Annotation of src/sys/netinet/ip_input.c, Revision 1.74
1.74 ! thorpej 1: /* $NetBSD: ip_input.c,v 1.73 1998/10/08 01:41:46 thorpej 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
1.71 tls 36: */
37:
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.
1.1 cgd 73: */
1.55 scottr 74:
1.62 matt 75: #include "opt_gateway.h"
1.69 mrg 76: #include "opt_pfil_hooks.h"
1.55 scottr 77: #include "opt_mrouting.h"
1.1 cgd 78:
1.5 mycroft 79: #include <sys/param.h>
80: #include <sys/systm.h>
81: #include <sys/malloc.h>
82: #include <sys/mbuf.h>
83: #include <sys/domain.h>
84: #include <sys/protosw.h>
85: #include <sys/socket.h>
1.44 thorpej 86: #include <sys/socketvar.h>
1.5 mycroft 87: #include <sys/errno.h>
88: #include <sys/time.h>
89: #include <sys/kernel.h>
1.28 christos 90: #include <sys/proc.h>
1.72 thorpej 91: #include <sys/pool.h>
1.28 christos 92:
93: #include <vm/vm.h>
94: #include <sys/sysctl.h>
1.1 cgd 95:
1.5 mycroft 96: #include <net/if.h>
1.44 thorpej 97: #include <net/if_dl.h>
1.5 mycroft 98: #include <net/route.h>
1.45 mrg 99: #include <net/pfil.h>
1.1 cgd 100:
1.5 mycroft 101: #include <netinet/in.h>
102: #include <netinet/in_systm.h>
103: #include <netinet/ip.h>
104: #include <netinet/in_pcb.h>
105: #include <netinet/in_var.h>
106: #include <netinet/ip_var.h>
107: #include <netinet/ip_icmp.h>
1.44 thorpej 108:
1.1 cgd 109: #ifndef IPFORWARDING
110: #ifdef GATEWAY
111: #define IPFORWARDING 1 /* forward IP packets not for us */
112: #else /* GATEWAY */
113: #define IPFORWARDING 0 /* don't forward IP packets not for us */
114: #endif /* GATEWAY */
115: #endif /* IPFORWARDING */
116: #ifndef IPSENDREDIRECTS
117: #define IPSENDREDIRECTS 1
118: #endif
1.26 thorpej 119: #ifndef IPFORWSRCRT
1.47 cjs 120: #define IPFORWSRCRT 1 /* forward source-routed packets */
121: #endif
122: #ifndef IPALLOWSRCRT
1.48 mrg 123: #define IPALLOWSRCRT 1 /* allow source-routed packets */
1.26 thorpej 124: #endif
1.53 kml 125: #ifndef IPMTUDISC
126: #define IPMTUDISC 0
127: #endif
1.60 kml 128: #ifndef IPMTUDISCTIMEOUT
1.61 kml 129: #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */
1.60 kml 130: #endif
1.53 kml 131:
1.27 thorpej 132: /*
133: * Note: DIRECTED_BROADCAST is handled this way so that previous
134: * configuration using this option will Just Work.
135: */
136: #ifndef IPDIRECTEDBCAST
137: #ifdef DIRECTED_BROADCAST
138: #define IPDIRECTEDBCAST 1
139: #else
140: #define IPDIRECTEDBCAST 0
141: #endif /* DIRECTED_BROADCAST */
142: #endif /* IPDIRECTEDBCAST */
1.1 cgd 143: int ipforwarding = IPFORWARDING;
144: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 145: int ip_defttl = IPDEFTTL;
1.26 thorpej 146: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 147: int ip_directedbcast = IPDIRECTEDBCAST;
1.47 cjs 148: int ip_allowsrcrt = IPALLOWSRCRT;
1.53 kml 149: int ip_mtudisc = IPMTUDISC;
1.60 kml 150: u_int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
1.1 cgd 151: #ifdef DIAGNOSTIC
152: int ipprintfs = 0;
153: #endif
154:
1.60 kml 155: struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
156:
1.1 cgd 157: extern struct domain inetdomain;
158: extern struct protosw inetsw[];
159: u_char ip_protox[IPPROTO_MAX];
160: int ipqmaxlen = IFQ_MAXLEN;
1.22 mycroft 161: struct in_ifaddrhead in_ifaddr;
1.57 tls 162: struct in_ifaddrhashhead *in_ifaddrhashtbl;
1.13 mycroft 163: struct ifqueue ipintrq;
1.63 matt 164: struct ipstat ipstat;
165: u_int16_t ip_id;
166: int ip_defttl;
167: struct ipqhead ipq;
1.1 cgd 168:
1.72 thorpej 169: struct pool ipqent_pool;
170:
1.1 cgd 171: /*
172: * We need to save the IP options in case a protocol wants to respond
173: * to an incoming packet over the same route if the packet got here
174: * using IP source routing. This allows connection establishment and
175: * maintenance when the remote end is on a network that is not known
176: * to us.
177: */
178: int ip_nhops = 0;
179: static struct ip_srcrt {
180: struct in_addr dst; /* final destination */
181: char nop; /* one NOP to align */
182: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
183: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
184: } ip_srcrt;
185:
1.13 mycroft 186: static void save_rte __P((u_char *, struct in_addr));
1.35 mycroft 187:
1.1 cgd 188: /*
189: * IP initialization: fill in IP protocol switch table.
190: * All protocols not implemented in kernel go to raw IP protocol handler.
