Annotation of src/sys/netinet/ip_input.c, Revision 1.282
1.282 ! minskim 1: /* $NetBSD: ip_input.c,v 1.281 2009/04/18 14:58:05 tsutsui 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>
1.282 ! minskim 94: __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.281 2009/04/18 14:58:05 tsutsui 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/malloc.h>
108: #include <sys/mbuf.h>
109: #include <sys/domain.h>
110: #include <sys/protosw.h>
111: #include <sys/socket.h>
1.44 thorpej 112: #include <sys/socketvar.h>
1.5 mycroft 113: #include <sys/errno.h>
114: #include <sys/time.h>
115: #include <sys/kernel.h>
1.72 thorpej 116: #include <sys/pool.h>
1.28 christos 117: #include <sys/sysctl.h>
1.230 elad 118: #include <sys/kauth.h>
1.1 cgd 119:
1.5 mycroft 120: #include <net/if.h>
1.44 thorpej 121: #include <net/if_dl.h>
1.5 mycroft 122: #include <net/route.h>
1.45 mrg 123: #include <net/pfil.h>
1.1 cgd 124:
1.5 mycroft 125: #include <netinet/in.h>
126: #include <netinet/in_systm.h>
127: #include <netinet/ip.h>
128: #include <netinet/in_pcb.h>
1.215 yamt 129: #include <netinet/in_proto.h>
1.5 mycroft 130: #include <netinet/in_var.h>
131: #include <netinet/ip_var.h>
1.266 thorpej 132: #include <netinet/ip_private.h>
1.5 mycroft 133: #include <netinet/ip_icmp.h>
1.89 itojun 134: /* just for gif_ttl */
135: #include <netinet/in_gif.h>
136: #include "gif.h"
1.144 martin 137: #include <net/if_gre.h>
138: #include "gre.h"
1.111 jdolecek 139:
140: #ifdef MROUTING
141: #include <netinet/ip_mroute.h>
142: #endif
1.89 itojun 143:
144: #ifdef IPSEC
145: #include <netinet6/ipsec.h>
1.267 thorpej 146: #include <netinet6/ipsec_private.h>
1.89 itojun 147: #include <netkey/key.h>
148: #endif
1.173 jonathan 149: #ifdef FAST_IPSEC
150: #include <netipsec/ipsec.h>
151: #include <netipsec/key.h>
152: #endif /* FAST_IPSEC*/
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:
1.278 christos 177: #ifdef COMPAT_50
178: #include <compat/sys/time.h>
179: #include <compat/sys/socket.h>
180: #endif
181:
1.27 thorpej 182: /*
183: * Note: DIRECTED_BROADCAST is handled this way so that previous
184: * configuration using this option will Just Work.
185: */
186: #ifndef IPDIRECTEDBCAST
187: #ifdef DIRECTED_BROADCAST
188: #define IPDIRECTEDBCAST 1
189: #else
190: #define IPDIRECTEDBCAST 0
191: #endif /* DIRECTED_BROADCAST */
192: #endif /* IPDIRECTEDBCAST */
1.1 cgd 193: int ipforwarding = IPFORWARDING;
194: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 195: int ip_defttl = IPDEFTTL;
1.26 thorpej 196: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 197: int ip_directedbcast = IPDIRECTEDBCAST;
1.47 cjs 198: int ip_allowsrcrt = IPALLOWSRCRT;
1.53 kml 199: int ip_mtudisc = IPMTUDISC;
1.156 itojun 200: int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
1.1 cgd 201: #ifdef DIAGNOSTIC
202: int ipprintfs = 0;
203: #endif
1.184 jonathan 204:
205: int ip_do_randomid = 0;
206:
1.165 christos 207: /*
208: * XXX - Setting ip_checkinterface mostly implements the receive side of
209: * the Strong ES model described in RFC 1122, but since the routing table
210: * and transmit implementation do not implement the Strong ES model,
211: * setting this to 1 results in an odd hybrid.
212: *
213: * XXX - ip_checkinterface currently must be disabled if you use ipnat
214: * to translate the destination address to another local interface.
215: *
216: * XXX - ip_checkinterface must be disabled if you add IP aliases
217: * to the loopback interface instead of the interface where the
218: * packets for those addresses are received.
219: */
220: int ip_checkinterface = 0;
221:
1.1 cgd 222:
1.60 kml 223: struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
224:
1.1 cgd 225: int ipqmaxlen = IFQ_MAXLEN;
1.150 matt 226: u_long in_ifaddrhash; /* size of hash table - 1 */
227: int in_ifaddrentries; /* total number of addrs */
1.212 perry 228: struct in_ifaddrhead in_ifaddrhead;
1.57 tls 229: struct in_ifaddrhashhead *in_ifaddrhashtbl;
1.166 matt 230: u_long in_multihash; /* size of hash table - 1 */
231: int in_multientries; /* total number of addrs */
232: struct in_multihashhead *in_multihashtbl;
1.13 mycroft 233: struct ifqueue ipintrq;
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:
1.194 jonathan 242: /*
243: * Cached copy of nmbclusters. If nbclusters is different,
244: * recalculate IP parameters derived from nmbclusters.
245: */
246: static int ip_nmbclusters; /* copy of nmbclusters */
1.210 perry 247: static void ip_nmbclusters_changed(void); /* recalc limits */
1.194 jonathan 248:
1.195 thorpej 249: #define CHECK_NMBCLUSTER_PARAMS() \
250: do { \
251: if (__predict_false(ip_nmbclusters != nmbclusters)) \
252: ip_nmbclusters_changed(); \
253: } while (/*CONSTCOND*/0)
1.194 jonathan 254:
1.190 jonathan 255: /* IP datagram reassembly queues (hashed) */
256: #define IPREASS_NHASH_LOG2 6
257: #define IPREASS_NHASH (1 << IPREASS_NHASH_LOG2)
258: #define IPREASS_HMASK (IPREASS_NHASH - 1)
259: #define IPREASS_HASH(x,y) \
260: (((((x) & 0xF) | ((((x) >> 8) & 0xF) << 4)) ^ (y)) & IPREASS_HMASK)
261: struct ipqhead ipq[IPREASS_NHASH];
1.75 thorpej 262: int ipq_locked;
1.212 perry 263: static int ip_nfragpackets; /* packets in reass queue */
1.194 jonathan 264: static int ip_nfrags; /* total fragments in reass queues */
265:
266: int ip_maxfragpackets = 200; /* limit on packets. XXX sysctl */
267: int ip_maxfrags; /* limit on fragments. XXX sysctl */
268:
269:
270: /*
271: * Additive-Increase/Multiplicative-Decrease (AIMD) strategy for
272: * IP reassembly queue buffer managment.
1.212 perry 273: *
1.194 jonathan 274: * We keep a count of total IP fragments (NB: not fragmented packets!)
275: * awaiting reassembly (ip_nfrags) and a limit (ip_maxfrags) on fragments.
276: * If ip_nfrags exceeds ip_maxfrags the limit, we drop half the
277: * total fragments in reassembly queues.This AIMD policy avoids
278: * repeatedly deleting single packets under heavy fragmentation load
279: * (e.g., from lossy NFS peers).
280: */
1.212 perry 281: static u_int ip_reass_ttl_decr(u_int ticks);
1.210 perry 282: static void ip_reass_drophalf(void);
1.194 jonathan 283:
1.75 thorpej 284:
1.223 perry 285: static inline int ipq_lock_try(void);
286: static inline void ipq_unlock(void);
1.75 thorpej 287:
1.223 perry 288: static inline int
1.211 perry 289: ipq_lock_try(void)
1.75 thorpej 290: {
291: int s;
292:
1.132 thorpej 293: /*
1.149 wiz 294: * Use splvm() -- we're blocking things that would cause
1.132 thorpej 295: * mbuf allocation.
296: */
297: s = splvm();
1.75 thorpej 298: if (ipq_locked) {
299: splx(s);
300: return (0);
301: }
302: ipq_locked = 1;
303: splx(s);
304: return (1);
305: }
306:
1.223 perry 307: static inline void
1.211 perry 308: ipq_unlock(void)
1.75 thorpej 309: {
310: int s;
311:
1.132 thorpej 312: s = splvm();
1.75 thorpej 313: ipq_locked = 0;
314: splx(s);
315: }
316:
317: #ifdef DIAGNOSTIC
318: #define IPQ_LOCK() \
319: do { \
320: if (ipq_lock_try() == 0) { \
321: printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \
322: panic("ipq_lock"); \
323: } \
1.159 perry 324: } while (/*CONSTCOND*/ 0)
1.75 thorpej 325: #define IPQ_LOCK_CHECK() \
326: do { \
327: if (ipq_locked == 0) { \
328: printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \
329: panic("ipq lock check"); \
330: } \
1.159 perry 331: } while (/*CONSTCOND*/ 0)
1.75 thorpej 332: #else
333: #define IPQ_LOCK() (void) ipq_lock_try()
334: #define IPQ_LOCK_CHECK() /* nothing */
335: #endif
336:
337: #define IPQ_UNLOCK() ipq_unlock()
1.1 cgd 338:
1.275 pooka 339: struct pool inmulti_pool;
340: struct pool ipqent_pool;
1.72 thorpej 341:
1.135 thorpej 342: #ifdef INET_CSUM_COUNTERS
343: #include <sys/device.h>
344:
345: struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
346: NULL, "inet", "hwcsum bad");
347: struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
348: NULL, "inet", "hwcsum ok");
349: struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
350: NULL, "inet", "swcsum");
351:
352: #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++
353:
1.201 matt 354: EVCNT_ATTACH_STATIC(ip_hwcsum_bad);
355: EVCNT_ATTACH_STATIC(ip_hwcsum_ok);
356: EVCNT_ATTACH_STATIC(ip_swcsum);
357:
1.135 thorpej 358: #else
359:
360: #define INET_CSUM_COUNTER_INCR(ev) /* nothing */
361:
362: #endif /* INET_CSUM_COUNTERS */
363:
1.1 cgd 364: /*
365: * We need to save the IP options in case a protocol wants to respond
366: * to an incoming packet over the same route if the packet got here
367: * using IP source routing. This allows connection establishment and
368: * maintenance when the remote end is on a network that is not known
369: * to us.
370: */
371: int ip_nhops = 0;
372: static struct ip_srcrt {
373: struct in_addr dst; /* final destination */
374: char nop; /* one NOP to align */
375: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
376: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
377: } ip_srcrt;
378:
1.210 perry 379: static void save_rte(u_char *, struct in_addr);
1.35 mycroft 380:
1.164 matt 381: #ifdef MBUFTRACE
1.234 dogcow 382: struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx");
383: struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx");
1.164 matt 384: #endif
385:
1.1 cgd 386: /*
1.194 jonathan 387: * Compute IP limits derived from the value of nmbclusters.
388: */
389: static void
390: ip_nmbclusters_changed(void)
391: {
392: ip_maxfrags = nmbclusters / 4;
393: ip_nmbclusters = nmbclusters;
394: }
395:
396: /*
1.1 cgd 397: * IP initialization: fill in IP protocol switch table.
398: * All protocols not implemented in kernel go to raw IP protocol handler.
