Annotation of src/sys/netinet/ip_input.c, Revision 1.262.6.5
1.262.6.1 mjf 1: /* $NetBSD$ */
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>
1.262.6.1 mjf 94: __KERNEL_RCSID(0, "$NetBSD$");
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>
1.262.6.2 mjf 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>
1.262.6.2 mjf 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:
1.262.6.2 mjf 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:
1.262.6.4 mjf 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:
1.262.6.4 mjf 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);
1.262.6.2 mjf 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 */
1.262.6.2 mjf 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:
1.262.6.2 mjf 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);
1.262.6.2 mjf 474: if (m == NULL)
475: break;
1.89 itojun 476: ip_input(m);
477: }
1.262.6.2 mjf 478: KERNEL_UNLOCK_ONE(NULL);
479: mutex_exit(softnet_lock);
1.89 itojun 480: }
481:
482: /*
1.1 cgd 483: * Ip input routine. Checksum and byte swap header. If fragmented
484: * try to reassemble. Process options. Pass to next level.
485: */
1.8 mycroft 486: void
1.89 itojun 487: ip_input(struct mbuf *m)
1.1 cgd 488: {
1.109 augustss 489: struct ip *ip = NULL;
490: struct ipq *fp;
491: struct in_ifaddr *ia;
492: struct ifaddr *ifa;
1.25 cgd 493: struct ipqent *ipqe;
1.89 itojun 494: int hlen = 0, mff, len;
1.100 itojun 495: int downmatch;
1.165 christos 496: int checkif;
1.169 itojun 497: int srcrt = 0;
1.233 tls 498: int s;
1.190 jonathan 499: u_int hash;
1.173 jonathan 500: #ifdef FAST_IPSEC
501: struct m_tag *mtag;
502: struct tdb_ident *tdbi;
503: struct secpolicy *sp;
1.233 tls 504: int error;
1.173 jonathan 505: #endif /* FAST_IPSEC */
1.1 cgd 506:
1.164 matt 507: MCLAIM(m, &ip_rx_mowner);
1.1 cgd 508: #ifdef DIAGNOSTIC
509: if ((m->m_flags & M_PKTHDR) == 0)
510: panic("ipintr no HDR");
1.89 itojun 511: #endif
1.164 matt 512:
1.1 cgd 513: /*
514: * If no IP addresses have been set yet but the interfaces
515: * are receiving, can't do anything with incoming packets yet.
516: */
1.181 jonathan 517: if (TAILQ_FIRST(&in_ifaddrhead) == 0)
1.1 cgd 518: goto bad;
1.262.6.2 mjf 519: IP_STATINC(IP_STAT_TOTAL);
1.154 thorpej 520: /*
521: * If the IP header is not aligned, slurp it up into a new
522: * mbuf with space for link headers, in the event we forward
523: * it. Otherwise, if it is aligned, make sure the entire
524: * base IP header is in the first mbuf of the chain.
525: */
1.244 christos 526: if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
1.154 thorpej 527: if ((m = m_copyup(m, sizeof(struct ip),
528: (max_linkhdr + 3) & ~3)) == NULL) {
529: /* XXXJRT new stat, please */
1.262.6.2 mjf 530: IP_STATINC(IP_STAT_TOOSMALL);
1.154 thorpej 531: return;
532: }
533: } else if (__predict_false(m->m_len < sizeof (struct ip))) {
534: if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
1.262.6.2 mjf 535: IP_STATINC(IP_STAT_TOOSMALL);
1.154 thorpej 536: return;
537: }
1.1 cgd 538: }
539: ip = mtod(m, struct ip *);
1.13 mycroft 540: if (ip->ip_v != IPVERSION) {
1.262.6.2 mjf 541: IP_STATINC(IP_STAT_BADVERS);
1.13 mycroft 542: goto bad;
543: }
1.1 cgd 544: hlen = ip->ip_hl << 2;
545: if (hlen < sizeof(struct ip)) { /* minimum header length */
1.262.6.2 mjf 546: IP_STATINC(IP_STAT_BADHLEN);
1.1 cgd 547: goto bad;
548: }
549: if (hlen > m->m_len) {
550: if ((m = m_pullup(m, hlen)) == 0) {
1.262.6.2 mjf 551: IP_STATINC(IP_STAT_BADHLEN);
1.89 itojun 552: return;
1.1 cgd 553: }
554: ip = mtod(m, struct ip *);
555: }
1.98 thorpej 556:
1.85 hwr 557: /*
1.99 thorpej 558: * RFC1122: packets with a multicast source address are
1.98 thorpej 559: * not allowed.
1.85 hwr 560: */
561: if (IN_MULTICAST(ip->ip_src.s_addr)) {
1.262.6.2 mjf 562: IP_STATINC(IP_STAT_BADADDR);
1.85 hwr 563: goto bad;
1.129 itojun 564: }
565:
566: /* 127/8 must not appear on wire - RFC1122 */
567: if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
568: (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
1.130 itojun 569: if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
1.262.6.2 mjf 570: IP_STATINC(IP_STAT_BADADDR);
1.129 itojun 571: goto bad;
1.130 itojun 572: }
1.85 hwr 573: }
574:
1.135 thorpej 575: switch (m->m_pkthdr.csum_flags &
1.137 thorpej 576: ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) |
1.135 thorpej 577: M_CSUM_IPv4_BAD)) {
578: case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
579: INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
580: goto badcsum;
581:
582: case M_CSUM_IPv4:
583: /* Checksum was okay. */
584: INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
585: break;
586:
587: default:
1.206 thorpej 588: /*
589: * Must compute it ourselves. Maybe skip checksum on
590: * loopback interfaces.
591: */
592: if (__predict_true(!(m->m_pkthdr.rcvif->if_flags &
593: IFF_LOOPBACK) || ip_do_loopback_cksum)) {
594: INET_CSUM_COUNTER_INCR(&ip_swcsum);
595: if (in_cksum(m, hlen) != 0)
596: goto badcsum;
597: }
1.135 thorpej 598: break;
1.1 cgd 599: }
600:
1.121 thorpej 601: /* Retrieve the packet length. */
602: len = ntohs(ip->ip_len);
1.81 proff 603:
604: /*
605: * Check for additional length bogosity
606: */
1.84 proff 607: if (len < hlen) {
1.262.6.2 mjf 608: IP_STATINC(IP_STAT_BADLEN);
1.81 proff 609: goto bad;
610: }
1.1 cgd 611:
612: /*
613: * Check that the amount of data in the buffers
614: * is as at least much as the IP header would have us expect.
615: * Trim mbufs if longer than we expect.
616: * Drop packet if shorter than we expect.
617: */
1.35 mycroft 618: if (m->m_pkthdr.len < len) {
1.262.6.2 mjf 619: IP_STATINC(IP_STAT_TOOSHORT);
1.1 cgd 620: goto bad;
621: }
1.35 mycroft 622: if (m->m_pkthdr.len > len) {
1.1 cgd 623: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 624: m->m_len = len;
625: m->m_pkthdr.len = len;
1.1 cgd 626: } else
1.35 mycroft 627: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 628: }
629:
1.193 scw 630: #if defined(IPSEC)
1.149 wiz 631: /* ipflow (IP fast forwarding) is not compatible with IPsec. */
1.94 itojun 632: m->m_flags &= ~M_CANFASTFWD;
633: #else
1.64 thorpej 634: /*
635: * Assume that we can create a fast-forward IP flow entry
636: * based on this packet.
637: */
638: m->m_flags |= M_CANFASTFWD;
1.94 itojun 639: #endif
1.64 thorpej 640:
1.36 mrg 641: #ifdef PFIL_HOOKS
1.33 mrg 642: /*
1.64 thorpej 643: * Run through list of hooks for input packets. If there are any
644: * filters which require that additional packets in the flow are
645: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
646: * Note that filters must _never_ set this flag, as another filter
647: * in the list may have previously cleared it.
1.33 mrg 648: */
1.127 itojun 649: /*
650: * let ipfilter look at packet on the wire,
651: * not the decapsulated packet.
652: */
653: #ifdef IPSEC
1.136 itojun 654: if (!ipsec_getnhist(m))
1.186 scw 655: #elif defined(FAST_IPSEC)
656: if (!ipsec_indone(m))
1.127 itojun 657: #else
658: if (1)
659: #endif
660: {
1.169 itojun 661: struct in_addr odst;
662:
663: odst = ip->ip_dst;
1.127 itojun 664: if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif,
1.168 itojun 665: PFIL_IN) != 0)
666: return;
1.127 itojun 667: if (m == NULL)
668: return;
669: ip = mtod(m, struct ip *);
1.142 darrenr 670: hlen = ip->ip_hl << 2;
1.205 darrenr 671: /*
672: * XXX The setting of "srcrt" here is to prevent ip_forward()
673: * from generating ICMP redirects for packets that have
674: * been redirected by a hook back out on to the same LAN that
675: * they came from and is not an indication that the packet
676: * is being inffluenced by source routing options. This
677: * allows things like
678: * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp"
679: * where tlp0 is both on the 1.1.1.0/24 network and is the
680: * default route for hosts on 1.1.1.0/24. Of course this
681: * also requires a "map tlp0 ..." to complete the story.
682: * One might argue whether or not this kind of network config.
1.212 perry 683: * should be supported in this manner...
1.205 darrenr 684: */
1.169 itojun 685: srcrt = (odst.s_addr != ip->ip_dst.s_addr);
1.127 itojun 686: }
1.36 mrg 687: #endif /* PFIL_HOOKS */
1.123 thorpej 688:
689: #ifdef ALTQ
690: /* XXX Temporary until ALTQ is changed to use a pfil hook */
691: if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) {
692: /* packet dropped by traffic conditioner */
693: return;
694: }
695: #endif
1.121 thorpej 696:
697: /*
1.1 cgd 698: * Process options and, if not destined for us,
699: * ship it on. ip_dooptions returns 1 when an
700: * error was detected (causing an icmp message
701: * to be sent and the original packet to be freed).
