Annotation of src/sys/netinet/ip_input.c, Revision 1.275.4.1.8.1
1.275.4.1.8.1! (matt 1:: /* $NetBSD: ip_input.c,v 1.275.4.1 2008/11/25 04:04:38 snj 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>
1.275.4.1.8.1! (matt 94:: __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.275.4.1 2008/11/25 04:04:38 snj Exp $");
1.55 scottr 95:
1.184 jonathan 96: #include "opt_inet.h"
1.62 matt 97: #include "opt_gateway.h"
1.69 mrg 98: #include "opt_pfil_hooks.h"
1.91 thorpej 99: #include "opt_ipsec.h"
1.55 scottr 100: #include "opt_mrouting.h"
1.167 martin 101: #include "opt_mbuftrace.h"
1.135 thorpej 102: #include "opt_inet_csum.h"
1.1 cgd 103:
1.5 mycroft 104: #include <sys/param.h>
105: #include <sys/systm.h>
106: #include <sys/malloc.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:
1.27 thorpej 176: /*
177: * Note: DIRECTED_BROADCAST is handled this way so that previous
178: * configuration using this option will Just Work.
179: */
180: #ifndef IPDIRECTEDBCAST
181: #ifdef DIRECTED_BROADCAST
182: #define IPDIRECTEDBCAST 1
183: #else
184: #define IPDIRECTEDBCAST 0
185: #endif /* DIRECTED_BROADCAST */
186: #endif /* IPDIRECTEDBCAST */
1.1 cgd 187: int ipforwarding = IPFORWARDING;
188: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 189: int ip_defttl = IPDEFTTL;
1.26 thorpej 190: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 191: int ip_directedbcast = IPDIRECTEDBCAST;
1.47 cjs 192: int ip_allowsrcrt = IPALLOWSRCRT;
1.53 kml 193: int ip_mtudisc = IPMTUDISC;
1.156 itojun 194: int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
1.1 cgd 195: #ifdef DIAGNOSTIC
196: int ipprintfs = 0;
197: #endif
1.184 jonathan 198:
199: int ip_do_randomid = 0;
200:
1.165 christos 201: /*
202: * XXX - Setting ip_checkinterface mostly implements the receive side of
203: * the Strong ES model described in RFC 1122, but since the routing table
204: * and transmit implementation do not implement the Strong ES model,
205: * setting this to 1 results in an odd hybrid.
206: *
207: * XXX - ip_checkinterface currently must be disabled if you use ipnat
208: * to translate the destination address to another local interface.
209: *
210: * XXX - ip_checkinterface must be disabled if you add IP aliases
211: * to the loopback interface instead of the interface where the
212: * packets for those addresses are received.
213: */
214: int ip_checkinterface = 0;
215:
1.1 cgd 216:
1.60 kml 217: struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
218:
1.1 cgd 219: int ipqmaxlen = IFQ_MAXLEN;
1.150 matt 220: u_long in_ifaddrhash; /* size of hash table - 1 */
221: int in_ifaddrentries; /* total number of addrs */
1.212 perry 222: struct in_ifaddrhead in_ifaddrhead;
1.57 tls 223: struct in_ifaddrhashhead *in_ifaddrhashtbl;
1.166 matt 224: u_long in_multihash; /* size of hash table - 1 */
225: int in_multientries; /* total number of addrs */
226: struct in_multihashhead *in_multihashtbl;
1.13 mycroft 227: struct ifqueue ipintrq;
1.183 jonathan 228: uint16_t ip_id;
1.75 thorpej 229:
1.266 thorpej 230: percpu_t *ipstat_percpu;
231:
1.121 thorpej 232: #ifdef PFIL_HOOKS
233: struct pfil_head inet_pfil_hook;
234: #endif
235:
1.194 jonathan 236: /*
237: * Cached copy of nmbclusters. If nbclusters is different,
238: * recalculate IP parameters derived from nmbclusters.
239: */
240: static int ip_nmbclusters; /* copy of nmbclusters */
1.210 perry 241: static void ip_nmbclusters_changed(void); /* recalc limits */
1.194 jonathan 242:
1.195 thorpej 243: #define CHECK_NMBCLUSTER_PARAMS() \
244: do { \
245: if (__predict_false(ip_nmbclusters != nmbclusters)) \
246: ip_nmbclusters_changed(); \
247: } while (/*CONSTCOND*/0)
1.194 jonathan 248:
1.190 jonathan 249: /* IP datagram reassembly queues (hashed) */
250: #define IPREASS_NHASH_LOG2 6
251: #define IPREASS_NHASH (1 << IPREASS_NHASH_LOG2)
252: #define IPREASS_HMASK (IPREASS_NHASH - 1)
253: #define IPREASS_HASH(x,y) \
254: (((((x) & 0xF) | ((((x) >> 8) & 0xF) << 4)) ^ (y)) & IPREASS_HMASK)
255: struct ipqhead ipq[IPREASS_NHASH];
1.75 thorpej 256: int ipq_locked;
1.212 perry 257: static int ip_nfragpackets; /* packets in reass queue */
1.194 jonathan 258: static int ip_nfrags; /* total fragments in reass queues */
259:
260: int ip_maxfragpackets = 200; /* limit on packets. XXX sysctl */
261: int ip_maxfrags; /* limit on fragments. XXX sysctl */
262:
263:
264: /*
265: * Additive-Increase/Multiplicative-Decrease (AIMD) strategy for
266: * IP reassembly queue buffer managment.
1.212 perry 267: *
1.194 jonathan 268: * We keep a count of total IP fragments (NB: not fragmented packets!)
269: * awaiting reassembly (ip_nfrags) and a limit (ip_maxfrags) on fragments.
270: * If ip_nfrags exceeds ip_maxfrags the limit, we drop half the
271: * total fragments in reassembly queues.This AIMD policy avoids
272: * repeatedly deleting single packets under heavy fragmentation load
273: * (e.g., from lossy NFS peers).
274: */
1.212 perry 275: static u_int ip_reass_ttl_decr(u_int ticks);
1.210 perry 276: static void ip_reass_drophalf(void);
1.194 jonathan 277:
1.75 thorpej 278:
1.223 perry 279: static inline int ipq_lock_try(void);
280: static inline void ipq_unlock(void);
1.75 thorpej 281:
1.223 perry 282: static inline int
1.211 perry 283: ipq_lock_try(void)
1.75 thorpej 284: {
285: int s;
286:
1.132 thorpej 287: /*
1.149 wiz 288: * Use splvm() -- we're blocking things that would cause
1.132 thorpej 289: * mbuf allocation.
290: */
291: s = splvm();
1.75 thorpej 292: if (ipq_locked) {
293: splx(s);
294: return (0);
295: }
296: ipq_locked = 1;
297: splx(s);
298: return (1);
299: }
300:
1.223 perry 301: static inline void
1.211 perry 302: ipq_unlock(void)
1.75 thorpej 303: {
304: int s;
305:
1.132 thorpej 306: s = splvm();
1.75 thorpej 307: ipq_locked = 0;
308: splx(s);
309: }
310:
311: #ifdef DIAGNOSTIC
312: #define IPQ_LOCK() \
313: do { \
314: if (ipq_lock_try() == 0) { \
315: printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \
316: panic("ipq_lock"); \
317: } \
1.159 perry 318: } while (/*CONSTCOND*/ 0)
1.75 thorpej 319: #define IPQ_LOCK_CHECK() \
320: do { \
321: if (ipq_locked == 0) { \
322: printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \
323: panic("ipq lock check"); \
324: } \
1.159 perry 325: } while (/*CONSTCOND*/ 0)
1.75 thorpej 326: #else
327: #define IPQ_LOCK() (void) ipq_lock_try()
328: #define IPQ_LOCK_CHECK() /* nothing */
329: #endif
330:
331: #define IPQ_UNLOCK() ipq_unlock()
1.1 cgd 332:
1.275 pooka 333: struct pool inmulti_pool;
334: struct pool ipqent_pool;
1.72 thorpej 335:
1.135 thorpej 336: #ifdef INET_CSUM_COUNTERS
337: #include <sys/device.h>
338:
339: struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
340: NULL, "inet", "hwcsum bad");
341: struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
342: NULL, "inet", "hwcsum ok");
343: struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
344: NULL, "inet", "swcsum");
345:
346: #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++
347:
1.201 matt 348: EVCNT_ATTACH_STATIC(ip_hwcsum_bad);
349: EVCNT_ATTACH_STATIC(ip_hwcsum_ok);
350: EVCNT_ATTACH_STATIC(ip_swcsum);
351:
1.135 thorpej 352: #else
353:
354: #define INET_CSUM_COUNTER_INCR(ev) /* nothing */
355:
356: #endif /* INET_CSUM_COUNTERS */
357:
1.1 cgd 358: /*
359: * We need to save the IP options in case a protocol wants to respond
360: * to an incoming packet over the same route if the packet got here
361: * using IP source routing. This allows connection establishment and
362: * maintenance when the remote end is on a network that is not known
363: * to us.
364: */
365: int ip_nhops = 0;
366: static struct ip_srcrt {
367: struct in_addr dst; /* final destination */
368: char nop; /* one NOP to align */
369: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
370: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
371: } ip_srcrt;
372:
1.210 perry 373: static void save_rte(u_char *, struct in_addr);
1.35 mycroft 374:
1.164 matt 375: #ifdef MBUFTRACE
1.234 dogcow 376: struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx");
377: struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx");
1.164 matt 378: #endif
379:
1.1 cgd 380: /*
1.194 jonathan 381: * Compute IP limits derived from the value of nmbclusters.
382: */
383: static void
384: ip_nmbclusters_changed(void)
385: {
386: ip_maxfrags = nmbclusters / 4;
387: ip_nmbclusters = nmbclusters;
388: }
389:
390: /*
1.1 cgd 391: * IP initialization: fill in IP protocol switch table.
392: * All protocols not implemented in kernel go to raw IP protocol handler.
