Annotation of src/sys/netinet/ip_input.c, Revision 1.288
1.288 ! rmind 1: /* $NetBSD: ip_input.c,v 1.287 2010/07/09 18:42:46 rmind 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: *
49: * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
50: * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
51: * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
52: * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
53: * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
54: * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
55: * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
56: * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
57: * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
58: * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
59: * POSSIBILITY OF SUCH DAMAGE.
60: */
1.14 cgd 61:
1.1 cgd 62: /*
1.13 mycroft 63: * Copyright (c) 1982, 1986, 1988, 1993
64: * The Regents of the University of California. All rights reserved.
1.1 cgd 65: *
66: * Redistribution and use in source and binary forms, with or without
67: * modification, are permitted provided that the following conditions
68: * are met:
69: * 1. Redistributions of source code must retain the above copyright
70: * notice, this list of conditions and the following disclaimer.
71: * 2. Redistributions in binary form must reproduce the above copyright
72: * notice, this list of conditions and the following disclaimer in the
73: * documentation and/or other materials provided with the distribution.
1.172 agc 74: * 3. Neither the name of the University nor the names of its contributors
1.1 cgd 75: * may be used to endorse or promote products derived from this software
76: * without specific prior written permission.
77: *
78: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
79: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
80: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
81: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
82: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
83: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
84: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
85: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
86: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
87: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
88: * SUCH DAMAGE.
89: *
1.14 cgd 90: * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
1.1 cgd 91: */
1.141 lukem 92:
93: #include <sys/cdefs.h>
1.288 ! rmind 94: __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.287 2010/07/09 18:42:46 rmind Exp $");
1.55 scottr 95:
1.184 jonathan 96: #include "opt_inet.h"
1.278 christos 97: #include "opt_compat_netbsd.h"
1.62 matt 98: #include "opt_gateway.h"
1.69 mrg 99: #include "opt_pfil_hooks.h"
1.91 thorpej 100: #include "opt_ipsec.h"
1.55 scottr 101: #include "opt_mrouting.h"
1.167 martin 102: #include "opt_mbuftrace.h"
1.135 thorpej 103: #include "opt_inet_csum.h"
1.1 cgd 104:
1.5 mycroft 105: #include <sys/param.h>
106: #include <sys/systm.h>
107: #include <sys/mbuf.h>
108: #include <sys/domain.h>
109: #include <sys/protosw.h>
110: #include <sys/socket.h>
1.44 thorpej 111: #include <sys/socketvar.h>
1.5 mycroft 112: #include <sys/errno.h>
113: #include <sys/time.h>
114: #include <sys/kernel.h>
1.72 thorpej 115: #include <sys/pool.h>
1.28 christos 116: #include <sys/sysctl.h>
1.230 elad 117: #include <sys/kauth.h>
1.1 cgd 118:
1.5 mycroft 119: #include <net/if.h>
1.44 thorpej 120: #include <net/if_dl.h>
1.5 mycroft 121: #include <net/route.h>
1.45 mrg 122: #include <net/pfil.h>
1.1 cgd 123:
1.5 mycroft 124: #include <netinet/in.h>
125: #include <netinet/in_systm.h>
126: #include <netinet/ip.h>
127: #include <netinet/in_pcb.h>
1.215 yamt 128: #include <netinet/in_proto.h>
1.5 mycroft 129: #include <netinet/in_var.h>
130: #include <netinet/ip_var.h>
1.266 thorpej 131: #include <netinet/ip_private.h>
1.5 mycroft 132: #include <netinet/ip_icmp.h>
1.89 itojun 133: /* just for gif_ttl */
134: #include <netinet/in_gif.h>
135: #include "gif.h"
1.144 martin 136: #include <net/if_gre.h>
137: #include "gre.h"
1.111 jdolecek 138:
139: #ifdef MROUTING
140: #include <netinet/ip_mroute.h>
141: #endif
1.89 itojun 142:
143: #ifdef IPSEC
144: #include <netinet6/ipsec.h>
1.267 thorpej 145: #include <netinet6/ipsec_private.h>
1.89 itojun 146: #include <netkey/key.h>
147: #endif
1.173 jonathan 148: #ifdef FAST_IPSEC
149: #include <netipsec/ipsec.h>
150: #include <netipsec/key.h>
151: #endif /* FAST_IPSEC*/
1.44 thorpej 152:
1.1 cgd 153: #ifndef IPFORWARDING
154: #ifdef GATEWAY
155: #define IPFORWARDING 1 /* forward IP packets not for us */
156: #else /* GATEWAY */
157: #define IPFORWARDING 0 /* don't forward IP packets not for us */
158: #endif /* GATEWAY */
159: #endif /* IPFORWARDING */
160: #ifndef IPSENDREDIRECTS
161: #define IPSENDREDIRECTS 1
162: #endif
1.26 thorpej 163: #ifndef IPFORWSRCRT
1.47 cjs 164: #define IPFORWSRCRT 1 /* forward source-routed packets */
165: #endif
166: #ifndef IPALLOWSRCRT
1.48 mrg 167: #define IPALLOWSRCRT 1 /* allow source-routed packets */
1.26 thorpej 168: #endif
1.53 kml 169: #ifndef IPMTUDISC
1.153 itojun 170: #define IPMTUDISC 1
1.53 kml 171: #endif
1.60 kml 172: #ifndef IPMTUDISCTIMEOUT
1.61 kml 173: #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */
1.60 kml 174: #endif
1.53 kml 175:
1.278 christos 176: #ifdef COMPAT_50
177: #include <compat/sys/time.h>
178: #include <compat/sys/socket.h>
179: #endif
180:
1.27 thorpej 181: /*
182: * Note: DIRECTED_BROADCAST is handled this way so that previous
183: * configuration using this option will Just Work.
184: */
185: #ifndef IPDIRECTEDBCAST
186: #ifdef DIRECTED_BROADCAST
187: #define IPDIRECTEDBCAST 1
188: #else
189: #define IPDIRECTEDBCAST 0
190: #endif /* DIRECTED_BROADCAST */
191: #endif /* IPDIRECTEDBCAST */
1.1 cgd 192: int ipforwarding = IPFORWARDING;
193: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 194: int ip_defttl = IPDEFTTL;
1.26 thorpej 195: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 196: int ip_directedbcast = IPDIRECTEDBCAST;
1.47 cjs 197: int ip_allowsrcrt = IPALLOWSRCRT;
1.53 kml 198: int ip_mtudisc = IPMTUDISC;
1.156 itojun 199: int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
1.1 cgd 200: #ifdef DIAGNOSTIC
201: int ipprintfs = 0;
202: #endif
1.184 jonathan 203:
204: int ip_do_randomid = 0;
205:
1.165 christos 206: /*
207: * XXX - Setting ip_checkinterface mostly implements the receive side of
208: * the Strong ES model described in RFC 1122, but since the routing table
209: * and transmit implementation do not implement the Strong ES model,
210: * setting this to 1 results in an odd hybrid.
211: *
212: * XXX - ip_checkinterface currently must be disabled if you use ipnat
213: * to translate the destination address to another local interface.
214: *
215: * XXX - ip_checkinterface must be disabled if you add IP aliases
216: * to the loopback interface instead of the interface where the
217: * packets for those addresses are received.
218: */
219: int ip_checkinterface = 0;
220:
1.1 cgd 221:
1.60 kml 222: struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
223:
1.1 cgd 224: int ipqmaxlen = IFQ_MAXLEN;
1.150 matt 225: u_long in_ifaddrhash; /* size of hash table - 1 */
226: int in_ifaddrentries; /* total number of addrs */
1.212 perry 227: struct in_ifaddrhead in_ifaddrhead;
1.57 tls 228: struct in_ifaddrhashhead *in_ifaddrhashtbl;
1.166 matt 229: u_long in_multihash; /* size of hash table - 1 */
230: int in_multientries; /* total number of addrs */
231: struct in_multihashhead *in_multihashtbl;
1.13 mycroft 232: struct ifqueue ipintrq;
1.286 tls 233:
1.183 jonathan 234: uint16_t ip_id;
1.75 thorpej 235:
1.266 thorpej 236: percpu_t *ipstat_percpu;
237:
1.121 thorpej 238: #ifdef PFIL_HOOKS
239: struct pfil_head inet_pfil_hook;
240: #endif
241:
1.275 pooka 242: struct pool inmulti_pool;
1.72 thorpej 243:
1.135 thorpej 244: #ifdef INET_CSUM_COUNTERS
245: #include <sys/device.h>
246:
247: struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
248: NULL, "inet", "hwcsum bad");
249: struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
250: NULL, "inet", "hwcsum ok");
251: struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
252: NULL, "inet", "swcsum");
253:
254: #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++
255:
1.201 matt 256: EVCNT_ATTACH_STATIC(ip_hwcsum_bad);
257: EVCNT_ATTACH_STATIC(ip_hwcsum_ok);
258: EVCNT_ATTACH_STATIC(ip_swcsum);
259:
1.135 thorpej 260: #else
261:
262: #define INET_CSUM_COUNTER_INCR(ev) /* nothing */
263:
264: #endif /* INET_CSUM_COUNTERS */
265:
1.1 cgd 266: /*
267: * We need to save the IP options in case a protocol wants to respond
268: * to an incoming packet over the same route if the packet got here
269: * using IP source routing. This allows connection establishment and
270: * maintenance when the remote end is on a network that is not known
271: * to us.
272: */
273: int ip_nhops = 0;
274: static struct ip_srcrt {
275: struct in_addr dst; /* final destination */
276: char nop; /* one NOP to align */
277: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
278: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
279: } ip_srcrt;
280:
1.210 perry 281: static void save_rte(u_char *, struct in_addr);
1.35 mycroft 282:
1.164 matt 283: #ifdef MBUFTRACE
1.234 dogcow 284: struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx");
285: struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx");
1.164 matt 286: #endif
287:
1.284 pooka 288: static void sysctl_net_inet_ip_setup(struct sysctllog **);
289:
1.1 cgd 290: /*
291: * IP initialization: fill in IP protocol switch table.