399: */
1.8 mycroft 400: void
1.211 perry 401: ip_init(void)
1.1 cgd 402: {
1.199 matt 403: const struct protosw *pr;
1.109 augustss 404: int i;
1.1 cgd 405:
1.275 pooka 406: pool_init(&inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl",
407: NULL, IPL_SOFTNET);
408: pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl",
409: NULL, IPL_VM);
410:
1.1 cgd 411: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
412: if (pr == 0)
413: panic("ip_init");
414: for (i = 0; i < IPPROTO_MAX; i++)
415: ip_protox[i] = pr - inetsw;
416: for (pr = inetdomain.dom_protosw;
417: pr < inetdomain.dom_protoswNPROTOSW; pr++)
418: if (pr->pr_domain->dom_family == PF_INET &&
419: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
420: ip_protox[pr->pr_protocol] = pr - inetsw;
1.192 jonathan 421:
1.190 jonathan 422: for (i = 0; i < IPREASS_NHASH; i++)
423: LIST_INIT(&ipq[i]);
424:
1.262 matt 425: ip_initid();
1.227 kardel 426: ip_id = time_second & 0xfffff;
1.194 jonathan 427:
1.1 cgd 428: ipintrq.ifq_maxlen = ipqmaxlen;
1.194 jonathan 429: ip_nmbclusters_changed();
430:
1.181 jonathan 431: TAILQ_INIT(&in_ifaddrhead);
1.272 ad 432: in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true,
433: &in_ifaddrhash);
434: in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true,
435: &in_multihash);
1.160 itojun 436: ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout);
1.73 thorpej 437: #ifdef GATEWAY
1.248 liamjfoy 438: ipflow_init(ip_hashsize);
1.73 thorpej 439: #endif
1.121 thorpej 440:
441: #ifdef PFIL_HOOKS
442: /* Register our Packet Filter hook. */
1.126 thorpej 443: inet_pfil_hook.ph_type = PFIL_TYPE_AF;
444: inet_pfil_hook.ph_af = AF_INET;
1.121 thorpej 445: i = pfil_head_register(&inet_pfil_hook);
446: if (i != 0)
447: printf("ip_init: WARNING: unable to register pfil hook, "
448: "error %d\n", i);
449: #endif /* PFIL_HOOKS */
1.135 thorpej 450:
1.164 matt 451: #ifdef MBUFTRACE
452: MOWNER_ATTACH(&ip_tx_mowner);
453: MOWNER_ATTACH(&ip_rx_mowner);
454: #endif /* MBUFTRACE */
1.266 thorpej 455:
456: ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS);
1.1 cgd 457: }
458:
1.229 christos 459: struct sockaddr_in ipaddr = {
460: .sin_len = sizeof(ipaddr),
461: .sin_family = AF_INET,
462: };
1.1 cgd 463: struct route ipforward_rt;
464:
465: /*
1.89 itojun 466: * IP software interrupt routine
467: */
468: void
1.211 perry 469: ipintr(void)
1.89 itojun 470: {
471: int s;
472: struct mbuf *m;
473:
1.268 ad 474: mutex_enter(softnet_lock);
475: KERNEL_LOCK(1, NULL);
1.241 ad 476: while (!IF_IS_EMPTY(&ipintrq)) {
1.132 thorpej 477: s = splnet();
1.89 itojun 478: IF_DEQUEUE(&ipintrq, m);
479: splx(s);
1.268 ad 480: if (m == NULL)
481: break;
1.89 itojun 482: ip_input(m);
483: }
1.268 ad 484: KERNEL_UNLOCK_ONE(NULL);
485: mutex_exit(softnet_lock);
1.89 itojun 486: }
487:
488: /*
1.1 cgd 489: * Ip input routine. Checksum and byte swap header. If fragmented
490: * try to reassemble. Process options. Pass to next level.
491: */
1.8 mycroft 492: void
1.89 itojun 493: ip_input(struct mbuf *m)
1.1 cgd 494: {
1.109 augustss 495: struct ip *ip = NULL;
496: struct ipq *fp;
497: struct in_ifaddr *ia;
498: struct ifaddr *ifa;
1.25 cgd 499: struct ipqent *ipqe;
1.89 itojun 500: int hlen = 0, mff, len;
1.100 itojun 501: int downmatch;
1.165 christos 502: int checkif;
1.169 itojun 503: int srcrt = 0;
1.233 tls 504: int s;
1.190 jonathan 505: u_int hash;
1.173 jonathan 506: #ifdef FAST_IPSEC
507: struct m_tag *mtag;
508: struct tdb_ident *tdbi;
509: struct secpolicy *sp;
1.233 tls 510: int error;
1.173 jonathan 511: #endif /* FAST_IPSEC */
1.1 cgd 512:
1.164 matt 513: MCLAIM(m, &ip_rx_mowner);
1.1 cgd 514: #ifdef DIAGNOSTIC
515: if ((m->m_flags & M_PKTHDR) == 0)
516: panic("ipintr no HDR");
1.89 itojun 517: #endif
1.164 matt 518:
1.1 cgd 519: /*
520: * If no IP addresses have been set yet but the interfaces
521: * are receiving, can't do anything with incoming packets yet.
522: */
1.181 jonathan 523: if (TAILQ_FIRST(&in_ifaddrhead) == 0)
1.1 cgd 524: goto bad;
1.266 thorpej 525: IP_STATINC(IP_STAT_TOTAL);
1.154 thorpej 526: /*
527: * If the IP header is not aligned, slurp it up into a new
528: * mbuf with space for link headers, in the event we forward
529: * it. Otherwise, if it is aligned, make sure the entire
530: * base IP header is in the first mbuf of the chain.
531: */
1.244 christos 532: if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
1.154 thorpej 533: if ((m = m_copyup(m, sizeof(struct ip),
534: (max_linkhdr + 3) & ~3)) == NULL) {
535: /* XXXJRT new stat, please */
1.266 thorpej 536: IP_STATINC(IP_STAT_TOOSMALL);
1.154 thorpej 537: return;
538: }
539: } else if (__predict_false(m->m_len < sizeof (struct ip))) {
540: if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
1.266 thorpej 541: IP_STATINC(IP_STAT_TOOSMALL);
1.154 thorpej 542: return;
543: }
1.1 cgd 544: }
545: ip = mtod(m, struct ip *);
1.13 mycroft 546: if (ip->ip_v != IPVERSION) {
1.266 thorpej 547: IP_STATINC(IP_STAT_BADVERS);
1.13 mycroft 548: goto bad;
549: }
1.1 cgd 550: hlen = ip->ip_hl << 2;
551: if (hlen < sizeof(struct ip)) { /* minimum header length */
1.266 thorpej 552: IP_STATINC(IP_STAT_BADHLEN);
1.1 cgd 553: goto bad;
554: }
555: if (hlen > m->m_len) {
556: if ((m = m_pullup(m, hlen)) == 0) {
1.266 thorpej 557: IP_STATINC(IP_STAT_BADHLEN);
1.89 itojun 558: return;
1.1 cgd 559: }
560: ip = mtod(m, struct ip *);
561: }
1.98 thorpej 562:
1.85 hwr 563: /*
1.99 thorpej 564: * RFC1122: packets with a multicast source address are
1.98 thorpej 565: * not allowed.
1.85 hwr 566: */
567: if (IN_MULTICAST(ip->ip_src.s_addr)) {
1.266 thorpej 568: IP_STATINC(IP_STAT_BADADDR);
1.85 hwr 569: goto bad;
1.129 itojun 570: }
571:
572: /* 127/8 must not appear on wire - RFC1122 */
573: if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
574: (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
1.130 itojun 575: if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
1.266 thorpej 576: IP_STATINC(IP_STAT_BADADDR);
1.129 itojun 577: goto bad;
1.130 itojun 578: }
1.85 hwr 579: }
580:
1.135 thorpej 581: switch (m->m_pkthdr.csum_flags &
1.137 thorpej 582: ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) |
1.135 thorpej 583: M_CSUM_IPv4_BAD)) {
584: case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
585: INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
586: goto badcsum;
587:
588: case M_CSUM_IPv4:
589: /* Checksum was okay. */
590: INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
591: break;
592:
593: default:
1.206 thorpej 594: /*
595: * Must compute it ourselves. Maybe skip checksum on
596: * loopback interfaces.
597: */
598: if (__predict_true(!(m->m_pkthdr.rcvif->if_flags &
599: IFF_LOOPBACK) || ip_do_loopback_cksum)) {
600: INET_CSUM_COUNTER_INCR(&ip_swcsum);
601: if (in_cksum(m, hlen) != 0)
602: goto badcsum;
603: }
1.135 thorpej 604: break;
1.1 cgd 605: }
606:
1.121 thorpej 607: /* Retrieve the packet length. */
608: len = ntohs(ip->ip_len);
1.81 proff 609:
610: /*
611: * Check for additional length bogosity
612: */
1.84 proff 613: if (len < hlen) {
1.266 thorpej 614: IP_STATINC(IP_STAT_BADLEN);
1.81 proff 615: goto bad;
616: }
1.1 cgd 617:
618: /*
619: * Check that the amount of data in the buffers
620: * is as at least much as the IP header would have us expect.
621: * Trim mbufs if longer than we expect.
622: * Drop packet if shorter than we expect.
623: */
1.35 mycroft 624: if (m->m_pkthdr.len < len) {
1.266 thorpej 625: IP_STATINC(IP_STAT_TOOSHORT);
1.1 cgd 626: goto bad;
627: }
1.35 mycroft 628: if (m->m_pkthdr.len > len) {
1.1 cgd 629: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 630: m->m_len = len;
631: m->m_pkthdr.len = len;
1.1 cgd 632: } else
1.35 mycroft 633: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 634: }
635:
1.193 scw 636: #if defined(IPSEC)
1.149 wiz 637: /* ipflow (IP fast forwarding) is not compatible with IPsec. */
1.94 itojun 638: m->m_flags &= ~M_CANFASTFWD;
639: #else
1.64 thorpej 640: /*
641: * Assume that we can create a fast-forward IP flow entry
642: * based on this packet.
643: */
644: m->m_flags |= M_CANFASTFWD;
1.94 itojun 645: #endif
1.64 thorpej 646:
1.36 mrg 647: #ifdef PFIL_HOOKS
1.33 mrg 648: /*
1.64 thorpej 649: * Run through list of hooks for input packets. If there are any
650: * filters which require that additional packets in the flow are
651: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
652: * Note that filters must _never_ set this flag, as another filter
653: * in the list may have previously cleared it.
1.33 mrg 654: */
1.127 itojun 655: /*
656: * let ipfilter look at packet on the wire,
657: * not the decapsulated packet.
658: */
659: #ifdef IPSEC
1.136 itojun 660: if (!ipsec_getnhist(m))
1.186 scw 661: #elif defined(FAST_IPSEC)
662: if (!ipsec_indone(m))
1.127 itojun 663: #else
664: if (1)
665: #endif
666: {
1.169 itojun 667: struct in_addr odst;
668:
669: odst = ip->ip_dst;
1.127 itojun 670: if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif,
1.168 itojun 671: PFIL_IN) != 0)
672: return;
1.127 itojun 673: if (m == NULL)
674: return;
675: ip = mtod(m, struct ip *);
1.142 darrenr 676: hlen = ip->ip_hl << 2;
1.205 darrenr 677: /*
678: * XXX The setting of "srcrt" here is to prevent ip_forward()
679: * from generating ICMP redirects for packets that have
680: * been redirected by a hook back out on to the same LAN that
681: * they came from and is not an indication that the packet
682: * is being inffluenced by source routing options. This
683: * allows things like
684: * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp"
685: * where tlp0 is both on the 1.1.1.0/24 network and is the
686: * default route for hosts on 1.1.1.0/24. Of course this
687: * also requires a "map tlp0 ..." to complete the story.
688: * One might argue whether or not this kind of network config.
1.212 perry 689: * should be supported in this manner...
1.205 darrenr 690: */
1.169 itojun 691: srcrt = (odst.s_addr != ip->ip_dst.s_addr);
1.127 itojun 692: }
1.36 mrg 693: #endif /* PFIL_HOOKS */
1.123 thorpej 694:
695: #ifdef ALTQ
696: /* XXX Temporary until ALTQ is changed to use a pfil hook */
697: if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) {
698: /* packet dropped by traffic conditioner */
699: return;
700: }
701: #endif
1.121 thorpej 702:
703: /*
1.1 cgd 704: * Process options and, if not destined for us,
705: * ship it on. ip_dooptions returns 1 when an
706: * error was detected (causing an icmp message
707: * to be sent and the original packet to be freed).