191: */
1.8 mycroft 192: void
1.1 cgd 193: ip_init()
194: {
195: register struct protosw *pr;
196: register int i;
197:
1.72 thorpej 198: pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl",
199: 0, NULL, NULL, M_IPQ);
200:
1.1 cgd 201: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
202: if (pr == 0)
203: panic("ip_init");
204: for (i = 0; i < IPPROTO_MAX; i++)
205: ip_protox[i] = pr - inetsw;
206: for (pr = inetdomain.dom_protosw;
207: pr < inetdomain.dom_protoswNPROTOSW; pr++)
208: if (pr->pr_domain->dom_family == PF_INET &&
209: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
210: ip_protox[pr->pr_protocol] = pr - inetsw;
1.25 cgd 211: LIST_INIT(&ipq);
1.1 cgd 212: ip_id = time.tv_sec & 0xffff;
213: ipintrq.ifq_maxlen = ipqmaxlen;
1.22 mycroft 214: TAILQ_INIT(&in_ifaddr);
1.57 tls 215: in_ifaddrhashtbl =
216: hashinit(IN_IFADDR_HASH_SIZE, M_IFADDR, M_WAITOK, &in_ifaddrhash);
1.60 kml 217: if (ip_mtudisc != 0)
218: ip_mtudisc_timeout_q =
219: rt_timer_queue_create(ip_mtudisc_timeout);
1.73 thorpej 220: #ifdef GATEWAY
221: ipflow_init();
222: #endif
1.1 cgd 223: }
224:
225: struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
226: struct route ipforward_rt;
227:
228: /*
229: * Ip input routine. Checksum and byte swap header. If fragmented
230: * try to reassemble. Process options. Pass to next level.
231: */
1.8 mycroft 232: void
1.1 cgd 233: ipintr()
234: {
1.33 mrg 235: register struct ip *ip = NULL;
1.1 cgd 236: register struct mbuf *m;
237: register struct ipq *fp;
238: register struct in_ifaddr *ia;
1.57 tls 239: register struct ifaddr *ifa;
1.25 cgd 240: struct ipqent *ipqe;
1.35 mycroft 241: int hlen = 0, mff, len, s;
1.36 mrg 242: #ifdef PFIL_HOOKS
1.33 mrg 243: struct packet_filter_hook *pfh;
244: struct mbuf *m0;
1.43 mrg 245: int rv;
1.36 mrg 246: #endif /* PFIL_HOOKS */
1.1 cgd 247:
248: next:
249: /*
250: * Get next datagram off input queue and get IP header
251: * in first mbuf.
252: */
253: s = splimp();
254: IF_DEQUEUE(&ipintrq, m);
255: splx(s);
1.13 mycroft 256: if (m == 0)
1.1 cgd 257: return;
258: #ifdef DIAGNOSTIC
259: if ((m->m_flags & M_PKTHDR) == 0)
260: panic("ipintr no HDR");
261: #endif
262: /*
263: * If no IP addresses have been set yet but the interfaces
264: * are receiving, can't do anything with incoming packets yet.
265: */
1.22 mycroft 266: if (in_ifaddr.tqh_first == 0)
1.1 cgd 267: goto bad;
268: ipstat.ips_total++;
269: if (m->m_len < sizeof (struct ip) &&
270: (m = m_pullup(m, sizeof (struct ip))) == 0) {
271: ipstat.ips_toosmall++;
272: goto next;
273: }
274: ip = mtod(m, struct ip *);
1.13 mycroft 275: if (ip->ip_v != IPVERSION) {
276: ipstat.ips_badvers++;
277: goto bad;
278: }
1.1 cgd 279: hlen = ip->ip_hl << 2;
280: if (hlen < sizeof(struct ip)) { /* minimum header length */
281: ipstat.ips_badhlen++;
282: goto bad;
283: }
284: if (hlen > m->m_len) {
285: if ((m = m_pullup(m, hlen)) == 0) {
286: ipstat.ips_badhlen++;
287: goto next;
288: }
289: ip = mtod(m, struct ip *);
290: }
1.28 christos 291: if ((ip->ip_sum = in_cksum(m, hlen)) != 0) {
1.1 cgd 292: ipstat.ips_badsum++;
293: goto bad;
294: }
295:
296: /*
297: * Convert fields to host representation.
298: */
299: NTOHS(ip->ip_len);
300: NTOHS(ip->ip_id);
301: NTOHS(ip->ip_off);
1.35 mycroft 302: len = ip->ip_len;
1.1 cgd 303:
304: /*
305: * Check that the amount of data in the buffers
306: * is as at least much as the IP header would have us expect.
307: * Trim mbufs if longer than we expect.
308: * Drop packet if shorter than we expect.
309: */
1.35 mycroft 310: if (m->m_pkthdr.len < len) {
1.1 cgd 311: ipstat.ips_tooshort++;
312: goto bad;
313: }
1.35 mycroft 314: if (m->m_pkthdr.len > len) {
1.1 cgd 315: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 316: m->m_len = len;
317: m->m_pkthdr.len = len;
1.1 cgd 318: } else
1.35 mycroft 319: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 320: }
321:
1.64 thorpej 322: /*
323: * Assume that we can create a fast-forward IP flow entry
324: * based on this packet.
325: */
326: m->m_flags |= M_CANFASTFWD;
327:
1.36 mrg 328: #ifdef PFIL_HOOKS
1.33 mrg 329: /*
1.64 thorpej 330: * Run through list of hooks for input packets. If there are any
331: * filters which require that additional packets in the flow are
332: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
333: * Note that filters must _never_ set this flag, as another filter
334: * in the list may have previously cleared it.
1.33 mrg 335: */
336: m0 = m;
1.59 mrg 337: for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next)
1.33 mrg 338: if (pfh->pfil_func) {
1.43 mrg 339: rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0);
340: if (rv)
1.40 veego 341: goto next;
1.68 sommerfe 342: m = m0;
343: if (m == NULL)
344: goto next;
345: ip = mtod(m, struct ip *);
1.33 mrg 346: }
1.36 mrg 347: #endif /* PFIL_HOOKS */
1.33 mrg 348:
1.1 cgd 349: /*
350: * Process options and, if not destined for us,
351: * ship it on. ip_dooptions returns 1 when an
352: * error was detected (causing an icmp message
353: * to be sent and the original packet to be freed).
354: */
355: ip_nhops = 0; /* for source routed packets */
356: if (hlen > sizeof (struct ip) && ip_dooptions(m))
357: goto next;
358:
359: /*
360: * Check our list of addresses, to see if the packet is for us.
