Annotation of src/sys/netinet/ip_input.c, Revision 1.174
1.174 ! itojun 1: /* $NetBSD: ip_input.c,v 1.173 2003/08/15 03:42:02 jonathan Exp $ */
1.89 itojun 2:
3: /*
4: * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5: * All rights reserved.
1.152 itojun 6: *
1.89 itojun 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. Neither the name of the project nor the names of its contributors
16: * may be used to endorse or promote products derived from this software
17: * without specific prior written permission.
1.152 itojun 18: *
1.89 itojun 19: * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22: * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29: * SUCH DAMAGE.
30: */
1.76 thorpej 31:
32: /*-
33: * Copyright (c) 1998 The NetBSD Foundation, Inc.
34: * All rights reserved.
35: *
36: * This code is derived from software contributed to The NetBSD Foundation
37: * by Public Access Networks Corporation ("Panix"). It was developed under
38: * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon.
39: *
40: * Redistribution and use in source and binary forms, with or without
41: * modification, are permitted provided that the following conditions
42: * are met:
43: * 1. Redistributions of source code must retain the above copyright
44: * notice, this list of conditions and the following disclaimer.
45: * 2. Redistributions in binary form must reproduce the above copyright
46: * notice, this list of conditions and the following disclaimer in the
47: * documentation and/or other materials provided with the distribution.
48: * 3. All advertising materials mentioning features or use of this software
49: * must display the following acknowledgement:
50: * This product includes software developed by the NetBSD
51: * Foundation, Inc. and its contributors.
52: * 4. Neither the name of The NetBSD Foundation nor the names of its
53: * contributors may be used to endorse or promote products derived
54: * from this software without specific prior written permission.
55: *
56: * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
57: * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
58: * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
59: * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
60: * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
61: * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
62: * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
63: * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
64: * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
65: * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
66: * POSSIBILITY OF SUCH DAMAGE.
67: */
1.14 cgd 68:
1.1 cgd 69: /*
1.13 mycroft 70: * Copyright (c) 1982, 1986, 1988, 1993
71: * The Regents of the University of California. All rights reserved.
1.1 cgd 72: *
73: * Redistribution and use in source and binary forms, with or without
74: * modification, are permitted provided that the following conditions
75: * are met:
76: * 1. Redistributions of source code must retain the above copyright
77: * notice, this list of conditions and the following disclaimer.
78: * 2. Redistributions in binary form must reproduce the above copyright
79: * notice, this list of conditions and the following disclaimer in the
80: * documentation and/or other materials provided with the distribution.
1.172 agc 81: * 3. Neither the name of the University nor the names of its contributors
1.1 cgd 82: * may be used to endorse or promote products derived from this software
83: * without specific prior written permission.
84: *
85: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
86: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
87: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
88: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
89: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
90: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
91: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
92: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
93: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
94: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
95: * SUCH DAMAGE.
96: *
1.14 cgd 97: * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
1.1 cgd 98: */
1.141 lukem 99:
100: #include <sys/cdefs.h>
1.174 ! itojun 101: __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.173 2003/08/15 03:42:02 jonathan Exp $");
1.55 scottr 102:
1.62 matt 103: #include "opt_gateway.h"
1.69 mrg 104: #include "opt_pfil_hooks.h"
1.91 thorpej 105: #include "opt_ipsec.h"
1.55 scottr 106: #include "opt_mrouting.h"
1.167 martin 107: #include "opt_mbuftrace.h"
1.135 thorpej 108: #include "opt_inet_csum.h"
1.1 cgd 109:
1.5 mycroft 110: #include <sys/param.h>
111: #include <sys/systm.h>
112: #include <sys/malloc.h>
113: #include <sys/mbuf.h>
114: #include <sys/domain.h>
115: #include <sys/protosw.h>
116: #include <sys/socket.h>
1.44 thorpej 117: #include <sys/socketvar.h>
1.5 mycroft 118: #include <sys/errno.h>
119: #include <sys/time.h>
120: #include <sys/kernel.h>
1.72 thorpej 121: #include <sys/pool.h>
1.28 christos 122: #include <sys/sysctl.h>
1.1 cgd 123:
1.5 mycroft 124: #include <net/if.h>
1.44 thorpej 125: #include <net/if_dl.h>
1.5 mycroft 126: #include <net/route.h>
1.45 mrg 127: #include <net/pfil.h>
1.1 cgd 128:
1.5 mycroft 129: #include <netinet/in.h>
130: #include <netinet/in_systm.h>
131: #include <netinet/ip.h>
132: #include <netinet/in_pcb.h>
133: #include <netinet/in_var.h>
134: #include <netinet/ip_var.h>
135: #include <netinet/ip_icmp.h>
1.89 itojun 136: /* just for gif_ttl */
137: #include <netinet/in_gif.h>
138: #include "gif.h"
1.144 martin 139: #include <net/if_gre.h>
140: #include "gre.h"
1.111 jdolecek 141:
142: #ifdef MROUTING
143: #include <netinet/ip_mroute.h>
144: #endif
1.89 itojun 145:
146: #ifdef IPSEC
147: #include <netinet6/ipsec.h>
148: #include <netkey/key.h>
149: #endif
1.173 jonathan 150: #ifdef FAST_IPSEC
151: #include <netipsec/ipsec.h>
152: #include <netipsec/key.h>
153: #endif /* FAST_IPSEC*/
1.44 thorpej 154:
1.1 cgd 155: #ifndef IPFORWARDING
156: #ifdef GATEWAY
157: #define IPFORWARDING 1 /* forward IP packets not for us */
158: #else /* GATEWAY */
159: #define IPFORWARDING 0 /* don't forward IP packets not for us */
160: #endif /* GATEWAY */
161: #endif /* IPFORWARDING */
162: #ifndef IPSENDREDIRECTS
163: #define IPSENDREDIRECTS 1
164: #endif
1.26 thorpej 165: #ifndef IPFORWSRCRT
1.47 cjs 166: #define IPFORWSRCRT 1 /* forward source-routed packets */
167: #endif
168: #ifndef IPALLOWSRCRT
1.48 mrg 169: #define IPALLOWSRCRT 1 /* allow source-routed packets */
1.26 thorpej 170: #endif
1.53 kml 171: #ifndef IPMTUDISC
1.153 itojun 172: #define IPMTUDISC 1
1.53 kml 173: #endif
1.60 kml 174: #ifndef IPMTUDISCTIMEOUT
1.61 kml 175: #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */
1.60 kml 176: #endif
1.53 kml 177:
1.27 thorpej 178: /*
179: * Note: DIRECTED_BROADCAST is handled this way so that previous
180: * configuration using this option will Just Work.
181: */
182: #ifndef IPDIRECTEDBCAST
183: #ifdef DIRECTED_BROADCAST
184: #define IPDIRECTEDBCAST 1
185: #else
186: #define IPDIRECTEDBCAST 0
187: #endif /* DIRECTED_BROADCAST */
188: #endif /* IPDIRECTEDBCAST */
1.1 cgd 189: int ipforwarding = IPFORWARDING;
190: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 191: int ip_defttl = IPDEFTTL;
1.26 thorpej 192: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 193: int ip_directedbcast = IPDIRECTEDBCAST;
1.47 cjs 194: int ip_allowsrcrt = IPALLOWSRCRT;
1.53 kml 195: int ip_mtudisc = IPMTUDISC;
1.156 itojun 196: int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
1.1 cgd 197: #ifdef DIAGNOSTIC
198: int ipprintfs = 0;
199: #endif
1.165 christos 200: /*
201: * XXX - Setting ip_checkinterface mostly implements the receive side of
202: * the Strong ES model described in RFC 1122, but since the routing table
203: * and transmit implementation do not implement the Strong ES model,
204: * setting this to 1 results in an odd hybrid.
205: *
206: * XXX - ip_checkinterface currently must be disabled if you use ipnat
207: * to translate the destination address to another local interface.
208: *
209: * XXX - ip_checkinterface must be disabled if you add IP aliases
210: * to the loopback interface instead of the interface where the
211: * packets for those addresses are received.
212: */
213: int ip_checkinterface = 0;
214:
1.1 cgd 215:
1.60 kml 216: struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
217:
1.1 cgd 218: extern struct domain inetdomain;
219: int ipqmaxlen = IFQ_MAXLEN;
1.150 matt 220: u_long in_ifaddrhash; /* size of hash table - 1 */
221: int in_ifaddrentries; /* total number of addrs */
1.22 mycroft 222: struct in_ifaddrhead in_ifaddr;
1.57 tls 223: struct in_ifaddrhashhead *in_ifaddrhashtbl;
1.166 matt 224: u_long in_multihash; /* size of hash table - 1 */
225: int in_multientries; /* total number of addrs */
226: struct in_multihashhead *in_multihashtbl;
1.13 mycroft 227: struct ifqueue ipintrq;
1.63 matt 228: struct ipstat ipstat;
229: u_int16_t ip_id;
1.75 thorpej 230:
1.121 thorpej 231: #ifdef PFIL_HOOKS
232: struct pfil_head inet_pfil_hook;
233: #endif
234:
1.63 matt 235: struct ipqhead ipq;
1.75 thorpej 236: int ipq_locked;
1.131 itojun 237: int ip_nfragpackets = 0;
1.133 itojun 238: int ip_maxfragpackets = 200;
1.75 thorpej 239:
240: static __inline int ipq_lock_try __P((void));
241: static __inline void ipq_unlock __P((void));
242:
243: static __inline int
244: ipq_lock_try()
245: {
246: int s;
247:
1.132 thorpej 248: /*
1.149 wiz 249: * Use splvm() -- we're blocking things that would cause
1.132 thorpej 250: * mbuf allocation.