708: */
709: ip_nhops = 0; /* for source routed packets */
710: if (hlen > sizeof (struct ip) && ip_dooptions(m))
1.89 itojun 711: return;
1.1 cgd 712:
713: /*
1.165 christos 714: * Enable a consistency check between the destination address
715: * and the arrival interface for a unicast packet (the RFC 1122
716: * strong ES model) if IP forwarding is disabled and the packet
717: * is not locally generated.
718: *
719: * XXX - Checking also should be disabled if the destination
720: * address is ipnat'ed to a different interface.
721: *
722: * XXX - Checking is incompatible with IP aliases added
723: * to the loopback interface instead of the interface where
724: * the packets are received.
725: *
726: * XXX - We need to add a per ifaddr flag for this so that
727: * we get finer grain control.
728: */
729: checkif = ip_checkinterface && (ipforwarding == 0) &&
730: (m->m_pkthdr.rcvif != NULL) &&
731: ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0);
732:
733: /*
1.1 cgd 734: * Check our list of addresses, to see if the packet is for us.
1.100 itojun 735: *
736: * Traditional 4.4BSD did not consult IFF_UP at all.
737: * The behavior here is to treat addresses on !IFF_UP interface
738: * as not mine.
1.1 cgd 739: */
1.100 itojun 740: downmatch = 0;
1.140 matt 741: LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) {
1.97 itojun 742: if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
1.165 christos 743: if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif)
744: continue;
1.97 itojun 745: if ((ia->ia_ifp->if_flags & IFF_UP) != 0)
746: break;
1.100 itojun 747: else
748: downmatch++;
1.97 itojun 749: }
750: }
1.86 thorpej 751: if (ia != NULL)
752: goto ours;
1.225 christos 753: if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
1.209 matt 754: IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) {
1.140 matt 755: if (ifa->ifa_addr->sa_family != AF_INET)
756: continue;
1.57 tls 757: ia = ifatoia(ifa);
1.35 mycroft 758: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
759: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 760: /*
761: * Look for all-0's host part (old broadcast addr),
762: * either for subnet or net.
763: */
764: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 765: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 766: goto ours;
1.57 tls 767: /*
768: * An interface with IP address zero accepts
769: * all packets that arrive on that interface.
770: */
771: if (in_nullhost(ia->ia_addr.sin_addr))
772: goto ours;
1.1 cgd 773: }
774: }
1.18 mycroft 775: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 776: struct in_multi *inm;
777: #ifdef MROUTING
778: extern struct socket *ip_mrouter;
1.10 brezak 779:
1.4 hpeyerl 780: if (ip_mrouter) {
781: /*
782: * If we are acting as a multicast router, all
783: * incoming multicast packets are passed to the
784: * kernel-level multicast forwarding function.
785: * The packet is returned (relatively) intact; if
786: * ip_mforward() returns a non-zero value, the packet
787: * must be discarded, else it may be accepted below.
788: *
789: * (The IP ident field is put in the same byte order
790: * as expected when ip_mforward() is called from
791: * ip_output().)
792: */
1.13 mycroft 793: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
1.266 thorpej 794: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 795: m_freem(m);
1.89 itojun 796: return;
1.4 hpeyerl 797: }
798:
799: /*
800: * The process-level routing demon needs to receive
801: * all multicast IGMP packets, whether or not this
802: * host belongs to their destination groups.
803: */
804: if (ip->ip_p == IPPROTO_IGMP)
805: goto ours;
1.266 thorpej 806: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 807: }
808: #endif
809: /*
810: * See if we belong to the destination multicast group on the
811: * arrival interface.
812: */
813: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
814: if (inm == NULL) {
1.266 thorpej 815: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 816: m_freem(m);
1.89 itojun 817: return;
1.4 hpeyerl 818: }
819: goto ours;
820: }
1.19 mycroft 821: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 822: in_nullhost(ip->ip_dst))
1.1 cgd 823: goto ours;
824:
825: /*
826: * Not for us; forward if possible and desirable.
827: */
828: if (ipforwarding == 0) {
1.266 thorpej 829: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 830: m_freem(m);
1.100 itojun 831: } else {
832: /*
833: * If ip_dst matched any of my address on !IFF_UP interface,
834: * and there's no IFF_UP interface that matches ip_dst,
835: * send icmp unreach. Forwarding it will result in in-kernel
836: * forwarding loop till TTL goes to 0.
837: */
838: if (downmatch) {
839: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
1.266 thorpej 840: IP_STATINC(IP_STAT_CANTFORWARD);
1.100 itojun 841: return;
842: }
1.145 itojun 843: #ifdef IPSEC
844: if (ipsec4_in_reject(m, NULL)) {
1.267 thorpej 845: IPSEC_STATINC(IPSEC_STAT_IN_POLVIO);
1.145 itojun 846: goto bad;
847: }
848: #endif
1.173 jonathan 849: #ifdef FAST_IPSEC
850: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
851: s = splsoftnet();
852: if (mtag != NULL) {
853: tdbi = (struct tdb_ident *)(mtag + 1);
854: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
855: } else {
856: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
1.212 perry 857: IP_FORWARDING, &error);
1.173 jonathan 858: }
859: if (sp == NULL) { /* NB: can happen if error */
860: splx(s);
861: /*XXX error stat???*/
862: DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/
863: goto bad;
864: }
865:
866: /*
867: * Check security policy against packet attributes.
868: */
869: error = ipsec_in_reject(sp, m);
870: KEY_FREESP(&sp);
871: splx(s);
872: if (error) {
1.266 thorpej 873: IP_STATINC(IP_STAT_CANTFORWARD);
1.173 jonathan 874: goto bad;
1.193 scw 875: }
876:
877: /*
878: * Peek at the outbound SP for this packet to determine if
879: * it's a Fast Forward candidate.
880: */
881: mtag = m_tag_find(m, PACKET_TAG_IPSEC_PENDING_TDB, NULL);
882: if (mtag != NULL)
883: m->m_flags &= ~M_CANFASTFWD;
884: else {
885: s = splsoftnet();
886: sp = ipsec4_checkpolicy(m, IPSEC_DIR_OUTBOUND,
887: (IP_FORWARDING |
888: (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
889: &error, NULL);
890: if (sp != NULL) {
891: m->m_flags &= ~M_CANFASTFWD;
892: KEY_FREESP(&sp);
893: }
894: splx(s);
1.173 jonathan 895: }
896: #endif /* FAST_IPSEC */
1.145 itojun 897:
1.169 itojun 898: ip_forward(m, srcrt);
1.100 itojun 899: }
1.89 itojun 900: return;
1.1 cgd 901:
902: ours:
903: /*
904: * If offset or IP_MF are set, must reassemble.
905: * Otherwise, nothing need be done.
906: * (We could look in the reassembly queue to see
907: * if the packet was previously fragmented,
908: * but it's not worth the time; just let them time out.)
909: */
1.155 itojun 910: if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
1.258 matt 911: uint16_t off;
912: /*
913: * Prevent TCP blind data attacks by not allowing non-initial
914: * fragments to start at less than 68 bytes (minimal fragment
1.259 matt 915: * size) and making sure the first fragment is at least 68
916: * bytes.
1.258 matt 917: */
1.260 matt 918: off = (ntohs(ip->ip_off) & IP_OFFMASK) << 3;
1.259 matt 919: if ((off > 0 ? off + hlen : len) < IP_MINFRAGSIZE - 1) {
1.266 thorpej 920: IP_STATINC(IP_STAT_BADFRAGS);
1.258 matt 921: goto bad;
922: }
1.1 cgd 923: /*
924: * Look for queue of fragments
925: * of this datagram.
926: */
1.75 thorpej 927: IPQ_LOCK();
1.190 jonathan 928: hash = IPREASS_HASH(ip->ip_src.s_addr, ip->ip_id);
1.250 dyoung 929: LIST_FOREACH(fp, &ipq[hash], ipq_q) {
1.1 cgd 930: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 931: in_hosteq(ip->ip_src, fp->ipq_src) &&
932: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.260 matt 933: ip->ip_p == fp->ipq_p) {
934: /*
935: * Make sure the TOS is matches previous
936: * fragments.
937: */
938: if (ip->ip_tos != fp->ipq_tos) {
1.266 thorpej 939: IP_STATINC(IP_STAT_BADFRAGS);
1.276 rmind 940: IPQ_UNLOCK();
1.260 matt 941: goto bad;
942: }
1.1 cgd 943: goto found;
1.260 matt 944: }
1.190 jonathan 945: }
1.1 cgd 946: fp = 0;
947: found:
948:
949: /*
950: * Adjust ip_len to not reflect header,
1.25 cgd 951: * set ipqe_mff if more fragments are expected,
1.1 cgd 952: * convert offset of this to bytes.
953: */
1.155 itojun 954: ip->ip_len = htons(ntohs(ip->ip_len) - hlen);
955: mff = (ip->ip_off & htons(IP_MF)) != 0;
1.25 cgd 956: if (mff) {
1.16 cgd 957: /*
958: * Make sure that fragments have a data length
959: * that's a non-zero multiple of 8 bytes.
960: */
1.155 itojun 961: if (ntohs(ip->ip_len) == 0 ||
962: (ntohs(ip->ip_len) & 0x7) != 0) {
1.266 thorpej 963: IP_STATINC(IP_STAT_BADFRAGS);
1.75 thorpej 964: IPQ_UNLOCK();
1.16 cgd 965: goto bad;
966: }
967: }
1.155 itojun 968: ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3);
1.1 cgd 969:
970: /*
971: * If datagram marked as having more fragments
972: * or if this is not the first fragment,
973: * attempt reassembly; if it succeeds, proceed.
974: */
1.155 itojun 975: if (mff || ip->ip_off != htons(0)) {
1.266 thorpej 976: IP_STATINC(IP_STAT_FRAGMENTS);
1.233 tls 977: s = splvm();
1.72 thorpej 978: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.233 tls 979: splx(s);
1.25 cgd 980: if (ipqe == NULL) {
1.266 thorpej 981: IP_STATINC(IP_STAT_RCVMEMDROP);
1.75 thorpej 982: IPQ_UNLOCK();
1.25 cgd 983: goto bad;
984: }
985: ipqe->ipqe_mff = mff;
1.50 thorpej 986: ipqe->ipqe_m = m;
1.25 cgd 987: ipqe->ipqe_ip = ip;
1.190 jonathan 988: m = ip_reass(ipqe, fp, &ipq[hash]);
1.75 thorpej 989: if (m == 0) {
990: IPQ_UNLOCK();
1.89 itojun 991: return;
1.75 thorpej 992: }
1.266 thorpej 993: IP_STATINC(IP_STAT_REASSEMBLED);
1.50 thorpej 994: ip = mtod(m, struct ip *);
1.74 thorpej 995: hlen = ip->ip_hl << 2;
1.155 itojun 996: ip->ip_len = htons(ntohs(ip->ip_len) + hlen);
1.1 cgd 997: } else
998: if (fp)
999: ip_freef(fp);
1.75 thorpej 1000: IPQ_UNLOCK();
1.79 mycroft 1001: }
1.128 itojun 1002:
1.173 jonathan 1003: #if defined(IPSEC)
1.128 itojun 1004: /*
1005: * enforce IPsec policy checking if we are seeing last header.
1006: * note that we do not visit this with protocols with pcb layer
1007: * code - like udp/tcp/raw ip.
1008: */
1009: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 &&
1010: ipsec4_in_reject(m, NULL)) {
1.267 thorpej 1011: IPSEC_STATINC(IPSEC_STAT_IN_POLVIO);
1.128 itojun 1012: goto bad;
1013: }
1014: #endif
1.226 liamjfoy 1015: #ifdef FAST_IPSEC
1.173 jonathan 1016: /*
1017: * enforce IPsec policy checking if we are seeing last header.
1018: * note that we do not visit this with protocols with pcb layer
1019: * code - like udp/tcp/raw ip.
1020: */
1021: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) {
1022: /*
1023: * Check if the packet has already had IPsec processing
1024: * done. If so, then just pass it along. This tag gets
1025: * set during AH, ESP, etc. input handling, before the
1026: * packet is returned to the ip input queue for delivery.
