Annotation of src/sys/netinet/ip_input.c, Revision 1.266
1.266 ! thorpej 1: /* $NetBSD: ip_input.c,v 1.265 2008/04/09 05:14:20 thorpej Exp $ */
1.89 itojun 2:
3: /*
4: * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5: * All rights reserved.
1.152 itojun 6: *
1.89 itojun 7: * Redistribution and use in source and binary forms, with or without
8: * modification, are permitted provided that the following conditions
9: * are met:
10: * 1. Redistributions of source code must retain the above copyright
11: * notice, this list of conditions and the following disclaimer.
12: * 2. Redistributions in binary form must reproduce the above copyright
13: * notice, this list of conditions and the following disclaimer in the
14: * documentation and/or other materials provided with the distribution.
15: * 3. Neither the name of the project nor the names of its contributors
16: * may be used to endorse or promote products derived from this software
17: * without specific prior written permission.
1.152 itojun 18: *
1.89 itojun 19: * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22: * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29: * SUCH DAMAGE.
30: */
1.76 thorpej 31:
32: /*-
33: * Copyright (c) 1998 The NetBSD Foundation, Inc.
34: * All rights reserved.
35: *
36: * This code is derived from software contributed to The NetBSD Foundation
37: * by Public Access Networks Corporation ("Panix"). It was developed under
38: * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon.
39: *
40: * Redistribution and use in source and binary forms, with or without
41: * modification, are permitted provided that the following conditions
42: * are met:
43: * 1. Redistributions of source code must retain the above copyright
44: * notice, this list of conditions and the following disclaimer.
45: * 2. Redistributions in binary form must reproduce the above copyright
46: * notice, this list of conditions and the following disclaimer in the
47: * documentation and/or other materials provided with the distribution.
48: * 3. All advertising materials mentioning features or use of this software
49: * must display the following acknowledgement:
50: * This product includes software developed by the NetBSD
51: * Foundation, Inc. and its contributors.
52: * 4. Neither the name of The NetBSD Foundation nor the names of its
53: * contributors may be used to endorse or promote products derived
54: * from this software without specific prior written permission.
55: *
56: * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
57: * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
58: * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
59: * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
60: * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
61: * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
62: * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
63: * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
64: * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
65: * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
66: * POSSIBILITY OF SUCH DAMAGE.
67: */
1.14 cgd 68:
1.1 cgd 69: /*
1.13 mycroft 70: * Copyright (c) 1982, 1986, 1988, 1993
71: * The Regents of the University of California. All rights reserved.
1.1 cgd 72: *
73: * Redistribution and use in source and binary forms, with or without
74: * modification, are permitted provided that the following conditions
75: * are met:
76: * 1. Redistributions of source code must retain the above copyright
77: * notice, this list of conditions and the following disclaimer.
78: * 2. Redistributions in binary form must reproduce the above copyright
79: * notice, this list of conditions and the following disclaimer in the
80: * documentation and/or other materials provided with the distribution.
1.172 agc 81: * 3. Neither the name of the University nor the names of its contributors
1.1 cgd 82: * may be used to endorse or promote products derived from this software
83: * without specific prior written permission.
84: *
85: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
86: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
87: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
88: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
89: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
90: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
91: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
92: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
93: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
94: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
95: * SUCH DAMAGE.
96: *
1.14 cgd 97: * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
1.1 cgd 98: */
1.141 lukem 99:
100: #include <sys/cdefs.h>
1.266 ! thorpej 101: __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.265 2008/04/09 05:14:20 thorpej Exp $");
1.55 scottr 102:
1.184 jonathan 103: #include "opt_inet.h"
1.62 matt 104: #include "opt_gateway.h"
1.69 mrg 105: #include "opt_pfil_hooks.h"
