Annotation of src/sys/netinet/ip_input.c, Revision 1.53.2.4
1.53.2.4! cgd 1: /* $NetBSD: ip_input.c,v 1.53.2.3 1998/10/01 17:57:42 cgd 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:
1.53.2.3 cgd 38: /*-
39: * Copyright (c) 1998 The NetBSD Foundation, Inc.
40: * All rights reserved.
41: *
42: * This code is derived from software contributed to The NetBSD Foundation
43: * by Public Access Networks Corporation ("Panix"). It was developed under
44: * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon.
45: *
46: * Redistribution and use in source and binary forms, with or without
47: * modification, are permitted provided that the following conditions
48: * are met:
49: * 1. Redistributions of source code must retain the above copyright
50: * notice, this list of conditions and the following disclaimer.
51: * 2. Redistributions in binary form must reproduce the above copyright
52: * notice, this list of conditions and the following disclaimer in the
53: * documentation and/or other materials provided with the distribution.
54: * 3. All advertising materials mentioning features or use of this software
55: * must display the following acknowledgement:
56: * This product includes software developed by the NetBSD
57: * Foundation, Inc. and its contributors.
58: * 4. Neither the name of The NetBSD Foundation nor the names of its
59: * contributors may be used to endorse or promote products derived
60: * from this software without specific prior written permission.
61: *
62: * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
63: * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
64: * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
65: * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
66: * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
67: * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
68: * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
69: * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
70: * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
71: * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
72: * POSSIBILITY OF SUCH DAMAGE.
73: */
74:
1.5 mycroft 75: #include <sys/param.h>
76: #include <sys/systm.h>
77: #include <sys/malloc.h>
78: #include <sys/mbuf.h>
79: #include <sys/domain.h>
80: #include <sys/protosw.h>
81: #include <sys/socket.h>
1.44 thorpej 82: #include <sys/socketvar.h>
1.5 mycroft 83: #include <sys/errno.h>
84: #include <sys/time.h>
85: #include <sys/kernel.h>
1.28 christos 86: #include <sys/proc.h>
87:
88: #include <vm/vm.h>
89: #include <sys/sysctl.h>
1.1 cgd 90:
1.5 mycroft 91: #include <net/if.h>
1.44 thorpej 92: #include <net/if_dl.h>
1.5 mycroft 93: #include <net/route.h>
1.45 mrg 94: #include <net/pfil.h>
1.1 cgd 95:
1.5 mycroft 96: #include <netinet/in.h>
97: #include <netinet/in_systm.h>
98: #include <netinet/ip.h>
99: #include <netinet/in_pcb.h>
100: #include <netinet/in_var.h>
101: #include <netinet/ip_var.h>
102: #include <netinet/ip_icmp.h>
1.36 mrg 103:
1.44 thorpej 104: /* XXX should really put this in libkern.h */
105: #define offsetof(type, member) ((size_t)(&((type *)0)->member))
106:
1.1 cgd 107: #ifndef IPFORWARDING
108: #ifdef GATEWAY
109: #define IPFORWARDING 1 /* forward IP packets not for us */
110: #else /* GATEWAY */
111: #define IPFORWARDING 0 /* don't forward IP packets not for us */
112: #endif /* GATEWAY */
113: #endif /* IPFORWARDING */
114: #ifndef IPSENDREDIRECTS
115: #define IPSENDREDIRECTS 1
116: #endif
1.26 thorpej 117: #ifndef IPFORWSRCRT
1.47 cjs 118: #define IPFORWSRCRT 1 /* forward source-routed packets */
119: #endif
120: #ifndef IPALLOWSRCRT
1.48 mrg 121: #define IPALLOWSRCRT 1 /* allow source-routed packets */
1.26 thorpej 122: #endif
1.53 kml 123: #ifndef IPMTUDISC
124: #define IPMTUDISC 0
125: #endif
1.53.2.1 mycroft 126: #ifndef IPMTUDISCTIMEOUT
127: #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */
128: #endif
129:
1.53 kml 130:
1.27 thorpej 131: /*
132: * Note: DIRECTED_BROADCAST is handled this way so that previous
133: * configuration using this option will Just Work.
134: */
135: #ifndef IPDIRECTEDBCAST
136: #ifdef DIRECTED_BROADCAST
137: #define IPDIRECTEDBCAST 1
138: #else
139: #define IPDIRECTEDBCAST 0
140: #endif /* DIRECTED_BROADCAST */
141: #endif /* IPDIRECTEDBCAST */
1.1 cgd 142: int ipforwarding = IPFORWARDING;
143: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 144: int ip_defttl = IPDEFTTL;
1.26 thorpej 145: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 146: int ip_directedbcast = IPDIRECTEDBCAST;
1.47 cjs 147: int ip_allowsrcrt = IPALLOWSRCRT;
1.53 kml 148: int ip_mtudisc = IPMTUDISC;
1.53.2.1 mycroft 149: u_int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
1.1 cgd 150: #ifdef DIAGNOSTIC
151: int ipprintfs = 0;
152: #endif
153:
1.53.2.1 mycroft 154: struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
155:
1.1 cgd 156: extern struct domain inetdomain;
157: extern struct protosw inetsw[];
158: u_char ip_protox[IPPROTO_MAX];
159: int ipqmaxlen = IFQ_MAXLEN;
1.22 mycroft 160: struct in_ifaddrhead in_ifaddr;
1.53.2.3 cgd 161: struct in_ifaddrhashhead *in_ifaddrhashtbl;
1.13 mycroft 162: struct ifqueue ipintrq;
1.1 cgd 163:
164: /*
165: * We need to save the IP options in case a protocol wants to respond
166: * to an incoming packet over the same route if the packet got here
167: * using IP source routing. This allows connection establishment and
168: * maintenance when the remote end is on a network that is not known
169: * to us.
