Annotation of src/sys/netinet/ip_input.c, Revision 1.30
1.30 ! christos 1: /* $NetBSD: ip_input.c,v 1.29 1996/02/26 23:17:06 mrg 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>
1.28 christos 48: #include <sys/proc.h>
49:
50: #include <vm/vm.h>
51: #include <sys/sysctl.h>
1.1 cgd 52:
1.5 mycroft 53: #include <net/if.h>
54: #include <net/route.h>
1.1 cgd 55:
1.5 mycroft 56: #include <netinet/in.h>
57: #include <netinet/in_systm.h>
58: #include <netinet/ip.h>
59: #include <netinet/in_pcb.h>
60: #include <netinet/in_var.h>
61: #include <netinet/ip_var.h>
62: #include <netinet/ip_icmp.h>
1.1 cgd 63:
64: #ifndef IPFORWARDING
65: #ifdef GATEWAY
66: #define IPFORWARDING 1 /* forward IP packets not for us */
67: #else /* GATEWAY */
68: #define IPFORWARDING 0 /* don't forward IP packets not for us */
69: #endif /* GATEWAY */
70: #endif /* IPFORWARDING */
71: #ifndef IPSENDREDIRECTS
72: #define IPSENDREDIRECTS 1
73: #endif
1.26 thorpej 74: #ifndef IPFORWSRCRT
75: #define IPFORWSRCRT 1 /* allow source-routed packets */
76: #endif
1.27 thorpej 77: /*
78: * Note: DIRECTED_BROADCAST is handled this way so that previous
79: * configuration using this option will Just Work.
80: */
81: #ifndef IPDIRECTEDBCAST
82: #ifdef DIRECTED_BROADCAST
83: #define IPDIRECTEDBCAST 1
84: #else
85: #define IPDIRECTEDBCAST 0
86: #endif /* DIRECTED_BROADCAST */
87: #endif /* IPDIRECTEDBCAST */
1.1 cgd 88: int ipforwarding = IPFORWARDING;
89: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 90: int ip_defttl = IPDEFTTL;
1.26 thorpej 91: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 92: int ip_directedbcast = IPDIRECTEDBCAST;
1.1 cgd 93: #ifdef DIAGNOSTIC
94: int ipprintfs = 0;
95: #endif
96:
97: extern struct domain inetdomain;
98: extern struct protosw inetsw[];
99: u_char ip_protox[IPPROTO_MAX];
100: int ipqmaxlen = IFQ_MAXLEN;
1.22 mycroft 101: struct in_ifaddrhead in_ifaddr;
1.13 mycroft 102: struct ifqueue ipintrq;
1.1 cgd 103:
104: /*
105: * We need to save the IP options in case a protocol wants to respond
106: * to an incoming packet over the same route if the packet got here
107: * using IP source routing. This allows connection establishment and
108: * maintenance when the remote end is on a network that is not known
109: * to us.
110: */
111: int ip_nhops = 0;
112: static struct ip_srcrt {
113: struct in_addr dst; /* final destination */
114: char nop; /* one NOP to align */
115: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
116: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
117: } ip_srcrt;
118:
1.13 mycroft 119: static void save_rte __P((u_char *, struct in_addr));
1.1 cgd 120: /*
121: * IP initialization: fill in IP protocol switch table.
122: * All protocols not implemented in kernel go to raw IP protocol handler.
123: */
1.8 mycroft 124: void
1.1 cgd 125: ip_init()
126: {
127: register struct protosw *pr;
128: register int i;
129:
130: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
131: if (pr == 0)
132: panic("ip_init");
133: for (i = 0; i < IPPROTO_MAX; i++)
134: ip_protox[i] = pr - inetsw;
135: for (pr = inetdomain.dom_protosw;
136: pr < inetdomain.dom_protoswNPROTOSW; pr++)
137: if (pr->pr_domain->dom_family == PF_INET &&
138: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
139: ip_protox[pr->pr_protocol] = pr - inetsw;
1.25 cgd 140: LIST_INIT(&ipq);
1.1 cgd 141: ip_id = time.tv_sec & 0xffff;
142: ipintrq.ifq_maxlen = ipqmaxlen;
1.22 mycroft 143: TAILQ_INIT(&in_ifaddr);
1.1 cgd 144: }
145:
146: struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
147: struct route ipforward_rt;
148:
149: /*
150: * Ip input routine. Checksum and byte swap header. If fragmented
151: * try to reassemble. Process options. Pass to next level.
152: */
1.8 mycroft 153: void
1.1 cgd 154: ipintr()
155: {
156: register struct ip *ip;
157: register struct mbuf *m;
158: register struct ipq *fp;
159: register struct in_ifaddr *ia;
1.25 cgd 160: struct ipqent *ipqe;
161: int hlen, mff, s;
1.1 cgd 162:
163: next:
164: /*
165: * Get next datagram off input queue and get IP header
166: * in first mbuf.