1.91 thorpej 106: #include "opt_ipsec.h"
1.55 scottr 107: #include "opt_mrouting.h"
1.167 martin 108: #include "opt_mbuftrace.h"
1.135 thorpej 109: #include "opt_inet_csum.h"
1.1 cgd 110:
1.5 mycroft 111: #include <sys/param.h>
112: #include <sys/systm.h>
113: #include <sys/malloc.h>
114: #include <sys/mbuf.h>
115: #include <sys/domain.h>
116: #include <sys/protosw.h>
117: #include <sys/socket.h>
1.44 thorpej 118: #include <sys/socketvar.h>
1.5 mycroft 119: #include <sys/errno.h>
120: #include <sys/time.h>
121: #include <sys/kernel.h>
1.72 thorpej 122: #include <sys/pool.h>
1.28 christos 123: #include <sys/sysctl.h>
1.230 elad 124: #include <sys/kauth.h>
1.1 cgd 125:
1.5 mycroft 126: #include <net/if.h>
1.44 thorpej 127: #include <net/if_dl.h>
1.5 mycroft 128: #include <net/route.h>
1.45 mrg 129: #include <net/pfil.h>
1.1 cgd 130:
1.5 mycroft 131: #include <netinet/in.h>
132: #include <netinet/in_systm.h>
133: #include <netinet/ip.h>
134: #include <netinet/in_pcb.h>
1.215 yamt 135: #include <netinet/in_proto.h>
1.5 mycroft 136: #include <netinet/in_var.h>
137: #include <netinet/ip_var.h>
1.266 ! thorpej 138: #include <netinet/ip_private.h>
1.5 mycroft 139: #include <netinet/ip_icmp.h>
1.89 itojun 140: /* just for gif_ttl */
141: #include <netinet/in_gif.h>
142: #include "gif.h"
1.144 martin 143: #include <net/if_gre.h>
144: #include "gre.h"
1.111 jdolecek 145:
146: #ifdef MROUTING
147: #include <netinet/ip_mroute.h>
148: #endif
1.89 itojun 149:
150: #ifdef IPSEC
151: #include <netinet6/ipsec.h>
152: #include <netkey/key.h>
153: #endif
1.173 jonathan 154: #ifdef FAST_IPSEC
155: #include <netipsec/ipsec.h>
156: #include <netipsec/key.h>
157: #endif /* FAST_IPSEC*/
1.44 thorpej 158:
1.1 cgd 159: #ifndef IPFORWARDING
160: #ifdef GATEWAY
161: #define IPFORWARDING 1 /* forward IP packets not for us */
162: #else /* GATEWAY */
163: #define IPFORWARDING 0 /* don't forward IP packets not for us */
164: #endif /* GATEWAY */
165: #endif /* IPFORWARDING */
166: #ifndef IPSENDREDIRECTS
167: #define IPSENDREDIRECTS 1
168: #endif
1.26 thorpej 169: #ifndef IPFORWSRCRT
1.47 cjs 170: #define IPFORWSRCRT 1 /* forward source-routed packets */
171: #endif
172: #ifndef IPALLOWSRCRT
1.48 mrg 173: #define IPALLOWSRCRT 1 /* allow source-routed packets */
1.26 thorpej 174: #endif
1.53 kml 175: #ifndef IPMTUDISC
1.153 itojun 176: #define IPMTUDISC 1
1.53 kml 177: #endif
1.60 kml 178: #ifndef IPMTUDISCTIMEOUT
1.61 kml 179: #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */
1.60 kml 180: #endif
1.53 kml 181:
1.27 thorpej 182: /*
183: * Note: DIRECTED_BROADCAST is handled this way so that previous
184: * configuration using this option will Just Work.
185: */
186: #ifndef IPDIRECTEDBCAST
187: #ifdef DIRECTED_BROADCAST
188: #define IPDIRECTEDBCAST 1
189: #else
190: #define IPDIRECTEDBCAST 0
191: #endif /* DIRECTED_BROADCAST */
192: #endif /* IPDIRECTEDBCAST */
1.1 cgd 193: int ipforwarding = IPFORWARDING;
194: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 195: int ip_defttl = IPDEFTTL;
1.26 thorpej 196: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 197: int ip_directedbcast = IPDIRECTEDBCAST;
1.47 cjs 198: int ip_allowsrcrt = IPALLOWSRCRT;
1.53 kml 199: int ip_mtudisc = IPMTUDISC;
1.156 itojun 200: int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
1.1 cgd 201: #ifdef DIAGNOSTIC
202: int ipprintfs = 0;
203: #endif
1.184 jonathan 204:
205: int ip_do_randomid = 0;
206:
1.165 christos 207: /*
208: * XXX - Setting ip_checkinterface mostly implements the receive side of
209: * the Strong ES model described in RFC 1122, but since the routing table
210: * and transmit implementation do not implement the Strong ES model,
211: * setting this to 1 results in an odd hybrid.
212: *
213: * XXX - ip_checkinterface currently must be disabled if you use ipnat
214: * to translate the destination address to another local interface.
215: *
216: * XXX - ip_checkinterface must be disabled if you add IP aliases
217: * to the loopback interface instead of the interface where the
218: * packets for those addresses are received.
219: */
220: int ip_checkinterface = 0;
221:
1.1 cgd 222:
1.60 kml 223: struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
224:
1.1 cgd 225: int ipqmaxlen = IFQ_MAXLEN;
1.150 matt 226: u_long in_ifaddrhash; /* size of hash table - 1 */
227: int in_ifaddrentries; /* total number of addrs */
1.212 perry 228: struct in_ifaddrhead in_ifaddrhead;
1.57 tls 229: struct in_ifaddrhashhead *in_ifaddrhashtbl;
1.166 matt 230: u_long in_multihash; /* size of hash table - 1 */
231: int in_multientries; /* total number of addrs */
232: struct in_multihashhead *in_multihashtbl;
1.13 mycroft 233: struct ifqueue ipintrq;
1.183 jonathan 234: uint16_t ip_id;
1.75 thorpej 235:
1.266 ! thorpej 236: percpu_t *ipstat_percpu;
! 237:
1.121 thorpej 238: #ifdef PFIL_HOOKS
239: struct pfil_head inet_pfil_hook;
240: #endif
241:
1.194 jonathan 242: /*
243: * Cached copy of nmbclusters. If nbclusters is different,
244: * recalculate IP parameters derived from nmbclusters.