1.212 perry 1027: */
1.173 jonathan 1028: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
1029: s = splsoftnet();
1030: if (mtag != NULL) {
1031: tdbi = (struct tdb_ident *)(mtag + 1);
1032: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
1033: } else {
1034: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
1.212 perry 1035: IP_FORWARDING, &error);
1.173 jonathan 1036: }
1037: if (sp != NULL) {
1038: /*
1039: * Check security policy against packet attributes.
1040: */
1041: error = ipsec_in_reject(sp, m);
1042: KEY_FREESP(&sp);
1043: } else {
1044: /* XXX error stat??? */
1045: error = EINVAL;
1046: DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/
1047: }
1048: splx(s);
1049: if (error)
1050: goto bad;
1051: }
1052: #endif /* FAST_IPSEC */
1.1 cgd 1053:
1054: /*
1055: * Switch out to protocol's input routine.
1056: */
1.82 aidan 1057: #if IFA_STATS
1.122 itojun 1058: if (ia && ip)
1.155 itojun 1059: ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
1.82 aidan 1060: #endif
1.266 thorpej 1061: IP_STATINC(IP_STAT_DELIVERED);
1.89 itojun 1062: {
1063: int off = hlen, nh = ip->ip_p;
1064:
1065: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
1066: return;
1067: }
1.1 cgd 1068: bad:
1069: m_freem(m);
1.135 thorpej 1070: return;
1071:
1072: badcsum:
1.266 thorpej 1073: IP_STATINC(IP_STAT_BADSUM);
1.135 thorpej 1074: m_freem(m);
1.1 cgd 1075: }
1076:
1077: /*
1078: * Take incoming datagram fragment and try to
1079: * reassemble it into whole datagram. If a chain for
1080: * reassembly of this datagram already exists, then it
1081: * is given as fp; otherwise have to make a chain.
1082: */
1.50 thorpej 1083: struct mbuf *
1.211 perry 1084: ip_reass(struct ipqent *ipqe, struct ipq *fp, struct ipqhead *ipqhead)
1.1 cgd 1085: {
1.109 augustss 1086: struct mbuf *m = ipqe->ipqe_m;
1087: struct ipqent *nq, *p, *q;
1.25 cgd 1088: struct ip *ip;
1.1 cgd 1089: struct mbuf *t;
1.25 cgd 1090: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.233 tls 1091: int i, next, s;
1.1 cgd 1092:
1.75 thorpej 1093: IPQ_LOCK_CHECK();
1094:
1.1 cgd 1095: /*
1096: * Presence of header sizes in mbufs
1097: * would confuse code below.
1098: */
1099: m->m_data += hlen;
1100: m->m_len -= hlen;
1101:
1.194 jonathan 1102: #ifdef notyet
1103: /* make sure fragment limit is up-to-date */
1104: CHECK_NMBCLUSTER_PARAMS();
1105:
1106: /* If we have too many fragments, drop the older half. */
1107: if (ip_nfrags >= ip_maxfrags)
1108: ip_reass_drophalf(void);
1109: #endif
1110:
1.1 cgd 1111: /*
1.192 jonathan 1112: * We are about to add a fragment; increment frag count.
1113: */
1114: ip_nfrags++;
1.212 perry 1115:
1.192 jonathan 1116: /*
1.1 cgd 1117: * If first fragment to arrive, create a reassembly queue.
1118: */
1119: if (fp == 0) {
1.131 itojun 1120: /*
1121: * Enforce upper bound on number of fragmented packets
1122: * for which we attempt reassembly;
1123: * If maxfrag is 0, never accept fragments.
1124: * If maxfrag is -1, accept all fragments without limitation.
1125: */
1126: if (ip_maxfragpackets < 0)
1127: ;
1128: else if (ip_nfragpackets >= ip_maxfragpackets)
1129: goto dropfrag;
1130: ip_nfragpackets++;
1.277 cegger 1131: fp = malloc(sizeof (struct ipq), M_FTABLE, M_NOWAIT);
1.50 thorpej 1132: if (fp == NULL)
1.1 cgd 1133: goto dropfrag;
1.190 jonathan 1134: LIST_INSERT_HEAD(ipqhead, fp, ipq_q);
1.192 jonathan 1135: fp->ipq_nfrags = 1;
1.1 cgd 1136: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 1137: fp->ipq_p = ipqe->ipqe_ip->ip_p;
1138: fp->ipq_id = ipqe->ipqe_ip->ip_id;
1.260 matt 1139: fp->ipq_tos = ipqe->ipqe_ip->ip_tos;
1.148 matt 1140: TAILQ_INIT(&fp->ipq_fragq);
1.25 cgd 1141: fp->ipq_src = ipqe->ipqe_ip->ip_src;
1142: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
1143: p = NULL;
1.1 cgd 1144: goto insert;
1.192 jonathan 1145: } else {
1146: fp->ipq_nfrags++;
1.1 cgd 1147: }
1148:
1149: /*
1150: * Find a segment which begins after this one does.
1151: */
1.148 matt 1152: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1153: p = q, q = TAILQ_NEXT(q, ipqe_q))
1.155 itojun 1154: if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off))
1.1 cgd 1155: break;
1156:
1157: /*
1158: * If there is a preceding segment, it may provide some of
1159: * our data already. If so, drop the data from the incoming
1160: * segment. If it provides all of our data, drop us.
1161: */
1.25 cgd 1162: if (p != NULL) {
1.155 itojun 1163: i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) -
1164: ntohs(ipqe->ipqe_ip->ip_off);
1.1 cgd 1165: if (i > 0) {
1.155 itojun 1166: if (i >= ntohs(ipqe->ipqe_ip->ip_len))
1.1 cgd 1167: goto dropfrag;
1.50 thorpej 1168: m_adj(ipqe->ipqe_m, i);
1.155 itojun 1169: ipqe->ipqe_ip->ip_off =
1170: htons(ntohs(ipqe->ipqe_ip->ip_off) + i);
1171: ipqe->ipqe_ip->ip_len =
1172: htons(ntohs(ipqe->ipqe_ip->ip_len) - i);
1.1 cgd 1173: }
1174: }
1175:
1176: /*
1177: * While we overlap succeeding segments trim them or,
1178: * if they are completely covered, dequeue them.
1179: */
1.155 itojun 1180: for (; q != NULL &&
1181: ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) >
1182: ntohs(q->ipqe_ip->ip_off); q = nq) {
1183: i = (ntohs(ipqe->ipqe_ip->ip_off) +
1184: ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off);
1185: if (i < ntohs(q->ipqe_ip->ip_len)) {
1186: q->ipqe_ip->ip_len =
1187: htons(ntohs(q->ipqe_ip->ip_len) - i);
1188: q->ipqe_ip->ip_off =
1189: htons(ntohs(q->ipqe_ip->ip_off) + i);
1.50 thorpej 1190: m_adj(q->ipqe_m, i);
1.1 cgd 1191: break;
1192: }
1.148 matt 1193: nq = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1194: m_freem(q->ipqe_m);
1.148 matt 1195: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.233 tls 1196: s = splvm();
1.72 thorpej 1197: pool_put(&ipqent_pool, q);
1.233 tls 1198: splx(s);
1.192 jonathan 1199: fp->ipq_nfrags--;
1200: ip_nfrags--;
1.1 cgd 1201: }
1202:
1203: insert:
1204: /*
1205: * Stick new segment in its place;
1206: * check for complete reassembly.
1207: */
1.25 cgd 1208: if (p == NULL) {
1.148 matt 1209: TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
1.25 cgd 1210: } else {
1.148 matt 1211: TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q);
1.25 cgd 1212: }
1.1 cgd 1213: next = 0;
1.148 matt 1214: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1215: p = q, q = TAILQ_NEXT(q, ipqe_q)) {
1.155 itojun 1216: if (ntohs(q->ipqe_ip->ip_off) != next)
1.1 cgd 1217: return (0);
1.155 itojun 1218: next += ntohs(q->ipqe_ip->ip_len);
1.1 cgd 1219: }
1.25 cgd 1220: if (p->ipqe_mff)
1.1 cgd 1221: return (0);
1222:
1223: /*
1.41 thorpej 1224: * Reassembly is complete. Check for a bogus message size and
1225: * concatenate fragments.
1.1 cgd 1226: */
1.148 matt 1227: q = TAILQ_FIRST(&fp->ipq_fragq);
1.25 cgd 1228: ip = q->ipqe_ip;
1.41 thorpej 1229: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
1.266 thorpej 1230: IP_STATINC(IP_STAT_TOOLONG);
1.41 thorpej 1231: ip_freef(fp);
1232: return (0);
1233: }
1.50 thorpej 1234: m = q->ipqe_m;
1.1 cgd 1235: t = m->m_next;
1236: m->m_next = 0;
1237: m_cat(m, t);
1.148 matt 1238: nq = TAILQ_NEXT(q, ipqe_q);
1.233 tls 1239: s = splvm();
1.72 thorpej 1240: pool_put(&ipqent_pool, q);
1.233 tls 1241: splx(s);
1.25 cgd 1242: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 1243: t = q->ipqe_m;
1.148 matt 1244: nq = TAILQ_NEXT(q, ipqe_q);
1.233 tls 1245: s = splvm();
1.72 thorpej 1246: pool_put(&ipqent_pool, q);
1.233 tls 1247: splx(s);
1.1 cgd 1248: m_cat(m, t);
1249: }
1.192 jonathan 1250: ip_nfrags -= fp->ipq_nfrags;
1.1 cgd 1251:
1252: /*
1253: * Create header for new ip packet by
1254: * modifying header of first packet;
1255: * dequeue and discard fragment reassembly header.
1256: * Make header visible.
1257: */
1.155 itojun 1258: ip->ip_len = htons(next);
1.25 cgd 1259: ip->ip_src = fp->ipq_src;
1260: ip->ip_dst = fp->ipq_dst;
1261: LIST_REMOVE(fp, ipq_q);
1.277 cegger 1262: free(fp, M_FTABLE);
1.131 itojun 1263: ip_nfragpackets--;
1.1 cgd 1264: m->m_len += (ip->ip_hl << 2);
1265: m->m_data -= (ip->ip_hl << 2);
1266: /* some debugging cruft by sklower, below, will go away soon */
1267: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
1.109 augustss 1268: int plen = 0;
1.50 thorpej 1269: for (t = m; t; t = t->m_next)
1270: plen += t->m_len;
1271: m->m_pkthdr.len = plen;
1.213 yamt 1272: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1273: }
1.50 thorpej 1274: return (m);
1.1 cgd 1275:
1276: dropfrag:
1.192 jonathan 1277: if (fp != 0)
1278: fp->ipq_nfrags--;
1279: ip_nfrags--;
1.266 thorpej 1280: IP_STATINC(IP_STAT_FRAGDROPPED);
1.1 cgd 1281: m_freem(m);
1.233 tls 1282: s = splvm();
1.72 thorpej 1283: pool_put(&ipqent_pool, ipqe);
1.233 tls 1284: splx(s);
1.1 cgd 1285: return (0);
1286: }
1287:
1288: /*
1289: * Free a fragment reassembly header and all
1290: * associated datagrams.
1291: */
1.8 mycroft 1292: void
1.211 perry 1293: ip_freef(struct ipq *fp)
1.1 cgd 1294: {
1.109 augustss 1295: struct ipqent *q, *p;
1.192 jonathan 1296: u_int nfrags = 0;
1.233 tls 1297: int s;
1.1 cgd 1298:
1.75 thorpej 1299: IPQ_LOCK_CHECK();
1300:
1.148 matt 1301: for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) {
1302: p = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1303: m_freem(q->ipqe_m);
1.192 jonathan 1304: nfrags++;
1.148 matt 1305: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.233 tls 1306: s = splvm();
1.72 thorpej 1307: pool_put(&ipqent_pool, q);
1.233 tls 1308: splx(s);
1.1 cgd 1309: }
1.192 jonathan 1310:
1311: if (nfrags != fp->ipq_nfrags)
1312: printf("ip_freef: nfrags %d != %d\n", fp->ipq_nfrags, nfrags);
1313: ip_nfrags -= nfrags;
1.25 cgd 1314: LIST_REMOVE(fp, ipq_q);
1.277 cegger 1315: free(fp, M_FTABLE);
1.131 itojun 1316: ip_nfragpackets--;
1.1 cgd 1317: }
1318:
1319: /*
1.194 jonathan 1320: * IP reassembly TTL machinery for multiplicative drop.