251: */
252: s = splvm();
1.75 thorpej 253: if (ipq_locked) {
254: splx(s);
255: return (0);
256: }
257: ipq_locked = 1;
258: splx(s);
259: return (1);
260: }
261:
262: static __inline void
263: ipq_unlock()
264: {
265: int s;
266:
1.132 thorpej 267: s = splvm();
1.75 thorpej 268: ipq_locked = 0;
269: splx(s);
270: }
271:
272: #ifdef DIAGNOSTIC
273: #define IPQ_LOCK() \
274: do { \
275: if (ipq_lock_try() == 0) { \
276: printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \
277: panic("ipq_lock"); \
278: } \
1.159 perry 279: } while (/*CONSTCOND*/ 0)
1.75 thorpej 280: #define IPQ_LOCK_CHECK() \
281: do { \
282: if (ipq_locked == 0) { \
283: printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \
284: panic("ipq lock check"); \
285: } \
1.159 perry 286: } while (/*CONSTCOND*/ 0)
1.75 thorpej 287: #else
288: #define IPQ_LOCK() (void) ipq_lock_try()
289: #define IPQ_LOCK_CHECK() /* nothing */
290: #endif
291:
292: #define IPQ_UNLOCK() ipq_unlock()
1.1 cgd 293:
1.166 matt 294: struct pool inmulti_pool;
1.72 thorpej 295: struct pool ipqent_pool;
296:
1.135 thorpej 297: #ifdef INET_CSUM_COUNTERS
298: #include <sys/device.h>
299:
300: struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
301: NULL, "inet", "hwcsum bad");
302: struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
303: NULL, "inet", "hwcsum ok");
304: struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
305: NULL, "inet", "swcsum");
306:
307: #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++
308:
309: #else
310:
311: #define INET_CSUM_COUNTER_INCR(ev) /* nothing */
312:
313: #endif /* INET_CSUM_COUNTERS */
314:
1.1 cgd 315: /*
316: * We need to save the IP options in case a protocol wants to respond
317: * to an incoming packet over the same route if the packet got here
318: * using IP source routing. This allows connection establishment and
319: * maintenance when the remote end is on a network that is not known
320: * to us.
321: */
322: int ip_nhops = 0;
323: static struct ip_srcrt {
324: struct in_addr dst; /* final destination */
325: char nop; /* one NOP to align */
326: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
327: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
328: } ip_srcrt;
329:
1.13 mycroft 330: static void save_rte __P((u_char *, struct in_addr));
1.35 mycroft 331:
1.164 matt 332: #ifdef MBUFTRACE
333: struct mowner ip_rx_mowner = { "internet", "rx" };
334: struct mowner ip_tx_mowner = { "internet", "tx" };
335: #endif
336:
1.1 cgd 337: /*
338: * IP initialization: fill in IP protocol switch table.
339: * All protocols not implemented in kernel go to raw IP protocol handler.
340: */
1.8 mycroft 341: void
1.1 cgd 342: ip_init()
343: {
1.109 augustss 344: struct protosw *pr;
345: int i;
1.1 cgd 346:
1.166 matt 347: pool_init(&inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl",
348: NULL);
1.72 thorpej 349: pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl",
1.146 thorpej 350: NULL);
1.72 thorpej 351:
1.1 cgd 352: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
353: if (pr == 0)
354: panic("ip_init");
355: for (i = 0; i < IPPROTO_MAX; i++)
356: ip_protox[i] = pr - inetsw;
357: for (pr = inetdomain.dom_protosw;
358: pr < inetdomain.dom_protoswNPROTOSW; pr++)
359: if (pr->pr_domain->dom_family == PF_INET &&
360: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
361: ip_protox[pr->pr_protocol] = pr - inetsw;
1.25 cgd 362: LIST_INIT(&ipq);
1.1 cgd 363: ip_id = time.tv_sec & 0xffff;
364: ipintrq.ifq_maxlen = ipqmaxlen;
1.22 mycroft 365: TAILQ_INIT(&in_ifaddr);
1.120 ad 366: in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IFADDR,
367: M_WAITOK, &in_ifaddrhash);
1.166 matt 368: in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IPMADDR,
369: M_WAITOK, &in_multihash);
1.160 itojun 370: ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout);
1.73 thorpej 371: #ifdef GATEWAY
372: ipflow_init();
373: #endif
1.121 thorpej 374:
375: #ifdef PFIL_HOOKS
376: /* Register our Packet Filter hook. */
1.126 thorpej 377: inet_pfil_hook.ph_type = PFIL_TYPE_AF;
378: inet_pfil_hook.ph_af = AF_INET;
1.121 thorpej 379: i = pfil_head_register(&inet_pfil_hook);
380: if (i != 0)
381: printf("ip_init: WARNING: unable to register pfil hook, "
382: "error %d\n", i);
383: #endif /* PFIL_HOOKS */
1.135 thorpej 384:
385: #ifdef INET_CSUM_COUNTERS
386: evcnt_attach_static(&ip_hwcsum_bad);
387: evcnt_attach_static(&ip_hwcsum_ok);
388: evcnt_attach_static(&ip_swcsum);
389: #endif /* INET_CSUM_COUNTERS */
1.164 matt 390:
391: #ifdef MBUFTRACE
392: MOWNER_ATTACH(&ip_tx_mowner);
393: MOWNER_ATTACH(&ip_rx_mowner);
394: #endif /* MBUFTRACE */
1.1 cgd 395: }
396:
397: struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
398: struct route ipforward_rt;
399:
400: /*
1.89 itojun 401: * IP software interrupt routine
402: */
403: void
404: ipintr()
405: {
406: int s;
407: struct mbuf *m;
408:
409: while (1) {
1.132 thorpej 410: s = splnet();
1.89 itojun 411: IF_DEQUEUE(&ipintrq, m);
412: splx(s);
413: if (m == 0)
414: return;
1.164 matt 415: MCLAIM(m, &ip_rx_mowner);
1.89 itojun 416: ip_input(m);
417: }
418: }
419:
420: /*
1.1 cgd 421: * Ip input routine. Checksum and byte swap header. If fragmented
422: * try to reassemble. Process options. Pass to next level.
423: */
1.8 mycroft 424: void
1.89 itojun 425: ip_input(struct mbuf *m)
1.1 cgd 426: {
1.109 augustss 427: struct ip *ip = NULL;
428: struct ipq *fp;
429: struct in_ifaddr *ia;
430: struct ifaddr *ifa;
1.25 cgd 431: struct ipqent *ipqe;
1.89 itojun 432: int hlen = 0, mff, len;
1.100 itojun 433: int downmatch;
1.165 christos 434: int checkif;
1.169 itojun 435: int srcrt = 0;
1.173 jonathan 436: #ifdef FAST_IPSEC
437: struct m_tag *mtag;
438: struct tdb_ident *tdbi;
439: struct secpolicy *sp;
440: int s, error;
441: #endif /* FAST_IPSEC */
1.1 cgd 442:
1.164 matt 443: MCLAIM(m, &ip_rx_mowner);
1.1 cgd 444: #ifdef DIAGNOSTIC
445: if ((m->m_flags & M_PKTHDR) == 0)
446: panic("ipintr no HDR");
447: #endif
1.89 itojun 448: #ifdef IPSEC
449: /*
450: * should the inner packet be considered authentic?
451: * see comment in ah4_input().
452: */
453: if (m) {
454: m->m_flags &= ~M_AUTHIPHDR;
455: m->m_flags &= ~M_AUTHIPDGM;
456: }
457: #endif
1.164 matt 458:
1.1 cgd 459: /*
460: * If no IP addresses have been set yet but the interfaces
461: * are receiving, can't do anything with incoming packets yet.
462: */
1.140 matt 463: if (TAILQ_FIRST(&in_ifaddr) == 0)
1.1 cgd 464: goto bad;
465: ipstat.ips_total++;
1.154 thorpej 466: /*
467: * If the IP header is not aligned, slurp it up into a new
468: * mbuf with space for link headers, in the event we forward
469: * it. Otherwise, if it is aligned, make sure the entire
470: * base IP header is in the first mbuf of the chain.
471: */
472: if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
473: if ((m = m_copyup(m, sizeof(struct ip),
474: (max_linkhdr + 3) & ~3)) == NULL) {
475: /* XXXJRT new stat, please */
476: ipstat.ips_toosmall++;
477: return;
478: }
479: } else if (__predict_false(m->m_len < sizeof (struct ip))) {
480: if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
481: ipstat.ips_toosmall++;
482: return;
483: }
1.1 cgd 484: }
485: ip = mtod(m, struct ip *);
1.13 mycroft 486: if (ip->ip_v != IPVERSION) {
487: ipstat.ips_badvers++;
488: goto bad;
489: }
1.1 cgd 490: hlen = ip->ip_hl << 2;
491: if (hlen < sizeof(struct ip)) { /* minimum header length */
492: ipstat.ips_badhlen++;
493: goto bad;
494: }
495: if (hlen > m->m_len) {
496: if ((m = m_pullup(m, hlen)) == 0) {
497: ipstat.ips_badhlen++;
1.89 itojun 498: return;
1.1 cgd 499: }
500: ip = mtod(m, struct ip *);
501: }
1.98 thorpej 502:
1.85 hwr 503: /*
1.99 thorpej 504: * RFC1122: packets with a multicast source address are
1.98 thorpej 505: * not allowed.