245: */
246: static int ip_nmbclusters; /* copy of nmbclusters */
1.210 perry 247: static void ip_nmbclusters_changed(void); /* recalc limits */
1.194 jonathan 248:
1.195 thorpej 249: #define CHECK_NMBCLUSTER_PARAMS() \
250: do { \
251: if (__predict_false(ip_nmbclusters != nmbclusters)) \
252: ip_nmbclusters_changed(); \
253: } while (/*CONSTCOND*/0)
1.194 jonathan 254:
1.190 jonathan 255: /* IP datagram reassembly queues (hashed) */
256: #define IPREASS_NHASH_LOG2 6
257: #define IPREASS_NHASH (1 << IPREASS_NHASH_LOG2)
258: #define IPREASS_HMASK (IPREASS_NHASH - 1)
259: #define IPREASS_HASH(x,y) \
260: (((((x) & 0xF) | ((((x) >> 8) & 0xF) << 4)) ^ (y)) & IPREASS_HMASK)
261: struct ipqhead ipq[IPREASS_NHASH];
1.75 thorpej 262: int ipq_locked;
1.212 perry 263: static int ip_nfragpackets; /* packets in reass queue */
1.194 jonathan 264: static int ip_nfrags; /* total fragments in reass queues */
265:
266: int ip_maxfragpackets = 200; /* limit on packets. XXX sysctl */
267: int ip_maxfrags; /* limit on fragments. XXX sysctl */
268:
269:
270: /*
271: * Additive-Increase/Multiplicative-Decrease (AIMD) strategy for
272: * IP reassembly queue buffer managment.
1.212 perry 273: *
1.194 jonathan 274: * We keep a count of total IP fragments (NB: not fragmented packets!)
275: * awaiting reassembly (ip_nfrags) and a limit (ip_maxfrags) on fragments.
276: * If ip_nfrags exceeds ip_maxfrags the limit, we drop half the
277: * total fragments in reassembly queues.This AIMD policy avoids
278: * repeatedly deleting single packets under heavy fragmentation load
279: * (e.g., from lossy NFS peers).
280: */
1.212 perry 281: static u_int ip_reass_ttl_decr(u_int ticks);
1.210 perry 282: static void ip_reass_drophalf(void);
1.194 jonathan 283:
1.75 thorpej 284:
1.223 perry 285: static inline int ipq_lock_try(void);
286: static inline void ipq_unlock(void);
1.75 thorpej 287:
1.223 perry 288: static inline int
1.211 perry 289: ipq_lock_try(void)
1.75 thorpej 290: {
291: int s;
292:
1.132 thorpej 293: /*
1.149 wiz 294: * Use splvm() -- we're blocking things that would cause
1.132 thorpej 295: * mbuf allocation.
296: */
297: s = splvm();
1.75 thorpej 298: if (ipq_locked) {
299: splx(s);
300: return (0);
301: }
302: ipq_locked = 1;
303: splx(s);
304: return (1);
305: }
306:
1.223 perry 307: static inline void
1.211 perry 308: ipq_unlock(void)
1.75 thorpej 309: {
310: int s;
311:
1.132 thorpej 312: s = splvm();
1.75 thorpej 313: ipq_locked = 0;
314: splx(s);
315: }
316:
317: #ifdef DIAGNOSTIC
318: #define IPQ_LOCK() \
319: do { \
320: if (ipq_lock_try() == 0) { \
321: printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \
322: panic("ipq_lock"); \
323: } \
1.159 perry 324: } while (/*CONSTCOND*/ 0)
1.75 thorpej 325: #define IPQ_LOCK_CHECK() \
326: do { \
327: if (ipq_locked == 0) { \
328: printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \
329: panic("ipq lock check"); \
330: } \
1.159 perry 331: } while (/*CONSTCOND*/ 0)
1.75 thorpej 332: #else
333: #define IPQ_LOCK() (void) ipq_lock_try()
334: #define IPQ_LOCK_CHECK() /* nothing */
335: #endif
336:
337: #define IPQ_UNLOCK() ipq_unlock()
1.1 cgd 338:
1.246 ad 339: POOL_INIT(inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl", NULL,
340: IPL_SOFTNET);
341: POOL_INIT(ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", NULL,
342: IPL_VM);
1.72 thorpej 343:
1.135 thorpej 344: #ifdef INET_CSUM_COUNTERS
345: #include <sys/device.h>
346:
347: struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
348: NULL, "inet", "hwcsum bad");
349: struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
350: NULL, "inet", "hwcsum ok");
351: struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
352: NULL, "inet", "swcsum");
353:
354: #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++
355:
1.201 matt 356: EVCNT_ATTACH_STATIC(ip_hwcsum_bad);
357: EVCNT_ATTACH_STATIC(ip_hwcsum_ok);
358: EVCNT_ATTACH_STATIC(ip_swcsum);
359:
1.135 thorpej 360: #else
361:
362: #define INET_CSUM_COUNTER_INCR(ev) /* nothing */
363:
364: #endif /* INET_CSUM_COUNTERS */
365:
1.1 cgd 366: /*
367: * We need to save the IP options in case a protocol wants to respond
368: * to an incoming packet over the same route if the packet got here
369: * using IP source routing. This allows connection establishment and
370: * maintenance when the remote end is on a network that is not known
371: * to us.
372: */
373: int ip_nhops = 0;
374: static struct ip_srcrt {
375: struct in_addr dst; /* final destination */
376: char nop; /* one NOP to align */
377: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
378: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
379: } ip_srcrt;
380:
1.210 perry 381: static void save_rte(u_char *, struct in_addr);
1.35 mycroft 382:
1.164 matt 383: #ifdef MBUFTRACE
1.234 dogcow 384: struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx");
385: struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx");
1.164 matt 386: #endif
387:
1.1 cgd 388: /*
1.194 jonathan 389: * Compute IP limits derived from the value of nmbclusters.
