Annotation of src/sys/netinet/ip_input.c, Revision 1.166
1.166 ! matt 1: /* $NetBSD: ip_input.c,v 1.165 2003/04/11 19:41:37 christos 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>
1.166 ! matt 105: __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.165 2003/04/11 19:41:37 christos 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"
1.135 thorpej 111: #include "opt_inet_csum.h"
1.1 cgd 112:
1.5 mycroft 113: #include <sys/param.h>
114: #include <sys/systm.h>
115: #include <sys/malloc.h>
116: #include <sys/mbuf.h>
117: #include <sys/domain.h>
118: #include <sys/protosw.h>
119: #include <sys/socket.h>
1.44 thorpej 120: #include <sys/socketvar.h>
1.5 mycroft 121: #include <sys/errno.h>
122: #include <sys/time.h>
123: #include <sys/kernel.h>
1.72 thorpej 124: #include <sys/pool.h>
1.28 christos 125: #include <sys/sysctl.h>
1.1 cgd 126:
1.5 mycroft 127: #include <net/if.h>
1.44 thorpej 128: #include <net/if_dl.h>
1.5 mycroft 129: #include <net/route.h>
1.45 mrg 130: #include <net/pfil.h>
1.1 cgd 131:
1.5 mycroft 132: #include <netinet/in.h>
133: #include <netinet/in_systm.h>
134: #include <netinet/ip.h>
135: #include <netinet/in_pcb.h>
136: #include <netinet/in_var.h>
137: #include <netinet/ip_var.h>
138: #include <netinet/ip_icmp.h>
1.89 itojun 139: /* just for gif_ttl */
140: #include <netinet/in_gif.h>
141: #include "gif.h"
1.144 martin 142: #include <net/if_gre.h>
143: #include "gre.h"
1.111 jdolecek 144:
145: #ifdef MROUTING
146: #include <netinet/ip_mroute.h>
147: #endif
1.89 itojun 148:
149: #ifdef IPSEC
150: #include <netinet6/ipsec.h>
151: #include <netkey/key.h>
152: #endif
1.44 thorpej 153:
1.1 cgd 154: #ifndef IPFORWARDING
155: #ifdef GATEWAY
156: #define IPFORWARDING 1 /* forward IP packets not for us */
157: #else /* GATEWAY */
158: #define IPFORWARDING 0 /* don't forward IP packets not for us */
159: #endif /* GATEWAY */
160: #endif /* IPFORWARDING */
161: #ifndef IPSENDREDIRECTS
162: #define IPSENDREDIRECTS 1
163: #endif
1.26 thorpej 164: #ifndef IPFORWSRCRT
1.47 cjs 165: #define IPFORWSRCRT 1 /* forward source-routed packets */
166: #endif
167: #ifndef IPALLOWSRCRT
1.48 mrg 168: #define IPALLOWSRCRT 1 /* allow source-routed packets */
1.26 thorpej 169: #endif
1.53 kml 170: #ifndef IPMTUDISC
1.153 itojun 171: #define IPMTUDISC 1
1.53 kml 172: #endif
1.60 kml 173: #ifndef IPMTUDISCTIMEOUT
1.61 kml 174: #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */
1.60 kml 175: #endif
1.53 kml 176:
1.27 thorpej 177: /*
178: * Note: DIRECTED_BROADCAST is handled this way so that previous
179: * configuration using this option will Just Work.
180: */
181: #ifndef IPDIRECTEDBCAST
182: #ifdef DIRECTED_BROADCAST
183: #define IPDIRECTEDBCAST 1
184: #else
185: #define IPDIRECTEDBCAST 0
186: #endif /* DIRECTED_BROADCAST */
187: #endif /* IPDIRECTEDBCAST */
1.1 cgd 188: int ipforwarding = IPFORWARDING;
189: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 190: int ip_defttl = IPDEFTTL;
1.26 thorpej 191: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 192: int ip_directedbcast = IPDIRECTEDBCAST;
1.47 cjs 193: int ip_allowsrcrt = IPALLOWSRCRT;
1.53 kml 194: int ip_mtudisc = IPMTUDISC;
1.156 itojun 195: int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
1.1 cgd 196: #ifdef DIAGNOSTIC
197: int ipprintfs = 0;
198: #endif
1.165 christos 199: /*
200: * XXX - Setting ip_checkinterface mostly implements the receive side of
201: * the Strong ES model described in RFC 1122, but since the routing table
202: * and transmit implementation do not implement the Strong ES model,
203: * setting this to 1 results in an odd hybrid.
204: *
205: * XXX - ip_checkinterface currently must be disabled if you use ipnat
206: * to translate the destination address to another local interface.
207: *
208: * XXX - ip_checkinterface must be disabled if you add IP aliases
209: * to the loopback interface instead of the interface where the
210: * packets for those addresses are received.
211: */
212: int ip_checkinterface = 0;
213:
1.1 cgd 214:
1.60 kml 215: struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
216:
1.1 cgd 217: extern struct domain inetdomain;
218: int ipqmaxlen = IFQ_MAXLEN;
1.150 matt 219: u_long in_ifaddrhash; /* size of hash table - 1 */
220: int in_ifaddrentries; /* total number of addrs */
1.22 mycroft 221: struct in_ifaddrhead in_ifaddr;
1.57 tls 222: struct in_ifaddrhashhead *in_ifaddrhashtbl;
1.166 ! matt 223: u_long in_multihash; /* size of hash table - 1 */
! 224: int in_multientries; /* total number of addrs */
! 225: struct in_multihead in_multi;
! 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;
1.1 cgd 435:
1.164 matt 436: MCLAIM(m, &ip_rx_mowner);
1.1 cgd 437: #ifdef DIAGNOSTIC
438: if ((m->m_flags & M_PKTHDR) == 0)
439: panic("ipintr no HDR");
440: #endif
1.89 itojun 441: #ifdef IPSEC
442: /*
443: * should the inner packet be considered authentic?
444: * see comment in ah4_input().
445: */
446: if (m) {
447: m->m_flags &= ~M_AUTHIPHDR;
448: m->m_flags &= ~M_AUTHIPDGM;
449: }
450: #endif
1.164 matt 451:
1.1 cgd 452: /*
453: * If no IP addresses have been set yet but the interfaces
454: * are receiving, can't do anything with incoming packets yet.
455: */
1.140 matt 456: if (TAILQ_FIRST(&in_ifaddr) == 0)
1.1 cgd 457: goto bad;
458: ipstat.ips_total++;
1.154 thorpej 459: /*
460: * If the IP header is not aligned, slurp it up into a new
461: * mbuf with space for link headers, in the event we forward
462: * it. Otherwise, if it is aligned, make sure the entire
463: * base IP header is in the first mbuf of the chain.
464: */
465: if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
466: if ((m = m_copyup(m, sizeof(struct ip),
467: (max_linkhdr + 3) & ~3)) == NULL) {
468: /* XXXJRT new stat, please */
469: ipstat.ips_toosmall++;
470: return;
471: }
472: } else if (__predict_false(m->m_len < sizeof (struct ip))) {
473: if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
474: ipstat.ips_toosmall++;
475: return;
476: }
1.1 cgd 477: }
478: ip = mtod(m, struct ip *);
1.13 mycroft 479: if (ip->ip_v != IPVERSION) {
480: ipstat.ips_badvers++;
481: goto bad;
482: }
1.1 cgd 483: hlen = ip->ip_hl << 2;
484: if (hlen < sizeof(struct ip)) { /* minimum header length */
485: ipstat.ips_badhlen++;
486: goto bad;
487: }
488: if (hlen > m->m_len) {
489: if ((m = m_pullup(m, hlen)) == 0) {
490: ipstat.ips_badhlen++;
1.89 itojun 491: return;
1.1 cgd 492: }
493: ip = mtod(m, struct ip *);
494: }
1.98 thorpej 495:
1.85 hwr 496: /*
1.99 thorpej 497: * RFC1122: packets with a multicast source address are
1.98 thorpej 498: * not allowed.