167: */
168: s = splimp();
169: IF_DEQUEUE(&ipintrq, m);
170: splx(s);
1.13 mycroft 171: if (m == 0)
1.1 cgd 172: return;
173: #ifdef DIAGNOSTIC
174: if ((m->m_flags & M_PKTHDR) == 0)
175: panic("ipintr no HDR");
176: #endif
177: /*
178: * If no IP addresses have been set yet but the interfaces
179: * are receiving, can't do anything with incoming packets yet.
180: */
1.22 mycroft 181: if (in_ifaddr.tqh_first == 0)
1.1 cgd 182: goto bad;
183: ipstat.ips_total++;
184: if (m->m_len < sizeof (struct ip) &&
185: (m = m_pullup(m, sizeof (struct ip))) == 0) {
186: ipstat.ips_toosmall++;
187: goto next;
188: }
189: ip = mtod(m, struct ip *);
1.13 mycroft 190: if (ip->ip_v != IPVERSION) {
191: ipstat.ips_badvers++;
192: goto bad;
193: }
1.1 cgd 194: hlen = ip->ip_hl << 2;
195: if (hlen < sizeof(struct ip)) { /* minimum header length */
196: ipstat.ips_badhlen++;
197: goto bad;
198: }
199: if (hlen > m->m_len) {
200: if ((m = m_pullup(m, hlen)) == 0) {
201: ipstat.ips_badhlen++;
202: goto next;
203: }
204: ip = mtod(m, struct ip *);
205: }
1.28 christos 206: if ((ip->ip_sum = in_cksum(m, hlen)) != 0) {
1.1 cgd 207: ipstat.ips_badsum++;
208: goto bad;
209: }
210:
211: /*
212: * Convert fields to host representation.
213: */
214: NTOHS(ip->ip_len);
215: if (ip->ip_len < hlen) {
216: ipstat.ips_badlen++;
217: goto bad;
218: }
219: NTOHS(ip->ip_id);
220: NTOHS(ip->ip_off);
221:
222: /*
223: * Check that the amount of data in the buffers
224: * is as at least much as the IP header would have us expect.
225: * Trim mbufs if longer than we expect.
226: * Drop packet if shorter than we expect.
227: */
228: if (m->m_pkthdr.len < ip->ip_len) {
229: ipstat.ips_tooshort++;
230: goto bad;
231: }
232: if (m->m_pkthdr.len > ip->ip_len) {
233: if (m->m_len == m->m_pkthdr.len) {
234: m->m_len = ip->ip_len;
235: m->m_pkthdr.len = ip->ip_len;
236: } else
237: m_adj(m, ip->ip_len - m->m_pkthdr.len);
238: }
239:
240: /*
241: * Process options and, if not destined for us,
242: * ship it on. ip_dooptions returns 1 when an
243: * error was detected (causing an icmp message
244: * to be sent and the original packet to be freed).
245: */
246: ip_nhops = 0; /* for source routed packets */
247: if (hlen > sizeof (struct ip) && ip_dooptions(m))
248: goto next;
249:
250: /*
251: * Check our list of addresses, to see if the packet is for us.
252: */
1.22 mycroft 253: for (ia = in_ifaddr.tqh_first; ia; ia = ia->ia_list.tqe_next) {
1.20 mycroft 254: if (ip->ip_dst.s_addr == ia->ia_addr.sin_addr.s_addr)
1.1 cgd 255: goto ours;
1.27 thorpej 256: if (((ip_directedbcast == 0) || (ip_directedbcast &&
257: ia->ia_ifp == m->m_pkthdr.rcvif)) &&
1.1 cgd 258: (ia->ia_ifp->if_flags & IFF_BROADCAST)) {
1.20 mycroft 259: if (ip->ip_dst.s_addr == ia->ia_broadaddr.sin_addr.s_addr ||
260: ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr ||
261: /*
262: * Look for all-0's host part (old broadcast addr),
263: * either for subnet or net.
264: */
265: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 266: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 267: goto ours;
268: }
269: }
1.18 mycroft 270: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 271: struct in_multi *inm;
272: #ifdef MROUTING
273: extern struct socket *ip_mrouter;
1.10 brezak 274:
275: if (m->m_flags & M_EXT) {
276: if ((m = m_pullup(m, hlen)) == 0) {
277: ipstat.ips_toosmall++;
278: goto next;
279: }
280: ip = mtod(m, struct ip *);
281: }
1.4 hpeyerl 282:
283: if (ip_mrouter) {
284: /*
285: * If we are acting as a multicast router, all
286: * incoming multicast packets are passed to the
287: * kernel-level multicast forwarding function.
288: * The packet is returned (relatively) intact; if
289: * ip_mforward() returns a non-zero value, the packet
290: * must be discarded, else it may be accepted below.
291: *
292: * (The IP ident field is put in the same byte order
293: * as expected when ip_mforward() is called from
294: * ip_output().)