390: */
391: static void
392: ip_nmbclusters_changed(void)
393: {
394: ip_maxfrags = nmbclusters / 4;
395: ip_nmbclusters = nmbclusters;
396: }
397:
398: /*
1.1 cgd 399: * IP initialization: fill in IP protocol switch table.
400: * All protocols not implemented in kernel go to raw IP protocol handler.
401: */
1.8 mycroft 402: void
1.211 perry 403: ip_init(void)
1.1 cgd 404: {
1.199 matt 405: const struct protosw *pr;
1.109 augustss 406: int i;
1.1 cgd 407:
408: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
409: if (pr == 0)
410: panic("ip_init");
411: for (i = 0; i < IPPROTO_MAX; i++)
412: ip_protox[i] = pr - inetsw;
413: for (pr = inetdomain.dom_protosw;
414: pr < inetdomain.dom_protoswNPROTOSW; pr++)
415: if (pr->pr_domain->dom_family == PF_INET &&
416: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
417: ip_protox[pr->pr_protocol] = pr - inetsw;
1.192 jonathan 418:
1.190 jonathan 419: for (i = 0; i < IPREASS_NHASH; i++)
420: LIST_INIT(&ipq[i]);
421:
1.262 matt 422: ip_initid();
1.227 kardel 423: ip_id = time_second & 0xfffff;
1.194 jonathan 424:
1.1 cgd 425: ipintrq.ifq_maxlen = ipqmaxlen;
1.194 jonathan 426: ip_nmbclusters_changed();
427:
1.181 jonathan 428: TAILQ_INIT(&in_ifaddrhead);
1.120 ad 429: in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IFADDR,
430: M_WAITOK, &in_ifaddrhash);
1.166 matt 431: in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IPMADDR,
432: M_WAITOK, &in_multihash);
1.160 itojun 433: ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout);
1.73 thorpej 434: #ifdef GATEWAY
1.248 liamjfoy 435: ipflow_init(ip_hashsize);
1.73 thorpej 436: #endif
1.121 thorpej 437:
438: #ifdef PFIL_HOOKS
439: /* Register our Packet Filter hook. */
1.126 thorpej 440: inet_pfil_hook.ph_type = PFIL_TYPE_AF;
441: inet_pfil_hook.ph_af = AF_INET;
1.121 thorpej 442: i = pfil_head_register(&inet_pfil_hook);
443: if (i != 0)
444: printf("ip_init: WARNING: unable to register pfil hook, "
445: "error %d\n", i);
446: #endif /* PFIL_HOOKS */
1.135 thorpej 447:
1.164 matt 448: #ifdef MBUFTRACE
449: MOWNER_ATTACH(&ip_tx_mowner);
450: MOWNER_ATTACH(&ip_rx_mowner);
451: #endif /* MBUFTRACE */
1.266 ! thorpej 452:
! 453: ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS);
1.1 cgd 454: }
455:
1.229 christos 456: struct sockaddr_in ipaddr = {
457: .sin_len = sizeof(ipaddr),
458: .sin_family = AF_INET,
459: };
1.1 cgd 460: struct route ipforward_rt;
461:
462: /*
1.89 itojun 463: * IP software interrupt routine
464: */
465: void
1.211 perry 466: ipintr(void)
1.89 itojun 467: {
468: int s;
469: struct mbuf *m;
470:
1.241 ad 471: while (!IF_IS_EMPTY(&ipintrq)) {
1.132 thorpej 472: s = splnet();
1.89 itojun 473: IF_DEQUEUE(&ipintrq, m);
474: splx(s);
475: if (m == 0)
476: return;
477: ip_input(m);
478: }
479: }
480:
481: /*
1.1 cgd 482: * Ip input routine. Checksum and byte swap header. If fragmented
483: * try to reassemble. Process options. Pass to next level.
484: */
1.8 mycroft 485: void
1.89 itojun 486: ip_input(struct mbuf *m)
1.1 cgd 487: {
1.109 augustss 488: struct ip *ip = NULL;
489: struct ipq *fp;
490: struct in_ifaddr *ia;
491: struct ifaddr *ifa;
1.25 cgd 492: struct ipqent *ipqe;
1.89 itojun 493: int hlen = 0, mff, len;
1.100 itojun 494: int downmatch;
1.165 christos 495: int checkif;
1.169 itojun 496: int srcrt = 0;
1.233 tls 497: int s;
1.190 jonathan 498: u_int hash;
1.173 jonathan 499: #ifdef FAST_IPSEC
500: struct m_tag *mtag;
501: struct tdb_ident *tdbi;
502: struct secpolicy *sp;
1.233 tls 503: int error;
1.173 jonathan 504: #endif /* FAST_IPSEC */
1.1 cgd 505:
1.164 matt 506: MCLAIM(m, &ip_rx_mowner);
1.1 cgd 507: #ifdef DIAGNOSTIC
508: if ((m->m_flags & M_PKTHDR) == 0)
509: panic("ipintr no HDR");
1.89 itojun 510: #endif
1.164 matt 511:
1.1 cgd 512: /*
513: * If no IP addresses have been set yet but the interfaces
514: * are receiving, can't do anything with incoming packets yet.