702: */
703: ip_nhops = 0; /* for source routed packets */
704: if (hlen > sizeof (struct ip) && ip_dooptions(m))
1.89 itojun 705: return;
1.1 cgd 706:
707: /*
1.165 christos 708: * Enable a consistency check between the destination address
709: * and the arrival interface for a unicast packet (the RFC 1122
710: * strong ES model) if IP forwarding is disabled and the packet
711: * is not locally generated.
712: *
713: * XXX - Checking also should be disabled if the destination
714: * address is ipnat'ed to a different interface.
715: *
716: * XXX - Checking is incompatible with IP aliases added
717: * to the loopback interface instead of the interface where
718: * the packets are received.
719: *
720: * XXX - We need to add a per ifaddr flag for this so that
721: * we get finer grain control.
722: */
723: checkif = ip_checkinterface && (ipforwarding == 0) &&
724: (m->m_pkthdr.rcvif != NULL) &&
725: ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0);
726:
727: /*
1.1 cgd 728: * Check our list of addresses, to see if the packet is for us.
1.100 itojun 729: *
730: * Traditional 4.4BSD did not consult IFF_UP at all.
731: * The behavior here is to treat addresses on !IFF_UP interface
732: * as not mine.
1.1 cgd 733: */
1.100 itojun 734: downmatch = 0;
1.140 matt 735: LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) {
1.97 itojun 736: if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
1.165 christos 737: if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif)
738: continue;
1.97 itojun 739: if ((ia->ia_ifp->if_flags & IFF_UP) != 0)
740: break;
1.100 itojun 741: else
742: downmatch++;
1.97 itojun 743: }
744: }
1.86 thorpej 745: if (ia != NULL)
746: goto ours;
1.225 christos 747: if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
1.209 matt 748: IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) {
1.140 matt 749: if (ifa->ifa_addr->sa_family != AF_INET)
750: continue;
1.57 tls 751: ia = ifatoia(ifa);
1.35 mycroft 752: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
753: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 754: /*
755: * Look for all-0's host part (old broadcast addr),
756: * either for subnet or net.
757: */
758: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 759: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 760: goto ours;
1.57 tls 761: /*
762: * An interface with IP address zero accepts
763: * all packets that arrive on that interface.
764: */
765: if (in_nullhost(ia->ia_addr.sin_addr))
766: goto ours;
1.1 cgd 767: }
768: }
1.18 mycroft 769: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 770: struct in_multi *inm;
771: #ifdef MROUTING
772: extern struct socket *ip_mrouter;
1.10 brezak 773:
1.4 hpeyerl 774: if (ip_mrouter) {
775: /*
776: * If we are acting as a multicast router, all
777: * incoming multicast packets are passed to the
778: * kernel-level multicast forwarding function.
779: * The packet is returned (relatively) intact; if
780: * ip_mforward() returns a non-zero value, the packet
781: * must be discarded, else it may be accepted below.
782: *
783: * (The IP ident field is put in the same byte order
784: * as expected when ip_mforward() is called from
785: * ip_output().)
786: */
1.13 mycroft 787: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
1.262.6.2 mjf 788: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 789: m_freem(m);
1.89 itojun 790: return;
1.4 hpeyerl 791: }
792:
793: /*
794: * The process-level routing demon needs to receive
795: * all multicast IGMP packets, whether or not this
796: * host belongs to their destination groups.
797: */
798: if (ip->ip_p == IPPROTO_IGMP)
799: goto ours;
1.262.6.2 mjf 800: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 801: }
802: #endif
803: /*
804: * See if we belong to the destination multicast group on the
805: * arrival interface.
806: */
807: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
808: if (inm == NULL) {
1.262.6.2 mjf 809: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 810: m_freem(m);
1.89 itojun 811: return;
1.4 hpeyerl 812: }
813: goto ours;
814: }
1.19 mycroft 815: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 816: in_nullhost(ip->ip_dst))
1.1 cgd 817: goto ours;
818:
819: /*
820: * Not for us; forward if possible and desirable.
821: */
822: if (ipforwarding == 0) {
1.262.6.2 mjf 823: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 824: m_freem(m);
1.100 itojun 825: } else {
826: /*
827: * If ip_dst matched any of my address on !IFF_UP interface,
828: * and there's no IFF_UP interface that matches ip_dst,
829: * send icmp unreach. Forwarding it will result in in-kernel
830: * forwarding loop till TTL goes to 0.
831: */
832: if (downmatch) {
833: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
1.262.6.2 mjf 834: IP_STATINC(IP_STAT_CANTFORWARD);
1.100 itojun 835: return;
836: }
1.145 itojun 837: #ifdef IPSEC
838: if (ipsec4_in_reject(m, NULL)) {
1.262.6.2 mjf 839: IPSEC_STATINC(IPSEC_STAT_IN_POLVIO);
1.145 itojun 840: goto bad;
841: }
842: #endif
1.173 jonathan 843: #ifdef FAST_IPSEC
844: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
845: s = splsoftnet();
846: if (mtag != NULL) {
847: tdbi = (struct tdb_ident *)(mtag + 1);
848: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
849: } else {
850: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
1.212 perry 851: IP_FORWARDING, &error);
1.173 jonathan 852: }
853: if (sp == NULL) { /* NB: can happen if error */
854: splx(s);
855: /*XXX error stat???*/
856: DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/
857: goto bad;
858: }
859:
860: /*
861: * Check security policy against packet attributes.
862: */
863: error = ipsec_in_reject(sp, m);
864: KEY_FREESP(&sp);
865: splx(s);
866: if (error) {
1.262.6.2 mjf 867: IP_STATINC(IP_STAT_CANTFORWARD);
1.173 jonathan 868: goto bad;
1.193 scw 869: }
870:
871: /*
872: * Peek at the outbound SP for this packet to determine if
873: * it's a Fast Forward candidate.
874: */
875: mtag = m_tag_find(m, PACKET_TAG_IPSEC_PENDING_TDB, NULL);
876: if (mtag != NULL)
877: m->m_flags &= ~M_CANFASTFWD;
878: else {
879: s = splsoftnet();
880: sp = ipsec4_checkpolicy(m, IPSEC_DIR_OUTBOUND,
881: (IP_FORWARDING |
882: (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
883: &error, NULL);
884: if (sp != NULL) {
885: m->m_flags &= ~M_CANFASTFWD;
886: KEY_FREESP(&sp);
887: }
888: splx(s);
1.173 jonathan 889: }
890: #endif /* FAST_IPSEC */
1.145 itojun 891:
1.169 itojun 892: ip_forward(m, srcrt);
1.100 itojun 893: }
1.89 itojun 894: return;
1.1 cgd 895:
896: ours:
897: /*
898: * If offset or IP_MF are set, must reassemble.
899: * Otherwise, nothing need be done.
900: * (We could look in the reassembly queue to see
901: * if the packet was previously fragmented,
902: * but it's not worth the time; just let them time out.)
903: */
1.155 itojun 904: if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
1.258 matt 905: uint16_t off;
906: /*
907: * Prevent TCP blind data attacks by not allowing non-initial
908: * fragments to start at less than 68 bytes (minimal fragment
1.259 matt 909: * size) and making sure the first fragment is at least 68
910: * bytes.
1.258 matt 911: */
1.260 matt 912: off = (ntohs(ip->ip_off) & IP_OFFMASK) << 3;
1.259 matt 913: if ((off > 0 ? off + hlen : len) < IP_MINFRAGSIZE - 1) {
1.262.6.2 mjf 914: IP_STATINC(IP_STAT_BADFRAGS);
1.258 matt 915: goto bad;
916: }
1.1 cgd 917: /*
918: * Look for queue of fragments
919: * of this datagram.
920: */
1.75 thorpej 921: IPQ_LOCK();
1.190 jonathan 922: hash = IPREASS_HASH(ip->ip_src.s_addr, ip->ip_id);
1.250 dyoung 923: LIST_FOREACH(fp, &ipq[hash], ipq_q) {
1.1 cgd 924: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 925: in_hosteq(ip->ip_src, fp->ipq_src) &&
926: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.260 matt 927: ip->ip_p == fp->ipq_p) {
928: /*
929: * Make sure the TOS is matches previous
930: * fragments.
931: */
932: if (ip->ip_tos != fp->ipq_tos) {
1.262.6.2 mjf 933: IP_STATINC(IP_STAT_BADFRAGS);
1.262.6.5! mjf 934: IPQ_UNLOCK();
1.260 matt 935: goto bad;
936: }
1.1 cgd 937: goto found;
1.260 matt 938: }
1.190 jonathan 939: }
1.1 cgd 940: fp = 0;
941: found:
942:
943: /*
944: * Adjust ip_len to not reflect header,
1.25 cgd 945: * set ipqe_mff if more fragments are expected,
1.1 cgd 946: * convert offset of this to bytes.
947: */
1.155 itojun 948: ip->ip_len = htons(ntohs(ip->ip_len) - hlen);
949: mff = (ip->ip_off & htons(IP_MF)) != 0;
1.25 cgd 950: if (mff) {
1.16 cgd 951: /*
952: * Make sure that fragments have a data length
953: * that's a non-zero multiple of 8 bytes.
954: */
1.155 itojun 955: if (ntohs(ip->ip_len) == 0 ||
956: (ntohs(ip->ip_len) & 0x7) != 0) {
1.262.6.2 mjf 957: IP_STATINC(IP_STAT_BADFRAGS);
1.75 thorpej 958: IPQ_UNLOCK();
1.16 cgd 959: goto bad;
960: }
961: }
1.155 itojun 962: ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3);
1.1 cgd 963:
964: /*
965: * If datagram marked as having more fragments
966: * or if this is not the first fragment,
967: * attempt reassembly; if it succeeds, proceed.