295: */
296: ip->ip_id = htons(ip->ip_id);
1.13 mycroft 297: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
298: ipstat.ips_cantforward++;
1.4 hpeyerl 299: m_freem(m);
300: goto next;
301: }
302: ip->ip_id = ntohs(ip->ip_id);
303:
304: /*
305: * The process-level routing demon needs to receive
306: * all multicast IGMP packets, whether or not this
307: * host belongs to their destination groups.
308: */
309: if (ip->ip_p == IPPROTO_IGMP)
310: goto ours;
1.13 mycroft 311: ipstat.ips_forward++;
1.4 hpeyerl 312: }
313: #endif
314: /*
315: * See if we belong to the destination multicast group on the
316: * arrival interface.
317: */
318: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
319: if (inm == NULL) {
1.13 mycroft 320: ipstat.ips_cantforward++;
1.4 hpeyerl 321: m_freem(m);
322: goto next;
323: }
324: goto ours;
325: }
1.19 mycroft 326: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
327: ip->ip_dst.s_addr == INADDR_ANY)
1.1 cgd 328: goto ours;
329:
330: /*
331: * Not for us; forward if possible and desirable.
332: */
333: if (ipforwarding == 0) {
334: ipstat.ips_cantforward++;
335: m_freem(m);
336: } else
337: ip_forward(m, 0);
338: goto next;
339:
340: ours:
341: /*
342: * If offset or IP_MF are set, must reassemble.
343: * Otherwise, nothing need be done.
344: * (We could look in the reassembly queue to see
345: * if the packet was previously fragmented,
346: * but it's not worth the time; just let them time out.)
347: */
348: if (ip->ip_off &~ IP_DF) {
349: if (m->m_flags & M_EXT) { /* XXX */
350: if ((m = m_pullup(m, sizeof (struct ip))) == 0) {
351: ipstat.ips_toosmall++;
352: goto next;
353: }
354: ip = mtod(m, struct ip *);
355: }
356: /*
357: * Look for queue of fragments
358: * of this datagram.
359: */
1.25 cgd 360: for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next)
1.1 cgd 361: if (ip->ip_id == fp->ipq_id &&
362: ip->ip_src.s_addr == fp->ipq_src.s_addr &&
363: ip->ip_dst.s_addr == fp->ipq_dst.s_addr &&
364: ip->ip_p == fp->ipq_p)
365: goto found;
366: fp = 0;
367: found:
368:
369: /*
370: * Adjust ip_len to not reflect header,
1.25 cgd 371: * set ipqe_mff if more fragments are expected,
1.1 cgd 372: * convert offset of this to bytes.
373: */
374: ip->ip_len -= hlen;
1.25 cgd 375: mff = (ip->ip_off & IP_MF) != 0;
376: if (mff) {
1.16 cgd 377: /*
378: * Make sure that fragments have a data length
379: * that's a non-zero multiple of 8 bytes.
380: */
1.17 cgd 381: if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
1.16 cgd 382: ipstat.ips_badfrags++;
383: goto bad;
384: }
385: }
1.1 cgd 386: ip->ip_off <<= 3;
387:
388: /*
389: * If datagram marked as having more fragments
390: * or if this is not the first fragment,
391: * attempt reassembly; if it succeeds, proceed.
392: */
1.25 cgd 393: if (mff || ip->ip_off) {
1.1 cgd 394: ipstat.ips_fragments++;
1.25 cgd 395: MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent),
396: M_IPQ, M_NOWAIT);
397: if (ipqe == NULL) {
398: ipstat.ips_rcvmemdrop++;
399: goto bad;
400: }
401: ipqe->ipqe_mff = mff;
402: ipqe->ipqe_ip = ip;
403: ip = ip_reass(ipqe, fp);
1.1 cgd 404: if (ip == 0)
405: goto next;
1.13 mycroft 406: ipstat.ips_reassembled++;
1.1 cgd 407: m = dtom(ip);
408: } else
409: if (fp)
410: ip_freef(fp);
411: } else
412: ip->ip_len -= hlen;
413:
414: /*
415: * Switch out to protocol's input routine.
416: */
417: ipstat.ips_delivered++;
418: (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
419: goto next;
420: bad:
421: m_freem(m);
422: goto next;
423: }
424:
425: /*
426: * Take incoming datagram fragment and try to
427: * reassemble it into whole datagram. If a chain for
428: * reassembly of this datagram already exists, then it
429: * is given as fp; otherwise have to make a chain.
430: */
431: struct ip *
1.25 cgd 432: ip_reass(ipqe, fp)
433: register struct ipqent *ipqe;
1.1 cgd 434: register struct ipq *fp;
435: {
1.25 cgd 436: register struct mbuf *m = dtom(ipqe->ipqe_ip);
437: register struct ipqent *nq, *p, *q;
438: struct ip *ip;
1.1 cgd 439: struct mbuf *t;
1.25 cgd 440: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 441: int i, next;
442:
443: /*
444: * Presence of header sizes in mbufs
445: * would confuse code below.