292: * All protocols not implemented in kernel go to raw IP protocol handler.
293: */
1.8 mycroft 294: void
1.211 perry 295: ip_init(void)
1.1 cgd 296: {
1.199 matt 297: const struct protosw *pr;
1.109 augustss 298: int i;
1.1 cgd 299:
1.284 pooka 300: sysctl_net_inet_ip_setup(NULL);
301:
1.275 pooka 302: pool_init(&inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl",
303: NULL, IPL_SOFTNET);
304:
1.1 cgd 305: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
306: if (pr == 0)
307: panic("ip_init");
308: for (i = 0; i < IPPROTO_MAX; i++)
309: ip_protox[i] = pr - inetsw;
310: for (pr = inetdomain.dom_protosw;
311: pr < inetdomain.dom_protoswNPROTOSW; pr++)
312: if (pr->pr_domain->dom_family == PF_INET &&
313: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
314: ip_protox[pr->pr_protocol] = pr - inetsw;
1.192 jonathan 315:
1.288 ! rmind 316: ip_reass_init();
1.190 jonathan 317:
1.262 matt 318: ip_initid();
1.227 kardel 319: ip_id = time_second & 0xfffff;
1.194 jonathan 320:
1.1 cgd 321: ipintrq.ifq_maxlen = ipqmaxlen;
1.194 jonathan 322:
1.181 jonathan 323: TAILQ_INIT(&in_ifaddrhead);
1.272 ad 324: in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true,
325: &in_ifaddrhash);
326: in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true,
327: &in_multihash);
1.160 itojun 328: ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout);
1.73 thorpej 329: #ifdef GATEWAY
1.248 liamjfoy 330: ipflow_init(ip_hashsize);
1.73 thorpej 331: #endif
1.121 thorpej 332:
333: #ifdef PFIL_HOOKS
334: /* Register our Packet Filter hook. */
1.126 thorpej 335: inet_pfil_hook.ph_type = PFIL_TYPE_AF;
336: inet_pfil_hook.ph_af = AF_INET;
1.121 thorpej 337: i = pfil_head_register(&inet_pfil_hook);
338: if (i != 0)
339: printf("ip_init: WARNING: unable to register pfil hook, "
340: "error %d\n", i);
341: #endif /* PFIL_HOOKS */
1.135 thorpej 342:
1.164 matt 343: #ifdef MBUFTRACE
344: MOWNER_ATTACH(&ip_tx_mowner);
345: MOWNER_ATTACH(&ip_rx_mowner);
346: #endif /* MBUFTRACE */
1.266 thorpej 347:
348: ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS);
1.1 cgd 349: }
350:
1.229 christos 351: struct sockaddr_in ipaddr = {
352: .sin_len = sizeof(ipaddr),
353: .sin_family = AF_INET,
354: };
1.1 cgd 355: struct route ipforward_rt;
356:
357: /*
1.89 itojun 358: * IP software interrupt routine
359: */
360: void
1.211 perry 361: ipintr(void)
1.89 itojun 362: {
363: int s;
364: struct mbuf *m;
1.286 tls 365: struct ifqueue lcl_intrq;
366:
367: memset(&lcl_intrq, 0, sizeof(lcl_intrq));
368: ipintrq.ifq_maxlen = ipqmaxlen;
1.89 itojun 369:
1.268 ad 370: mutex_enter(softnet_lock);
371: KERNEL_LOCK(1, NULL);
1.286 tls 372: if (!IF_IS_EMPTY(&ipintrq)) {
1.132 thorpej 373: s = splnet();
1.286 tls 374:
375: /* Take existing queue onto stack */
376: lcl_intrq = ipintrq;
377:
378: /* Zero out global queue, preserving maxlen and drops */
379: ipintrq.ifq_head = NULL;
380: ipintrq.ifq_tail = NULL;
381: ipintrq.ifq_len = 0;
382: ipintrq.ifq_maxlen = lcl_intrq.ifq_maxlen;
383: ipintrq.ifq_drops = lcl_intrq.ifq_drops;
384:
1.89 itojun 385: splx(s);
1.286 tls 386: }
387: KERNEL_UNLOCK_ONE(NULL);
388: while (!IF_IS_EMPTY(&lcl_intrq)) {
389: IF_DEQUEUE(&lcl_intrq, m);
1.268 ad 390: if (m == NULL)
391: break;
1.89 itojun 392: ip_input(m);
393: }
1.268 ad 394: mutex_exit(softnet_lock);
1.89 itojun 395: }
396:
397: /*
1.1 cgd 398: * Ip input routine. Checksum and byte swap header. If fragmented
399: * try to reassemble. Process options. Pass to next level.
400: */
1.8 mycroft 401: void
1.89 itojun 402: ip_input(struct mbuf *m)
1.1 cgd 403: {
1.109 augustss 404: struct ip *ip = NULL;
405: struct in_ifaddr *ia;
406: struct ifaddr *ifa;
1.288 ! rmind 407: int hlen = 0, len;
1.100 itojun 408: int downmatch;
1.165 christos 409: int checkif;
1.169 itojun 410: int srcrt = 0;
1.173 jonathan 411: #ifdef FAST_IPSEC
412: struct m_tag *mtag;
413: struct tdb_ident *tdbi;
414: struct secpolicy *sp;
1.233 tls 415: int error;
1.173 jonathan 416: #endif /* FAST_IPSEC */
1.1 cgd 417:
1.164 matt 418: MCLAIM(m, &ip_rx_mowner);
1.1 cgd 419: #ifdef DIAGNOSTIC
420: if ((m->m_flags & M_PKTHDR) == 0)
421: panic("ipintr no HDR");
1.89 itojun 422: #endif
1.164 matt 423:
1.1 cgd 424: /*
425: * If no IP addresses have been set yet but the interfaces
426: * are receiving, can't do anything with incoming packets yet.
427: */
1.181 jonathan 428: if (TAILQ_FIRST(&in_ifaddrhead) == 0)
1.1 cgd 429: goto bad;
1.266 thorpej 430: IP_STATINC(IP_STAT_TOTAL);
1.154 thorpej 431: /*
432: * If the IP header is not aligned, slurp it up into a new
433: * mbuf with space for link headers, in the event we forward
434: * it. Otherwise, if it is aligned, make sure the entire
435: * base IP header is in the first mbuf of the chain.
436: */
1.244 christos 437: if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
1.154 thorpej 438: if ((m = m_copyup(m, sizeof(struct ip),
439: (max_linkhdr + 3) & ~3)) == NULL) {
440: /* XXXJRT new stat, please */
1.266 thorpej 441: IP_STATINC(IP_STAT_TOOSMALL);
1.154 thorpej 442: return;
443: }
444: } else if (__predict_false(m->m_len < sizeof (struct ip))) {
445: if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
1.266 thorpej 446: IP_STATINC(IP_STAT_TOOSMALL);
1.154 thorpej 447: return;
448: }
1.1 cgd 449: }
450: ip = mtod(m, struct ip *);
1.13 mycroft 451: if (ip->ip_v != IPVERSION) {
1.266 thorpej 452: IP_STATINC(IP_STAT_BADVERS);
1.13 mycroft 453: goto bad;
454: }
1.1 cgd 455: hlen = ip->ip_hl << 2;
456: if (hlen < sizeof(struct ip)) { /* minimum header length */
1.266 thorpej 457: IP_STATINC(IP_STAT_BADHLEN);
1.1 cgd 458: goto bad;
459: }
460: if (hlen > m->m_len) {
461: if ((m = m_pullup(m, hlen)) == 0) {
1.266 thorpej 462: IP_STATINC(IP_STAT_BADHLEN);
1.89 itojun 463: return;
1.1 cgd 464: }
465: ip = mtod(m, struct ip *);
466: }
1.98 thorpej 467:
1.85 hwr 468: /*
1.99 thorpej 469: * RFC1122: packets with a multicast source address are
1.98 thorpej 470: * not allowed.
1.85 hwr 471: */
472: if (IN_MULTICAST(ip->ip_src.s_addr)) {
1.266 thorpej 473: IP_STATINC(IP_STAT_BADADDR);
1.85 hwr 474: goto bad;
1.129 itojun 475: }
476:
477: /* 127/8 must not appear on wire - RFC1122 */
478: if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
479: (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
1.130 itojun 480: if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
1.266 thorpej 481: IP_STATINC(IP_STAT_BADADDR);
1.129 itojun 482: goto bad;
1.130 itojun 483: }
1.85 hwr 484: }
485:
1.135 thorpej 486: switch (m->m_pkthdr.csum_flags &
1.137 thorpej 487: ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) |
1.135 thorpej 488: M_CSUM_IPv4_BAD)) {
489: case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
490: INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
491: goto badcsum;
492:
493: case M_CSUM_IPv4:
494: /* Checksum was okay. */
495: INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
496: break;
497:
498: default:
1.206 thorpej 499: /*
500: * Must compute it ourselves. Maybe skip checksum on
501: * loopback interfaces.
502: */
503: if (__predict_true(!(m->m_pkthdr.rcvif->if_flags &
504: IFF_LOOPBACK) || ip_do_loopback_cksum)) {
505: INET_CSUM_COUNTER_INCR(&ip_swcsum);
506: if (in_cksum(m, hlen) != 0)
507: goto badcsum;
508: }
1.135 thorpej 509: break;
1.1 cgd 510: }
511:
1.121 thorpej 512: /* Retrieve the packet length. */
513: len = ntohs(ip->ip_len);
1.81 proff 514:
515: /*
516: * Check for additional length bogosity
517: */
1.84 proff 518: if (len < hlen) {
1.266 thorpej 519: IP_STATINC(IP_STAT_BADLEN);
1.81 proff 520: goto bad;
521: }
1.1 cgd 522:
523: /*
524: * Check that the amount of data in the buffers
525: * is as at least much as the IP header would have us expect.