515: */
1.181 jonathan 516: if (TAILQ_FIRST(&in_ifaddrhead) == 0)
1.1 cgd 517: goto bad;
1.266 ! thorpej 518: IP_STATINC(IP_STAT_TOTAL);
1.154 thorpej 519: /*
520: * If the IP header is not aligned, slurp it up into a new
521: * mbuf with space for link headers, in the event we forward
522: * it. Otherwise, if it is aligned, make sure the entire
523: * base IP header is in the first mbuf of the chain.
524: */
1.244 christos 525: if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
1.154 thorpej 526: if ((m = m_copyup(m, sizeof(struct ip),
527: (max_linkhdr + 3) & ~3)) == NULL) {
528: /* XXXJRT new stat, please */
1.266 ! thorpej 529: IP_STATINC(IP_STAT_TOOSMALL);
1.154 thorpej 530: return;
531: }
532: } else if (__predict_false(m->m_len < sizeof (struct ip))) {
533: if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
1.266 ! thorpej 534: IP_STATINC(IP_STAT_TOOSMALL);
1.154 thorpej 535: return;
536: }
1.1 cgd 537: }
538: ip = mtod(m, struct ip *);
1.13 mycroft 539: if (ip->ip_v != IPVERSION) {
1.266 ! thorpej 540: IP_STATINC(IP_STAT_BADVERS);
1.13 mycroft 541: goto bad;
542: }
1.1 cgd 543: hlen = ip->ip_hl << 2;
544: if (hlen < sizeof(struct ip)) { /* minimum header length */
1.266 ! thorpej 545: IP_STATINC(IP_STAT_BADHLEN);
1.1 cgd 546: goto bad;
547: }
548: if (hlen > m->m_len) {
549: if ((m = m_pullup(m, hlen)) == 0) {
1.266 ! thorpej 550: IP_STATINC(IP_STAT_BADHLEN);
1.89 itojun 551: return;
1.1 cgd 552: }
553: ip = mtod(m, struct ip *);
554: }
1.98 thorpej 555:
1.85 hwr 556: /*
1.99 thorpej 557: * RFC1122: packets with a multicast source address are
1.98 thorpej 558: * not allowed.
1.85 hwr 559: */
560: if (IN_MULTICAST(ip->ip_src.s_addr)) {
1.266 ! thorpej 561: IP_STATINC(IP_STAT_BADADDR);
1.85 hwr 562: goto bad;
1.129 itojun 563: }
564:
565: /* 127/8 must not appear on wire - RFC1122 */
566: if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
567: (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
1.130 itojun 568: if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
1.266 ! thorpej 569: IP_STATINC(IP_STAT_BADADDR);
1.129 itojun 570: goto bad;
1.130 itojun 571: }
1.85 hwr 572: }
573:
1.135 thorpej 574: switch (m->m_pkthdr.csum_flags &
1.137 thorpej 575: ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) |
1.135 thorpej 576: M_CSUM_IPv4_BAD)) {
577: case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
578: INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
579: goto badcsum;
580:
581: case M_CSUM_IPv4:
582: /* Checksum was okay. */
583: INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
584: break;
585:
586: default:
1.206 thorpej 587: /*
588: * Must compute it ourselves. Maybe skip checksum on
589: * loopback interfaces.
590: */
591: if (__predict_true(!(m->m_pkthdr.rcvif->if_flags &
592: IFF_LOOPBACK) || ip_do_loopback_cksum)) {
593: INET_CSUM_COUNTER_INCR(&ip_swcsum);
594: if (in_cksum(m, hlen) != 0)
595: goto badcsum;
596: }
1.135 thorpej 597: break;
1.1 cgd 598: }
599:
1.121 thorpej 600: /* Retrieve the packet length. */
601: len = ntohs(ip->ip_len);
1.81 proff 602:
603: /*
604: * Check for additional length bogosity
605: */
1.84 proff 606: if (len < hlen) {
1.266 ! thorpej 607: IP_STATINC(IP_STAT_BADLEN);
1.81 proff 608: goto bad;
609: }
1.1 cgd 610:
611: /*
612: * Check that the amount of data in the buffers
613: * is as at least much as the IP header would have us expect.
614: * Trim mbufs if longer than we expect.
615: * Drop packet if shorter than we expect.
616: */
1.35 mycroft 617: if (m->m_pkthdr.len < len) {
1.266 ! thorpej 618: IP_STATINC(IP_STAT_TOOSHORT);
1.1 cgd 619: goto bad;
620: }
1.35 mycroft 621: if (m->m_pkthdr.len > len) {
1.1 cgd 622: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 623: m->m_len = len;
624: m->m_pkthdr.len = len;
1.1 cgd 625: } else
1.35 mycroft 626: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 627: }
628:
1.193 scw 629: #if defined(IPSEC)
1.149 wiz 630: /* ipflow (IP fast forwarding) is not compatible with IPsec. */
1.94 itojun 631: m->m_flags &= ~M_CANFASTFWD;
632: #else
1.64 thorpej 633: /*
634: * Assume that we can create a fast-forward IP flow entry
635: * based on this packet.