1.85 hwr 506: */
507: if (IN_MULTICAST(ip->ip_src.s_addr)) {
1.130 itojun 508: ipstat.ips_badaddr++;
1.85 hwr 509: goto bad;
1.129 itojun 510: }
511:
512: /* 127/8 must not appear on wire - RFC1122 */
513: if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
514: (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
1.130 itojun 515: if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
516: ipstat.ips_badaddr++;
1.129 itojun 517: goto bad;
1.130 itojun 518: }
1.85 hwr 519: }
520:
1.135 thorpej 521: switch (m->m_pkthdr.csum_flags &
1.137 thorpej 522: ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) |
1.135 thorpej 523: M_CSUM_IPv4_BAD)) {
524: case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
525: INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
526: goto badcsum;
527:
528: case M_CSUM_IPv4:
529: /* Checksum was okay. */
530: INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
531: break;
532:
533: default:
534: /* Must compute it ourselves. */
535: INET_CSUM_COUNTER_INCR(&ip_swcsum);
536: if (in_cksum(m, hlen) != 0)
537: goto bad;
538: break;
1.1 cgd 539: }
540:
1.121 thorpej 541: /* Retrieve the packet length. */
542: len = ntohs(ip->ip_len);
1.81 proff 543:
544: /*
545: * Check for additional length bogosity
546: */
1.84 proff 547: if (len < hlen) {
1.81 proff 548: ipstat.ips_badlen++;
549: goto bad;
550: }
1.1 cgd 551:
552: /*
553: * Check that the amount of data in the buffers
554: * is as at least much as the IP header would have us expect.
555: * Trim mbufs if longer than we expect.
556: * Drop packet if shorter than we expect.
557: */
1.35 mycroft 558: if (m->m_pkthdr.len < len) {
1.1 cgd 559: ipstat.ips_tooshort++;
560: goto bad;
561: }
1.35 mycroft 562: if (m->m_pkthdr.len > len) {
1.1 cgd 563: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 564: m->m_len = len;
565: m->m_pkthdr.len = len;
1.1 cgd 566: } else
1.35 mycroft 567: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 568: }
569:
1.94 itojun 570: #ifdef IPSEC
1.149 wiz 571: /* ipflow (IP fast forwarding) is not compatible with IPsec. */
1.94 itojun 572: m->m_flags &= ~M_CANFASTFWD;
573: #else
1.64 thorpej 574: /*
575: * Assume that we can create a fast-forward IP flow entry
576: * based on this packet.
577: */
578: m->m_flags |= M_CANFASTFWD;
1.94 itojun 579: #endif
1.64 thorpej 580:
1.36 mrg 581: #ifdef PFIL_HOOKS
1.33 mrg 582: /*
1.64 thorpej 583: * Run through list of hooks for input packets. If there are any
584: * filters which require that additional packets in the flow are
585: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
586: * Note that filters must _never_ set this flag, as another filter
587: * in the list may have previously cleared it.
1.33 mrg 588: */
1.127 itojun 589: /*
590: * let ipfilter look at packet on the wire,
591: * not the decapsulated packet.
592: */
593: #ifdef IPSEC
1.136 itojun 594: if (!ipsec_getnhist(m))
1.127 itojun 595: #else
596: if (1)
597: #endif
598: {
1.169 itojun 599: struct in_addr odst;
600:
601: odst = ip->ip_dst;
1.127 itojun 602: if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif,
1.168 itojun 603: PFIL_IN) != 0)
604: return;
1.127 itojun 605: if (m == NULL)
606: return;
607: ip = mtod(m, struct ip *);
1.142 darrenr 608: hlen = ip->ip_hl << 2;
1.169 itojun 609: srcrt = (odst.s_addr != ip->ip_dst.s_addr);
1.127 itojun 610: }
1.36 mrg 611: #endif /* PFIL_HOOKS */
1.123 thorpej 612:
613: #ifdef ALTQ
614: /* XXX Temporary until ALTQ is changed to use a pfil hook */
615: if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) {
616: /* packet dropped by traffic conditioner */
617: return;
618: }
619: #endif
1.121 thorpej 620:
621: /*
1.1 cgd 622: * Process options and, if not destined for us,
623: * ship it on. ip_dooptions returns 1 when an
624: * error was detected (causing an icmp message
625: * to be sent and the original packet to be freed).
626: */
627: ip_nhops = 0; /* for source routed packets */
628: if (hlen > sizeof (struct ip) && ip_dooptions(m))
1.89 itojun 629: return;
1.1 cgd 630:
631: /*
1.165 christos 632: * Enable a consistency check between the destination address
633: * and the arrival interface for a unicast packet (the RFC 1122
634: * strong ES model) if IP forwarding is disabled and the packet
635: * is not locally generated.
636: *
637: * XXX - Checking also should be disabled if the destination
638: * address is ipnat'ed to a different interface.
639: *
640: * XXX - Checking is incompatible with IP aliases added
641: * to the loopback interface instead of the interface where
642: * the packets are received.
643: *
644: * XXX - We need to add a per ifaddr flag for this so that
645: * we get finer grain control.
646: */
647: checkif = ip_checkinterface && (ipforwarding == 0) &&
648: (m->m_pkthdr.rcvif != NULL) &&
649: ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0);
650:
651: /*
1.1 cgd 652: * Check our list of addresses, to see if the packet is for us.
1.100 itojun 653: *
654: * Traditional 4.4BSD did not consult IFF_UP at all.
655: * The behavior here is to treat addresses on !IFF_UP interface
656: * as not mine.
1.1 cgd 657: */
1.100 itojun 658: downmatch = 0;
1.140 matt 659: LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) {
1.97 itojun 660: if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
1.165 christos 661: if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif)
662: continue;
1.97 itojun 663: if ((ia->ia_ifp->if_flags & IFF_UP) != 0)
664: break;
1.100 itojun 665: else
666: downmatch++;
1.97 itojun 667: }
668: }
1.86 thorpej 669: if (ia != NULL)
670: goto ours;
1.57 tls 671: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
1.140 matt 672: TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) {
673: if (ifa->ifa_addr->sa_family != AF_INET)
674: continue;
1.57 tls 675: ia = ifatoia(ifa);
1.35 mycroft 676: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
677: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 678: /*
679: * Look for all-0's host part (old broadcast addr),
680: * either for subnet or net.
681: */
682: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 683: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 684: goto ours;
1.57 tls 685: /*
686: * An interface with IP address zero accepts
687: * all packets that arrive on that interface.
688: */
689: if (in_nullhost(ia->ia_addr.sin_addr))
690: goto ours;
1.1 cgd 691: }
692: }
1.18 mycroft 693: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 694: struct in_multi *inm;
695: #ifdef MROUTING
696: extern struct socket *ip_mrouter;
1.10 brezak 697:
1.147 matt 698: if (M_READONLY(m)) {
1.10 brezak 699: if ((m = m_pullup(m, hlen)) == 0) {
700: ipstat.ips_toosmall++;
1.89 itojun 701: return;
1.10 brezak 702: }
703: ip = mtod(m, struct ip *);
704: }
1.4 hpeyerl 705:
706: if (ip_mrouter) {
707: /*
708: * If we are acting as a multicast router, all
709: * incoming multicast packets are passed to the
710: * kernel-level multicast forwarding function.
711: * The packet is returned (relatively) intact; if
712: * ip_mforward() returns a non-zero value, the packet
713: * must be discarded, else it may be accepted below.
714: *
715: * (The IP ident field is put in the same byte order
716: * as expected when ip_mforward() is called from
717: * ip_output().)
718: */
1.13 mycroft 719: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
720: ipstat.ips_cantforward++;
1.4 hpeyerl 721: m_freem(m);
1.89 itojun 722: return;
1.4 hpeyerl 723: }
724:
725: /*
726: * The process-level routing demon needs to receive
727: * all multicast IGMP packets, whether or not this
728: * host belongs to their destination groups.
729: */
730: if (ip->ip_p == IPPROTO_IGMP)
731: goto ours;
1.13 mycroft 732: ipstat.ips_forward++;
1.4 hpeyerl 733: }
734: #endif
735: /*
736: * See if we belong to the destination multicast group on the
737: * arrival interface.
738: */
739: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
740: if (inm == NULL) {
1.13 mycroft 741: ipstat.ips_cantforward++;
1.4 hpeyerl 742: m_freem(m);
1.89 itojun 743: return;
1.4 hpeyerl 744: }
745: goto ours;
746: }
1.19 mycroft 747: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 748: in_nullhost(ip->ip_dst))
1.1 cgd 749: goto ours;
750:
751: /*
752: * Not for us; forward if possible and desirable.
753: */
754: if (ipforwarding == 0) {
755: ipstat.ips_cantforward++;
756: m_freem(m);
1.100 itojun 757: } else {
758: /*
759: * If ip_dst matched any of my address on !IFF_UP interface,
760: * and there's no IFF_UP interface that matches ip_dst,
761: * send icmp unreach. Forwarding it will result in in-kernel
762: * forwarding loop till TTL goes to 0.
763: */
764: if (downmatch) {
765: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
766: ipstat.ips_cantforward++;
767: return;
768: }
1.145 itojun 769: #ifdef IPSEC
770: if (ipsec4_in_reject(m, NULL)) {
771: ipsecstat.in_polvio++;
772: goto bad;
773: }
774: #endif
1.173 jonathan 775: #ifdef FAST_IPSEC
776: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
777: s = splsoftnet();
778: if (mtag != NULL) {
779: tdbi = (struct tdb_ident *)(mtag + 1);
780: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
781: } else {
782: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
783: IP_FORWARDING, &error);
784: }
785: if (sp == NULL) { /* NB: can happen if error */
786: splx(s);
787: /*XXX error stat???*/
788: DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/
789: goto bad;
790: }
791:
792: /*
793: * Check security policy against packet attributes.