526: * Trim mbufs if longer than we expect.
527: * Drop packet if shorter than we expect.
528: */
1.35 mycroft 529: if (m->m_pkthdr.len < len) {
1.266 thorpej 530: IP_STATINC(IP_STAT_TOOSHORT);
1.1 cgd 531: goto bad;
532: }
1.35 mycroft 533: if (m->m_pkthdr.len > len) {
1.1 cgd 534: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 535: m->m_len = len;
536: m->m_pkthdr.len = len;
1.1 cgd 537: } else
1.35 mycroft 538: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 539: }
540:
1.193 scw 541: #if defined(IPSEC)
1.149 wiz 542: /* ipflow (IP fast forwarding) is not compatible with IPsec. */
1.94 itojun 543: m->m_flags &= ~M_CANFASTFWD;
544: #else
1.64 thorpej 545: /*
546: * Assume that we can create a fast-forward IP flow entry
547: * based on this packet.
548: */
549: m->m_flags |= M_CANFASTFWD;
1.94 itojun 550: #endif
1.64 thorpej 551:
1.36 mrg 552: #ifdef PFIL_HOOKS
1.33 mrg 553: /*
1.64 thorpej 554: * Run through list of hooks for input packets. If there are any
555: * filters which require that additional packets in the flow are
556: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
557: * Note that filters must _never_ set this flag, as another filter
558: * in the list may have previously cleared it.
1.33 mrg 559: */
1.127 itojun 560: /*
561: * let ipfilter look at packet on the wire,
562: * not the decapsulated packet.
563: */
564: #ifdef IPSEC
1.136 itojun 565: if (!ipsec_getnhist(m))
1.186 scw 566: #elif defined(FAST_IPSEC)
567: if (!ipsec_indone(m))
1.127 itojun 568: #else
569: if (1)
570: #endif
571: {
1.169 itojun 572: struct in_addr odst;
573:
574: odst = ip->ip_dst;
1.127 itojun 575: if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif,
1.168 itojun 576: PFIL_IN) != 0)
577: return;
1.127 itojun 578: if (m == NULL)
579: return;
580: ip = mtod(m, struct ip *);
1.142 darrenr 581: hlen = ip->ip_hl << 2;
1.205 darrenr 582: /*
583: * XXX The setting of "srcrt" here is to prevent ip_forward()
584: * from generating ICMP redirects for packets that have
585: * been redirected by a hook back out on to the same LAN that
586: * they came from and is not an indication that the packet
587: * is being inffluenced by source routing options. This
588: * allows things like
589: * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp"
590: * where tlp0 is both on the 1.1.1.0/24 network and is the
591: * default route for hosts on 1.1.1.0/24. Of course this
592: * also requires a "map tlp0 ..." to complete the story.
593: * One might argue whether or not this kind of network config.
1.212 perry 594: * should be supported in this manner...
1.205 darrenr 595: */
1.169 itojun 596: srcrt = (odst.s_addr != ip->ip_dst.s_addr);
1.127 itojun 597: }
1.36 mrg 598: #endif /* PFIL_HOOKS */
1.123 thorpej 599:
600: #ifdef ALTQ
601: /* XXX Temporary until ALTQ is changed to use a pfil hook */
602: if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) {
603: /* packet dropped by traffic conditioner */
604: return;
605: }
606: #endif
1.121 thorpej 607:
608: /*
1.1 cgd 609: * Process options and, if not destined for us,
610: * ship it on. ip_dooptions returns 1 when an
611: * error was detected (causing an icmp message
612: * to be sent and the original packet to be freed).
613: */
614: ip_nhops = 0; /* for source routed packets */
615: if (hlen > sizeof (struct ip) && ip_dooptions(m))
1.89 itojun 616: return;
1.1 cgd 617:
618: /*
1.165 christos 619: * Enable a consistency check between the destination address
620: * and the arrival interface for a unicast packet (the RFC 1122
621: * strong ES model) if IP forwarding is disabled and the packet
622: * is not locally generated.
623: *
624: * XXX - Checking also should be disabled if the destination
625: * address is ipnat'ed to a different interface.
626: *
627: * XXX - Checking is incompatible with IP aliases added
628: * to the loopback interface instead of the interface where
629: * the packets are received.
630: *
631: * XXX - We need to add a per ifaddr flag for this so that
632: * we get finer grain control.
633: */
634: checkif = ip_checkinterface && (ipforwarding == 0) &&
635: (m->m_pkthdr.rcvif != NULL) &&
636: ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0);
637:
638: /*
1.1 cgd 639: * Check our list of addresses, to see if the packet is for us.
1.100 itojun 640: *
641: * Traditional 4.4BSD did not consult IFF_UP at all.
642: * The behavior here is to treat addresses on !IFF_UP interface
643: * as not mine.
1.1 cgd 644: */
1.100 itojun 645: downmatch = 0;
1.140 matt 646: LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) {
1.97 itojun 647: if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
1.165 christos 648: if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif)
649: continue;
1.97 itojun 650: if ((ia->ia_ifp->if_flags & IFF_UP) != 0)
651: break;
1.100 itojun 652: else
653: downmatch++;
1.97 itojun 654: }
655: }
1.86 thorpej 656: if (ia != NULL)
657: goto ours;
1.225 christos 658: if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
1.209 matt 659: IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) {
1.140 matt 660: if (ifa->ifa_addr->sa_family != AF_INET)
661: continue;
1.57 tls 662: ia = ifatoia(ifa);
1.35 mycroft 663: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
664: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 665: /*
666: * Look for all-0's host part (old broadcast addr),
667: * either for subnet or net.
668: */
669: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 670: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 671: goto ours;
1.57 tls 672: /*
673: * An interface with IP address zero accepts
674: * all packets that arrive on that interface.
675: */
676: if (in_nullhost(ia->ia_addr.sin_addr))
677: goto ours;
1.1 cgd 678: }
679: }
1.18 mycroft 680: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 681: struct in_multi *inm;
682: #ifdef MROUTING
683: extern struct socket *ip_mrouter;
1.10 brezak 684:
1.4 hpeyerl 685: if (ip_mrouter) {
686: /*
687: * If we are acting as a multicast router, all
688: * incoming multicast packets are passed to the
689: * kernel-level multicast forwarding function.
690: * The packet is returned (relatively) intact; if
691: * ip_mforward() returns a non-zero value, the packet
692: * must be discarded, else it may be accepted below.
693: *
694: * (The IP ident field is put in the same byte order
695: * as expected when ip_mforward() is called from
696: * ip_output().)
697: */
1.13 mycroft 698: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
1.266 thorpej 699: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 700: m_freem(m);
1.89 itojun 701: return;
1.4 hpeyerl 702: }
703:
704: /*
705: * The process-level routing demon needs to receive
706: * all multicast IGMP packets, whether or not this
707: * host belongs to their destination groups.
708: */
709: if (ip->ip_p == IPPROTO_IGMP)
710: goto ours;
1.266 thorpej 711: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 712: }
713: #endif
714: /*
715: * See if we belong to the destination multicast group on the
716: * arrival interface.
717: */
718: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
719: if (inm == NULL) {
1.266 thorpej 720: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 721: m_freem(m);
1.89 itojun 722: return;
1.4 hpeyerl 723: }
724: goto ours;
725: }
1.19 mycroft 726: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 727: in_nullhost(ip->ip_dst))
1.1 cgd 728: goto ours;
729:
730: /*
731: * Not for us; forward if possible and desirable.
732: */
733: if (ipforwarding == 0) {
1.266 thorpej 734: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 735: m_freem(m);
1.100 itojun 736: } else {
737: /*
738: * If ip_dst matched any of my address on !IFF_UP interface,
739: * and there's no IFF_UP interface that matches ip_dst,
740: * send icmp unreach. Forwarding it will result in in-kernel
741: * forwarding loop till TTL goes to 0.
742: */
743: if (downmatch) {
744: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
1.266 thorpej 745: IP_STATINC(IP_STAT_CANTFORWARD);
1.100 itojun 746: return;
747: }
1.145 itojun 748: #ifdef IPSEC
749: if (ipsec4_in_reject(m, NULL)) {
1.267 thorpej 750: IPSEC_STATINC(IPSEC_STAT_IN_POLVIO);
1.145 itojun 751: goto bad;
752: }
753: #endif
1.173 jonathan 754: #ifdef FAST_IPSEC
755: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
756: s = splsoftnet();
757: if (mtag != NULL) {
758: tdbi = (struct tdb_ident *)(mtag + 1);
759: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
760: } else {
761: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
1.212 perry 762: IP_FORWARDING, &error);
1.173 jonathan 763: }
764: if (sp == NULL) { /* NB: can happen if error */
765: splx(s);
766: /*XXX error stat???*/
767: DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/
768: goto bad;
769: }
770:
771: /*
772: * Check security policy against packet attributes.
773: */
774: error = ipsec_in_reject(sp, m);
775: KEY_FREESP(&sp);
776: splx(s);
777: if (error) {
1.266 thorpej 778: IP_STATINC(IP_STAT_CANTFORWARD);
1.173 jonathan 779: goto bad;
1.193 scw 780: }
781:
782: /*
783: * Peek at the outbound SP for this packet to determine if
784: * it's a Fast Forward candidate.