1.85 hwr 499: */
500: if (IN_MULTICAST(ip->ip_src.s_addr)) {
1.130 itojun 501: ipstat.ips_badaddr++;
1.85 hwr 502: goto bad;
1.129 itojun 503: }
504:
505: /* 127/8 must not appear on wire - RFC1122 */
506: if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
507: (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
1.130 itojun 508: if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
509: ipstat.ips_badaddr++;
1.129 itojun 510: goto bad;
1.130 itojun 511: }
1.85 hwr 512: }
513:
1.135 thorpej 514: switch (m->m_pkthdr.csum_flags &
1.137 thorpej 515: ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) |
1.135 thorpej 516: M_CSUM_IPv4_BAD)) {
517: case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
518: INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
519: goto badcsum;
520:
521: case M_CSUM_IPv4:
522: /* Checksum was okay. */
523: INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
524: break;
525:
526: default:
527: /* Must compute it ourselves. */
528: INET_CSUM_COUNTER_INCR(&ip_swcsum);
529: if (in_cksum(m, hlen) != 0)
530: goto bad;
531: break;
1.1 cgd 532: }
533:
1.121 thorpej 534: /* Retrieve the packet length. */
535: len = ntohs(ip->ip_len);
1.81 proff 536:
537: /*
538: * Check for additional length bogosity
539: */
1.84 proff 540: if (len < hlen) {
1.81 proff 541: ipstat.ips_badlen++;
542: goto bad;
543: }
1.1 cgd 544:
545: /*
546: * Check that the amount of data in the buffers
547: * is as at least much as the IP header would have us expect.
548: * Trim mbufs if longer than we expect.
549: * Drop packet if shorter than we expect.
550: */
1.35 mycroft 551: if (m->m_pkthdr.len < len) {
1.1 cgd 552: ipstat.ips_tooshort++;
553: goto bad;
554: }
1.35 mycroft 555: if (m->m_pkthdr.len > len) {
1.1 cgd 556: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 557: m->m_len = len;
558: m->m_pkthdr.len = len;
1.1 cgd 559: } else
1.35 mycroft 560: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 561: }
562:
1.94 itojun 563: #ifdef IPSEC
1.149 wiz 564: /* ipflow (IP fast forwarding) is not compatible with IPsec. */
1.94 itojun 565: m->m_flags &= ~M_CANFASTFWD;
566: #else
1.64 thorpej 567: /*
568: * Assume that we can create a fast-forward IP flow entry
569: * based on this packet.
570: */
571: m->m_flags |= M_CANFASTFWD;
1.94 itojun 572: #endif
1.64 thorpej 573:
1.36 mrg 574: #ifdef PFIL_HOOKS
1.33 mrg 575: /*
1.64 thorpej 576: * Run through list of hooks for input packets. If there are any
577: * filters which require that additional packets in the flow are
578: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
579: * Note that filters must _never_ set this flag, as another filter
580: * in the list may have previously cleared it.
1.33 mrg 581: */
1.127 itojun 582: /*
583: * let ipfilter look at packet on the wire,
584: * not the decapsulated packet.
585: */
586: #ifdef IPSEC
1.136 itojun 587: if (!ipsec_getnhist(m))
1.127 itojun 588: #else
589: if (1)
590: #endif
591: {
592: if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif,
593: PFIL_IN) != 0)
1.121 thorpej 594: return;
1.127 itojun 595: if (m == NULL)
596: return;
597: ip = mtod(m, struct ip *);
1.142 darrenr 598: hlen = ip->ip_hl << 2;
1.127 itojun 599: }
1.36 mrg 600: #endif /* PFIL_HOOKS */
1.123 thorpej 601:
602: #ifdef ALTQ
603: /* XXX Temporary until ALTQ is changed to use a pfil hook */
604: if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) {
605: /* packet dropped by traffic conditioner */
606: return;
607: }
608: #endif
1.121 thorpej 609:
610: /*
1.1 cgd 611: * Process options and, if not destined for us,
612: * ship it on. ip_dooptions returns 1 when an
613: * error was detected (causing an icmp message
614: * to be sent and the original packet to be freed).
615: */
616: ip_nhops = 0; /* for source routed packets */
617: if (hlen > sizeof (struct ip) && ip_dooptions(m))
1.89 itojun 618: return;
1.1 cgd 619:
620: /*
1.165 christos 621: * Enable a consistency check between the destination address
622: * and the arrival interface for a unicast packet (the RFC 1122
623: * strong ES model) if IP forwarding is disabled and the packet
624: * is not locally generated.
625: *
626: * XXX - Checking also should be disabled if the destination
627: * address is ipnat'ed to a different interface.
628: *
629: * XXX - Checking is incompatible with IP aliases added
630: * to the loopback interface instead of the interface where
631: * the packets are received.
632: *
633: * XXX - We need to add a per ifaddr flag for this so that
634: * we get finer grain control.
635: */
636: checkif = ip_checkinterface && (ipforwarding == 0) &&
637: (m->m_pkthdr.rcvif != NULL) &&
638: ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0);
639:
640: /*
1.1 cgd 641: * Check our list of addresses, to see if the packet is for us.
1.100 itojun 642: *
643: * Traditional 4.4BSD did not consult IFF_UP at all.
644: * The behavior here is to treat addresses on !IFF_UP interface
645: * as not mine.
1.1 cgd 646: */
1.100 itojun 647: downmatch = 0;
1.140 matt 648: LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) {
1.97 itojun 649: if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
1.165 christos 650: if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif)
651: continue;
1.97 itojun 652: if ((ia->ia_ifp->if_flags & IFF_UP) != 0)
653: break;
1.100 itojun 654: else
655: downmatch++;
1.97 itojun 656: }
657: }
1.86 thorpej 658: if (ia != NULL)
659: goto ours;
1.57 tls 660: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
1.140 matt 661: TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) {
662: if (ifa->ifa_addr->sa_family != AF_INET)
663: continue;
1.57 tls 664: ia = ifatoia(ifa);
1.35 mycroft 665: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
666: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 667: /*
668: * Look for all-0's host part (old broadcast addr),
669: * either for subnet or net.
670: */
671: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 672: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 673: goto ours;
1.57 tls 674: /*
675: * An interface with IP address zero accepts
676: * all packets that arrive on that interface.
677: */
678: if (in_nullhost(ia->ia_addr.sin_addr))
679: goto ours;
1.1 cgd 680: }
681: }
1.18 mycroft 682: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 683: struct in_multi *inm;
684: #ifdef MROUTING
685: extern struct socket *ip_mrouter;
1.10 brezak 686:
1.147 matt 687: if (M_READONLY(m)) {
1.10 brezak 688: if ((m = m_pullup(m, hlen)) == 0) {
689: ipstat.ips_toosmall++;
1.89 itojun 690: return;
1.10 brezak 691: }
692: ip = mtod(m, struct ip *);
693: }
1.4 hpeyerl 694:
695: if (ip_mrouter) {
696: /*
697: * If we are acting as a multicast router, all
698: * incoming multicast packets are passed to the
699: * kernel-level multicast forwarding function.
700: * The packet is returned (relatively) intact; if
701: * ip_mforward() returns a non-zero value, the packet
702: * must be discarded, else it may be accepted below.
703: *
704: * (The IP ident field is put in the same byte order
705: * as expected when ip_mforward() is called from
706: * ip_output().)
707: */
1.13 mycroft 708: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
709: ipstat.ips_cantforward++;
1.4 hpeyerl 710: m_freem(m);
1.89 itojun 711: return;
1.4 hpeyerl 712: }
713:
714: /*
715: * The process-level routing demon needs to receive
716: * all multicast IGMP packets, whether or not this
717: * host belongs to their destination groups.
