Annotation of src/sys/netinet/ip_input.c, Revision 1.171
1.171 ! itojun 1: /* $NetBSD: ip_input.c,v 1.170 2003/07/03 05:03:53 itojun 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.
81: * 3. All advertising materials mentioning features or use of this software
82: * must display the following acknowledgement:
83: * This product includes software developed by the University of
84: * California, Berkeley and its contributors.
85: * 4. Neither the name of the University nor the names of its contributors
86: * may be used to endorse or promote products derived from this software
87: * without specific prior written permission.
88: *
89: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
90: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
91: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
92: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
93: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
94: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
95: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
96: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
97: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
98: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
99: * SUCH DAMAGE.
100: *
1.14 cgd 101: * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
1.1 cgd 102: */
1.141 lukem 103:
104: #include <sys/cdefs.h>
1.171 ! itojun 105: __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.170 2003/07/03 05:03:53 itojun Exp $");
1.55 scottr 106:
1.62 matt 107: #include "opt_gateway.h"
1.69 mrg 108: #include "opt_pfil_hooks.h"
1.91 thorpej 109: #include "opt_ipsec.h"
1.55 scottr 110: #include "opt_mrouting.h"
1.167 martin 111: #include "opt_mbuftrace.h"
1.135 thorpej 112: #include "opt_inet_csum.h"
1.1 cgd 113:
1.5 mycroft 114: #include <sys/param.h>
115: #include <sys/systm.h>
116: #include <sys/malloc.h>
117: #include <sys/mbuf.h>
118: #include <sys/domain.h>
119: #include <sys/protosw.h>
120: #include <sys/socket.h>
1.44 thorpej 121: #include <sys/socketvar.h>
1.5 mycroft 122: #include <sys/errno.h>
123: #include <sys/time.h>
124: #include <sys/kernel.h>
1.72 thorpej 125: #include <sys/pool.h>
1.28 christos 126: #include <sys/sysctl.h>
1.1 cgd 127:
1.5 mycroft 128: #include <net/if.h>
1.44 thorpej 129: #include <net/if_dl.h>
1.5 mycroft 130: #include <net/route.h>
1.45 mrg 131: #include <net/pfil.h>
1.1 cgd 132:
1.5 mycroft 133: #include <netinet/in.h>
134: #include <netinet/in_systm.h>
135: #include <netinet/ip.h>
136: #include <netinet/in_pcb.h>
137: #include <netinet/in_var.h>
138: #include <netinet/ip_var.h>
139: #include <netinet/ip_icmp.h>
1.89 itojun 140: /* just for gif_ttl */
141: #include <netinet/in_gif.h>
142: #include "gif.h"
1.144 martin 143: #include <net/if_gre.h>
144: #include "gre.h"
1.111 jdolecek 145:
146: #ifdef MROUTING
147: #include <netinet/ip_mroute.h>
148: #endif
1.89 itojun 149:
150: #ifdef IPSEC
151: #include <netinet6/ipsec.h>
152: #include <netkey/key.h>
153: #endif
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;
1.1 cgd 436:
1.164 matt 437: MCLAIM(m, &ip_rx_mowner);
1.1 cgd 438: #ifdef DIAGNOSTIC
439: if ((m->m_flags & M_PKTHDR) == 0)
440: panic("ipintr no HDR");
441: #endif
1.89 itojun 442: #ifdef IPSEC
443: /*
444: * should the inner packet be considered authentic?
445: * see comment in ah4_input().
446: */
447: if (m) {
448: m->m_flags &= ~M_AUTHIPHDR;
449: m->m_flags &= ~M_AUTHIPDGM;
450: }
451: #endif
1.164 matt 452:
1.1 cgd 453: /*
454: * If no IP addresses have been set yet but the interfaces
455: * are receiving, can't do anything with incoming packets yet.
456: */
1.140 matt 457: if (TAILQ_FIRST(&in_ifaddr) == 0)
1.1 cgd 458: goto bad;
459: ipstat.ips_total++;
1.154 thorpej 460: /*
461: * If the IP header is not aligned, slurp it up into a new
462: * mbuf with space for link headers, in the event we forward
463: * it. Otherwise, if it is aligned, make sure the entire
464: * base IP header is in the first mbuf of the chain.
465: */
466: if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
467: if ((m = m_copyup(m, sizeof(struct ip),
468: (max_linkhdr + 3) & ~3)) == NULL) {
469: /* XXXJRT new stat, please */
470: ipstat.ips_toosmall++;
471: return;
472: }
473: } else if (__predict_false(m->m_len < sizeof (struct ip))) {
474: if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
475: ipstat.ips_toosmall++;
476: return;
477: }
1.1 cgd 478: }
479: ip = mtod(m, struct ip *);
1.13 mycroft 480: if (ip->ip_v != IPVERSION) {
481: ipstat.ips_badvers++;
482: goto bad;
483: }
1.1 cgd 484: hlen = ip->ip_hl << 2;
485: if (hlen < sizeof(struct ip)) { /* minimum header length */
486: ipstat.ips_badhlen++;
487: goto bad;
488: }
489: if (hlen > m->m_len) {
490: if ((m = m_pullup(m, hlen)) == 0) {
491: ipstat.ips_badhlen++;
1.89 itojun 492: return;
1.1 cgd 493: }
494: ip = mtod(m, struct ip *);
495: }
1.98 thorpej 496:
1.85 hwr 497: /*
1.99 thorpej 498: * RFC1122: packets with a multicast source address are
1.98 thorpej 499: * not allowed.
1.85 hwr 500: */
501: if (IN_MULTICAST(ip->ip_src.s_addr)) {
1.130 itojun 502: ipstat.ips_badaddr++;
1.85 hwr 503: goto bad;
1.129 itojun 504: }
505:
506: /* 127/8 must not appear on wire - RFC1122 */
507: if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
508: (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
1.130 itojun 509: if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
510: ipstat.ips_badaddr++;
1.129 itojun 511: goto bad;
1.130 itojun 512: }
1.85 hwr 513: }
514:
1.135 thorpej 515: switch (m->m_pkthdr.csum_flags &
1.137 thorpej 516: ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) |
1.135 thorpej 517: M_CSUM_IPv4_BAD)) {
518: case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
519: INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
520: goto badcsum;
521:
522: case M_CSUM_IPv4:
523: /* Checksum was okay. */
524: INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
525: break;
526:
527: default:
528: /* Must compute it ourselves. */
529: INET_CSUM_COUNTER_INCR(&ip_swcsum);
530: if (in_cksum(m, hlen) != 0)
531: goto bad;
532: break;
1.1 cgd 533: }
534:
1.121 thorpej 535: /* Retrieve the packet length. */
536: len = ntohs(ip->ip_len);
1.81 proff 537:
538: /*
539: * Check for additional length bogosity
540: */
1.84 proff 541: if (len < hlen) {
1.81 proff 542: ipstat.ips_badlen++;
543: goto bad;
544: }
1.1 cgd 545:
546: /*
547: * Check that the amount of data in the buffers
548: * is as at least much as the IP header would have us expect.
549: * Trim mbufs if longer than we expect.
550: * Drop packet if shorter than we expect.
551: */
1.35 mycroft 552: if (m->m_pkthdr.len < len) {
1.1 cgd 553: ipstat.ips_tooshort++;
554: goto bad;
555: }
1.35 mycroft 556: if (m->m_pkthdr.len > len) {
1.1 cgd 557: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 558: m->m_len = len;
559: m->m_pkthdr.len = len;
1.1 cgd 560: } else
1.35 mycroft 561: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 562: }
563:
1.94 itojun 564: #ifdef IPSEC
1.149 wiz 565: /* ipflow (IP fast forwarding) is not compatible with IPsec. */
1.94 itojun 566: m->m_flags &= ~M_CANFASTFWD;
567: #else
1.64 thorpej 568: /*
569: * Assume that we can create a fast-forward IP flow entry
570: * based on this packet.