968: */
1.155 itojun 969: if (mff || ip->ip_off != htons(0)) {
1.262.6.2 mjf 970: IP_STATINC(IP_STAT_FRAGMENTS);
1.233 tls 971: s = splvm();
1.72 thorpej 972: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.233 tls 973: splx(s);
1.25 cgd 974: if (ipqe == NULL) {
1.262.6.2 mjf 975: IP_STATINC(IP_STAT_RCVMEMDROP);
1.75 thorpej 976: IPQ_UNLOCK();
1.25 cgd 977: goto bad;
978: }
979: ipqe->ipqe_mff = mff;
1.50 thorpej 980: ipqe->ipqe_m = m;
1.25 cgd 981: ipqe->ipqe_ip = ip;
1.190 jonathan 982: m = ip_reass(ipqe, fp, &ipq[hash]);
1.75 thorpej 983: if (m == 0) {
984: IPQ_UNLOCK();
1.89 itojun 985: return;
1.75 thorpej 986: }
1.262.6.2 mjf 987: IP_STATINC(IP_STAT_REASSEMBLED);
1.50 thorpej 988: ip = mtod(m, struct ip *);
1.74 thorpej 989: hlen = ip->ip_hl << 2;
1.155 itojun 990: ip->ip_len = htons(ntohs(ip->ip_len) + hlen);
1.1 cgd 991: } else
992: if (fp)
993: ip_freef(fp);
1.75 thorpej 994: IPQ_UNLOCK();
1.79 mycroft 995: }
1.128 itojun 996:
1.173 jonathan 997: #if defined(IPSEC)
1.128 itojun 998: /*
999: * enforce IPsec policy checking if we are seeing last header.
1000: * note that we do not visit this with protocols with pcb layer
1001: * code - like udp/tcp/raw ip.
1002: */
1003: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 &&
1004: ipsec4_in_reject(m, NULL)) {
1.262.6.2 mjf 1005: IPSEC_STATINC(IPSEC_STAT_IN_POLVIO);
1.128 itojun 1006: goto bad;
1007: }
1008: #endif
1.226 liamjfoy 1009: #ifdef FAST_IPSEC
1.173 jonathan 1010: /*
1011: * enforce IPsec policy checking if we are seeing last header.
1012: * note that we do not visit this with protocols with pcb layer
1013: * code - like udp/tcp/raw ip.
1014: */
1015: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) {
1016: /*
1017: * Check if the packet has already had IPsec processing
1018: * done. If so, then just pass it along. This tag gets
1019: * set during AH, ESP, etc. input handling, before the
1020: * packet is returned to the ip input queue for delivery.
1.212 perry 1021: */
1.173 jonathan 1022: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
1023: s = splsoftnet();
1024: if (mtag != NULL) {
1025: tdbi = (struct tdb_ident *)(mtag + 1);
1026: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
1027: } else {
1028: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
1.212 perry 1029: IP_FORWARDING, &error);
1.173 jonathan 1030: }
1031: if (sp != NULL) {
1032: /*
1033: * Check security policy against packet attributes.
1034: */
1035: error = ipsec_in_reject(sp, m);
1036: KEY_FREESP(&sp);
1037: } else {
1038: /* XXX error stat??? */
1039: error = EINVAL;
1040: DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/
1041: }
1042: splx(s);
1043: if (error)
1044: goto bad;
1045: }
1046: #endif /* FAST_IPSEC */
1.1 cgd 1047:
1048: /*
1049: * Switch out to protocol's input routine.
1050: */
1.82 aidan 1051: #if IFA_STATS
1.122 itojun 1052: if (ia && ip)
1.155 itojun 1053: ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
1.82 aidan 1054: #endif
1.262.6.2 mjf 1055: IP_STATINC(IP_STAT_DELIVERED);
1.89 itojun 1056: {
1057: int off = hlen, nh = ip->ip_p;
1058:
1059: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
1060: return;
1061: }
1.1 cgd 1062: bad:
1063: m_freem(m);
1.135 thorpej 1064: return;
1065:
1066: badcsum:
1.262.6.2 mjf 1067: IP_STATINC(IP_STAT_BADSUM);
1.135 thorpej 1068: m_freem(m);
1.1 cgd 1069: }
1070:
1071: /*
1072: * Take incoming datagram fragment and try to
1073: * reassemble it into whole datagram. If a chain for
1074: * reassembly of this datagram already exists, then it
1075: * is given as fp; otherwise have to make a chain.
1076: */
1.50 thorpej 1077: struct mbuf *
1.211 perry 1078: ip_reass(struct ipqent *ipqe, struct ipq *fp, struct ipqhead *ipqhead)
1.1 cgd 1079: {
1.109 augustss 1080: struct mbuf *m = ipqe->ipqe_m;
1081: struct ipqent *nq, *p, *q;
1.25 cgd 1082: struct ip *ip;
1.1 cgd 1083: struct mbuf *t;
1.25 cgd 1084: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.233 tls 1085: int i, next, s;
1.1 cgd 1086:
1.75 thorpej 1087: IPQ_LOCK_CHECK();
1088:
1.1 cgd 1089: /*
1090: * Presence of header sizes in mbufs
1091: * would confuse code below.
1092: */
1093: m->m_data += hlen;
1094: m->m_len -= hlen;
1095:
1.194 jonathan 1096: #ifdef notyet
1097: /* make sure fragment limit is up-to-date */
1098: CHECK_NMBCLUSTER_PARAMS();
1099:
1100: /* If we have too many fragments, drop the older half. */
1101: if (ip_nfrags >= ip_maxfrags)
1102: ip_reass_drophalf(void);
1103: #endif
1104:
1.1 cgd 1105: /*
1.192 jonathan 1106: * We are about to add a fragment; increment frag count.
1107: */
1108: ip_nfrags++;
1.212 perry 1109:
1.192 jonathan 1110: /*
1.1 cgd 1111: * If first fragment to arrive, create a reassembly queue.
1112: */
1113: if (fp == 0) {
1.131 itojun 1114: /*
1115: * Enforce upper bound on number of fragmented packets
1116: * for which we attempt reassembly;
1117: * If maxfrag is 0, never accept fragments.
1118: * If maxfrag is -1, accept all fragments without limitation.
1119: */
1120: if (ip_maxfragpackets < 0)
1121: ;
1122: else if (ip_nfragpackets >= ip_maxfragpackets)
1123: goto dropfrag;
1124: ip_nfragpackets++;
1.262.6.5! mjf 1125: fp = malloc(sizeof (struct ipq), M_FTABLE, M_NOWAIT);
1.50 thorpej 1126: if (fp == NULL)
1.1 cgd 1127: goto dropfrag;
1.190 jonathan 1128: LIST_INSERT_HEAD(ipqhead, fp, ipq_q);
1.192 jonathan 1129: fp->ipq_nfrags = 1;
1.1 cgd 1130: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 1131: fp->ipq_p = ipqe->ipqe_ip->ip_p;
1132: fp->ipq_id = ipqe->ipqe_ip->ip_id;
1.260 matt 1133: fp->ipq_tos = ipqe->ipqe_ip->ip_tos;
1.148 matt 1134: TAILQ_INIT(&fp->ipq_fragq);
1.25 cgd 1135: fp->ipq_src = ipqe->ipqe_ip->ip_src;
1136: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
1137: p = NULL;
1.1 cgd 1138: goto insert;
1.192 jonathan 1139: } else {
1140: fp->ipq_nfrags++;
1.1 cgd 1141: }
1142:
1143: /*
1144: * Find a segment which begins after this one does.
1145: */
1.148 matt 1146: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1147: p = q, q = TAILQ_NEXT(q, ipqe_q))
1.155 itojun 1148: if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off))
1.1 cgd 1149: break;
1150:
1151: /*
1152: * If there is a preceding segment, it may provide some of
1153: * our data already. If so, drop the data from the incoming
1154: * segment. If it provides all of our data, drop us.
1155: */
1.25 cgd 1156: if (p != NULL) {
1.155 itojun 1157: i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) -
1158: ntohs(ipqe->ipqe_ip->ip_off);
1.1 cgd 1159: if (i > 0) {
1.155 itojun 1160: if (i >= ntohs(ipqe->ipqe_ip->ip_len))
1.1 cgd 1161: goto dropfrag;
1.50 thorpej 1162: m_adj(ipqe->ipqe_m, i);
1.155 itojun 1163: ipqe->ipqe_ip->ip_off =
1164: htons(ntohs(ipqe->ipqe_ip->ip_off) + i);
1165: ipqe->ipqe_ip->ip_len =
1166: htons(ntohs(ipqe->ipqe_ip->ip_len) - i);
1.1 cgd 1167: }
1168: }
1169:
1170: /*
1171: * While we overlap succeeding segments trim them or,
1172: * if they are completely covered, dequeue them.
1173: */
1.155 itojun 1174: for (; q != NULL &&
1175: ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) >
1176: ntohs(q->ipqe_ip->ip_off); q = nq) {
1177: i = (ntohs(ipqe->ipqe_ip->ip_off) +
1178: ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off);
1179: if (i < ntohs(q->ipqe_ip->ip_len)) {
1180: q->ipqe_ip->ip_len =
1181: htons(ntohs(q->ipqe_ip->ip_len) - i);
1182: q->ipqe_ip->ip_off =
1183: htons(ntohs(q->ipqe_ip->ip_off) + i);
1.50 thorpej 1184: m_adj(q->ipqe_m, i);
1.1 cgd 1185: break;
1186: }
1.148 matt 1187: nq = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1188: m_freem(q->ipqe_m);
1.148 matt 1189: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.233 tls 1190: s = splvm();
1.72 thorpej 1191: pool_put(&ipqent_pool, q);
1.233 tls 1192: splx(s);
1.192 jonathan 1193: fp->ipq_nfrags--;
1194: ip_nfrags--;
1.1 cgd 1195: }
1196:
1197: insert:
1198: /*
1199: * Stick new segment in its place;
1200: * check for complete reassembly.
1201: */
1.25 cgd 1202: if (p == NULL) {
1.148 matt 1203: TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
1.25 cgd 1204: } else {
1.148 matt 1205: TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q);
1.25 cgd 1206: }
1.1 cgd 1207: next = 0;
1.148 matt 1208: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1209: p = q, q = TAILQ_NEXT(q, ipqe_q)) {
1.155 itojun 1210: if (ntohs(q->ipqe_ip->ip_off) != next)
1.1 cgd 1211: return (0);
1.155 itojun 1212: next += ntohs(q->ipqe_ip->ip_len);
1.1 cgd 1213: }
1.25 cgd 1214: if (p->ipqe_mff)
1.1 cgd 1215: return (0);
1216:
1217: /*
1.41 thorpej 1218: * Reassembly is complete. Check for a bogus message size and
1219: * concatenate fragments.