446: */
447: m->m_data += hlen;
448: m->m_len -= hlen;
449:
450: /*
451: * If first fragment to arrive, create a reassembly queue.
452: */
453: if (fp == 0) {
454: if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL)
455: goto dropfrag;
456: fp = mtod(t, struct ipq *);
1.25 cgd 457: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 458: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 459: fp->ipq_p = ipqe->ipqe_ip->ip_p;
460: fp->ipq_id = ipqe->ipqe_ip->ip_id;
461: LIST_INIT(&fp->ipq_fragq);
462: fp->ipq_src = ipqe->ipqe_ip->ip_src;
463: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
464: p = NULL;
1.1 cgd 465: goto insert;
466: }
467:
468: /*
469: * Find a segment which begins after this one does.
470: */
1.25 cgd 471: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
472: p = q, q = q->ipqe_q.le_next)
473: if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off)
1.1 cgd 474: break;
475:
476: /*
477: * If there is a preceding segment, it may provide some of
478: * our data already. If so, drop the data from the incoming
479: * segment. If it provides all of our data, drop us.
480: */
1.25 cgd 481: if (p != NULL) {
482: i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len -
483: ipqe->ipqe_ip->ip_off;
1.1 cgd 484: if (i > 0) {
1.25 cgd 485: if (i >= ipqe->ipqe_ip->ip_len)
1.1 cgd 486: goto dropfrag;
1.25 cgd 487: m_adj(dtom(ipqe->ipqe_ip), i);
488: ipqe->ipqe_ip->ip_off += i;
489: ipqe->ipqe_ip->ip_len -= i;
1.1 cgd 490: }
491: }
492:
493: /*
494: * While we overlap succeeding segments trim them or,
495: * if they are completely covered, dequeue them.
496: */
1.25 cgd 497: for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len >
498: q->ipqe_ip->ip_off; q = nq) {
499: i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) -
500: q->ipqe_ip->ip_off;
501: if (i < q->ipqe_ip->ip_len) {
502: q->ipqe_ip->ip_len -= i;
503: q->ipqe_ip->ip_off += i;
504: m_adj(dtom(q->ipqe_ip), i);
1.1 cgd 505: break;
506: }
1.25 cgd 507: nq = q->ipqe_q.le_next;
508: m_freem(dtom(q->ipqe_ip));
509: LIST_REMOVE(q, ipqe_q);
510: FREE(q, M_IPQ);
1.1 cgd 511: }
512:
513: insert:
514: /*
515: * Stick new segment in its place;
516: * check for complete reassembly.
517: */
1.25 cgd 518: if (p == NULL) {
519: LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
520: } else {
521: LIST_INSERT_AFTER(p, ipqe, ipqe_q);
522: }
1.1 cgd 523: next = 0;
1.25 cgd 524: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
525: p = q, q = q->ipqe_q.le_next) {
526: if (q->ipqe_ip->ip_off != next)
1.1 cgd 527: return (0);
1.25 cgd 528: next += q->ipqe_ip->ip_len;
1.1 cgd 529: }
1.25 cgd 530: if (p->ipqe_mff)
1.1 cgd 531: return (0);
532:
533: /*
534: * Reassembly is complete; concatenate fragments.
535: */
1.25 cgd 536: q = fp->ipq_fragq.lh_first;
537: ip = q->ipqe_ip;
538: m = dtom(q->ipqe_ip);
1.1 cgd 539: t = m->m_next;
540: m->m_next = 0;
541: m_cat(m, t);
1.25 cgd 542: nq = q->ipqe_q.le_next;
543: FREE(q, M_IPQ);
544: for (q = nq; q != NULL; q = nq) {
545: t = dtom(q->ipqe_ip);
546: nq = q->ipqe_q.le_next;
547: FREE(q, M_IPQ);
1.1 cgd 548: m_cat(m, t);
549: }
550:
551: /*
552: * Create header for new ip packet by
553: * modifying header of first packet;
554: * dequeue and discard fragment reassembly header.
555: * Make header visible.
556: */
557: ip->ip_len = next;
1.25 cgd 558: ip->ip_src = fp->ipq_src;
559: ip->ip_dst = fp->ipq_dst;
560: LIST_REMOVE(fp, ipq_q);
1.1 cgd 561: (void) m_free(dtom(fp));
562: m->m_len += (ip->ip_hl << 2);
563: m->m_data -= (ip->ip_hl << 2);
564: /* some debugging cruft by sklower, below, will go away soon */
565: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
566: register int plen = 0;
567: for (t = m; m; m = m->m_next)
568: plen += m->m_len;
569: t->m_pkthdr.len = plen;
570: }
1.25 cgd 571: return (ip);
1.1 cgd 572:
573: dropfrag:
574: ipstat.ips_fragdropped++;
575: m_freem(m);
1.25 cgd 576: FREE(ipqe, M_IPQ);
1.1 cgd 577: return (0);
578: }
579:
580: /*
581: * Free a fragment reassembly header and all
582: * associated datagrams.