571: */
572: m->m_flags |= M_CANFASTFWD;
1.94 itojun 573: #endif
1.64 thorpej 574:
1.36 mrg 575: #ifdef PFIL_HOOKS
1.33 mrg 576: /*
1.64 thorpej 577: * Run through list of hooks for input packets. If there are any
578: * filters which require that additional packets in the flow are
579: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
580: * Note that filters must _never_ set this flag, as another filter
581: * in the list may have previously cleared it.
1.33 mrg 582: */
1.127 itojun 583: /*
584: * let ipfilter look at packet on the wire,
585: * not the decapsulated packet.
586: */
587: #ifdef IPSEC
1.136 itojun 588: if (!ipsec_getnhist(m))
1.127 itojun 589: #else
590: if (1)
591: #endif
592: {
1.169 itojun 593: struct in_addr odst;
594:
595: odst = ip->ip_dst;
1.127 itojun 596: if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif,
1.168 itojun 597: PFIL_IN) != 0)
598: return;
1.127 itojun 599: if (m == NULL)
600: return;
601: ip = mtod(m, struct ip *);
1.142 darrenr 602: hlen = ip->ip_hl << 2;
1.169 itojun 603: srcrt = (odst.s_addr != ip->ip_dst.s_addr);
1.127 itojun 604: }
1.36 mrg 605: #endif /* PFIL_HOOKS */
1.123 thorpej 606:
607: #ifdef ALTQ
608: /* XXX Temporary until ALTQ is changed to use a pfil hook */
609: if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) {
610: /* packet dropped by traffic conditioner */
611: return;
612: }
613: #endif
1.121 thorpej 614:
615: /*
1.1 cgd 616: * Process options and, if not destined for us,
617: * ship it on. ip_dooptions returns 1 when an
618: * error was detected (causing an icmp message
619: * to be sent and the original packet to be freed).
620: */
621: ip_nhops = 0; /* for source routed packets */
622: if (hlen > sizeof (struct ip) && ip_dooptions(m))
1.89 itojun 623: return;
1.1 cgd 624:
625: /*
1.165 christos 626: * Enable a consistency check between the destination address
627: * and the arrival interface for a unicast packet (the RFC 1122
628: * strong ES model) if IP forwarding is disabled and the packet
629: * is not locally generated.
630: *
631: * XXX - Checking also should be disabled if the destination
632: * address is ipnat'ed to a different interface.
633: *
634: * XXX - Checking is incompatible with IP aliases added
635: * to the loopback interface instead of the interface where
636: * the packets are received.
637: *
638: * XXX - We need to add a per ifaddr flag for this so that
639: * we get finer grain control.
640: */
641: checkif = ip_checkinterface && (ipforwarding == 0) &&
642: (m->m_pkthdr.rcvif != NULL) &&
643: ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0);
644:
645: /*
1.1 cgd 646: * Check our list of addresses, to see if the packet is for us.
1.100 itojun 647: *
648: * Traditional 4.4BSD did not consult IFF_UP at all.
649: * The behavior here is to treat addresses on !IFF_UP interface
650: * as not mine.
1.1 cgd 651: */
1.100 itojun 652: downmatch = 0;
1.140 matt 653: LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) {
1.97 itojun 654: if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
1.165 christos 655: if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif)
656: continue;
1.97 itojun 657: if ((ia->ia_ifp->if_flags & IFF_UP) != 0)
658: break;
1.100 itojun 659: else
660: downmatch++;
1.97 itojun 661: }
662: }
1.86 thorpej 663: if (ia != NULL)
664: goto ours;
1.57 tls 665: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
1.140 matt 666: TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) {
667: if (ifa->ifa_addr->sa_family != AF_INET)
668: continue;
1.57 tls 669: ia = ifatoia(ifa);
1.35 mycroft 670: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
671: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 672: /*
673: * Look for all-0's host part (old broadcast addr),
674: * either for subnet or net.
675: */
676: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 677: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 678: goto ours;
1.57 tls 679: /*
680: * An interface with IP address zero accepts
681: * all packets that arrive on that interface.
682: */
683: if (in_nullhost(ia->ia_addr.sin_addr))
684: goto ours;
1.1 cgd 685: }
686: }
1.18 mycroft 687: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 688: struct in_multi *inm;
689: #ifdef MROUTING
690: extern struct socket *ip_mrouter;
1.10 brezak 691:
1.147 matt 692: if (M_READONLY(m)) {
1.10 brezak 693: if ((m = m_pullup(m, hlen)) == 0) {
694: ipstat.ips_toosmall++;
1.89 itojun 695: return;
1.10 brezak 696: }
697: ip = mtod(m, struct ip *);
698: }
1.4 hpeyerl 699:
700: if (ip_mrouter) {
701: /*
702: * If we are acting as a multicast router, all
703: * incoming multicast packets are passed to the
704: * kernel-level multicast forwarding function.
705: * The packet is returned (relatively) intact; if
706: * ip_mforward() returns a non-zero value, the packet
707: * must be discarded, else it may be accepted below.
708: *
709: * (The IP ident field is put in the same byte order
710: * as expected when ip_mforward() is called from
711: * ip_output().)
712: */
1.13 mycroft 713: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
714: ipstat.ips_cantforward++;
1.4 hpeyerl 715: m_freem(m);
1.89 itojun 716: return;
1.4 hpeyerl 717: }
718:
719: /*
720: * The process-level routing demon needs to receive
721: * all multicast IGMP packets, whether or not this
722: * host belongs to their destination groups.
723: */
724: if (ip->ip_p == IPPROTO_IGMP)
725: goto ours;
1.13 mycroft 726: ipstat.ips_forward++;
1.4 hpeyerl 727: }
728: #endif
729: /*
730: * See if we belong to the destination multicast group on the
731: * arrival interface.
732: */
733: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
734: if (inm == NULL) {
1.13 mycroft 735: ipstat.ips_cantforward++;
1.4 hpeyerl 736: m_freem(m);
1.89 itojun 737: return;
1.4 hpeyerl 738: }
739: goto ours;
740: }
1.19 mycroft 741: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 742: in_nullhost(ip->ip_dst))
1.1 cgd 743: goto ours;
744:
745: /*
746: * Not for us; forward if possible and desirable.
747: */
748: if (ipforwarding == 0) {
749: ipstat.ips_cantforward++;
750: m_freem(m);
1.100 itojun 751: } else {
752: /*
753: * If ip_dst matched any of my address on !IFF_UP interface,
754: * and there's no IFF_UP interface that matches ip_dst,
755: * send icmp unreach. Forwarding it will result in in-kernel
756: * forwarding loop till TTL goes to 0.
757: */
758: if (downmatch) {
759: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
760: ipstat.ips_cantforward++;
761: return;
762: }
1.145 itojun 763: #ifdef IPSEC
764: if (ipsec4_in_reject(m, NULL)) {
765: ipsecstat.in_polvio++;
766: goto bad;
767: }
768: #endif
769:
1.169 itojun 770: ip_forward(m, srcrt);
1.100 itojun 771: }
1.89 itojun 772: return;
1.1 cgd 773:
774: ours:
775: /*
776: * If offset or IP_MF are set, must reassemble.
777: * Otherwise, nothing need be done.
778: * (We could look in the reassembly queue to see
779: * if the packet was previously fragmented,
780: * but it's not worth the time; just let them time out.)