636: */
637: m->m_flags |= M_CANFASTFWD;
1.94 itojun 638: #endif
1.64 thorpej 639:
1.36 mrg 640: #ifdef PFIL_HOOKS
1.33 mrg 641: /*
1.64 thorpej 642: * Run through list of hooks for input packets. If there are any
643: * filters which require that additional packets in the flow are
644: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
645: * Note that filters must _never_ set this flag, as another filter
646: * in the list may have previously cleared it.
1.33 mrg 647: */
1.127 itojun 648: /*
649: * let ipfilter look at packet on the wire,
650: * not the decapsulated packet.
651: */
652: #ifdef IPSEC
1.136 itojun 653: if (!ipsec_getnhist(m))
1.186 scw 654: #elif defined(FAST_IPSEC)
655: if (!ipsec_indone(m))
1.127 itojun 656: #else
657: if (1)
658: #endif
659: {
1.169 itojun 660: struct in_addr odst;
661:
662: odst = ip->ip_dst;
1.127 itojun 663: if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif,
1.168 itojun 664: PFIL_IN) != 0)
665: return;
1.127 itojun 666: if (m == NULL)
667: return;
668: ip = mtod(m, struct ip *);
1.142 darrenr 669: hlen = ip->ip_hl << 2;
1.205 darrenr 670: /*
671: * XXX The setting of "srcrt" here is to prevent ip_forward()
672: * from generating ICMP redirects for packets that have
673: * been redirected by a hook back out on to the same LAN that
674: * they came from and is not an indication that the packet
675: * is being inffluenced by source routing options. This
676: * allows things like
677: * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp"
678: * where tlp0 is both on the 1.1.1.0/24 network and is the
679: * default route for hosts on 1.1.1.0/24. Of course this
680: * also requires a "map tlp0 ..." to complete the story.
681: * One might argue whether or not this kind of network config.
1.212 perry 682: * should be supported in this manner...
1.205 darrenr 683: */
1.169 itojun 684: srcrt = (odst.s_addr != ip->ip_dst.s_addr);
1.127 itojun 685: }
1.36 mrg 686: #endif /* PFIL_HOOKS */
1.123 thorpej 687:
688: #ifdef ALTQ
689: /* XXX Temporary until ALTQ is changed to use a pfil hook */
690: if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) {
691: /* packet dropped by traffic conditioner */
692: return;
693: }
694: #endif
1.121 thorpej 695:
696: /*
1.1 cgd 697: * Process options and, if not destined for us,
698: * ship it on. ip_dooptions returns 1 when an
699: * error was detected (causing an icmp message
700: * to be sent and the original packet to be freed).
701: */
702: ip_nhops = 0; /* for source routed packets */
703: if (hlen > sizeof (struct ip) && ip_dooptions(m))
1.89 itojun 704: return;
1.1 cgd 705:
706: /*
1.165 christos 707: * Enable a consistency check between the destination address
708: * and the arrival interface for a unicast packet (the RFC 1122
709: * strong ES model) if IP forwarding is disabled and the packet
710: * is not locally generated.
711: *
712: * XXX - Checking also should be disabled if the destination
713: * address is ipnat'ed to a different interface.
714: *
715: * XXX - Checking is incompatible with IP aliases added
716: * to the loopback interface instead of the interface where
717: * the packets are received.
718: *
719: * XXX - We need to add a per ifaddr flag for this so that
720: * we get finer grain control.
721: */
722: checkif = ip_checkinterface && (ipforwarding == 0) &&
723: (m->m_pkthdr.rcvif != NULL) &&
724: ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0);
725:
726: /*
1.1 cgd 727: * Check our list of addresses, to see if the packet is for us.
1.100 itojun 728: *
729: * Traditional 4.4BSD did not consult IFF_UP at all.
730: * The behavior here is to treat addresses on !IFF_UP interface
731: * as not mine.
1.1 cgd 732: */
1.100 itojun 733: downmatch = 0;
1.140 matt 734: LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) {
1.97 itojun 735: if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
1.165 christos 736: if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif)
737: continue;
1.97 itojun 738: if ((ia->ia_ifp->if_flags & IFF_UP) != 0)
739: break;
1.100 itojun 740: else
741: downmatch++;
1.97 itojun 742: }
743: }
1.86 thorpej 744: if (ia != NULL)
745: goto ours;
1.225 christos 746: if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
1.209 matt 747: IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) {
1.140 matt 748: if (ifa->ifa_addr->sa_family != AF_INET)
749: continue;
1.57 tls 750: ia = ifatoia(ifa);
1.35 mycroft 751: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
752: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 753: /*
754: * Look for all-0's host part (old broadcast addr),
755: * either for subnet or net.
756: */
757: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 758: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 759: goto ours;
1.57 tls 760: /*
761: * An interface with IP address zero accepts
762: * all packets that arrive on that interface.
763: */
764: if (in_nullhost(ia->ia_addr.sin_addr))
765: goto ours;
1.1 cgd 766: }
767: }
1.18 mycroft 768: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 769: struct in_multi *inm;
770: #ifdef MROUTING
771: extern struct socket *ip_mrouter;
1.10 brezak 772:
1.4 hpeyerl 773: if (ip_mrouter) {
774: /*
775: * If we are acting as a multicast router, all
776: * incoming multicast packets are passed to the
777: * kernel-level multicast forwarding function.