785: */
786: mtag = m_tag_find(m, PACKET_TAG_IPSEC_PENDING_TDB, NULL);
787: if (mtag != NULL)
788: m->m_flags &= ~M_CANFASTFWD;
789: else {
790: s = splsoftnet();
791: sp = ipsec4_checkpolicy(m, IPSEC_DIR_OUTBOUND,
792: (IP_FORWARDING |
793: (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
794: &error, NULL);
795: if (sp != NULL) {
796: m->m_flags &= ~M_CANFASTFWD;
797: KEY_FREESP(&sp);
798: }
799: splx(s);
1.173 jonathan 800: }
801: #endif /* FAST_IPSEC */
1.145 itojun 802:
1.169 itojun 803: ip_forward(m, srcrt);
1.100 itojun 804: }
1.89 itojun 805: return;
1.1 cgd 806:
807: ours:
808: /*
809: * If offset or IP_MF are set, must reassemble.
810: */
1.155 itojun 811: if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
1.288 ! rmind 812: struct ipq *fp;
! 813: u_int off, hash;
! 814: bool mff;
! 815:
1.258 matt 816: /*
817: * Prevent TCP blind data attacks by not allowing non-initial
818: * fragments to start at less than 68 bytes (minimal fragment
1.259 matt 819: * size) and making sure the first fragment is at least 68
820: * bytes.
1.258 matt 821: */
1.260 matt 822: off = (ntohs(ip->ip_off) & IP_OFFMASK) << 3;
1.259 matt 823: if ((off > 0 ? off + hlen : len) < IP_MINFRAGSIZE - 1) {
1.266 thorpej 824: IP_STATINC(IP_STAT_BADFRAGS);
1.258 matt 825: goto bad;
826: }
1.1 cgd 827:
828: /*
1.288 ! rmind 829: * Adjust total IP length to not reflect header. Set 'mff'
! 830: * indicator, if more fragments are expected. Convert offset
! 831: * of this to bytes.
1.1 cgd 832: */
1.155 itojun 833: ip->ip_len = htons(ntohs(ip->ip_len) - hlen);
834: mff = (ip->ip_off & htons(IP_MF)) != 0;
1.25 cgd 835: if (mff) {
1.287 rmind 836: /*
837: * Make sure that fragments have a data length
1.288 ! rmind 838: * which is non-zero and multiple of 8 bytes.
1.287 rmind 839: */
1.155 itojun 840: if (ntohs(ip->ip_len) == 0 ||
841: (ntohs(ip->ip_len) & 0x7) != 0) {
1.266 thorpej 842: IP_STATINC(IP_STAT_BADFRAGS);
1.287 rmind 843: goto bad;
844: }
845: }
846: ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3);
847:
1.288 ! rmind 848: /* Look for queue of fragments of this datagram. */
! 849: fp = ip_reass_lookup(ip, &hash);
! 850:
! 851: /* Make sure the TOS matches previous fragments. */
! 852: if (fp && fp->ipq_tos != ip->ip_tos) {
! 853: IP_STATINC(IP_STAT_BADFRAGS);
! 854: ip_reass_unlock();
! 855: goto bad;
1.16 cgd 856: }
1.1 cgd 857:
858: /*
859: * If datagram marked as having more fragments
860: * or if this is not the first fragment,
861: * attempt reassembly; if it succeeds, proceed.
862: */
1.155 itojun 863: if (mff || ip->ip_off != htons(0)) {
1.288 ! rmind 864: struct ipqent *ipqe;
! 865:
! 866: ipqe = ip_reass_getent();
1.25 cgd 867: if (ipqe == NULL) {
1.266 thorpej 868: IP_STATINC(IP_STAT_RCVMEMDROP);
1.288 ! rmind 869: ip_reass_unlock();
1.25 cgd 870: goto bad;
871: }
872: ipqe->ipqe_mff = mff;
1.50 thorpej 873: ipqe->ipqe_m = m;
1.25 cgd 874: ipqe->ipqe_ip = ip;
1.288 ! rmind 875: m = ip_reass(ipqe, fp, hash);
1.287 rmind 876: if (m == NULL) {
1.89 itojun 877: return;
1.75 thorpej 878: }
1.266 thorpej 879: IP_STATINC(IP_STAT_REASSEMBLED);
1.50 thorpej 880: ip = mtod(m, struct ip *);
1.74 thorpej 881: hlen = ip->ip_hl << 2;
1.155 itojun 882: ip->ip_len = htons(ntohs(ip->ip_len) + hlen);
1.287 rmind 883: } else if (fp) {
884: ip_freef(fp);
1.288 ! rmind 885: ip_reass_unlock();
1.287 rmind 886: }
1.79 mycroft 887: }
1.128 itojun 888:
1.173 jonathan 889: #if defined(IPSEC)
1.128 itojun 890: /*
891: * enforce IPsec policy checking if we are seeing last header.
892: * note that we do not visit this with protocols with pcb layer
893: * code - like udp/tcp/raw ip.
894: */
895: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 &&
896: ipsec4_in_reject(m, NULL)) {
1.267 thorpej 897: IPSEC_STATINC(IPSEC_STAT_IN_POLVIO);
1.128 itojun 898: goto bad;
899: }
900: #endif
1.226 liamjfoy 901: #ifdef FAST_IPSEC
1.173 jonathan 902: /*
903: * enforce IPsec policy checking if we are seeing last header.
904: * note that we do not visit this with protocols with pcb layer
905: * code - like udp/tcp/raw ip.
906: */
907: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) {
908: /*
909: * Check if the packet has already had IPsec processing
910: * done. If so, then just pass it along. This tag gets
911: * set during AH, ESP, etc. input handling, before the
912: * packet is returned to the ip input queue for delivery.
1.212 perry 913: */
1.173 jonathan 914: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
915: s = splsoftnet();
916: if (mtag != NULL) {
917: tdbi = (struct tdb_ident *)(mtag + 1);
918: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
919: } else {
920: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
1.212 perry 921: IP_FORWARDING, &error);
1.173 jonathan 922: }
923: if (sp != NULL) {
924: /*
925: * Check security policy against packet attributes.
926: */
927: error = ipsec_in_reject(sp, m);
928: KEY_FREESP(&sp);
929: } else {
930: /* XXX error stat??? */
931: error = EINVAL;
932: DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/
933: }
934: splx(s);
935: if (error)
936: goto bad;
937: }
938: #endif /* FAST_IPSEC */
1.1 cgd 939:
940: /*
941: * Switch out to protocol's input routine.
942: */
1.82 aidan 943: #if IFA_STATS
1.122 itojun 944: if (ia && ip)
1.155 itojun 945: ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
1.82 aidan 946: #endif
1.266 thorpej 947: IP_STATINC(IP_STAT_DELIVERED);
1.89 itojun 948: {
949: int off = hlen, nh = ip->ip_p;
950:
951: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
952: return;
953: }
1.1 cgd 954: bad:
955: m_freem(m);
1.135 thorpej 956: return;
957:
958: badcsum:
1.266 thorpej 959: IP_STATINC(IP_STAT_BADSUM);
1.135 thorpej 960: m_freem(m);
1.1 cgd 961: }
962:
963: /*
1.288 ! rmind 964: * IP timer processing.
1.1 cgd 965: */
1.8 mycroft 966: void
1.211 perry 967: ip_slowtimo(void)
1.1 cgd 968: {
1.268 ad 969:
970: mutex_enter(softnet_lock);
971: KERNEL_LOCK(1, NULL);
1.1 cgd 972:
1.288 ! rmind 973: ip_reass_slowtimo();
1.268 ad 974:
975: KERNEL_UNLOCK_ONE(NULL);
976: mutex_exit(softnet_lock);
1.1 cgd 977: }
978:
979: /*
1.288 ! rmind 980: * IP drain processing.
1.1 cgd 981: */
1.8 mycroft 982: void
1.211 perry 983: ip_drain(void)
1.1 cgd 984: {
985:
1.268 ad 986: KERNEL_LOCK(1, NULL);
1.288 ! rmind 987: ip_reass_drain();
1.268 ad 988: KERNEL_UNLOCK_ONE(NULL);
1.1 cgd 989: }
990:
991: /*
992: * Do option processing on a datagram,
993: * possibly discarding it if bad options are encountered,
994: * or forwarding it if source-routed.
995: * Returns 1 if packet has been forwarded/freed,
996: * 0 if the packet should be processed further.
997: */
1.8 mycroft 998: int
1.211 perry 999: ip_dooptions(struct mbuf *m)
1.1 cgd 1000: {
1.109 augustss 1001: struct ip *ip = mtod(m, struct ip *);
1002: u_char *cp, *cp0;
1003: struct ip_timestamp *ipt;
1004: struct in_ifaddr *ia;
1.1 cgd 1005: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 1006: struct in_addr dst;
1.1 cgd 1007: n_time ntime;
1008:
1.13 mycroft 1009: dst = ip->ip_dst;
1.1 cgd 1010: cp = (u_char *)(ip + 1);
1011: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
1012: for (; cnt > 0; cnt -= optlen, cp += optlen) {
1013: opt = cp[IPOPT_OPTVAL];
1014: if (opt == IPOPT_EOL)
1015: break;
1016: if (opt == IPOPT_NOP)
1017: optlen = 1;
1018: else {
1.113 itojun 1019: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1020: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1021: goto bad;
1022: }
1.1 cgd 1023: optlen = cp[IPOPT_OLEN];
1.114 itojun 1024: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 1025: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1026: goto bad;
1027: }
1028: }
1029: switch (opt) {
1030:
1031: default:
1032: break;
1033:
1034: /*
1035: * Source routing with record.
1036: * Find interface with current destination address.
1037: * If none on this machine then drop if strictly routed,
1038: * or do nothing if loosely routed.
1039: * Record interface address and bring up next address
1040: * component. If strictly routed make sure next
1041: * address is on directly accessible net.