781: */
1.155 itojun 782: if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
783: if (M_READONLY(m)) {
784: if ((m = m_pullup(m, hlen)) == NULL) {
785: ipstat.ips_toosmall++;
786: goto bad;
787: }
788: ip = mtod(m, struct ip *);
789: }
790:
1.1 cgd 791: /*
792: * Look for queue of fragments
793: * of this datagram.
794: */
1.75 thorpej 795: IPQ_LOCK();
1.140 matt 796: LIST_FOREACH(fp, &ipq, ipq_q)
1.1 cgd 797: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 798: in_hosteq(ip->ip_src, fp->ipq_src) &&
799: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 800: ip->ip_p == fp->ipq_p)
801: goto found;
802: fp = 0;
803: found:
804:
805: /*
806: * Adjust ip_len to not reflect header,
1.25 cgd 807: * set ipqe_mff if more fragments are expected,
1.1 cgd 808: * convert offset of this to bytes.
809: */
1.155 itojun 810: ip->ip_len = htons(ntohs(ip->ip_len) - hlen);
811: mff = (ip->ip_off & htons(IP_MF)) != 0;
1.25 cgd 812: if (mff) {
1.16 cgd 813: /*
814: * Make sure that fragments have a data length
815: * that's a non-zero multiple of 8 bytes.
816: */
1.155 itojun 817: if (ntohs(ip->ip_len) == 0 ||
818: (ntohs(ip->ip_len) & 0x7) != 0) {
1.16 cgd 819: ipstat.ips_badfrags++;
1.75 thorpej 820: IPQ_UNLOCK();
1.16 cgd 821: goto bad;
822: }
823: }
1.155 itojun 824: ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3);
1.1 cgd 825:
826: /*
827: * If datagram marked as having more fragments
828: * or if this is not the first fragment,
829: * attempt reassembly; if it succeeds, proceed.
830: */
1.155 itojun 831: if (mff || ip->ip_off != htons(0)) {
1.1 cgd 832: ipstat.ips_fragments++;
1.72 thorpej 833: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.25 cgd 834: if (ipqe == NULL) {
835: ipstat.ips_rcvmemdrop++;
1.75 thorpej 836: IPQ_UNLOCK();
1.25 cgd 837: goto bad;
838: }
839: ipqe->ipqe_mff = mff;
1.50 thorpej 840: ipqe->ipqe_m = m;
1.25 cgd 841: ipqe->ipqe_ip = ip;
1.50 thorpej 842: m = ip_reass(ipqe, fp);
1.75 thorpej 843: if (m == 0) {
844: IPQ_UNLOCK();
1.89 itojun 845: return;
1.75 thorpej 846: }
1.13 mycroft 847: ipstat.ips_reassembled++;
1.50 thorpej 848: ip = mtod(m, struct ip *);
1.74 thorpej 849: hlen = ip->ip_hl << 2;
1.155 itojun 850: ip->ip_len = htons(ntohs(ip->ip_len) + hlen);
1.1 cgd 851: } else
852: if (fp)
853: ip_freef(fp);
1.75 thorpej 854: IPQ_UNLOCK();
1.79 mycroft 855: }
1.128 itojun 856:
857: #ifdef IPSEC
858: /*
859: * enforce IPsec policy checking if we are seeing last header.
860: * note that we do not visit this with protocols with pcb layer
861: * code - like udp/tcp/raw ip.
862: */
863: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 &&
864: ipsec4_in_reject(m, NULL)) {
865: ipsecstat.in_polvio++;
866: goto bad;
867: }
868: #endif
1.1 cgd 869:
870: /*
871: * Switch out to protocol's input routine.
872: */
1.82 aidan 873: #if IFA_STATS
1.122 itojun 874: if (ia && ip)
1.155 itojun 875: ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
1.82 aidan 876: #endif
1.1 cgd 877: ipstat.ips_delivered++;
1.89 itojun 878: {
879: int off = hlen, nh = ip->ip_p;
880:
881: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
882: return;
883: }
1.1 cgd 884: bad:
885: m_freem(m);
1.135 thorpej 886: return;
887:
888: badcsum:
889: ipstat.ips_badsum++;
890: m_freem(m);
1.1 cgd 891: }
892:
893: /*
894: * Take incoming datagram fragment and try to
895: * reassemble it into whole datagram. If a chain for
896: * reassembly of this datagram already exists, then it
897: * is given as fp; otherwise have to make a chain.
898: */
1.50 thorpej 899: struct mbuf *
1.25 cgd 900: ip_reass(ipqe, fp)
1.109 augustss 901: struct ipqent *ipqe;
902: struct ipq *fp;
1.1 cgd 903: {
1.109 augustss 904: struct mbuf *m = ipqe->ipqe_m;
905: struct ipqent *nq, *p, *q;
1.25 cgd 906: struct ip *ip;
1.1 cgd 907: struct mbuf *t;
1.25 cgd 908: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 909: int i, next;
910:
1.75 thorpej 911: IPQ_LOCK_CHECK();
912:
1.1 cgd 913: /*
914: * Presence of header sizes in mbufs
915: * would confuse code below.
916: */
917: m->m_data += hlen;
918: m->m_len -= hlen;
919:
920: /*
921: * If first fragment to arrive, create a reassembly queue.
922: */
923: if (fp == 0) {
1.131 itojun 924: /*
925: * Enforce upper bound on number of fragmented packets
926: * for which we attempt reassembly;
927: * If maxfrag is 0, never accept fragments.
928: * If maxfrag is -1, accept all fragments without limitation.
929: */
930: if (ip_maxfragpackets < 0)
931: ;
932: else if (ip_nfragpackets >= ip_maxfragpackets)
933: goto dropfrag;
934: ip_nfragpackets++;
1.50 thorpej 935: MALLOC(fp, struct ipq *, sizeof (struct ipq),
936: M_FTABLE, M_NOWAIT);
937: if (fp == NULL)
1.1 cgd 938: goto dropfrag;
1.25 cgd 939: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 940: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 941: fp->ipq_p = ipqe->ipqe_ip->ip_p;
942: fp->ipq_id = ipqe->ipqe_ip->ip_id;
1.148 matt 943: TAILQ_INIT(&fp->ipq_fragq);
1.25 cgd 944: fp->ipq_src = ipqe->ipqe_ip->ip_src;
945: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
946: p = NULL;
1.1 cgd 947: goto insert;
948: }
949:
950: /*
951: * Find a segment which begins after this one does.
952: */
1.148 matt 953: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
954: p = q, q = TAILQ_NEXT(q, ipqe_q))
1.155 itojun 955: if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off))
1.1 cgd 956: break;
957:
958: /*
959: * If there is a preceding segment, it may provide some of
960: * our data already. If so, drop the data from the incoming
961: * segment. If it provides all of our data, drop us.
962: */
1.25 cgd 963: if (p != NULL) {
1.155 itojun 964: i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) -
965: ntohs(ipqe->ipqe_ip->ip_off);
1.1 cgd 966: if (i > 0) {
1.155 itojun 967: if (i >= ntohs(ipqe->ipqe_ip->ip_len))
1.1 cgd 968: goto dropfrag;
1.50 thorpej 969: m_adj(ipqe->ipqe_m, i);
1.155 itojun 970: ipqe->ipqe_ip->ip_off =
971: htons(ntohs(ipqe->ipqe_ip->ip_off) + i);
972: ipqe->ipqe_ip->ip_len =
973: htons(ntohs(ipqe->ipqe_ip->ip_len) - i);
1.1 cgd 974: }
975: }
976:
977: /*
978: * While we overlap succeeding segments trim them or,
979: * if they are completely covered, dequeue them.