393: */
1.8 mycroft 394: void
1.211 perry 395: ip_init(void)
1.1 cgd 396: {
1.199 matt 397: const struct protosw *pr;
1.109 augustss 398: int i;
1.1 cgd 399:
1.275 pooka 400: pool_init(&inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl",
401: NULL, IPL_SOFTNET);
402: pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl",
403: NULL, IPL_VM);
404:
1.1 cgd 405: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
406: if (pr == 0)
407: panic("ip_init");
408: for (i = 0; i < IPPROTO_MAX; i++)
409: ip_protox[i] = pr - inetsw;
410: for (pr = inetdomain.dom_protosw;
411: pr < inetdomain.dom_protoswNPROTOSW; pr++)
412: if (pr->pr_domain->dom_family == PF_INET &&
413: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
414: ip_protox[pr->pr_protocol] = pr - inetsw;
1.192 jonathan 415:
1.190 jonathan 416: for (i = 0; i < IPREASS_NHASH; i++)
417: LIST_INIT(&ipq[i]);
418:
1.262 matt 419: ip_initid();
1.227 kardel 420: ip_id = time_second & 0xfffff;
1.194 jonathan 421:
1.1 cgd 422: ipintrq.ifq_maxlen = ipqmaxlen;
1.194 jonathan 423: ip_nmbclusters_changed();
424:
1.181 jonathan 425: TAILQ_INIT(&in_ifaddrhead);
1.272 ad 426: in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true,
427: &in_ifaddrhash);
428: in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true,
429: &in_multihash);
1.160 itojun 430: ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout);
1.73 thorpej 431: #ifdef GATEWAY
1.248 liamjfoy 432: ipflow_init(ip_hashsize);
1.73 thorpej 433: #endif
1.121 thorpej 434:
435: #ifdef PFIL_HOOKS
436: /* Register our Packet Filter hook. */
1.126 thorpej 437: inet_pfil_hook.ph_type = PFIL_TYPE_AF;
438: inet_pfil_hook.ph_af = AF_INET;
1.121 thorpej 439: i = pfil_head_register(&inet_pfil_hook);
440: if (i != 0)
441: printf("ip_init: WARNING: unable to register pfil hook, "
442: "error %d\n", i);
443: #endif /* PFIL_HOOKS */
1.135 thorpej 444:
1.164 matt 445: #ifdef MBUFTRACE
446: MOWNER_ATTACH(&ip_tx_mowner);
447: MOWNER_ATTACH(&ip_rx_mowner);
448: #endif /* MBUFTRACE */
1.266 thorpej 449:
450: ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS);
1.1 cgd 451: }
452:
1.229 christos 453: struct sockaddr_in ipaddr = {
454: .sin_len = sizeof(ipaddr),
455: .sin_family = AF_INET,
456: };
1.1 cgd 457: struct route ipforward_rt;
458:
459: /*
1.89 itojun 460: * IP software interrupt routine
461: */
462: void
1.211 perry 463: ipintr(void)
1.89 itojun 464: {
465: int s;
466: struct mbuf *m;
467:
1.268 ad 468: mutex_enter(softnet_lock);
469: KERNEL_LOCK(1, NULL);
1.241 ad 470: while (!IF_IS_EMPTY(&ipintrq)) {
1.132 thorpej 471: s = splnet();
1.89 itojun 472: IF_DEQUEUE(&ipintrq, m);
473: splx(s);
1.268 ad 474: if (m == NULL)
475: break;
1.275.4.1.8.1! (matt 476:: KERNEL_UNLOCK_ONE(NULL);
1.89 itojun 477: ip_input(m);
1.275.4.1.8.1! (matt 478:: KERNEL_LOCK(1, NULL);
1.89 itojun 479: }
1.268 ad 480: KERNEL_UNLOCK_ONE(NULL);
481: mutex_exit(softnet_lock);
1.89 itojun 482: }
483:
484: /*
1.1 cgd 485: * Ip input routine. Checksum and byte swap header. If fragmented
486: * try to reassemble. Process options. Pass to next level.
487: */
1.8 mycroft 488: void
1.89 itojun 489: ip_input(struct mbuf *m)
1.1 cgd 490: {
1.109 augustss 491: struct ip *ip = NULL;
492: struct ipq *fp;
493: struct in_ifaddr *ia;
494: struct ifaddr *ifa;
1.25 cgd 495: struct ipqent *ipqe;
1.89 itojun 496: int hlen = 0, mff, len;
1.100 itojun 497: int downmatch;
1.165 christos 498: int checkif;
1.169 itojun 499: int srcrt = 0;
1.233 tls 500: int s;
1.190 jonathan 501: u_int hash;
1.173 jonathan 502: #ifdef FAST_IPSEC
503: struct m_tag *mtag;
504: struct tdb_ident *tdbi;
505: struct secpolicy *sp;
1.233 tls 506: int error;
1.173 jonathan 507: #endif /* FAST_IPSEC */
1.1 cgd 508:
1.164 matt 509: MCLAIM(m, &ip_rx_mowner);
1.1 cgd 510: #ifdef DIAGNOSTIC
511: if ((m->m_flags & M_PKTHDR) == 0)
512: panic("ipintr no HDR");
1.89 itojun 513: #endif
1.164 matt 514:
1.1 cgd 515: /*
516: * If no IP addresses have been set yet but the interfaces
517: * are receiving, can't do anything with incoming packets yet.
518: */
1.181 jonathan 519: if (TAILQ_FIRST(&in_ifaddrhead) == 0)
1.1 cgd 520: goto bad;
1.266 thorpej 521: IP_STATINC(IP_STAT_TOTAL);
1.154 thorpej 522: /*
523: * If the IP header is not aligned, slurp it up into a new
524: * mbuf with space for link headers, in the event we forward
525: * it. Otherwise, if it is aligned, make sure the entire
526: * base IP header is in the first mbuf of the chain.
527: */
1.244 christos 528: if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
1.154 thorpej 529: if ((m = m_copyup(m, sizeof(struct ip),
530: (max_linkhdr + 3) & ~3)) == NULL) {
531: /* XXXJRT new stat, please */
1.266 thorpej 532: IP_STATINC(IP_STAT_TOOSMALL);
1.154 thorpej 533: return;
534: }
535: } else if (__predict_false(m->m_len < sizeof (struct ip))) {
536: if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
1.266 thorpej 537: IP_STATINC(IP_STAT_TOOSMALL);
1.154 thorpej 538: return;
539: }
1.1 cgd 540: }
541: ip = mtod(m, struct ip *);
1.13 mycroft 542: if (ip->ip_v != IPVERSION) {
1.266 thorpej 543: IP_STATINC(IP_STAT_BADVERS);
1.13 mycroft 544: goto bad;
545: }
1.1 cgd 546: hlen = ip->ip_hl << 2;
547: if (hlen < sizeof(struct ip)) { /* minimum header length */
1.266 thorpej 548: IP_STATINC(IP_STAT_BADHLEN);
1.1 cgd 549: goto bad;
550: }
551: if (hlen > m->m_len) {
552: if ((m = m_pullup(m, hlen)) == 0) {
1.266 thorpej 553: IP_STATINC(IP_STAT_BADHLEN);
1.89 itojun 554: return;
1.1 cgd 555: }
556: ip = mtod(m, struct ip *);
557: }
1.98 thorpej 558:
1.85 hwr 559: /*
1.99 thorpej 560: * RFC1122: packets with a multicast source address are
1.98 thorpej 561: * not allowed.
1.85 hwr 562: */
563: if (IN_MULTICAST(ip->ip_src.s_addr)) {
1.266 thorpej 564: IP_STATINC(IP_STAT_BADADDR);
1.85 hwr 565: goto bad;
1.129 itojun 566: }
567:
568: /* 127/8 must not appear on wire - RFC1122 */
569: if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
570: (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
1.130 itojun 571: if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
1.266 thorpej 572: IP_STATINC(IP_STAT_BADADDR);
1.129 itojun 573: goto bad;
1.130 itojun 574: }
1.85 hwr 575: }
576:
1.135 thorpej 577: switch (m->m_pkthdr.csum_flags &
1.137 thorpej 578: ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) |
1.135 thorpej 579: M_CSUM_IPv4_BAD)) {
580: case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
581: INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
582: goto badcsum;
583:
584: case M_CSUM_IPv4:
585: /* Checksum was okay. */
586: INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
587: break;
588:
589: default:
1.206 thorpej 590: /*
591: * Must compute it ourselves. Maybe skip checksum on
592: * loopback interfaces.
593: */
594: if (__predict_true(!(m->m_pkthdr.rcvif->if_flags &
595: IFF_LOOPBACK) || ip_do_loopback_cksum)) {
596: INET_CSUM_COUNTER_INCR(&ip_swcsum);
597: if (in_cksum(m, hlen) != 0)
598: goto badcsum;
599: }
1.135 thorpej 600: break;
1.1 cgd 601: }
602:
1.121 thorpej 603: /* Retrieve the packet length. */
604: len = ntohs(ip->ip_len);
1.81 proff 605:
606: /*
607: * Check for additional length bogosity
608: */
1.84 proff 609: if (len < hlen) {
1.266 thorpej 610: IP_STATINC(IP_STAT_BADLEN);
1.81 proff 611: goto bad;
612: }
1.1 cgd 613:
614: /*
615: * Check that the amount of data in the buffers
616: * is as at least much as the IP header would have us expect.
617: * Trim mbufs if longer than we expect.
618: * Drop packet if shorter than we expect.
619: */
1.35 mycroft 620: if (m->m_pkthdr.len < len) {
1.266 thorpej 621: IP_STATINC(IP_STAT_TOOSHORT);
1.1 cgd 622: goto bad;
623: }
1.35 mycroft 624: if (m->m_pkthdr.len > len) {
1.1 cgd 625: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 626: m->m_len = len;
627: m->m_pkthdr.len = len;
1.1 cgd 628: } else
1.35 mycroft 629: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 630: }
631:
1.193 scw 632: #if defined(IPSEC)
1.149 wiz 633: /* ipflow (IP fast forwarding) is not compatible with IPsec. */
1.94 itojun 634: m->m_flags &= ~M_CANFASTFWD;
635: #else
1.64 thorpej 636: /*
637: * Assume that we can create a fast-forward IP flow entry
638: * based on this packet.
639: */
640: m->m_flags |= M_CANFASTFWD;
1.94 itojun 641: #endif
1.64 thorpej 642:
1.36 mrg 643: #ifdef PFIL_HOOKS
1.33 mrg 644: /*
1.64 thorpej 645: * Run through list of hooks for input packets. If there are any
646: * filters which require that additional packets in the flow are
647: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
648: * Note that filters must _never_ set this flag, as another filter
649: * in the list may have previously cleared it.
1.33 mrg 650: */
1.127 itojun 651: /*
652: * let ipfilter look at packet on the wire,
653: * not the decapsulated packet.
654: */
655: #ifdef IPSEC
1.136 itojun 656: if (!ipsec_getnhist(m))
1.186 scw 657: #elif defined(FAST_IPSEC)
658: if (!ipsec_indone(m))
1.127 itojun 659: #else
660: if (1)
661: #endif
662: {
1.169 itojun 663: struct in_addr odst;
664:
665: odst = ip->ip_dst;
1.127 itojun 666: if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif,
1.168 itojun 667: PFIL_IN) != 0)
668: return;
1.127 itojun 669: if (m == NULL)
670: return;
671: ip = mtod(m, struct ip *);
1.142 darrenr 672: hlen = ip->ip_hl << 2;
1.205 darrenr 673: /*
674: * XXX The setting of "srcrt" here is to prevent ip_forward()
675: * from generating ICMP redirects for packets that have
676: * been redirected by a hook back out on to the same LAN that
677: * they came from and is not an indication that the packet
678: * is being inffluenced by source routing options. This
679: * allows things like
680: * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp"
681: * where tlp0 is both on the 1.1.1.0/24 network and is the
682: * default route for hosts on 1.1.1.0/24. Of course this
683: * also requires a "map tlp0 ..." to complete the story.
684: * One might argue whether or not this kind of network config.
1.212 perry 685: * should be supported in this manner...
1.205 darrenr 686: */
1.169 itojun 687: srcrt = (odst.s_addr != ip->ip_dst.s_addr);
1.127 itojun 688: }
1.36 mrg 689: #endif /* PFIL_HOOKS */
1.123 thorpej 690:
691: #ifdef ALTQ
692: /* XXX Temporary until ALTQ is changed to use a pfil hook */
693: if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) {
694: /* packet dropped by traffic conditioner */
695: return;
696: }
697: #endif
1.121 thorpej 698:
699: /*
1.1 cgd 700: * Process options and, if not destined for us,
701: * ship it on. ip_dooptions returns 1 when an
702: * error was detected (causing an icmp message
703: * to be sent and the original packet to be freed).
704: */
705: ip_nhops = 0; /* for source routed packets */
706: if (hlen > sizeof (struct ip) && ip_dooptions(m))
1.89 itojun 707: return;
1.1 cgd 708:
709: /*
1.165 christos 710: * Enable a consistency check between the destination address
711: * and the arrival interface for a unicast packet (the RFC 1122
712: * strong ES model) if IP forwarding is disabled and the packet
713: * is not locally generated.
714: *
715: * XXX - Checking also should be disabled if the destination
716: * address is ipnat'ed to a different interface.
717: *
718: * XXX - Checking is incompatible with IP aliases added
719: * to the loopback interface instead of the interface where
720: * the packets are received.