1.1 cgd 1220: */
1.148 matt 1221: q = TAILQ_FIRST(&fp->ipq_fragq);
1.25 cgd 1222: ip = q->ipqe_ip;
1.41 thorpej 1223: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
1.262.6.2 mjf 1224: IP_STATINC(IP_STAT_TOOLONG);
1.41 thorpej 1225: ip_freef(fp);
1226: return (0);
1227: }
1.50 thorpej 1228: m = q->ipqe_m;
1.1 cgd 1229: t = m->m_next;
1230: m->m_next = 0;
1231: m_cat(m, t);
1.148 matt 1232: nq = TAILQ_NEXT(q, ipqe_q);
1.233 tls 1233: s = splvm();
1.72 thorpej 1234: pool_put(&ipqent_pool, q);
1.233 tls 1235: splx(s);
1.25 cgd 1236: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 1237: t = q->ipqe_m;
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);
1.1 cgd 1242: m_cat(m, t);
1243: }
1.192 jonathan 1244: ip_nfrags -= fp->ipq_nfrags;
1.1 cgd 1245:
1246: /*
1247: * Create header for new ip packet by
1248: * modifying header of first packet;
1249: * dequeue and discard fragment reassembly header.
1250: * Make header visible.
1251: */
1.155 itojun 1252: ip->ip_len = htons(next);
1.25 cgd 1253: ip->ip_src = fp->ipq_src;
1254: ip->ip_dst = fp->ipq_dst;
1255: LIST_REMOVE(fp, ipq_q);
1.262.6.5! mjf 1256: free(fp, M_FTABLE);
1.131 itojun 1257: ip_nfragpackets--;
1.1 cgd 1258: m->m_len += (ip->ip_hl << 2);
1259: m->m_data -= (ip->ip_hl << 2);
1260: /* some debugging cruft by sklower, below, will go away soon */
1261: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
1.109 augustss 1262: int plen = 0;
1.50 thorpej 1263: for (t = m; t; t = t->m_next)
1264: plen += t->m_len;
1265: m->m_pkthdr.len = plen;
1.213 yamt 1266: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1267: }
1.50 thorpej 1268: return (m);
1.1 cgd 1269:
1270: dropfrag:
1.192 jonathan 1271: if (fp != 0)
1272: fp->ipq_nfrags--;
1273: ip_nfrags--;
1.262.6.2 mjf 1274: IP_STATINC(IP_STAT_FRAGDROPPED);
1.1 cgd 1275: m_freem(m);
1.233 tls 1276: s = splvm();
1.72 thorpej 1277: pool_put(&ipqent_pool, ipqe);
1.233 tls 1278: splx(s);
1.1 cgd 1279: return (0);
1280: }
1281:
1282: /*
1283: * Free a fragment reassembly header and all
1284: * associated datagrams.
1285: */
1.8 mycroft 1286: void
1.211 perry 1287: ip_freef(struct ipq *fp)
1.1 cgd 1288: {
1.109 augustss 1289: struct ipqent *q, *p;
1.192 jonathan 1290: u_int nfrags = 0;
1.233 tls 1291: int s;
1.1 cgd 1292:
1.75 thorpej 1293: IPQ_LOCK_CHECK();
1294:
1.148 matt 1295: for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) {
1296: p = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1297: m_freem(q->ipqe_m);
1.192 jonathan 1298: nfrags++;
1.148 matt 1299: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.233 tls 1300: s = splvm();
1.72 thorpej 1301: pool_put(&ipqent_pool, q);
1.233 tls 1302: splx(s);
1.1 cgd 1303: }
1.192 jonathan 1304:
1305: if (nfrags != fp->ipq_nfrags)
1306: printf("ip_freef: nfrags %d != %d\n", fp->ipq_nfrags, nfrags);
1307: ip_nfrags -= nfrags;
1.25 cgd 1308: LIST_REMOVE(fp, ipq_q);
1.262.6.5! mjf 1309: free(fp, M_FTABLE);
1.131 itojun 1310: ip_nfragpackets--;
1.1 cgd 1311: }
1312:
1313: /*
1.194 jonathan 1314: * IP reassembly TTL machinery for multiplicative drop.
1315: */
1316: static u_int fragttl_histo[(IPFRAGTTL+1)];
1317:
1318:
1319: /*
1320: * Decrement TTL of all reasembly queue entries by `ticks'.
1321: * Count number of distinct fragments (as opposed to partial, fragmented
1322: * datagrams) in the reassembly queue. While we traverse the entire
1323: * reassembly queue, compute and return the median TTL over all fragments.
1324: */
1325: static u_int
1326: ip_reass_ttl_decr(u_int ticks)
1327: {
1.198 matt 1328: u_int nfrags, median, dropfraction, keepfraction;
1.194 jonathan 1329: struct ipq *fp, *nfp;
1.198 matt 1330: int i;
1.212 perry 1331:
1.194 jonathan 1332: nfrags = 0;
1333: memset(fragttl_histo, 0, sizeof fragttl_histo);
1.212 perry 1334:
1.194 jonathan 1335: for (i = 0; i < IPREASS_NHASH; i++) {
1336: for (fp = LIST_FIRST(&ipq[i]); fp != NULL; fp = nfp) {
1337: fp->ipq_ttl = ((fp->ipq_ttl <= ticks) ?
1338: 0 : fp->ipq_ttl - ticks);
1339: nfp = LIST_NEXT(fp, ipq_q);
1340: if (fp->ipq_ttl == 0) {
1.262.6.2 mjf 1341: IP_STATINC(IP_STAT_FRAGTIMEOUT);
1.194 jonathan 1342: ip_freef(fp);
1343: } else {
1344: nfrags += fp->ipq_nfrags;
1345: fragttl_histo[fp->ipq_ttl] += fp->ipq_nfrags;
1346: }
1347: }
1348: }
1349:
1350: KASSERT(ip_nfrags == nfrags);
1351:
1352: /* Find median (or other drop fraction) in histogram. */
1353: dropfraction = (ip_nfrags / 2);
1354: keepfraction = ip_nfrags - dropfraction;
1355: for (i = IPFRAGTTL, median = 0; i >= 0; i--) {
1356: median += fragttl_histo[i];
1357: if (median >= keepfraction)
1358: break;
1359: }
1360:
1361: /* Return TTL of median (or other fraction). */
1362: return (u_int)i;
1363: }
1364:
1365: void
1366: ip_reass_drophalf(void)
1367: {
1368:
1369: u_int median_ticks;
1370: /*
1371: * Compute median TTL of all fragments, and count frags
1372: * with that TTL or lower (roughly half of all fragments).
1373: */
1374: median_ticks = ip_reass_ttl_decr(0);
1375:
1376: /* Drop half. */
1377: median_ticks = ip_reass_ttl_decr(median_ticks);
1378:
1379: }
1380:
1381: /*
1.1 cgd 1382: * IP timer processing;
1383: * if a timer expires on a reassembly
1384: * queue, discard it.
1385: */
1.8 mycroft 1386: void
1.211 perry 1387: ip_slowtimo(void)
1.1 cgd 1388: {
1.191 jonathan 1389: static u_int dropscanidx = 0;
1390: u_int i;
1.194 jonathan 1391: u_int median_ttl;
1.262.6.2 mjf 1392:
1393: mutex_enter(softnet_lock);
1394: KERNEL_LOCK(1, NULL);
1.1 cgd 1395:
1.75 thorpej 1396: IPQ_LOCK();
1.194 jonathan 1397:
1398: /* Age TTL of all fragments by 1 tick .*/
1399: median_ttl = ip_reass_ttl_decr(1);
1400:
1401: /* make sure fragment limit is up-to-date */
1402: CHECK_NMBCLUSTER_PARAMS();
1403:
1404: /* If we have too many fragments, drop the older half. */
1405: if (ip_nfrags > ip_maxfrags)
1406: ip_reass_ttl_decr(median_ttl);
1407:
1.131 itojun 1408: /*
1.194 jonathan 1409: * If we are over the maximum number of fragmented packets
1.131 itojun 1410: * (due to the limit being lowered), drain off
1.190 jonathan 1411: * enough to get down to the new limit. Start draining
1412: * from the reassembly hashqueue most recently drained.
1.131 itojun 1413: */
1414: if (ip_maxfragpackets < 0)
1415: ;
1416: else {
1.190 jonathan 1417: int wrapped = 0;
1418:
1419: i = dropscanidx;
1420: while (ip_nfragpackets > ip_maxfragpackets && wrapped == 0) {
1421: while (LIST_FIRST(&ipq[i]) != NULL)
1422: ip_freef(LIST_FIRST(&ipq[i]));
1423: if (++i >= IPREASS_NHASH) {
1424: i = 0;
1425: }
1426: /*
1427: * Dont scan forever even if fragment counters are
1428: * wrong: stop after scanning entire reassembly queue.
1429: */
1430: if (i == dropscanidx)
1431: wrapped = 1;
1432: }
1433: dropscanidx = i;
1.131 itojun 1434: }
1.75 thorpej 1435: IPQ_UNLOCK();
1.262.6.2 mjf 1436:
1437: KERNEL_UNLOCK_ONE(NULL);
1438: mutex_exit(softnet_lock);
1.1 cgd 1439: }
1440:
1441: /*
1.262.6.2 mjf 1442: * Drain off all datagram fragments. Don't acquire softnet_lock as
1443: * can be called from hardware interrupt context.
1.1 cgd 1444: */
1.8 mycroft 1445: void
1.211 perry 1446: ip_drain(void)
1.1 cgd 1447: {
1448:
1.262.6.2 mjf 1449: KERNEL_LOCK(1, NULL);
1450:
1.75 thorpej 1451: /*
1452: * We may be called from a device's interrupt context. If
1453: * the ipq is already busy, just bail out now.
1454: */
1.262.6.2 mjf 1455: if (ipq_lock_try() != 0) {
1456: /*
1457: * Drop half the total fragments now. If more mbufs are
1458: * needed, we will be called again soon.