361: */
1.57 tls 362: INADDR_TO_IA(ip->ip_dst, ia);
363: if (ia != NULL) goto ours;
364: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
365: for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
366: ifa != NULL; ifa = ifa->ifa_list.tqe_next) {
367: if (ifa->ifa_addr->sa_family != AF_INET) continue;
368: ia = ifatoia(ifa);
1.35 mycroft 369: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
370: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 371: /*
372: * Look for all-0's host part (old broadcast addr),
373: * either for subnet or net.
374: */
375: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 376: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 377: goto ours;
1.57 tls 378: /*
379: * An interface with IP address zero accepts
380: * all packets that arrive on that interface.
381: */
382: if (in_nullhost(ia->ia_addr.sin_addr))
383: goto ours;
1.1 cgd 384: }
385: }
1.18 mycroft 386: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 387: struct in_multi *inm;
388: #ifdef MROUTING
389: extern struct socket *ip_mrouter;
1.10 brezak 390:
391: if (m->m_flags & M_EXT) {
392: if ((m = m_pullup(m, hlen)) == 0) {
393: ipstat.ips_toosmall++;
394: goto next;
395: }
396: ip = mtod(m, struct ip *);
397: }
1.4 hpeyerl 398:
399: if (ip_mrouter) {
400: /*
401: * If we are acting as a multicast router, all
402: * incoming multicast packets are passed to the
403: * kernel-level multicast forwarding function.
404: * The packet is returned (relatively) intact; if
405: * ip_mforward() returns a non-zero value, the packet
406: * must be discarded, else it may be accepted below.
407: *
408: * (The IP ident field is put in the same byte order
409: * as expected when ip_mforward() is called from
410: * ip_output().)
411: */
412: ip->ip_id = htons(ip->ip_id);
1.13 mycroft 413: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
414: ipstat.ips_cantforward++;
1.4 hpeyerl 415: m_freem(m);
416: goto next;
417: }
418: ip->ip_id = ntohs(ip->ip_id);
419:
420: /*
421: * The process-level routing demon needs to receive
422: * all multicast IGMP packets, whether or not this
423: * host belongs to their destination groups.
424: */
425: if (ip->ip_p == IPPROTO_IGMP)
426: goto ours;
1.13 mycroft 427: ipstat.ips_forward++;
1.4 hpeyerl 428: }
429: #endif
430: /*
431: * See if we belong to the destination multicast group on the
432: * arrival interface.
433: */
434: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
435: if (inm == NULL) {
1.13 mycroft 436: ipstat.ips_cantforward++;
1.4 hpeyerl 437: m_freem(m);
438: goto next;
439: }
440: goto ours;
441: }
1.19 mycroft 442: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 443: in_nullhost(ip->ip_dst))
1.1 cgd 444: goto ours;
445:
446: /*
447: * Not for us; forward if possible and desirable.
448: */
449: if (ipforwarding == 0) {
450: ipstat.ips_cantforward++;
451: m_freem(m);
452: } else
453: ip_forward(m, 0);
454: goto next;
455:
456: ours:
457: /*
458: * If offset or IP_MF are set, must reassemble.
459: * Otherwise, nothing need be done.
460: * (We could look in the reassembly queue to see
461: * if the packet was previously fragmented,
462: * but it's not worth the time; just let them time out.)
463: */
1.37 perry 464: if (ip->ip_off & ~(IP_DF|IP_RF)) {
1.1 cgd 465: /*
466: * Look for queue of fragments
467: * of this datagram.
468: */
1.25 cgd 469: for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next)
1.1 cgd 470: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 471: in_hosteq(ip->ip_src, fp->ipq_src) &&
472: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 473: ip->ip_p == fp->ipq_p)
474: goto found;
475: fp = 0;
476: found:
477:
478: /*
479: * Adjust ip_len to not reflect header,
1.25 cgd 480: * set ipqe_mff if more fragments are expected,
1.1 cgd 481: * convert offset of this to bytes.
482: */
483: ip->ip_len -= hlen;
1.25 cgd 484: mff = (ip->ip_off & IP_MF) != 0;
485: if (mff) {
1.16 cgd 486: /*
487: * Make sure that fragments have a data length
488: * that's a non-zero multiple of 8 bytes.
489: */
1.17 cgd 490: if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
1.16 cgd 491: ipstat.ips_badfrags++;
492: goto bad;
493: }
494: }
1.1 cgd 495: ip->ip_off <<= 3;
496:
497: /*
498: * If datagram marked as having more fragments
499: * or if this is not the first fragment,
500: * attempt reassembly; if it succeeds, proceed.
501: */
1.25 cgd 502: if (mff || ip->ip_off) {
1.1 cgd 503: ipstat.ips_fragments++;
1.72 thorpej 504: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.25 cgd 505: if (ipqe == NULL) {
506: ipstat.ips_rcvmemdrop++;
507: goto bad;
508: }
509: ipqe->ipqe_mff = mff;
1.50 thorpej 510: ipqe->ipqe_m = m;
1.25 cgd 511: ipqe->ipqe_ip = ip;
1.50 thorpej 512: m = ip_reass(ipqe, fp);
513: if (m == 0)
1.1 cgd 514: goto next;
1.13 mycroft 515: ipstat.ips_reassembled++;
1.50 thorpej 516: ip = mtod(m, struct ip *);
1.74 ! thorpej 517: hlen = ip->ip_hl << 2;
1.1 cgd 518: } else
519: if (fp)
520: ip_freef(fp);
521: } else
522: ip->ip_len -= hlen;
523:
524: /*
525: * Switch out to protocol's input routine.
526: */
527: ipstat.ips_delivered++;
528: (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
529: goto next;
530: bad:
531: m_freem(m);
532: goto next;
533: }
534:
535: /*
536: * Take incoming datagram fragment and try to
537: * reassemble it into whole datagram. If a chain for
538: * reassembly of this datagram already exists, then it
539: * is given as fp; otherwise have to make a chain.
540: */
1.50 thorpej 541: struct mbuf *
1.25 cgd 542: ip_reass(ipqe, fp)
543: register struct ipqent *ipqe;
1.1 cgd 544: register struct ipq *fp;
545: {
1.50 thorpej 546: register struct mbuf *m = ipqe->ipqe_m;
1.25 cgd 547: register struct ipqent *nq, *p, *q;
548: struct ip *ip;
1.1 cgd 549: struct mbuf *t;
1.25 cgd 550: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 551: int i, next;
552:
553: /*
554: * Presence of header sizes in mbufs
555: * would confuse code below.