1042: */
1043: case IPOPT_LSRR:
1044: case IPOPT_SSRR:
1.47 cjs 1045: if (ip_allowsrcrt == 0) {
1046: type = ICMP_UNREACH;
1047: code = ICMP_UNREACH_NET_PROHIB;
1048: goto bad;
1049: }
1.114 itojun 1050: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1051: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1052: goto bad;
1053: }
1.1 cgd 1054: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1055: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1056: goto bad;
1057: }
1058: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 1059: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 1060: if (ia == 0) {
1061: if (opt == IPOPT_SSRR) {
1062: type = ICMP_UNREACH;
1063: code = ICMP_UNREACH_SRCFAIL;
1064: goto bad;
1065: }
1066: /*
1067: * Loose routing, and not at next destination
1068: * yet; nothing to do except forward.
1069: */
1070: break;
1071: }
1072: off--; /* 0 origin */
1.112 sommerfe 1073: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 1074: /*
1075: * End of source route. Should be for us.
1076: */
1077: save_rte(cp, ip->ip_src);
1078: break;
1079: }
1080: /*
1081: * locate outgoing interface
1082: */
1.281 tsutsui 1083: memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off),
1.1 cgd 1084: sizeof(ipaddr.sin_addr));
1.96 thorpej 1085: if (opt == IPOPT_SSRR)
1.196 itojun 1086: ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr)));
1.96 thorpej 1087: else
1.1 cgd 1088: ia = ip_rtaddr(ipaddr.sin_addr);
1089: if (ia == 0) {
1090: type = ICMP_UNREACH;
1091: code = ICMP_UNREACH_SRCFAIL;
1092: goto bad;
1093: }
1094: ip->ip_dst = ipaddr.sin_addr;
1.244 christos 1095: bcopy((void *)&ia->ia_addr.sin_addr,
1096: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1097: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1098: /*
1099: * Let ip_intr's mcast routing check handle mcast pkts
1100: */
1.18 mycroft 1101: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1102: break;
1103:
1104: case IPOPT_RR:
1.114 itojun 1105: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1106: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1107: goto bad;
1108: }
1.1 cgd 1109: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1110: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1111: goto bad;
1112: }
1113: /*
1114: * If no space remains, ignore.
1115: */
1116: off--; /* 0 origin */
1.112 sommerfe 1117: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1118: break;
1.281 tsutsui 1119: memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst),
1.1 cgd 1120: sizeof(ipaddr.sin_addr));
1121: /*
1122: * locate outgoing interface; if we're the destination,
1123: * use the incoming interface (should be same).
1124: */
1.96 thorpej 1125: if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))))
1126: == NULL &&
1127: (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
1.1 cgd 1128: type = ICMP_UNREACH;
1129: code = ICMP_UNREACH_HOST;
1130: goto bad;
1131: }
1.244 christos 1132: bcopy((void *)&ia->ia_addr.sin_addr,
1133: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1134: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1135: break;
1136:
1137: case IPOPT_TS:
1138: code = cp - (u_char *)ip;
1139: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1140: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1141: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1142: goto bad;
1.114 itojun 1143: }
1144: if (ipt->ipt_ptr < 5) {
1145: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1146: goto bad;
1147: }
1.15 cgd 1148: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1149: if (++ipt->ipt_oflw == 0) {
1150: code = (u_char *)&ipt->ipt_ptr -
1151: (u_char *)ip;
1.1 cgd 1152: goto bad;
1.114 itojun 1153: }
1.1 cgd 1154: break;
1155: }
1.104 thorpej 1156: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1157: switch (ipt->ipt_flg) {
1158:
1159: case IPOPT_TS_TSONLY:
1160: break;
1161:
1162: case IPOPT_TS_TSANDADDR:
1.66 thorpej 1163: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1164: sizeof(struct in_addr) > ipt->ipt_len) {
1165: code = (u_char *)&ipt->ipt_ptr -
1166: (u_char *)ip;
1.1 cgd 1167: goto bad;
1.114 itojun 1168: }
1.13 mycroft 1169: ipaddr.sin_addr = dst;
1.96 thorpej 1170: ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr),
1171: m->m_pkthdr.rcvif));
1.13 mycroft 1172: if (ia == 0)
1173: continue;
1.104 thorpej 1174: bcopy(&ia->ia_addr.sin_addr,
1175: cp0, sizeof(struct in_addr));
1.1 cgd 1176: ipt->ipt_ptr += sizeof(struct in_addr);
1177: break;
1178:
1179: case IPOPT_TS_PRESPEC:
1.66 thorpej 1180: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1181: sizeof(struct in_addr) > ipt->ipt_len) {
1182: code = (u_char *)&ipt->ipt_ptr -
1183: (u_char *)ip;
1.1 cgd 1184: goto bad;
1.114 itojun 1185: }
1.281 tsutsui 1186: memcpy(&ipaddr.sin_addr, cp0,
1.1 cgd 1187: sizeof(struct in_addr));
1.96 thorpej 1188: if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))
1189: == NULL)
1.1 cgd 1190: continue;
1191: ipt->ipt_ptr += sizeof(struct in_addr);
1192: break;
1193:
1194: default:
1.114 itojun 1195: /* XXX can't take &ipt->ipt_flg */
1196: code = (u_char *)&ipt->ipt_ptr -
1197: (u_char *)ip + 1;
1.1 cgd 1198: goto bad;
1199: }
1200: ntime = iptime();
1.107 thorpej 1201: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1.244 christos 1202: memmove((char *)cp + ipt->ipt_ptr - 1, cp0,
1.1 cgd 1203: sizeof(n_time));
1204: ipt->ipt_ptr += sizeof(n_time);
1205: }
1206: }
1207: if (forward) {
1.26 thorpej 1208: if (ip_forwsrcrt == 0) {
1209: type = ICMP_UNREACH;
1210: code = ICMP_UNREACH_SRCFAIL;
1211: goto bad;
1212: }
1.1 cgd 1213: ip_forward(m, 1);
1214: return (1);
1.13 mycroft 1215: }
1216: return (0);
1.1 cgd 1217: bad:
1.13 mycroft 1218: icmp_error(m, type, code, 0, 0);
1.266 thorpej 1219: IP_STATINC(IP_STAT_BADOPTIONS);
1.1 cgd 1220: return (1);
1221: }
1222:
1223: /*
1224: * Given address of next destination (final or next hop),
1225: * return internet address info of interface to be used to get there.
1226: */
1227: struct in_ifaddr *
1.211 perry 1228: ip_rtaddr(struct in_addr dst)
1.1 cgd 1229: {
1.249 dyoung 1230: struct rtentry *rt;
1231: union {
1232: struct sockaddr dst;
1233: struct sockaddr_in dst4;
1234: } u;
1235:
1236: sockaddr_in_init(&u.dst4, &dst, 0);
1237:
1238: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL)
1239: return NULL;
1240:
1241: return ifatoia(rt->rt_ifa);
1.1 cgd 1242: }
1243:
1244: /*
1245: * Save incoming source route for use in replies,
1246: * to be picked up later by ip_srcroute if the receiver is interested.
1247: */
1.13 mycroft 1248: void
1.211 perry 1249: save_rte(u_char *option, struct in_addr dst)
1.1 cgd 1250: {
1251: unsigned olen;
1252:
1253: olen = option[IPOPT_OLEN];
1254: #ifdef DIAGNOSTIC
1255: if (ipprintfs)
1.39 christos 1256: printf("save_rte: olen %d\n", olen);
1.89 itojun 1257: #endif /* 0 */
1.1 cgd 1258: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1259: return;
1.281 tsutsui 1260: memcpy((void *)ip_srcrt.srcopt, (void *)option, olen);
1.1 cgd 1261: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1262: ip_srcrt.dst = dst;
1263: }
1264:
1265: /*
1266: * Retrieve incoming source route for use in replies,
1267: * in the same form used by setsockopt.
1268: * The first hop is placed before the options, will be removed later.
1269: */
1270: struct mbuf *
1.211 perry 1271: ip_srcroute(void)
1.1 cgd 1272: {
1.109 augustss 1273: struct in_addr *p, *q;
1274: struct mbuf *m;
1.1 cgd 1275:
1276: if (ip_nhops == 0)
1.237 dyoung 1277: return NULL;
1.1 cgd 1278: m = m_get(M_DONTWAIT, MT_SOOPTS);
1279: if (m == 0)
1.237 dyoung 1280: return NULL;
1.1 cgd 1281:
1.164 matt 1282: MCLAIM(m, &inetdomain.dom_mowner);
1.13 mycroft 1283: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1284:
1285: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1286: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1287: OPTSIZ;
1288: #ifdef DIAGNOSTIC
1289: if (ipprintfs)
1.39 christos 1290: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1291: #endif
1292:
1293: /*
1294: * First save first hop for return route
1295: */
1296: p = &ip_srcrt.route[ip_nhops - 1];
1297: *(mtod(m, struct in_addr *)) = *p--;
1298: #ifdef DIAGNOSTIC
1299: if (ipprintfs)
1.39 christos 1300: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1301: #endif
1302:
1303: /*
1304: * Copy option fields and padding (nop) to mbuf.
1305: */
1306: ip_srcrt.nop = IPOPT_NOP;
1307: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1.244 christos 1308: memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop,
1309: OPTSIZ);
1310: q = (struct in_addr *)(mtod(m, char *) +
1.1 cgd 1311: sizeof(struct in_addr) + OPTSIZ);
1312: #undef OPTSIZ
1313: /*
1314: * Record return path as an IP source route,
1315: * reversing the path (pointers are now aligned).
1316: */
1317: while (p >= ip_srcrt.route) {
1318: #ifdef DIAGNOSTIC
1319: if (ipprintfs)
1.39 christos 1320: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1321: #endif
1322: *q++ = *p--;
1323: }
1324: /*
1325: * Last hop goes to final destination.