170: */
171: int ip_nhops = 0;
172: static struct ip_srcrt {
173: struct in_addr dst; /* final destination */
174: char nop; /* one NOP to align */
175: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
176: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
177: } ip_srcrt;
178:
1.13 mycroft 179: static void save_rte __P((u_char *, struct in_addr));
1.35 mycroft 180:
1.1 cgd 181: /*
182: * IP initialization: fill in IP protocol switch table.
183: * All protocols not implemented in kernel go to raw IP protocol handler.
184: */
1.8 mycroft 185: void
1.1 cgd 186: ip_init()
187: {
188: register struct protosw *pr;
189: register int i;
190:
191: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
192: if (pr == 0)
193: panic("ip_init");
194: for (i = 0; i < IPPROTO_MAX; i++)
195: ip_protox[i] = pr - inetsw;
196: for (pr = inetdomain.dom_protosw;
197: pr < inetdomain.dom_protoswNPROTOSW; pr++)
198: if (pr->pr_domain->dom_family == PF_INET &&
199: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
200: ip_protox[pr->pr_protocol] = pr - inetsw;
1.25 cgd 201: LIST_INIT(&ipq);
1.1 cgd 202: ip_id = time.tv_sec & 0xffff;
203: ipintrq.ifq_maxlen = ipqmaxlen;
1.22 mycroft 204: TAILQ_INIT(&in_ifaddr);
1.53.2.3 cgd 205: in_ifaddrhashtbl =
206: hashinit(IN_IFADDR_HASH_SIZE, M_IFADDR, &in_ifaddrhash);
1.53.2.1 mycroft 207: if (ip_mtudisc != 0)
208: ip_mtudisc_timeout_q =
209: rt_timer_queue_create(ip_mtudisc_timeout);
1.1 cgd 210: }
211:
212: struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
213: struct route ipforward_rt;
214:
215: /*
216: * Ip input routine. Checksum and byte swap header. If fragmented
217: * try to reassemble. Process options. Pass to next level.
218: */
1.8 mycroft 219: void
1.1 cgd 220: ipintr()
221: {
1.33 mrg 222: register struct ip *ip = NULL;
1.1 cgd 223: register struct mbuf *m;
224: register struct ipq *fp;
225: register struct in_ifaddr *ia;
1.53.2.3 cgd 226: register struct ifaddr *ifa;
1.25 cgd 227: struct ipqent *ipqe;
1.35 mycroft 228: int hlen = 0, mff, len, s;
1.36 mrg 229: #ifdef PFIL_HOOKS
1.33 mrg 230: struct packet_filter_hook *pfh;
231: struct mbuf *m0;
1.43 mrg 232: int rv;
1.36 mrg 233: #endif /* PFIL_HOOKS */
1.1 cgd 234:
235: next:
236: /*
237: * Get next datagram off input queue and get IP header
238: * in first mbuf.
239: */
240: s = splimp();
241: IF_DEQUEUE(&ipintrq, m);
242: splx(s);
1.13 mycroft 243: if (m == 0)
1.1 cgd 244: return;
245: #ifdef DIAGNOSTIC
246: if ((m->m_flags & M_PKTHDR) == 0)
247: panic("ipintr no HDR");
248: #endif
249: /*
250: * If no IP addresses have been set yet but the interfaces
251: * are receiving, can't do anything with incoming packets yet.
252: */
1.22 mycroft 253: if (in_ifaddr.tqh_first == 0)
1.1 cgd 254: goto bad;
255: ipstat.ips_total++;
256: if (m->m_len < sizeof (struct ip) &&
257: (m = m_pullup(m, sizeof (struct ip))) == 0) {
258: ipstat.ips_toosmall++;
259: goto next;
260: }
261: ip = mtod(m, struct ip *);
1.13 mycroft 262: if (ip->ip_v != IPVERSION) {
263: ipstat.ips_badvers++;
264: goto bad;
265: }
1.1 cgd 266: hlen = ip->ip_hl << 2;
267: if (hlen < sizeof(struct ip)) { /* minimum header length */
268: ipstat.ips_badhlen++;
269: goto bad;
270: }
271: if (hlen > m->m_len) {
272: if ((m = m_pullup(m, hlen)) == 0) {
273: ipstat.ips_badhlen++;
274: goto next;
275: }
276: ip = mtod(m, struct ip *);
277: }
1.28 christos 278: if ((ip->ip_sum = in_cksum(m, hlen)) != 0) {
1.1 cgd 279: ipstat.ips_badsum++;
280: goto bad;
281: }
282:
283: /*
284: * Convert fields to host representation.
285: */
286: NTOHS(ip->ip_len);
287: NTOHS(ip->ip_id);
288: NTOHS(ip->ip_off);
1.35 mycroft 289: len = ip->ip_len;
1.1 cgd 290:
291: /*
292: * Check that the amount of data in the buffers
293: * is as at least much as the IP header would have us expect.
294: * Trim mbufs if longer than we expect.
295: * Drop packet if shorter than we expect.
296: */
1.35 mycroft 297: if (m->m_pkthdr.len < len) {
1.1 cgd 298: ipstat.ips_tooshort++;
299: goto bad;
300: }
1.35 mycroft 301: if (m->m_pkthdr.len > len) {
1.1 cgd 302: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 303: m->m_len = len;
304: m->m_pkthdr.len = len;
1.1 cgd 305: } else
1.35 mycroft 306: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 307: }
308:
1.36 mrg 309: #ifdef PFIL_HOOKS
1.33 mrg 310: /*
311: * Run through list of hooks for input packets.
312: */
313: m0 = m;
1.53.2.2 mellon 314: for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next)
1.33 mrg 315: if (pfh->pfil_func) {
1.43 mrg 316: rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0);
317: if (rv)
1.40 veego 318: goto next;
1.53.2.2 mellon 319: m = m0;
320: if (m == NULL)
321: goto next;
322: ip = mtod(m, struct ip *);
1.33 mrg 323: }
1.36 mrg 324: #endif /* PFIL_HOOKS */
1.33 mrg 325:
1.1 cgd 326: /*
327: * Process options and, if not destined for us,
328: * ship it on. ip_dooptions returns 1 when an
329: * error was detected (causing an icmp message
330: * to be sent and the original packet to be freed).