794: */
795: error = ipsec_in_reject(sp, m);
796: KEY_FREESP(&sp);
797: splx(s);
798: if (error) {
799: ipstat.ips_cantforward++;
800: goto bad;
801: }
802: #endif /* FAST_IPSEC */
1.145 itojun 803:
1.169 itojun 804: ip_forward(m, srcrt);
1.100 itojun 805: }
1.89 itojun 806: return;
1.1 cgd 807:
808: ours:
809: /*
810: * If offset or IP_MF are set, must reassemble.
811: * Otherwise, nothing need be done.
812: * (We could look in the reassembly queue to see
813: * if the packet was previously fragmented,
814: * but it's not worth the time; just let them time out.)
815: */
1.155 itojun 816: if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
817: if (M_READONLY(m)) {
818: if ((m = m_pullup(m, hlen)) == NULL) {
819: ipstat.ips_toosmall++;
820: goto bad;
821: }
822: ip = mtod(m, struct ip *);
823: }
824:
1.1 cgd 825: /*
826: * Look for queue of fragments
827: * of this datagram.
828: */
1.75 thorpej 829: IPQ_LOCK();
1.140 matt 830: LIST_FOREACH(fp, &ipq, ipq_q)
1.1 cgd 831: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 832: in_hosteq(ip->ip_src, fp->ipq_src) &&
833: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 834: ip->ip_p == fp->ipq_p)
835: goto found;
836: fp = 0;
837: found:
838:
839: /*
840: * Adjust ip_len to not reflect header,
1.25 cgd 841: * set ipqe_mff if more fragments are expected,
1.1 cgd 842: * convert offset of this to bytes.
843: */
1.155 itojun 844: ip->ip_len = htons(ntohs(ip->ip_len) - hlen);
845: mff = (ip->ip_off & htons(IP_MF)) != 0;
1.25 cgd 846: if (mff) {
1.16 cgd 847: /*
848: * Make sure that fragments have a data length
849: * that's a non-zero multiple of 8 bytes.
850: */
1.155 itojun 851: if (ntohs(ip->ip_len) == 0 ||
852: (ntohs(ip->ip_len) & 0x7) != 0) {
1.16 cgd 853: ipstat.ips_badfrags++;
1.75 thorpej 854: IPQ_UNLOCK();
1.16 cgd 855: goto bad;
856: }
857: }
1.155 itojun 858: ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3);
1.1 cgd 859:
860: /*
861: * If datagram marked as having more fragments
862: * or if this is not the first fragment,
863: * attempt reassembly; if it succeeds, proceed.
864: */
1.155 itojun 865: if (mff || ip->ip_off != htons(0)) {
1.1 cgd 866: ipstat.ips_fragments++;
1.72 thorpej 867: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.25 cgd 868: if (ipqe == NULL) {
869: ipstat.ips_rcvmemdrop++;
1.75 thorpej 870: IPQ_UNLOCK();
1.25 cgd 871: goto bad;
872: }
873: ipqe->ipqe_mff = mff;
1.50 thorpej 874: ipqe->ipqe_m = m;
1.25 cgd 875: ipqe->ipqe_ip = ip;
1.50 thorpej 876: m = ip_reass(ipqe, fp);
1.75 thorpej 877: if (m == 0) {
878: IPQ_UNLOCK();
1.89 itojun 879: return;
1.75 thorpej 880: }
1.13 mycroft 881: ipstat.ips_reassembled++;
1.50 thorpej 882: ip = mtod(m, struct ip *);
1.74 thorpej 883: hlen = ip->ip_hl << 2;
1.155 itojun 884: ip->ip_len = htons(ntohs(ip->ip_len) + hlen);
1.1 cgd 885: } else
886: if (fp)
887: ip_freef(fp);
1.75 thorpej 888: IPQ_UNLOCK();
1.79 mycroft 889: }
1.128 itojun 890:
1.173 jonathan 891: #if defined(IPSEC)
1.128 itojun 892: /*
893: * enforce IPsec policy checking if we are seeing last header.
894: * note that we do not visit this with protocols with pcb layer
895: * code - like udp/tcp/raw ip.
896: */
897: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 &&
898: ipsec4_in_reject(m, NULL)) {
899: ipsecstat.in_polvio++;
900: goto bad;
901: }
902: #endif
1.173 jonathan 903: #if FAST_IPSEC
904: /*
905: * enforce IPsec policy checking if we are seeing last header.
906: * note that we do not visit this with protocols with pcb layer
907: * code - like udp/tcp/raw ip.
908: */
909: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) {
910: /*
911: * Check if the packet has already had IPsec processing
912: * done. If so, then just pass it along. This tag gets
913: * set during AH, ESP, etc. input handling, before the
914: * packet is returned to the ip input queue for delivery.
915: */
916: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
917: s = splsoftnet();
918: if (mtag != NULL) {
919: tdbi = (struct tdb_ident *)(mtag + 1);
920: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
921: } else {
922: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
923: IP_FORWARDING, &error);
924: }
925: if (sp != NULL) {
926: /*
927: * Check security policy against packet attributes.
928: */
929: error = ipsec_in_reject(sp, m);
930: KEY_FREESP(&sp);
931: } else {
932: /* XXX error stat??? */
933: error = EINVAL;
934: DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/
935: goto bad;
936: }
937: splx(s);
938: if (error)
939: goto bad;
940: }
941: #endif /* FAST_IPSEC */
1.1 cgd 942:
943: /*
944: * Switch out to protocol's input routine.
945: */
1.82 aidan 946: #if IFA_STATS
1.122 itojun 947: if (ia && ip)
1.155 itojun 948: ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
1.82 aidan 949: #endif
1.1 cgd 950: ipstat.ips_delivered++;
1.89 itojun 951: {
952: int off = hlen, nh = ip->ip_p;
953:
954: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
955: return;
956: }
1.1 cgd 957: bad:
958: m_freem(m);
1.135 thorpej 959: return;
960:
961: badcsum:
962: ipstat.ips_badsum++;
963: m_freem(m);
1.1 cgd 964: }
965:
966: /*
967: * Take incoming datagram fragment and try to
968: * reassemble it into whole datagram. If a chain for
969: * reassembly of this datagram already exists, then it
970: * is given as fp; otherwise have to make a chain.
971: */
1.50 thorpej 972: struct mbuf *
1.25 cgd 973: ip_reass(ipqe, fp)
1.109 augustss 974: struct ipqent *ipqe;
975: struct ipq *fp;
1.1 cgd 976: {
1.109 augustss 977: struct mbuf *m = ipqe->ipqe_m;
978: struct ipqent *nq, *p, *q;
1.25 cgd 979: struct ip *ip;
1.1 cgd 980: struct mbuf *t;
1.25 cgd 981: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 982: int i, next;
983:
1.75 thorpej 984: IPQ_LOCK_CHECK();
985:
1.1 cgd 986: /*
987: * Presence of header sizes in mbufs
988: * would confuse code below.
989: */
990: m->m_data += hlen;
991: m->m_len -= hlen;
992:
993: /*
994: * If first fragment to arrive, create a reassembly queue.
995: */
996: if (fp == 0) {
1.131 itojun 997: /*
998: * Enforce upper bound on number of fragmented packets
999: * for which we attempt reassembly;
1000: * If maxfrag is 0, never accept fragments.
1001: * If maxfrag is -1, accept all fragments without limitation.
1002: */
1003: if (ip_maxfragpackets < 0)
1004: ;
1005: else if (ip_nfragpackets >= ip_maxfragpackets)
1006: goto dropfrag;
1007: ip_nfragpackets++;
1.50 thorpej 1008: MALLOC(fp, struct ipq *, sizeof (struct ipq),
1009: M_FTABLE, M_NOWAIT);
1010: if (fp == NULL)
1.1 cgd 1011: goto dropfrag;
1.25 cgd 1012: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 1013: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 1014: fp->ipq_p = ipqe->ipqe_ip->ip_p;
1015: fp->ipq_id = ipqe->ipqe_ip->ip_id;
1.148 matt 1016: TAILQ_INIT(&fp->ipq_fragq);
1.25 cgd 1017: fp->ipq_src = ipqe->ipqe_ip->ip_src;
1018: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
1019: p = NULL;
1.1 cgd 1020: goto insert;
1021: }
1022:
1023: /*
1024: * Find a segment which begins after this one does.
1025: */
1.148 matt 1026: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1027: p = q, q = TAILQ_NEXT(q, ipqe_q))
1.155 itojun 1028: if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off))
1.1 cgd 1029: break;
1030:
1031: /*
1032: * If there is a preceding segment, it may provide some of
1033: * our data already. If so, drop the data from the incoming
1034: * segment. If it provides all of our data, drop us.
1035: */
1.25 cgd 1036: if (p != NULL) {
1.155 itojun 1037: i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) -
1038: ntohs(ipqe->ipqe_ip->ip_off);
1.1 cgd 1039: if (i > 0) {
1.155 itojun 1040: if (i >= ntohs(ipqe->ipqe_ip->ip_len))
1.1 cgd 1041: goto dropfrag;
1.50 thorpej 1042: m_adj(ipqe->ipqe_m, i);
1.155 itojun 1043: ipqe->ipqe_ip->ip_off =
1044: htons(ntohs(ipqe->ipqe_ip->ip_off) + i);
1045: ipqe->ipqe_ip->ip_len =
1046: htons(ntohs(ipqe->ipqe_ip->ip_len) - i);
1.1 cgd 1047: }
1048: }
1049:
1050: /*
1051: * While we overlap succeeding segments trim them or,
1052: * if they are completely covered, dequeue them.