721: *
722: * XXX - We need to add a per ifaddr flag for this so that
723: * we get finer grain control.
724: */
725: checkif = ip_checkinterface && (ipforwarding == 0) &&
726: (m->m_pkthdr.rcvif != NULL) &&
727: ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0);
728:
729: /*
1.1 cgd 730: * Check our list of addresses, to see if the packet is for us.
1.100 itojun 731: *
732: * Traditional 4.4BSD did not consult IFF_UP at all.
733: * The behavior here is to treat addresses on !IFF_UP interface
734: * as not mine.
1.1 cgd 735: */
1.100 itojun 736: downmatch = 0;
1.140 matt 737: LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) {
1.97 itojun 738: if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
1.165 christos 739: if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif)
740: continue;
1.97 itojun 741: if ((ia->ia_ifp->if_flags & IFF_UP) != 0)
742: break;
1.100 itojun 743: else
744: downmatch++;
1.97 itojun 745: }
746: }
1.86 thorpej 747: if (ia != NULL)
748: goto ours;
1.225 christos 749: if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
1.209 matt 750: IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) {
1.140 matt 751: if (ifa->ifa_addr->sa_family != AF_INET)
752: continue;
1.57 tls 753: ia = ifatoia(ifa);
1.35 mycroft 754: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
755: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 756: /*
757: * Look for all-0's host part (old broadcast addr),
758: * either for subnet or net.
759: */
760: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 761: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 762: goto ours;
1.57 tls 763: /*
764: * An interface with IP address zero accepts
765: * all packets that arrive on that interface.
766: */
767: if (in_nullhost(ia->ia_addr.sin_addr))
768: goto ours;
1.1 cgd 769: }
770: }
1.18 mycroft 771: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 772: struct in_multi *inm;
773: #ifdef MROUTING
774: extern struct socket *ip_mrouter;
1.10 brezak 775:
1.4 hpeyerl 776: if (ip_mrouter) {
777: /*
778: * If we are acting as a multicast router, all
779: * incoming multicast packets are passed to the
780: * kernel-level multicast forwarding function.
781: * The packet is returned (relatively) intact; if
782: * ip_mforward() returns a non-zero value, the packet
783: * must be discarded, else it may be accepted below.
784: *
785: * (The IP ident field is put in the same byte order
786: * as expected when ip_mforward() is called from
787: * ip_output().)
788: */
1.13 mycroft 789: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
1.266 thorpej 790: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 791: m_freem(m);
1.89 itojun 792: return;
1.4 hpeyerl 793: }
794:
795: /*
796: * The process-level routing demon needs to receive
797: * all multicast IGMP packets, whether or not this
798: * host belongs to their destination groups.
799: */
800: if (ip->ip_p == IPPROTO_IGMP)
801: goto ours;
1.266 thorpej 802: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 803: }
804: #endif
805: /*
806: * See if we belong to the destination multicast group on the
807: * arrival interface.
808: */
809: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
810: if (inm == NULL) {
1.266 thorpej 811: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 812: m_freem(m);
1.89 itojun 813: return;
1.4 hpeyerl 814: }
815: goto ours;
816: }
1.19 mycroft 817: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 818: in_nullhost(ip->ip_dst))
1.1 cgd 819: goto ours;
820:
821: /*
822: * Not for us; forward if possible and desirable.
823: */
824: if (ipforwarding == 0) {
1.266 thorpej 825: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 826: m_freem(m);
1.100 itojun 827: } else {
828: /*
829: * If ip_dst matched any of my address on !IFF_UP interface,
830: * and there's no IFF_UP interface that matches ip_dst,
831: * send icmp unreach. Forwarding it will result in in-kernel
832: * forwarding loop till TTL goes to 0.
833: */
834: if (downmatch) {
835: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
1.266 thorpej 836: IP_STATINC(IP_STAT_CANTFORWARD);
1.100 itojun 837: return;
838: }
1.145 itojun 839: #ifdef IPSEC
840: if (ipsec4_in_reject(m, NULL)) {
1.267 thorpej 841: IPSEC_STATINC(IPSEC_STAT_IN_POLVIO);
1.145 itojun 842: goto bad;
843: }
844: #endif
1.173 jonathan 845: #ifdef FAST_IPSEC
846: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
847: s = splsoftnet();
848: if (mtag != NULL) {
849: tdbi = (struct tdb_ident *)(mtag + 1);
850: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
851: } else {
852: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
1.212 perry 853: IP_FORWARDING, &error);
1.173 jonathan 854: }
855: if (sp == NULL) { /* NB: can happen if error */
856: splx(s);
857: /*XXX error stat???*/
858: DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/
859: goto bad;
860: }
861:
862: /*
863: * Check security policy against packet attributes.
864: */
865: error = ipsec_in_reject(sp, m);
866: KEY_FREESP(&sp);
867: splx(s);
868: if (error) {
1.266 thorpej 869: IP_STATINC(IP_STAT_CANTFORWARD);
1.173 jonathan 870: goto bad;
1.193 scw 871: }
872:
873: /*
874: * Peek at the outbound SP for this packet to determine if
875: * it's a Fast Forward candidate.
876: */
877: mtag = m_tag_find(m, PACKET_TAG_IPSEC_PENDING_TDB, NULL);
878: if (mtag != NULL)
879: m->m_flags &= ~M_CANFASTFWD;
880: else {
881: s = splsoftnet();
882: sp = ipsec4_checkpolicy(m, IPSEC_DIR_OUTBOUND,
883: (IP_FORWARDING |
884: (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
885: &error, NULL);
886: if (sp != NULL) {
887: m->m_flags &= ~M_CANFASTFWD;
888: KEY_FREESP(&sp);
889: }
890: splx(s);
1.173 jonathan 891: }
892: #endif /* FAST_IPSEC */
1.145 itojun 893:
1.169 itojun 894: ip_forward(m, srcrt);
1.100 itojun 895: }
1.89 itojun 896: return;
1.1 cgd 897:
898: ours:
899: /*
900: * If offset or IP_MF are set, must reassemble.
901: * Otherwise, nothing need be done.
902: * (We could look in the reassembly queue to see
903: * if the packet was previously fragmented,
904: * but it's not worth the time; just let them time out.)
905: */
1.155 itojun 906: if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
1.258 matt 907: uint16_t off;
908: /*
909: * Prevent TCP blind data attacks by not allowing non-initial
910: * fragments to start at less than 68 bytes (minimal fragment
1.259 matt 911: * size) and making sure the first fragment is at least 68
912: * bytes.
1.258 matt 913: */
1.260 matt 914: off = (ntohs(ip->ip_off) & IP_OFFMASK) << 3;
1.259 matt 915: if ((off > 0 ? off + hlen : len) < IP_MINFRAGSIZE - 1) {
1.266 thorpej 916: IP_STATINC(IP_STAT_BADFRAGS);
1.258 matt 917: goto bad;
918: }
1.1 cgd 919: /*
920: * Look for queue of fragments
921: * of this datagram.
922: */
1.75 thorpej 923: IPQ_LOCK();
1.190 jonathan 924: hash = IPREASS_HASH(ip->ip_src.s_addr, ip->ip_id);
1.250 dyoung 925: LIST_FOREACH(fp, &ipq[hash], ipq_q) {
1.1 cgd 926: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 927: in_hosteq(ip->ip_src, fp->ipq_src) &&
928: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.260 matt 929: ip->ip_p == fp->ipq_p) {
930: /*
931: * Make sure the TOS is matches previous
932: * fragments.
933: */
934: if (ip->ip_tos != fp->ipq_tos) {
1.266 thorpej 935: IP_STATINC(IP_STAT_BADFRAGS);
1.275.4.1 snj 936: IPQ_UNLOCK();
1.260 matt 937: goto bad;
938: }
1.1 cgd 939: goto found;
1.260 matt 940: }
1.190 jonathan 941: }
1.1 cgd 942: fp = 0;
943: found:
944:
945: /*
946: * Adjust ip_len to not reflect header,
1.25 cgd 947: * set ipqe_mff if more fragments are expected,
1.1 cgd 948: * convert offset of this to bytes.
949: */
1.155 itojun 950: ip->ip_len = htons(ntohs(ip->ip_len) - hlen);
951: mff = (ip->ip_off & htons(IP_MF)) != 0;
1.25 cgd 952: if (mff) {
1.16 cgd 953: /*
954: * Make sure that fragments have a data length
955: * that's a non-zero multiple of 8 bytes.
956: */
1.155 itojun 957: if (ntohs(ip->ip_len) == 0 ||
958: (ntohs(ip->ip_len) & 0x7) != 0) {
1.266 thorpej 959: IP_STATINC(IP_STAT_BADFRAGS);
1.75 thorpej 960: IPQ_UNLOCK();
1.16 cgd 961: goto bad;
962: }
963: }
1.155 itojun 964: ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3);
1.1 cgd 965:
966: /*
967: * If datagram marked as having more fragments
968: * or if this is not the first fragment,
969: * attempt reassembly; if it succeeds, proceed.
970: */
1.155 itojun 971: if (mff || ip->ip_off != htons(0)) {
1.266 thorpej 972: IP_STATINC(IP_STAT_FRAGMENTS);
1.233 tls 973: s = splvm();
1.72 thorpej 974: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.233 tls 975: splx(s);
1.25 cgd 976: if (ipqe == NULL) {
1.266 thorpej 977: IP_STATINC(IP_STAT_RCVMEMDROP);
1.75 thorpej 978: IPQ_UNLOCK();
1.25 cgd 979: goto bad;
980: }
981: ipqe->ipqe_mff = mff;
1.50 thorpej 982: ipqe->ipqe_m = m;
1.25 cgd 983: ipqe->ipqe_ip = ip;
1.190 jonathan 984: m = ip_reass(ipqe, fp, &ipq[hash]);
1.75 thorpej 985: if (m == 0) {
986: IPQ_UNLOCK();
1.89 itojun 987: return;
1.75 thorpej 988: }
1.266 thorpej 989: IP_STATINC(IP_STAT_REASSEMBLED);
1.50 thorpej 990: ip = mtod(m, struct ip *);
1.74 thorpej 991: hlen = ip->ip_hl << 2;
1.155 itojun 992: ip->ip_len = htons(ntohs(ip->ip_len) + hlen);
1.1 cgd 993: } else
994: if (fp)
995: ip_freef(fp);
1.75 thorpej 996: IPQ_UNLOCK();
1.79 mycroft 997: }
1.128 itojun 998:
1.173 jonathan 999: #if defined(IPSEC)
1.128 itojun 1000: /*
1001: * enforce IPsec policy checking if we are seeing last header.
1002: * note that we do not visit this with protocols with pcb layer
1003: * code - like udp/tcp/raw ip.
1004: */
1005: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 &&
1006: ipsec4_in_reject(m, NULL)) {
1.267 thorpej 1007: IPSEC_STATINC(IPSEC_STAT_IN_POLVIO);
1.128 itojun 1008: goto bad;
1009: }
1010: #endif
1.226 liamjfoy 1011: #ifdef FAST_IPSEC
1.173 jonathan 1012: /*
1013: * enforce IPsec policy checking if we are seeing last header.
1014: * note that we do not visit this with protocols with pcb layer
1015: * code - like udp/tcp/raw ip.
1016: */
1017: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) {
1018: /*
1019: * Check if the packet has already had IPsec processing
1020: * done. If so, then just pass it along. This tag gets
1021: * set during AH, ESP, etc. input handling, before the
1022: * packet is returned to the ip input queue for delivery.