718: */
719: if (ip->ip_p == IPPROTO_IGMP)
720: goto ours;
1.13 mycroft 721: ipstat.ips_forward++;
1.4 hpeyerl 722: }
723: #endif
724: /*
725: * See if we belong to the destination multicast group on the
726: * arrival interface.
727: */
728: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
729: if (inm == NULL) {
1.13 mycroft 730: ipstat.ips_cantforward++;
1.4 hpeyerl 731: m_freem(m);
1.89 itojun 732: return;
1.4 hpeyerl 733: }
734: goto ours;
735: }
1.19 mycroft 736: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 737: in_nullhost(ip->ip_dst))
1.1 cgd 738: goto ours;
739:
740: /*
741: * Not for us; forward if possible and desirable.
742: */
743: if (ipforwarding == 0) {
744: ipstat.ips_cantforward++;
745: m_freem(m);
1.100 itojun 746: } else {
747: /*
748: * If ip_dst matched any of my address on !IFF_UP interface,
749: * and there's no IFF_UP interface that matches ip_dst,
750: * send icmp unreach. Forwarding it will result in in-kernel
751: * forwarding loop till TTL goes to 0.
752: */
753: if (downmatch) {
754: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
755: ipstat.ips_cantforward++;
756: return;
757: }
1.145 itojun 758: #ifdef IPSEC
759: if (ipsec4_in_reject(m, NULL)) {
760: ipsecstat.in_polvio++;
761: goto bad;
762: }
763: #endif
764:
1.1 cgd 765: ip_forward(m, 0);
1.100 itojun 766: }
1.89 itojun 767: return;
1.1 cgd 768:
769: ours:
770: /*
771: * If offset or IP_MF are set, must reassemble.
772: * Otherwise, nothing need be done.
773: * (We could look in the reassembly queue to see
774: * if the packet was previously fragmented,
775: * but it's not worth the time; just let them time out.)
776: */
1.155 itojun 777: if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
778: if (M_READONLY(m)) {
779: if ((m = m_pullup(m, hlen)) == NULL) {
780: ipstat.ips_toosmall++;
781: goto bad;
782: }
783: ip = mtod(m, struct ip *);
784: }
785:
1.1 cgd 786: /*
787: * Look for queue of fragments
788: * of this datagram.
789: */
1.75 thorpej 790: IPQ_LOCK();
1.140 matt 791: LIST_FOREACH(fp, &ipq, ipq_q)
1.1 cgd 792: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 793: in_hosteq(ip->ip_src, fp->ipq_src) &&
794: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 795: ip->ip_p == fp->ipq_p)
796: goto found;
797: fp = 0;
798: found:
799:
800: /*
801: * Adjust ip_len to not reflect header,
1.25 cgd 802: * set ipqe_mff if more fragments are expected,
1.1 cgd 803: * convert offset of this to bytes.
804: */
1.155 itojun 805: ip->ip_len = htons(ntohs(ip->ip_len) - hlen);
806: mff = (ip->ip_off & htons(IP_MF)) != 0;
1.25 cgd 807: if (mff) {
1.16 cgd 808: /*
809: * Make sure that fragments have a data length
810: * that's a non-zero multiple of 8 bytes.
811: */
1.155 itojun 812: if (ntohs(ip->ip_len) == 0 ||
813: (ntohs(ip->ip_len) & 0x7) != 0) {
1.16 cgd 814: ipstat.ips_badfrags++;
1.75 thorpej 815: IPQ_UNLOCK();
1.16 cgd 816: goto bad;
817: }
818: }
1.155 itojun 819: ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3);
1.1 cgd 820:
821: /*
822: * If datagram marked as having more fragments
823: * or if this is not the first fragment,
824: * attempt reassembly; if it succeeds, proceed.
825: */
1.155 itojun 826: if (mff || ip->ip_off != htons(0)) {
1.1 cgd 827: ipstat.ips_fragments++;
1.72 thorpej 828: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.25 cgd 829: if (ipqe == NULL) {
830: ipstat.ips_rcvmemdrop++;
1.75 thorpej 831: IPQ_UNLOCK();
1.25 cgd 832: goto bad;
833: }
834: ipqe->ipqe_mff = mff;
1.50 thorpej 835: ipqe->ipqe_m = m;
1.25 cgd 836: ipqe->ipqe_ip = ip;
1.50 thorpej 837: m = ip_reass(ipqe, fp);
1.75 thorpej 838: if (m == 0) {
839: IPQ_UNLOCK();
1.89 itojun 840: return;
1.75 thorpej 841: }
1.13 mycroft 842: ipstat.ips_reassembled++;
1.50 thorpej 843: ip = mtod(m, struct ip *);
1.74 thorpej 844: hlen = ip->ip_hl << 2;
1.155 itojun 845: ip->ip_len = htons(ntohs(ip->ip_len) + hlen);
1.1 cgd 846: } else
847: if (fp)
848: ip_freef(fp);
1.75 thorpej 849: IPQ_UNLOCK();
1.79 mycroft 850: }
1.128 itojun 851:
852: #ifdef IPSEC
853: /*
854: * enforce IPsec policy checking if we are seeing last header.
855: * note that we do not visit this with protocols with pcb layer
856: * code - like udp/tcp/raw ip.
857: */
858: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 &&
859: ipsec4_in_reject(m, NULL)) {
860: ipsecstat.in_polvio++;
861: goto bad;
862: }
863: #endif
1.1 cgd 864:
865: /*
866: * Switch out to protocol's input routine.
867: */
1.82 aidan 868: #if IFA_STATS
1.122 itojun 869: if (ia && ip)
1.155 itojun 870: ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
1.82 aidan 871: #endif
1.1 cgd 872: ipstat.ips_delivered++;
1.89 itojun 873: {
874: int off = hlen, nh = ip->ip_p;
875:
876: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
877: return;
878: }
1.1 cgd 879: bad:
880: m_freem(m);
1.135 thorpej 881: return;
882:
883: badcsum:
884: ipstat.ips_badsum++;
885: m_freem(m);
1.1 cgd 886: }
887:
888: /*
889: * Take incoming datagram fragment and try to
890: * reassemble it into whole datagram. If a chain for
891: * reassembly of this datagram already exists, then it
892: * is given as fp; otherwise have to make a chain.
893: */
1.50 thorpej 894: struct mbuf *
1.25 cgd 895: ip_reass(ipqe, fp)
1.109 augustss 896: struct ipqent *ipqe;
897: struct ipq *fp;
1.1 cgd 898: {
1.109 augustss 899: struct mbuf *m = ipqe->ipqe_m;
900: struct ipqent *nq, *p, *q;
1.25 cgd 901: struct ip *ip;
1.1 cgd 902: struct mbuf *t;
1.25 cgd 903: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 904: int i, next;
905:
1.75 thorpej 906: IPQ_LOCK_CHECK();
907:
1.1 cgd 908: /*
909: * Presence of header sizes in mbufs
910: * would confuse code below.
911: */
912: m->m_data += hlen;
913: m->m_len -= hlen;
914:
915: /*
916: * If first fragment to arrive, create a reassembly queue.
917: */
918: if (fp == 0) {
1.131 itojun 919: /*
920: * Enforce upper bound on number of fragmented packets
921: * for which we attempt reassembly;
922: * If maxfrag is 0, never accept fragments.
923: * If maxfrag is -1, accept all fragments without limitation.
924: */
925: if (ip_maxfragpackets < 0)
926: ;
927: else if (ip_nfragpackets >= ip_maxfragpackets)
928: goto dropfrag;
929: ip_nfragpackets++;
1.50 thorpej 930: MALLOC(fp, struct ipq *, sizeof (struct ipq),
931: M_FTABLE, M_NOWAIT);
932: if (fp == NULL)
1.1 cgd 933: goto dropfrag;
1.25 cgd 934: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 935: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 936: fp->ipq_p = ipqe->ipqe_ip->ip_p;
937: fp->ipq_id = ipqe->ipqe_ip->ip_id;
1.148 matt 938: TAILQ_INIT(&fp->ipq_fragq);
1.25 cgd 939: fp->ipq_src = ipqe->ipqe_ip->ip_src;
940: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
941: p = NULL;
1.1 cgd 942: goto insert;
943: }
944:
945: /*
946: * Find a segment which begins after this one does.