556: */
557: m->m_data += hlen;
558: m->m_len -= hlen;
559:
560: /*
561: * If first fragment to arrive, create a reassembly queue.
562: */
563: if (fp == 0) {
1.50 thorpej 564: MALLOC(fp, struct ipq *, sizeof (struct ipq),
565: M_FTABLE, M_NOWAIT);
566: if (fp == NULL)
1.1 cgd 567: goto dropfrag;
1.25 cgd 568: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 569: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 570: fp->ipq_p = ipqe->ipqe_ip->ip_p;
571: fp->ipq_id = ipqe->ipqe_ip->ip_id;
572: LIST_INIT(&fp->ipq_fragq);
573: fp->ipq_src = ipqe->ipqe_ip->ip_src;
574: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
575: p = NULL;
1.1 cgd 576: goto insert;
577: }
578:
579: /*
580: * Find a segment which begins after this one does.
581: */
1.25 cgd 582: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
583: p = q, q = q->ipqe_q.le_next)
584: if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off)
1.1 cgd 585: break;
586:
587: /*
588: * If there is a preceding segment, it may provide some of
589: * our data already. If so, drop the data from the incoming
590: * segment. If it provides all of our data, drop us.
591: */
1.25 cgd 592: if (p != NULL) {
593: i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len -
594: ipqe->ipqe_ip->ip_off;
1.1 cgd 595: if (i > 0) {
1.25 cgd 596: if (i >= ipqe->ipqe_ip->ip_len)
1.1 cgd 597: goto dropfrag;
1.50 thorpej 598: m_adj(ipqe->ipqe_m, i);
1.25 cgd 599: ipqe->ipqe_ip->ip_off += i;
600: ipqe->ipqe_ip->ip_len -= i;
1.1 cgd 601: }
602: }
603:
604: /*
605: * While we overlap succeeding segments trim them or,
606: * if they are completely covered, dequeue them.
607: */
1.25 cgd 608: for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len >
609: q->ipqe_ip->ip_off; q = nq) {
610: i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) -
611: q->ipqe_ip->ip_off;
612: if (i < q->ipqe_ip->ip_len) {
613: q->ipqe_ip->ip_len -= i;
614: q->ipqe_ip->ip_off += i;
1.50 thorpej 615: m_adj(q->ipqe_m, i);
1.1 cgd 616: break;
617: }
1.25 cgd 618: nq = q->ipqe_q.le_next;
1.50 thorpej 619: m_freem(q->ipqe_m);
1.25 cgd 620: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 621: pool_put(&ipqent_pool, q);
1.1 cgd 622: }
623:
624: insert:
625: /*
626: * Stick new segment in its place;
627: * check for complete reassembly.
628: */
1.25 cgd 629: if (p == NULL) {
630: LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
631: } else {
632: LIST_INSERT_AFTER(p, ipqe, ipqe_q);
633: }
1.1 cgd 634: next = 0;
1.25 cgd 635: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
636: p = q, q = q->ipqe_q.le_next) {
637: if (q->ipqe_ip->ip_off != next)
1.1 cgd 638: return (0);
1.25 cgd 639: next += q->ipqe_ip->ip_len;
1.1 cgd 640: }
1.25 cgd 641: if (p->ipqe_mff)
1.1 cgd 642: return (0);
643:
644: /*
1.41 thorpej 645: * Reassembly is complete. Check for a bogus message size and
646: * concatenate fragments.
1.1 cgd 647: */
1.25 cgd 648: q = fp->ipq_fragq.lh_first;
649: ip = q->ipqe_ip;
1.41 thorpej 650: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
651: ipstat.ips_toolong++;
652: ip_freef(fp);
653: return (0);
654: }
1.50 thorpej 655: m = q->ipqe_m;
1.1 cgd 656: t = m->m_next;
657: m->m_next = 0;
658: m_cat(m, t);
1.25 cgd 659: nq = q->ipqe_q.le_next;
1.72 thorpej 660: pool_put(&ipqent_pool, q);
1.25 cgd 661: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 662: t = q->ipqe_m;
1.25 cgd 663: nq = q->ipqe_q.le_next;
1.72 thorpej 664: pool_put(&ipqent_pool, q);
1.1 cgd 665: m_cat(m, t);
666: }
667:
668: /*
669: * Create header for new ip packet by
670: * modifying header of first packet;
671: * dequeue and discard fragment reassembly header.
672: * Make header visible.
673: */
674: ip->ip_len = next;
1.25 cgd 675: ip->ip_src = fp->ipq_src;
676: ip->ip_dst = fp->ipq_dst;
677: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 678: FREE(fp, M_FTABLE);
1.1 cgd 679: m->m_len += (ip->ip_hl << 2);
680: m->m_data -= (ip->ip_hl << 2);
681: /* some debugging cruft by sklower, below, will go away soon */
682: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
683: register int plen = 0;
1.50 thorpej 684: for (t = m; t; t = t->m_next)
685: plen += t->m_len;
686: m->m_pkthdr.len = plen;
1.1 cgd 687: }
1.50 thorpej 688: return (m);
1.1 cgd 689:
690: dropfrag:
691: ipstat.ips_fragdropped++;
692: m_freem(m);
1.72 thorpej 693: pool_put(&ipqent_pool, ipqe);
1.1 cgd 694: return (0);
695: }
696:
697: /*
698: * Free a fragment reassembly header and all
699: * associated datagrams.
700: */
1.8 mycroft 701: void
1.1 cgd 702: ip_freef(fp)
703: struct ipq *fp;
704: {
1.25 cgd 705: register struct ipqent *q, *p;
1.1 cgd 706:
1.25 cgd 707: for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) {
708: p = q->ipqe_q.le_next;
1.50 thorpej 709: m_freem(q->ipqe_m);
1.25 cgd 710: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 711: pool_put(&ipqent_pool, q);
1.1 cgd 712: }
1.25 cgd 713: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 714: FREE(fp, M_FTABLE);
1.1 cgd 715: }
716:
717: /*
718: * IP timer processing;
719: * if a timer expires on a reassembly
720: * queue, discard it.