331: */
332: ip_nhops = 0; /* for source routed packets */
333: if (hlen > sizeof (struct ip) && ip_dooptions(m))
334: goto next;
335:
336: /*
337: * Check our list of addresses, to see if the packet is for us.
338: */
1.53.2.3 cgd 339: INADDR_TO_IA(ip->ip_dst, ia);
340: if (ia != NULL) goto ours;
341: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
342: for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
343: ifa != NULL; ifa = ifa->ifa_list.tqe_next) {
344: if (ifa->ifa_addr->sa_family != AF_INET) continue;
345: ia = ifatoia(ifa);
1.35 mycroft 346: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
347: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 348: /*
349: * Look for all-0's host part (old broadcast addr),
350: * either for subnet or net.
351: */
352: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 353: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 354: goto ours;
1.53.2.3 cgd 355: /*
356: * An interface with IP address zero accepts
357: * all packets that arrive on that interface.
358: */
359: if (in_nullhost(ia->ia_addr.sin_addr))
360: goto ours;
1.1 cgd 361: }
362: }
1.18 mycroft 363: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 364: struct in_multi *inm;
365: #ifdef MROUTING
366: extern struct socket *ip_mrouter;
1.10 brezak 367:
368: if (m->m_flags & M_EXT) {
369: if ((m = m_pullup(m, hlen)) == 0) {
370: ipstat.ips_toosmall++;
371: goto next;
372: }
373: ip = mtod(m, struct ip *);
374: }
1.4 hpeyerl 375:
376: if (ip_mrouter) {
377: /*
378: * If we are acting as a multicast router, all
379: * incoming multicast packets are passed to the
380: * kernel-level multicast forwarding function.
381: * The packet is returned (relatively) intact; if
382: * ip_mforward() returns a non-zero value, the packet
383: * must be discarded, else it may be accepted below.
384: *
385: * (The IP ident field is put in the same byte order
386: * as expected when ip_mforward() is called from
387: * ip_output().)
388: */
389: ip->ip_id = htons(ip->ip_id);
1.13 mycroft 390: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
391: ipstat.ips_cantforward++;
1.4 hpeyerl 392: m_freem(m);
393: goto next;
394: }
395: ip->ip_id = ntohs(ip->ip_id);
396:
397: /*
398: * The process-level routing demon needs to receive
399: * all multicast IGMP packets, whether or not this
400: * host belongs to their destination groups.
401: */
402: if (ip->ip_p == IPPROTO_IGMP)
403: goto ours;
1.13 mycroft 404: ipstat.ips_forward++;
1.4 hpeyerl 405: }
406: #endif
407: /*
408: * See if we belong to the destination multicast group on the
409: * arrival interface.
410: */
411: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
412: if (inm == NULL) {
1.13 mycroft 413: ipstat.ips_cantforward++;
1.4 hpeyerl 414: m_freem(m);
415: goto next;
416: }
417: goto ours;
418: }
1.19 mycroft 419: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 420: in_nullhost(ip->ip_dst))
1.1 cgd 421: goto ours;
422:
423: /*
424: * Not for us; forward if possible and desirable.
425: */
426: if (ipforwarding == 0) {
427: ipstat.ips_cantforward++;
428: m_freem(m);
429: } else
430: ip_forward(m, 0);
431: goto next;
432:
433: ours:
434: /*
435: * If offset or IP_MF are set, must reassemble.
436: * Otherwise, nothing need be done.
437: * (We could look in the reassembly queue to see
438: * if the packet was previously fragmented,
439: * but it's not worth the time; just let them time out.)
440: */
1.37 perry 441: if (ip->ip_off & ~(IP_DF|IP_RF)) {
1.1 cgd 442: /*
443: * Look for queue of fragments
444: * of this datagram.
445: */
1.25 cgd 446: for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next)
1.1 cgd 447: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 448: in_hosteq(ip->ip_src, fp->ipq_src) &&
449: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 450: ip->ip_p == fp->ipq_p)
451: goto found;
452: fp = 0;
453: found:
454:
455: /*
456: * Adjust ip_len to not reflect header,
1.25 cgd 457: * set ipqe_mff if more fragments are expected,
1.1 cgd 458: * convert offset of this to bytes.
459: */
460: ip->ip_len -= hlen;
1.25 cgd 461: mff = (ip->ip_off & IP_MF) != 0;
462: if (mff) {
1.16 cgd 463: /*
464: * Make sure that fragments have a data length
465: * that's a non-zero multiple of 8 bytes.
466: */
1.17 cgd 467: if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
1.16 cgd 468: ipstat.ips_badfrags++;
469: goto bad;
470: }
471: }
1.1 cgd 472: ip->ip_off <<= 3;
473:
474: /*
475: * If datagram marked as having more fragments
476: * or if this is not the first fragment,
477: * attempt reassembly; if it succeeds, proceed.
478: */
1.25 cgd 479: if (mff || ip->ip_off) {
1.1 cgd 480: ipstat.ips_fragments++;
1.25 cgd 481: MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent),
482: M_IPQ, M_NOWAIT);
483: if (ipqe == NULL) {
484: ipstat.ips_rcvmemdrop++;
485: goto bad;
486: }
487: ipqe->ipqe_mff = mff;
1.50 thorpej 488: ipqe->ipqe_m = m;
1.25 cgd 489: ipqe->ipqe_ip = ip;
1.50 thorpej 490: m = ip_reass(ipqe, fp);
491: if (m == 0)
1.1 cgd 492: goto next;
1.13 mycroft 493: ipstat.ips_reassembled++;
1.50 thorpej 494: ip = mtod(m, struct ip *);
1.53.2.4! cgd 495: hlen = ip->ip_hl << 2;
1.1 cgd 496: } else
497: if (fp)
498: ip_freef(fp);
499: } else
500: ip->ip_len -= hlen;
501:
502: /*
503: * Switch out to protocol's input routine.