583: */
1.8 mycroft 584: void
1.1 cgd 585: ip_freef(fp)
586: struct ipq *fp;
587: {
1.25 cgd 588: register struct ipqent *q, *p;
1.1 cgd 589:
1.25 cgd 590: for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) {
591: p = q->ipqe_q.le_next;
592: m_freem(dtom(q->ipqe_ip));
593: LIST_REMOVE(q, ipqe_q);
594: FREE(q, M_IPQ);
1.1 cgd 595: }
1.25 cgd 596: LIST_REMOVE(fp, ipq_q);
1.1 cgd 597: (void) m_free(dtom(fp));
598: }
599:
600: /*
601: * IP timer processing;
602: * if a timer expires on a reassembly
603: * queue, discard it.
604: */
1.8 mycroft 605: void
1.1 cgd 606: ip_slowtimo()
607: {
1.25 cgd 608: register struct ipq *fp, *nfp;
1.24 mycroft 609: int s = splsoftnet();
1.1 cgd 610:
1.25 cgd 611: for (fp = ipq.lh_first; fp != NULL; fp = nfp) {
612: nfp = fp->ipq_q.le_next;
613: if (--fp->ipq_ttl == 0) {
1.1 cgd 614: ipstat.ips_fragtimeout++;
1.25 cgd 615: ip_freef(fp);
1.1 cgd 616: }
617: }
618: splx(s);
619: }
620:
621: /*
622: * Drain off all datagram fragments.
623: */
1.8 mycroft 624: void
1.1 cgd 625: ip_drain()
626: {
627:
1.25 cgd 628: while (ipq.lh_first != NULL) {
1.1 cgd 629: ipstat.ips_fragdropped++;
1.25 cgd 630: ip_freef(ipq.lh_first);
1.1 cgd 631: }
632: }
633:
634: /*
635: * Do option processing on a datagram,
636: * possibly discarding it if bad options are encountered,
637: * or forwarding it if source-routed.
638: * Returns 1 if packet has been forwarded/freed,
639: * 0 if the packet should be processed further.
640: */
1.8 mycroft 641: int
1.1 cgd 642: ip_dooptions(m)
643: struct mbuf *m;
644: {
645: register struct ip *ip = mtod(m, struct ip *);
646: register u_char *cp;
647: register struct ip_timestamp *ipt;
648: register struct in_ifaddr *ia;
649: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.13 mycroft 650: struct in_addr *sin, dst;
1.1 cgd 651: n_time ntime;
652:
1.13 mycroft 653: dst = ip->ip_dst;
1.1 cgd 654: cp = (u_char *)(ip + 1);
655: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
656: for (; cnt > 0; cnt -= optlen, cp += optlen) {
657: opt = cp[IPOPT_OPTVAL];
658: if (opt == IPOPT_EOL)
659: break;
660: if (opt == IPOPT_NOP)
661: optlen = 1;
662: else {
663: optlen = cp[IPOPT_OLEN];
664: if (optlen <= 0 || optlen > cnt) {
665: code = &cp[IPOPT_OLEN] - (u_char *)ip;
666: goto bad;
667: }
668: }
669: switch (opt) {
670:
671: default:
672: break;
673:
674: /*
675: * Source routing with record.
676: * Find interface with current destination address.
677: * If none on this machine then drop if strictly routed,
678: * or do nothing if loosely routed.
679: * Record interface address and bring up next address
680: * component. If strictly routed make sure next
681: * address is on directly accessible net.
682: */
683: case IPOPT_LSRR:
684: case IPOPT_SSRR:
685: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
686: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
687: goto bad;
688: }
689: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 690: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 691: if (ia == 0) {
692: if (opt == IPOPT_SSRR) {
693: type = ICMP_UNREACH;
694: code = ICMP_UNREACH_SRCFAIL;
695: goto bad;
696: }
697: /*
698: * Loose routing, and not at next destination
699: * yet; nothing to do except forward.
700: */
701: break;
702: }
703: off--; /* 0 origin */
704: if (off > optlen - sizeof(struct in_addr)) {
705: /*
706: * End of source route. Should be for us.
707: */
708: save_rte(cp, ip->ip_src);
709: break;
710: }
711: /*
712: * locate outgoing interface
713: */
714: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
715: sizeof(ipaddr.sin_addr));
716: if (opt == IPOPT_SSRR) {
717: #define INA struct in_ifaddr *
718: #define SA struct sockaddr *
1.29 mrg 719: ia = (INA)ifa_ifwithladdr((SA)&ipaddr);
1.1 cgd 720: } else
721: ia = ip_rtaddr(ipaddr.sin_addr);
722: if (ia == 0) {
723: type = ICMP_UNREACH;
724: code = ICMP_UNREACH_SRCFAIL;
725: goto bad;
726: }
727: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 728: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 729: (caddr_t)(cp + off), sizeof(struct in_addr));
730: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 731: /*
732: * Let ip_intr's mcast routing check handle mcast pkts
733: */
1.18 mycroft 734: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 735: break;
736:
737: case IPOPT_RR:
738: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
739: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
740: goto bad;
741: }
742: /*
743: * If no space remains, ignore.