721: */
1.8 mycroft 722: void
1.1 cgd 723: ip_slowtimo()
724: {
1.25 cgd 725: register struct ipq *fp, *nfp;
1.24 mycroft 726: int s = splsoftnet();
1.1 cgd 727:
1.25 cgd 728: for (fp = ipq.lh_first; fp != NULL; fp = nfp) {
729: nfp = fp->ipq_q.le_next;
730: if (--fp->ipq_ttl == 0) {
1.1 cgd 731: ipstat.ips_fragtimeout++;
1.25 cgd 732: ip_freef(fp);
1.1 cgd 733: }
734: }
1.63 matt 735: #ifdef GATEWAY
736: ipflow_slowtimo();
737: #endif
1.1 cgd 738: splx(s);
739: }
740:
741: /*
742: * Drain off all datagram fragments.
743: */
1.8 mycroft 744: void
1.1 cgd 745: ip_drain()
746: {
747:
1.25 cgd 748: while (ipq.lh_first != NULL) {
1.1 cgd 749: ipstat.ips_fragdropped++;
1.25 cgd 750: ip_freef(ipq.lh_first);
1.1 cgd 751: }
752: }
753:
754: /*
755: * Do option processing on a datagram,
756: * possibly discarding it if bad options are encountered,
757: * or forwarding it if source-routed.
758: * Returns 1 if packet has been forwarded/freed,
759: * 0 if the packet should be processed further.
760: */
1.8 mycroft 761: int
1.1 cgd 762: ip_dooptions(m)
763: struct mbuf *m;
764: {
765: register struct ip *ip = mtod(m, struct ip *);
766: register u_char *cp;
767: register struct ip_timestamp *ipt;
768: register struct in_ifaddr *ia;
769: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.13 mycroft 770: struct in_addr *sin, dst;
1.1 cgd 771: n_time ntime;
772:
1.13 mycroft 773: dst = ip->ip_dst;
1.1 cgd 774: cp = (u_char *)(ip + 1);
775: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
776: for (; cnt > 0; cnt -= optlen, cp += optlen) {
777: opt = cp[IPOPT_OPTVAL];
778: if (opt == IPOPT_EOL)
779: break;
780: if (opt == IPOPT_NOP)
781: optlen = 1;
782: else {
783: optlen = cp[IPOPT_OLEN];
784: if (optlen <= 0 || optlen > cnt) {
785: code = &cp[IPOPT_OLEN] - (u_char *)ip;
786: goto bad;
787: }
788: }
789: switch (opt) {
790:
791: default:
792: break;
793:
794: /*
795: * Source routing with record.
796: * Find interface with current destination address.
797: * If none on this machine then drop if strictly routed,
798: * or do nothing if loosely routed.
799: * Record interface address and bring up next address
800: * component. If strictly routed make sure next
801: * address is on directly accessible net.
802: */
803: case IPOPT_LSRR:
804: case IPOPT_SSRR:
1.47 cjs 805: if (ip_allowsrcrt == 0) {
806: type = ICMP_UNREACH;
807: code = ICMP_UNREACH_NET_PROHIB;
808: goto bad;
809: }
1.1 cgd 810: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
811: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
812: goto bad;
813: }
814: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 815: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 816: if (ia == 0) {
817: if (opt == IPOPT_SSRR) {
818: type = ICMP_UNREACH;
819: code = ICMP_UNREACH_SRCFAIL;
820: goto bad;
821: }
822: /*
823: * Loose routing, and not at next destination
824: * yet; nothing to do except forward.
825: */
826: break;
827: }
828: off--; /* 0 origin */
829: if (off > optlen - sizeof(struct in_addr)) {
830: /*
831: * End of source route. Should be for us.
832: */
833: save_rte(cp, ip->ip_src);
834: break;
835: }
836: /*
837: * locate outgoing interface
838: */
839: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
840: sizeof(ipaddr.sin_addr));
841: if (opt == IPOPT_SSRR) {
842: #define INA struct in_ifaddr *
843: #define SA struct sockaddr *
1.29 mrg 844: ia = (INA)ifa_ifwithladdr((SA)&ipaddr);
1.1 cgd 845: } else
846: ia = ip_rtaddr(ipaddr.sin_addr);
847: if (ia == 0) {
848: type = ICMP_UNREACH;
849: code = ICMP_UNREACH_SRCFAIL;
850: goto bad;
851: }
852: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 853: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 854: (caddr_t)(cp + off), sizeof(struct in_addr));
855: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 856: /*
857: * Let ip_intr's mcast routing check handle mcast pkts
858: */
1.18 mycroft 859: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 860: break;
861:
862: case IPOPT_RR:
863: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
864: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
865: goto bad;
866: }
867: /*
868: * If no space remains, ignore.
869: */
870: off--; /* 0 origin */
871: if (off > optlen - sizeof(struct in_addr))
872: break;
873: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
874: sizeof(ipaddr.sin_addr));
875: /*
876: * locate outgoing interface; if we're the destination,
877: * use the incoming interface (should be same).