504: */
505: ipstat.ips_delivered++;
506: (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
507: goto next;
508: bad:
509: m_freem(m);
510: goto next;
511: }
512:
513: /*
514: * Take incoming datagram fragment and try to
515: * reassemble it into whole datagram. If a chain for
516: * reassembly of this datagram already exists, then it
517: * is given as fp; otherwise have to make a chain.
518: */
1.50 thorpej 519: struct mbuf *
1.25 cgd 520: ip_reass(ipqe, fp)
521: register struct ipqent *ipqe;
1.1 cgd 522: register struct ipq *fp;
523: {
1.50 thorpej 524: register struct mbuf *m = ipqe->ipqe_m;
1.25 cgd 525: register struct ipqent *nq, *p, *q;
526: struct ip *ip;
1.1 cgd 527: struct mbuf *t;
1.25 cgd 528: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 529: int i, next;
530:
531: /*
532: * Presence of header sizes in mbufs
533: * would confuse code below.
534: */
535: m->m_data += hlen;
536: m->m_len -= hlen;
537:
538: /*
539: * If first fragment to arrive, create a reassembly queue.
540: */
541: if (fp == 0) {
1.50 thorpej 542: MALLOC(fp, struct ipq *, sizeof (struct ipq),
543: M_FTABLE, M_NOWAIT);
544: if (fp == NULL)
1.1 cgd 545: goto dropfrag;
1.25 cgd 546: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 547: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 548: fp->ipq_p = ipqe->ipqe_ip->ip_p;
549: fp->ipq_id = ipqe->ipqe_ip->ip_id;
550: LIST_INIT(&fp->ipq_fragq);
551: fp->ipq_src = ipqe->ipqe_ip->ip_src;
552: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
553: p = NULL;
1.1 cgd 554: goto insert;
555: }
556:
557: /*
558: * Find a segment which begins after this one does.
559: */
1.25 cgd 560: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
561: p = q, q = q->ipqe_q.le_next)
562: if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off)
1.1 cgd 563: break;
564:
565: /*
566: * If there is a preceding segment, it may provide some of
567: * our data already. If so, drop the data from the incoming
568: * segment. If it provides all of our data, drop us.
569: */
1.25 cgd 570: if (p != NULL) {
571: i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len -
572: ipqe->ipqe_ip->ip_off;
1.1 cgd 573: if (i > 0) {
1.25 cgd 574: if (i >= ipqe->ipqe_ip->ip_len)
1.1 cgd 575: goto dropfrag;
1.50 thorpej 576: m_adj(ipqe->ipqe_m, i);
1.25 cgd 577: ipqe->ipqe_ip->ip_off += i;
578: ipqe->ipqe_ip->ip_len -= i;
1.1 cgd 579: }
580: }
581:
582: /*
583: * While we overlap succeeding segments trim them or,
584: * if they are completely covered, dequeue them.
585: */
1.25 cgd 586: for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len >
587: q->ipqe_ip->ip_off; q = nq) {
588: i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) -
589: q->ipqe_ip->ip_off;
590: if (i < q->ipqe_ip->ip_len) {
591: q->ipqe_ip->ip_len -= i;
592: q->ipqe_ip->ip_off += i;
1.50 thorpej 593: m_adj(q->ipqe_m, i);
1.1 cgd 594: break;
595: }
1.25 cgd 596: nq = q->ipqe_q.le_next;
1.50 thorpej 597: m_freem(q->ipqe_m);
1.25 cgd 598: LIST_REMOVE(q, ipqe_q);
599: FREE(q, M_IPQ);
1.1 cgd 600: }
601:
602: insert:
603: /*
604: * Stick new segment in its place;
605: * check for complete reassembly.
606: */
1.25 cgd 607: if (p == NULL) {
608: LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
609: } else {
610: LIST_INSERT_AFTER(p, ipqe, ipqe_q);
611: }
1.1 cgd 612: next = 0;
1.25 cgd 613: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
614: p = q, q = q->ipqe_q.le_next) {
615: if (q->ipqe_ip->ip_off != next)
1.1 cgd 616: return (0);
1.25 cgd 617: next += q->ipqe_ip->ip_len;
1.1 cgd 618: }
1.25 cgd 619: if (p->ipqe_mff)
1.1 cgd 620: return (0);
621:
622: /*
1.41 thorpej 623: * Reassembly is complete. Check for a bogus message size and
624: * concatenate fragments.
1.1 cgd 625: */
1.25 cgd 626: q = fp->ipq_fragq.lh_first;
627: ip = q->ipqe_ip;
1.41 thorpej 628: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
629: ipstat.ips_toolong++;
630: ip_freef(fp);
631: return (0);
632: }
1.50 thorpej 633: m = q->ipqe_m;
1.1 cgd 634: t = m->m_next;
635: m->m_next = 0;
636: m_cat(m, t);
1.25 cgd 637: nq = q->ipqe_q.le_next;
638: FREE(q, M_IPQ);
639: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 640: t = q->ipqe_m;
1.25 cgd 641: nq = q->ipqe_q.le_next;
642: FREE(q, M_IPQ);
1.1 cgd 643: m_cat(m, t);
644: }
645:
646: /*
647: * Create header for new ip packet by
648: * modifying header of first packet;
649: * dequeue and discard fragment reassembly header.
650: * Make header visible.
651: */
652: ip->ip_len = next;
1.25 cgd 653: ip->ip_src = fp->ipq_src;
654: ip->ip_dst = fp->ipq_dst;
655: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 656: FREE(fp, M_FTABLE);
1.1 cgd 657: m->m_len += (ip->ip_hl << 2);
658: m->m_data -= (ip->ip_hl << 2);
659: /* some debugging cruft by sklower, below, will go away soon */
660: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
661: register int plen = 0;
1.50 thorpej 662: for (t = m; t; t = t->m_next)
663: plen += t->m_len;
664: m->m_pkthdr.len = plen;
1.1 cgd 665: }
1.50 thorpej 666: return (m);
1.1 cgd 667:
668: dropfrag:
669: ipstat.ips_fragdropped++;
670: m_freem(m);
1.25 cgd 671: FREE(ipqe, M_IPQ);
1.1 cgd 672: return (0);
673: }
674:
675: /*
676: * Free a fragment reassembly header and all
677: * associated datagrams.