1053: */
1.155 itojun 1054: for (; q != NULL &&
1055: ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) >
1056: ntohs(q->ipqe_ip->ip_off); q = nq) {
1057: i = (ntohs(ipqe->ipqe_ip->ip_off) +
1058: ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off);
1059: if (i < ntohs(q->ipqe_ip->ip_len)) {
1060: q->ipqe_ip->ip_len =
1061: htons(ntohs(q->ipqe_ip->ip_len) - i);
1062: q->ipqe_ip->ip_off =
1063: htons(ntohs(q->ipqe_ip->ip_off) + i);
1.50 thorpej 1064: m_adj(q->ipqe_m, i);
1.1 cgd 1065: break;
1066: }
1.148 matt 1067: nq = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1068: m_freem(q->ipqe_m);
1.148 matt 1069: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.72 thorpej 1070: pool_put(&ipqent_pool, q);
1.1 cgd 1071: }
1072:
1073: insert:
1074: /*
1075: * Stick new segment in its place;
1076: * check for complete reassembly.
1077: */
1.25 cgd 1078: if (p == NULL) {
1.148 matt 1079: TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
1.25 cgd 1080: } else {
1.148 matt 1081: TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q);
1.25 cgd 1082: }
1.1 cgd 1083: next = 0;
1.148 matt 1084: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1085: p = q, q = TAILQ_NEXT(q, ipqe_q)) {
1.155 itojun 1086: if (ntohs(q->ipqe_ip->ip_off) != next)
1.1 cgd 1087: return (0);
1.155 itojun 1088: next += ntohs(q->ipqe_ip->ip_len);
1.1 cgd 1089: }
1.25 cgd 1090: if (p->ipqe_mff)
1.1 cgd 1091: return (0);
1092:
1093: /*
1.41 thorpej 1094: * Reassembly is complete. Check for a bogus message size and
1095: * concatenate fragments.
1.1 cgd 1096: */
1.148 matt 1097: q = TAILQ_FIRST(&fp->ipq_fragq);
1.25 cgd 1098: ip = q->ipqe_ip;
1.41 thorpej 1099: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
1100: ipstat.ips_toolong++;
1101: ip_freef(fp);
1102: return (0);
1103: }
1.50 thorpej 1104: m = q->ipqe_m;
1.1 cgd 1105: t = m->m_next;
1106: m->m_next = 0;
1107: m_cat(m, t);
1.148 matt 1108: nq = TAILQ_NEXT(q, ipqe_q);
1.72 thorpej 1109: pool_put(&ipqent_pool, q);
1.25 cgd 1110: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 1111: t = q->ipqe_m;
1.148 matt 1112: nq = TAILQ_NEXT(q, ipqe_q);
1.72 thorpej 1113: pool_put(&ipqent_pool, q);
1.1 cgd 1114: m_cat(m, t);
1115: }
1116:
1117: /*
1118: * Create header for new ip packet by
1119: * modifying header of first packet;
1120: * dequeue and discard fragment reassembly header.
1121: * Make header visible.
1122: */
1.155 itojun 1123: ip->ip_len = htons(next);
1.25 cgd 1124: ip->ip_src = fp->ipq_src;
1125: ip->ip_dst = fp->ipq_dst;
1126: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1127: FREE(fp, M_FTABLE);
1.131 itojun 1128: ip_nfragpackets--;
1.1 cgd 1129: m->m_len += (ip->ip_hl << 2);
1130: m->m_data -= (ip->ip_hl << 2);
1131: /* some debugging cruft by sklower, below, will go away soon */
1132: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
1.109 augustss 1133: int plen = 0;
1.50 thorpej 1134: for (t = m; t; t = t->m_next)
1135: plen += t->m_len;
1136: m->m_pkthdr.len = plen;
1.1 cgd 1137: }
1.50 thorpej 1138: return (m);
1.1 cgd 1139:
1140: dropfrag:
1141: ipstat.ips_fragdropped++;
1142: m_freem(m);
1.72 thorpej 1143: pool_put(&ipqent_pool, ipqe);
1.1 cgd 1144: return (0);
1145: }
1146:
1147: /*
1148: * Free a fragment reassembly header and all
1149: * associated datagrams.
1150: */
1.8 mycroft 1151: void
1.1 cgd 1152: ip_freef(fp)
1153: struct ipq *fp;
1154: {
1.109 augustss 1155: struct ipqent *q, *p;
1.1 cgd 1156:
1.75 thorpej 1157: IPQ_LOCK_CHECK();
1158:
1.148 matt 1159: for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) {
1160: p = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1161: m_freem(q->ipqe_m);
1.148 matt 1162: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.72 thorpej 1163: pool_put(&ipqent_pool, q);
1.1 cgd 1164: }
1.25 cgd 1165: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1166: FREE(fp, M_FTABLE);
1.131 itojun 1167: ip_nfragpackets--;
1.1 cgd 1168: }
1169:
1170: /*
1171: * IP timer processing;
1172: * if a timer expires on a reassembly
1173: * queue, discard it.
1174: */
1.8 mycroft 1175: void
1.1 cgd 1176: ip_slowtimo()
1177: {
1.109 augustss 1178: struct ipq *fp, *nfp;
1.24 mycroft 1179: int s = splsoftnet();
1.1 cgd 1180:
1.75 thorpej 1181: IPQ_LOCK();
1.140 matt 1182: for (fp = LIST_FIRST(&ipq); fp != NULL; fp = nfp) {
1183: nfp = LIST_NEXT(fp, ipq_q);
1.25 cgd 1184: if (--fp->ipq_ttl == 0) {
1.1 cgd 1185: ipstat.ips_fragtimeout++;
1.25 cgd 1186: ip_freef(fp);
1.1 cgd 1187: }
1188: }
1.131 itojun 1189: /*
1190: * If we are over the maximum number of fragments
1191: * (due to the limit being lowered), drain off
1192: * enough to get down to the new limit.
1193: */
1194: if (ip_maxfragpackets < 0)
1195: ;
1196: else {
1.140 matt 1197: while (ip_nfragpackets > ip_maxfragpackets && LIST_FIRST(&ipq))
1198: ip_freef(LIST_FIRST(&ipq));
1.131 itojun 1199: }
1.75 thorpej 1200: IPQ_UNLOCK();
1.63 matt 1201: #ifdef GATEWAY
1202: ipflow_slowtimo();
1203: #endif
1.1 cgd 1204: splx(s);
1205: }
1206:
1207: /*
1208: * Drain off all datagram fragments.
1209: */
1.8 mycroft 1210: void
1.1 cgd 1211: ip_drain()
1212: {
1213:
1.75 thorpej 1214: /*
1215: * We may be called from a device's interrupt context. If
1216: * the ipq is already busy, just bail out now.
1217: */
1218: if (ipq_lock_try() == 0)
1219: return;
1220:
1.140 matt 1221: while (LIST_FIRST(&ipq) != NULL) {
1.1 cgd 1222: ipstat.ips_fragdropped++;
1.140 matt 1223: ip_freef(LIST_FIRST(&ipq));
1.1 cgd 1224: }
1.75 thorpej 1225:
1226: IPQ_UNLOCK();
1.1 cgd 1227: }
1228:
1229: /*
1230: * Do option processing on a datagram,
1231: * possibly discarding it if bad options are encountered,
1232: * or forwarding it if source-routed.
1233: * Returns 1 if packet has been forwarded/freed,
1234: * 0 if the packet should be processed further.
1235: */
1.8 mycroft 1236: int
1.1 cgd 1237: ip_dooptions(m)
1238: struct mbuf *m;
1239: {
1.109 augustss 1240: struct ip *ip = mtod(m, struct ip *);
1241: u_char *cp, *cp0;
1242: struct ip_timestamp *ipt;
1243: struct in_ifaddr *ia;
1.1 cgd 1244: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 1245: struct in_addr dst;
1.1 cgd 1246: n_time ntime;
1247:
1.13 mycroft 1248: dst = ip->ip_dst;
1.1 cgd 1249: cp = (u_char *)(ip + 1);
1250: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
1251: for (; cnt > 0; cnt -= optlen, cp += optlen) {
1252: opt = cp[IPOPT_OPTVAL];
1253: if (opt == IPOPT_EOL)
1254: break;
1255: if (opt == IPOPT_NOP)
1256: optlen = 1;
1257: else {
1.113 itojun 1258: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1259: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1260: goto bad;
1261: }
1.1 cgd 1262: optlen = cp[IPOPT_OLEN];
1.114 itojun 1263: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 1264: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1265: goto bad;
1266: }
1267: }
1268: switch (opt) {
1269:
1270: default:
1271: break;
1272:
1273: /*
1274: * Source routing with record.
1275: * Find interface with current destination address.
1276: * If none on this machine then drop if strictly routed,
1277: * or do nothing if loosely routed.
1278: * Record interface address and bring up next address
1279: * component. If strictly routed make sure next
1280: * address is on directly accessible net.
1281: */
1282: case IPOPT_LSRR:
1283: case IPOPT_SSRR:
1.47 cjs 1284: if (ip_allowsrcrt == 0) {
1285: type = ICMP_UNREACH;
1286: code = ICMP_UNREACH_NET_PROHIB;
1287: goto bad;
1288: }
1.114 itojun 1289: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1290: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1291: goto bad;
1292: }
1.1 cgd 1293: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1294: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1295: goto bad;
1296: }
1297: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 1298: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 1299: if (ia == 0) {
1300: if (opt == IPOPT_SSRR) {
1301: type = ICMP_UNREACH;
1302: code = ICMP_UNREACH_SRCFAIL;
1303: goto bad;
1304: }
1305: /*
1306: * Loose routing, and not at next destination
1307: * yet; nothing to do except forward.