744: */
745: off--; /* 0 origin */
746: if (off > optlen - sizeof(struct in_addr))
747: break;
748: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
749: sizeof(ipaddr.sin_addr));
750: /*
751: * locate outgoing interface; if we're the destination,
752: * use the incoming interface (should be same).
753: */
754: if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
755: (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
756: type = ICMP_UNREACH;
757: code = ICMP_UNREACH_HOST;
758: goto bad;
759: }
1.20 mycroft 760: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 761: (caddr_t)(cp + off), sizeof(struct in_addr));
762: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
763: break;
764:
765: case IPOPT_TS:
766: code = cp - (u_char *)ip;
767: ipt = (struct ip_timestamp *)cp;
768: if (ipt->ipt_len < 5)
769: goto bad;
1.15 cgd 770: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.1 cgd 771: if (++ipt->ipt_oflw == 0)
772: goto bad;
773: break;
774: }
775: sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
776: switch (ipt->ipt_flg) {
777:
778: case IPOPT_TS_TSONLY:
779: break;
780:
781: case IPOPT_TS_TSANDADDR:
782: if (ipt->ipt_ptr + sizeof(n_time) +
783: sizeof(struct in_addr) > ipt->ipt_len)
784: goto bad;
1.13 mycroft 785: ipaddr.sin_addr = dst;
786: ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
787: m->m_pkthdr.rcvif);
788: if (ia == 0)
789: continue;
1.20 mycroft 790: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 791: (caddr_t)sin, sizeof(struct in_addr));
792: ipt->ipt_ptr += sizeof(struct in_addr);
793: break;
794:
795: case IPOPT_TS_PRESPEC:
796: if (ipt->ipt_ptr + sizeof(n_time) +
797: sizeof(struct in_addr) > ipt->ipt_len)
798: goto bad;
799: bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
800: sizeof(struct in_addr));
801: if (ifa_ifwithaddr((SA)&ipaddr) == 0)
802: continue;
803: ipt->ipt_ptr += sizeof(struct in_addr);
804: break;
805:
806: default:
807: goto bad;
808: }
809: ntime = iptime();
810: bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
811: sizeof(n_time));
812: ipt->ipt_ptr += sizeof(n_time);
813: }
814: }
815: if (forward) {
1.26 thorpej 816: if (ip_forwsrcrt == 0) {
817: type = ICMP_UNREACH;
818: code = ICMP_UNREACH_SRCFAIL;
819: goto bad;
820: }
1.1 cgd 821: ip_forward(m, 1);
822: return (1);
1.13 mycroft 823: }
824: return (0);
1.1 cgd 825: bad:
1.13 mycroft 826: ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */
827: icmp_error(m, type, code, 0, 0);
828: ipstat.ips_badoptions++;
1.1 cgd 829: return (1);
830: }
831:
832: /*
833: * Given address of next destination (final or next hop),
834: * return internet address info of interface to be used to get there.
835: */
836: struct in_ifaddr *
837: ip_rtaddr(dst)
838: struct in_addr dst;
839: {
840: register struct sockaddr_in *sin;
841:
1.19 mycroft 842: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 843:
844: if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) {
845: if (ipforward_rt.ro_rt) {
846: RTFREE(ipforward_rt.ro_rt);
847: ipforward_rt.ro_rt = 0;
848: }
849: sin->sin_family = AF_INET;
850: sin->sin_len = sizeof(*sin);
851: sin->sin_addr = dst;
852:
853: rtalloc(&ipforward_rt);
854: }
855: if (ipforward_rt.ro_rt == 0)
856: return ((struct in_ifaddr *)0);
1.19 mycroft 857: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 858: }
859:
860: /*
861: * Save incoming source route for use in replies,
862: * to be picked up later by ip_srcroute if the receiver is interested.
863: */
1.13 mycroft 864: void
1.1 cgd 865: save_rte(option, dst)
866: u_char *option;
867: struct in_addr dst;
868: {
869: unsigned olen;
870:
871: olen = option[IPOPT_OLEN];
872: #ifdef DIAGNOSTIC
873: if (ipprintfs)
874: printf("save_rte: olen %d\n", olen);
875: #endif
876: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
877: return;
878: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
879: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
880: ip_srcrt.dst = dst;
881: }
882:
883: /*
884: * Retrieve incoming source route for use in replies,
885: * in the same form used by setsockopt.
886: * The first hop is placed before the options, will be removed later.