678: */
1.8 mycroft 679: void
1.1 cgd 680: ip_freef(fp)
681: struct ipq *fp;
682: {
1.25 cgd 683: register struct ipqent *q, *p;
1.1 cgd 684:
1.25 cgd 685: for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) {
686: p = q->ipqe_q.le_next;
1.50 thorpej 687: m_freem(q->ipqe_m);
1.25 cgd 688: LIST_REMOVE(q, ipqe_q);
689: FREE(q, M_IPQ);
1.1 cgd 690: }
1.25 cgd 691: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 692: FREE(fp, M_FTABLE);
1.1 cgd 693: }
694:
695: /*
696: * IP timer processing;
697: * if a timer expires on a reassembly
698: * queue, discard it.
699: */
1.8 mycroft 700: void
1.1 cgd 701: ip_slowtimo()
702: {
1.25 cgd 703: register struct ipq *fp, *nfp;
1.24 mycroft 704: int s = splsoftnet();
1.1 cgd 705:
1.25 cgd 706: for (fp = ipq.lh_first; fp != NULL; fp = nfp) {
707: nfp = fp->ipq_q.le_next;
708: if (--fp->ipq_ttl == 0) {
1.1 cgd 709: ipstat.ips_fragtimeout++;
1.25 cgd 710: ip_freef(fp);
1.1 cgd 711: }
712: }
713: splx(s);
714: }
715:
716: /*
717: * Drain off all datagram fragments.
718: */
1.8 mycroft 719: void
1.1 cgd 720: ip_drain()
721: {
722:
1.25 cgd 723: while (ipq.lh_first != NULL) {
1.1 cgd 724: ipstat.ips_fragdropped++;
1.25 cgd 725: ip_freef(ipq.lh_first);
1.1 cgd 726: }
727: }
728:
729: /*
730: * Do option processing on a datagram,
731: * possibly discarding it if bad options are encountered,
732: * or forwarding it if source-routed.
733: * Returns 1 if packet has been forwarded/freed,
734: * 0 if the packet should be processed further.
735: */
1.8 mycroft 736: int
1.1 cgd 737: ip_dooptions(m)
738: struct mbuf *m;
739: {
740: register struct ip *ip = mtod(m, struct ip *);
741: register u_char *cp;
742: register struct ip_timestamp *ipt;
743: register struct in_ifaddr *ia;
744: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.13 mycroft 745: struct in_addr *sin, dst;
1.1 cgd 746: n_time ntime;
747:
1.13 mycroft 748: dst = ip->ip_dst;
1.1 cgd 749: cp = (u_char *)(ip + 1);
750: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
751: for (; cnt > 0; cnt -= optlen, cp += optlen) {
752: opt = cp[IPOPT_OPTVAL];
753: if (opt == IPOPT_EOL)
754: break;
755: if (opt == IPOPT_NOP)
756: optlen = 1;
757: else {
758: optlen = cp[IPOPT_OLEN];
759: if (optlen <= 0 || optlen > cnt) {
760: code = &cp[IPOPT_OLEN] - (u_char *)ip;
761: goto bad;
762: }
763: }
764: switch (opt) {
765:
766: default:
767: break;
768:
769: /*
770: * Source routing with record.
771: * Find interface with current destination address.
772: * If none on this machine then drop if strictly routed,
773: * or do nothing if loosely routed.
774: * Record interface address and bring up next address
775: * component. If strictly routed make sure next
776: * address is on directly accessible net.
777: */
778: case IPOPT_LSRR:
779: case IPOPT_SSRR:
1.47 cjs 780: if (ip_allowsrcrt == 0) {
781: type = ICMP_UNREACH;
782: code = ICMP_UNREACH_NET_PROHIB;
783: goto bad;
784: }
1.1 cgd 785: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
786: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
787: goto bad;
788: }
789: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 790: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 791: if (ia == 0) {
792: if (opt == IPOPT_SSRR) {
793: type = ICMP_UNREACH;
794: code = ICMP_UNREACH_SRCFAIL;
795: goto bad;
796: }
797: /*
798: * Loose routing, and not at next destination
799: * yet; nothing to do except forward.
800: */
801: break;
802: }
803: off--; /* 0 origin */
804: if (off > optlen - sizeof(struct in_addr)) {
805: /*
806: * End of source route. Should be for us.
807: */
808: save_rte(cp, ip->ip_src);
809: break;
810: }
811: /*
812: * locate outgoing interface
813: */
814: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
815: sizeof(ipaddr.sin_addr));
816: if (opt == IPOPT_SSRR) {
817: #define INA struct in_ifaddr *
818: #define SA struct sockaddr *
1.29 mrg 819: ia = (INA)ifa_ifwithladdr((SA)&ipaddr);
1.1 cgd 820: } else
821: ia = ip_rtaddr(ipaddr.sin_addr);
822: if (ia == 0) {
823: type = ICMP_UNREACH;
824: code = ICMP_UNREACH_SRCFAIL;
825: goto bad;
826: }
827: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 828: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 829: (caddr_t)(cp + off), sizeof(struct in_addr));
830: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 831: /*
832: * Let ip_intr's mcast routing check handle mcast pkts
833: */
1.18 mycroft 834: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 835: break;
836:
837: case IPOPT_RR:
838: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
839: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
840: goto bad;
841: }
842: /*
843: * If no space remains, ignore.
844: */
845: off--; /* 0 origin */
846: if (off > optlen - sizeof(struct in_addr))
847: break;
848: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
849: sizeof(ipaddr.sin_addr));
850: /*
851: * locate outgoing interface; if we're the destination,
852: * use the incoming interface (should be same).