1321: */
1322: static u_int fragttl_histo[(IPFRAGTTL+1)];
1323:
1324:
1325: /*
1326: * Decrement TTL of all reasembly queue entries by `ticks'.
1327: * Count number of distinct fragments (as opposed to partial, fragmented
1328: * datagrams) in the reassembly queue. While we traverse the entire
1329: * reassembly queue, compute and return the median TTL over all fragments.
1330: */
1331: static u_int
1332: ip_reass_ttl_decr(u_int ticks)
1333: {
1.198 matt 1334: u_int nfrags, median, dropfraction, keepfraction;
1.194 jonathan 1335: struct ipq *fp, *nfp;
1.198 matt 1336: int i;
1.212 perry 1337:
1.194 jonathan 1338: nfrags = 0;
1339: memset(fragttl_histo, 0, sizeof fragttl_histo);
1.212 perry 1340:
1.194 jonathan 1341: for (i = 0; i < IPREASS_NHASH; i++) {
1342: for (fp = LIST_FIRST(&ipq[i]); fp != NULL; fp = nfp) {
1343: fp->ipq_ttl = ((fp->ipq_ttl <= ticks) ?
1344: 0 : fp->ipq_ttl - ticks);
1345: nfp = LIST_NEXT(fp, ipq_q);
1346: if (fp->ipq_ttl == 0) {
1.266 thorpej 1347: IP_STATINC(IP_STAT_FRAGTIMEOUT);
1.194 jonathan 1348: ip_freef(fp);
1349: } else {
1350: nfrags += fp->ipq_nfrags;
1351: fragttl_histo[fp->ipq_ttl] += fp->ipq_nfrags;
1352: }
1353: }
1354: }
1355:
1356: KASSERT(ip_nfrags == nfrags);
1357:
1358: /* Find median (or other drop fraction) in histogram. */
1359: dropfraction = (ip_nfrags / 2);
1360: keepfraction = ip_nfrags - dropfraction;
1361: for (i = IPFRAGTTL, median = 0; i >= 0; i--) {
1362: median += fragttl_histo[i];
1363: if (median >= keepfraction)
1364: break;
1365: }
1366:
1367: /* Return TTL of median (or other fraction). */
1368: return (u_int)i;
1369: }
1370:
1371: void
1372: ip_reass_drophalf(void)
1373: {
1374:
1375: u_int median_ticks;
1376: /*
1377: * Compute median TTL of all fragments, and count frags
1378: * with that TTL or lower (roughly half of all fragments).
1379: */
1380: median_ticks = ip_reass_ttl_decr(0);
1381:
1382: /* Drop half. */
1383: median_ticks = ip_reass_ttl_decr(median_ticks);
1384:
1385: }
1386:
1387: /*
1.1 cgd 1388: * IP timer processing;
1389: * if a timer expires on a reassembly
1390: * queue, discard it.
1391: */
1.8 mycroft 1392: void
1.211 perry 1393: ip_slowtimo(void)
1.1 cgd 1394: {
1.191 jonathan 1395: static u_int dropscanidx = 0;
1396: u_int i;
1.194 jonathan 1397: u_int median_ttl;
1.268 ad 1398:
1399: mutex_enter(softnet_lock);
1400: KERNEL_LOCK(1, NULL);
1.1 cgd 1401:
1.75 thorpej 1402: IPQ_LOCK();
1.194 jonathan 1403:
1404: /* Age TTL of all fragments by 1 tick .*/
1405: median_ttl = ip_reass_ttl_decr(1);
1406:
1407: /* make sure fragment limit is up-to-date */
1408: CHECK_NMBCLUSTER_PARAMS();
1409:
1410: /* If we have too many fragments, drop the older half. */
1411: if (ip_nfrags > ip_maxfrags)
1412: ip_reass_ttl_decr(median_ttl);
1413:
1.131 itojun 1414: /*
1.194 jonathan 1415: * If we are over the maximum number of fragmented packets
1.131 itojun 1416: * (due to the limit being lowered), drain off
1.190 jonathan 1417: * enough to get down to the new limit. Start draining
1418: * from the reassembly hashqueue most recently drained.
1.131 itojun 1419: */
1420: if (ip_maxfragpackets < 0)
1421: ;
1422: else {
1.190 jonathan 1423: int wrapped = 0;
1424:
1425: i = dropscanidx;
1426: while (ip_nfragpackets > ip_maxfragpackets && wrapped == 0) {
1427: while (LIST_FIRST(&ipq[i]) != NULL)
1428: ip_freef(LIST_FIRST(&ipq[i]));
1429: if (++i >= IPREASS_NHASH) {
1430: i = 0;
1431: }
1432: /*
1433: * Dont scan forever even if fragment counters are
1434: * wrong: stop after scanning entire reassembly queue.
1435: */
1436: if (i == dropscanidx)
1437: wrapped = 1;
1438: }
1439: dropscanidx = i;
1.131 itojun 1440: }
1.75 thorpej 1441: IPQ_UNLOCK();
1.268 ad 1442:
1443: KERNEL_UNLOCK_ONE(NULL);
1444: mutex_exit(softnet_lock);
1.1 cgd 1445: }
1446:
1447: /*
1.270 ad 1448: * Drain off all datagram fragments. Don't acquire softnet_lock as
1449: * can be called from hardware interrupt context.
1.1 cgd 1450: */
1.8 mycroft 1451: void
1.211 perry 1452: ip_drain(void)
1.1 cgd 1453: {
1454:
1.268 ad 1455: KERNEL_LOCK(1, NULL);
1456:
1.75 thorpej 1457: /*
1458: * We may be called from a device's interrupt context. If
1459: * the ipq is already busy, just bail out now.
1460: */
1.268 ad 1461: if (ipq_lock_try() != 0) {
1462: /*
1463: * Drop half the total fragments now. If more mbufs are
1464: * needed, we will be called again soon.
1465: */
1466: ip_reass_drophalf();
1467: IPQ_UNLOCK();
1468: }
1.75 thorpej 1469:
1.268 ad 1470: KERNEL_UNLOCK_ONE(NULL);
1.1 cgd 1471: }
1472:
1473: /*
1474: * Do option processing on a datagram,
1475: * possibly discarding it if bad options are encountered,
1476: * or forwarding it if source-routed.
1477: * Returns 1 if packet has been forwarded/freed,
1478: * 0 if the packet should be processed further.
1479: */
1.8 mycroft 1480: int
1.211 perry 1481: ip_dooptions(struct mbuf *m)
1.1 cgd 1482: {
1.109 augustss 1483: struct ip *ip = mtod(m, struct ip *);
1484: u_char *cp, *cp0;
1485: struct ip_timestamp *ipt;
1486: struct in_ifaddr *ia;
1.1 cgd 1487: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 1488: struct in_addr dst;
1.1 cgd 1489: n_time ntime;
1490:
1.13 mycroft 1491: dst = ip->ip_dst;
1.1 cgd 1492: cp = (u_char *)(ip + 1);
1493: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
1494: for (; cnt > 0; cnt -= optlen, cp += optlen) {
1495: opt = cp[IPOPT_OPTVAL];
1496: if (opt == IPOPT_EOL)
1497: break;
1498: if (opt == IPOPT_NOP)
1499: optlen = 1;
1500: else {
1.113 itojun 1501: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1502: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1503: goto bad;
1504: }
1.1 cgd 1505: optlen = cp[IPOPT_OLEN];
1.114 itojun 1506: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 1507: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1508: goto bad;
1509: }
1510: }
1511: switch (opt) {
1512:
1513: default:
1514: break;
1515:
1516: /*
1517: * Source routing with record.
1518: * Find interface with current destination address.
1519: * If none on this machine then drop if strictly routed,
1520: * or do nothing if loosely routed.
1521: * Record interface address and bring up next address
1522: * component. If strictly routed make sure next
1523: * address is on directly accessible net.
1524: */
1525: case IPOPT_LSRR:
1526: case IPOPT_SSRR:
1.47 cjs 1527: if (ip_allowsrcrt == 0) {
1528: type = ICMP_UNREACH;
1529: code = ICMP_UNREACH_NET_PROHIB;
1530: goto bad;
1531: }
1.114 itojun 1532: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1533: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1534: goto bad;
1535: }
1.1 cgd 1536: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1537: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1538: goto bad;
1539: }
1540: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 1541: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 1542: if (ia == 0) {
1543: if (opt == IPOPT_SSRR) {
1544: type = ICMP_UNREACH;
1545: code = ICMP_UNREACH_SRCFAIL;
1546: goto bad;
1547: }
1548: /*
1549: * Loose routing, and not at next destination
1550: * yet; nothing to do except forward.
1551: */
1552: break;
1553: }
1554: off--; /* 0 origin */
1.112 sommerfe 1555: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 1556: /*
1557: * End of source route. Should be for us.
1558: */
1559: save_rte(cp, ip->ip_src);
1560: break;
1561: }
1562: /*
1563: * locate outgoing interface
1564: */
1.281 tsutsui 1565: memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off),
1.1 cgd 1566: sizeof(ipaddr.sin_addr));
1.96 thorpej 1567: if (opt == IPOPT_SSRR)
1.196 itojun 1568: ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr)));
1.96 thorpej 1569: else
1.1 cgd 1570: ia = ip_rtaddr(ipaddr.sin_addr);
1571: if (ia == 0) {
1572: type = ICMP_UNREACH;
1573: code = ICMP_UNREACH_SRCFAIL;
1574: goto bad;
1575: }
1576: ip->ip_dst = ipaddr.sin_addr;
1.244 christos 1577: bcopy((void *)&ia->ia_addr.sin_addr,
1578: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1579: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1580: /*
1581: * Let ip_intr's mcast routing check handle mcast pkts
1582: */
1.18 mycroft 1583: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1584: break;
1585:
1586: case IPOPT_RR:
1.114 itojun 1587: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1588: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1589: goto bad;
1590: }
1.1 cgd 1591: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1592: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1593: goto bad;
1594: }
1595: /*
1596: * If no space remains, ignore.
1597: */
1598: off--; /* 0 origin */
1.112 sommerfe 1599: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1600: break;
1.281 tsutsui 1601: memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst),
1.1 cgd 1602: sizeof(ipaddr.sin_addr));
1603: /*
1604: * locate outgoing interface; if we're the destination,
1605: * use the incoming interface (should be same).