980: */
1.155 itojun 981: for (; q != NULL &&
982: ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) >
983: ntohs(q->ipqe_ip->ip_off); q = nq) {
984: i = (ntohs(ipqe->ipqe_ip->ip_off) +
985: ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off);
986: if (i < ntohs(q->ipqe_ip->ip_len)) {
987: q->ipqe_ip->ip_len =
988: htons(ntohs(q->ipqe_ip->ip_len) - i);
989: q->ipqe_ip->ip_off =
990: htons(ntohs(q->ipqe_ip->ip_off) + i);
1.50 thorpej 991: m_adj(q->ipqe_m, i);
1.1 cgd 992: break;
993: }
1.148 matt 994: nq = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 995: m_freem(q->ipqe_m);
1.148 matt 996: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.72 thorpej 997: pool_put(&ipqent_pool, q);
1.1 cgd 998: }
999:
1000: insert:
1001: /*
1002: * Stick new segment in its place;
1003: * check for complete reassembly.
1004: */
1.25 cgd 1005: if (p == NULL) {
1.148 matt 1006: TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
1.25 cgd 1007: } else {
1.148 matt 1008: TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q);
1.25 cgd 1009: }
1.1 cgd 1010: next = 0;
1.148 matt 1011: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1012: p = q, q = TAILQ_NEXT(q, ipqe_q)) {
1.155 itojun 1013: if (ntohs(q->ipqe_ip->ip_off) != next)
1.1 cgd 1014: return (0);
1.155 itojun 1015: next += ntohs(q->ipqe_ip->ip_len);
1.1 cgd 1016: }
1.25 cgd 1017: if (p->ipqe_mff)
1.1 cgd 1018: return (0);
1019:
1020: /*
1.41 thorpej 1021: * Reassembly is complete. Check for a bogus message size and
1022: * concatenate fragments.
1.1 cgd 1023: */
1.148 matt 1024: q = TAILQ_FIRST(&fp->ipq_fragq);
1.25 cgd 1025: ip = q->ipqe_ip;
1.41 thorpej 1026: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
1027: ipstat.ips_toolong++;
1028: ip_freef(fp);
1029: return (0);
1030: }
1.50 thorpej 1031: m = q->ipqe_m;
1.1 cgd 1032: t = m->m_next;
1033: m->m_next = 0;
1034: m_cat(m, t);
1.148 matt 1035: nq = TAILQ_NEXT(q, ipqe_q);
1.72 thorpej 1036: pool_put(&ipqent_pool, q);
1.25 cgd 1037: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 1038: t = q->ipqe_m;
1.148 matt 1039: nq = TAILQ_NEXT(q, ipqe_q);
1.72 thorpej 1040: pool_put(&ipqent_pool, q);
1.1 cgd 1041: m_cat(m, t);
1042: }
1043:
1044: /*
1045: * Create header for new ip packet by
1046: * modifying header of first packet;
1047: * dequeue and discard fragment reassembly header.
1048: * Make header visible.
1049: */
1.155 itojun 1050: ip->ip_len = htons(next);
1.25 cgd 1051: ip->ip_src = fp->ipq_src;
1052: ip->ip_dst = fp->ipq_dst;
1053: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1054: FREE(fp, M_FTABLE);
1.131 itojun 1055: ip_nfragpackets--;
1.1 cgd 1056: m->m_len += (ip->ip_hl << 2);
1057: m->m_data -= (ip->ip_hl << 2);
1058: /* some debugging cruft by sklower, below, will go away soon */
1059: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
1.109 augustss 1060: int plen = 0;
1.50 thorpej 1061: for (t = m; t; t = t->m_next)
1062: plen += t->m_len;
1063: m->m_pkthdr.len = plen;
1.1 cgd 1064: }
1.50 thorpej 1065: return (m);
1.1 cgd 1066:
1067: dropfrag:
1068: ipstat.ips_fragdropped++;
1069: m_freem(m);
1.72 thorpej 1070: pool_put(&ipqent_pool, ipqe);
1.1 cgd 1071: return (0);
1072: }
1073:
1074: /*
1075: * Free a fragment reassembly header and all
1076: * associated datagrams.
1077: */
1.8 mycroft 1078: void
1.1 cgd 1079: ip_freef(fp)
1080: struct ipq *fp;
1081: {
1.109 augustss 1082: struct ipqent *q, *p;
1.1 cgd 1083:
1.75 thorpej 1084: IPQ_LOCK_CHECK();
1085:
1.148 matt 1086: for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) {
1087: p = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1088: m_freem(q->ipqe_m);
1.148 matt 1089: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.72 thorpej 1090: pool_put(&ipqent_pool, q);
1.1 cgd 1091: }
1.25 cgd 1092: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1093: FREE(fp, M_FTABLE);
1.131 itojun 1094: ip_nfragpackets--;
1.1 cgd 1095: }
1096:
1097: /*
1098: * IP timer processing;
1099: * if a timer expires on a reassembly
1100: * queue, discard it.
1101: */
1.8 mycroft 1102: void
1.1 cgd 1103: ip_slowtimo()
1104: {
1.109 augustss 1105: struct ipq *fp, *nfp;
1.24 mycroft 1106: int s = splsoftnet();
1.1 cgd 1107:
1.75 thorpej 1108: IPQ_LOCK();
1.140 matt 1109: for (fp = LIST_FIRST(&ipq); fp != NULL; fp = nfp) {
1110: nfp = LIST_NEXT(fp, ipq_q);
1.25 cgd 1111: if (--fp->ipq_ttl == 0) {
1.1 cgd 1112: ipstat.ips_fragtimeout++;
1.25 cgd 1113: ip_freef(fp);
1.1 cgd 1114: }
1115: }
1.131 itojun 1116: /*
1117: * If we are over the maximum number of fragments
1118: * (due to the limit being lowered), drain off
1119: * enough to get down to the new limit.
1120: */
1121: if (ip_maxfragpackets < 0)
1122: ;
1123: else {
1.140 matt 1124: while (ip_nfragpackets > ip_maxfragpackets && LIST_FIRST(&ipq))
1125: ip_freef(LIST_FIRST(&ipq));
1.131 itojun 1126: }
1.75 thorpej 1127: IPQ_UNLOCK();
1.63 matt 1128: #ifdef GATEWAY
1129: ipflow_slowtimo();
1130: #endif
1.1 cgd 1131: splx(s);
1132: }
1133:
1134: /*
1135: * Drain off all datagram fragments.
1136: */
1.8 mycroft 1137: void
1.1 cgd 1138: ip_drain()
1139: {
1140:
1.75 thorpej 1141: /*
1142: * We may be called from a device's interrupt context. If
1143: * the ipq is already busy, just bail out now.
1144: */
1145: if (ipq_lock_try() == 0)
1146: return;
1147:
1.140 matt 1148: while (LIST_FIRST(&ipq) != NULL) {
1.1 cgd 1149: ipstat.ips_fragdropped++;
1.140 matt 1150: ip_freef(LIST_FIRST(&ipq));
1.1 cgd 1151: }
1.75 thorpej 1152:
1153: IPQ_UNLOCK();
1.1 cgd 1154: }
1155:
1156: /*
1157: * Do option processing on a datagram,
1158: * possibly discarding it if bad options are encountered,
1159: * or forwarding it if source-routed.
1160: * Returns 1 if packet has been forwarded/freed,
1161: * 0 if the packet should be processed further.