1308: */
1309: break;
1310: }
1311: off--; /* 0 origin */
1.112 sommerfe 1312: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 1313: /*
1314: * End of source route. Should be for us.
1315: */
1316: save_rte(cp, ip->ip_src);
1317: break;
1318: }
1319: /*
1320: * locate outgoing interface
1321: */
1322: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
1323: sizeof(ipaddr.sin_addr));
1.96 thorpej 1324: if (opt == IPOPT_SSRR)
1325: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1326: else
1.1 cgd 1327: ia = ip_rtaddr(ipaddr.sin_addr);
1328: if (ia == 0) {
1329: type = ICMP_UNREACH;
1330: code = ICMP_UNREACH_SRCFAIL;
1331: goto bad;
1332: }
1333: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 1334: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 1335: (caddr_t)(cp + off), sizeof(struct in_addr));
1336: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1337: /*
1338: * Let ip_intr's mcast routing check handle mcast pkts
1339: */
1.18 mycroft 1340: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1341: break;
1342:
1343: case IPOPT_RR:
1.114 itojun 1344: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1345: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1346: goto bad;
1347: }
1.1 cgd 1348: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1349: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1350: goto bad;
1351: }
1352: /*
1353: * If no space remains, ignore.
1354: */
1355: off--; /* 0 origin */
1.112 sommerfe 1356: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1357: break;
1358: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
1359: sizeof(ipaddr.sin_addr));
1360: /*
1361: * locate outgoing interface; if we're the destination,
1362: * use the incoming interface (should be same).
1363: */
1.96 thorpej 1364: if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))))
1365: == NULL &&
1366: (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
1.1 cgd 1367: type = ICMP_UNREACH;
1368: code = ICMP_UNREACH_HOST;
1369: goto bad;
1370: }
1.20 mycroft 1371: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 1372: (caddr_t)(cp + off), sizeof(struct in_addr));
1373: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1374: break;
1375:
1376: case IPOPT_TS:
1377: code = cp - (u_char *)ip;
1378: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1379: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1380: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1381: goto bad;
1.114 itojun 1382: }
1383: if (ipt->ipt_ptr < 5) {
1384: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1385: goto bad;
1386: }
1.15 cgd 1387: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1388: if (++ipt->ipt_oflw == 0) {
1389: code = (u_char *)&ipt->ipt_ptr -
1390: (u_char *)ip;
1.1 cgd 1391: goto bad;
1.114 itojun 1392: }
1.1 cgd 1393: break;
1394: }
1.104 thorpej 1395: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1396: switch (ipt->ipt_flg) {
1397:
1398: case IPOPT_TS_TSONLY:
1399: break;
1400:
1401: case IPOPT_TS_TSANDADDR:
1.66 thorpej 1402: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1403: sizeof(struct in_addr) > ipt->ipt_len) {
1404: code = (u_char *)&ipt->ipt_ptr -
1405: (u_char *)ip;
1.1 cgd 1406: goto bad;
1.114 itojun 1407: }
1.13 mycroft 1408: ipaddr.sin_addr = dst;
1.96 thorpej 1409: ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr),
1410: m->m_pkthdr.rcvif));
1.13 mycroft 1411: if (ia == 0)
1412: continue;
1.104 thorpej 1413: bcopy(&ia->ia_addr.sin_addr,
1414: cp0, sizeof(struct in_addr));
1.1 cgd 1415: ipt->ipt_ptr += sizeof(struct in_addr);
1416: break;
1417:
1418: case IPOPT_TS_PRESPEC:
1.66 thorpej 1419: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1420: sizeof(struct in_addr) > ipt->ipt_len) {
1421: code = (u_char *)&ipt->ipt_ptr -
1422: (u_char *)ip;
1.1 cgd 1423: goto bad;
1.114 itojun 1424: }
1.104 thorpej 1425: bcopy(cp0, &ipaddr.sin_addr,
1.1 cgd 1426: sizeof(struct in_addr));
1.96 thorpej 1427: if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))
1428: == NULL)
1.1 cgd 1429: continue;
1430: ipt->ipt_ptr += sizeof(struct in_addr);
1431: break;
1432:
1433: default:
1.114 itojun 1434: /* XXX can't take &ipt->ipt_flg */
1435: code = (u_char *)&ipt->ipt_ptr -
1436: (u_char *)ip + 1;
1.1 cgd 1437: goto bad;
1438: }
1439: ntime = iptime();
1.107 thorpej 1440: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1441: bcopy(cp0, (caddr_t)cp + ipt->ipt_ptr - 1,
1.1 cgd 1442: sizeof(n_time));
1443: ipt->ipt_ptr += sizeof(n_time);
1444: }
1445: }
1446: if (forward) {
1.26 thorpej 1447: if (ip_forwsrcrt == 0) {
1448: type = ICMP_UNREACH;
1449: code = ICMP_UNREACH_SRCFAIL;
1450: goto bad;
1451: }
1.1 cgd 1452: ip_forward(m, 1);
1453: return (1);
1.13 mycroft 1454: }
1455: return (0);
1.1 cgd 1456: bad:
1.13 mycroft 1457: icmp_error(m, type, code, 0, 0);
1458: ipstat.ips_badoptions++;
1.1 cgd 1459: return (1);
1460: }
1461:
1462: /*
1463: * Given address of next destination (final or next hop),
1464: * return internet address info of interface to be used to get there.
1465: */
1466: struct in_ifaddr *
1467: ip_rtaddr(dst)
1468: struct in_addr dst;
1469: {
1.109 augustss 1470: struct sockaddr_in *sin;
1.1 cgd 1471:
1.19 mycroft 1472: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1473:
1.35 mycroft 1474: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 1475: if (ipforward_rt.ro_rt) {
1476: RTFREE(ipforward_rt.ro_rt);
1477: ipforward_rt.ro_rt = 0;
1478: }
1479: sin->sin_family = AF_INET;
1480: sin->sin_len = sizeof(*sin);
1481: sin->sin_addr = dst;
1482:
1483: rtalloc(&ipforward_rt);
1484: }
1485: if (ipforward_rt.ro_rt == 0)
1486: return ((struct in_ifaddr *)0);
1.19 mycroft 1487: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 1488: }
1489:
1490: /*
1491: * Save incoming source route for use in replies,
1492: * to be picked up later by ip_srcroute if the receiver is interested.
1493: */
1.13 mycroft 1494: void
1.1 cgd 1495: save_rte(option, dst)
1496: u_char *option;
1497: struct in_addr dst;
1498: {
1499: unsigned olen;
1500:
1501: olen = option[IPOPT_OLEN];
1502: #ifdef DIAGNOSTIC
1503: if (ipprintfs)
1.39 christos 1504: printf("save_rte: olen %d\n", olen);
1.89 itojun 1505: #endif /* 0 */
1.1 cgd 1506: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1507: return;
1508: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
1509: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1510: ip_srcrt.dst = dst;
1511: }
1512:
1513: /*
1514: * Retrieve incoming source route for use in replies,
1515: * in the same form used by setsockopt.
1516: * The first hop is placed before the options, will be removed later.
1517: */
1518: struct mbuf *
1519: ip_srcroute()
1520: {
1.109 augustss 1521: struct in_addr *p, *q;
1522: struct mbuf *m;
1.1 cgd 1523:
1524: if (ip_nhops == 0)
1525: return ((struct mbuf *)0);
1526: m = m_get(M_DONTWAIT, MT_SOOPTS);
1527: if (m == 0)
1528: return ((struct mbuf *)0);
1529:
1.164 matt 1530: MCLAIM(m, &inetdomain.dom_mowner);
1.13 mycroft 1531: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1532:
1533: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1534: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1535: OPTSIZ;
1536: #ifdef DIAGNOSTIC
1537: if (ipprintfs)
1.39 christos 1538: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1539: #endif
1540:
1541: /*
1542: * First save first hop for return route
1543: */
1544: p = &ip_srcrt.route[ip_nhops - 1];
1545: *(mtod(m, struct in_addr *)) = *p--;
1546: #ifdef DIAGNOSTIC
1547: if (ipprintfs)
1.39 christos 1548: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1549: #endif
1550:
1551: /*
1552: * Copy option fields and padding (nop) to mbuf.
1553: */
1554: ip_srcrt.nop = IPOPT_NOP;
1555: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1556: bcopy((caddr_t)&ip_srcrt.nop,
1557: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1558: q = (struct in_addr *)(mtod(m, caddr_t) +
1559: sizeof(struct in_addr) + OPTSIZ);
1560: #undef OPTSIZ
1561: /*
1562: * Record return path as an IP source route,
1563: * reversing the path (pointers are now aligned).
1564: */
1565: while (p >= ip_srcrt.route) {
1566: #ifdef DIAGNOSTIC
1567: if (ipprintfs)
1.39 christos 1568: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1569: #endif
1570: *q++ = *p--;
1571: }
1572: /*
1573: * Last hop goes to final destination.
1574: */
1575: *q = ip_srcrt.dst;
1576: #ifdef DIAGNOSTIC
1577: if (ipprintfs)
1.39 christos 1578: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1579: #endif
1580: return (m);
1581: }
1582:
1583: /*
1584: * Strip out IP options, at higher
1585: * level protocol in the kernel.
1586: * Second argument is buffer to which options
1587: * will be moved, and return value is their length.
1588: * XXX should be deleted; last arg currently ignored.