1606: */
1.96 thorpej 1607: if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))))
1608: == NULL &&
1609: (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
1.1 cgd 1610: type = ICMP_UNREACH;
1611: code = ICMP_UNREACH_HOST;
1612: goto bad;
1613: }
1.244 christos 1614: bcopy((void *)&ia->ia_addr.sin_addr,
1615: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1616: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1617: break;
1618:
1619: case IPOPT_TS:
1620: code = cp - (u_char *)ip;
1621: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1622: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1623: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1624: goto bad;
1.114 itojun 1625: }
1626: if (ipt->ipt_ptr < 5) {
1627: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1628: goto bad;
1629: }
1.15 cgd 1630: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1631: if (++ipt->ipt_oflw == 0) {
1632: code = (u_char *)&ipt->ipt_ptr -
1633: (u_char *)ip;
1.1 cgd 1634: goto bad;
1.114 itojun 1635: }
1.1 cgd 1636: break;
1637: }
1.104 thorpej 1638: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1639: switch (ipt->ipt_flg) {
1640:
1641: case IPOPT_TS_TSONLY:
1642: break;
1643:
1644: case IPOPT_TS_TSANDADDR:
1.66 thorpej 1645: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1646: sizeof(struct in_addr) > ipt->ipt_len) {
1647: code = (u_char *)&ipt->ipt_ptr -
1648: (u_char *)ip;
1.1 cgd 1649: goto bad;
1.114 itojun 1650: }
1.13 mycroft 1651: ipaddr.sin_addr = dst;
1.96 thorpej 1652: ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr),
1653: m->m_pkthdr.rcvif));
1.13 mycroft 1654: if (ia == 0)
1655: continue;
1.104 thorpej 1656: bcopy(&ia->ia_addr.sin_addr,
1657: cp0, sizeof(struct in_addr));
1.1 cgd 1658: ipt->ipt_ptr += sizeof(struct in_addr);
1659: break;
1660:
1661: case IPOPT_TS_PRESPEC:
1.66 thorpej 1662: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1663: sizeof(struct in_addr) > ipt->ipt_len) {
1664: code = (u_char *)&ipt->ipt_ptr -
1665: (u_char *)ip;
1.1 cgd 1666: goto bad;
1.114 itojun 1667: }
1.281 tsutsui 1668: memcpy(&ipaddr.sin_addr, cp0,
1.1 cgd 1669: sizeof(struct in_addr));
1.96 thorpej 1670: if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))
1671: == NULL)
1.1 cgd 1672: continue;
1673: ipt->ipt_ptr += sizeof(struct in_addr);
1674: break;
1675:
1676: default:
1.114 itojun 1677: /* XXX can't take &ipt->ipt_flg */
1678: code = (u_char *)&ipt->ipt_ptr -
1679: (u_char *)ip + 1;
1.1 cgd 1680: goto bad;
1681: }
1682: ntime = iptime();
1.107 thorpej 1683: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1.244 christos 1684: memmove((char *)cp + ipt->ipt_ptr - 1, cp0,
1.1 cgd 1685: sizeof(n_time));
1686: ipt->ipt_ptr += sizeof(n_time);
1687: }
1688: }
1689: if (forward) {
1.26 thorpej 1690: if (ip_forwsrcrt == 0) {
1691: type = ICMP_UNREACH;
1692: code = ICMP_UNREACH_SRCFAIL;
1693: goto bad;
1694: }
1.1 cgd 1695: ip_forward(m, 1);
1696: return (1);
1.13 mycroft 1697: }
1698: return (0);
1.1 cgd 1699: bad:
1.13 mycroft 1700: icmp_error(m, type, code, 0, 0);
1.266 thorpej 1701: IP_STATINC(IP_STAT_BADOPTIONS);
1.1 cgd 1702: return (1);
1703: }
1704:
1705: /*
1706: * Given address of next destination (final or next hop),
1707: * return internet address info of interface to be used to get there.
1708: */
1709: struct in_ifaddr *
1.211 perry 1710: ip_rtaddr(struct in_addr dst)
1.1 cgd 1711: {
1.249 dyoung 1712: struct rtentry *rt;
1713: union {
1714: struct sockaddr dst;
1715: struct sockaddr_in dst4;
1716: } u;
1717:
1718: sockaddr_in_init(&u.dst4, &dst, 0);
1719:
1720: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL)
1721: return NULL;
1722:
1723: return ifatoia(rt->rt_ifa);
1.1 cgd 1724: }
1725:
1726: /*
1727: * Save incoming source route for use in replies,
1728: * to be picked up later by ip_srcroute if the receiver is interested.
1729: */
1.13 mycroft 1730: void
1.211 perry 1731: save_rte(u_char *option, struct in_addr dst)
1.1 cgd 1732: {
1733: unsigned olen;
1734:
1735: olen = option[IPOPT_OLEN];
1736: #ifdef DIAGNOSTIC
1737: if (ipprintfs)
1.39 christos 1738: printf("save_rte: olen %d\n", olen);
1.89 itojun 1739: #endif /* 0 */
1.1 cgd 1740: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1741: return;
1.281 tsutsui 1742: memcpy((void *)ip_srcrt.srcopt, (void *)option, olen);
1.1 cgd 1743: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1744: ip_srcrt.dst = dst;
1745: }
1746:
1747: /*
1748: * Retrieve incoming source route for use in replies,
1749: * in the same form used by setsockopt.
1750: * The first hop is placed before the options, will be removed later.
1751: */
1752: struct mbuf *
1.211 perry 1753: ip_srcroute(void)
1.1 cgd 1754: {
1.109 augustss 1755: struct in_addr *p, *q;
1756: struct mbuf *m;
1.1 cgd 1757:
1758: if (ip_nhops == 0)
1.237 dyoung 1759: return NULL;
1.1 cgd 1760: m = m_get(M_DONTWAIT, MT_SOOPTS);
1761: if (m == 0)
1.237 dyoung 1762: return NULL;
1.1 cgd 1763:
1.164 matt 1764: MCLAIM(m, &inetdomain.dom_mowner);
1.13 mycroft 1765: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1766:
1767: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1768: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1769: OPTSIZ;
1770: #ifdef DIAGNOSTIC
1771: if (ipprintfs)
1.39 christos 1772: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1773: #endif
1774:
1775: /*
1776: * First save first hop for return route
1777: */
1778: p = &ip_srcrt.route[ip_nhops - 1];
1779: *(mtod(m, struct in_addr *)) = *p--;
1780: #ifdef DIAGNOSTIC
1781: if (ipprintfs)
1.39 christos 1782: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1783: #endif
1784:
1785: /*
1786: * Copy option fields and padding (nop) to mbuf.
1787: */
1788: ip_srcrt.nop = IPOPT_NOP;
1789: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1.244 christos 1790: memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop,
1791: OPTSIZ);
1792: q = (struct in_addr *)(mtod(m, char *) +
1.1 cgd 1793: sizeof(struct in_addr) + OPTSIZ);
1794: #undef OPTSIZ
1795: /*
1796: * Record return path as an IP source route,
1797: * reversing the path (pointers are now aligned).
1798: */
1799: while (p >= ip_srcrt.route) {
1800: #ifdef DIAGNOSTIC
1801: if (ipprintfs)
1.39 christos 1802: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1803: #endif
1804: *q++ = *p--;
1805: }
1806: /*
1807: * Last hop goes to final destination.
1808: */
1809: *q = ip_srcrt.dst;
1810: #ifdef DIAGNOSTIC
1811: if (ipprintfs)
1.39 christos 1812: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1813: #endif
1814: return (m);
1815: }
1816:
1.139 matt 1817: const int inetctlerrmap[PRC_NCMDS] = {
1.256 yamt 1818: [PRC_MSGSIZE] = EMSGSIZE,
1819: [PRC_HOSTDEAD] = EHOSTDOWN,
1820: [PRC_HOSTUNREACH] = EHOSTUNREACH,
1821: [PRC_UNREACH_NET] = EHOSTUNREACH,
1822: [PRC_UNREACH_HOST] = EHOSTUNREACH,
1823: [PRC_UNREACH_PROTOCOL] = ECONNREFUSED,
1824: [PRC_UNREACH_PORT] = ECONNREFUSED,
1825: [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH,
1826: [PRC_PARAMPROB] = ENOPROTOOPT,
1.1 cgd 1827: };
1828:
1829: /*
1830: * Forward a packet. If some error occurs return the sender
1831: * an icmp packet. Note we can't always generate a meaningful
1832: * icmp message because icmp doesn't have a large enough repertoire
1833: * of codes and types.
1834: *
1835: * If not forwarding, just drop the packet. This could be confusing
1836: * if ipforwarding was zero but some routing protocol was advancing
1837: * us as a gateway to somewhere. However, we must let the routing
1838: * protocol deal with that.
1839: *
1840: * The srcrt parameter indicates whether the packet is being forwarded
1841: * via a source route.
1842: */
1.13 mycroft 1843: void
1.211 perry 1844: ip_forward(struct mbuf *m, int srcrt)
1.1 cgd 1845: {
1.109 augustss 1846: struct ip *ip = mtod(m, struct ip *);
1847: struct rtentry *rt;
1.220 christos 1848: int error, type = 0, code = 0, destmtu = 0;
1.1 cgd 1849: struct mbuf *mcopy;
1.13 mycroft 1850: n_long dest;
1.249 dyoung 1851: union {
1852: struct sockaddr dst;
1853: struct sockaddr_in dst4;
1854: } u;
1.164 matt 1855:
1856: /*
1857: * We are now in the output path.
1858: */
1859: MCLAIM(m, &ip_tx_mowner);
1.135 thorpej 1860:
1861: /*
1862: * Clear any in-bound checksum flags for this packet.
1863: */
1864: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1865:
1.13 mycroft 1866: dest = 0;
1.1 cgd 1867: #ifdef DIAGNOSTIC
1.224 joerg 1868: if (ipprintfs) {
1869: printf("forward: src %s ", inet_ntoa(ip->ip_src));
1870: printf("dst %s ttl %x\n", inet_ntoa(ip->ip_dst), ip->ip_ttl);
1871: }
1.1 cgd 1872: #endif
1.93 sommerfe 1873: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.266 thorpej 1874: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 1875: m_freem(m);
1876: return;
1877: }
1878: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1879: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1880: return;
1881: }
1882:
1.249 dyoung 1883: sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
1884: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) {
1885: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0);
1886: return;
1.1 cgd 1887: }
1888:
1889: /*
1.34 mycroft 1890: * Save at most 68 bytes of the packet in case
1.1 cgd 1891: * we need to generate an ICMP message to the src.
1.119 itojun 1892: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1893: */
1.155 itojun 1894: mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1.119 itojun 1895: if (mcopy)
1896: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1897:
1.221 christos 1898: ip->ip_ttl -= IPTTLDEC;
1899:
1.1 cgd 1900: /*
1901: * If forwarding packet using same interface that it came in on,
1902: * perhaps should send a redirect to sender to shortcut a hop.
1903: * Only send redirect if source is sending directly to us,
1904: * and if packet was not source routed (or has any options).
1905: * Also, don't send redirect if forwarding using a default route
1906: * or a route modified by a redirect.
1907: */
1908: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1909: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.250 dyoung 1910: !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) &&
1.1 cgd 1911: ipsendredirects && !srcrt) {
1.19 mycroft 1912: if (rt->rt_ifa &&
1913: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1914: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1915: if (rt->rt_flags & RTF_GATEWAY)
1916: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1917: else
1918: dest = ip->ip_dst.s_addr;
1919: /*
1920: * Router requirements says to only send host
1921: * redirects.
1922: */
1923: type = ICMP_REDIRECT;
1924: code = ICMP_REDIRECT_HOST;
1.1 cgd 1925: #ifdef DIAGNOSTIC
1.77 thorpej 1926: if (ipprintfs)
1927: printf("redirect (%d) to %x\n", code,
1928: (u_int32_t)dest);
1.1 cgd 1929: #endif
1930: }
1931: }
1932:
1.238 dyoung 1933: error = ip_output(m, NULL, &ipforward_rt,
1.173 jonathan 1934: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
1.174 itojun 1935: (struct ip_moptions *)NULL, (struct socket *)NULL);
1.173 jonathan 1936:
1.1 cgd 1937: if (error)
1.266 thorpej 1938: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 1939: else {
1.266 thorpej 1940: uint64_t *ips = IP_STAT_GETREF();
1941: ips[IP_STAT_FORWARD]++;
1942: if (type) {
1943: ips[IP_STAT_REDIRECTSENT]++;
1944: IP_STAT_PUTREF();
1945: } else {
1946: IP_STAT_PUTREF();
1.63 matt 1947: if (mcopy) {
1948: #ifdef GATEWAY
1.64 thorpej 1949: if (mcopy->m_flags & M_CANFASTFWD)
1950: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1951: #endif
1.1 cgd 1952: m_freem(mcopy);
1.63 matt 1953: }
1.1 cgd 1954: return;
1955: }
1956: }
1957: if (mcopy == NULL)
1958: return;
1.13 mycroft 1959:
1.1 cgd 1960: switch (error) {
1961:
1962: case 0: /* forwarded, but need redirect */
1963: /* type, code set above */
1964: break;
1965:
1966: case ENETUNREACH: /* shouldn't happen, checked above */
1967: case EHOSTUNREACH:
1968: case ENETDOWN:
1969: case EHOSTDOWN:
1970: default:
1971: type = ICMP_UNREACH;
1972: code = ICMP_UNREACH_HOST;
1973: break;
1974:
1975: case EMSGSIZE:
1976: type = ICMP_UNREACH;
1977: code = ICMP_UNREACH_NEEDFRAG;
1.263 cube 1978:
1.274 seanb 1979: if ((rt = rtcache_validate(&ipforward_rt)) != NULL)
1980: destmtu = rt->rt_ifp->if_mtu;
1.263 cube 1981:
1982: #if defined(IPSEC) || defined(FAST_IPSEC)
1.274 seanb 1983: {
1.263 cube 1984: /*
1985: * If the packet is routed over IPsec tunnel, tell the
1986: * originator the tunnel MTU.