1162: */
1.8 mycroft 1163: int
1.1 cgd 1164: ip_dooptions(m)
1165: struct mbuf *m;
1166: {
1.109 augustss 1167: struct ip *ip = mtod(m, struct ip *);
1168: u_char *cp, *cp0;
1169: struct ip_timestamp *ipt;
1170: struct in_ifaddr *ia;
1.1 cgd 1171: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 1172: struct in_addr dst;
1.1 cgd 1173: n_time ntime;
1174:
1.13 mycroft 1175: dst = ip->ip_dst;
1.1 cgd 1176: cp = (u_char *)(ip + 1);
1177: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
1178: for (; cnt > 0; cnt -= optlen, cp += optlen) {
1179: opt = cp[IPOPT_OPTVAL];
1180: if (opt == IPOPT_EOL)
1181: break;
1182: if (opt == IPOPT_NOP)
1183: optlen = 1;
1184: else {
1.113 itojun 1185: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1186: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1187: goto bad;
1188: }
1.1 cgd 1189: optlen = cp[IPOPT_OLEN];
1.114 itojun 1190: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 1191: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1192: goto bad;
1193: }
1194: }
1195: switch (opt) {
1196:
1197: default:
1198: break;
1199:
1200: /*
1201: * Source routing with record.
1202: * Find interface with current destination address.
1203: * If none on this machine then drop if strictly routed,
1204: * or do nothing if loosely routed.
1205: * Record interface address and bring up next address
1206: * component. If strictly routed make sure next
1207: * address is on directly accessible net.
1208: */
1209: case IPOPT_LSRR:
1210: case IPOPT_SSRR:
1.47 cjs 1211: if (ip_allowsrcrt == 0) {
1212: type = ICMP_UNREACH;
1213: code = ICMP_UNREACH_NET_PROHIB;
1214: goto bad;
1215: }
1.114 itojun 1216: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1217: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1218: goto bad;
1219: }
1.1 cgd 1220: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1221: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1222: goto bad;
1223: }
1224: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 1225: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 1226: if (ia == 0) {
1227: if (opt == IPOPT_SSRR) {
1228: type = ICMP_UNREACH;
1229: code = ICMP_UNREACH_SRCFAIL;
1230: goto bad;
1231: }
1232: /*
1233: * Loose routing, and not at next destination
1234: * yet; nothing to do except forward.
1235: */
1236: break;
1237: }
1238: off--; /* 0 origin */
1.112 sommerfe 1239: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 1240: /*
1241: * End of source route. Should be for us.
1242: */
1243: save_rte(cp, ip->ip_src);
1244: break;
1245: }
1246: /*
1247: * locate outgoing interface
1248: */
1249: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
1250: sizeof(ipaddr.sin_addr));
1.96 thorpej 1251: if (opt == IPOPT_SSRR)
1252: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1253: else
1.1 cgd 1254: ia = ip_rtaddr(ipaddr.sin_addr);
1255: if (ia == 0) {
1256: type = ICMP_UNREACH;
1257: code = ICMP_UNREACH_SRCFAIL;
1258: goto bad;
1259: }
1260: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 1261: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 1262: (caddr_t)(cp + off), sizeof(struct in_addr));
1263: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1264: /*
1265: * Let ip_intr's mcast routing check handle mcast pkts
1266: */
1.18 mycroft 1267: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1268: break;
1269:
1270: case IPOPT_RR:
1.114 itojun 1271: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1272: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1273: goto bad;
1274: }
1.1 cgd 1275: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1276: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1277: goto bad;
1278: }
1279: /*
1280: * If no space remains, ignore.
1281: */
1282: off--; /* 0 origin */
1.112 sommerfe 1283: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1284: break;
1285: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
1286: sizeof(ipaddr.sin_addr));
1287: /*
1288: * locate outgoing interface; if we're the destination,
1289: * use the incoming interface (should be same).
1290: */
1.96 thorpej 1291: if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))))
1292: == NULL &&
1293: (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
1.1 cgd 1294: type = ICMP_UNREACH;
1295: code = ICMP_UNREACH_HOST;
1296: goto bad;
1297: }
1.20 mycroft 1298: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 1299: (caddr_t)(cp + off), sizeof(struct in_addr));
1300: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1301: break;
1302:
1303: case IPOPT_TS:
1304: code = cp - (u_char *)ip;
1305: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1306: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1307: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1308: goto bad;
1.114 itojun 1309: }
1310: if (ipt->ipt_ptr < 5) {
1311: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1312: goto bad;
1313: }
1.15 cgd 1314: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1315: if (++ipt->ipt_oflw == 0) {
1316: code = (u_char *)&ipt->ipt_ptr -
1317: (u_char *)ip;
1.1 cgd 1318: goto bad;
1.114 itojun 1319: }
1.1 cgd 1320: break;
1321: }
1.104 thorpej 1322: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1323: switch (ipt->ipt_flg) {
1324:
1325: case IPOPT_TS_TSONLY:
1326: break;
1327:
1328: case IPOPT_TS_TSANDADDR:
1.66 thorpej 1329: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1330: sizeof(struct in_addr) > ipt->ipt_len) {
1331: code = (u_char *)&ipt->ipt_ptr -
1332: (u_char *)ip;
1.1 cgd 1333: goto bad;
1.114 itojun 1334: }
1.13 mycroft 1335: ipaddr.sin_addr = dst;
1.96 thorpej 1336: ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr),
1337: m->m_pkthdr.rcvif));
1.13 mycroft 1338: if (ia == 0)
1339: continue;
1.104 thorpej 1340: bcopy(&ia->ia_addr.sin_addr,
1341: cp0, sizeof(struct in_addr));
1.1 cgd 1342: ipt->ipt_ptr += sizeof(struct in_addr);
1343: break;
1344:
1345: case IPOPT_TS_PRESPEC:
1.66 thorpej 1346: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1347: sizeof(struct in_addr) > ipt->ipt_len) {
1348: code = (u_char *)&ipt->ipt_ptr -
1349: (u_char *)ip;
1.1 cgd 1350: goto bad;
1.114 itojun 1351: }
1.104 thorpej 1352: bcopy(cp0, &ipaddr.sin_addr,
1.1 cgd 1353: sizeof(struct in_addr));
1.96 thorpej 1354: if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))
1355: == NULL)
1.1 cgd 1356: continue;
1357: ipt->ipt_ptr += sizeof(struct in_addr);
1358: break;
1359:
1360: default:
1.114 itojun 1361: /* XXX can't take &ipt->ipt_flg */
1362: code = (u_char *)&ipt->ipt_ptr -
1363: (u_char *)ip + 1;
1.1 cgd 1364: goto bad;
1365: }
1366: ntime = iptime();
1.107 thorpej 1367: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1368: bcopy(cp0, (caddr_t)cp + ipt->ipt_ptr - 1,
1.1 cgd 1369: sizeof(n_time));
1370: ipt->ipt_ptr += sizeof(n_time);
1371: }
1372: }
1373: if (forward) {
1.26 thorpej 1374: if (ip_forwsrcrt == 0) {
1375: type = ICMP_UNREACH;
1376: code = ICMP_UNREACH_SRCFAIL;
1377: goto bad;
1378: }
1.1 cgd 1379: ip_forward(m, 1);
1380: return (1);
1.13 mycroft 1381: }
1382: return (0);
1.1 cgd 1383: bad:
1.13 mycroft 1384: icmp_error(m, type, code, 0, 0);
1385: ipstat.ips_badoptions++;
1.1 cgd 1386: return (1);
1387: }
1388:
1389: /*
1390: * Given address of next destination (final or next hop),
1391: * return internet address info of interface to be used to get there.