1459: */
1460: ip_reass_drophalf();
1461: IPQ_UNLOCK();
1462: }
1.75 thorpej 1463:
1.262.6.2 mjf 1464: KERNEL_UNLOCK_ONE(NULL);
1.1 cgd 1465: }
1466:
1467: /*
1468: * Do option processing on a datagram,
1469: * possibly discarding it if bad options are encountered,
1470: * or forwarding it if source-routed.
1471: * Returns 1 if packet has been forwarded/freed,
1472: * 0 if the packet should be processed further.
1473: */
1.8 mycroft 1474: int
1.211 perry 1475: ip_dooptions(struct mbuf *m)
1.1 cgd 1476: {
1.109 augustss 1477: struct ip *ip = mtod(m, struct ip *);
1478: u_char *cp, *cp0;
1479: struct ip_timestamp *ipt;
1480: struct in_ifaddr *ia;
1.1 cgd 1481: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 1482: struct in_addr dst;
1.1 cgd 1483: n_time ntime;
1484:
1.13 mycroft 1485: dst = ip->ip_dst;
1.1 cgd 1486: cp = (u_char *)(ip + 1);
1487: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
1488: for (; cnt > 0; cnt -= optlen, cp += optlen) {
1489: opt = cp[IPOPT_OPTVAL];
1490: if (opt == IPOPT_EOL)
1491: break;
1492: if (opt == IPOPT_NOP)
1493: optlen = 1;
1494: else {
1.113 itojun 1495: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1496: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1497: goto bad;
1498: }
1.1 cgd 1499: optlen = cp[IPOPT_OLEN];
1.114 itojun 1500: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 1501: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1502: goto bad;
1503: }
1504: }
1505: switch (opt) {
1506:
1507: default:
1508: break;
1509:
1510: /*
1511: * Source routing with record.
1512: * Find interface with current destination address.
1513: * If none on this machine then drop if strictly routed,
1514: * or do nothing if loosely routed.
1515: * Record interface address and bring up next address
1516: * component. If strictly routed make sure next
1517: * address is on directly accessible net.
1518: */
1519: case IPOPT_LSRR:
1520: case IPOPT_SSRR:
1.47 cjs 1521: if (ip_allowsrcrt == 0) {
1522: type = ICMP_UNREACH;
1523: code = ICMP_UNREACH_NET_PROHIB;
1524: goto bad;
1525: }
1.114 itojun 1526: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1527: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1528: goto bad;
1529: }
1.1 cgd 1530: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1531: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1532: goto bad;
1533: }
1534: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 1535: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 1536: if (ia == 0) {
1537: if (opt == IPOPT_SSRR) {
1538: type = ICMP_UNREACH;
1539: code = ICMP_UNREACH_SRCFAIL;
1540: goto bad;
1541: }
1542: /*
1543: * Loose routing, and not at next destination
1544: * yet; nothing to do except forward.
1545: */
1546: break;
1547: }
1548: off--; /* 0 origin */
1.112 sommerfe 1549: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 1550: /*
1551: * End of source route. Should be for us.
1552: */
1553: save_rte(cp, ip->ip_src);
1554: break;
1555: }
1556: /*
1557: * locate outgoing interface
1558: */
1.244 christos 1559: bcopy((void *)(cp + off), (void *)&ipaddr.sin_addr,
1.1 cgd 1560: sizeof(ipaddr.sin_addr));
1.96 thorpej 1561: if (opt == IPOPT_SSRR)
1.196 itojun 1562: ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr)));
1.96 thorpej 1563: else
1.1 cgd 1564: ia = ip_rtaddr(ipaddr.sin_addr);
1565: if (ia == 0) {
1566: type = ICMP_UNREACH;
1567: code = ICMP_UNREACH_SRCFAIL;
1568: goto bad;
1569: }
1570: ip->ip_dst = ipaddr.sin_addr;
1.244 christos 1571: bcopy((void *)&ia->ia_addr.sin_addr,
1572: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1573: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1574: /*
1575: * Let ip_intr's mcast routing check handle mcast pkts
1576: */
1.18 mycroft 1577: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1578: break;
1579:
1580: case IPOPT_RR:
1.114 itojun 1581: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1582: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1583: goto bad;
1584: }
1.1 cgd 1585: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1586: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1587: goto bad;
1588: }
1589: /*
1590: * If no space remains, ignore.
1591: */
1592: off--; /* 0 origin */
1.112 sommerfe 1593: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1594: break;
1.244 christos 1595: bcopy((void *)(&ip->ip_dst), (void *)&ipaddr.sin_addr,
1.1 cgd 1596: sizeof(ipaddr.sin_addr));
1597: /*
1598: * locate outgoing interface; if we're the destination,
1599: * use the incoming interface (should be same).
1600: */
1.96 thorpej 1601: if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))))
1602: == NULL &&
1603: (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
1.1 cgd 1604: type = ICMP_UNREACH;
1605: code = ICMP_UNREACH_HOST;
1606: goto bad;
1607: }
1.244 christos 1608: bcopy((void *)&ia->ia_addr.sin_addr,
1609: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1610: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1611: break;
1612:
1613: case IPOPT_TS:
1614: code = cp - (u_char *)ip;
1615: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1616: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1617: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1618: goto bad;
1.114 itojun 1619: }
1620: if (ipt->ipt_ptr < 5) {
1621: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1622: goto bad;
1623: }
1.15 cgd 1624: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1625: if (++ipt->ipt_oflw == 0) {
1626: code = (u_char *)&ipt->ipt_ptr -
1627: (u_char *)ip;
1.1 cgd 1628: goto bad;
1.114 itojun 1629: }
1.1 cgd 1630: break;
1631: }
1.104 thorpej 1632: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1633: switch (ipt->ipt_flg) {
1634:
1635: case IPOPT_TS_TSONLY:
1636: break;
1637:
1638: case IPOPT_TS_TSANDADDR:
1.66 thorpej 1639: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1640: sizeof(struct in_addr) > ipt->ipt_len) {
1641: code = (u_char *)&ipt->ipt_ptr -
1642: (u_char *)ip;
1.1 cgd 1643: goto bad;
1.114 itojun 1644: }
1.13 mycroft 1645: ipaddr.sin_addr = dst;
1.96 thorpej 1646: ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr),
1647: m->m_pkthdr.rcvif));
1.13 mycroft 1648: if (ia == 0)
1649: continue;
1.104 thorpej 1650: bcopy(&ia->ia_addr.sin_addr,
1651: cp0, sizeof(struct in_addr));
1.1 cgd 1652: ipt->ipt_ptr += sizeof(struct in_addr);
1653: break;
1654:
1655: case IPOPT_TS_PRESPEC:
1.66 thorpej 1656: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1657: sizeof(struct in_addr) > ipt->ipt_len) {
1658: code = (u_char *)&ipt->ipt_ptr -
1659: (u_char *)ip;
1.1 cgd 1660: goto bad;
1.114 itojun 1661: }
1.104 thorpej 1662: bcopy(cp0, &ipaddr.sin_addr,
1.1 cgd 1663: sizeof(struct in_addr));
1.96 thorpej 1664: if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))
1665: == NULL)
1.1 cgd 1666: continue;
1667: ipt->ipt_ptr += sizeof(struct in_addr);
1668: break;
1669:
1670: default:
1.114 itojun 1671: /* XXX can't take &ipt->ipt_flg */
1672: code = (u_char *)&ipt->ipt_ptr -
1673: (u_char *)ip + 1;
1.1 cgd 1674: goto bad;
1675: }
1676: ntime = iptime();
1.107 thorpej 1677: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1.244 christos 1678: memmove((char *)cp + ipt->ipt_ptr - 1, cp0,
1.1 cgd 1679: sizeof(n_time));
1680: ipt->ipt_ptr += sizeof(n_time);
1681: }
1682: }
1683: if (forward) {
1.26 thorpej 1684: if (ip_forwsrcrt == 0) {
1685: type = ICMP_UNREACH;
1686: code = ICMP_UNREACH_SRCFAIL;
1687: goto bad;
1688: }
1.1 cgd 1689: ip_forward(m, 1);
1690: return (1);
1.13 mycroft 1691: }
1692: return (0);
1.1 cgd 1693: bad:
1.13 mycroft 1694: icmp_error(m, type, code, 0, 0);
1.262.6.2 mjf 1695: IP_STATINC(IP_STAT_BADOPTIONS);
1.1 cgd 1696: return (1);
1697: }
1698:
1699: /*
1700: * Given address of next destination (final or next hop),
1701: * return internet address info of interface to be used to get there.
1702: */
1703: struct in_ifaddr *
1.211 perry 1704: ip_rtaddr(struct in_addr dst)
1.1 cgd 1705: {
1.249 dyoung 1706: struct rtentry *rt;
1707: union {
1708: struct sockaddr dst;
1709: struct sockaddr_in dst4;
1710: } u;
1711:
1712: sockaddr_in_init(&u.dst4, &dst, 0);
1713:
1714: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL)
1715: return NULL;
1716:
1717: return ifatoia(rt->rt_ifa);
1.1 cgd 1718: }
1719:
1720: /*
1721: * Save incoming source route for use in replies,
1722: * to be picked up later by ip_srcroute if the receiver is interested.
1723: */
1.13 mycroft 1724: void
1.211 perry 1725: save_rte(u_char *option, struct in_addr dst)
1.1 cgd 1726: {
1727: unsigned olen;
1728:
1729: olen = option[IPOPT_OLEN];
1730: #ifdef DIAGNOSTIC
1731: if (ipprintfs)
1.39 christos 1732: printf("save_rte: olen %d\n", olen);
1.89 itojun 1733: #endif /* 0 */
1.1 cgd 1734: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1735: return;
1.244 christos 1736: bcopy((void *)option, (void *)ip_srcrt.srcopt, olen);
1.1 cgd 1737: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1738: ip_srcrt.dst = dst;
1739: }
1740:
1741: /*
1742: * Retrieve incoming source route for use in replies,
1743: * in the same form used by setsockopt.
1744: * The first hop is placed before the options, will be removed later.