1987: * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1988: * XXX quickhack!!!
1989: */
1990:
1.89 itojun 1991: struct secpolicy *sp;
1992: int ipsecerror;
1.95 itojun 1993: size_t ipsechdr;
1.89 itojun 1994: struct route *ro;
1995:
1996: sp = ipsec4_getpolicybyaddr(mcopy,
1.170 itojun 1997: IPSEC_DIR_OUTBOUND, IP_FORWARDING,
1998: &ipsecerror);
1.89 itojun 1999:
1.263 cube 2000: if (sp != NULL) {
1.89 itojun 2001: /* count IPsec header size */
1.95 itojun 2002: ipsechdr = ipsec4_hdrsiz(mcopy,
1.170 itojun 2003: IPSEC_DIR_OUTBOUND, NULL);
1.89 itojun 2004:
2005: /*
2006: * find the correct route for outer IPv4
2007: * header, compute tunnel MTU.
2008: */
1.220 christos 2009:
1.89 itojun 2010: if (sp->req != NULL
1.95 itojun 2011: && sp->req->sav != NULL
2012: && sp->req->sav->sah != NULL) {
2013: ro = &sp->req->sav->sah->sa_route;
1.274 seanb 2014: rt = rtcache_validate(ro);
1.257 dyoung 2015: if (rt && rt->rt_ifp) {
1.220 christos 2016: destmtu =
1.257 dyoung 2017: rt->rt_rmx.rmx_mtu ?
2018: rt->rt_rmx.rmx_mtu :
2019: rt->rt_ifp->if_mtu;
1.220 christos 2020: destmtu -= ipsechdr;
1.89 itojun 2021: }
2022: }
2023:
1.173 jonathan 2024: #ifdef IPSEC
1.89 itojun 2025: key_freesp(sp);
1.173 jonathan 2026: #else
2027: KEY_FREESP(&sp);
2028: #endif
1.89 itojun 2029: }
1.274 seanb 2030: }
1.263 cube 2031: #endif /*defined(IPSEC) || defined(FAST_IPSEC)*/
1.266 thorpej 2032: IP_STATINC(IP_STAT_CANTFRAG);
1.1 cgd 2033: break;
2034:
2035: case ENOBUFS:
1.143 itojun 2036: #if 1
2037: /*
2038: * a router should not generate ICMP_SOURCEQUENCH as
2039: * required in RFC1812 Requirements for IP Version 4 Routers.
2040: * source quench could be a big problem under DoS attacks,
1.149 wiz 2041: * or if the underlying interface is rate-limited.
1.143 itojun 2042: */
2043: if (mcopy)
2044: m_freem(mcopy);
2045: return;
2046: #else
1.1 cgd 2047: type = ICMP_SOURCEQUENCH;
2048: code = 0;
2049: break;
1.143 itojun 2050: #endif
1.1 cgd 2051: }
1.220 christos 2052: icmp_error(mcopy, type, code, dest, destmtu);
1.44 thorpej 2053: }
2054:
2055: void
1.211 perry 2056: ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
2057: struct mbuf *m)
1.44 thorpej 2058: {
2059:
1.278 christos 2060: if (inp->inp_socket->so_options & SO_TIMESTAMP
2061: #ifdef SO_OTIMESTAMP
2062: || inp->inp_socket->so_options & SO_OTIMESTAMP
2063: #endif
2064: ) {
1.44 thorpej 2065: struct timeval tv;
2066:
2067: microtime(&tv);
1.278 christos 2068: #ifdef SO_OTIMESTAMP
2069: if (inp->inp_socket->so_options & SO_OTIMESTAMP) {
2070: struct timeval50 tv50;
2071: timeval_to_timeval50(&tv, &tv50);
2072: *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50),
2073: SCM_OTIMESTAMP, SOL_SOCKET);
2074: } else
2075: #endif
1.244 christos 2076: *mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1.44 thorpej 2077: SCM_TIMESTAMP, SOL_SOCKET);
2078: if (*mp)
2079: mp = &(*mp)->m_next;
2080: }
2081: if (inp->inp_flags & INP_RECVDSTADDR) {
1.244 christos 2082: *mp = sbcreatecontrol((void *) &ip->ip_dst,
1.44 thorpej 2083: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
2084: if (*mp)
2085: mp = &(*mp)->m_next;
2086: }
2087: #ifdef notyet
2088: /*
2089: * XXX
2090: * Moving these out of udp_input() made them even more broken
2091: * than they already were.
2092: * - fenner@parc.xerox.com
2093: */
2094: /* options were tossed already */
2095: if (inp->inp_flags & INP_RECVOPTS) {
1.244 christos 2096: *mp = sbcreatecontrol((void *) opts_deleted_above,
1.44 thorpej 2097: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
2098: if (*mp)
2099: mp = &(*mp)->m_next;
2100: }
2101: /* ip_srcroute doesn't do what we want here, need to fix */
2102: if (inp->inp_flags & INP_RECVRETOPTS) {
1.244 christos 2103: *mp = sbcreatecontrol((void *) ip_srcroute(),
1.44 thorpej 2104: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
2105: if (*mp)
2106: mp = &(*mp)->m_next;
2107: }
2108: #endif
2109: if (inp->inp_flags & INP_RECVIF) {
2110: struct sockaddr_dl sdl;
2111:
1.252 dyoung 2112: sockaddr_dl_init(&sdl, sizeof(sdl),
2113: (m->m_pkthdr.rcvif != NULL)
2114: ? m->m_pkthdr.rcvif->if_index
2115: : 0,
2116: 0, NULL, 0, NULL, 0);
1.251 dyoung 2117: *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP);
1.44 thorpej 2118: if (*mp)
2119: mp = &(*mp)->m_next;
2120: }
1.282 ! minskim 2121: if (inp->inp_flags & INP_RECVTTL) {
! 2122: *mp = sbcreatecontrol((void *) &ip->ip_ttl,
! 2123: sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP);
! 2124: if (*mp)
! 2125: mp = &(*mp)->m_next;
! 2126: }
1.13 mycroft 2127: }
2128:
1.189 atatat 2129: /*
1.228 elad 2130: * sysctl helper routine for net.inet.ip.forwsrcrt.
2131: */
2132: static int
2133: sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS)
2134: {
2135: int error, tmp;
2136: struct sysctlnode node;
2137:
2138: node = *rnode;
2139: tmp = ip_forwsrcrt;
2140: node.sysctl_data = &tmp;
2141: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2142: if (error || newp == NULL)
2143: return (error);
2144:
1.280 elad 2145: error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT,
2146: 0, NULL, NULL, NULL);
2147: if (error)
2148: return (error);
1.228 elad 2149:
2150: ip_forwsrcrt = tmp;
2151:
2152: return (0);
2153: }
2154:
2155: /*
1.189 atatat 2156: * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the
2157: * range of the new value and tweaks timers if it changes.
2158: */
2159: static int
2160: sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS)
1.13 mycroft 2161: {
1.189 atatat 2162: int error, tmp;
2163: struct sysctlnode node;
2164:
2165: node = *rnode;
2166: tmp = ip_mtudisc_timeout;
2167: node.sysctl_data = &tmp;
2168: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2169: if (error || newp == NULL)
2170: return (error);
2171: if (tmp < 0)
2172: return (EINVAL);
1.52 thorpej 2173:
1.273 matt 2174: mutex_enter(softnet_lock);
2175:
1.189 atatat 2176: ip_mtudisc_timeout = tmp;
2177: rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout);
2178:
1.273 matt 2179: mutex_exit(softnet_lock);
2180:
1.189 atatat 2181: return (0);
2182: }
1.54 lukem 2183:
1.65 matt 2184: #ifdef GATEWAY
1.189 atatat 2185: /*
1.247 liamjfoy 2186: * sysctl helper routine for net.inet.ip.maxflows.