947: */
1.148 matt 948: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
949: p = q, q = TAILQ_NEXT(q, ipqe_q))
1.155 itojun 950: if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off))
1.1 cgd 951: break;
952:
953: /*
954: * If there is a preceding segment, it may provide some of
955: * our data already. If so, drop the data from the incoming
956: * segment. If it provides all of our data, drop us.
957: */
1.25 cgd 958: if (p != NULL) {
1.155 itojun 959: i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) -
960: ntohs(ipqe->ipqe_ip->ip_off);
1.1 cgd 961: if (i > 0) {
1.155 itojun 962: if (i >= ntohs(ipqe->ipqe_ip->ip_len))
1.1 cgd 963: goto dropfrag;
1.50 thorpej 964: m_adj(ipqe->ipqe_m, i);
1.155 itojun 965: ipqe->ipqe_ip->ip_off =
966: htons(ntohs(ipqe->ipqe_ip->ip_off) + i);
967: ipqe->ipqe_ip->ip_len =
968: htons(ntohs(ipqe->ipqe_ip->ip_len) - i);
1.1 cgd 969: }
970: }
971:
972: /*
973: * While we overlap succeeding segments trim them or,
974: * if they are completely covered, dequeue them.
975: */
1.155 itojun 976: for (; q != NULL &&
977: ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) >
978: ntohs(q->ipqe_ip->ip_off); q = nq) {
979: i = (ntohs(ipqe->ipqe_ip->ip_off) +
980: ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off);
981: if (i < ntohs(q->ipqe_ip->ip_len)) {
982: q->ipqe_ip->ip_len =
983: htons(ntohs(q->ipqe_ip->ip_len) - i);
984: q->ipqe_ip->ip_off =
985: htons(ntohs(q->ipqe_ip->ip_off) + i);
1.50 thorpej 986: m_adj(q->ipqe_m, i);
1.1 cgd 987: break;
988: }
1.148 matt 989: nq = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 990: m_freem(q->ipqe_m);
1.148 matt 991: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.72 thorpej 992: pool_put(&ipqent_pool, q);
1.1 cgd 993: }
994:
995: insert:
996: /*
997: * Stick new segment in its place;
998: * check for complete reassembly.
999: */
1.25 cgd 1000: if (p == NULL) {
1.148 matt 1001: TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
1.25 cgd 1002: } else {
1.148 matt 1003: TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q);
1.25 cgd 1004: }
1.1 cgd 1005: next = 0;
1.148 matt 1006: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1007: p = q, q = TAILQ_NEXT(q, ipqe_q)) {
1.155 itojun 1008: if (ntohs(q->ipqe_ip->ip_off) != next)
1.1 cgd 1009: return (0);
1.155 itojun 1010: next += ntohs(q->ipqe_ip->ip_len);
1.1 cgd 1011: }
1.25 cgd 1012: if (p->ipqe_mff)
1.1 cgd 1013: return (0);
1014:
1015: /*
1.41 thorpej 1016: * Reassembly is complete. Check for a bogus message size and
1017: * concatenate fragments.
1.1 cgd 1018: */
1.148 matt 1019: q = TAILQ_FIRST(&fp->ipq_fragq);
1.25 cgd 1020: ip = q->ipqe_ip;
1.41 thorpej 1021: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
1022: ipstat.ips_toolong++;
1023: ip_freef(fp);
1024: return (0);
1025: }
1.50 thorpej 1026: m = q->ipqe_m;
1.1 cgd 1027: t = m->m_next;
1028: m->m_next = 0;
1029: m_cat(m, t);
1.148 matt 1030: nq = TAILQ_NEXT(q, ipqe_q);
1.72 thorpej 1031: pool_put(&ipqent_pool, q);
1.25 cgd 1032: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 1033: t = q->ipqe_m;
1.148 matt 1034: nq = TAILQ_NEXT(q, ipqe_q);
1.72 thorpej 1035: pool_put(&ipqent_pool, q);
1.1 cgd 1036: m_cat(m, t);
1037: }
1038:
1039: /*
1040: * Create header for new ip packet by
1041: * modifying header of first packet;
1042: * dequeue and discard fragment reassembly header.
1043: * Make header visible.
1044: */
1.155 itojun 1045: ip->ip_len = htons(next);
1.25 cgd 1046: ip->ip_src = fp->ipq_src;
1047: ip->ip_dst = fp->ipq_dst;
1048: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1049: FREE(fp, M_FTABLE);
1.131 itojun 1050: ip_nfragpackets--;
1.1 cgd 1051: m->m_len += (ip->ip_hl << 2);
1052: m->m_data -= (ip->ip_hl << 2);
1053: /* some debugging cruft by sklower, below, will go away soon */
1054: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
1.109 augustss 1055: int plen = 0;
1.50 thorpej 1056: for (t = m; t; t = t->m_next)
1057: plen += t->m_len;
1058: m->m_pkthdr.len = plen;
1.1 cgd 1059: }
1.50 thorpej 1060: return (m);
1.1 cgd 1061:
1062: dropfrag:
1063: ipstat.ips_fragdropped++;
1064: m_freem(m);
1.72 thorpej 1065: pool_put(&ipqent_pool, ipqe);
1.1 cgd 1066: return (0);
1067: }
1068:
1069: /*
1070: * Free a fragment reassembly header and all
1071: * associated datagrams.
1072: */
1.8 mycroft 1073: void
1.1 cgd 1074: ip_freef(fp)
1075: struct ipq *fp;
1076: {
1.109 augustss 1077: struct ipqent *q, *p;
1.1 cgd 1078:
1.75 thorpej 1079: IPQ_LOCK_CHECK();
1080:
1.148 matt 1081: for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) {
1082: p = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1083: m_freem(q->ipqe_m);
1.148 matt 1084: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.72 thorpej 1085: pool_put(&ipqent_pool, q);
1.1 cgd 1086: }
1.25 cgd 1087: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1088: FREE(fp, M_FTABLE);
1.131 itojun 1089: ip_nfragpackets--;
1.1 cgd 1090: }
1091:
1092: /*
1093: * IP timer processing;
1094: * if a timer expires on a reassembly
1095: * queue, discard it.
1096: */
1.8 mycroft 1097: void
1.1 cgd 1098: ip_slowtimo()
1099: {
1.109 augustss 1100: struct ipq *fp, *nfp;
1.24 mycroft 1101: int s = splsoftnet();
1.1 cgd 1102:
1.75 thorpej 1103: IPQ_LOCK();
1.140 matt 1104: for (fp = LIST_FIRST(&ipq); fp != NULL; fp = nfp) {
1105: nfp = LIST_NEXT(fp, ipq_q);
1.25 cgd 1106: if (--fp->ipq_ttl == 0) {
1.1 cgd 1107: ipstat.ips_fragtimeout++;
1.25 cgd 1108: ip_freef(fp);
1.1 cgd 1109: }
1110: }
1.131 itojun 1111: /*
1112: * If we are over the maximum number of fragments
1113: * (due to the limit being lowered), drain off
1114: * enough to get down to the new limit.
1115: */
1116: if (ip_maxfragpackets < 0)
1117: ;
1118: else {
1.140 matt 1119: while (ip_nfragpackets > ip_maxfragpackets && LIST_FIRST(&ipq))
1120: ip_freef(LIST_FIRST(&ipq));
1.131 itojun 1121: }
1.75 thorpej 1122: IPQ_UNLOCK();
1.63 matt 1123: #ifdef GATEWAY
1124: ipflow_slowtimo();
1125: #endif
1.1 cgd 1126: splx(s);
1127: }
1128:
1129: /*
1130: * Drain off all datagram fragments.