887: */
888: struct mbuf *
889: ip_srcroute()
890: {
891: register struct in_addr *p, *q;
892: register struct mbuf *m;
893:
894: if (ip_nhops == 0)
895: return ((struct mbuf *)0);
896: m = m_get(M_DONTWAIT, MT_SOOPTS);
897: if (m == 0)
898: return ((struct mbuf *)0);
899:
1.13 mycroft 900: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 901:
902: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
903: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
904: OPTSIZ;
905: #ifdef DIAGNOSTIC
906: if (ipprintfs)
907: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
908: #endif
909:
910: /*
911: * First save first hop for return route
912: */
913: p = &ip_srcrt.route[ip_nhops - 1];
914: *(mtod(m, struct in_addr *)) = *p--;
915: #ifdef DIAGNOSTIC
916: if (ipprintfs)
1.30 ! christos 917: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 918: #endif
919:
920: /*
921: * Copy option fields and padding (nop) to mbuf.
922: */
923: ip_srcrt.nop = IPOPT_NOP;
924: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
925: bcopy((caddr_t)&ip_srcrt.nop,
926: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
927: q = (struct in_addr *)(mtod(m, caddr_t) +
928: sizeof(struct in_addr) + OPTSIZ);
929: #undef OPTSIZ
930: /*
931: * Record return path as an IP source route,
932: * reversing the path (pointers are now aligned).
933: */
934: while (p >= ip_srcrt.route) {
935: #ifdef DIAGNOSTIC
936: if (ipprintfs)
1.30 ! christos 937: printf(" %x", ntohl(q->s_addr));
1.1 cgd 938: #endif
939: *q++ = *p--;
940: }
941: /*
942: * Last hop goes to final destination.
943: */
944: *q = ip_srcrt.dst;
945: #ifdef DIAGNOSTIC
946: if (ipprintfs)
1.30 ! christos 947: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 948: #endif
949: return (m);
950: }
951:
952: /*
953: * Strip out IP options, at higher
954: * level protocol in the kernel.
955: * Second argument is buffer to which options
956: * will be moved, and return value is their length.
957: * XXX should be deleted; last arg currently ignored.
958: */
1.8 mycroft 959: void
1.1 cgd 960: ip_stripoptions(m, mopt)
961: register struct mbuf *m;
962: struct mbuf *mopt;
963: {
964: register int i;
965: struct ip *ip = mtod(m, struct ip *);
966: register caddr_t opts;
967: int olen;
968:
969: olen = (ip->ip_hl<<2) - sizeof (struct ip);
970: opts = (caddr_t)(ip + 1);
971: i = m->m_len - (sizeof (struct ip) + olen);
972: bcopy(opts + olen, opts, (unsigned)i);
973: m->m_len -= olen;
974: if (m->m_flags & M_PKTHDR)
975: m->m_pkthdr.len -= olen;
976: ip->ip_hl = sizeof(struct ip) >> 2;
977: }
978:
1.23 mycroft 979: int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 980: 0, 0, 0, 0,
981: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
982: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
983: EMSGSIZE, EHOSTUNREACH, 0, 0,
984: 0, 0, 0, 0,
985: ENOPROTOOPT
986: };
987:
988: /*
989: * Forward a packet. If some error occurs return the sender
990: * an icmp packet. Note we can't always generate a meaningful
991: * icmp message because icmp doesn't have a large enough repertoire
992: * of codes and types.
993: *
994: * If not forwarding, just drop the packet. This could be confusing
995: * if ipforwarding was zero but some routing protocol was advancing
996: * us as a gateway to somewhere. However, we must let the routing
997: * protocol deal with that.
998: *
999: * The srcrt parameter indicates whether the packet is being forwarded
1000: * via a source route.
1001: */
1.13 mycroft 1002: void
1.1 cgd 1003: ip_forward(m, srcrt)
1004: struct mbuf *m;
1005: int srcrt;
1006: {
1007: register struct ip *ip = mtod(m, struct ip *);
1008: register struct sockaddr_in *sin;
1009: register struct rtentry *rt;
1.28 christos 1010: int error, type = 0, code = 0;
1.1 cgd 1011: struct mbuf *mcopy;
1.13 mycroft 1012: n_long dest;
1013: struct ifnet *destifp;
1.1 cgd 1014:
1.13 mycroft 1015: dest = 0;
1.1 cgd 1016: #ifdef DIAGNOSTIC
1017: if (ipprintfs)
1.28 christos 1018: printf("forward: src %x dst %x ttl %x\n",
1019: ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl);
1.1 cgd 1020: #endif
1021: if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1022: ipstat.ips_cantforward++;
1023: m_freem(m);
1024: return;
1025: }
1026: HTONS(ip->ip_id);
1027: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1028: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1029: return;
1030: }
1031: ip->ip_ttl -= IPTTLDEC;
1032:
1.19 mycroft 1033: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1034: if ((rt = ipforward_rt.ro_rt) == 0 ||
1035: ip->ip_dst.s_addr != sin->sin_addr.s_addr) {
1036: if (ipforward_rt.ro_rt) {
1037: RTFREE(ipforward_rt.ro_rt);
1038: ipforward_rt.ro_rt = 0;
1039: }
1040: sin->sin_family = AF_INET;
1041: sin->sin_len = sizeof(*sin);
1042: sin->sin_addr = ip->ip_dst;
1043:
1044: rtalloc(&ipforward_rt);
1045: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1046: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1047: return;
1048: }
1049: rt = ipforward_rt.ro_rt;
1050: }
1051:
1052: /*
1053: * Save at most 64 bytes of the packet in case
1054: * we need to generate an ICMP message to the src.