1589: */
1.8 mycroft 1590: void
1.1 cgd 1591: ip_stripoptions(m, mopt)
1.109 augustss 1592: struct mbuf *m;
1.1 cgd 1593: struct mbuf *mopt;
1594: {
1.109 augustss 1595: int i;
1.1 cgd 1596: struct ip *ip = mtod(m, struct ip *);
1.109 augustss 1597: caddr_t opts;
1.1 cgd 1598: int olen;
1599:
1.79 mycroft 1600: olen = (ip->ip_hl << 2) - sizeof (struct ip);
1.1 cgd 1601: opts = (caddr_t)(ip + 1);
1602: i = m->m_len - (sizeof (struct ip) + olen);
1603: bcopy(opts + olen, opts, (unsigned)i);
1604: m->m_len -= olen;
1605: if (m->m_flags & M_PKTHDR)
1606: m->m_pkthdr.len -= olen;
1.155 itojun 1607: ip->ip_len = htons(ntohs(ip->ip_len) - olen);
1.79 mycroft 1608: ip->ip_hl = sizeof (struct ip) >> 2;
1.1 cgd 1609: }
1610:
1.139 matt 1611: const int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1612: 0, 0, 0, 0,
1613: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1614: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1615: EMSGSIZE, EHOSTUNREACH, 0, 0,
1616: 0, 0, 0, 0,
1617: ENOPROTOOPT
1618: };
1619:
1620: /*
1621: * Forward a packet. If some error occurs return the sender
1622: * an icmp packet. Note we can't always generate a meaningful
1623: * icmp message because icmp doesn't have a large enough repertoire
1624: * of codes and types.
1625: *
1626: * If not forwarding, just drop the packet. This could be confusing
1627: * if ipforwarding was zero but some routing protocol was advancing
1628: * us as a gateway to somewhere. However, we must let the routing
1629: * protocol deal with that.
1630: *
1631: * The srcrt parameter indicates whether the packet is being forwarded
1632: * via a source route.
1633: */
1.13 mycroft 1634: void
1.1 cgd 1635: ip_forward(m, srcrt)
1636: struct mbuf *m;
1637: int srcrt;
1638: {
1.109 augustss 1639: struct ip *ip = mtod(m, struct ip *);
1640: struct sockaddr_in *sin;
1641: struct rtentry *rt;
1.28 christos 1642: int error, type = 0, code = 0;
1.1 cgd 1643: struct mbuf *mcopy;
1.13 mycroft 1644: n_long dest;
1645: struct ifnet *destifp;
1.173 jonathan 1646: #if defined(IPSEC) || defined(FAST_IPSEC)
1.89 itojun 1647: struct ifnet dummyifp;
1648: #endif
1.164 matt 1649:
1650: /*
1651: * We are now in the output path.
1652: */
1653: MCLAIM(m, &ip_tx_mowner);
1.135 thorpej 1654:
1655: /*
1656: * Clear any in-bound checksum flags for this packet.
1657: */
1658: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1659:
1.13 mycroft 1660: dest = 0;
1.1 cgd 1661: #ifdef DIAGNOSTIC
1662: if (ipprintfs)
1.70 thorpej 1663: printf("forward: src %2.2x dst %2.2x ttl %x\n",
1664: ntohl(ip->ip_src.s_addr),
1665: ntohl(ip->ip_dst.s_addr), ip->ip_ttl);
1.1 cgd 1666: #endif
1.93 sommerfe 1667: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.1 cgd 1668: ipstat.ips_cantforward++;
1669: m_freem(m);
1670: return;
1671: }
1672: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1673: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1674: return;
1675: }
1676: ip->ip_ttl -= IPTTLDEC;
1677:
1.19 mycroft 1678: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1679: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1680: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1681: if (ipforward_rt.ro_rt) {
1682: RTFREE(ipforward_rt.ro_rt);
1683: ipforward_rt.ro_rt = 0;
1684: }
1685: sin->sin_family = AF_INET;
1.35 mycroft 1686: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1687: sin->sin_addr = ip->ip_dst;
1688:
1689: rtalloc(&ipforward_rt);
1690: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1691: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1692: return;
1693: }
1694: rt = ipforward_rt.ro_rt;
1695: }
1696:
1697: /*
1.34 mycroft 1698: * Save at most 68 bytes of the packet in case
1.1 cgd 1699: * we need to generate an ICMP message to the src.
1.119 itojun 1700: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1701: */
1.155 itojun 1702: mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1.119 itojun 1703: if (mcopy)
1704: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1705:
1706: /*
1707: * If forwarding packet using same interface that it came in on,
1708: * perhaps should send a redirect to sender to shortcut a hop.
1709: * Only send redirect if source is sending directly to us,
1710: * and if packet was not source routed (or has any options).
1711: * Also, don't send redirect if forwarding using a default route
1712: * or a route modified by a redirect.
1713: */
1714: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1715: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1716: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1717: ipsendredirects && !srcrt) {
1.19 mycroft 1718: if (rt->rt_ifa &&
1719: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1720: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1721: if (rt->rt_flags & RTF_GATEWAY)
1722: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1723: else
1724: dest = ip->ip_dst.s_addr;
1725: /*
1726: * Router requirements says to only send host
1727: * redirects.
1728: */
1729: type = ICMP_REDIRECT;
1730: code = ICMP_REDIRECT_HOST;
1.1 cgd 1731: #ifdef DIAGNOSTIC
1.77 thorpej 1732: if (ipprintfs)
1733: printf("redirect (%d) to %x\n", code,
1734: (u_int32_t)dest);
1.1 cgd 1735: #endif
1736: }
1737: }
1738:
1.89 itojun 1739: #ifdef IPSEC
1.134 lukem 1740: /* Don't lookup socket in forwarding case */
1.127 itojun 1741: (void)ipsec_setsocket(m, NULL);
1.103 itojun 1742: #endif
1.27 thorpej 1743: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1.173 jonathan 1744: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
1.174 ! itojun 1745: (struct ip_moptions *)NULL, (struct socket *)NULL);
1.173 jonathan 1746:
1.1 cgd 1747: if (error)
1748: ipstat.ips_cantforward++;
1749: else {
1750: ipstat.ips_forward++;
1751: if (type)
1752: ipstat.ips_redirectsent++;
1753: else {
1.63 matt 1754: if (mcopy) {
1755: #ifdef GATEWAY
1.64 thorpej 1756: if (mcopy->m_flags & M_CANFASTFWD)
1757: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1758: #endif
1.1 cgd 1759: m_freem(mcopy);
1.63 matt 1760: }
1.1 cgd 1761: return;
1762: }
1763: }
1764: if (mcopy == NULL)
1765: return;
1.13 mycroft 1766: destifp = NULL;
1767:
1.1 cgd 1768: switch (error) {
1769:
1770: case 0: /* forwarded, but need redirect */
1771: /* type, code set above */
1772: break;
1773:
1774: case ENETUNREACH: /* shouldn't happen, checked above */
1775: case EHOSTUNREACH:
1776: case ENETDOWN:
1777: case EHOSTDOWN:
1778: default:
1779: type = ICMP_UNREACH;
1780: code = ICMP_UNREACH_HOST;
1781: break;
1782:
1783: case EMSGSIZE:
1784: type = ICMP_UNREACH;
1785: code = ICMP_UNREACH_NEEDFRAG;
1.173 jonathan 1786: #if !defined(IPSEC) && !defined(FAST_IPSEC)
1.13 mycroft 1787: if (ipforward_rt.ro_rt)
1788: destifp = ipforward_rt.ro_rt->rt_ifp;
1.89 itojun 1789: #else
1790: /*
1791: * If the packet is routed over IPsec tunnel, tell the
1792: * originator the tunnel MTU.
1793: * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1794: * XXX quickhack!!!
1795: */
1796: if (ipforward_rt.ro_rt) {
1797: struct secpolicy *sp;
1798: int ipsecerror;
1.95 itojun 1799: size_t ipsechdr;
1.89 itojun 1800: struct route *ro;
1801:
1802: sp = ipsec4_getpolicybyaddr(mcopy,
1.170 itojun 1803: IPSEC_DIR_OUTBOUND, IP_FORWARDING,
1804: &ipsecerror);
1.89 itojun 1805:
1806: if (sp == NULL)
1807: destifp = ipforward_rt.ro_rt->rt_ifp;
1808: else {
1809: /* count IPsec header size */
1.95 itojun 1810: ipsechdr = ipsec4_hdrsiz(mcopy,
1.170 itojun 1811: IPSEC_DIR_OUTBOUND, NULL);
1.89 itojun 1812:
1813: /*
1814: * find the correct route for outer IPv4
1815: * header, compute tunnel MTU.
1816: *
1817: * XXX BUG ALERT
1818: * The "dummyifp" code relies upon the fact
1819: * that icmp_error() touches only ifp->if_mtu.
1820: */
1821: /*XXX*/
1822: destifp = NULL;
1823: if (sp->req != NULL
1.95 itojun 1824: && sp->req->sav != NULL
1825: && sp->req->sav->sah != NULL) {
1826: ro = &sp->req->sav->sah->sa_route;
1.89 itojun 1827: if (ro->ro_rt && ro->ro_rt->rt_ifp) {
1828: dummyifp.if_mtu =
1.151 itojun 1829: ro->ro_rt->rt_rmx.rmx_mtu ?