1131: */
1.8 mycroft 1132: void
1.1 cgd 1133: ip_drain()
1134: {
1135:
1.75 thorpej 1136: /*
1137: * We may be called from a device's interrupt context. If
1138: * the ipq is already busy, just bail out now.
1139: */
1140: if (ipq_lock_try() == 0)
1141: return;
1142:
1.140 matt 1143: while (LIST_FIRST(&ipq) != NULL) {
1.1 cgd 1144: ipstat.ips_fragdropped++;
1.140 matt 1145: ip_freef(LIST_FIRST(&ipq));
1.1 cgd 1146: }
1.75 thorpej 1147:
1148: IPQ_UNLOCK();
1.1 cgd 1149: }
1150:
1151: /*
1152: * Do option processing on a datagram,
1153: * possibly discarding it if bad options are encountered,
1154: * or forwarding it if source-routed.
1155: * Returns 1 if packet has been forwarded/freed,
1156: * 0 if the packet should be processed further.
1157: */
1.8 mycroft 1158: int
1.1 cgd 1159: ip_dooptions(m)
1160: struct mbuf *m;
1161: {
1.109 augustss 1162: struct ip *ip = mtod(m, struct ip *);
1163: u_char *cp, *cp0;
1164: struct ip_timestamp *ipt;
1165: struct in_ifaddr *ia;
1.1 cgd 1166: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 1167: struct in_addr dst;
1.1 cgd 1168: n_time ntime;
1169:
1.13 mycroft 1170: dst = ip->ip_dst;
1.1 cgd 1171: cp = (u_char *)(ip + 1);
1172: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
1173: for (; cnt > 0; cnt -= optlen, cp += optlen) {
1174: opt = cp[IPOPT_OPTVAL];
1175: if (opt == IPOPT_EOL)
1176: break;
1177: if (opt == IPOPT_NOP)
1178: optlen = 1;
1179: else {
1.113 itojun 1180: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1181: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1182: goto bad;
1183: }
1.1 cgd 1184: optlen = cp[IPOPT_OLEN];
1.114 itojun 1185: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 1186: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1187: goto bad;
1188: }
1189: }
1190: switch (opt) {
1191:
1192: default:
1193: break;
1194:
1195: /*
1196: * Source routing with record.
1197: * Find interface with current destination address.
1198: * If none on this machine then drop if strictly routed,
1199: * or do nothing if loosely routed.
1200: * Record interface address and bring up next address
1201: * component. If strictly routed make sure next
1202: * address is on directly accessible net.
1203: */
1204: case IPOPT_LSRR:
1205: case IPOPT_SSRR:
1.47 cjs 1206: if (ip_allowsrcrt == 0) {
1207: type = ICMP_UNREACH;
1208: code = ICMP_UNREACH_NET_PROHIB;
1209: goto bad;
1210: }
1.114 itojun 1211: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1212: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1213: goto bad;
1214: }
1.1 cgd 1215: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1216: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1217: goto bad;
1218: }
1219: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 1220: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 1221: if (ia == 0) {
1222: if (opt == IPOPT_SSRR) {
1223: type = ICMP_UNREACH;
1224: code = ICMP_UNREACH_SRCFAIL;
1225: goto bad;
1226: }
1227: /*
1228: * Loose routing, and not at next destination
1229: * yet; nothing to do except forward.
1230: */
1231: break;
1232: }
1233: off--; /* 0 origin */
1.112 sommerfe 1234: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 1235: /*
1236: * End of source route. Should be for us.
1237: */
1238: save_rte(cp, ip->ip_src);
1239: break;
1240: }
1241: /*
1242: * locate outgoing interface
1243: */
1244: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
1245: sizeof(ipaddr.sin_addr));
1.96 thorpej 1246: if (opt == IPOPT_SSRR)
1247: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1248: else
1.1 cgd 1249: ia = ip_rtaddr(ipaddr.sin_addr);
1250: if (ia == 0) {
1251: type = ICMP_UNREACH;
1252: code = ICMP_UNREACH_SRCFAIL;
1253: goto bad;
1254: }
1255: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 1256: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 1257: (caddr_t)(cp + off), sizeof(struct in_addr));
1258: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1259: /*
1260: * Let ip_intr's mcast routing check handle mcast pkts
1261: */
1.18 mycroft 1262: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1263: break;
1264:
1265: case IPOPT_RR:
1.114 itojun 1266: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1267: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1268: goto bad;
1269: }
1.1 cgd 1270: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1271: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1272: goto bad;
1273: }
1274: /*
1275: * If no space remains, ignore.
1276: */
1277: off--; /* 0 origin */
1.112 sommerfe 1278: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1279: break;
1280: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
1281: sizeof(ipaddr.sin_addr));
1282: /*
1283: * locate outgoing interface; if we're the destination,
1284: * use the incoming interface (should be same).
1285: */
1.96 thorpej 1286: if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))))
1287: == NULL &&
1288: (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
1.1 cgd 1289: type = ICMP_UNREACH;
1290: code = ICMP_UNREACH_HOST;
1291: goto bad;
1292: }
1.20 mycroft 1293: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 1294: (caddr_t)(cp + off), sizeof(struct in_addr));
1295: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1296: break;
1297:
1298: case IPOPT_TS:
1299: code = cp - (u_char *)ip;
1300: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1301: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1302: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1303: goto bad;
1.114 itojun 1304: }
1305: if (ipt->ipt_ptr < 5) {
1306: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1307: goto bad;
1308: }
1.15 cgd 1309: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1310: if (++ipt->ipt_oflw == 0) {
1311: code = (u_char *)&ipt->ipt_ptr -
1312: (u_char *)ip;
1.1 cgd 1313: goto bad;
1.114 itojun 1314: }
1.1 cgd 1315: break;
1316: }
1.104 thorpej 1317: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1318: switch (ipt->ipt_flg) {
1319:
1320: case IPOPT_TS_TSONLY:
1321: break;
1322:
1323: case IPOPT_TS_TSANDADDR:
1.66 thorpej 1324: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1325: sizeof(struct in_addr) > ipt->ipt_len) {
1326: code = (u_char *)&ipt->ipt_ptr -
1327: (u_char *)ip;
1.1 cgd 1328: goto bad;
1.114 itojun 1329: }
1.13 mycroft 1330: ipaddr.sin_addr = dst;
1.96 thorpej 1331: ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr),
1332: m->m_pkthdr.rcvif));
1.13 mycroft 1333: if (ia == 0)
1334: continue;
1.104 thorpej 1335: bcopy(&ia->ia_addr.sin_addr,
1336: cp0, sizeof(struct in_addr));
1.1 cgd 1337: ipt->ipt_ptr += sizeof(struct in_addr);
1338: break;
1339:
1340: case IPOPT_TS_PRESPEC:
1.66 thorpej 1341: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1342: sizeof(struct in_addr) > ipt->ipt_len) {
1343: code = (u_char *)&ipt->ipt_ptr -
1344: (u_char *)ip;
1.1 cgd 1345: goto bad;
1.114 itojun 1346: }
1.104 thorpej 1347: bcopy(cp0, &ipaddr.sin_addr,
1.1 cgd 1348: sizeof(struct in_addr));
1.96 thorpej 1349: if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))
1350: == NULL)
1.1 cgd 1351: continue;
1352: ipt->ipt_ptr += sizeof(struct in_addr);
1353: break;
1354:
1355: default:
1.114 itojun 1356: /* XXX can't take &ipt->ipt_flg */
1357: code = (u_char *)&ipt->ipt_ptr -
1358: (u_char *)ip + 1;
1.1 cgd 1359: goto bad;
1360: }
1361: ntime = iptime();
1.107 thorpej 1362: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1363: bcopy(cp0, (caddr_t)cp + ipt->ipt_ptr - 1,
1.1 cgd 1364: sizeof(n_time));
1365: ipt->ipt_ptr += sizeof(n_time);
1366: }
1367: }
1368: if (forward) {
1.26 thorpej 1369: if (ip_forwsrcrt == 0) {
1370: type = ICMP_UNREACH;
1371: code = ICMP_UNREACH_SRCFAIL;
1372: goto bad;
1373: }
1.1 cgd 1374: ip_forward(m, 1);
1375: return (1);
1.13 mycroft 1376: }
1377: return (0);
1.1 cgd 1378: bad:
1.13 mycroft 1379: icmp_error(m, type, code, 0, 0);
1380: ipstat.ips_badoptions++;
1.1 cgd 1381: return (1);
1382: }
1383:
1384: /*
1385: * Given address of next destination (final or next hop),
1386: * return internet address info of interface to be used to get there.