878: */
879: if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
880: (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
881: type = ICMP_UNREACH;
882: code = ICMP_UNREACH_HOST;
883: goto bad;
884: }
1.20 mycroft 885: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 886: (caddr_t)(cp + off), sizeof(struct in_addr));
887: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
888: break;
889:
890: case IPOPT_TS:
891: code = cp - (u_char *)ip;
892: ipt = (struct ip_timestamp *)cp;
893: if (ipt->ipt_len < 5)
894: goto bad;
1.15 cgd 895: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.1 cgd 896: if (++ipt->ipt_oflw == 0)
897: goto bad;
898: break;
899: }
900: sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
901: switch (ipt->ipt_flg) {
902:
903: case IPOPT_TS_TSONLY:
904: break;
905:
906: case IPOPT_TS_TSANDADDR:
1.66 thorpej 907: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 908: sizeof(struct in_addr) > ipt->ipt_len)
909: goto bad;
1.13 mycroft 910: ipaddr.sin_addr = dst;
911: ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
912: m->m_pkthdr.rcvif);
913: if (ia == 0)
914: continue;
1.20 mycroft 915: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 916: (caddr_t)sin, sizeof(struct in_addr));
917: ipt->ipt_ptr += sizeof(struct in_addr);
918: break;
919:
920: case IPOPT_TS_PRESPEC:
1.66 thorpej 921: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 922: sizeof(struct in_addr) > ipt->ipt_len)
923: goto bad;
924: bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
925: sizeof(struct in_addr));
926: if (ifa_ifwithaddr((SA)&ipaddr) == 0)
927: continue;
928: ipt->ipt_ptr += sizeof(struct in_addr);
929: break;
930:
931: default:
932: goto bad;
933: }
934: ntime = iptime();
935: bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
936: sizeof(n_time));
937: ipt->ipt_ptr += sizeof(n_time);
938: }
939: }
940: if (forward) {
1.26 thorpej 941: if (ip_forwsrcrt == 0) {
942: type = ICMP_UNREACH;
943: code = ICMP_UNREACH_SRCFAIL;
944: goto bad;
945: }
1.1 cgd 946: ip_forward(m, 1);
947: return (1);
1.13 mycroft 948: }
949: return (0);
1.1 cgd 950: bad:
1.13 mycroft 951: ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */
952: icmp_error(m, type, code, 0, 0);
953: ipstat.ips_badoptions++;
1.1 cgd 954: return (1);
955: }
956:
957: /*
958: * Given address of next destination (final or next hop),
959: * return internet address info of interface to be used to get there.
960: */
961: struct in_ifaddr *
962: ip_rtaddr(dst)
963: struct in_addr dst;
964: {
965: register struct sockaddr_in *sin;
966:
1.19 mycroft 967: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 968:
1.35 mycroft 969: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 970: if (ipforward_rt.ro_rt) {
971: RTFREE(ipforward_rt.ro_rt);
972: ipforward_rt.ro_rt = 0;
973: }
974: sin->sin_family = AF_INET;
975: sin->sin_len = sizeof(*sin);
976: sin->sin_addr = dst;
977:
978: rtalloc(&ipforward_rt);
979: }
980: if (ipforward_rt.ro_rt == 0)
981: return ((struct in_ifaddr *)0);
1.19 mycroft 982: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 983: }
984:
985: /*
986: * Save incoming source route for use in replies,
987: * to be picked up later by ip_srcroute if the receiver is interested.
988: */
1.13 mycroft 989: void
1.1 cgd 990: save_rte(option, dst)
991: u_char *option;
992: struct in_addr dst;
993: {
994: unsigned olen;
995:
996: olen = option[IPOPT_OLEN];
997: #ifdef DIAGNOSTIC
998: if (ipprintfs)
1.39 christos 999: printf("save_rte: olen %d\n", olen);
1.1 cgd 1000: #endif
1001: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1002: return;
1003: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
1004: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1005: ip_srcrt.dst = dst;
1006: }
1007:
1008: /*
1009: * Retrieve incoming source route for use in replies,
1010: * in the same form used by setsockopt.
1011: * The first hop is placed before the options, will be removed later.
1012: */
1013: struct mbuf *
1014: ip_srcroute()
1015: {
1016: register struct in_addr *p, *q;
1017: register struct mbuf *m;
1018:
1019: if (ip_nhops == 0)
1020: return ((struct mbuf *)0);
1021: m = m_get(M_DONTWAIT, MT_SOOPTS);
1022: if (m == 0)
1023: return ((struct mbuf *)0);
1024:
1.13 mycroft 1025: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1026:
1027: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1028: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1029: OPTSIZ;
1030: #ifdef DIAGNOSTIC
1031: if (ipprintfs)
1.39 christos 1032: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1033: #endif
1034:
1035: /*
1036: * First save first hop for return route
1037: */
1038: p = &ip_srcrt.route[ip_nhops - 1];
1039: *(mtod(m, struct in_addr *)) = *p--;
1040: #ifdef DIAGNOSTIC
1041: if (ipprintfs)
1.39 christos 1042: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1043: #endif
1044:
1045: /*
1046: * Copy option fields and padding (nop) to mbuf.
1047: */
1048: ip_srcrt.nop = IPOPT_NOP;
1049: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1050: bcopy((caddr_t)&ip_srcrt.nop,
1051: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1052: q = (struct in_addr *)(mtod(m, caddr_t) +
1053: sizeof(struct in_addr) + OPTSIZ);
1054: #undef OPTSIZ
1055: /*
1056: * Record return path as an IP source route,
1057: * reversing the path (pointers are now aligned).
1058: */
1059: while (p >= ip_srcrt.route) {
1060: #ifdef DIAGNOSTIC
1061: if (ipprintfs)
1.39 christos 1062: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1063: #endif
1064: *q++ = *p--;
1065: }
1066: /*
1067: * Last hop goes to final destination.
1068: */
1069: *q = ip_srcrt.dst;
1070: #ifdef DIAGNOSTIC
1071: if (ipprintfs)
1.39 christos 1072: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1073: #endif
1074: return (m);
1075: }
1076:
1077: /*
1078: * Strip out IP options, at higher
1079: * level protocol in the kernel.
1080: * Second argument is buffer to which options
1081: * will be moved, and return value is their length.
1082: * XXX should be deleted; last arg currently ignored.
1083: */
1.8 mycroft 1084: void
1.1 cgd 1085: ip_stripoptions(m, mopt)
1086: register struct mbuf *m;
1087: struct mbuf *mopt;
1088: {
1089: register int i;
1090: struct ip *ip = mtod(m, struct ip *);
1091: register caddr_t opts;
1092: int olen;
1093:
1094: olen = (ip->ip_hl<<2) - sizeof (struct ip);
1095: opts = (caddr_t)(ip + 1);
1096: i = m->m_len - (sizeof (struct ip) + olen);
1097: bcopy(opts + olen, opts, (unsigned)i);
1098: m->m_len -= olen;
1099: if (m->m_flags & M_PKTHDR)
1100: m->m_pkthdr.len -= olen;
1101: ip->ip_hl = sizeof(struct ip) >> 2;
1102: }
1103:
1.23 mycroft 1104: int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1105: 0, 0, 0, 0,
1106: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1107: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1108: EMSGSIZE, EHOSTUNREACH, 0, 0,
1109: 0, 0, 0, 0,
1110: ENOPROTOOPT
1111: };
1112:
1113: /*
1114: * Forward a packet. If some error occurs return the sender
1115: * an icmp packet. Note we can't always generate a meaningful
1116: * icmp message because icmp doesn't have a large enough repertoire
1117: * of codes and types.