1745: */
1746: struct mbuf *
1.211 perry 1747: ip_srcroute(void)
1.1 cgd 1748: {
1.109 augustss 1749: struct in_addr *p, *q;
1750: struct mbuf *m;
1.1 cgd 1751:
1752: if (ip_nhops == 0)
1.237 dyoung 1753: return NULL;
1.1 cgd 1754: m = m_get(M_DONTWAIT, MT_SOOPTS);
1755: if (m == 0)
1.237 dyoung 1756: return NULL;
1.1 cgd 1757:
1.164 matt 1758: MCLAIM(m, &inetdomain.dom_mowner);
1.13 mycroft 1759: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1760:
1761: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1762: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1763: OPTSIZ;
1764: #ifdef DIAGNOSTIC
1765: if (ipprintfs)
1.39 christos 1766: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1767: #endif
1768:
1769: /*
1770: * First save first hop for return route
1771: */
1772: p = &ip_srcrt.route[ip_nhops - 1];
1773: *(mtod(m, struct in_addr *)) = *p--;
1774: #ifdef DIAGNOSTIC
1775: if (ipprintfs)
1.39 christos 1776: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1777: #endif
1778:
1779: /*
1780: * Copy option fields and padding (nop) to mbuf.
1781: */
1782: ip_srcrt.nop = IPOPT_NOP;
1783: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1.244 christos 1784: memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop,
1785: OPTSIZ);
1786: q = (struct in_addr *)(mtod(m, char *) +
1.1 cgd 1787: sizeof(struct in_addr) + OPTSIZ);
1788: #undef OPTSIZ
1789: /*
1790: * Record return path as an IP source route,
1791: * reversing the path (pointers are now aligned).
1792: */
1793: while (p >= ip_srcrt.route) {
1794: #ifdef DIAGNOSTIC
1795: if (ipprintfs)
1.39 christos 1796: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1797: #endif
1798: *q++ = *p--;
1799: }
1800: /*
1801: * Last hop goes to final destination.
1802: */
1803: *q = ip_srcrt.dst;
1804: #ifdef DIAGNOSTIC
1805: if (ipprintfs)
1.39 christos 1806: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1807: #endif
1808: return (m);
1809: }
1810:
1.139 matt 1811: const int inetctlerrmap[PRC_NCMDS] = {
1.256 yamt 1812: [PRC_MSGSIZE] = EMSGSIZE,
1813: [PRC_HOSTDEAD] = EHOSTDOWN,
1814: [PRC_HOSTUNREACH] = EHOSTUNREACH,
1815: [PRC_UNREACH_NET] = EHOSTUNREACH,
1816: [PRC_UNREACH_HOST] = EHOSTUNREACH,
1817: [PRC_UNREACH_PROTOCOL] = ECONNREFUSED,
1818: [PRC_UNREACH_PORT] = ECONNREFUSED,
1819: [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH,
1820: [PRC_PARAMPROB] = ENOPROTOOPT,
1.1 cgd 1821: };
1822:
1823: /*
1824: * Forward a packet. If some error occurs return the sender
1825: * an icmp packet. Note we can't always generate a meaningful
1826: * icmp message because icmp doesn't have a large enough repertoire
1827: * of codes and types.
1828: *
1829: * If not forwarding, just drop the packet. This could be confusing
1830: * if ipforwarding was zero but some routing protocol was advancing
1831: * us as a gateway to somewhere. However, we must let the routing
1832: * protocol deal with that.
1833: *
1834: * The srcrt parameter indicates whether the packet is being forwarded
1835: * via a source route.
1836: */
1.13 mycroft 1837: void
1.211 perry 1838: ip_forward(struct mbuf *m, int srcrt)
1.1 cgd 1839: {
1.109 augustss 1840: struct ip *ip = mtod(m, struct ip *);
1841: struct rtentry *rt;
1.220 christos 1842: int error, type = 0, code = 0, destmtu = 0;
1.1 cgd 1843: struct mbuf *mcopy;
1.13 mycroft 1844: n_long dest;
1.249 dyoung 1845: union {
1846: struct sockaddr dst;
1847: struct sockaddr_in dst4;
1848: } u;
1.164 matt 1849:
1850: /*
1851: * We are now in the output path.
1852: */
1853: MCLAIM(m, &ip_tx_mowner);
1.135 thorpej 1854:
1855: /*
1856: * Clear any in-bound checksum flags for this packet.
1857: */
1858: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1859:
1.13 mycroft 1860: dest = 0;
1.1 cgd 1861: #ifdef DIAGNOSTIC
1.224 joerg 1862: if (ipprintfs) {
1863: printf("forward: src %s ", inet_ntoa(ip->ip_src));
1864: printf("dst %s ttl %x\n", inet_ntoa(ip->ip_dst), ip->ip_ttl);
1865: }
1.1 cgd 1866: #endif
1.93 sommerfe 1867: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.262.6.2 mjf 1868: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 1869: m_freem(m);
1870: return;
1871: }
1872: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1873: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1874: return;
1875: }
1876:
1.249 dyoung 1877: sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
1878: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) {
1879: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0);
1880: return;
1.1 cgd 1881: }
1882:
1883: /*
1.34 mycroft 1884: * Save at most 68 bytes of the packet in case
1.1 cgd 1885: * we need to generate an ICMP message to the src.
1.119 itojun 1886: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1887: */
1.155 itojun 1888: mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1.119 itojun 1889: if (mcopy)
1890: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1891:
1.221 christos 1892: ip->ip_ttl -= IPTTLDEC;
1893:
1.1 cgd 1894: /*
1895: * If forwarding packet using same interface that it came in on,
1896: * perhaps should send a redirect to sender to shortcut a hop.
1897: * Only send redirect if source is sending directly to us,
1898: * and if packet was not source routed (or has any options).
1899: * Also, don't send redirect if forwarding using a default route
1900: * or a route modified by a redirect.
1901: */
1902: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1903: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.250 dyoung 1904: !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) &&
1.1 cgd 1905: ipsendredirects && !srcrt) {
1.19 mycroft 1906: if (rt->rt_ifa &&
1907: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1908: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1909: if (rt->rt_flags & RTF_GATEWAY)
1910: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1911: else
1912: dest = ip->ip_dst.s_addr;
1913: /*
1914: * Router requirements says to only send host
1915: * redirects.
1916: */
1917: type = ICMP_REDIRECT;
1918: code = ICMP_REDIRECT_HOST;
1.1 cgd 1919: #ifdef DIAGNOSTIC
1.77 thorpej 1920: if (ipprintfs)
1921: printf("redirect (%d) to %x\n", code,
1922: (u_int32_t)dest);
1.1 cgd 1923: #endif
1924: }
1925: }
1926:
1.238 dyoung 1927: error = ip_output(m, NULL, &ipforward_rt,
1.173 jonathan 1928: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
1.174 itojun 1929: (struct ip_moptions *)NULL, (struct socket *)NULL);
1.173 jonathan 1930:
1.1 cgd 1931: if (error)
1.262.6.2 mjf 1932: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 1933: else {
1.262.6.2 mjf 1934: uint64_t *ips = IP_STAT_GETREF();
1935: ips[IP_STAT_FORWARD]++;
1936: if (type) {
1937: ips[IP_STAT_REDIRECTSENT]++;
1938: IP_STAT_PUTREF();
1939: } else {
1940: IP_STAT_PUTREF();
1.63 matt 1941: if (mcopy) {
1942: #ifdef GATEWAY
1.64 thorpej 1943: if (mcopy->m_flags & M_CANFASTFWD)
1944: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1945: #endif
1.1 cgd 1946: m_freem(mcopy);
1.63 matt 1947: }
1.1 cgd 1948: return;
1949: }
1950: }
1951: if (mcopy == NULL)
1952: return;
1.13 mycroft 1953:
1.1 cgd 1954: switch (error) {
1955:
1956: case 0: /* forwarded, but need redirect */
1957: /* type, code set above */
1958: break;
1959:
1960: case ENETUNREACH: /* shouldn't happen, checked above */
1961: case EHOSTUNREACH:
1962: case ENETDOWN:
1963: case EHOSTDOWN:
1964: default:
1965: type = ICMP_UNREACH;
1966: code = ICMP_UNREACH_HOST;
1967: break;
1968:
1969: case EMSGSIZE:
1970: type = ICMP_UNREACH;
1971: code = ICMP_UNREACH_NEEDFRAG;
1.262.6.1 mjf 1972:
1.262.6.3 mjf 1973: if ((rt = rtcache_validate(&ipforward_rt)) != NULL)
1974: destmtu = rt->rt_ifp->if_mtu;
1.262.6.1 mjf 1975:
1976: #if defined(IPSEC) || defined(FAST_IPSEC)
1.262.6.3 mjf 1977: {
1.262.6.1 mjf 1978: /*
1979: * If the packet is routed over IPsec tunnel, tell the
1980: * originator the tunnel MTU.
1981: * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1982: * XXX quickhack!!!
1983: */
1984:
1.89 itojun 1985: struct secpolicy *sp;
1986: int ipsecerror;
1.95 itojun 1987: size_t ipsechdr;
1.89 itojun 1988: struct route *ro;
1989:
1990: sp = ipsec4_getpolicybyaddr(mcopy,
1.170 itojun 1991: IPSEC_DIR_OUTBOUND, IP_FORWARDING,
1992: &ipsecerror);
1.89 itojun 1993:
1.262.6.1 mjf 1994: if (sp != NULL) {
1.89 itojun 1995: /* count IPsec header size */
1.95 itojun 1996: ipsechdr = ipsec4_hdrsiz(mcopy,
1.170 itojun 1997: IPSEC_DIR_OUTBOUND, NULL);
1.89 itojun 1998:
1999: /*
2000: * find the correct route for outer IPv4
2001: * header, compute tunnel MTU.
2002: */
1.220 christos 2003:
1.89 itojun 2004: if (sp->req != NULL
1.95 itojun 2005: && sp->req->sav != NULL
2006: && sp->req->sav->sah != NULL) {
2007: ro = &sp->req->sav->sah->sa_route;
1.262.6.3 mjf 2008: rt = rtcache_validate(ro);
1.257 dyoung 2009: if (rt && rt->rt_ifp) {
1.220 christos 2010: destmtu =
1.257 dyoung 2011: rt->rt_rmx.rmx_mtu ?