1326: */
1327: *q = ip_srcrt.dst;
1328: #ifdef DIAGNOSTIC
1329: if (ipprintfs)
1.39 christos 1330: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1331: #endif
1332: return (m);
1333: }
1334:
1.139 matt 1335: const int inetctlerrmap[PRC_NCMDS] = {
1.256 yamt 1336: [PRC_MSGSIZE] = EMSGSIZE,
1337: [PRC_HOSTDEAD] = EHOSTDOWN,
1338: [PRC_HOSTUNREACH] = EHOSTUNREACH,
1339: [PRC_UNREACH_NET] = EHOSTUNREACH,
1340: [PRC_UNREACH_HOST] = EHOSTUNREACH,
1341: [PRC_UNREACH_PROTOCOL] = ECONNREFUSED,
1342: [PRC_UNREACH_PORT] = ECONNREFUSED,
1343: [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH,
1344: [PRC_PARAMPROB] = ENOPROTOOPT,
1.1 cgd 1345: };
1346:
1347: /*
1348: * Forward a packet. If some error occurs return the sender
1349: * an icmp packet. Note we can't always generate a meaningful
1350: * icmp message because icmp doesn't have a large enough repertoire
1351: * of codes and types.
1352: *
1353: * If not forwarding, just drop the packet. This could be confusing
1354: * if ipforwarding was zero but some routing protocol was advancing
1355: * us as a gateway to somewhere. However, we must let the routing
1356: * protocol deal with that.
1357: *
1358: * The srcrt parameter indicates whether the packet is being forwarded
1359: * via a source route.
1360: */
1.13 mycroft 1361: void
1.211 perry 1362: ip_forward(struct mbuf *m, int srcrt)
1.1 cgd 1363: {
1.109 augustss 1364: struct ip *ip = mtod(m, struct ip *);
1365: struct rtentry *rt;
1.220 christos 1366: int error, type = 0, code = 0, destmtu = 0;
1.1 cgd 1367: struct mbuf *mcopy;
1.13 mycroft 1368: n_long dest;
1.249 dyoung 1369: union {
1370: struct sockaddr dst;
1371: struct sockaddr_in dst4;
1372: } u;
1.164 matt 1373:
1374: /*
1375: * We are now in the output path.
1376: */
1377: MCLAIM(m, &ip_tx_mowner);
1.135 thorpej 1378:
1379: /*
1380: * Clear any in-bound checksum flags for this packet.
1381: */
1382: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1383:
1.13 mycroft 1384: dest = 0;
1.1 cgd 1385: #ifdef DIAGNOSTIC
1.224 joerg 1386: if (ipprintfs) {
1387: printf("forward: src %s ", inet_ntoa(ip->ip_src));
1388: printf("dst %s ttl %x\n", inet_ntoa(ip->ip_dst), ip->ip_ttl);
1389: }
1.1 cgd 1390: #endif
1.93 sommerfe 1391: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.266 thorpej 1392: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 1393: m_freem(m);
1394: return;
1395: }
1396: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1397: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1398: return;
1399: }
1400:
1.249 dyoung 1401: sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
1402: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) {
1403: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0);
1404: return;
1.1 cgd 1405: }
1406:
1407: /*
1.34 mycroft 1408: * Save at most 68 bytes of the packet in case
1.1 cgd 1409: * we need to generate an ICMP message to the src.
1.119 itojun 1410: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1411: */
1.155 itojun 1412: mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1.119 itojun 1413: if (mcopy)
1414: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1415:
1.221 christos 1416: ip->ip_ttl -= IPTTLDEC;
1417:
1.1 cgd 1418: /*
1419: * If forwarding packet using same interface that it came in on,
1420: * perhaps should send a redirect to sender to shortcut a hop.
1421: * Only send redirect if source is sending directly to us,
1422: * and if packet was not source routed (or has any options).
1423: * Also, don't send redirect if forwarding using a default route
1424: * or a route modified by a redirect.
1425: */
1426: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1427: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.250 dyoung 1428: !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) &&
1.1 cgd 1429: ipsendredirects && !srcrt) {
1.19 mycroft 1430: if (rt->rt_ifa &&
1431: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1432: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1433: if (rt->rt_flags & RTF_GATEWAY)
1434: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1435: else
1436: dest = ip->ip_dst.s_addr;
1437: /*
1438: * Router requirements says to only send host
1439: * redirects.
1440: */
1441: type = ICMP_REDIRECT;
1442: code = ICMP_REDIRECT_HOST;
1.1 cgd 1443: #ifdef DIAGNOSTIC
1.77 thorpej 1444: if (ipprintfs)
1445: printf("redirect (%d) to %x\n", code,
1446: (u_int32_t)dest);
1.1 cgd 1447: #endif
1448: }
1449: }
1450:
1.238 dyoung 1451: error = ip_output(m, NULL, &ipforward_rt,
1.173 jonathan 1452: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
1.174 itojun 1453: (struct ip_moptions *)NULL, (struct socket *)NULL);
1.173 jonathan 1454:
1.1 cgd 1455: if (error)
1.266 thorpej 1456: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 1457: else {
1.266 thorpej 1458: uint64_t *ips = IP_STAT_GETREF();
1459: ips[IP_STAT_FORWARD]++;
1460: if (type) {
1461: ips[IP_STAT_REDIRECTSENT]++;
1462: IP_STAT_PUTREF();
1463: } else {
1464: IP_STAT_PUTREF();
1.63 matt 1465: if (mcopy) {
1466: #ifdef GATEWAY
1.64 thorpej 1467: if (mcopy->m_flags & M_CANFASTFWD)
1468: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1469: #endif
1.1 cgd 1470: m_freem(mcopy);
1.63 matt 1471: }
1.1 cgd 1472: return;
1473: }
1474: }
1475: if (mcopy == NULL)
1476: return;
1.13 mycroft 1477:
1.1 cgd 1478: switch (error) {
1479:
1480: case 0: /* forwarded, but need redirect */
1481: /* type, code set above */
1482: break;
1483:
1484: case ENETUNREACH: /* shouldn't happen, checked above */
1485: case EHOSTUNREACH:
1486: case ENETDOWN:
1487: case EHOSTDOWN:
1488: default:
1489: type = ICMP_UNREACH;
1490: code = ICMP_UNREACH_HOST;
1491: break;
1492:
1493: case EMSGSIZE:
1494: type = ICMP_UNREACH;
1495: code = ICMP_UNREACH_NEEDFRAG;
1.263 cube 1496:
1.274 seanb 1497: if ((rt = rtcache_validate(&ipforward_rt)) != NULL)
1498: destmtu = rt->rt_ifp->if_mtu;
1.263 cube 1499:
1500: #if defined(IPSEC) || defined(FAST_IPSEC)
1.274 seanb 1501: {
1.263 cube 1502: /*
1503: * If the packet is routed over IPsec tunnel, tell the
1504: * originator the tunnel MTU.
1505: * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1506: * XXX quickhack!!!
1507: */
1508:
1.89 itojun 1509: struct secpolicy *sp;
1510: int ipsecerror;
1.95 itojun 1511: size_t ipsechdr;
1.89 itojun 1512: struct route *ro;
1513:
1514: sp = ipsec4_getpolicybyaddr(mcopy,
1.170 itojun 1515: IPSEC_DIR_OUTBOUND, IP_FORWARDING,
1516: &ipsecerror);
1.89 itojun 1517:
1.263 cube 1518: if (sp != NULL) {
1.89 itojun 1519: /* count IPsec header size */
1.95 itojun 1520: ipsechdr = ipsec4_hdrsiz(mcopy,
1.170 itojun 1521: IPSEC_DIR_OUTBOUND, NULL);
1.89 itojun 1522:
1523: /*
1524: * find the correct route for outer IPv4
1525: * header, compute tunnel MTU.
1526: */
1.220 christos 1527:
1.89 itojun 1528: if (sp->req != NULL
1.95 itojun 1529: && sp->req->sav != NULL
1530: && sp->req->sav->sah != NULL) {
1531: ro = &sp->req->sav->sah->sa_route;
1.274 seanb 1532: rt = rtcache_validate(ro);
1.257 dyoung 1533: if (rt && rt->rt_ifp) {
1.220 christos 1534: destmtu =
1.257 dyoung 1535: rt->rt_rmx.rmx_mtu ?
1536: rt->rt_rmx.rmx_mtu :
1537: rt->rt_ifp->if_mtu;
1.220 christos 1538: destmtu -= ipsechdr;
1.89 itojun 1539: }
1540: }
1541:
1.173 jonathan 1542: #ifdef IPSEC
1.89 itojun 1543: key_freesp(sp);
1.173 jonathan 1544: #else
1545: KEY_FREESP(&sp);
1546: #endif
1.89 itojun 1547: }
1.274 seanb 1548: }
1.263 cube 1549: #endif /*defined(IPSEC) || defined(FAST_IPSEC)*/
1.266 thorpej 1550: IP_STATINC(IP_STAT_CANTFRAG);
1.1 cgd 1551: break;
1552:
1553: case ENOBUFS:
1.143 itojun 1554: #if 1
1555: /*
1556: * a router should not generate ICMP_SOURCEQUENCH as
1557: * required in RFC1812 Requirements for IP Version 4 Routers.
1558: * source quench could be a big problem under DoS attacks,
1.149 wiz 1559: * or if the underlying interface is rate-limited.