853: */
854: if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
855: (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
856: type = ICMP_UNREACH;
857: code = ICMP_UNREACH_HOST;
858: goto bad;
859: }
1.20 mycroft 860: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 861: (caddr_t)(cp + off), sizeof(struct in_addr));
862: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
863: break;
864:
865: case IPOPT_TS:
866: code = cp - (u_char *)ip;
867: ipt = (struct ip_timestamp *)cp;
868: if (ipt->ipt_len < 5)
869: goto bad;
1.15 cgd 870: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.1 cgd 871: if (++ipt->ipt_oflw == 0)
872: goto bad;
873: break;
874: }
875: sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
876: switch (ipt->ipt_flg) {
877:
878: case IPOPT_TS_TSONLY:
879: break;
880:
881: case IPOPT_TS_TSANDADDR:
882: if (ipt->ipt_ptr + sizeof(n_time) +
883: sizeof(struct in_addr) > ipt->ipt_len)
884: goto bad;
1.13 mycroft 885: ipaddr.sin_addr = dst;
886: ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
887: m->m_pkthdr.rcvif);
888: if (ia == 0)
889: continue;
1.20 mycroft 890: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 891: (caddr_t)sin, sizeof(struct in_addr));
892: ipt->ipt_ptr += sizeof(struct in_addr);
893: break;
894:
895: case IPOPT_TS_PRESPEC:
896: if (ipt->ipt_ptr + sizeof(n_time) +
897: sizeof(struct in_addr) > ipt->ipt_len)
898: goto bad;
899: bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
900: sizeof(struct in_addr));
901: if (ifa_ifwithaddr((SA)&ipaddr) == 0)
902: continue;
903: ipt->ipt_ptr += sizeof(struct in_addr);
904: break;
905:
906: default:
907: goto bad;
908: }
909: ntime = iptime();
910: bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
911: sizeof(n_time));
912: ipt->ipt_ptr += sizeof(n_time);
913: }
914: }
915: if (forward) {
1.26 thorpej 916: if (ip_forwsrcrt == 0) {
917: type = ICMP_UNREACH;
918: code = ICMP_UNREACH_SRCFAIL;
919: goto bad;
920: }
1.1 cgd 921: ip_forward(m, 1);
922: return (1);
1.13 mycroft 923: }
924: return (0);
1.1 cgd 925: bad:
1.13 mycroft 926: ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */
927: icmp_error(m, type, code, 0, 0);
928: ipstat.ips_badoptions++;
1.1 cgd 929: return (1);
930: }
931:
932: /*
933: * Given address of next destination (final or next hop),
934: * return internet address info of interface to be used to get there.
935: */
936: struct in_ifaddr *
937: ip_rtaddr(dst)
938: struct in_addr dst;
939: {
940: register struct sockaddr_in *sin;
941:
1.19 mycroft 942: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 943:
1.35 mycroft 944: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 945: if (ipforward_rt.ro_rt) {
946: RTFREE(ipforward_rt.ro_rt);
947: ipforward_rt.ro_rt = 0;
948: }
949: sin->sin_family = AF_INET;
950: sin->sin_len = sizeof(*sin);
951: sin->sin_addr = dst;
952:
953: rtalloc(&ipforward_rt);
954: }
955: if (ipforward_rt.ro_rt == 0)
956: return ((struct in_ifaddr *)0);
1.19 mycroft 957: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 958: }
959:
960: /*
961: * Save incoming source route for use in replies,
962: * to be picked up later by ip_srcroute if the receiver is interested.
963: */
1.13 mycroft 964: void
1.1 cgd 965: save_rte(option, dst)
966: u_char *option;
967: struct in_addr dst;
968: {
969: unsigned olen;
970:
971: olen = option[IPOPT_OLEN];
972: #ifdef DIAGNOSTIC
973: if (ipprintfs)
1.39 christos 974: printf("save_rte: olen %d\n", olen);
1.1 cgd 975: #endif
976: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
977: return;
978: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
979: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
980: ip_srcrt.dst = dst;
981: }
982:
983: /*
984: * Retrieve incoming source route for use in replies,
985: * in the same form used by setsockopt.
986: * The first hop is placed before the options, will be removed later.
987: */
988: struct mbuf *
989: ip_srcroute()
990: {
991: register struct in_addr *p, *q;
992: register struct mbuf *m;
993:
994: if (ip_nhops == 0)
995: return ((struct mbuf *)0);
996: m = m_get(M_DONTWAIT, MT_SOOPTS);
997: if (m == 0)
998: return ((struct mbuf *)0);
999:
1.13 mycroft 1000: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1001:
1002: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1003: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1004: OPTSIZ;
1005: #ifdef DIAGNOSTIC
1006: if (ipprintfs)
1.39 christos 1007: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1008: #endif
1009:
1010: /*
1011: * First save first hop for return route
1012: */
1013: p = &ip_srcrt.route[ip_nhops - 1];
1014: *(mtod(m, struct in_addr *)) = *p--;
1015: #ifdef DIAGNOSTIC
1016: if (ipprintfs)
1.39 christos 1017: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1018: #endif
1019:
1020: /*
1021: * Copy option fields and padding (nop) to mbuf.
1022: */
1023: ip_srcrt.nop = IPOPT_NOP;
1024: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1025: bcopy((caddr_t)&ip_srcrt.nop,
1026: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1027: q = (struct in_addr *)(mtod(m, caddr_t) +
1028: sizeof(struct in_addr) + OPTSIZ);
1029: #undef OPTSIZ
1030: /*
1031: * Record return path as an IP source route,
1032: * reversing the path (pointers are now aligned).
1033: */
1034: while (p >= ip_srcrt.route) {
1035: #ifdef DIAGNOSTIC
1036: if (ipprintfs)
1.39 christos 1037: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1038: #endif
1039: *q++ = *p--;
1040: }
1041: /*
1042: * Last hop goes to final destination.