1118: *
1119: * If not forwarding, just drop the packet. This could be confusing
1120: * if ipforwarding was zero but some routing protocol was advancing
1121: * us as a gateway to somewhere. However, we must let the routing
1122: * protocol deal with that.
1123: *
1124: * The srcrt parameter indicates whether the packet is being forwarded
1125: * via a source route.
1126: */
1.13 mycroft 1127: void
1.1 cgd 1128: ip_forward(m, srcrt)
1129: struct mbuf *m;
1130: int srcrt;
1131: {
1132: register struct ip *ip = mtod(m, struct ip *);
1133: register struct sockaddr_in *sin;
1134: register struct rtentry *rt;
1.28 christos 1135: int error, type = 0, code = 0;
1.1 cgd 1136: struct mbuf *mcopy;
1.13 mycroft 1137: n_long dest;
1138: struct ifnet *destifp;
1.1 cgd 1139:
1.13 mycroft 1140: dest = 0;
1.1 cgd 1141: #ifdef DIAGNOSTIC
1142: if (ipprintfs)
1.70 thorpej 1143: printf("forward: src %2.2x dst %2.2x ttl %x\n",
1144: ntohl(ip->ip_src.s_addr),
1145: ntohl(ip->ip_dst.s_addr), ip->ip_ttl);
1.1 cgd 1146: #endif
1147: if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1148: ipstat.ips_cantforward++;
1149: m_freem(m);
1150: return;
1151: }
1152: HTONS(ip->ip_id);
1153: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1154: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1155: return;
1156: }
1157: ip->ip_ttl -= IPTTLDEC;
1158:
1.19 mycroft 1159: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1160: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1161: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1162: if (ipforward_rt.ro_rt) {
1163: RTFREE(ipforward_rt.ro_rt);
1164: ipforward_rt.ro_rt = 0;
1165: }
1166: sin->sin_family = AF_INET;
1.35 mycroft 1167: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1168: sin->sin_addr = ip->ip_dst;
1169:
1170: rtalloc(&ipforward_rt);
1171: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1172: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1173: return;
1174: }
1175: rt = ipforward_rt.ro_rt;
1176: }
1177:
1178: /*
1.34 mycroft 1179: * Save at most 68 bytes of the packet in case
1.1 cgd 1180: * we need to generate an ICMP message to the src.
1181: */
1.34 mycroft 1182: mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68));
1.1 cgd 1183:
1184: /*
1185: * If forwarding packet using same interface that it came in on,
1186: * perhaps should send a redirect to sender to shortcut a hop.
1187: * Only send redirect if source is sending directly to us,
1188: * and if packet was not source routed (or has any options).
1189: * Also, don't send redirect if forwarding using a default route
1190: * or a route modified by a redirect.
1191: */
1192: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1193: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1194: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1195: ipsendredirects && !srcrt) {
1.19 mycroft 1196: if (rt->rt_ifa &&
1197: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1198: ifatoia(rt->rt_ifa)->ia_subnet) {
1.1 cgd 1199: if (rt->rt_flags & RTF_GATEWAY)
1.13 mycroft 1200: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1.1 cgd 1201: else
1.13 mycroft 1202: dest = ip->ip_dst.s_addr;
1203: /* Router requirements says to only send host redirects */
1.1 cgd 1204: type = ICMP_REDIRECT;
1.13 mycroft 1205: code = ICMP_REDIRECT_HOST;
1.1 cgd 1206: #ifdef DIAGNOSTIC
1207: if (ipprintfs)
1.39 christos 1208: printf("redirect (%d) to %x\n", code, (u_int32_t)dest);
1.1 cgd 1209: #endif
1210: }
1211: }
1212:
1.27 thorpej 1213: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1214: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1.1 cgd 1215: if (error)
1216: ipstat.ips_cantforward++;
1217: else {
1218: ipstat.ips_forward++;
1219: if (type)
1220: ipstat.ips_redirectsent++;
1221: else {
1.63 matt 1222: if (mcopy) {
1223: #ifdef GATEWAY
1.64 thorpej 1224: if (mcopy->m_flags & M_CANFASTFWD)
1225: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1226: #endif
1.1 cgd 1227: m_freem(mcopy);
1.63 matt 1228: }
1.1 cgd 1229: return;
1230: }
1231: }
1232: if (mcopy == NULL)
1233: return;
1.13 mycroft 1234: destifp = NULL;
1235:
1.1 cgd 1236: switch (error) {
1237:
1238: case 0: /* forwarded, but need redirect */
1239: /* type, code set above */
1240: break;
1241:
1242: case ENETUNREACH: /* shouldn't happen, checked above */
1243: case EHOSTUNREACH:
1244: case ENETDOWN:
1245: case EHOSTDOWN:
1246: default:
1247: type = ICMP_UNREACH;
1248: code = ICMP_UNREACH_HOST;
1249: break;
1250:
1251: case EMSGSIZE:
1252: type = ICMP_UNREACH;
1253: code = ICMP_UNREACH_NEEDFRAG;
1.13 mycroft 1254: if (ipforward_rt.ro_rt)
1255: destifp = ipforward_rt.ro_rt->rt_ifp;
1.1 cgd 1256: ipstat.ips_cantfrag++;
1257: break;
1258:
1259: case ENOBUFS:
1260: type = ICMP_SOURCEQUENCH;
1261: code = 0;
1262: break;
1263: }
1.13 mycroft 1264: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 1265: }
1266:
1267: void
1268: ip_savecontrol(inp, mp, ip, m)
1269: register struct inpcb *inp;
1270: register struct mbuf **mp;
1271: register struct ip *ip;
1272: register struct mbuf *m;
1273: {
1274:
1275: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1276: struct timeval tv;
1277:
1278: microtime(&tv);
1279: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1280: SCM_TIMESTAMP, SOL_SOCKET);
1281: if (*mp)
1282: mp = &(*mp)->m_next;
1283: }
1284: if (inp->inp_flags & INP_RECVDSTADDR) {
1285: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1286: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1287: if (*mp)
1288: mp = &(*mp)->m_next;
1289: }
1290: #ifdef notyet
1291: /*
1292: * XXX
1293: * Moving these out of udp_input() made them even more broken
1294: * than they already were.