2012: rt->rt_rmx.rmx_mtu :
2013: rt->rt_ifp->if_mtu;
1.220 christos 2014: destmtu -= ipsechdr;
1.89 itojun 2015: }
2016: }
2017:
1.173 jonathan 2018: #ifdef IPSEC
1.89 itojun 2019: key_freesp(sp);
1.173 jonathan 2020: #else
2021: KEY_FREESP(&sp);
2022: #endif
1.89 itojun 2023: }
2024: }
1.262.6.3 mjf 2025: #endif /*defined(IPSEC) || defined(FAST_IPSEC)*/
1.262.6.2 mjf 2026: IP_STATINC(IP_STAT_CANTFRAG);
1.1 cgd 2027: break;
2028:
2029: case ENOBUFS:
1.143 itojun 2030: #if 1
2031: /*
2032: * a router should not generate ICMP_SOURCEQUENCH as
2033: * required in RFC1812 Requirements for IP Version 4 Routers.
2034: * source quench could be a big problem under DoS attacks,
1.149 wiz 2035: * or if the underlying interface is rate-limited.
1.143 itojun 2036: */
2037: if (mcopy)
2038: m_freem(mcopy);
2039: return;
2040: #else
1.1 cgd 2041: type = ICMP_SOURCEQUENCH;
2042: code = 0;
2043: break;
1.143 itojun 2044: #endif
1.1 cgd 2045: }
1.220 christos 2046: icmp_error(mcopy, type, code, dest, destmtu);
1.44 thorpej 2047: }
2048:
2049: void
1.211 perry 2050: ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
2051: struct mbuf *m)
1.44 thorpej 2052: {
2053:
2054: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
2055: struct timeval tv;
2056:
2057: microtime(&tv);
1.244 christos 2058: *mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1.44 thorpej 2059: SCM_TIMESTAMP, SOL_SOCKET);
2060: if (*mp)
2061: mp = &(*mp)->m_next;
2062: }
2063: if (inp->inp_flags & INP_RECVDSTADDR) {
1.244 christos 2064: *mp = sbcreatecontrol((void *) &ip->ip_dst,
1.44 thorpej 2065: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
2066: if (*mp)
2067: mp = &(*mp)->m_next;
2068: }
2069: #ifdef notyet
2070: /*
2071: * XXX
2072: * Moving these out of udp_input() made them even more broken
2073: * than they already were.
2074: * - fenner@parc.xerox.com
2075: */
2076: /* options were tossed already */
2077: if (inp->inp_flags & INP_RECVOPTS) {
1.244 christos 2078: *mp = sbcreatecontrol((void *) opts_deleted_above,
1.44 thorpej 2079: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
2080: if (*mp)
2081: mp = &(*mp)->m_next;
2082: }
2083: /* ip_srcroute doesn't do what we want here, need to fix */
2084: if (inp->inp_flags & INP_RECVRETOPTS) {
1.244 christos 2085: *mp = sbcreatecontrol((void *) ip_srcroute(),
1.44 thorpej 2086: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
2087: if (*mp)
2088: mp = &(*mp)->m_next;
2089: }
2090: #endif
2091: if (inp->inp_flags & INP_RECVIF) {
2092: struct sockaddr_dl sdl;
2093:
1.252 dyoung 2094: sockaddr_dl_init(&sdl, sizeof(sdl),
2095: (m->m_pkthdr.rcvif != NULL)
2096: ? m->m_pkthdr.rcvif->if_index
2097: : 0,
2098: 0, NULL, 0, NULL, 0);
1.251 dyoung 2099: *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP);
1.44 thorpej 2100: if (*mp)
2101: mp = &(*mp)->m_next;
2102: }
1.13 mycroft 2103: }
2104:
1.189 atatat 2105: /*
1.228 elad 2106: * sysctl helper routine for net.inet.ip.forwsrcrt.
2107: */
2108: static int
2109: sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS)
2110: {
2111: int error, tmp;
2112: struct sysctlnode node;
2113:
2114: node = *rnode;
2115: tmp = ip_forwsrcrt;
2116: node.sysctl_data = &tmp;
2117: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2118: if (error || newp == NULL)
2119: return (error);
2120:
1.230 elad 2121: if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT,
1.232 elad 2122: 0, NULL, NULL, NULL))
1.228 elad 2123: return (EPERM);
2124:
2125: ip_forwsrcrt = tmp;
2126:
2127: return (0);
2128: }
2129:
2130: /*
1.189 atatat 2131: * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the
2132: * range of the new value and tweaks timers if it changes.
2133: */
2134: static int
2135: sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS)
1.13 mycroft 2136: {
1.189 atatat 2137: int error, tmp;
2138: struct sysctlnode node;
2139:
2140: node = *rnode;
2141: tmp = ip_mtudisc_timeout;
2142: node.sysctl_data = &tmp;
2143: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2144: if (error || newp == NULL)
2145: return (error);
2146: if (tmp < 0)
2147: return (EINVAL);
1.52 thorpej 2148:
1.262.6.3 mjf 2149: mutex_enter(softnet_lock);
2150:
1.189 atatat 2151: ip_mtudisc_timeout = tmp;
2152: rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout);
2153:
1.262.6.3 mjf 2154: mutex_exit(softnet_lock);
2155:
1.189 atatat 2156: return (0);
2157: }
1.54 lukem 2158:
1.65 matt 2159: #ifdef GATEWAY
1.189 atatat 2160: /*
1.247 liamjfoy 2161: * sysctl helper routine for net.inet.ip.maxflows.
1.189 atatat 2162: */
2163: static int
2164: sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS)
2165: {
1.262.6.3 mjf 2166: int error;
1.67 thorpej 2167:
1.262.6.3 mjf 2168: error = sysctl_lookup(SYSCTLFN_CALL(rnode));
2169: if (error || newp == NULL)
2170: return (error);
2171:
2172: mutex_enter(softnet_lock);
2173: KERNEL_LOCK(1, NULL);
1.212 perry 2174:
1.262.6.2 mjf 2175: ipflow_prune();
1.262.6.3 mjf 2176:
2177: KERNEL_UNLOCK_ONE(NULL);
2178: mutex_exit(softnet_lock);
1.144 martin 2179:
1.189 atatat 2180: return (0);
2181: }
1.248 liamjfoy 2182:
2183: static int
2184: sysctl_net_inet_ip_hashsize(SYSCTLFN_ARGS)
2185: {
2186: int error, tmp;
2187: struct sysctlnode node;
2188:
2189: node = *rnode;
2190: tmp = ip_hashsize;
2191: node.sysctl_data = &tmp;
2192: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2193: if (error || newp == NULL)
2194: return (error);
2195:
2196: if ((tmp & (tmp - 1)) == 0 && tmp != 0) {
2197: /*
2198: * Can only fail due to malloc()
2199: */
1.262.6.3 mjf 2200: mutex_enter(softnet_lock);
2201: KERNEL_LOCK(1, NULL);
2202:
2203: error = ipflow_invalidate_all(tmp);
2204:
2205: KERNEL_UNLOCK_ONE(NULL);
2206: mutex_exit(softnet_lock);
2207:
1.248 liamjfoy 2208: } else {
2209: /*
2210: * EINVAL if not a power of 2
2211: */
1.262.6.3 mjf 2212: error = EINVAL;
1.248 liamjfoy 2213: }
2214:
1.262.6.3 mjf 2215: return error;
1.248 liamjfoy 2216: }
1.189 atatat 2217: #endif /* GATEWAY */
1.117 tron 2218:
1.262.6.2 mjf 2219: static int
2220: sysctl_net_inet_ip_stats(SYSCTLFN_ARGS)
2221: {
2222:
2223: return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS));
2224: }
1.131 itojun 2225:
1.189 atatat 2226: SYSCTL_SETUP(sysctl_net_inet_ip_setup, "sysctl net.inet.ip subtree setup")
2227: {
2228: extern int subnetsarelocal, hostzeroisbroadcast;
1.180 jonathan 2229:
1.197 atatat 2230: sysctl_createv(clog, 0, NULL, NULL,
2231: CTLFLAG_PERMANENT,
1.189 atatat 2232: CTLTYPE_NODE, "net", NULL,
2233: NULL, 0, NULL, 0,
2234: CTL_NET, CTL_EOL);
1.197 atatat 2235: sysctl_createv(clog, 0, NULL, NULL,
2236: CTLFLAG_PERMANENT,
1.203 atatat 2237: CTLTYPE_NODE, "inet",
2238: SYSCTL_DESCR("PF_INET related settings"),
1.189 atatat 2239: NULL, 0, NULL, 0,
2240: CTL_NET, PF_INET, CTL_EOL);
1.197 atatat 2241: sysctl_createv(clog, 0, NULL, NULL,
2242: CTLFLAG_PERMANENT,
1.203 atatat 2243: CTLTYPE_NODE, "ip",
2244: SYSCTL_DESCR("IPv4 related settings"),
1.189 atatat 2245: NULL, 0, NULL, 0,
2246: CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL);
1.212 perry 2247:
1.197 atatat 2248: sysctl_createv(clog, 0, NULL, NULL,
2249: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2250: CTLTYPE_INT, "forwarding",
2251: SYSCTL_DESCR("Enable forwarding of INET datagrams"),
1.189 atatat 2252: NULL, 0, &ipforwarding, 0,
2253: CTL_NET, PF_INET, IPPROTO_IP,
2254: IPCTL_FORWARDING, CTL_EOL);
1.197 atatat 2255: sysctl_createv(clog, 0, NULL, NULL,
2256: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2257: CTLTYPE_INT, "redirect",
2258: SYSCTL_DESCR("Enable sending of ICMP redirect messages"),
1.189 atatat 2259: NULL, 0, &ipsendredirects, 0,
2260: CTL_NET, PF_INET, IPPROTO_IP,
2261: IPCTL_SENDREDIRECTS, CTL_EOL);
1.197 atatat 2262: sysctl_createv(clog, 0, NULL, NULL,