1.143 itojun 1560: */
1561: if (mcopy)
1562: m_freem(mcopy);
1563: return;
1564: #else
1.1 cgd 1565: type = ICMP_SOURCEQUENCH;
1566: code = 0;
1567: break;
1.143 itojun 1568: #endif
1.1 cgd 1569: }
1.220 christos 1570: icmp_error(mcopy, type, code, dest, destmtu);
1.44 thorpej 1571: }
1572:
1573: void
1.211 perry 1574: ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
1575: struct mbuf *m)
1.44 thorpej 1576: {
1577:
1.283 minskim 1578: if (inp->inp_socket->so_options & SO_TIMESTAMP
1.278 christos 1579: #ifdef SO_OTIMESTAMP
1.283 minskim 1580: || inp->inp_socket->so_options & SO_OTIMESTAMP
1.278 christos 1581: #endif
1582: ) {
1.44 thorpej 1583: struct timeval tv;
1584:
1585: microtime(&tv);
1.278 christos 1586: #ifdef SO_OTIMESTAMP
1587: if (inp->inp_socket->so_options & SO_OTIMESTAMP) {
1588: struct timeval50 tv50;
1589: timeval_to_timeval50(&tv, &tv50);
1590: *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50),
1591: SCM_OTIMESTAMP, SOL_SOCKET);
1592: } else
1593: #endif
1.244 christos 1594: *mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1.44 thorpej 1595: SCM_TIMESTAMP, SOL_SOCKET);
1596: if (*mp)
1597: mp = &(*mp)->m_next;
1598: }
1599: if (inp->inp_flags & INP_RECVDSTADDR) {
1.244 christos 1600: *mp = sbcreatecontrol((void *) &ip->ip_dst,
1.44 thorpej 1601: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1602: if (*mp)
1603: mp = &(*mp)->m_next;
1604: }
1605: #ifdef notyet
1606: /*
1607: * XXX
1608: * Moving these out of udp_input() made them even more broken
1609: * than they already were.
1610: * - fenner@parc.xerox.com
1611: */
1612: /* options were tossed already */
1613: if (inp->inp_flags & INP_RECVOPTS) {
1.244 christos 1614: *mp = sbcreatecontrol((void *) opts_deleted_above,
1.44 thorpej 1615: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1616: if (*mp)
1617: mp = &(*mp)->m_next;
1618: }
1619: /* ip_srcroute doesn't do what we want here, need to fix */
1620: if (inp->inp_flags & INP_RECVRETOPTS) {
1.244 christos 1621: *mp = sbcreatecontrol((void *) ip_srcroute(),
1.44 thorpej 1622: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1623: if (*mp)
1624: mp = &(*mp)->m_next;
1625: }
1626: #endif
1627: if (inp->inp_flags & INP_RECVIF) {
1628: struct sockaddr_dl sdl;
1629:
1.252 dyoung 1630: sockaddr_dl_init(&sdl, sizeof(sdl),
1631: (m->m_pkthdr.rcvif != NULL)
1632: ? m->m_pkthdr.rcvif->if_index
1633: : 0,
1634: 0, NULL, 0, NULL, 0);
1.251 dyoung 1635: *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP);
1.44 thorpej 1636: if (*mp)
1637: mp = &(*mp)->m_next;
1638: }
1.282 minskim 1639: if (inp->inp_flags & INP_RECVTTL) {
1640: *mp = sbcreatecontrol((void *) &ip->ip_ttl,
1641: sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP);
1642: if (*mp)
1643: mp = &(*mp)->m_next;
1644: }
1.13 mycroft 1645: }
1646:
1.189 atatat 1647: /*
1.228 elad 1648: * sysctl helper routine for net.inet.ip.forwsrcrt.
1649: */
1650: static int
1651: sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS)
1652: {
1653: int error, tmp;
1654: struct sysctlnode node;
1655:
1656: node = *rnode;
1657: tmp = ip_forwsrcrt;
1658: node.sysctl_data = &tmp;
1659: error = sysctl_lookup(SYSCTLFN_CALL(&node));
1660: if (error || newp == NULL)
1661: return (error);
1662:
1.280 elad 1663: error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT,
1664: 0, NULL, NULL, NULL);
1665: if (error)
1666: return (error);
1.228 elad 1667:
1668: ip_forwsrcrt = tmp;
1669:
1670: return (0);
1671: }
1672:
1673: /*
1.189 atatat 1674: * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the
1675: * range of the new value and tweaks timers if it changes.
1676: */
1677: static int
1678: sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS)
1.13 mycroft 1679: {
1.189 atatat 1680: int error, tmp;
1681: struct sysctlnode node;
1682:
1683: node = *rnode;
1684: tmp = ip_mtudisc_timeout;
1685: node.sysctl_data = &tmp;
1686: error = sysctl_lookup(SYSCTLFN_CALL(&node));
1687: if (error || newp == NULL)
1688: return (error);
1689: if (tmp < 0)
1690: return (EINVAL);
1.52 thorpej 1691:
1.273 matt 1692: mutex_enter(softnet_lock);
1693:
1.189 atatat 1694: ip_mtudisc_timeout = tmp;
1695: rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout);
1696:
1.273 matt 1697: mutex_exit(softnet_lock);
1698:
1.189 atatat 1699: return (0);
1700: }
1.54 lukem 1701:
1.65 matt 1702: #ifdef GATEWAY
1.189 atatat 1703: /*
1.247 liamjfoy 1704: * sysctl helper routine for net.inet.ip.maxflows.
1.189 atatat 1705: */
1706: static int
1707: sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS)
1708: {
1.273 matt 1709: int error;
1.67 thorpej 1710:
1.273 matt 1711: error = sysctl_lookup(SYSCTLFN_CALL(rnode));
1712: if (error || newp == NULL)
1713: return (error);
1714:
1715: mutex_enter(softnet_lock);
1716: KERNEL_LOCK(1, NULL);
1.212 perry 1717:
1.265 thorpej 1718: ipflow_prune();
1.273 matt 1719:
1720: KERNEL_UNLOCK_ONE(NULL);
1721: mutex_exit(softnet_lock);
1.144 martin 1722:
1.189 atatat 1723: return (0);
1724: }
1.248 liamjfoy 1725:
1726: static int
1727: sysctl_net_inet_ip_hashsize(SYSCTLFN_ARGS)
1.283 minskim 1728: {
1.248 liamjfoy 1729: int error, tmp;
1730: struct sysctlnode node;
1731:
1732: node = *rnode;
1733: tmp = ip_hashsize;
1734: node.sysctl_data = &tmp;
1735: error = sysctl_lookup(SYSCTLFN_CALL(&node));
1736: if (error || newp == NULL)
1737: return (error);
1738:
1739: if ((tmp & (tmp - 1)) == 0 && tmp != 0) {
1740: /*
1741: * Can only fail due to malloc()
1742: */
1.273 matt 1743: mutex_enter(softnet_lock);
1744: KERNEL_LOCK(1, NULL);
1745:
1746: error = ipflow_invalidate_all(tmp);
1747:
1748: KERNEL_UNLOCK_ONE(NULL);
1749: mutex_exit(softnet_lock);
1750:
1.248 liamjfoy 1751: } else {
1752: /*
1753: * EINVAL if not a power of 2
1754: */
1.273 matt 1755: error = EINVAL;
1.283 minskim 1756: }
1.248 liamjfoy 1757:
1.273 matt 1758: return error;
1.248 liamjfoy 1759: }
1.189 atatat 1760: #endif /* GATEWAY */
1.117 tron 1761:
1.266 thorpej 1762: static int
1763: sysctl_net_inet_ip_stats(SYSCTLFN_ARGS)
1764: {
1765:
1.271 thorpej 1766: return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS));
1.266 thorpej 1767: }
1.131 itojun 1768:
1.284 pooka 1769: static void
1770: sysctl_net_inet_ip_setup(struct sysctllog **clog)
1.189 atatat 1771: {
1772: extern int subnetsarelocal, hostzeroisbroadcast;
1.180 jonathan 1773:
1.197 atatat 1774: sysctl_createv(clog, 0, NULL, NULL,
1775: CTLFLAG_PERMANENT,
1.189 atatat 1776: CTLTYPE_NODE, "net", NULL,
1777: NULL, 0, NULL, 0,
1778: CTL_NET, CTL_EOL);
1.197 atatat 1779: sysctl_createv(clog, 0, NULL, NULL,
1780: CTLFLAG_PERMANENT,
1.203 atatat 1781: CTLTYPE_NODE, "inet",
1782: SYSCTL_DESCR("PF_INET related settings"),
1.189 atatat 1783: NULL, 0, NULL, 0,
1784: CTL_NET, PF_INET, CTL_EOL);
1.197 atatat 1785: sysctl_createv(clog, 0, NULL, NULL,
1786: CTLFLAG_PERMANENT,
1.203 atatat 1787: CTLTYPE_NODE, "ip",
1788: SYSCTL_DESCR("IPv4 related settings"),
1.189 atatat 1789: NULL, 0, NULL, 0,
1790: CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL);
1.212 perry 1791:
1.197 atatat 1792: sysctl_createv(clog, 0, NULL, NULL,
1793: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1794: CTLTYPE_INT, "forwarding",
1795: SYSCTL_DESCR("Enable forwarding of INET datagrams"),
1.189 atatat 1796: NULL, 0, &ipforwarding, 0,
1797: CTL_NET, PF_INET, IPPROTO_IP,
1798: IPCTL_FORWARDING, CTL_EOL);
1.197 atatat 1799: sysctl_createv(clog, 0, NULL, NULL,
1800: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1801: CTLTYPE_INT, "redirect",
1802: SYSCTL_DESCR("Enable sending of ICMP redirect messages"),
1.189 atatat 1803: NULL, 0, &ipsendredirects, 0,