1295: * - fenner@parc.xerox.com
1296: */
1297: /* options were tossed already */
1298: if (inp->inp_flags & INP_RECVOPTS) {
1299: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1300: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1301: if (*mp)
1302: mp = &(*mp)->m_next;
1303: }
1304: /* ip_srcroute doesn't do what we want here, need to fix */
1305: if (inp->inp_flags & INP_RECVRETOPTS) {
1306: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1307: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1308: if (*mp)
1309: mp = &(*mp)->m_next;
1310: }
1311: #endif
1312: if (inp->inp_flags & INP_RECVIF) {
1313: struct sockaddr_dl sdl;
1314:
1315: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1316: sdl.sdl_family = AF_LINK;
1317: sdl.sdl_index = m->m_pkthdr.rcvif ?
1318: m->m_pkthdr.rcvif->if_index : 0;
1319: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1320: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1321: IP_RECVIF, IPPROTO_IP);
1322: if (*mp)
1323: mp = &(*mp)->m_next;
1324: }
1.13 mycroft 1325: }
1326:
1327: int
1328: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1329: int *name;
1330: u_int namelen;
1331: void *oldp;
1332: size_t *oldlenp;
1333: void *newp;
1334: size_t newlen;
1335: {
1.52 thorpej 1336: extern int subnetsarelocal;
1337:
1.54 lukem 1338: int error, old;
1339:
1.13 mycroft 1340: /* All sysctl names at this level are terminal. */
1341: if (namelen != 1)
1342: return (ENOTDIR);
1343:
1344: switch (name[0]) {
1345: case IPCTL_FORWARDING:
1346: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1347: case IPCTL_SENDREDIRECTS:
1348: return (sysctl_int(oldp, oldlenp, newp, newlen,
1349: &ipsendredirects));
1350: case IPCTL_DEFTTL:
1351: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1352: #ifdef notyet
1353: case IPCTL_DEFMTU:
1354: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1355: #endif
1.26 thorpej 1356: case IPCTL_FORWSRCRT:
1.47 cjs 1357: /* Don't allow this to change in a secure environment. */
1.26 thorpej 1358: if (securelevel > 0)
1.46 cjs 1359: return (sysctl_rdint(oldp, oldlenp, newp,
1360: ip_forwsrcrt));
1361: else
1362: return (sysctl_int(oldp, oldlenp, newp, newlen,
1363: &ip_forwsrcrt));
1.27 thorpej 1364: case IPCTL_DIRECTEDBCAST:
1365: return (sysctl_int(oldp, oldlenp, newp, newlen,
1366: &ip_directedbcast));
1.47 cjs 1367: case IPCTL_ALLOWSRCRT:
1368: return (sysctl_int(oldp, oldlenp, newp, newlen,
1369: &ip_allowsrcrt));
1.52 thorpej 1370: case IPCTL_SUBNETSARELOCAL:
1371: return (sysctl_int(oldp, oldlenp, newp, newlen,
1372: &subnetsarelocal));
1.53 kml 1373: case IPCTL_MTUDISC:
1.60 kml 1374: error = sysctl_int(oldp, oldlenp, newp, newlen,
1375: &ip_mtudisc);
1376: if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) {
1377: ip_mtudisc_timeout_q =
1378: rt_timer_queue_create(ip_mtudisc_timeout);
1379: } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) {
1380: rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE);
1381: ip_mtudisc_timeout_q = NULL;
1382: }
1383: return error;
1.54 lukem 1384: case IPCTL_ANONPORTMIN:
1385: old = anonportmin;
1386: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1387: if (anonportmin >= anonportmax || anonportmin > 65535
1388: #ifndef IPNOPRIVPORTS
1389: || anonportmin < IPPORT_RESERVED
1390: #endif
1391: ) {
1392: anonportmin = old;
1393: return (EINVAL);
1394: }
1395: return (error);
1396: case IPCTL_ANONPORTMAX:
1397: old = anonportmax;
1398: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1399: if (anonportmin >= anonportmax || anonportmax > 65535
1400: #ifndef IPNOPRIVPORTS
1401: || anonportmax < IPPORT_RESERVED
1402: #endif
1403: ) {
1404: anonportmax = old;
1405: return (EINVAL);
1406: }
1.60 kml 1407: return (error);
1408: case IPCTL_MTUDISCTIMEOUT:
1409: error = sysctl_int(oldp, oldlenp, newp, newlen,
1410: &ip_mtudisc_timeout);
1411: if (ip_mtudisc_timeout_q != NULL)
1412: rt_timer_queue_change(ip_mtudisc_timeout_q,
1413: ip_mtudisc_timeout);
1.54 lukem 1414: return (error);
1.65 matt 1415: #ifdef GATEWAY
1416: case IPCTL_MAXFLOWS:
1.67 thorpej 1417: {
1418: int s;
1419:
1.65 matt 1420: error = sysctl_int(oldp, oldlenp, newp, newlen,
1421: &ip_maxflows);
1.67 thorpej 1422: s = splsoftnet();
1.65 matt 1423: ipflow_reap(0);
1.67 thorpej 1424: splx(s);
1.65 matt 1425: return (error);
1.67 thorpej 1426: }
1.65 matt 1427: #endif
1.13 mycroft 1428: default:
1429: return (EOPNOTSUPP);
1430: }
1431: /* NOTREACHED */
1.1 cgd 1432: }
CVSweb <webmaster@jp.NetBSD.org>