1392: */
1393: struct in_ifaddr *
1394: ip_rtaddr(dst)
1395: struct in_addr dst;
1396: {
1.109 augustss 1397: struct sockaddr_in *sin;
1.1 cgd 1398:
1.19 mycroft 1399: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1400:
1.35 mycroft 1401: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 1402: if (ipforward_rt.ro_rt) {
1403: RTFREE(ipforward_rt.ro_rt);
1404: ipforward_rt.ro_rt = 0;
1405: }
1406: sin->sin_family = AF_INET;
1407: sin->sin_len = sizeof(*sin);
1408: sin->sin_addr = dst;
1409:
1410: rtalloc(&ipforward_rt);
1411: }
1412: if (ipforward_rt.ro_rt == 0)
1413: return ((struct in_ifaddr *)0);
1.19 mycroft 1414: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 1415: }
1416:
1417: /*
1418: * Save incoming source route for use in replies,
1419: * to be picked up later by ip_srcroute if the receiver is interested.
1420: */
1.13 mycroft 1421: void
1.1 cgd 1422: save_rte(option, dst)
1423: u_char *option;
1424: struct in_addr dst;
1425: {
1426: unsigned olen;
1427:
1428: olen = option[IPOPT_OLEN];
1429: #ifdef DIAGNOSTIC
1430: if (ipprintfs)
1.39 christos 1431: printf("save_rte: olen %d\n", olen);
1.89 itojun 1432: #endif /* 0 */
1.1 cgd 1433: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1434: return;
1435: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
1436: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1437: ip_srcrt.dst = dst;
1438: }
1439:
1440: /*
1441: * Retrieve incoming source route for use in replies,
1442: * in the same form used by setsockopt.
1443: * The first hop is placed before the options, will be removed later.
1444: */
1445: struct mbuf *
1446: ip_srcroute()
1447: {
1.109 augustss 1448: struct in_addr *p, *q;
1449: struct mbuf *m;
1.1 cgd 1450:
1451: if (ip_nhops == 0)
1452: return ((struct mbuf *)0);
1453: m = m_get(M_DONTWAIT, MT_SOOPTS);
1454: if (m == 0)
1455: return ((struct mbuf *)0);
1456:
1.164 matt 1457: MCLAIM(m, &inetdomain.dom_mowner);
1.13 mycroft 1458: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1459:
1460: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1461: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1462: OPTSIZ;
1463: #ifdef DIAGNOSTIC
1464: if (ipprintfs)
1.39 christos 1465: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1466: #endif
1467:
1468: /*
1469: * First save first hop for return route
1470: */
1471: p = &ip_srcrt.route[ip_nhops - 1];
1472: *(mtod(m, struct in_addr *)) = *p--;
1473: #ifdef DIAGNOSTIC
1474: if (ipprintfs)
1.39 christos 1475: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1476: #endif
1477:
1478: /*
1479: * Copy option fields and padding (nop) to mbuf.
1480: */
1481: ip_srcrt.nop = IPOPT_NOP;
1482: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1483: bcopy((caddr_t)&ip_srcrt.nop,
1484: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1485: q = (struct in_addr *)(mtod(m, caddr_t) +
1486: sizeof(struct in_addr) + OPTSIZ);
1487: #undef OPTSIZ
1488: /*
1489: * Record return path as an IP source route,
1490: * reversing the path (pointers are now aligned).
1491: */
1492: while (p >= ip_srcrt.route) {
1493: #ifdef DIAGNOSTIC
1494: if (ipprintfs)
1.39 christos 1495: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1496: #endif
1497: *q++ = *p--;
1498: }
1499: /*
1500: * Last hop goes to final destination.
1501: */
1502: *q = ip_srcrt.dst;
1503: #ifdef DIAGNOSTIC
1504: if (ipprintfs)
1.39 christos 1505: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1506: #endif
1507: return (m);
1508: }
1509:
1510: /*
1511: * Strip out IP options, at higher
1512: * level protocol in the kernel.
1513: * Second argument is buffer to which options
1514: * will be moved, and return value is their length.
1515: * XXX should be deleted; last arg currently ignored.
1516: */
1.8 mycroft 1517: void
1.1 cgd 1518: ip_stripoptions(m, mopt)
1.109 augustss 1519: struct mbuf *m;
1.1 cgd 1520: struct mbuf *mopt;
1521: {
1.109 augustss 1522: int i;
1.1 cgd 1523: struct ip *ip = mtod(m, struct ip *);
1.109 augustss 1524: caddr_t opts;
1.1 cgd 1525: int olen;
1526:
1.79 mycroft 1527: olen = (ip->ip_hl << 2) - sizeof (struct ip);
1.1 cgd 1528: opts = (caddr_t)(ip + 1);
1529: i = m->m_len - (sizeof (struct ip) + olen);
1530: bcopy(opts + olen, opts, (unsigned)i);
1531: m->m_len -= olen;
1532: if (m->m_flags & M_PKTHDR)
1533: m->m_pkthdr.len -= olen;
1.155 itojun 1534: ip->ip_len = htons(ntohs(ip->ip_len) - olen);
1.79 mycroft 1535: ip->ip_hl = sizeof (struct ip) >> 2;
1.1 cgd 1536: }
1537:
1.139 matt 1538: const int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1539: 0, 0, 0, 0,
1540: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1541: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1542: EMSGSIZE, EHOSTUNREACH, 0, 0,
1543: 0, 0, 0, 0,
1544: ENOPROTOOPT
1545: };
1546:
1547: /*
1548: * Forward a packet. If some error occurs return the sender
1549: * an icmp packet. Note we can't always generate a meaningful
1550: * icmp message because icmp doesn't have a large enough repertoire
1551: * of codes and types.
1552: *
1553: * If not forwarding, just drop the packet. This could be confusing
1554: * if ipforwarding was zero but some routing protocol was advancing
1555: * us as a gateway to somewhere. However, we must let the routing
1556: * protocol deal with that.
1557: *
1558: * The srcrt parameter indicates whether the packet is being forwarded
1559: * via a source route.
1560: */
1.13 mycroft 1561: void
1.1 cgd 1562: ip_forward(m, srcrt)
1563: struct mbuf *m;
1564: int srcrt;
1565: {
1.109 augustss 1566: struct ip *ip = mtod(m, struct ip *);
1567: struct sockaddr_in *sin;
1568: struct rtentry *rt;
1.28 christos 1569: int error, type = 0, code = 0;
1.1 cgd 1570: struct mbuf *mcopy;
1.13 mycroft 1571: n_long dest;
1572: struct ifnet *destifp;
1.89 itojun 1573: #ifdef IPSEC
1574: struct ifnet dummyifp;
1575: #endif
1.164 matt 1576:
1577: /*
1578: * We are now in the output path.
1579: */
1580: MCLAIM(m, &ip_tx_mowner);
1.135 thorpej 1581:
1582: /*
1583: * Clear any in-bound checksum flags for this packet.
1584: */
1585: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1586:
1.13 mycroft 1587: dest = 0;
1.1 cgd 1588: #ifdef DIAGNOSTIC
1589: if (ipprintfs)
1.70 thorpej 1590: printf("forward: src %2.2x dst %2.2x ttl %x\n",
1591: ntohl(ip->ip_src.s_addr),
1592: ntohl(ip->ip_dst.s_addr), ip->ip_ttl);
1.1 cgd 1593: #endif
1.93 sommerfe 1594: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.1 cgd 1595: ipstat.ips_cantforward++;
1596: m_freem(m);
1597: return;
1598: }
1599: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1600: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1601: return;
1602: }
1603: ip->ip_ttl -= IPTTLDEC;
1604:
1.19 mycroft 1605: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1606: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1607: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1608: if (ipforward_rt.ro_rt) {
1609: RTFREE(ipforward_rt.ro_rt);
1610: ipforward_rt.ro_rt = 0;
1611: }
1612: sin->sin_family = AF_INET;
1.35 mycroft 1613: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1614: sin->sin_addr = ip->ip_dst;
1615:
1616: rtalloc(&ipforward_rt);
1617: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1618: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1619: return;
1620: }
1621: rt = ipforward_rt.ro_rt;
1622: }
1623:
1624: /*
1.34 mycroft 1625: * Save at most 68 bytes of the packet in case
1.1 cgd 1626: * we need to generate an ICMP message to the src.