1804: CTL_NET, PF_INET, IPPROTO_IP,
1805: IPCTL_SENDREDIRECTS, CTL_EOL);
1.197 atatat 1806: sysctl_createv(clog, 0, NULL, NULL,
1807: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1808: CTLTYPE_INT, "ttl",
1809: SYSCTL_DESCR("Default TTL for an INET datagram"),
1.189 atatat 1810: NULL, 0, &ip_defttl, 0,
1811: CTL_NET, PF_INET, IPPROTO_IP,
1812: IPCTL_DEFTTL, CTL_EOL);
1813: #ifdef IPCTL_DEFMTU
1.197 atatat 1814: sysctl_createv(clog, 0, NULL, NULL,
1815: CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */,
1.203 atatat 1816: CTLTYPE_INT, "mtu",
1817: SYSCTL_DESCR("Default MTA for an INET route"),
1.189 atatat 1818: NULL, 0, &ip_mtu, 0,
1819: CTL_NET, PF_INET, IPPROTO_IP,
1820: IPCTL_DEFMTU, CTL_EOL);
1821: #endif /* IPCTL_DEFMTU */
1.197 atatat 1822: sysctl_createv(clog, 0, NULL, NULL,
1.228 elad 1823: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1824: CTLTYPE_INT, "forwsrcrt",
1825: SYSCTL_DESCR("Enable forwarding of source-routed "
1826: "datagrams"),
1.228 elad 1827: sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0,
1.189 atatat 1828: CTL_NET, PF_INET, IPPROTO_IP,
1829: IPCTL_FORWSRCRT, CTL_EOL);
1.197 atatat 1830: sysctl_createv(clog, 0, NULL, NULL,
1831: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1832: CTLTYPE_INT, "directed-broadcast",
1833: SYSCTL_DESCR("Enable forwarding of broadcast datagrams"),
1.189 atatat 1834: NULL, 0, &ip_directedbcast, 0,
1835: CTL_NET, PF_INET, IPPROTO_IP,
1836: IPCTL_DIRECTEDBCAST, CTL_EOL);
1.197 atatat 1837: sysctl_createv(clog, 0, NULL, NULL,
1838: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1839: CTLTYPE_INT, "allowsrcrt",
1840: SYSCTL_DESCR("Accept source-routed datagrams"),
1.189 atatat 1841: NULL, 0, &ip_allowsrcrt, 0,
1842: CTL_NET, PF_INET, IPPROTO_IP,
1843: IPCTL_ALLOWSRCRT, CTL_EOL);
1.197 atatat 1844: sysctl_createv(clog, 0, NULL, NULL,
1845: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1846: CTLTYPE_INT, "subnetsarelocal",
1847: SYSCTL_DESCR("Whether logical subnets are considered "
1848: "local"),
1.189 atatat 1849: NULL, 0, &subnetsarelocal, 0,
1850: CTL_NET, PF_INET, IPPROTO_IP,
1851: IPCTL_SUBNETSARELOCAL, CTL_EOL);
1.197 atatat 1852: sysctl_createv(clog, 0, NULL, NULL,
1853: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1854: CTLTYPE_INT, "mtudisc",
1855: SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"),
1.189 atatat 1856: NULL, 0, &ip_mtudisc, 0,
1857: CTL_NET, PF_INET, IPPROTO_IP,
1858: IPCTL_MTUDISC, CTL_EOL);
1.197 atatat 1859: sysctl_createv(clog, 0, NULL, NULL,
1860: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1861: CTLTYPE_INT, "anonportmin",
1862: SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1.189 atatat 1863: sysctl_net_inet_ip_ports, 0, &anonportmin, 0,
1864: CTL_NET, PF_INET, IPPROTO_IP,
1865: IPCTL_ANONPORTMIN, CTL_EOL);
1.197 atatat 1866: sysctl_createv(clog, 0, NULL, NULL,
1867: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1868: CTLTYPE_INT, "anonportmax",
1869: SYSCTL_DESCR("Highest ephemeral port number to assign"),
1.189 atatat 1870: sysctl_net_inet_ip_ports, 0, &anonportmax, 0,
1871: CTL_NET, PF_INET, IPPROTO_IP,
1872: IPCTL_ANONPORTMAX, CTL_EOL);
1.197 atatat 1873: sysctl_createv(clog, 0, NULL, NULL,
1874: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1875: CTLTYPE_INT, "mtudisctimeout",
1876: SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"),
1.189 atatat 1877: sysctl_net_inet_ip_pmtudto, 0, &ip_mtudisc_timeout, 0,
1878: CTL_NET, PF_INET, IPPROTO_IP,
1879: IPCTL_MTUDISCTIMEOUT, CTL_EOL);
1880: #ifdef GATEWAY
1.197 atatat 1881: sysctl_createv(clog, 0, NULL, NULL,
1882: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1883: CTLTYPE_INT, "maxflows",
1884: SYSCTL_DESCR("Number of flows for fast forwarding"),
1.189 atatat 1885: sysctl_net_inet_ip_maxflows, 0, &ip_maxflows, 0,
1886: CTL_NET, PF_INET, IPPROTO_IP,
1887: IPCTL_MAXFLOWS, CTL_EOL);
1.248 liamjfoy 1888: sysctl_createv(clog, 0, NULL, NULL,
1889: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1890: CTLTYPE_INT, "hashsize",
1891: SYSCTL_DESCR("Size of hash table for fast forwarding (IPv4)"),
1892: sysctl_net_inet_ip_hashsize, 0, &ip_hashsize, 0,
1893: CTL_NET, PF_INET, IPPROTO_IP,
1894: CTL_CREATE, CTL_EOL);
1.189 atatat 1895: #endif /* GATEWAY */
1.197 atatat 1896: sysctl_createv(clog, 0, NULL, NULL,
1897: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1898: CTLTYPE_INT, "hostzerobroadcast",
1899: SYSCTL_DESCR("All zeroes address is broadcast address"),
1.189 atatat 1900: NULL, 0, &hostzeroisbroadcast, 0,
1901: CTL_NET, PF_INET, IPPROTO_IP,
1902: IPCTL_HOSTZEROBROADCAST, CTL_EOL);
1903: #if NGIF > 0
1.197 atatat 1904: sysctl_createv(clog, 0, NULL, NULL,
1905: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1906: CTLTYPE_INT, "gifttl",
1907: SYSCTL_DESCR("Default TTL for a gif tunnel datagram"),
1.189 atatat 1908: NULL, 0, &ip_gif_ttl, 0,
1909: CTL_NET, PF_INET, IPPROTO_IP,
1910: IPCTL_GIF_TTL, CTL_EOL);
1911: #endif /* NGIF */
1912: #ifndef IPNOPRIVPORTS
1.197 atatat 1913: sysctl_createv(clog, 0, NULL, NULL,
1914: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1915: CTLTYPE_INT, "lowportmin",
1916: SYSCTL_DESCR("Lowest privileged ephemeral port number "
1917: "to assign"),
1.189 atatat 1918: sysctl_net_inet_ip_ports, 0, &lowportmin, 0,
1919: CTL_NET, PF_INET, IPPROTO_IP,
1920: IPCTL_LOWPORTMIN, CTL_EOL);
1.197 atatat 1921: sysctl_createv(clog, 0, NULL, NULL,
1922: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1923: CTLTYPE_INT, "lowportmax",
1924: SYSCTL_DESCR("Highest privileged ephemeral port number "
1925: "to assign"),
1.189 atatat 1926: sysctl_net_inet_ip_ports, 0, &lowportmax, 0,
1927: CTL_NET, PF_INET, IPPROTO_IP,
1928: IPCTL_LOWPORTMAX, CTL_EOL);
1929: #endif /* IPNOPRIVPORTS */
1930: #if NGRE > 0
1.197 atatat 1931: sysctl_createv(clog, 0, NULL, NULL,
1932: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1933: CTLTYPE_INT, "grettl",
1934: SYSCTL_DESCR("Default TTL for a gre tunnel datagram"),
1.189 atatat 1935: NULL, 0, &ip_gre_ttl, 0,
1936: CTL_NET, PF_INET, IPPROTO_IP,
1937: IPCTL_GRE_TTL, CTL_EOL);
1938: #endif /* NGRE */
1.197 atatat 1939: sysctl_createv(clog, 0, NULL, NULL,
1940: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1941: CTLTYPE_INT, "checkinterface",
1942: SYSCTL_DESCR("Enable receive side of Strong ES model "
1943: "from RFC1122"),
1.189 atatat 1944: NULL, 0, &ip_checkinterface, 0,
1945: CTL_NET, PF_INET, IPPROTO_IP,
1946: IPCTL_CHECKINTERFACE, CTL_EOL);
1.197 atatat 1947: sysctl_createv(clog, 0, NULL, NULL,
1948: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1949: CTLTYPE_INT, "random_id",
1950: SYSCTL_DESCR("Assign random ip_id values"),
1.189 atatat 1951: NULL, 0, &ip_do_randomid, 0,
1952: CTL_NET, PF_INET, IPPROTO_IP,
1953: IPCTL_RANDOMID, CTL_EOL);
1.206 thorpej 1954: sysctl_createv(clog, 0, NULL, NULL,
1955: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1956: CTLTYPE_INT, "do_loopback_cksum",
1957: SYSCTL_DESCR("Perform IP checksum on loopback"),
1958: NULL, 0, &ip_do_loopback_cksum, 0,
1959: CTL_NET, PF_INET, IPPROTO_IP,
1960: IPCTL_LOOPBACKCKSUM, CTL_EOL);
1.219 elad 1961: sysctl_createv(clog, 0, NULL, NULL,
1962: CTLFLAG_PERMANENT,
1963: CTLTYPE_STRUCT, "stats",
1964: SYSCTL_DESCR("IP statistics"),
1.266 thorpej 1965: sysctl_net_inet_ip_stats, 0, NULL, 0,
1.219 elad 1966: CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS,
1967: CTL_EOL);
1.1 cgd 1968: }
1.266 thorpej 1969:
1970: void
1971: ip_statinc(u_int stat)
1972: {
1973:
1974: KASSERT(stat < IP_NSTATS);
1975: IP_STATINC(stat);
1976: }
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