1.212 perry 1023: */
1.173 jonathan 1024: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
1025: s = splsoftnet();
1026: if (mtag != NULL) {
1027: tdbi = (struct tdb_ident *)(mtag + 1);
1028: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
1029: } else {
1030: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
1.212 perry 1031: IP_FORWARDING, &error);
1.173 jonathan 1032: }
1033: if (sp != NULL) {
1034: /*
1035: * Check security policy against packet attributes.
1036: */
1037: error = ipsec_in_reject(sp, m);
1038: KEY_FREESP(&sp);
1039: } else {
1040: /* XXX error stat??? */
1041: error = EINVAL;
1042: DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/
1043: }
1044: splx(s);
1045: if (error)
1046: goto bad;
1047: }
1048: #endif /* FAST_IPSEC */
1.1 cgd 1049:
1050: /*
1051: * Switch out to protocol's input routine.
1052: */
1.82 aidan 1053: #if IFA_STATS
1.122 itojun 1054: if (ia && ip)
1.155 itojun 1055: ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
1.82 aidan 1056: #endif
1.266 thorpej 1057: IP_STATINC(IP_STAT_DELIVERED);
1.89 itojun 1058: {
1059: int off = hlen, nh = ip->ip_p;
1060:
1061: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
1062: return;
1063: }
1.1 cgd 1064: bad:
1065: m_freem(m);
1.135 thorpej 1066: return;
1067:
1068: badcsum:
1.266 thorpej 1069: IP_STATINC(IP_STAT_BADSUM);
1.135 thorpej 1070: m_freem(m);
1.1 cgd 1071: }
1072:
1073: /*
1074: * Take incoming datagram fragment and try to
1075: * reassemble it into whole datagram. If a chain for
1076: * reassembly of this datagram already exists, then it
1077: * is given as fp; otherwise have to make a chain.
1078: */
1.50 thorpej 1079: struct mbuf *
1.211 perry 1080: ip_reass(struct ipqent *ipqe, struct ipq *fp, struct ipqhead *ipqhead)
1.1 cgd 1081: {
1.109 augustss 1082: struct mbuf *m = ipqe->ipqe_m;
1083: struct ipqent *nq, *p, *q;
1.25 cgd 1084: struct ip *ip;
1.1 cgd 1085: struct mbuf *t;
1.25 cgd 1086: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.233 tls 1087: int i, next, s;
1.1 cgd 1088:
1.75 thorpej 1089: IPQ_LOCK_CHECK();
1090:
1.1 cgd 1091: /*
1092: * Presence of header sizes in mbufs
1093: * would confuse code below.
1094: */
1095: m->m_data += hlen;
1096: m->m_len -= hlen;
1097:
1.194 jonathan 1098: #ifdef notyet
1099: /* make sure fragment limit is up-to-date */
1100: CHECK_NMBCLUSTER_PARAMS();
1101:
1102: /* If we have too many fragments, drop the older half. */
1103: if (ip_nfrags >= ip_maxfrags)
1104: ip_reass_drophalf(void);
1105: #endif
1106:
1.1 cgd 1107: /*
1.192 jonathan 1108: * We are about to add a fragment; increment frag count.
1109: */
1110: ip_nfrags++;
1.212 perry 1111:
1.192 jonathan 1112: /*
1.1 cgd 1113: * If first fragment to arrive, create a reassembly queue.
1114: */
1115: if (fp == 0) {
1.131 itojun 1116: /*
1117: * Enforce upper bound on number of fragmented packets
1118: * for which we attempt reassembly;
1119: * If maxfrag is 0, never accept fragments.
1120: * If maxfrag is -1, accept all fragments without limitation.
1121: */
1122: if (ip_maxfragpackets < 0)
1123: ;
1124: else if (ip_nfragpackets >= ip_maxfragpackets)
1125: goto dropfrag;
1126: ip_nfragpackets++;
1.50 thorpej 1127: MALLOC(fp, struct ipq *, sizeof (struct ipq),
1128: M_FTABLE, M_NOWAIT);
1129: if (fp == NULL)
1.1 cgd 1130: goto dropfrag;
1.190 jonathan 1131: LIST_INSERT_HEAD(ipqhead, fp, ipq_q);
1.192 jonathan 1132: fp->ipq_nfrags = 1;
1.1 cgd 1133: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 1134: fp->ipq_p = ipqe->ipqe_ip->ip_p;
1135: fp->ipq_id = ipqe->ipqe_ip->ip_id;
1.260 matt 1136: fp->ipq_tos = ipqe->ipqe_ip->ip_tos;
1.148 matt 1137: TAILQ_INIT(&fp->ipq_fragq);
1.25 cgd 1138: fp->ipq_src = ipqe->ipqe_ip->ip_src;
1139: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
1140: p = NULL;
1.1 cgd 1141: goto insert;
1.192 jonathan 1142: } else {
1143: fp->ipq_nfrags++;
1.1 cgd 1144: }
1145:
1146: /*
1147: * Find a segment which begins after this one does.
1148: */
1.148 matt 1149: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1150: p = q, q = TAILQ_NEXT(q, ipqe_q))
1.155 itojun 1151: if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off))
1.1 cgd 1152: break;
1153:
1154: /*
1155: * If there is a preceding segment, it may provide some of
1156: * our data already. If so, drop the data from the incoming
1157: * segment. If it provides all of our data, drop us.
1158: */
1.25 cgd 1159: if (p != NULL) {
1.155 itojun 1160: i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) -
1161: ntohs(ipqe->ipqe_ip->ip_off);
1.1 cgd 1162: if (i > 0) {
1.155 itojun 1163: if (i >= ntohs(ipqe->ipqe_ip->ip_len))
1.1 cgd 1164: goto dropfrag;
1.50 thorpej 1165: m_adj(ipqe->ipqe_m, i);
1.155 itojun 1166: ipqe->ipqe_ip->ip_off =
1167: htons(ntohs(ipqe->ipqe_ip->ip_off) + i);
1168: ipqe->ipqe_ip->ip_len =
1169: htons(ntohs(ipqe->ipqe_ip->ip_len) - i);
1.1 cgd 1170: }
1171: }
1172:
1173: /*
1174: * While we overlap succeeding segments trim them or,
1175: * if they are completely covered, dequeue them.
1176: */
1.155 itojun 1177: for (; q != NULL &&
1178: ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) >
1179: ntohs(q->ipqe_ip->ip_off); q = nq) {
1180: i = (ntohs(ipqe->ipqe_ip->ip_off) +
1181: ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off);
1182: if (i < ntohs(q->ipqe_ip->ip_len)) {
1183: q->ipqe_ip->ip_len =
1184: htons(ntohs(q->ipqe_ip->ip_len) - i);
1185: q->ipqe_ip->ip_off =
1186: htons(ntohs(q->ipqe_ip->ip_off) + i);
1.50 thorpej 1187: m_adj(q->ipqe_m, i);
1.1 cgd 1188: break;
1189: }
1.148 matt 1190: nq = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1191: m_freem(q->ipqe_m);
1.148 matt 1192: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.233 tls 1193: s = splvm();
1.72 thorpej 1194: pool_put(&ipqent_pool, q);
1.233 tls 1195: splx(s);
1.192 jonathan 1196: fp->ipq_nfrags--;
1197: ip_nfrags--;
1.1 cgd 1198: }
1199:
1200: insert:
1201: /*
1202: * Stick new segment in its place;
1203: * check for complete reassembly.
1204: */
1.25 cgd 1205: if (p == NULL) {
1.148 matt 1206: TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
1.25 cgd 1207: } else {
1.148 matt 1208: TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q);
1.25 cgd 1209: }
1.1 cgd 1210: next = 0;
1.148 matt 1211: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1212: p = q, q = TAILQ_NEXT(q, ipqe_q)) {
1.155 itojun 1213: if (ntohs(q->ipqe_ip->ip_off) != next)
1.1 cgd 1214: return (0);
1.155 itojun 1215: next += ntohs(q->ipqe_ip->ip_len);
1.1 cgd 1216: }
1.25 cgd 1217: if (p->ipqe_mff)
1.1 cgd 1218: return (0);
1219:
1220: /*
1.41 thorpej 1221: * Reassembly is complete. Check for a bogus message size and
1222: * concatenate fragments.
1.1 cgd 1223: */
1.148 matt 1224: q = TAILQ_FIRST(&fp->ipq_fragq);
1.25 cgd 1225: ip = q->ipqe_ip;
1.41 thorpej 1226: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
1.266 thorpej 1227: IP_STATINC(IP_STAT_TOOLONG);
1.41 thorpej 1228: ip_freef(fp);
1229: return (0);
1230: }
1.50 thorpej 1231: m = q->ipqe_m;
1.1 cgd 1232: t = m->m_next;
1233: m->m_next = 0;
1234: m_cat(m, t);
1.148 matt 1235: nq = TAILQ_NEXT(q, ipqe_q);
1.233 tls 1236: s = splvm();
1.72 thorpej 1237: pool_put(&ipqent_pool, q);
1.233 tls 1238: splx(s);
1.25 cgd 1239: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 1240: t = q->ipqe_m;
1.148 matt 1241: nq = TAILQ_NEXT(q, ipqe_q);
1.233 tls 1242: s = splvm();
1.72 thorpej 1243: pool_put(&ipqent_pool, q);
1.233 tls 1244: splx(s);
1.1 cgd 1245: m_cat(m, t);
1246: }
1.192 jonathan 1247: ip_nfrags -= fp->ipq_nfrags;
1.1 cgd 1248:
1249: /*
1250: * Create header for new ip packet by
1251: * modifying header of first packet;
1252: * dequeue and discard fragment reassembly header.
1253: * Make header visible.
1254: */
1.155 itojun 1255: ip->ip_len = htons(next);
1.25 cgd 1256: ip->ip_src = fp->ipq_src;
1257: ip->ip_dst = fp->ipq_dst;
1258: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1259: FREE(fp, M_FTABLE);
1.131 itojun 1260: ip_nfragpackets--;
1.1 cgd 1261: m->m_len += (ip->ip_hl << 2);
1262: m->m_data -= (ip->ip_hl << 2);
1263: /* some debugging cruft by sklower, below, will go away soon */
1264: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
1.109 augustss 1265: int plen = 0;
1.50 thorpej 1266: for (t = m; t; t = t->m_next)
1267: plen += t->m_len;
1268: m->m_pkthdr.len = plen;
1.213 yamt 1269: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1270: }
1.50 thorpej 1271: return (m);
1.1 cgd 1272:
1273: dropfrag:
1.192 jonathan 1274: if (fp != 0)
1275: fp->ipq_nfrags--;
1276: ip_nfrags--;
1.266 thorpej 1277: IP_STATINC(IP_STAT_FRAGDROPPED);
1.1 cgd 1278: m_freem(m);
1.233 tls 1279: s = splvm();
1.72 thorpej 1280: pool_put(&ipqent_pool, ipqe);
1.233 tls 1281: splx(s);
1.1 cgd 1282: return (0);
1283: }
1284:
1285: /*
1286: * Free a fragment reassembly header and all
1287: * associated datagrams.
1288: */
1.8 mycroft 1289: void
1.211 perry 1290: ip_freef(struct ipq *fp)
1.1 cgd 1291: {
1.109 augustss 1292: struct ipqent *q, *p;
1.192 jonathan 1293: u_int nfrags = 0;
1.233 tls 1294: int s;
1.1 cgd 1295:
1.75 thorpej 1296: IPQ_LOCK_CHECK();
1297:
1.148 matt 1298: for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) {
1299: p = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1300: m_freem(q->ipqe_m);
1.192 jonathan 1301: nfrags++;
1.148 matt 1302: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.233 tls 1303: s = splvm();
1.72 thorpej 1304: pool_put(&ipqent_pool, q);
1.233 tls 1305: splx(s);
1.1 cgd 1306: }
1.192 jonathan 1307:
1308: if (nfrags != fp->ipq_nfrags)
1309: printf("ip_freef: nfrags %d != %d\n", fp->ipq_nfrags, nfrags);
1310: ip_nfrags -= nfrags;
1.25 cgd 1311: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1312: FREE(fp, M_FTABLE);
1.131 itojun 1313: ip_nfragpackets--;
1.1 cgd 1314: }
1315:
1316: /*
1.194 jonathan 1317: * IP reassembly TTL machinery for multiplicative drop.
