while (p >= ip_srcrt.route) {
914: #ifdef DIAGNOSTIC
915: if (ipprintfs)
916: printf(" %lx", ntohl(q->s_addr));
917: #endif
918: *q++ = *p--;
919: }
920: /*
921: * Last hop goes to final destination.
922: */
923: *q = ip_srcrt.dst;
924: #ifdef DIAGNOSTIC
925: if (ipprintfs)
926: printf(" %lx\n", ntohl(q->s_addr));
927: #endif
928: return (m);
929: }
930:
931: /*
932: * Strip out IP options, at higher
933: * level protocol in the kernel.
934: * Second argument is buffer to which options
935: * will be moved, and return value is their length.
936: * XXX should be deleted; last arg currently ignored.
937: */
938: ip_stripoptions(m, mopt)
939: register struct mbuf *m;
940: struct mbuf *mopt;
941: {
942: register int i;
943: struct ip *ip = mtod(m, struct ip *);
944: register caddr_t opts;
945: int olen;
946:
947: olen = (ip->ip_hl<<2) - sizeof (struct ip);
948: opts = (caddr_t)(ip + 1);
949: i = m->m_len - (sizeof (struct ip) + olen);
950: bcopy(opts + olen, opts, (unsigned)i);
951: m->m_len -= olen;
952: if (m->m_flags & M_PKTHDR)
953: m->m_pkthdr.len -= olen;
954: ip->ip_hl = sizeof(struct ip) >> 2;
955: }
956:
957: u_char inetctlerrmap[PRC_NCMDS] = {
958: 0, 0, 0, 0,
959: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
960: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
961: EMSGSIZE, EHOSTUNREACH, 0, 0,
962: 0, 0, 0, 0,
963: ENOPROTOOPT
964: };
965:
966: /*
967: * Forward a packet. If some error occurs return the sender
968: * an icmp packet. Note we can't always generate a meaningful
969: * icmp message because icmp doesn't have a large enough repertoire
970: * of codes and types.
971: *
972: * If not forwarding, just drop the packet. This could be confusing
973: * if ipforwarding was zero but some routing protocol was advancing
974: * us as a gateway to somewhere. However, we must let the routing
975: * protocol deal with that.
976: *
977: * The srcrt parameter indicates whether the packet is being forwarded
978: * via a source route.
979: */
980: ip_forward(m, srcrt)
981: struct mbuf *m;
982: int srcrt;
983: {
984: register struct ip *ip = mtod(m, struct ip *);
985: register struct sockaddr_in *sin;
986: register struct rtentry *rt;
987: int error, type = 0, code;
988: struct mbuf *mcopy;
989: struct in_addr dest;
990:
991: dest.s_addr = 0;
992: #ifdef DIAGNOSTIC
993: if (ipprintfs)
994: printf("forward: src %x dst %x ttl %x\n", ip->ip_src,
995: ip->ip_dst, ip->ip_ttl);
996: #endif
997: if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
998: ipstat.ips_cantforward++;
999: m_freem(m);
1000: return;
1001: }
1002: HTONS(ip->ip_id);
1003: if (ip->ip_ttl <= IPTTLDEC) {
1004: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest);
1005: return;
1006: }
1007: ip->ip_ttl -= IPTTLDEC;
1008:
1009: sin = (struct sockaddr_in *)&ipforward_rt.ro_dst;
1010: if ((rt = ipforward_rt.ro_rt) == 0 ||
1011: ip->ip_dst.s_addr != sin->sin_addr.s_addr) {
1012: if (ipforward_rt.ro_rt) {
1013: RTFREE(ipforward_rt.ro_rt);
1014: ipforward_rt.ro_rt = 0;
1015: }
1016: sin->sin_family = AF_INET;
1017: sin->sin_len = sizeof(*sin);
1018: sin->sin_addr = ip->ip_dst;
1019:
1020: rtalloc(&ipforward_rt);
1021: if (ipforward_rt.ro_rt == 0) {
1022: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest);
1023: return;
1024: }
1025: rt = ipforward_rt.ro_rt;
1026: }
1027:
1028: /*
1029: * Save at most 64 bytes of the packet in case
1030: * we need to generate an ICMP message to the src.