1.189 atatat 2187: */
2188: static int
2189: sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS)
2190: {
1.273 matt 2191: int error;
1.67 thorpej 2192:
1.273 matt 2193: error = sysctl_lookup(SYSCTLFN_CALL(rnode));
2194: if (error || newp == NULL)
2195: return (error);
2196:
2197: mutex_enter(softnet_lock);
2198: KERNEL_LOCK(1, NULL);
1.212 perry 2199:
1.265 thorpej 2200: ipflow_prune();
1.273 matt 2201:
2202: KERNEL_UNLOCK_ONE(NULL);
2203: mutex_exit(softnet_lock);
1.144 martin 2204:
1.189 atatat 2205: return (0);
2206: }
1.248 liamjfoy 2207:
2208: static int
2209: sysctl_net_inet_ip_hashsize(SYSCTLFN_ARGS)
2210: {
2211: int error, tmp;
2212: struct sysctlnode node;
2213:
2214: node = *rnode;
2215: tmp = ip_hashsize;
2216: node.sysctl_data = &tmp;
2217: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2218: if (error || newp == NULL)
2219: return (error);
2220:
2221: if ((tmp & (tmp - 1)) == 0 && tmp != 0) {
2222: /*
2223: * Can only fail due to malloc()
2224: */
1.273 matt 2225: mutex_enter(softnet_lock);
2226: KERNEL_LOCK(1, NULL);
2227:
2228: error = ipflow_invalidate_all(tmp);
2229:
2230: KERNEL_UNLOCK_ONE(NULL);
2231: mutex_exit(softnet_lock);
2232:
1.248 liamjfoy 2233: } else {
2234: /*
2235: * EINVAL if not a power of 2
2236: */
1.273 matt 2237: error = EINVAL;
1.248 liamjfoy 2238: }
2239:
1.273 matt 2240: return error;
1.248 liamjfoy 2241: }
1.189 atatat 2242: #endif /* GATEWAY */
1.117 tron 2243:
1.266 thorpej 2244: static int
2245: sysctl_net_inet_ip_stats(SYSCTLFN_ARGS)
2246: {
2247:
1.271 thorpej 2248: return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS));
1.266 thorpej 2249: }
1.131 itojun 2250:
1.189 atatat 2251: SYSCTL_SETUP(sysctl_net_inet_ip_setup, "sysctl net.inet.ip subtree setup")
2252: {
2253: extern int subnetsarelocal, hostzeroisbroadcast;
1.180 jonathan 2254:
1.197 atatat 2255: sysctl_createv(clog, 0, NULL, NULL,
2256: CTLFLAG_PERMANENT,
1.189 atatat 2257: CTLTYPE_NODE, "net", NULL,
2258: NULL, 0, NULL, 0,
2259: CTL_NET, CTL_EOL);
1.197 atatat 2260: sysctl_createv(clog, 0, NULL, NULL,
2261: CTLFLAG_PERMANENT,
1.203 atatat 2262: CTLTYPE_NODE, "inet",
2263: SYSCTL_DESCR("PF_INET related settings"),
1.189 atatat 2264: NULL, 0, NULL, 0,
2265: CTL_NET, PF_INET, CTL_EOL);
1.197 atatat 2266: sysctl_createv(clog, 0, NULL, NULL,
2267: CTLFLAG_PERMANENT,
1.203 atatat 2268: CTLTYPE_NODE, "ip",
2269: SYSCTL_DESCR("IPv4 related settings"),
1.189 atatat 2270: NULL, 0, NULL, 0,
2271: CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL);
1.212 perry 2272:
1.197 atatat 2273: sysctl_createv(clog, 0, NULL, NULL,
2274: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2275: CTLTYPE_INT, "forwarding",
2276: SYSCTL_DESCR("Enable forwarding of INET datagrams"),
1.189 atatat 2277: NULL, 0, &ipforwarding, 0,
2278: CTL_NET, PF_INET, IPPROTO_IP,
2279: IPCTL_FORWARDING, CTL_EOL);
1.197 atatat 2280: sysctl_createv(clog, 0, NULL, NULL,
2281: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2282: CTLTYPE_INT, "redirect",
2283: SYSCTL_DESCR("Enable sending of ICMP redirect messages"),
1.189 atatat 2284: NULL, 0, &ipsendredirects, 0,
2285: CTL_NET, PF_INET, IPPROTO_IP,
2286: IPCTL_SENDREDIRECTS, CTL_EOL);
1.197 atatat 2287: sysctl_createv(clog, 0, NULL, NULL,
2288: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2289: CTLTYPE_INT, "ttl",
2290: SYSCTL_DESCR("Default TTL for an INET datagram"),
1.189 atatat 2291: NULL, 0, &ip_defttl, 0,
2292: CTL_NET, PF_INET, IPPROTO_IP,
2293: IPCTL_DEFTTL, CTL_EOL);
2294: #ifdef IPCTL_DEFMTU
1.197 atatat 2295: sysctl_createv(clog, 0, NULL, NULL,
2296: CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */,
1.203 atatat 2297: CTLTYPE_INT, "mtu",
2298: SYSCTL_DESCR("Default MTA for an INET route"),
1.189 atatat 2299: NULL, 0, &ip_mtu, 0,
2300: CTL_NET, PF_INET, IPPROTO_IP,
2301: IPCTL_DEFMTU, CTL_EOL);
2302: #endif /* IPCTL_DEFMTU */
1.197 atatat 2303: sysctl_createv(clog, 0, NULL, NULL,
1.228 elad 2304: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2305: CTLTYPE_INT, "forwsrcrt",
2306: SYSCTL_DESCR("Enable forwarding of source-routed "
2307: "datagrams"),
1.228 elad 2308: sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0,
1.189 atatat 2309: CTL_NET, PF_INET, IPPROTO_IP,
2310: IPCTL_FORWSRCRT, CTL_EOL);
1.197 atatat 2311: sysctl_createv(clog, 0, NULL, NULL,
2312: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2313: CTLTYPE_INT, "directed-broadcast",
2314: SYSCTL_DESCR("Enable forwarding of broadcast datagrams"),
1.189 atatat 2315: NULL, 0, &ip_directedbcast, 0,
2316: CTL_NET, PF_INET, IPPROTO_IP,
2317: IPCTL_DIRECTEDBCAST, CTL_EOL);
1.197 atatat 2318: sysctl_createv(clog, 0, NULL, NULL,
2319: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2320: CTLTYPE_INT, "allowsrcrt",
2321: SYSCTL_DESCR("Accept source-routed datagrams"),
1.189 atatat 2322: NULL, 0, &ip_allowsrcrt, 0,
2323: CTL_NET, PF_INET, IPPROTO_IP,
2324: IPCTL_ALLOWSRCRT, CTL_EOL);
1.197 atatat 2325: sysctl_createv(clog, 0, NULL, NULL,
2326: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2327: CTLTYPE_INT, "subnetsarelocal",
2328: SYSCTL_DESCR("Whether logical subnets are considered "
2329: "local"),
1.189 atatat 2330: NULL, 0, &subnetsarelocal, 0,
2331: CTL_NET, PF_INET, IPPROTO_IP,
2332: IPCTL_SUBNETSARELOCAL, CTL_EOL);
1.197 atatat 2333: sysctl_createv(clog, 0, NULL, NULL,
2334: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2335: CTLTYPE_INT, "mtudisc",
2336: SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"),
1.189 atatat 2337: NULL, 0, &ip_mtudisc, 0,
2338: CTL_NET, PF_INET, IPPROTO_IP,
2339: IPCTL_MTUDISC, CTL_EOL);
1.197 atatat 2340: sysctl_createv(clog, 0, NULL, NULL,
2341: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2342: CTLTYPE_INT, "anonportmin",
2343: SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1.189 atatat 2344: sysctl_net_inet_ip_ports, 0, &anonportmin, 0,
2345: CTL_NET, PF_INET, IPPROTO_IP,
2346: IPCTL_ANONPORTMIN, CTL_EOL);
1.197 atatat 2347: sysctl_createv(clog, 0, NULL, NULL,
2348: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2349: CTLTYPE_INT, "anonportmax",
2350: SYSCTL_DESCR("Highest ephemeral port number to assign"),
1.189 atatat 2351: sysctl_net_inet_ip_ports, 0, &anonportmax, 0,
2352: CTL_NET, PF_INET, IPPROTO_IP,
2353: IPCTL_ANONPORTMAX, CTL_EOL);
1.197 atatat 2354: sysctl_createv(clog, 0, NULL, NULL,
2355: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2356: CTLTYPE_INT, "mtudisctimeout",
2357: SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"),
1.189 atatat 2358: sysctl_net_inet_ip_pmtudto, 0, &ip_mtudisc_timeout, 0,
2359: CTL_NET, PF_INET, IPPROTO_IP,
2360: IPCTL_MTUDISCTIMEOUT, CTL_EOL);
2361: #ifdef GATEWAY
1.197 atatat 2362: sysctl_createv(clog, 0, NULL, NULL,
2363: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2364: CTLTYPE_INT, "maxflows",
2365: SYSCTL_DESCR("Number of flows for fast forwarding"),
1.189 atatat 2366: sysctl_net_inet_ip_maxflows, 0, &ip_maxflows, 0,
2367: CTL_NET, PF_INET, IPPROTO_IP,
2368: IPCTL_MAXFLOWS, CTL_EOL);
1.248 liamjfoy 2369: sysctl_createv(clog, 0, NULL, NULL,
2370: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2371: CTLTYPE_INT, "hashsize",
2372: SYSCTL_DESCR("Size of hash table for fast forwarding (IPv4)"),
2373: sysctl_net_inet_ip_hashsize, 0, &ip_hashsize, 0,
2374: CTL_NET, PF_INET, IPPROTO_IP,
2375: CTL_CREATE, CTL_EOL);
1.189 atatat 2376: #endif /* GATEWAY */
1.197 atatat 2377: sysctl_createv(clog, 0, NULL, NULL,
2378: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2379: CTLTYPE_INT, "hostzerobroadcast",
2380: SYSCTL_DESCR("All zeroes address is broadcast address"),
1.189 atatat 2381: NULL, 0, &hostzeroisbroadcast, 0,
2382: CTL_NET, PF_INET, IPPROTO_IP,
2383: IPCTL_HOSTZEROBROADCAST, CTL_EOL);
2384: #if NGIF > 0
1.197 atatat 2385: sysctl_createv(clog, 0, NULL, NULL,
2386: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2387: CTLTYPE_INT, "gifttl",
2388: SYSCTL_DESCR("Default TTL for a gif tunnel datagram"),
1.189 atatat 2389: NULL, 0, &ip_gif_ttl, 0,
2390: CTL_NET, PF_INET, IPPROTO_IP,
2391: IPCTL_GIF_TTL, CTL_EOL);
2392: #endif /* NGIF */
2393: #ifndef IPNOPRIVPORTS
1.197 atatat 2394: sysctl_createv(clog, 0, NULL, NULL,
2395: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2396: CTLTYPE_INT, "lowportmin",
2397: SYSCTL_DESCR("Lowest privileged ephemeral port number "
2398: "to assign"),
1.189 atatat 2399: sysctl_net_inet_ip_ports, 0, &lowportmin, 0,
2400: CTL_NET, PF_INET, IPPROTO_IP,
2401: IPCTL_LOWPORTMIN, CTL_EOL);
1.197 atatat 2402: sysctl_createv(clog, 0, NULL, NULL,
2403: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2404: CTLTYPE_INT, "lowportmax",
2405: SYSCTL_DESCR("Highest privileged ephemeral port number "
2406: "to assign"),
1.189 atatat 2407: sysctl_net_inet_ip_ports, 0, &lowportmax, 0,
2408: CTL_NET, PF_INET, IPPROTO_IP,
2409: IPCTL_LOWPORTMAX, CTL_EOL);
2410: #endif /* IPNOPRIVPORTS */
1.197 atatat 2411: sysctl_createv(clog, 0, NULL, NULL,
2412: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2413: CTLTYPE_INT, "maxfragpackets",
2414: SYSCTL_DESCR("Maximum number of fragments to retain for "
2415: "possible reassembly"),
1.189 atatat 2416: NULL, 0, &ip_maxfragpackets, 0,
2417: CTL_NET, PF_INET, IPPROTO_IP,
2418: IPCTL_MAXFRAGPACKETS, CTL_EOL);
2419: #if NGRE > 0
1.197 atatat 2420: sysctl_createv(clog, 0, NULL, NULL,
2421: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2422: CTLTYPE_INT, "grettl",
2423: SYSCTL_DESCR("Default TTL for a gre tunnel datagram"),
1.189 atatat 2424: NULL, 0, &ip_gre_ttl, 0,
2425: CTL_NET, PF_INET, IPPROTO_IP,
2426: IPCTL_GRE_TTL, CTL_EOL);
2427: #endif /* NGRE */
1.197 atatat 2428: sysctl_createv(clog, 0, NULL, NULL,
2429: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2430: CTLTYPE_INT, "checkinterface",
2431: SYSCTL_DESCR("Enable receive side of Strong ES model "
2432: "from RFC1122"),
1.189 atatat 2433: NULL, 0, &ip_checkinterface, 0,
2434: CTL_NET, PF_INET, IPPROTO_IP,
2435: IPCTL_CHECKINTERFACE, CTL_EOL);
1.197 atatat 2436: sysctl_createv(clog, 0, NULL, NULL,
2437: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2438: CTLTYPE_INT, "random_id",
2439: SYSCTL_DESCR("Assign random ip_id values"),
1.189 atatat 2440: NULL, 0, &ip_do_randomid, 0,
2441: CTL_NET, PF_INET, IPPROTO_IP,
2442: IPCTL_RANDOMID, CTL_EOL);
1.206 thorpej 2443: sysctl_createv(clog, 0, NULL, NULL,
2444: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2445: CTLTYPE_INT, "do_loopback_cksum",
2446: SYSCTL_DESCR("Perform IP checksum on loopback"),
2447: NULL, 0, &ip_do_loopback_cksum, 0,
2448: CTL_NET, PF_INET, IPPROTO_IP,
2449: IPCTL_LOOPBACKCKSUM, CTL_EOL);
1.219 elad 2450: sysctl_createv(clog, 0, NULL, NULL,
2451: CTLFLAG_PERMANENT,
2452: CTLTYPE_STRUCT, "stats",
2453: SYSCTL_DESCR("IP statistics"),
1.266 thorpej 2454: sysctl_net_inet_ip_stats, 0, NULL, 0,
1.219 elad 2455: CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS,
2456: CTL_EOL);
1.1 cgd 2457: }
1.266 thorpej 2458:
2459: void
2460: ip_statinc(u_int stat)
2461: {
2462:
2463: KASSERT(stat < IP_NSTATS);
2464: IP_STATINC(stat);
2465: }
CVSweb <webmaster@jp.NetBSD.org>