1318: */
1319: static u_int fragttl_histo[(IPFRAGTTL+1)];
1320:
1321:
1322: /*
1323: * Decrement TTL of all reasembly queue entries by `ticks'.
1324: * Count number of distinct fragments (as opposed to partial, fragmented
1325: * datagrams) in the reassembly queue. While we traverse the entire
1326: * reassembly queue, compute and return the median TTL over all fragments.
1327: */
1328: static u_int
1329: ip_reass_ttl_decr(u_int ticks)
1330: {
1.198 matt 1331: u_int nfrags, median, dropfraction, keepfraction;
1.194 jonathan 1332: struct ipq *fp, *nfp;
1.198 matt 1333: int i;
1.212 perry 1334:
1.194 jonathan 1335: nfrags = 0;
1336: memset(fragttl_histo, 0, sizeof fragttl_histo);
1.212 perry 1337:
1.194 jonathan 1338: for (i = 0; i < IPREASS_NHASH; i++) {
1339: for (fp = LIST_FIRST(&ipq[i]); fp != NULL; fp = nfp) {
1340: fp->ipq_ttl = ((fp->ipq_ttl <= ticks) ?
1341: 0 : fp->ipq_ttl - ticks);
1342: nfp = LIST_NEXT(fp, ipq_q);
1343: if (fp->ipq_ttl == 0) {
1.266 thorpej 1344: IP_STATINC(IP_STAT_FRAGTIMEOUT);
1.194 jonathan 1345: ip_freef(fp);
1346: } else {
1347: nfrags += fp->ipq_nfrags;
1348: fragttl_histo[fp->ipq_ttl] += fp->ipq_nfrags;
1349: }
1350: }
1351: }
1352:
1353: KASSERT(ip_nfrags == nfrags);
1354:
1355: /* Find median (or other drop fraction) in histogram. */
1356: dropfraction = (ip_nfrags / 2);
1357: keepfraction = ip_nfrags - dropfraction;
1358: for (i = IPFRAGTTL, median = 0; i >= 0; i--) {
1359: median += fragttl_histo[i];
1360: if (median >= keepfraction)
1361: break;
1362: }
1363:
1364: /* Return TTL of median (or other fraction). */
1365: return (u_int)i;
1366: }
1367:
1368: void
1369: ip_reass_drophalf(void)
1370: {
1371:
1372: u_int median_ticks;
1373: /*
1374: * Compute median TTL of all fragments, and count frags
1375: * with that TTL or lower (roughly half of all fragments).
1376: */
1377: median_ticks = ip_reass_ttl_decr(0);
1378:
1379: /* Drop half. */
1380: median_ticks = ip_reass_ttl_decr(median_ticks);
1381:
1382: }
1383:
1384: /*
1.1 cgd 1385: * IP timer processing;
1386: * if a timer expires on a reassembly
1387: * queue, discard it.
1388: */
1.8 mycroft 1389: void
1.211 perry 1390: ip_slowtimo(void)
1.1 cgd 1391: {
1.191 jonathan 1392: static u_int dropscanidx = 0;
1393: u_int i;
1.194 jonathan 1394: u_int median_ttl;
1.268 ad 1395:
1396: mutex_enter(softnet_lock);
1397: KERNEL_LOCK(1, NULL);
1.1 cgd 1398:
1.75 thorpej 1399: IPQ_LOCK();
1.194 jonathan 1400:
1401: /* Age TTL of all fragments by 1 tick .*/
1402: median_ttl = ip_reass_ttl_decr(1);
1403:
1404: /* make sure fragment limit is up-to-date */
1405: CHECK_NMBCLUSTER_PARAMS();
1406:
1407: /* If we have too many fragments, drop the older half. */
1408: if (ip_nfrags > ip_maxfrags)
1409: ip_reass_ttl_decr(median_ttl);
1410:
1.131 itojun 1411: /*
1.194 jonathan 1412: * If we are over the maximum number of fragmented packets
1.131 itojun 1413: * (due to the limit being lowered), drain off
1.190 jonathan 1414: * enough to get down to the new limit. Start draining
1415: * from the reassembly hashqueue most recently drained.
1.131 itojun 1416: */
1417: if (ip_maxfragpackets < 0)
1418: ;
1419: else {
1.190 jonathan 1420: int wrapped = 0;
1421:
1422: i = dropscanidx;
1423: while (ip_nfragpackets > ip_maxfragpackets && wrapped == 0) {
1424: while (LIST_FIRST(&ipq[i]) != NULL)
1425: ip_freef(LIST_FIRST(&ipq[i]));
1426: if (++i >= IPREASS_NHASH) {
1427: i = 0;
1428: }
1429: /*
1430: * Dont scan forever even if fragment counters are
1431: * wrong: stop after scanning entire reassembly queue.
1432: */
1433: if (i == dropscanidx)
1434: wrapped = 1;
1435: }
1436: dropscanidx = i;
1.131 itojun 1437: }
1.75 thorpej 1438: IPQ_UNLOCK();
1.268 ad 1439:
1440: KERNEL_UNLOCK_ONE(NULL);
1441: mutex_exit(softnet_lock);
1.1 cgd 1442: }
1443:
1444: /*
1.270 ad 1445: * Drain off all datagram fragments. Don't acquire softnet_lock as
1446: * can be called from hardware interrupt context.
1.1 cgd 1447: */
1.8 mycroft 1448: void
1.211 perry 1449: ip_drain(void)
1.1 cgd 1450: {
1451:
1.268 ad 1452: KERNEL_LOCK(1, NULL);
1453:
1.75 thorpej 1454: /*
1455: * We may be called from a device's interrupt context. If
1456: * the ipq is already busy, just bail out now.
1457: */
1.268 ad 1458: if (ipq_lock_try() != 0) {
1459: /*
1460: * Drop half the total fragments now. If more mbufs are
1461: * needed, we will be called again soon.
1462: */
1463: ip_reass_drophalf();
1464: IPQ_UNLOCK();
1465: }
1.75 thorpej 1466:
1.268 ad 1467: KERNEL_UNLOCK_ONE(NULL);
1.1 cgd 1468: }
1469:
1470: /*
1471: * Do option processing on a datagram,
1472: * possibly discarding it if bad options are encountered,
1473: * or forwarding it if source-routed.
1474: * Returns 1 if packet has been forwarded/freed,
1475: * 0 if the packet should be processed further.
1476: */
1.8 mycroft 1477: int
1.211 perry 1478: ip_dooptions(struct mbuf *m)
1.1 cgd 1479: {
1.109 augustss 1480: struct ip *ip = mtod(m, struct ip *);
1481: u_char *cp, *cp0;
1482: struct ip_timestamp *ipt;
1483: struct in_ifaddr *ia;
1.1 cgd 1484: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 1485: struct in_addr dst;
1.1 cgd 1486: n_time ntime;
1487:
1.13 mycroft 1488: dst = ip->ip_dst;
1.1 cgd 1489: cp = (u_char *)(ip + 1);
1490: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
1491: for (; cnt > 0; cnt -= optlen, cp += optlen) {
1492: opt = cp[IPOPT_OPTVAL];
1493: if (opt == IPOPT_EOL)
1494: break;
1495: if (opt == IPOPT_NOP)
1496: optlen = 1;
1497: else {
1.113 itojun 1498: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1499: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1500: goto bad;
1501: }
1.1 cgd 1502: optlen = cp[IPOPT_OLEN];
1.114 itojun 1503: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 1504: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1505: goto bad;
1506: }
1507: }
1508: switch (opt) {
1509:
1510: default:
1511: break;
1512:
1513: /*
1514: * Source routing with record.
1515: * Find interface with current destination address.
1516: * If none on this machine then drop if strictly routed,
1517: * or do nothing if loosely routed.
1518: * Record interface address and bring up next address
1519: * component. If strictly routed make sure next
1520: * address is on directly accessible net.
1521: */
1522: case IPOPT_LSRR:
1523: case IPOPT_SSRR:
1.47 cjs 1524: if (ip_allowsrcrt == 0) {
1525: type = ICMP_UNREACH;
1526: code = ICMP_UNREACH_NET_PROHIB;
1527: goto bad;
1528: }
1.114 itojun 1529: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1530: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1531: goto bad;
1532: }
1.1 cgd 1533: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1534: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1535: goto bad;
1536: }
1537: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 1538: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 1539: if (ia == 0) {
1540: if (opt == IPOPT_SSRR) {
1541: type = ICMP_UNREACH;
1542: code = ICMP_UNREACH_SRCFAIL;
1543: goto bad;
1544: }
1545: /*
1546: * Loose routing, and not at next destination
1547: * yet; nothing to do except forward.
1548: */
1549: break;
1550: }
1551: off--; /* 0 origin */
1.112 sommerfe 1552: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 1553: /*
1554: * End of source route. Should be for us.
1555: */
1556: save_rte(cp, ip->ip_src);
1557: break;
1558: }
1559: /*
1560: * locate outgoing interface
1561: */
1.244 christos 1562: bcopy((void *)(cp + off), (void *)&ipaddr.sin_addr,
1.1 cgd 1563: sizeof(ipaddr.sin_addr));
1.96 thorpej 1564: if (opt == IPOPT_SSRR)
1.196 itojun 1565: ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr)));
1.96 thorpej 1566: else
1.1 cgd 1567: ia = ip_rtaddr(ipaddr.sin_addr);
1568: if (ia == 0) {
1569: type = ICMP_UNREACH;
1570: code = ICMP_UNREACH_SRCFAIL;
1571: goto bad;
1572: }
1573: ip->ip_dst = ipaddr.sin_addr;
1.244 christos 1574: bcopy((void *)&ia->ia_addr.sin_addr,
1575: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1576: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1577: /*
1578: * Let ip_intr's mcast routing check handle mcast pkts
1579: */
1.18 mycroft 1580: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1581: break;
1582:
1583: case IPOPT_RR:
1.114 itojun 1584: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1585: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1586: goto bad;
1587: }
1.1 cgd 1588: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1589: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1590: goto bad;
1591: }
1592: /*
1593: * If no space remains, ignore.
1594: */
1595: off--; /* 0 origin */
1.112 sommerfe 1596: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1597: break;
1.244 christos 1598: bcopy((void *)(&ip->ip_dst), (void *)&ipaddr.sin_addr,
1.1 cgd 1599: sizeof(ipaddr.sin_addr));
1600: /*
1601: * locate outgoing interface; if we're the destination,
1602: * use the incoming interface (should be same).