1043: */
1044: *q = ip_srcrt.dst;
1045: #ifdef DIAGNOSTIC
1046: if (ipprintfs)
1.39 christos 1047: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1048: #endif
1049: return (m);
1050: }
1051:
1052: /*
1053: * Strip out IP options, at higher
1054: * level protocol in the kernel.
1055: * Second argument is buffer to which options
1056: * will be moved, and return value is their length.
1057: * XXX should be deleted; last arg currently ignored.
1058: */
1.8 mycroft 1059: void
1.1 cgd 1060: ip_stripoptions(m, mopt)
1061: register struct mbuf *m;
1062: struct mbuf *mopt;
1063: {
1064: register int i;
1065: struct ip *ip = mtod(m, struct ip *);
1066: register caddr_t opts;
1067: int olen;
1068:
1069: olen = (ip->ip_hl<<2) - sizeof (struct ip);
1070: opts = (caddr_t)(ip + 1);
1071: i = m->m_len - (sizeof (struct ip) + olen);
1072: bcopy(opts + olen, opts, (unsigned)i);
1073: m->m_len -= olen;
1074: if (m->m_flags & M_PKTHDR)
1075: m->m_pkthdr.len -= olen;
1076: ip->ip_hl = sizeof(struct ip) >> 2;
1077: }
1078:
1.23 mycroft 1079: int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1080: 0, 0, 0, 0,
1081: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1082: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1083: EMSGSIZE, EHOSTUNREACH, 0, 0,
1084: 0, 0, 0, 0,
1085: ENOPROTOOPT
1086: };
1087:
1088: /*
1089: * Forward a packet. If some error occurs return the sender
1090: * an icmp packet. Note we can't always generate a meaningful
1091: * icmp message because icmp doesn't have a large enough repertoire
1092: * of codes and types.
1093: *
1094: * If not forwarding, just drop the packet. This could be confusing
1095: * if ipforwarding was zero but some routing protocol was advancing
1096: * us as a gateway to somewhere. However, we must let the routing
1097: * protocol deal with that.
1098: *
1099: * The srcrt parameter indicates whether the packet is being forwarded
1100: * via a source route.
1101: */
1.13 mycroft 1102: void
1.1 cgd 1103: ip_forward(m, srcrt)
1104: struct mbuf *m;
1105: int srcrt;
1106: {
1107: register struct ip *ip = mtod(m, struct ip *);
1108: register struct sockaddr_in *sin;
1109: register struct rtentry *rt;
1.28 christos 1110: int error, type = 0, code = 0;
1.1 cgd 1111: struct mbuf *mcopy;
1.13 mycroft 1112: n_long dest;
1113: struct ifnet *destifp;
1.1 cgd 1114:
1.13 mycroft 1115: dest = 0;
1.1 cgd 1116: #ifdef DIAGNOSTIC
1117: if (ipprintfs)
1.39 christos 1118: printf("forward: src %x dst %x ttl %x\n",
1.35 mycroft 1119: ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl);
1.1 cgd 1120: #endif
1121: if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1122: ipstat.ips_cantforward++;
1123: m_freem(m);
1124: return;
1125: }
1126: HTONS(ip->ip_id);
1127: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1128: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1129: return;
1130: }
1131: ip->ip_ttl -= IPTTLDEC;
1132:
1.19 mycroft 1133: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1134: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1135: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1136: if (ipforward_rt.ro_rt) {
1137: RTFREE(ipforward_rt.ro_rt);
1138: ipforward_rt.ro_rt = 0;
1139: }
1140: sin->sin_family = AF_INET;
1.35 mycroft 1141: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1142: sin->sin_addr = ip->ip_dst;
1143:
1144: rtalloc(&ipforward_rt);
1145: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1146: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1147: return;
1148: }
1149: rt = ipforward_rt.ro_rt;
1150: }
1151:
1152: /*
1.34 mycroft 1153: * Save at most 68 bytes of the packet in case
1.1 cgd 1154: * we need to generate an ICMP message to the src.
1155: */
1.34 mycroft 1156: mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68));
1.1 cgd 1157:
1158: /*
1159: * If forwarding packet using same interface that it came in on,
1160: * perhaps should send a redirect to sender to shortcut a hop.
1161: * Only send redirect if source is sending directly to us,
1162: * and if packet was not source routed (or has any options).
1163: * Also, don't send redirect if forwarding using a default route
1164: * or a route modified by a redirect.