1031: */
1032: mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64));
1033:
1034: #ifdef GATEWAY
1035: ip_ifmatrix[rt->rt_ifp->if_index +
1036: if_index * m->m_pkthdr.rcvif->if_index]++;
1037: #endif
1038: /*
1039: * If forwarding packet using same interface that it came in on,
1040: * perhaps should send a redirect to sender to shortcut a hop.
1041: * Only send redirect if source is sending directly to us,
1042: * and if packet was not source routed (or has any options).
1043: * Also, don't send redirect if forwarding using a default route
1044: * or a route modified by a redirect.
1045: */
1046: #define satosin(sa) ((struct sockaddr_in *)(sa))
1047: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1048: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1049: satosin(rt_key(rt))->sin_addr.s_addr != 0 &&
1050: ipsendredirects && !srcrt) {
1051: struct in_ifaddr *ia;
1052: u_long src = ntohl(ip->ip_src.s_addr);
1053: u_long dst = ntohl(ip->ip_dst.s_addr);
1054:
1055: if ((ia = ifptoia(m->m_pkthdr.rcvif)) &&
1056: (src & ia->ia_subnetmask) == ia->ia_subnet) {
1057: if (rt->rt_flags & RTF_GATEWAY)
1058: dest = satosin(rt->rt_gateway)->sin_addr;
1059: else
1060: dest = ip->ip_dst;
1061: /*
1062: * If the destination is reached by a route to host,
1063: * is on a subnet of a local net, or is directly
1064: * on the attached net (!), use host redirect.
1065: * (We may be the correct first hop for other subnets.)
1066: */
1067: #define RTA(rt) ((struct in_ifaddr *)(rt->rt_ifa))
1068: type = ICMP_REDIRECT;
1069: if ((rt->rt_flags & RTF_HOST) ||
1070: (rt->rt_flags & RTF_GATEWAY) == 0)
1071: code = ICMP_REDIRECT_HOST;
1072: else if (RTA(rt)->ia_subnetmask != RTA(rt)->ia_netmask &&
1073: (dst & RTA(rt)->ia_netmask) == RTA(rt)->ia_net)
1074: code = ICMP_REDIRECT_HOST;
1075: else
1076: code = ICMP_REDIRECT_NET;
1077: #ifdef DIAGNOSTIC
1078: if (ipprintfs)
1079: printf("redirect (%d) to %x\n", code, dest.s_addr);
1080: #endif
1081: }
1082: }
1083:
1084: error = ip_output(m, (struct mbuf *)0, &ipforward_rt, IP_FORWARDING);
1085: if (error)
1086: ipstat.ips_cantforward++;
1087: else {
1088: ipstat.ips_forward++;
1089: if (type)
1090: ipstat.ips_redirectsent++;
1091: else {
1092: if (mcopy)
1093: m_freem(mcopy);
1094: return;
1095: }
1096: }
1097: if (mcopy == NULL)
1098: return;
1099: switch (error) {
1100:
1101: case 0: /* forwarded, but need redirect */
1102: /* type, code set above */
1103: break;
1104:
1105: case ENETUNREACH: /* shouldn't happen, checked above */
1106: case EHOSTUNREACH:
1107: case ENETDOWN:
1108: case EHOSTDOWN:
1109: default:
1110: type = ICMP_UNREACH;
1111: code = ICMP_UNREACH_HOST;
1112: break;
1113:
1114: case EMSGSIZE:
1115: type = ICMP_UNREACH;
1116: code = ICMP_UNREACH_NEEDFRAG;
1117: ipstat.ips_cantfrag++;
1118: break;
1119:
1120: case ENOBUFS:
1121: type = ICMP_SOURCEQUENCH;
1122: code = 0;
1123: break;
1124: }
1125: icmp_error(mcopy, type, code, dest);
1126: }
CVSweb <webmaster@jp.NetBSD.org>