1603: */
1.96 thorpej 1604: if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))))
1605: == NULL &&
1606: (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
1.1 cgd 1607: type = ICMP_UNREACH;
1608: code = ICMP_UNREACH_HOST;
1609: goto bad;
1610: }
1.244 christos 1611: bcopy((void *)&ia->ia_addr.sin_addr,
1612: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1613: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1614: break;
1615:
1616: case IPOPT_TS:
1617: code = cp - (u_char *)ip;
1618: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1619: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1620: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1621: goto bad;
1.114 itojun 1622: }
1623: if (ipt->ipt_ptr < 5) {
1624: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1625: goto bad;
1626: }
1.15 cgd 1627: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1628: if (++ipt->ipt_oflw == 0) {
1629: code = (u_char *)&ipt->ipt_ptr -
1630: (u_char *)ip;
1.1 cgd 1631: goto bad;
1.114 itojun 1632: }
1.1 cgd 1633: break;
1634: }
1.104 thorpej 1635: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1636: switch (ipt->ipt_flg) {
1637:
1638: case IPOPT_TS_TSONLY:
1639: break;
1640:
1641: case IPOPT_TS_TSANDADDR:
1.66 thorpej 1642: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1643: sizeof(struct in_addr) > ipt->ipt_len) {
1644: code = (u_char *)&ipt->ipt_ptr -
1645: (u_char *)ip;
1.1 cgd 1646: goto bad;
1.114 itojun 1647: }
1.13 mycroft 1648: ipaddr.sin_addr = dst;
1.96 thorpej 1649: ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr),
1650: m->m_pkthdr.rcvif));
1.13 mycroft 1651: if (ia == 0)
1652: continue;
1.104 thorpej 1653: bcopy(&ia->ia_addr.sin_addr,
1654: cp0, sizeof(struct in_addr));
1.1 cgd 1655: ipt->ipt_ptr += sizeof(struct in_addr);
1656: break;
1657:
1658: case IPOPT_TS_PRESPEC:
1.66 thorpej 1659: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1660: sizeof(struct in_addr) > ipt->ipt_len) {
1661: code = (u_char *)&ipt->ipt_ptr -
1662: (u_char *)ip;
1.1 cgd 1663: goto bad;
1.114 itojun 1664: }
1.104 thorpej 1665: bcopy(cp0, &ipaddr.sin_addr,
1.1 cgd 1666: sizeof(struct in_addr));
1.96 thorpej 1667: if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))
1668: == NULL)
1.1 cgd 1669: continue;
1670: ipt->ipt_ptr += sizeof(struct in_addr);
1671: break;
1672:
1673: default:
1.114 itojun 1674: /* XXX can't take &ipt->ipt_flg */
1675: code = (u_char *)&ipt->ipt_ptr -
1676: (u_char *)ip + 1;
1.1 cgd 1677: goto bad;
1678: }
1679: ntime = iptime();
1.107 thorpej 1680: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1.244 christos 1681: memmove((char *)cp + ipt->ipt_ptr - 1, cp0,
1.1 cgd 1682: sizeof(n_time));
1683: ipt->ipt_ptr += sizeof(n_time);
1684: }
1685: }
1686: if (forward) {
1.26 thorpej 1687: if (ip_forwsrcrt == 0) {
1688: type = ICMP_UNREACH;
1689: code = ICMP_UNREACH_SRCFAIL;
1690: goto bad;
1691: }
1.1 cgd 1692: ip_forward(m, 1);
1693: return (1);
1.13 mycroft 1694: }
1695: return (0);
1.1 cgd 1696: bad:
1.13 mycroft 1697: icmp_error(m, type, code, 0, 0);
1.266 thorpej 1698: IP_STATINC(IP_STAT_BADOPTIONS);
1.1 cgd 1699: return (1);
1700: }
1701:
1702: /*
1703: * Given address of next destination (final or next hop),
1704: * return internet address info of interface to be used to get there.
1705: */
1706: struct in_ifaddr *
1.211 perry 1707: ip_rtaddr(struct in_addr dst)
1.1 cgd 1708: {
1.249 dyoung 1709: struct rtentry *rt;
1710: union {
1711: struct sockaddr dst;
1712: struct sockaddr_in dst4;
1713: } u;
1714:
1715: sockaddr_in_init(&u.dst4, &dst, 0);
1716:
1717: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL)
1718: return NULL;
1719:
1720: return ifatoia(rt->rt_ifa);
1.1 cgd 1721: }
1722:
1723: /*
1724: * Save incoming source route for use in replies,
1725: * to be picked up later by ip_srcroute if the receiver is interested.
1726: */
1.13 mycroft 1727: void
1.211 perry 1728: save_rte(u_char *option, struct in_addr dst)
1.1 cgd 1729: {
1730: unsigned olen;
1731:
1732: olen = option[IPOPT_OLEN];
1733: #ifdef DIAGNOSTIC
1734: if (ipprintfs)
1.39 christos 1735: printf("save_rte: olen %d\n", olen);
1.89 itojun 1736: #endif /* 0 */
1.1 cgd 1737: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1738: return;
1.244 christos 1739: bcopy((void *)option, (void *)ip_srcrt.srcopt, olen);
1.1 cgd 1740: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1741: ip_srcrt.dst = dst;
1742: }
1743:
1744: /*
1745: * Retrieve incoming source route for use in replies,
1746: * in the same form used by setsockopt.
1747: * The first hop is placed before the options, will be removed later.
1748: */
1749: struct mbuf *
1.211 perry 1750: ip_srcroute(void)
1.1 cgd 1751: {
1.109 augustss 1752: struct in_addr *p, *q;
1753: struct mbuf *m;
1.1 cgd 1754:
1755: if (ip_nhops == 0)
1.237 dyoung 1756: return NULL;
1.1 cgd 1757: m = m_get(M_DONTWAIT, MT_SOOPTS);
1758: if (m == 0)
1.237 dyoung 1759: return NULL;
1.1 cgd 1760:
1.164 matt 1761: MCLAIM(m, &inetdomain.dom_mowner);
1.13 mycroft 1762: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1763:
1764: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1765: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1766: OPTSIZ;
1767: #ifdef DIAGNOSTIC
1768: if (ipprintfs)
1.39 christos 1769: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1770: #endif
1771:
1772: /*
1773: * First save first hop for return route
1774: */
1775: p = &ip_srcrt.route[ip_nhops - 1];
1776: *(mtod(m, struct in_addr *)) = *p--;
1777: #ifdef DIAGNOSTIC
1778: if (ipprintfs)
1.39 christos 1779: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1780: #endif
1781:
1782: /*
1783: * Copy option fields and padding (nop) to mbuf.
1784: */
1785: ip_srcrt.nop = IPOPT_NOP;
1786: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1.244 christos 1787: memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop,
1788: OPTSIZ);
1789: q = (struct in_addr *)(mtod(m, char *) +
1.1 cgd 1790: sizeof(struct in_addr) + OPTSIZ);
1791: #undef OPTSIZ
1792: /*
1793: * Record return path as an IP source route,
1794: * reversing the path (pointers are now aligned).
1795: */
1796: while (p >= ip_srcrt.route) {
1797: #ifdef DIAGNOSTIC
1798: if (ipprintfs)
1.39 christos 1799: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1800: #endif
1801: *q++ = *p--;
1802: }
1803: /*
1804: * Last hop goes to final destination.
1805: */
1806: *q = ip_srcrt.dst;
1807: #ifdef DIAGNOSTIC
1808: if (ipprintfs)
1.39 christos 1809: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1810: #endif
1811: return (m);
1812: }
1813:
1.139 matt 1814: const int inetctlerrmap[PRC_NCMDS] = {
1.256 yamt 1815: [PRC_MSGSIZE] = EMSGSIZE,
1816: [PRC_HOSTDEAD] = EHOSTDOWN,
1817: [PRC_HOSTUNREACH] = EHOSTUNREACH,
1818: [PRC_UNREACH_NET] = EHOSTUNREACH,
1819: [PRC_UNREACH_HOST] = EHOSTUNREACH,
1820: [PRC_UNREACH_PROTOCOL] = ECONNREFUSED,
1821: [PRC_UNREACH_PORT] = ECONNREFUSED,
1822: [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH,
1823: [PRC_PARAMPROB] = ENOPROTOOPT,
1.1 cgd 1824: };
1825:
1826: /*
1827: * Forward a packet. If some error occurs return the sender
1828: * an icmp packet. Note we can't always generate a meaningful
1829: * icmp message because icmp doesn't have a large enough repertoire
1830: * of codes and types.
1831: *
1832: * If not forwarding, just drop the packet. This could be confusing
1833: * if ipforwarding was zero but some routing protocol was advancing
1834: * us as a gateway to somewhere. However, we must let the routing
1835: * protocol deal with that.
1836: *
1837: * The srcrt parameter indicates whether the packet is being forwarded
1838: * via a source route.
1839: */
1.13 mycroft 1840: void
1.211 perry 1841: ip_forward(struct mbuf *m, int srcrt)
1.1 cgd 1842: {
1.109 augustss 1843: struct ip *ip = mtod(m, struct ip *);
1844: struct rtentry *rt;
1.220 christos 1845: int error, type = 0, code = 0, destmtu = 0;
1.1 cgd 1846: struct mbuf *mcopy;
1.13 mycroft 1847: n_long dest;
1.249 dyoung 1848: union {
1849: struct sockaddr dst;
1850: struct sockaddr_in dst4;
1851: } u;
1.164 matt 1852:
1853: /*
1854: * We are now in the output path.
1855: */
1856: MCLAIM(m, &ip_tx_mowner);
1.135 thorpej 1857:
1858: /*
1859: * Clear any in-bound checksum flags for this packet.
1860: */
1861: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1862:
1.13 mycroft 1863: dest = 0;
1.1 cgd 1864: #ifdef DIAGNOSTIC
1.224 joerg 1865: if (ipprintfs) {
1866: printf("forward: src %s ", inet_ntoa(ip->ip_src));
1867: printf("dst %s ttl %x\n", inet_ntoa(ip->ip_dst), ip->ip_ttl);
1868: }
1.1 cgd 1869: #endif
1.93 sommerfe 1870: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.266 thorpej 1871: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 1872: m_freem(m);
1873: return;
1874: }
1875: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1876: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1877: return;
1878: }
1879:
1.249 dyoung 1880: sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
1881: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) {
1882: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0);
1883: return;
1.1 cgd 1884: }
1885:
1886: /*
1.34 mycroft 1887: * Save at most 68 bytes of the packet in case
1.1 cgd 1888: * we need to generate an ICMP message to the src.
1.119 itojun 1889: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1890: */
1.155 itojun 1891: mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1.119 itojun 1892: if (mcopy)
1893: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1894:
1.221 christos 1895: ip->ip_ttl -= IPTTLDEC;
1896:
1.1 cgd 1897: /*
1898: * If forwarding packet using same interface that it came in on,
1899: * perhaps should send a redirect to sender to shortcut a hop.
1900: * Only send redirect if source is sending directly to us,
1901: * and if packet was not source routed (or has any options).
1902: * Also, don't send redirect if forwarding using a default route
1903: * or a route modified by a redirect.
1904: */
1905: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1906: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.250 dyoung 1907: !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) &&
1.1 cgd 1908: ipsendredirects && !srcrt) {
1.19 mycroft 1909: if (rt->rt_ifa &&
1910: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1911: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1912: if (rt->rt_flags & RTF_GATEWAY)
1913: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1914: else
1915: dest = ip->ip_dst.s_addr;
1916: /*
1917: * Router requirements says to only send host
1918: * redirects.