1.119 itojun 1627: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1628: */
1.155 itojun 1629: mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1.119 itojun 1630: if (mcopy)
1631: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1632:
1633: /*
1634: * If forwarding packet using same interface that it came in on,
1635: * perhaps should send a redirect to sender to shortcut a hop.
1636: * Only send redirect if source is sending directly to us,
1637: * and if packet was not source routed (or has any options).
1638: * Also, don't send redirect if forwarding using a default route
1639: * or a route modified by a redirect.
1640: */
1641: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1642: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1643: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1644: ipsendredirects && !srcrt) {
1.19 mycroft 1645: if (rt->rt_ifa &&
1646: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1647: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1648: if (rt->rt_flags & RTF_GATEWAY)
1649: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1650: else
1651: dest = ip->ip_dst.s_addr;
1652: /*
1653: * Router requirements says to only send host
1654: * redirects.
1655: */
1656: type = ICMP_REDIRECT;
1657: code = ICMP_REDIRECT_HOST;
1.1 cgd 1658: #ifdef DIAGNOSTIC
1.77 thorpej 1659: if (ipprintfs)
1660: printf("redirect (%d) to %x\n", code,
1661: (u_int32_t)dest);
1.1 cgd 1662: #endif
1663: }
1664: }
1665:
1.89 itojun 1666: #ifdef IPSEC
1.134 lukem 1667: /* Don't lookup socket in forwarding case */
1.127 itojun 1668: (void)ipsec_setsocket(m, NULL);
1.103 itojun 1669: #endif
1.27 thorpej 1670: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1671: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1.1 cgd 1672: if (error)
1673: ipstat.ips_cantforward++;
1674: else {
1675: ipstat.ips_forward++;
1676: if (type)
1677: ipstat.ips_redirectsent++;
1678: else {
1.63 matt 1679: if (mcopy) {
1680: #ifdef GATEWAY
1.64 thorpej 1681: if (mcopy->m_flags & M_CANFASTFWD)
1682: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1683: #endif
1.1 cgd 1684: m_freem(mcopy);
1.63 matt 1685: }
1.1 cgd 1686: return;
1687: }
1688: }
1689: if (mcopy == NULL)
1690: return;
1.13 mycroft 1691: destifp = NULL;
1692:
1.1 cgd 1693: switch (error) {
1694:
1695: case 0: /* forwarded, but need redirect */
1696: /* type, code set above */
1697: break;
1698:
1699: case ENETUNREACH: /* shouldn't happen, checked above */
1700: case EHOSTUNREACH:
1701: case ENETDOWN:
1702: case EHOSTDOWN:
1703: default:
1704: type = ICMP_UNREACH;
1705: code = ICMP_UNREACH_HOST;
1706: break;
1707:
1708: case EMSGSIZE:
1709: type = ICMP_UNREACH;
1710: code = ICMP_UNREACH_NEEDFRAG;
1.89 itojun 1711: #ifndef IPSEC
1.13 mycroft 1712: if (ipforward_rt.ro_rt)
1713: destifp = ipforward_rt.ro_rt->rt_ifp;
1.89 itojun 1714: #else
1715: /*
1716: * If the packet is routed over IPsec tunnel, tell the
1717: * originator the tunnel MTU.
1718: * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1719: * XXX quickhack!!!
1720: */
1721: if (ipforward_rt.ro_rt) {
1722: struct secpolicy *sp;
1723: int ipsecerror;
1.95 itojun 1724: size_t ipsechdr;
1.89 itojun 1725: struct route *ro;
1726:
1727: sp = ipsec4_getpolicybyaddr(mcopy,
1.170 itojun 1728: IPSEC_DIR_OUTBOUND, IP_FORWARDING,
1729: &ipsecerror);
1.89 itojun 1730:
1731: if (sp == NULL)
1732: destifp = ipforward_rt.ro_rt->rt_ifp;
1733: else {
1734: /* count IPsec header size */
1.95 itojun 1735: ipsechdr = ipsec4_hdrsiz(mcopy,
1.170 itojun 1736: IPSEC_DIR_OUTBOUND, NULL);
1.89 itojun 1737:
1738: /*
1739: * find the correct route for outer IPv4
1740: * header, compute tunnel MTU.
1741: *
1742: * XXX BUG ALERT
1743: * The "dummyifp" code relies upon the fact
1744: * that icmp_error() touches only ifp->if_mtu.
1745: */
1746: /*XXX*/
1747: destifp = NULL;
1748: if (sp->req != NULL
1.95 itojun 1749: && sp->req->sav != NULL
1750: && sp->req->sav->sah != NULL) {
1751: ro = &sp->req->sav->sah->sa_route;
1.89 itojun 1752: if (ro->ro_rt && ro->ro_rt->rt_ifp) {
1753: dummyifp.if_mtu =
1.151 itojun 1754: ro->ro_rt->rt_rmx.rmx_mtu ?
1755: ro->ro_rt->rt_rmx.rmx_mtu :
1.89 itojun 1756: ro->ro_rt->rt_ifp->if_mtu;
1757: dummyifp.if_mtu -= ipsechdr;
1758: destifp = &dummyifp;
1759: }
1760: }
1761:
1762: key_freesp(sp);
1763: }
1764: }
1765: #endif /*IPSEC*/
1.1 cgd 1766: ipstat.ips_cantfrag++;
1767: break;
1768:
1769: case ENOBUFS:
1.143 itojun 1770: #if 1
1771: /*
1772: * a router should not generate ICMP_SOURCEQUENCH as
1773: * required in RFC1812 Requirements for IP Version 4 Routers.
1774: * source quench could be a big problem under DoS attacks,
1.149 wiz 1775: * or if the underlying interface is rate-limited.
1.143 itojun 1776: */
1777: if (mcopy)
1778: m_freem(mcopy);
1779: return;
1780: #else
1.1 cgd 1781: type = ICMP_SOURCEQUENCH;
1782: code = 0;
1783: break;
1.143 itojun 1784: #endif
1.1 cgd 1785: }
1.13 mycroft 1786: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 1787: }
1788:
1789: void
1790: ip_savecontrol(inp, mp, ip, m)
1.109 augustss 1791: struct inpcb *inp;
1792: struct mbuf **mp;
1793: struct ip *ip;
1794: struct mbuf *m;
1.44 thorpej 1795: {
1796:
1797: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1798: struct timeval tv;
1799:
1800: microtime(&tv);
1801: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1802: SCM_TIMESTAMP, SOL_SOCKET);
1803: if (*mp)
1804: mp = &(*mp)->m_next;
1805: }
1806: if (inp->inp_flags & INP_RECVDSTADDR) {
1807: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1808: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1809: if (*mp)
1810: mp = &(*mp)->m_next;
1811: }
1812: #ifdef notyet
1813: /*
1814: * XXX
1815: * Moving these out of udp_input() made them even more broken
1816: * than they already were.