2263: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2264: CTLTYPE_INT, "ttl",
2265: SYSCTL_DESCR("Default TTL for an INET datagram"),
1.189 atatat 2266: NULL, 0, &ip_defttl, 0,
2267: CTL_NET, PF_INET, IPPROTO_IP,
2268: IPCTL_DEFTTL, CTL_EOL);
2269: #ifdef IPCTL_DEFMTU
1.197 atatat 2270: sysctl_createv(clog, 0, NULL, NULL,
2271: CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */,
1.203 atatat 2272: CTLTYPE_INT, "mtu",
2273: SYSCTL_DESCR("Default MTA for an INET route"),
1.189 atatat 2274: NULL, 0, &ip_mtu, 0,
2275: CTL_NET, PF_INET, IPPROTO_IP,
2276: IPCTL_DEFMTU, CTL_EOL);
2277: #endif /* IPCTL_DEFMTU */
1.197 atatat 2278: sysctl_createv(clog, 0, NULL, NULL,
1.228 elad 2279: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2280: CTLTYPE_INT, "forwsrcrt",
2281: SYSCTL_DESCR("Enable forwarding of source-routed "
2282: "datagrams"),
1.228 elad 2283: sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0,
1.189 atatat 2284: CTL_NET, PF_INET, IPPROTO_IP,
2285: IPCTL_FORWSRCRT, CTL_EOL);
1.197 atatat 2286: sysctl_createv(clog, 0, NULL, NULL,
2287: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2288: CTLTYPE_INT, "directed-broadcast",
2289: SYSCTL_DESCR("Enable forwarding of broadcast datagrams"),
1.189 atatat 2290: NULL, 0, &ip_directedbcast, 0,
2291: CTL_NET, PF_INET, IPPROTO_IP,
2292: IPCTL_DIRECTEDBCAST, CTL_EOL);
1.197 atatat 2293: sysctl_createv(clog, 0, NULL, NULL,
2294: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2295: CTLTYPE_INT, "allowsrcrt",
2296: SYSCTL_DESCR("Accept source-routed datagrams"),
1.189 atatat 2297: NULL, 0, &ip_allowsrcrt, 0,
2298: CTL_NET, PF_INET, IPPROTO_IP,
2299: IPCTL_ALLOWSRCRT, CTL_EOL);
1.197 atatat 2300: sysctl_createv(clog, 0, NULL, NULL,
2301: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2302: CTLTYPE_INT, "subnetsarelocal",
2303: SYSCTL_DESCR("Whether logical subnets are considered "
2304: "local"),
1.189 atatat 2305: NULL, 0, &subnetsarelocal, 0,
2306: CTL_NET, PF_INET, IPPROTO_IP,
2307: IPCTL_SUBNETSARELOCAL, CTL_EOL);
1.197 atatat 2308: sysctl_createv(clog, 0, NULL, NULL,
2309: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2310: CTLTYPE_INT, "mtudisc",
2311: SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"),
1.189 atatat 2312: NULL, 0, &ip_mtudisc, 0,
2313: CTL_NET, PF_INET, IPPROTO_IP,
2314: IPCTL_MTUDISC, CTL_EOL);
1.197 atatat 2315: sysctl_createv(clog, 0, NULL, NULL,
2316: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2317: CTLTYPE_INT, "anonportmin",
2318: SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1.189 atatat 2319: sysctl_net_inet_ip_ports, 0, &anonportmin, 0,
2320: CTL_NET, PF_INET, IPPROTO_IP,
2321: IPCTL_ANONPORTMIN, CTL_EOL);
1.197 atatat 2322: sysctl_createv(clog, 0, NULL, NULL,
2323: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2324: CTLTYPE_INT, "anonportmax",
2325: SYSCTL_DESCR("Highest ephemeral port number to assign"),
1.189 atatat 2326: sysctl_net_inet_ip_ports, 0, &anonportmax, 0,
2327: CTL_NET, PF_INET, IPPROTO_IP,
2328: IPCTL_ANONPORTMAX, CTL_EOL);
1.197 atatat 2329: sysctl_createv(clog, 0, NULL, NULL,
2330: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2331: CTLTYPE_INT, "mtudisctimeout",
2332: SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"),
1.189 atatat 2333: sysctl_net_inet_ip_pmtudto, 0, &ip_mtudisc_timeout, 0,
2334: CTL_NET, PF_INET, IPPROTO_IP,
2335: IPCTL_MTUDISCTIMEOUT, CTL_EOL);
2336: #ifdef GATEWAY
1.197 atatat 2337: sysctl_createv(clog, 0, NULL, NULL,
2338: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2339: CTLTYPE_INT, "maxflows",
2340: SYSCTL_DESCR("Number of flows for fast forwarding"),
1.189 atatat 2341: sysctl_net_inet_ip_maxflows, 0, &ip_maxflows, 0,
2342: CTL_NET, PF_INET, IPPROTO_IP,
2343: IPCTL_MAXFLOWS, CTL_EOL);
1.248 liamjfoy 2344: sysctl_createv(clog, 0, NULL, NULL,
2345: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2346: CTLTYPE_INT, "hashsize",
2347: SYSCTL_DESCR("Size of hash table for fast forwarding (IPv4)"),
2348: sysctl_net_inet_ip_hashsize, 0, &ip_hashsize, 0,
2349: CTL_NET, PF_INET, IPPROTO_IP,
2350: CTL_CREATE, CTL_EOL);
1.189 atatat 2351: #endif /* GATEWAY */
1.197 atatat 2352: sysctl_createv(clog, 0, NULL, NULL,
2353: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2354: CTLTYPE_INT, "hostzerobroadcast",
2355: SYSCTL_DESCR("All zeroes address is broadcast address"),
1.189 atatat 2356: NULL, 0, &hostzeroisbroadcast, 0,
2357: CTL_NET, PF_INET, IPPROTO_IP,
2358: IPCTL_HOSTZEROBROADCAST, CTL_EOL);
2359: #if NGIF > 0
1.197 atatat 2360: sysctl_createv(clog, 0, NULL, NULL,
2361: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2362: CTLTYPE_INT, "gifttl",
2363: SYSCTL_DESCR("Default TTL for a gif tunnel datagram"),
1.189 atatat 2364: NULL, 0, &ip_gif_ttl, 0,
2365: CTL_NET, PF_INET, IPPROTO_IP,
2366: IPCTL_GIF_TTL, CTL_EOL);
2367: #endif /* NGIF */
2368: #ifndef IPNOPRIVPORTS
1.197 atatat 2369: sysctl_createv(clog, 0, NULL, NULL,
2370: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2371: CTLTYPE_INT, "lowportmin",
2372: SYSCTL_DESCR("Lowest privileged ephemeral port number "
2373: "to assign"),
1.189 atatat 2374: sysctl_net_inet_ip_ports, 0, &lowportmin, 0,
2375: CTL_NET, PF_INET, IPPROTO_IP,
2376: IPCTL_LOWPORTMIN, CTL_EOL);
1.197 atatat 2377: sysctl_createv(clog, 0, NULL, NULL,
2378: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2379: CTLTYPE_INT, "lowportmax",
2380: SYSCTL_DESCR("Highest privileged ephemeral port number "
2381: "to assign"),
1.189 atatat 2382: sysctl_net_inet_ip_ports, 0, &lowportmax, 0,
2383: CTL_NET, PF_INET, IPPROTO_IP,
2384: IPCTL_LOWPORTMAX, CTL_EOL);
2385: #endif /* IPNOPRIVPORTS */
1.197 atatat 2386: sysctl_createv(clog, 0, NULL, NULL,
2387: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2388: CTLTYPE_INT, "maxfragpackets",
2389: SYSCTL_DESCR("Maximum number of fragments to retain for "
2390: "possible reassembly"),
1.189 atatat 2391: NULL, 0, &ip_maxfragpackets, 0,
2392: CTL_NET, PF_INET, IPPROTO_IP,
2393: IPCTL_MAXFRAGPACKETS, CTL_EOL);
2394: #if NGRE > 0
1.197 atatat 2395: sysctl_createv(clog, 0, NULL, NULL,
2396: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2397: CTLTYPE_INT, "grettl",
2398: SYSCTL_DESCR("Default TTL for a gre tunnel datagram"),
1.189 atatat 2399: NULL, 0, &ip_gre_ttl, 0,
2400: CTL_NET, PF_INET, IPPROTO_IP,
2401: IPCTL_GRE_TTL, CTL_EOL);
2402: #endif /* NGRE */
1.197 atatat 2403: sysctl_createv(clog, 0, NULL, NULL,
2404: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2405: CTLTYPE_INT, "checkinterface",
2406: SYSCTL_DESCR("Enable receive side of Strong ES model "
2407: "from RFC1122"),
1.189 atatat 2408: NULL, 0, &ip_checkinterface, 0,
2409: CTL_NET, PF_INET, IPPROTO_IP,
2410: IPCTL_CHECKINTERFACE, CTL_EOL);
1.197 atatat 2411: sysctl_createv(clog, 0, NULL, NULL,
2412: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2413: CTLTYPE_INT, "random_id",
2414: SYSCTL_DESCR("Assign random ip_id values"),
1.189 atatat 2415: NULL, 0, &ip_do_randomid, 0,
2416: CTL_NET, PF_INET, IPPROTO_IP,
2417: IPCTL_RANDOMID, CTL_EOL);
1.206 thorpej 2418: sysctl_createv(clog, 0, NULL, NULL,
2419: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2420: CTLTYPE_INT, "do_loopback_cksum",
2421: SYSCTL_DESCR("Perform IP checksum on loopback"),
2422: NULL, 0, &ip_do_loopback_cksum, 0,
2423: CTL_NET, PF_INET, IPPROTO_IP,
2424: IPCTL_LOOPBACKCKSUM, CTL_EOL);
1.219 elad 2425: sysctl_createv(clog, 0, NULL, NULL,
2426: CTLFLAG_PERMANENT,
2427: CTLTYPE_STRUCT, "stats",
2428: SYSCTL_DESCR("IP statistics"),
1.262.6.2 mjf 2429: sysctl_net_inet_ip_stats, 0, NULL, 0,
1.219 elad 2430: CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS,
2431: CTL_EOL);
1.1 cgd 2432: }
1.262.6.2 mjf 2433:
2434: void
2435: ip_statinc(u_int stat)
2436: {
2437:
2438: KASSERT(stat < IP_NSTATS);
2439: IP_STATINC(stat);
2440: }
CVSweb <webmaster@jp.NetBSD.org>