1919: */
1920: type = ICMP_REDIRECT;
1921: code = ICMP_REDIRECT_HOST;
1.1 cgd 1922: #ifdef DIAGNOSTIC
1.77 thorpej 1923: if (ipprintfs)
1924: printf("redirect (%d) to %x\n", code,
1925: (u_int32_t)dest);
1.1 cgd 1926: #endif
1927: }
1928: }
1929:
1.238 dyoung 1930: error = ip_output(m, NULL, &ipforward_rt,
1.173 jonathan 1931: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
1.174 itojun 1932: (struct ip_moptions *)NULL, (struct socket *)NULL);
1.173 jonathan 1933:
1.1 cgd 1934: if (error)
1.266 thorpej 1935: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 1936: else {
1.266 thorpej 1937: uint64_t *ips = IP_STAT_GETREF();
1938: ips[IP_STAT_FORWARD]++;
1939: if (type) {
1940: ips[IP_STAT_REDIRECTSENT]++;
1941: IP_STAT_PUTREF();
1942: } else {
1943: IP_STAT_PUTREF();
1.63 matt 1944: if (mcopy) {
1945: #ifdef GATEWAY
1.64 thorpej 1946: if (mcopy->m_flags & M_CANFASTFWD)
1947: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1948: #endif
1.1 cgd 1949: m_freem(mcopy);
1.63 matt 1950: }
1.1 cgd 1951: return;
1952: }
1953: }
1954: if (mcopy == NULL)
1955: return;
1.13 mycroft 1956:
1.1 cgd 1957: switch (error) {
1958:
1959: case 0: /* forwarded, but need redirect */
1960: /* type, code set above */
1961: break;
1962:
1963: case ENETUNREACH: /* shouldn't happen, checked above */
1964: case EHOSTUNREACH:
1965: case ENETDOWN:
1966: case EHOSTDOWN:
1967: default:
1968: type = ICMP_UNREACH;
1969: code = ICMP_UNREACH_HOST;
1970: break;
1971:
1972: case EMSGSIZE:
1973: type = ICMP_UNREACH;
1974: code = ICMP_UNREACH_NEEDFRAG;
1.263 cube 1975:
1.274 seanb 1976: if ((rt = rtcache_validate(&ipforward_rt)) != NULL)
1977: destmtu = rt->rt_ifp->if_mtu;
1.263 cube 1978:
1979: #if defined(IPSEC) || defined(FAST_IPSEC)
1.274 seanb 1980: {
1.263 cube 1981: /*
1982: * If the packet is routed over IPsec tunnel, tell the
1983: * originator the tunnel MTU.
1984: * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1985: * XXX quickhack!!!
1986: */
1987:
1.89 itojun 1988: struct secpolicy *sp;
1989: int ipsecerror;
1.95 itojun 1990: size_t ipsechdr;
1.89 itojun 1991: struct route *ro;
1992:
1993: sp = ipsec4_getpolicybyaddr(mcopy,
1.170 itojun 1994: IPSEC_DIR_OUTBOUND, IP_FORWARDING,
1995: &ipsecerror);
1.89 itojun 1996:
1.263 cube 1997: if (sp != NULL) {
1.89 itojun 1998: /* count IPsec header size */
1.95 itojun 1999: ipsechdr = ipsec4_hdrsiz(mcopy,
1.170 itojun 2000: IPSEC_DIR_OUTBOUND, NULL);
1.89 itojun 2001:
2002: /*
2003: * find the correct route for outer IPv4
2004: * header, compute tunnel MTU.
2005: */
1.220 christos 2006:
1.89 itojun 2007: if (sp->req != NULL
1.95 itojun 2008: && sp->req->sav != NULL
2009: && sp->req->sav->sah != NULL) {
2010: ro = &sp->req->sav->sah->sa_route;
1.274 seanb 2011: rt = rtcache_validate(ro);
1.257 dyoung 2012: if (rt && rt->rt_ifp) {
1.220 christos 2013: destmtu =
1.257 dyoung 2014: rt->rt_rmx.rmx_mtu ?
2015: rt->rt_rmx.rmx_mtu :
2016: rt->rt_ifp->if_mtu;
1.220 christos 2017: destmtu -= ipsechdr;
1.89 itojun 2018: }
2019: }
2020:
1.173 jonathan 2021: #ifdef IPSEC
1.89 itojun 2022: key_freesp(sp);
1.173 jonathan 2023: #else
2024: KEY_FREESP(&sp);
2025: #endif
1.89 itojun 2026: }
1.274 seanb 2027: }
1.263 cube 2028: #endif /*defined(IPSEC) || defined(FAST_IPSEC)*/
1.266 thorpej 2029: IP_STATINC(IP_STAT_CANTFRAG);
1.1 cgd 2030: break;
2031:
2032: case ENOBUFS:
1.143 itojun 2033: #if 1
2034: /*
2035: * a router should not generate ICMP_SOURCEQUENCH as
2036: * required in RFC1812 Requirements for IP Version 4 Routers.
2037: * source quench could be a big problem under DoS attacks,
1.149 wiz 2038: * or if the underlying interface is rate-limited.
1.143 itojun 2039: */
2040: if (mcopy)
2041: m_freem(mcopy);
2042: return;
2043: #else
1.1 cgd 2044: type = ICMP_SOURCEQUENCH;
2045: code = 0;
2046: break;
1.143 itojun 2047: #endif
1.1 cgd 2048: }
1.220 christos 2049: icmp_error(mcopy, type, code, dest, destmtu);
1.44 thorpej 2050: }
2051:
2052: void
1.211 perry 2053: ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
2054: struct mbuf *m)
1.44 thorpej 2055: {
2056:
2057: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
2058: struct timeval tv;
2059:
2060: microtime(&tv);
1.244 christos 2061: *mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1.44 thorpej 2062: SCM_TIMESTAMP, SOL_SOCKET);
2063: if (*mp)
2064: mp = &(*mp)->m_next;
2065: }
2066: if (inp->inp_flags & INP_RECVDSTADDR) {
1.244 christos 2067: *mp = sbcreatecontrol((void *) &ip->ip_dst,
1.44 thorpej 2068: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
2069: if (*mp)
2070: mp = &(*mp)->m_next;
2071: }
2072: #ifdef notyet
2073: /*
2074: * XXX
2075: * Moving these out of udp_input() made them even more broken
2076: * than they already were.
2077: * - fenner@parc.xerox.com
2078: */
2079: /* options were tossed already */
2080: if (inp->inp_flags & INP_RECVOPTS) {
1.244 christos 2081: *mp = sbcreatecontrol((void *) opts_deleted_above,
1.44 thorpej 2082: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
2083: if (*mp)
2084: mp = &(*mp)->m_next;
2085: }
2086: /* ip_srcroute doesn't do what we want here, need to fix */
2087: if (inp->inp_flags & INP_RECVRETOPTS) {
1.244 christos 2088: *mp = sbcreatecontrol((void *) ip_srcroute(),
1.44 thorpej 2089: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
2090: if (*mp)
2091: mp = &(*mp)->m_next;
2092: }
2093: #endif
2094: if (inp->inp_flags & INP_RECVIF) {
2095: struct sockaddr_dl sdl;
2096:
1.252 dyoung 2097: sockaddr_dl_init(&sdl, sizeof(sdl),
2098: (m->m_pkthdr.rcvif != NULL)
2099: ? m->m_pkthdr.rcvif->if_index
2100: : 0,
2101: 0, NULL, 0, NULL, 0);
1.251 dyoung 2102: *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP);
1.44 thorpej 2103: if (*mp)
2104: mp = &(*mp)->m_next;
2105: }
1.13 mycroft 2106: }
2107:
1.189 atatat 2108: /*
1.228 elad 2109: * sysctl helper routine for net.inet.ip.forwsrcrt.
2110: */
2111: static int
2112: sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS)
2113: {
2114: int error, tmp;
2115: struct sysctlnode node;
2116:
2117: node = *rnode;
2118: tmp = ip_forwsrcrt;
2119: node.sysctl_data = &tmp;
2120: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2121: if (error || newp == NULL)
2122: return (error);
2123:
1.230 elad 2124: if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT,
1.232 elad 2125: 0, NULL, NULL, NULL))
1.228 elad 2126: return (EPERM);
2127:
2128: ip_forwsrcrt = tmp;
2129:
2130: return (0);
2131: }
2132:
2133: /*
1.189 atatat 2134: * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the
2135: * range of the new value and tweaks timers if it changes.
2136: */
2137: static int
2138: sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS)
1.13 mycroft 2139: {
1.189 atatat 2140: int error, tmp;
2141: struct sysctlnode node;
2142:
2143: node = *rnode;
2144: tmp = ip_mtudisc_timeout;
2145: node.sysctl_data = &tmp;
2146: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2147: if (error || newp == NULL)
2148: return (error);
2149: if (tmp < 0)
2150: return (EINVAL);
1.52 thorpej 2151:
1.273 matt 2152: mutex_enter(softnet_lock);
2153:
1.189 atatat 2154: ip_mtudisc_timeout = tmp;
2155: rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout);
2156:
1.273 matt 2157: mutex_exit(softnet_lock);
2158:
1.189 atatat 2159: return (0);
2160: }
1.54 lukem 2161:
1.65 matt 2162: #ifdef GATEWAY
1.189 atatat 2163: /*
1.247 liamjfoy 2164: * sysctl helper routine for net.inet.ip.maxflows.