1055: */
1056: mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64));
1057:
1058: /*
1059: * If forwarding packet using same interface that it came in on,
1060: * perhaps should send a redirect to sender to shortcut a hop.
1061: * Only send redirect if source is sending directly to us,
1062: * and if packet was not source routed (or has any options).
1063: * Also, don't send redirect if forwarding using a default route
1064: * or a route modified by a redirect.
1065: */
1066: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1067: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1068: satosin(rt_key(rt))->sin_addr.s_addr != 0 &&
1069: ipsendredirects && !srcrt) {
1.19 mycroft 1070: if (rt->rt_ifa &&
1071: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1072: ifatoia(rt->rt_ifa)->ia_subnet) {
1.1 cgd 1073: if (rt->rt_flags & RTF_GATEWAY)
1.13 mycroft 1074: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1.1 cgd 1075: else
1.13 mycroft 1076: dest = ip->ip_dst.s_addr;
1077: /* Router requirements says to only send host redirects */
1.1 cgd 1078: type = ICMP_REDIRECT;
1.13 mycroft 1079: code = ICMP_REDIRECT_HOST;
1.1 cgd 1080: #ifdef DIAGNOSTIC
1081: if (ipprintfs)
1.30 ! christos 1082: printf("redirect (%d) to %x\n", code, (u_int32_t)dest);
1.1 cgd 1083: #endif
1084: }
1085: }
1086:
1.27 thorpej 1087: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1088: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1.1 cgd 1089: if (error)
1090: ipstat.ips_cantforward++;
1091: else {
1092: ipstat.ips_forward++;
1093: if (type)
1094: ipstat.ips_redirectsent++;
1095: else {
1096: if (mcopy)
1097: m_freem(mcopy);
1098: return;
1099: }
1100: }
1101: if (mcopy == NULL)
1102: return;
1.13 mycroft 1103: destifp = NULL;
1104:
1.1 cgd 1105: switch (error) {
1106:
1107: case 0: /* forwarded, but need redirect */
1108: /* type, code set above */
1109: break;
1110:
1111: case ENETUNREACH: /* shouldn't happen, checked above */
1112: case EHOSTUNREACH:
1113: case ENETDOWN:
1114: case EHOSTDOWN:
1115: default:
1116: type = ICMP_UNREACH;
1117: code = ICMP_UNREACH_HOST;
1118: break;
1119:
1120: case EMSGSIZE:
1121: type = ICMP_UNREACH;
1122: code = ICMP_UNREACH_NEEDFRAG;
1.13 mycroft 1123: if (ipforward_rt.ro_rt)
1124: destifp = ipforward_rt.ro_rt->rt_ifp;
1.1 cgd 1125: ipstat.ips_cantfrag++;
1126: break;
1127:
1128: case ENOBUFS:
1129: type = ICMP_SOURCEQUENCH;
1130: code = 0;
1131: break;
1132: }
1.13 mycroft 1133: icmp_error(mcopy, type, code, dest, destifp);
1134: }
1135:
1136: int
1137: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1138: int *name;
1139: u_int namelen;
1140: void *oldp;
1141: size_t *oldlenp;
1142: void *newp;
1143: size_t newlen;
1144: {
1145: /* All sysctl names at this level are terminal. */
1146: if (namelen != 1)
1147: return (ENOTDIR);
1148:
1149: switch (name[0]) {
1150: case IPCTL_FORWARDING:
1151: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1152: case IPCTL_SENDREDIRECTS:
1153: return (sysctl_int(oldp, oldlenp, newp, newlen,
1154: &ipsendredirects));
1155: case IPCTL_DEFTTL:
1156: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1157: #ifdef notyet
1158: case IPCTL_DEFMTU:
1159: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1160: #endif
1.26 thorpej 1161: case IPCTL_FORWSRCRT:
1162: /*
1163: * Don't allow this to change in a secure environment.
1164: */
1165: if (securelevel > 0)
1166: return (EPERM);
1167: return (sysctl_int(oldp, oldlenp, newp, newlen,
1168: &ip_forwsrcrt));
1.27 thorpej 1169: case IPCTL_DIRECTEDBCAST:
1170: return (sysctl_int(oldp, oldlenp, newp, newlen,
1171: &ip_directedbcast));
1.13 mycroft 1172: default:
1173: return (EOPNOTSUPP);
1174: }
1175: /* NOTREACHED */
1.1 cgd 1176: }
CVSweb <webmaster@jp.NetBSD.org>