778: * The packet is returned (relatively) intact; if
779: * ip_mforward() returns a non-zero value, the packet
780: * must be discarded, else it may be accepted below.
781: *
782: * (The IP ident field is put in the same byte order
783: * as expected when ip_mforward() is called from
784: * ip_output().)
785: */
1.13 mycroft 786: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
1.266 ! thorpej 787: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 788: m_freem(m);
1.89 itojun 789: return;
1.4 hpeyerl 790: }
791:
792: /*
793: * The process-level routing demon needs to receive
794: * all multicast IGMP packets, whether or not this
795: * host belongs to their destination groups.
796: */
797: if (ip->ip_p == IPPROTO_IGMP)
798: goto ours;
1.266 ! thorpej 799: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 800: }
801: #endif
802: /*
803: * See if we belong to the destination multicast group on the
804: * arrival interface.
805: */
806: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
807: if (inm == NULL) {
1.266 ! thorpej 808: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 809: m_freem(m);
1.89 itojun 810: return;
1.4 hpeyerl 811: }
812: goto ours;
813: }
1.19 mycroft 814: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 815: in_nullhost(ip->ip_dst))
1.1 cgd 816: goto ours;
817:
818: /*
819: * Not for us; forward if possible and desirable.
820: */
821: if (ipforwarding == 0) {
1.266 ! thorpej 822: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 823: m_freem(m);
1.100 itojun 824: } else {
825: /*
826: * If ip_dst matched any of my address on !IFF_UP interface,
827: * and there's no IFF_UP interface that matches ip_dst,
828: * send icmp unreach. Forwarding it will result in in-kernel
829: * forwarding loop till TTL goes to 0.
830: */
831: if (downmatch) {
832: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
1.266 ! thorpej 833: IP_STATINC(IP_STAT_CANTFORWARD);
1.100 itojun 834: return;
835: }
1.145 itojun 836: #ifdef IPSEC
837: if (ipsec4_in_reject(m, NULL)) {
838: ipsecstat.in_polvio++;
839: goto bad;
840: }
841: #endif
1.173 jonathan 842: #ifdef FAST_IPSEC
843: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
844: s = splsoftnet();
845: if (mtag != NULL) {
846: tdbi = (struct tdb_ident *)(mtag + 1);
847: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
848: } else {
849: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
1.212 perry 850: IP_FORWARDING, &error);
1.173 jonathan 851: }
852: if (sp == NULL) { /* NB: can happen if error */
853: splx(s);
854: /*XXX error stat???*/
855: DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/
856: goto bad;
857: }
858:
859: /*
860: * Check security policy against packet attributes.
861: */
862: error = ipsec_in_reject(sp, m);
863: KEY_FREESP(&sp);
864: splx(s);
865: if (error) {
1.266 ! thorpej 866: IP_STATINC(IP_STAT_CANTFORWARD);
1.173 jonathan 867: goto bad;
1.193 scw 868: }
869:
870: /*
871: * Peek at the outbound SP for this packet to determine if
872: * it's a Fast Forward candidate.
873: */
874: mtag = m_tag_find(m, PACKET_TAG_IPSEC_PENDING_TDB, NULL);
875: if (mtag != NULL)
876: m->m_flags &= ~M_CANFASTFWD;
877: else {
878: s = splsoftnet();
879: sp = ipsec4_checkpolicy(m, IPSEC_DIR_OUTBOUND,
880: (IP_FORWARDING |
881: (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
882: &error, NULL);
883: if (sp != NULL) {
884: m->m_flags &= ~M_CANFASTFWD;
885: KEY_FREESP(&sp);
886: }
887: splx(s);
1.173 jonathan 888: }
889: #endif /* FAST_IPSEC */
1.145 itojun 890:
1.169 itojun 891: ip_forward(m, srcrt);
1.100 itojun 892: }
1.89 itojun 893: return;
1.1 cgd 894:
895: ours:
896: /*
897: * If offset or IP_MF are set, must reassemble.
898: * Otherwise, nothing need be done.
899: * (We could look in the reassembly queue to see
900: * if the packet was previously fragmented,
901: * but it's not worth the time; just let them time out.)
902: */
1.155 itojun 903: if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
1.258 matt 904: uint16_t off;
905: /*
906: * Prevent TCP blind data attacks by not allowing non-initial
907: * fragments to start at less than 68 bytes (minimal fragment
1.259 matt 908: * size) and making sure the first fragment is at least 68
909: * bytes.
1.258 matt 910: */
1.260 matt 911: off = (ntohs(ip->ip_off) & IP_OFFMASK) << 3;
1.259 matt 912: if ((off > 0 ? off + hlen : len) < IP_MINFRAGSIZE - 1) {
1.266 ! thorpej 913: IP_STATINC(IP_STAT_BADFRAGS);
1.258 matt 914: goto bad;
915: }
1.1 cgd 916: /*
917: * Look for queue of fragments
918: * of this datagram.