1.189 atatat 2165: */
2166: static int
2167: sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS)
2168: {
1.273 matt 2169: int error;
1.67 thorpej 2170:
1.273 matt 2171: error = sysctl_lookup(SYSCTLFN_CALL(rnode));
2172: if (error || newp == NULL)
2173: return (error);
2174:
2175: mutex_enter(softnet_lock);
2176: KERNEL_LOCK(1, NULL);
1.212 perry 2177:
1.265 thorpej 2178: ipflow_prune();
1.273 matt 2179:
2180: KERNEL_UNLOCK_ONE(NULL);
2181: mutex_exit(softnet_lock);
1.144 martin 2182:
1.189 atatat 2183: return (0);
2184: }
1.248 liamjfoy 2185:
2186: static int
2187: sysctl_net_inet_ip_hashsize(SYSCTLFN_ARGS)
2188: {
2189: int error, tmp;
2190: struct sysctlnode node;
2191:
2192: node = *rnode;
2193: tmp = ip_hashsize;
2194: node.sysctl_data = &tmp;
2195: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2196: if (error || newp == NULL)
2197: return (error);
2198:
2199: if ((tmp & (tmp - 1)) == 0 && tmp != 0) {
2200: /*
2201: * Can only fail due to malloc()
2202: */
1.273 matt 2203: mutex_enter(softnet_lock);
2204: KERNEL_LOCK(1, NULL);
2205:
2206: error = ipflow_invalidate_all(tmp);
2207:
2208: KERNEL_UNLOCK_ONE(NULL);
2209: mutex_exit(softnet_lock);
2210:
1.248 liamjfoy 2211: } else {
2212: /*
2213: * EINVAL if not a power of 2
2214: */
1.273 matt 2215: error = EINVAL;
1.248 liamjfoy 2216: }
2217:
1.273 matt 2218: return error;
1.248 liamjfoy 2219: }
1.189 atatat 2220: #endif /* GATEWAY */
1.117 tron 2221:
1.266 thorpej 2222: static int
2223: sysctl_net_inet_ip_stats(SYSCTLFN_ARGS)
2224: {
2225:
1.271 thorpej 2226: return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS));
1.266 thorpej 2227: }
1.131 itojun 2228:
1.189 atatat 2229: SYSCTL_SETUP(sysctl_net_inet_ip_setup, "sysctl net.inet.ip subtree setup")
2230: {
2231: extern int subnetsarelocal, hostzeroisbroadcast;
1.180 jonathan 2232:
1.197 atatat 2233: sysctl_createv(clog, 0, NULL, NULL,
2234: CTLFLAG_PERMANENT,
1.189 atatat 2235: CTLTYPE_NODE, "net", NULL,
2236: NULL, 0, NULL, 0,
2237: CTL_NET, CTL_EOL);
1.197 atatat 2238: sysctl_createv(clog, 0, NULL, NULL,
2239: CTLFLAG_PERMANENT,
1.203 atatat 2240: CTLTYPE_NODE, "inet",
2241: SYSCTL_DESCR("PF_INET related settings"),
1.189 atatat 2242: NULL, 0, NULL, 0,
2243: CTL_NET, PF_INET, CTL_EOL);
1.197 atatat 2244: sysctl_createv(clog, 0, NULL, NULL,
2245: CTLFLAG_PERMANENT,
1.203 atatat 2246: CTLTYPE_NODE, "ip",
2247: SYSCTL_DESCR("IPv4 related settings"),
1.189 atatat 2248: NULL, 0, NULL, 0,
2249: CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL);
1.212 perry 2250:
1.197 atatat 2251: sysctl_createv(clog, 0, NULL, NULL,
2252: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2253: CTLTYPE_INT, "forwarding",
2254: SYSCTL_DESCR("Enable forwarding of INET datagrams"),
1.189 atatat 2255: NULL, 0, &ipforwarding, 0,
2256: CTL_NET, PF_INET, IPPROTO_IP,
2257: IPCTL_FORWARDING, CTL_EOL);
1.197 atatat 2258: sysctl_createv(clog, 0, NULL, NULL,
2259: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2260: CTLTYPE_INT, "redirect",
2261: SYSCTL_DESCR("Enable sending of ICMP redirect messages"),
1.189 atatat 2262: NULL, 0, &ipsendredirects, 0,
2263: CTL_NET, PF_INET, IPPROTO_IP,
2264: IPCTL_SENDREDIRECTS, CTL_EOL);
1.197 atatat 2265: sysctl_createv(clog, 0, NULL, NULL,
2266: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2267: CTLTYPE_INT, "ttl",
2268: SYSCTL_DESCR("Default TTL for an INET datagram"),
1.189 atatat 2269: NULL, 0, &ip_defttl, 0,
2270: CTL_NET, PF_INET, IPPROTO_IP,
2271: IPCTL_DEFTTL, CTL_EOL);
2272: #ifdef IPCTL_DEFMTU
1.197 atatat 2273: sysctl_createv(clog, 0, NULL, NULL,
2274: CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */,
1.203 atatat 2275: CTLTYPE_INT, "mtu",
2276: SYSCTL_DESCR("Default MTA for an INET route"),
1.189 atatat 2277: NULL, 0, &ip_mtu, 0,
2278: CTL_NET, PF_INET, IPPROTO_IP,
2279: IPCTL_DEFMTU, CTL_EOL);
2280: #endif /* IPCTL_DEFMTU */
1.197 atatat 2281: sysctl_createv(clog, 0, NULL, NULL,
1.228 elad 2282: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2283: CTLTYPE_INT, "forwsrcrt",
2284: SYSCTL_DESCR("Enable forwarding of source-routed "
2285: "datagrams"),
1.228 elad 2286: sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0,
1.189 atatat 2287: CTL_NET, PF_INET, IPPROTO_IP,
2288: IPCTL_FORWSRCRT, CTL_EOL);
1.197 atatat 2289: sysctl_createv(clog, 0, NULL, NULL,
2290: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2291: CTLTYPE_INT, "directed-broadcast",
2292: SYSCTL_DESCR("Enable forwarding of broadcast datagrams"),
1.189 atatat 2293: NULL, 0, &ip_directedbcast, 0,
2294: CTL_NET, PF_INET, IPPROTO_IP,
2295: IPCTL_DIRECTEDBCAST, CTL_EOL);
1.197 atatat 2296: sysctl_createv(clog, 0, NULL, NULL,
2297: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2298: CTLTYPE_INT, "allowsrcrt",
2299: SYSCTL_DESCR("Accept source-routed datagrams"),
1.189 atatat 2300: NULL, 0, &ip_allowsrcrt, 0,
2301: CTL_NET, PF_INET, IPPROTO_IP,
2302: IPCTL_ALLOWSRCRT, CTL_EOL);
1.197 atatat 2303: sysctl_createv(clog, 0, NULL, NULL,
2304: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2305: CTLTYPE_INT, "subnetsarelocal",
2306: SYSCTL_DESCR("Whether logical subnets are considered "
2307: "local"),
1.189 atatat 2308: NULL, 0, &subnetsarelocal, 0,
2309: CTL_NET, PF_INET, IPPROTO_IP,
2310: IPCTL_SUBNETSARELOCAL, CTL_EOL);
1.197 atatat 2311: sysctl_createv(clog, 0, NULL, NULL,
2312: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2313: CTLTYPE_INT, "mtudisc",
2314: SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"),
1.189 atatat 2315: NULL, 0, &ip_mtudisc, 0,
2316: CTL_NET, PF_INET, IPPROTO_IP,
2317: IPCTL_MTUDISC, CTL_EOL);
1.197 atatat 2318: sysctl_createv(clog, 0, NULL, NULL,
2319: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2320: CTLTYPE_INT, "anonportmin",
2321: SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1.189 atatat 2322: sysctl_net_inet_ip_ports, 0, &anonportmin, 0,
2323: CTL_NET, PF_INET, IPPROTO_IP,
2324: IPCTL_ANONPORTMIN, CTL_EOL);
1.197 atatat 2325: sysctl_createv(clog, 0, NULL, NULL,
2326: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2327: CTLTYPE_INT, "anonportmax",
2328: SYSCTL_DESCR("Highest ephemeral port number to assign"),
1.189 atatat 2329: sysctl_net_inet_ip_ports, 0, &anonportmax, 0,
2330: CTL_NET, PF_INET, IPPROTO_IP,
2331: IPCTL_ANONPORTMAX, CTL_EOL);
1.197 atatat 2332: sysctl_createv(clog, 0, NULL, NULL,
2333: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2334: CTLTYPE_INT, "mtudisctimeout",
2335: SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"),
1.189 atatat 2336: sysctl_net_inet_ip_pmtudto, 0, &ip_mtudisc_timeout, 0,
2337: CTL_NET, PF_INET, IPPROTO_IP,
2338: IPCTL_MTUDISCTIMEOUT, CTL_EOL);
2339: #ifdef GATEWAY
1.197 atatat 2340: sysctl_createv(clog, 0, NULL, NULL,
2341: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2342: CTLTYPE_INT, "maxflows",
2343: SYSCTL_DESCR("Number of flows for fast forwarding"),
1.189 atatat 2344: sysctl_net_inet_ip_maxflows, 0, &ip_maxflows, 0,
2345: CTL_NET, PF_INET, IPPROTO_IP,
2346: IPCTL_MAXFLOWS, CTL_EOL);
1.248 liamjfoy 2347: sysctl_createv(clog, 0, NULL, NULL,
2348: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2349: CTLTYPE_INT, "hashsize",
2350: SYSCTL_DESCR("Size of hash table for fast forwarding (IPv4)"),
2351: sysctl_net_inet_ip_hashsize, 0, &ip_hashsize, 0,
2352: CTL_NET, PF_INET, IPPROTO_IP,
2353: CTL_CREATE, CTL_EOL);
1.189 atatat 2354: #endif /* GATEWAY */
1.197 atatat 2355: sysctl_createv(clog, 0, NULL, NULL,
2356: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2357: CTLTYPE_INT, "hostzerobroadcast",
2358: SYSCTL_DESCR("All zeroes address is broadcast address"),
1.189 atatat 2359: NULL, 0, &hostzeroisbroadcast, 0,
2360: CTL_NET, PF_INET, IPPROTO_IP,
2361: IPCTL_HOSTZEROBROADCAST, CTL_EOL);
2362: #if NGIF > 0
1.197 atatat 2363: sysctl_createv(clog, 0, NULL, NULL,
2364: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2365: CTLTYPE_INT, "gifttl",
2366: SYSCTL_DESCR("Default TTL for a gif tunnel datagram"),
1.189 atatat 2367: NULL, 0, &ip_gif_ttl, 0,
2368: CTL_NET, PF_INET, IPPROTO_IP,
2369: IPCTL_GIF_TTL, CTL_EOL);
2370: #endif /* NGIF */
2371: #ifndef IPNOPRIVPORTS
1.197 atatat 2372: sysctl_createv(clog, 0, NULL, NULL,
2373: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2374: CTLTYPE_INT, "lowportmin",
2375: SYSCTL_DESCR("Lowest privileged ephemeral port number "
2376: "to assign"),
1.189 atatat 2377: sysctl_net_inet_ip_ports, 0, &lowportmin, 0,
2378: CTL_NET, PF_INET, IPPROTO_IP,
2379: IPCTL_LOWPORTMIN, CTL_EOL);
1.197 atatat 2380: sysctl_createv(clog, 0, NULL, NULL,
2381: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2382: CTLTYPE_INT, "lowportmax",
2383: SYSCTL_DESCR("Highest privileged ephemeral port number "
2384: "to assign"),
1.189 atatat 2385: sysctl_net_inet_ip_ports, 0, &lowportmax, 0,
2386: CTL_NET, PF_INET, IPPROTO_IP,
2387: IPCTL_LOWPORTMAX, CTL_EOL);
2388: #endif /* IPNOPRIVPORTS */
1.197 atatat 2389: sysctl_createv(clog, 0, NULL, NULL,
2390: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2391: CTLTYPE_INT, "maxfragpackets",
2392: SYSCTL_DESCR("Maximum number of fragments to retain for "
2393: "possible reassembly"),
1.189 atatat 2394: NULL, 0, &ip_maxfragpackets, 0,
2395: CTL_NET, PF_INET, IPPROTO_IP,
2396: IPCTL_MAXFRAGPACKETS, CTL_EOL);
2397: #if NGRE > 0
1.197 atatat 2398: sysctl_createv(clog, 0, NULL, NULL,
2399: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2400: CTLTYPE_INT, "grettl",
2401: SYSCTL_DESCR("Default TTL for a gre tunnel datagram"),
1.189 atatat 2402: NULL, 0, &ip_gre_ttl, 0,
2403: CTL_NET, PF_INET, IPPROTO_IP,
2404: IPCTL_GRE_TTL, CTL_EOL);
2405: #endif /* NGRE */
1.197 atatat 2406: sysctl_createv(clog, 0, NULL, NULL,
2407: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2408: CTLTYPE_INT, "checkinterface",
2409: SYSCTL_DESCR("Enable receive side of Strong ES model "
2410: "from RFC1122"),
1.189 atatat 2411: NULL, 0, &ip_checkinterface, 0,
2412: CTL_NET, PF_INET, IPPROTO_IP,
2413: IPCTL_CHECKINTERFACE, CTL_EOL);
1.197 atatat 2414: sysctl_createv(clog, 0, NULL, NULL,
2415: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2416: CTLTYPE_INT, "random_id",
2417: SYSCTL_DESCR("Assign random ip_id values"),
1.189 atatat 2418: NULL, 0, &ip_do_randomid, 0,
2419: CTL_NET, PF_INET, IPPROTO_IP,
2420: IPCTL_RANDOMID, CTL_EOL);
1.206 thorpej 2421: sysctl_createv(clog, 0, NULL, NULL,
2422: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2423: CTLTYPE_INT, "do_loopback_cksum",
2424: SYSCTL_DESCR("Perform IP checksum on loopback"),
2425: NULL, 0, &ip_do_loopback_cksum, 0,
2426: CTL_NET, PF_INET, IPPROTO_IP,
2427: IPCTL_LOOPBACKCKSUM, CTL_EOL);
1.219 elad 2428: sysctl_createv(clog, 0, NULL, NULL,
2429: CTLFLAG_PERMANENT,
2430: CTLTYPE_STRUCT, "stats",
2431: SYSCTL_DESCR("IP statistics"),
1.266 thorpej 2432: sysctl_net_inet_ip_stats, 0, NULL, 0,
1.219 elad 2433: CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS,
2434: CTL_EOL);
1.1 cgd 2435: }
1.266 thorpej 2436:
2437: void
2438: ip_statinc(u_int stat)
2439: {
2440:
2441: KASSERT(stat < IP_NSTATS);
2442: IP_STATINC(stat);
2443: }
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