1817: * - fenner@parc.xerox.com
1818: */
1819: /* options were tossed already */
1820: if (inp->inp_flags & INP_RECVOPTS) {
1821: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1822: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1823: if (*mp)
1824: mp = &(*mp)->m_next;
1825: }
1826: /* ip_srcroute doesn't do what we want here, need to fix */
1827: if (inp->inp_flags & INP_RECVRETOPTS) {
1828: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1829: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1830: if (*mp)
1831: mp = &(*mp)->m_next;
1832: }
1833: #endif
1834: if (inp->inp_flags & INP_RECVIF) {
1835: struct sockaddr_dl sdl;
1836:
1837: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1838: sdl.sdl_family = AF_LINK;
1839: sdl.sdl_index = m->m_pkthdr.rcvif ?
1840: m->m_pkthdr.rcvif->if_index : 0;
1841: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1842: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1843: IP_RECVIF, IPPROTO_IP);
1844: if (*mp)
1845: mp = &(*mp)->m_next;
1846: }
1.13 mycroft 1847: }
1848:
1849: int
1850: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1851: int *name;
1852: u_int namelen;
1853: void *oldp;
1854: size_t *oldlenp;
1855: void *newp;
1856: size_t newlen;
1857: {
1.88 sommerfe 1858: extern int subnetsarelocal, hostzeroisbroadcast;
1.52 thorpej 1859:
1.54 lukem 1860: int error, old;
1861:
1.13 mycroft 1862: /* All sysctl names at this level are terminal. */
1863: if (namelen != 1)
1864: return (ENOTDIR);
1865:
1866: switch (name[0]) {
1867: case IPCTL_FORWARDING:
1868: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1869: case IPCTL_SENDREDIRECTS:
1870: return (sysctl_int(oldp, oldlenp, newp, newlen,
1871: &ipsendredirects));
1872: case IPCTL_DEFTTL:
1873: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1874: #ifdef notyet
1875: case IPCTL_DEFMTU:
1876: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1877: #endif
1.26 thorpej 1878: case IPCTL_FORWSRCRT:
1.47 cjs 1879: /* Don't allow this to change in a secure environment. */
1.26 thorpej 1880: if (securelevel > 0)
1.46 cjs 1881: return (sysctl_rdint(oldp, oldlenp, newp,
1882: ip_forwsrcrt));
1883: else
1884: return (sysctl_int(oldp, oldlenp, newp, newlen,
1885: &ip_forwsrcrt));
1.27 thorpej 1886: case IPCTL_DIRECTEDBCAST:
1887: return (sysctl_int(oldp, oldlenp, newp, newlen,
1888: &ip_directedbcast));
1.47 cjs 1889: case IPCTL_ALLOWSRCRT:
1890: return (sysctl_int(oldp, oldlenp, newp, newlen,
1891: &ip_allowsrcrt));
1.52 thorpej 1892: case IPCTL_SUBNETSARELOCAL:
1893: return (sysctl_int(oldp, oldlenp, newp, newlen,
1894: &subnetsarelocal));
1.53 kml 1895: case IPCTL_MTUDISC:
1.60 kml 1896: error = sysctl_int(oldp, oldlenp, newp, newlen,
1897: &ip_mtudisc);
1.163 itojun 1898: if (error == 0 && ip_mtudisc == 0)
1899: rt_timer_queue_remove_all(ip_mtudisc_timeout_q, TRUE);
1.60 kml 1900: return error;
1.54 lukem 1901: case IPCTL_ANONPORTMIN:
1902: old = anonportmin;
1903: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1.118 itojun 1904: if (anonportmin >= anonportmax || anonportmin < 0
1905: || anonportmin > 65535
1.54 lukem 1906: #ifndef IPNOPRIVPORTS
1907: || anonportmin < IPPORT_RESERVED
1908: #endif
1909: ) {
1910: anonportmin = old;
1911: return (EINVAL);
1912: }
1913: return (error);
1914: case IPCTL_ANONPORTMAX:
1915: old = anonportmax;
1916: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1.118 itojun 1917: if (anonportmin >= anonportmax || anonportmax < 0
1918: || anonportmax > 65535
1.54 lukem 1919: #ifndef IPNOPRIVPORTS
1920: || anonportmax < IPPORT_RESERVED
1921: #endif
1922: ) {
1923: anonportmax = old;
1924: return (EINVAL);
1925: }
1.60 kml 1926: return (error);
1927: case IPCTL_MTUDISCTIMEOUT:
1.158 itojun 1928: old = ip_mtudisc_timeout;
1.60 kml 1929: error = sysctl_int(oldp, oldlenp, newp, newlen,
1930: &ip_mtudisc_timeout);
1.158 itojun 1931: if (ip_mtudisc_timeout < 0) {
1932: ip_mtudisc_timeout = old;
1.156 itojun 1933: return (EINVAL);
1.158 itojun 1934: }
1.163 itojun 1935: if (error == 0)
1.162 itojun 1936: rt_timer_queue_change(ip_mtudisc_timeout_q,
1937: ip_mtudisc_timeout);
1.54 lukem 1938: return (error);
1.65 matt 1939: #ifdef GATEWAY
1940: case IPCTL_MAXFLOWS:
1.67 thorpej 1941: {
1942: int s;
1943:
1.65 matt 1944: error = sysctl_int(oldp, oldlenp, newp, newlen,
1945: &ip_maxflows);
1.67 thorpej 1946: s = splsoftnet();
1.65 matt 1947: ipflow_reap(0);
1.67 thorpej 1948: splx(s);
1.65 matt 1949: return (error);
1.67 thorpej 1950: }
1.89 itojun 1951: #endif
1.90 itojun 1952: case IPCTL_HOSTZEROBROADCAST:
1953: return (sysctl_int(oldp, oldlenp, newp, newlen,
1954: &hostzeroisbroadcast));
1.89 itojun 1955: #if NGIF > 0
1956: case IPCTL_GIF_TTL:
1.157 itojun 1957: return (sysctl_int(oldp, oldlenp, newp, newlen,
1.90 itojun 1958: &ip_gif_ttl));
1.144 martin 1959: #endif
1960:
1961: #if NGRE > 0
1962: case IPCTL_GRE_TTL:
1.157 itojun 1963: return (sysctl_int(oldp, oldlenp, newp, newlen,
1.144 martin 1964: &ip_gre_ttl));
1.117 tron 1965: #endif
1966:
1967: #ifndef IPNOPRIVPORTS
1968: case IPCTL_LOWPORTMIN:
1969: old = lowportmin;
1970: error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmin);
1971: if (lowportmin >= lowportmax
1972: || lowportmin > IPPORT_RESERVEDMAX
1973: || lowportmin < IPPORT_RESERVEDMIN
1974: ) {
1975: lowportmin = old;
1976: return (EINVAL);
1977: }
1978: return (error);
1979: case IPCTL_LOWPORTMAX:
1980: old = lowportmax;
1981: error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmax);
1982: if (lowportmin >= lowportmax
1983: || lowportmax > IPPORT_RESERVEDMAX
1984: || lowportmax < IPPORT_RESERVEDMIN
1985: ) {
1986: lowportmax = old;
1987: return (EINVAL);
1988: }
1989: return (error);
1.65 matt 1990: #endif
1.131 itojun 1991:
1992: case IPCTL_MAXFRAGPACKETS:
1993: return (sysctl_int(oldp, oldlenp, newp, newlen,
1994: &ip_maxfragpackets));
1.88 sommerfe 1995:
1.165 christos 1996: case IPCTL_CHECKINTERFACE:
1997: return (sysctl_int(oldp, oldlenp, newp, newlen,
1998: &ip_checkinterface));
1.13 mycroft 1999: default:
2000: return (EOPNOTSUPP);
2001: }
2002: /* NOTREACHED */
1.1 cgd 2003: }
CVSweb <webmaster@jp.NetBSD.org>
![[BACK]](/icons/back.gif){kind=icon}
Return to ip_input.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [cvs.NetBSD.org] / src / sys / netinet