1165: */
1166: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1167: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1168: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1169: ipsendredirects && !srcrt) {
1.19 mycroft 1170: if (rt->rt_ifa &&
1171: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1172: ifatoia(rt->rt_ifa)->ia_subnet) {
1.1 cgd 1173: if (rt->rt_flags & RTF_GATEWAY)
1.13 mycroft 1174: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1.1 cgd 1175: else
1.13 mycroft 1176: dest = ip->ip_dst.s_addr;
1177: /* Router requirements says to only send host redirects */
1.1 cgd 1178: type = ICMP_REDIRECT;
1.13 mycroft 1179: code = ICMP_REDIRECT_HOST;
1.1 cgd 1180: #ifdef DIAGNOSTIC
1181: if (ipprintfs)
1.39 christos 1182: printf("redirect (%d) to %x\n", code, (u_int32_t)dest);
1.1 cgd 1183: #endif
1184: }
1185: }
1186:
1.27 thorpej 1187: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1188: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1.1 cgd 1189: if (error)
1190: ipstat.ips_cantforward++;
1191: else {
1192: ipstat.ips_forward++;
1193: if (type)
1194: ipstat.ips_redirectsent++;
1195: else {
1196: if (mcopy)
1197: m_freem(mcopy);
1198: return;
1199: }
1200: }
1201: if (mcopy == NULL)
1202: return;
1.13 mycroft 1203: destifp = NULL;
1204:
1.1 cgd 1205: switch (error) {
1206:
1207: case 0: /* forwarded, but need redirect */
1208: /* type, code set above */
1209: break;
1210:
1211: case ENETUNREACH: /* shouldn't happen, checked above */
1212: case EHOSTUNREACH:
1213: case ENETDOWN:
1214: case EHOSTDOWN:
1215: default:
1216: type = ICMP_UNREACH;
1217: code = ICMP_UNREACH_HOST;
1218: break;
1219:
1220: case EMSGSIZE:
1221: type = ICMP_UNREACH;
1222: code = ICMP_UNREACH_NEEDFRAG;
1.13 mycroft 1223: if (ipforward_rt.ro_rt)
1224: destifp = ipforward_rt.ro_rt->rt_ifp;
1.1 cgd 1225: ipstat.ips_cantfrag++;
1226: break;
1227:
1228: case ENOBUFS:
1229: type = ICMP_SOURCEQUENCH;
1230: code = 0;
1231: break;
1232: }
1.13 mycroft 1233: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 1234: }
1235:
1236: void
1237: ip_savecontrol(inp, mp, ip, m)
1238: register struct inpcb *inp;
1239: register struct mbuf **mp;
1240: register struct ip *ip;
1241: register struct mbuf *m;
1242: {
1243:
1244: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1245: struct timeval tv;
1246:
1247: microtime(&tv);
1248: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1249: SCM_TIMESTAMP, SOL_SOCKET);
1250: if (*mp)
1251: mp = &(*mp)->m_next;
1252: }
1253: if (inp->inp_flags & INP_RECVDSTADDR) {
1254: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1255: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1256: if (*mp)
1257: mp = &(*mp)->m_next;
1258: }
1259: #ifdef notyet
1260: /*
1261: * XXX
1262: * Moving these out of udp_input() made them even more broken
1263: * than they already were.
1264: * - fenner@parc.xerox.com
1265: */
1266: /* options were tossed already */
1267: if (inp->inp_flags & INP_RECVOPTS) {
1268: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1269: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1270: if (*mp)
1271: mp = &(*mp)->m_next;
1272: }
1273: /* ip_srcroute doesn't do what we want here, need to fix */
1274: if (inp->inp_flags & INP_RECVRETOPTS) {
1275: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1276: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1277: if (*mp)
1278: mp = &(*mp)->m_next;
1279: }
1280: #endif
1281: if (inp->inp_flags & INP_RECVIF) {
1282: struct sockaddr_dl sdl;
1283:
1284: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1285: sdl.sdl_family = AF_LINK;
1286: sdl.sdl_index = m->m_pkthdr.rcvif ?
1287: m->m_pkthdr.rcvif->if_index : 0;
1288: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1289: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1290: IP_RECVIF, IPPROTO_IP);
1291: if (*mp)
1292: mp = &(*mp)->m_next;
1293: }
1.13 mycroft 1294: }
1295:
1296: int
1297: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1298: int *name;
1299: u_int namelen;
1300: void *oldp;
1301: size_t *oldlenp;
1302: void *newp;
1303: size_t newlen;
1304: {
1.52 thorpej 1305: extern int subnetsarelocal;
1.53.2.1 mycroft 1306: int error;
1.52 thorpej 1307:
1.13 mycroft 1308: /* All sysctl names at this level are terminal. */
1309: if (namelen != 1)
1310: return (ENOTDIR);
1311:
1312: switch (name[0]) {
1313: case IPCTL_FORWARDING:
1314: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1315: case IPCTL_SENDREDIRECTS:
1316: return (sysctl_int(oldp, oldlenp, newp, newlen,
1317: &ipsendredirects));
1318: case IPCTL_DEFTTL:
1319: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1320: #ifdef notyet
1321: case IPCTL_DEFMTU:
1322: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1323: #endif
1.26 thorpej 1324: case IPCTL_FORWSRCRT:
1.47 cjs 1325: /* Don't allow this to change in a secure environment. */
1.26 thorpej 1326: if (securelevel > 0)
1.46 cjs 1327: return (sysctl_rdint(oldp, oldlenp, newp,
1328: ip_forwsrcrt));
1329: else
1330: return (sysctl_int(oldp, oldlenp, newp, newlen,
1331: &ip_forwsrcrt));
1.27 thorpej 1332: case IPCTL_DIRECTEDBCAST:
1333: return (sysctl_int(oldp, oldlenp, newp, newlen,
1334: &ip_directedbcast));
1.47 cjs 1335: case IPCTL_ALLOWSRCRT:
1336: return (sysctl_int(oldp, oldlenp, newp, newlen,
1337: &ip_allowsrcrt));
1.52 thorpej 1338: case IPCTL_SUBNETSARELOCAL:
1339: return (sysctl_int(oldp, oldlenp, newp, newlen,
1340: &subnetsarelocal));
1.53 kml 1341: case IPCTL_MTUDISC:
1.53.2.1 mycroft 1342: error = sysctl_int(oldp, oldlenp, newp, newlen,
1343: &ip_mtudisc);
1344: if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) {
1345: ip_mtudisc_timeout_q =
1346: rt_timer_queue_create(ip_mtudisc_timeout);
1347: } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) {
1348: rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE);
1349: ip_mtudisc_timeout_q = NULL;
1350: }
1351: return error;
1352: case IPCTL_MTUDISCTIMEOUT:
1353: error = sysctl_int(oldp, oldlenp, newp, newlen,
1354: &ip_mtudisc_timeout);
1355: if (ip_mtudisc_timeout_q != NULL)
1356: rt_timer_queue_change(ip_mtudisc_timeout_q,
1357: ip_mtudisc_timeout);
1358: return (error);
1.13 mycroft 1359: default:
1360: return (EOPNOTSUPP);
1361: }
1362: /* NOTREACHED */
1.1 cgd 1363: }
CVSweb <webmaster@jp.NetBSD.org>
![[BACK]](/icons/back.gif){kind=icon}
Return to ip_input.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [cvs.NetBSD.org] / src / sys / netinet