1387: */
1388: struct in_ifaddr *
1389: ip_rtaddr(dst)
1390: struct in_addr dst;
1391: {
1.109 augustss 1392: struct sockaddr_in *sin;
1.1 cgd 1393:
1.19 mycroft 1394: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1395:
1.35 mycroft 1396: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 1397: if (ipforward_rt.ro_rt) {
1398: RTFREE(ipforward_rt.ro_rt);
1399: ipforward_rt.ro_rt = 0;
1400: }
1401: sin->sin_family = AF_INET;
1402: sin->sin_len = sizeof(*sin);
1403: sin->sin_addr = dst;
1404:
1405: rtalloc(&ipforward_rt);
1406: }
1407: if (ipforward_rt.ro_rt == 0)
1408: return ((struct in_ifaddr *)0);
1.19 mycroft 1409: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 1410: }
1411:
1412: /*
1413: * Save incoming source route for use in replies,
1414: * to be picked up later by ip_srcroute if the receiver is interested.
1415: */
1.13 mycroft 1416: void
1.1 cgd 1417: save_rte(option, dst)
1418: u_char *option;
1419: struct in_addr dst;
1420: {
1421: unsigned olen;
1422:
1423: olen = option[IPOPT_OLEN];
1424: #ifdef DIAGNOSTIC
1425: if (ipprintfs)
1.39 christos 1426: printf("save_rte: olen %d\n", olen);
1.89 itojun 1427: #endif /* 0 */
1.1 cgd 1428: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1429: return;
1430: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
1431: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1432: ip_srcrt.dst = dst;
1433: }
1434:
1435: /*
1436: * Retrieve incoming source route for use in replies,
1437: * in the same form used by setsockopt.
1438: * The first hop is placed before the options, will be removed later.
1439: */
1440: struct mbuf *
1441: ip_srcroute()
1442: {
1.109 augustss 1443: struct in_addr *p, *q;
1444: struct mbuf *m;
1.1 cgd 1445:
1446: if (ip_nhops == 0)
1447: return ((struct mbuf *)0);
1448: m = m_get(M_DONTWAIT, MT_SOOPTS);
1449: if (m == 0)
1450: return ((struct mbuf *)0);
1451:
1.164 matt 1452: MCLAIM(m, &inetdomain.dom_mowner);
1.13 mycroft 1453: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1454:
1455: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1456: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1457: OPTSIZ;
1458: #ifdef DIAGNOSTIC
1459: if (ipprintfs)
1.39 christos 1460: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1461: #endif
1462:
1463: /*
1464: * First save first hop for return route
1465: */
1466: p = &ip_srcrt.route[ip_nhops - 1];
1467: *(mtod(m, struct in_addr *)) = *p--;
1468: #ifdef DIAGNOSTIC
1469: if (ipprintfs)
1.39 christos 1470: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1471: #endif
1472:
1473: /*
1474: * Copy option fields and padding (nop) to mbuf.
1475: */
1476: ip_srcrt.nop = IPOPT_NOP;
1477: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1478: bcopy((caddr_t)&ip_srcrt.nop,
1479: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1480: q = (struct in_addr *)(mtod(m, caddr_t) +
1481: sizeof(struct in_addr) + OPTSIZ);
1482: #undef OPTSIZ
1483: /*
1484: * Record return path as an IP source route,
1485: * reversing the path (pointers are now aligned).
1486: */
1487: while (p >= ip_srcrt.route) {
1488: #ifdef DIAGNOSTIC
1489: if (ipprintfs)
1.39 christos 1490: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1491: #endif
1492: *q++ = *p--;
1493: }
1494: /*
1495: * Last hop goes to final destination.
1496: */
1497: *q = ip_srcrt.dst;
1498: #ifdef DIAGNOSTIC
1499: if (ipprintfs)
1.39 christos 1500: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1501: #endif
1502: return (m);
1503: }
1504:
1505: /*
1506: * Strip out IP options, at higher
1507: * level protocol in the kernel.
1508: * Second argument is buffer to which options
1509: * will be moved, and return value is their length.
1510: * XXX should be deleted; last arg currently ignored.
1511: */
1.8 mycroft 1512: void
1.1 cgd 1513: ip_stripoptions(m, mopt)
1.109 augustss 1514: struct mbuf *m;
1.1 cgd 1515: struct mbuf *mopt;
1516: {
1.109 augustss 1517: int i;
1.1 cgd 1518: struct ip *ip = mtod(m, struct ip *);
1.109 augustss 1519: caddr_t opts;
1.1 cgd 1520: int olen;
1521:
1.79 mycroft 1522: olen = (ip->ip_hl << 2) - sizeof (struct ip);
1.1 cgd 1523: opts = (caddr_t)(ip + 1);
1524: i = m->m_len - (sizeof (struct ip) + olen);
1525: bcopy(opts + olen, opts, (unsigned)i);
1526: m->m_len -= olen;
1527: if (m->m_flags & M_PKTHDR)
1528: m->m_pkthdr.len -= olen;
1.155 itojun 1529: ip->ip_len = htons(ntohs(ip->ip_len) - olen);
1.79 mycroft 1530: ip->ip_hl = sizeof (struct ip) >> 2;
1.1 cgd 1531: }
1532:
1.139 matt 1533: const int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1534: 0, 0, 0, 0,
1535: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1536: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1537: EMSGSIZE, EHOSTUNREACH, 0, 0,
1538: 0, 0, 0, 0,
1539: ENOPROTOOPT
1540: };
1541:
1542: /*
1543: * Forward a packet. If some error occurs return the sender
1544: * an icmp packet. Note we can't always generate a meaningful
1545: * icmp message because icmp doesn't have a large enough repertoire
1546: * of codes and types.
1547: *
1548: * If not forwarding, just drop the packet. This could be confusing
1549: * if ipforwarding was zero but some routing protocol was advancing
1550: * us as a gateway to somewhere. However, we must let the routing
1551: * protocol deal with that.
1552: *
1553: * The srcrt parameter indicates whether the packet is being forwarded
1554: * via a source route.
1555: */
1.13 mycroft 1556: void
1.1 cgd 1557: ip_forward(m, srcrt)
1558: struct mbuf *m;
1559: int srcrt;
1560: {
1.109 augustss 1561: struct ip *ip = mtod(m, struct ip *);
1562: struct sockaddr_in *sin;
1563: struct rtentry *rt;
1.28 christos 1564: int error, type = 0, code = 0;
1.1 cgd 1565: struct mbuf *mcopy;
1.13 mycroft 1566: n_long dest;
1567: struct ifnet *destifp;
1.89 itojun 1568: #ifdef IPSEC
1569: struct ifnet dummyifp;
1570: #endif
1.164 matt 1571:
1572: /*
1573: * We are now in the output path.
1574: */
1575: MCLAIM(m, &ip_tx_mowner);
1.135 thorpej 1576:
1577: /*
1578: * Clear any in-bound checksum flags for this packet.