1830: ro->ro_rt->rt_rmx.rmx_mtu :
1.89 itojun 1831: ro->ro_rt->rt_ifp->if_mtu;
1832: dummyifp.if_mtu -= ipsechdr;
1833: destifp = &dummyifp;
1834: }
1835: }
1836:
1.173 jonathan 1837: #ifdef IPSEC
1.89 itojun 1838: key_freesp(sp);
1.173 jonathan 1839: #else
1840: KEY_FREESP(&sp);
1841: #endif
1.89 itojun 1842: }
1843: }
1844: #endif /*IPSEC*/
1.1 cgd 1845: ipstat.ips_cantfrag++;
1846: break;
1847:
1848: case ENOBUFS:
1.143 itojun 1849: #if 1
1850: /*
1851: * a router should not generate ICMP_SOURCEQUENCH as
1852: * required in RFC1812 Requirements for IP Version 4 Routers.
1853: * source quench could be a big problem under DoS attacks,
1.149 wiz 1854: * or if the underlying interface is rate-limited.
1.143 itojun 1855: */
1856: if (mcopy)
1857: m_freem(mcopy);
1858: return;
1859: #else
1.1 cgd 1860: type = ICMP_SOURCEQUENCH;
1861: code = 0;
1862: break;
1.143 itojun 1863: #endif
1.1 cgd 1864: }
1.13 mycroft 1865: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 1866: }
1867:
1868: void
1869: ip_savecontrol(inp, mp, ip, m)
1.109 augustss 1870: struct inpcb *inp;
1871: struct mbuf **mp;
1872: struct ip *ip;
1873: struct mbuf *m;
1.44 thorpej 1874: {
1875:
1876: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1877: struct timeval tv;
1878:
1879: microtime(&tv);
1880: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1881: SCM_TIMESTAMP, SOL_SOCKET);
1882: if (*mp)
1883: mp = &(*mp)->m_next;
1884: }
1885: if (inp->inp_flags & INP_RECVDSTADDR) {
1886: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1887: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1888: if (*mp)
1889: mp = &(*mp)->m_next;
1890: }
1891: #ifdef notyet
1892: /*
1893: * XXX
1894: * Moving these out of udp_input() made them even more broken
1895: * than they already were.
1896: * - fenner@parc.xerox.com
1897: */
1898: /* options were tossed already */
1899: if (inp->inp_flags & INP_RECVOPTS) {
1900: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1901: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1902: if (*mp)
1903: mp = &(*mp)->m_next;
1904: }
1905: /* ip_srcroute doesn't do what we want here, need to fix */
1906: if (inp->inp_flags & INP_RECVRETOPTS) {
1907: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1908: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1909: if (*mp)
1910: mp = &(*mp)->m_next;
1911: }
1912: #endif
1913: if (inp->inp_flags & INP_RECVIF) {
1914: struct sockaddr_dl sdl;
1915:
1916: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1917: sdl.sdl_family = AF_LINK;
1918: sdl.sdl_index = m->m_pkthdr.rcvif ?
1919: m->m_pkthdr.rcvif->if_index : 0;
1920: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1921: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1922: IP_RECVIF, IPPROTO_IP);
1923: if (*mp)
1924: mp = &(*mp)->m_next;
1925: }
1.13 mycroft 1926: }
1927:
1928: int
1929: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1930: int *name;
1931: u_int namelen;
1932: void *oldp;
1933: size_t *oldlenp;
1934: void *newp;
1935: size_t newlen;
1936: {
1.88 sommerfe 1937: extern int subnetsarelocal, hostzeroisbroadcast;
1.52 thorpej 1938:
1.54 lukem 1939: int error, old;
1940:
1.13 mycroft 1941: /* All sysctl names at this level are terminal. */
1942: if (namelen != 1)
1943: return (ENOTDIR);
1944:
1945: switch (name[0]) {
1946: case IPCTL_FORWARDING:
1947: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1948: case IPCTL_SENDREDIRECTS:
1949: return (sysctl_int(oldp, oldlenp, newp, newlen,
1950: &ipsendredirects));
1951: case IPCTL_DEFTTL:
1952: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1953: #ifdef notyet
1954: case IPCTL_DEFMTU:
1955: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1956: #endif
1.26 thorpej 1957: case IPCTL_FORWSRCRT:
1.47 cjs 1958: /* Don't allow this to change in a secure environment. */
1.26 thorpej 1959: if (securelevel > 0)
1.46 cjs 1960: return (sysctl_rdint(oldp, oldlenp, newp,
1961: ip_forwsrcrt));
1962: else
1963: return (sysctl_int(oldp, oldlenp, newp, newlen,
1964: &ip_forwsrcrt));
1.27 thorpej 1965: case IPCTL_DIRECTEDBCAST:
1966: return (sysctl_int(oldp, oldlenp, newp, newlen,
1967: &ip_directedbcast));
1.47 cjs 1968: case IPCTL_ALLOWSRCRT:
1969: return (sysctl_int(oldp, oldlenp, newp, newlen,
1970: &ip_allowsrcrt));
1.52 thorpej 1971: case IPCTL_SUBNETSARELOCAL:
1972: return (sysctl_int(oldp, oldlenp, newp, newlen,
1973: &subnetsarelocal));
1.53 kml 1974: case IPCTL_MTUDISC:
1.60 kml 1975: error = sysctl_int(oldp, oldlenp, newp, newlen,
1976: &ip_mtudisc);
1.163 itojun 1977: if (error == 0 && ip_mtudisc == 0)
1978: rt_timer_queue_remove_all(ip_mtudisc_timeout_q, TRUE);
1.60 kml 1979: return error;
1.54 lukem 1980: case IPCTL_ANONPORTMIN:
1981: old = anonportmin;
1982: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1.118 itojun 1983: if (anonportmin >= anonportmax || anonportmin < 0
1984: || anonportmin > 65535
1.54 lukem 1985: #ifndef IPNOPRIVPORTS
1986: || anonportmin < IPPORT_RESERVED
1987: #endif
1988: ) {
1989: anonportmin = old;
1990: return (EINVAL);
1991: }
1992: return (error);
1993: case IPCTL_ANONPORTMAX:
1994: old = anonportmax;
1995: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1.118 itojun 1996: if (anonportmin >= anonportmax || anonportmax < 0
1997: || anonportmax > 65535
1.54 lukem 1998: #ifndef IPNOPRIVPORTS
1999: || anonportmax < IPPORT_RESERVED
2000: #endif
2001: ) {
2002: anonportmax = old;
2003: return (EINVAL);
2004: }
1.60 kml 2005: return (error);
2006: case IPCTL_MTUDISCTIMEOUT:
1.158 itojun 2007: old = ip_mtudisc_timeout;
1.60 kml 2008: error = sysctl_int(oldp, oldlenp, newp, newlen,
2009: &ip_mtudisc_timeout);
1.158 itojun 2010: if (ip_mtudisc_timeout < 0) {
2011: ip_mtudisc_timeout = old;
1.156 itojun 2012: return (EINVAL);
1.158 itojun 2013: }
1.163 itojun 2014: if (error == 0)
1.162 itojun 2015: rt_timer_queue_change(ip_mtudisc_timeout_q,
2016: ip_mtudisc_timeout);
1.54 lukem 2017: return (error);
1.65 matt 2018: #ifdef GATEWAY
2019: case IPCTL_MAXFLOWS:
1.67 thorpej 2020: {
2021: int s;
2022:
1.65 matt 2023: error = sysctl_int(oldp, oldlenp, newp, newlen,
2024: &ip_maxflows);
1.67 thorpej 2025: s = splsoftnet();
1.65 matt 2026: ipflow_reap(0);
1.67 thorpej 2027: splx(s);
1.65 matt 2028: return (error);
1.67 thorpej 2029: }
1.89 itojun 2030: #endif
1.90 itojun 2031: case IPCTL_HOSTZEROBROADCAST:
2032: return (sysctl_int(oldp, oldlenp, newp, newlen,
2033: &hostzeroisbroadcast));
1.89 itojun 2034: #if NGIF > 0
2035: case IPCTL_GIF_TTL:
1.157 itojun 2036: return (sysctl_int(oldp, oldlenp, newp, newlen,
1.90 itojun 2037: &ip_gif_ttl));
1.144 martin 2038: #endif
2039:
2040: #if NGRE > 0
2041: case IPCTL_GRE_TTL:
1.157 itojun 2042: return (sysctl_int(oldp, oldlenp, newp, newlen,
1.144 martin 2043: &ip_gre_ttl));
1.117 tron 2044: #endif
2045:
2046: #ifndef IPNOPRIVPORTS
2047: case IPCTL_LOWPORTMIN:
2048: old = lowportmin;
2049: error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmin);
2050: if (lowportmin >= lowportmax
2051: || lowportmin > IPPORT_RESERVEDMAX
2052: || lowportmin < IPPORT_RESERVEDMIN
2053: ) {
2054: lowportmin = old;
2055: return (EINVAL);
2056: }
2057: return (error);
2058: case IPCTL_LOWPORTMAX:
2059: old = lowportmax;
2060: error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmax);
2061: if (lowportmin >= lowportmax
2062: || lowportmax > IPPORT_RESERVEDMAX
2063: || lowportmax < IPPORT_RESERVEDMIN
2064: ) {
2065: lowportmax = old;
2066: return (EINVAL);
2067: }
2068: return (error);
1.65 matt 2069: #endif
1.131 itojun 2070:
2071: case IPCTL_MAXFRAGPACKETS:
2072: return (sysctl_int(oldp, oldlenp, newp, newlen,
2073: &ip_maxfragpackets));
1.88 sommerfe 2074:
1.165 christos 2075: case IPCTL_CHECKINTERFACE:
2076: return (sysctl_int(oldp, oldlenp, newp, newlen,
2077: &ip_checkinterface));
1.13 mycroft 2078: default:
2079: return (EOPNOTSUPP);
2080: }
2081: /* NOTREACHED */
1.1 cgd 2082: }
CVSweb <webmaster@jp.NetBSD.org>