1579: */
1580: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1581:
1.13 mycroft 1582: dest = 0;
1.1 cgd 1583: #ifdef DIAGNOSTIC
1584: if (ipprintfs)
1.70 thorpej 1585: printf("forward: src %2.2x dst %2.2x ttl %x\n",
1586: ntohl(ip->ip_src.s_addr),
1587: ntohl(ip->ip_dst.s_addr), ip->ip_ttl);
1.1 cgd 1588: #endif
1.93 sommerfe 1589: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.1 cgd 1590: ipstat.ips_cantforward++;
1591: m_freem(m);
1592: return;
1593: }
1594: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1595: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1596: return;
1597: }
1598: ip->ip_ttl -= IPTTLDEC;
1599:
1.19 mycroft 1600: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1601: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1602: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1603: if (ipforward_rt.ro_rt) {
1604: RTFREE(ipforward_rt.ro_rt);
1605: ipforward_rt.ro_rt = 0;
1606: }
1607: sin->sin_family = AF_INET;
1.35 mycroft 1608: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1609: sin->sin_addr = ip->ip_dst;
1610:
1611: rtalloc(&ipforward_rt);
1612: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1613: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1614: return;
1615: }
1616: rt = ipforward_rt.ro_rt;
1617: }
1618:
1619: /*
1.34 mycroft 1620: * Save at most 68 bytes of the packet in case
1.1 cgd 1621: * we need to generate an ICMP message to the src.
1.119 itojun 1622: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1623: */
1.155 itojun 1624: mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1.119 itojun 1625: if (mcopy)
1626: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1627:
1628: /*
1629: * If forwarding packet using same interface that it came in on,
1630: * perhaps should send a redirect to sender to shortcut a hop.
1631: * Only send redirect if source is sending directly to us,
1632: * and if packet was not source routed (or has any options).
1633: * Also, don't send redirect if forwarding using a default route
1634: * or a route modified by a redirect.
1635: */
1636: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1637: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1638: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1639: ipsendredirects && !srcrt) {
1.19 mycroft 1640: if (rt->rt_ifa &&
1641: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1642: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1643: if (rt->rt_flags & RTF_GATEWAY)
1644: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1645: else
1646: dest = ip->ip_dst.s_addr;
1647: /*
1648: * Router requirements says to only send host
1649: * redirects.
1650: */
1651: type = ICMP_REDIRECT;
1652: code = ICMP_REDIRECT_HOST;
1.1 cgd 1653: #ifdef DIAGNOSTIC
1.77 thorpej 1654: if (ipprintfs)
1655: printf("redirect (%d) to %x\n", code,
1656: (u_int32_t)dest);
1.1 cgd 1657: #endif
1658: }
1659: }
1660:
1.89 itojun 1661: #ifdef IPSEC
1.134 lukem 1662: /* Don't lookup socket in forwarding case */
1.127 itojun 1663: (void)ipsec_setsocket(m, NULL);
1.103 itojun 1664: #endif
1.27 thorpej 1665: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1666: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1.1 cgd 1667: if (error)
1668: ipstat.ips_cantforward++;
1669: else {
1670: ipstat.ips_forward++;
1671: if (type)
1672: ipstat.ips_redirectsent++;
1673: else {
1.63 matt 1674: if (mcopy) {
1675: #ifdef GATEWAY
1.64 thorpej 1676: if (mcopy->m_flags & M_CANFASTFWD)
1677: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1678: #endif
1.1 cgd 1679: m_freem(mcopy);
1.63 matt 1680: }
1.1 cgd 1681: return;
1682: }
1683: }
1684: if (mcopy == NULL)
1685: return;
1.13 mycroft 1686: destifp = NULL;
1687:
1.1 cgd 1688: switch (error) {
1689:
1690: case 0: /* forwarded, but need redirect */
1691: /* type, code set above */
1692: break;
1693:
1694: case ENETUNREACH: /* shouldn't happen, checked above */
1695: case EHOSTUNREACH:
1696: case ENETDOWN:
1697: case EHOSTDOWN:
1698: default:
1699: type = ICMP_UNREACH;
1700: code = ICMP_UNREACH_HOST;
1701: break;
1702:
1703: case EMSGSIZE:
1704: type = ICMP_UNREACH;
1705: code = ICMP_UNREACH_NEEDFRAG;
1.89 itojun 1706: #ifndef IPSEC
1.13 mycroft 1707: if (ipforward_rt.ro_rt)
1708: destifp = ipforward_rt.ro_rt->rt_ifp;
1.89 itojun 1709: #else
1710: /*
1711: * If the packet is routed over IPsec tunnel, tell the
1712: * originator the tunnel MTU.
1713: * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1714: * XXX quickhack!!!
1715: */
1716: if (ipforward_rt.ro_rt) {
1717: struct secpolicy *sp;
1718: int ipsecerror;
1.95 itojun 1719: size_t ipsechdr;
1.89 itojun 1720: struct route *ro;
1721:
1722: sp = ipsec4_getpolicybyaddr(mcopy,
1.95 itojun 1723: IPSEC_DIR_OUTBOUND,
1724: IP_FORWARDING,
1725: &ipsecerror);
1.89 itojun 1726:
1727: if (sp == NULL)
1728: destifp = ipforward_rt.ro_rt->rt_ifp;
1729: else {
1730: /* count IPsec header size */
1.95 itojun 1731: ipsechdr = ipsec4_hdrsiz(mcopy,
1732: IPSEC_DIR_OUTBOUND,
1733: NULL);
1.89 itojun 1734:
1735: /*
1736: * find the correct route for outer IPv4
1737: * header, compute tunnel MTU.
1738: *
1739: * XXX BUG ALERT
1740: * The "dummyifp" code relies upon the fact
1741: * that icmp_error() touches only ifp->if_mtu.
1742: */
1743: /*XXX*/
1744: destifp = NULL;
1745: if (sp->req != NULL
1.95 itojun 1746: && sp->req->sav != NULL
1747: && sp->req->sav->sah != NULL) {
1748: ro = &sp->req->sav->sah->sa_route;
1.89 itojun 1749: if (ro->ro_rt && ro->ro_rt->rt_ifp) {
1750: dummyifp.if_mtu =
1.151 itojun 1751: ro->ro_rt->rt_rmx.rmx_mtu ?
1752: ro->ro_rt->rt_rmx.rmx_mtu :
1.89 itojun 1753: ro->ro_rt->rt_ifp->if_mtu;
1754: dummyifp.if_mtu -= ipsechdr;
1755: destifp = &dummyifp;
1756: }
1757: }
1758:
1759: key_freesp(sp);
1760: }
1761: }
1762: #endif /*IPSEC*/
1.1 cgd 1763: ipstat.ips_cantfrag++;
1764: break;
1765:
1766: case ENOBUFS:
1.143 itojun 1767: #if 1
1768: /*
1769: * a router should not generate ICMP_SOURCEQUENCH as
1770: * required in RFC1812 Requirements for IP Version 4 Routers.
1771: * source quench could be a big problem under DoS attacks,
1.149 wiz 1772: * or if the underlying interface is rate-limited.
1.143 itojun 1773: */
1774: if (mcopy)
1775: m_freem(mcopy);
1776: return;
1777: #else
1.1 cgd 1778: type = ICMP_SOURCEQUENCH;
1779: code = 0;
1780: break;
1.143 itojun 1781: #endif
1.1 cgd 1782: }
1.13 mycroft 1783: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 1784: }
1785:
1786: void
1787: ip_savecontrol(inp, mp, ip, m)
1.109 augustss 1788: struct inpcb *inp;
1789: struct mbuf **mp;
1790: struct ip *ip;
1791: struct mbuf *m;
1.44 thorpej 1792: {
1793:
1794: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1795: struct timeval tv;
1796:
1797: microtime(&tv);
1798: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1799: SCM_TIMESTAMP, SOL_SOCKET);
1800: if (*mp)
1801: mp = &(*mp)->m_next;
1802: }
1803: if (inp->inp_flags & INP_RECVDSTADDR) {
1804: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1805: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1806: if (*mp)
1807: mp = &(*mp)->m_next;
1808: }
1809: #ifdef notyet
1810: /*
1811: * XXX
1812: * Moving these out of udp_input() made them even more broken
1813: * than they already were.
1814: * - fenner@parc.xerox.com
1815: */
1816: /* options were tossed already */
1817: if (inp->inp_flags & INP_RECVOPTS) {
1818: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1819: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1820: if (*mp)
1821: mp = &(*mp)->m_next;
1822: }
1823: /* ip_srcroute doesn't do what we want here, need to fix */
1824: if (inp->inp_flags & INP_RECVRETOPTS) {
1825: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1826: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1827: if (*mp)
1828: mp = &(*mp)->m_next;
1829: }
1830: #endif
1831: if (inp->inp_flags & INP_RECVIF) {
1832: struct sockaddr_dl sdl;
1833:
1834: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1835: sdl.sdl_family = AF_LINK;
1836: sdl.sdl_index = m->m_pkthdr.rcvif ?
1837: m->m_pkthdr.rcvif->if_index : 0;
1838: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1839: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1840: IP_RECVIF, IPPROTO_IP);
1841: if (*mp)
1842: mp = &(*mp)->m_next;
1843: }
1.13 mycroft 1844: }
1845:
1846: int
1847: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1848: int *name;
1849: u_int namelen;
1850: void *oldp;
1851: size_t *oldlenp;
1852: void *newp;
1853: size_t newlen;
1854: {
1.88 sommerfe 1855: extern int subnetsarelocal, hostzeroisbroadcast;
1.52 thorpej 1856:
1.54 lukem 1857: int error, old;
1858:
1.13 mycroft 1859: /* All sysctl names at this level are terminal. */
1860: if (namelen != 1)
1861: return (ENOTDIR);
1862:
1863: switch (name[0]) {
1864: case IPCTL_FORWARDING:
1865: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1866: case IPCTL_SENDREDIRECTS:
1867: return (sysctl_int(oldp, oldlenp, newp, newlen,
1868: &ipsendredirects));
1869: case IPCTL_DEFTTL:
1870: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1871: #ifdef notyet
1872: case IPCTL_DEFMTU:
1873: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1874: #endif
1.26 thorpej 1875: case IPCTL_FORWSRCRT:
1.47 cjs 1876: /* Don't allow this to change in a secure environment. */
1.26 thorpej 1877: if (securelevel > 0)
1.46 cjs 1878: return (sysctl_rdint(oldp, oldlenp, newp,
1879: ip_forwsrcrt));
1880: else
1881: return (sysctl_int(oldp, oldlenp, newp, newlen,
1882: &ip_forwsrcrt));
1.27 thorpej 1883: case IPCTL_DIRECTEDBCAST:
1884: return (sysctl_int(oldp, oldlenp, newp, newlen,
1885: &ip_directedbcast));
1.47 cjs 1886: case IPCTL_ALLOWSRCRT:
1887: return (sysctl_int(oldp, oldlenp, newp, newlen,
1888: &ip_allowsrcrt));
1.52 thorpej 1889: case IPCTL_SUBNETSARELOCAL:
1890: return (sysctl_int(oldp, oldlenp, newp, newlen,
1891: &subnetsarelocal));
1.53 kml 1892: case IPCTL_MTUDISC:
1.60 kml 1893: error = sysctl_int(oldp, oldlenp, newp, newlen,
1894: &ip_mtudisc);
1.163 itojun 1895: if (error == 0 && ip_mtudisc == 0)
1896: rt_timer_queue_remove_all(ip_mtudisc_timeout_q, TRUE);
1.60 kml 1897: return error;
1.54 lukem 1898: case IPCTL_ANONPORTMIN:
1899: old = anonportmin;
1900: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1.118 itojun 1901: if (anonportmin >= anonportmax || anonportmin < 0
1902: || anonportmin > 65535
1.54 lukem 1903: #ifndef IPNOPRIVPORTS
1904: || anonportmin < IPPORT_RESERVED
1905: #endif
1906: ) {
1907: anonportmin = old;
1908: return (EINVAL);
1909: }
1910: return (error);
1911: case IPCTL_ANONPORTMAX:
1912: old = anonportmax;
1913: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1.118 itojun 1914: if (anonportmin >= anonportmax || anonportmax < 0
1915: || anonportmax > 65535
1.54 lukem 1916: #ifndef IPNOPRIVPORTS
1917: || anonportmax < IPPORT_RESERVED
1918: #endif
1919: ) {
1920: anonportmax = old;
1921: return (EINVAL);
1922: }
1.60 kml 1923: return (error);
1924: case IPCTL_MTUDISCTIMEOUT:
1.158 itojun 1925: old = ip_mtudisc_timeout;
1.60 kml 1926: error = sysctl_int(oldp, oldlenp, newp, newlen,
1927: &ip_mtudisc_timeout);
1.158 itojun 1928: if (ip_mtudisc_timeout < 0) {
1929: ip_mtudisc_timeout = old;
1.156 itojun 1930: return (EINVAL);
1.158 itojun 1931: }
1.163 itojun 1932: if (error == 0)
1.162 itojun 1933: rt_timer_queue_change(ip_mtudisc_timeout_q,
1934: ip_mtudisc_timeout);
1.54 lukem 1935: return (error);
1.65 matt 1936: #ifdef GATEWAY
1937: case IPCTL_MAXFLOWS:
1.67 thorpej 1938: {
1939: int s;
1940:
1.65 matt 1941: error = sysctl_int(oldp, oldlenp, newp, newlen,
1942: &ip_maxflows);
1.67 thorpej 1943: s = splsoftnet();
1.65 matt 1944: ipflow_reap(0);
1.67 thorpej 1945: splx(s);
1.65 matt 1946: return (error);
1.67 thorpej 1947: }
1.89 itojun 1948: #endif
1.90 itojun 1949: case IPCTL_HOSTZEROBROADCAST:
1950: return (sysctl_int(oldp, oldlenp, newp, newlen,
1951: &hostzeroisbroadcast));
1.89 itojun 1952: #if NGIF > 0
1953: case IPCTL_GIF_TTL:
1.157 itojun 1954: return (sysctl_int(oldp, oldlenp, newp, newlen,
1.90 itojun 1955: &ip_gif_ttl));
1.144 martin 1956: #endif
1957:
1958: #if NGRE > 0
1959: case IPCTL_GRE_TTL:
1.157 itojun 1960: return (sysctl_int(oldp, oldlenp, newp, newlen,
1.144 martin 1961: &ip_gre_ttl));
1.117 tron 1962: #endif
1963:
1964: #ifndef IPNOPRIVPORTS
1965: case IPCTL_LOWPORTMIN:
1966: old = lowportmin;
1967: error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmin);
1968: if (lowportmin >= lowportmax
1969: || lowportmin > IPPORT_RESERVEDMAX
1970: || lowportmin < IPPORT_RESERVEDMIN
1971: ) {
1972: lowportmin = old;
1973: return (EINVAL);
1974: }
1975: return (error);
1976: case IPCTL_LOWPORTMAX:
1977: old = lowportmax;
1978: error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmax);
1979: if (lowportmin >= lowportmax
1980: || lowportmax > IPPORT_RESERVEDMAX
1981: || lowportmax < IPPORT_RESERVEDMIN
1982: ) {
1983: lowportmax = old;
1984: return (EINVAL);
1985: }
1986: return (error);
1.65 matt 1987: #endif
1.131 itojun 1988:
1989: case IPCTL_MAXFRAGPACKETS:
1990: return (sysctl_int(oldp, oldlenp, newp, newlen,
1991: &ip_maxfragpackets));
1.88 sommerfe 1992:
1.165 christos 1993: case IPCTL_CHECKINTERFACE:
1994: return (sysctl_int(oldp, oldlenp, newp, newlen,
1995: &ip_checkinterface));
1.13 mycroft 1996: default:
1997: return (EOPNOTSUPP);
1998: }
1999: /* NOTREACHED */
1.1 cgd 2000: }
CVSweb <webmaster@jp.NetBSD.org>