919: */
1.75 thorpej 920: IPQ_LOCK();
1.190 jonathan 921: hash = IPREASS_HASH(ip->ip_src.s_addr, ip->ip_id);
1.250 dyoung 922: LIST_FOREACH(fp, &ipq[hash], ipq_q) {
1.1 cgd 923: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 924: in_hosteq(ip->ip_src, fp->ipq_src) &&
925: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.260 matt 926: ip->ip_p == fp->ipq_p) {
927: /*
928: * Make sure the TOS is matches previous
929: * fragments.
930: */
931: if (ip->ip_tos != fp->ipq_tos) {
1.266 ! thorpej 932: IP_STATINC(IP_STAT_BADFRAGS);
1.260 matt 933: goto bad;
934: }
1.1 cgd 935: goto found;
1.260 matt 936: }
1.190 jonathan 937: }
1.1 cgd 938: fp = 0;
939: found:
940:
941: /*
942: * Adjust ip_len to not reflect header,
1.25 cgd 943: * set ipqe_mff if more fragments are expected,
1.1 cgd 944: * convert offset of this to bytes.
945: */
1.155 itojun 946: ip->ip_len = htons(ntohs(ip->ip_len) - hlen);
947: mff = (ip->ip_off & htons(IP_MF)) != 0;
1.25 cgd 948: if (mff) {
1.16 cgd 949: /*
950: * Make sure that fragments have a data length
951: * that's a non-zero multiple of 8 bytes.
952: */
1.155 itojun 953: if (ntohs(ip->ip_len) == 0 ||
954: (ntohs(ip->ip_len) & 0x7) != 0) {
1.266 ! thorpej 955: IP_STATINC(IP_STAT_BADFRAGS);
1.75 thorpej 956: IPQ_UNLOCK();
1.16 cgd 957: goto bad;
958: }
959: }
1.155 itojun 960: ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3);
1.1 cgd 961:
962: /*
963: * If datagram marked as having more fragments
964: * or if this is not the first fragment,
965: * attempt reassembly; if it succeeds, proceed.
966: */
1.155 itojun 967: if (mff || ip->ip_off != htons(0)) {
1.266 ! thorpej 968: IP_STATINC(IP_STAT_FRAGMENTS);
1.233 tls 969: s = splvm();
1.72 thorpej 970: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.233 tls 971: splx(s);
1.25 cgd 972: if (ipqe == NULL) {
1.266 ! thorpej 973: IP_STATINC(IP_STAT_RCVMEMDROP);
1.75 thorpej 974: IPQ_UNLOCK();
1.25 cgd 975: goto bad;
976: }
977: ipqe->ipqe_mff = mff;
1.50 thorpej 978: ipqe->ipqe_m = m;
1.25 cgd 979: ipqe->ipqe_ip = ip;
1.190 jonathan 980: m = ip_reass(ipqe, fp, &ipq[hash]);
1.75 thorpej 981: if (m == 0) {
982: IPQ_UNLOCK();
1.89 itojun 983: return;
1.75 thorpej 984: }
1.266 ! thorpej 985: IP_STATINC(IP_STAT_REASSEMBLED);
1.50 thorpej 986: ip = mtod(m, struct ip *);
1.74 thorpej 987: hlen = ip->ip_hl << 2;
1.155 itojun 988: ip->ip_len = htons(ntohs(ip->ip_len) + hlen);
1.1 cgd 989: } else
990: if (fp)
991: ip_freef(fp);
1.75 thorpej 992: IPQ_UNLOCK();
1.79 mycroft 993: }
1.128 itojun 994:
1.173 jonathan 995: #if defined(IPSEC)
1.128 itojun 996: /*
997: * enforce IPsec policy checking if we are seeing last header.
998: * note that we do not visit this with protocols with pcb layer
999: * code - like udp/tcp/raw ip.
1000: */
1001: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 &&
1002: ipsec4_in_reject(m, NULL)) {
1003: ipsecstat.in_polvio++;
1004: goto bad;
1005: }
1006: #endif
1.226 liamjfoy 1007: #ifdef FAST_IPSEC
1.173 jonathan 1008: /*
1009: * enforce IPsec policy checking if we are seeing last header.
1010: * note that we do not visit this with protocols with pcb layer
1011: * code - like udp/tcp/raw ip.
1012: */
1013: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) {
1014: /*
1015: * Check if the packet has already had IPsec processing
1016: * done. If so, then just pass it along. This tag gets
1017: * set during AH, ESP, etc. input handling, before the
1018: * packet is returned to the ip input queue for delivery.
1.212 perry 1019: */
1.173 jonathan 1020: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
1021: s = splsoftnet();
1022: if (mtag != NULL) {
1023: tdbi = (struct tdb_ident *)(mtag + 1);
1024: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
1025: } else {
1026: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
1.212 perry 1027: IP_FORWARDING, &error);
1.173 jonathan 1028: }
1029: if (sp != NULL) {
1030: /*
1031: * Check security policy against packet attributes.
1032: */
1033: error = ipsec_in_reject(sp, m);
1034: KEY_FREESP(&sp);
1035: } else {
1036: /* XXX error stat??? */
1037: error = EINVAL;
1038: DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/
1039: }
1040: splx(s);
1041: if (error)
1042: goto bad;
1043: }
1044: #endif /* FAST_IPSEC */
1.1 cgd 1045:
1046: /*
1047: * Switch out to protocol's input routine.
1048: */
1.82 aidan 1049: #if IFA_STATS
1.122 itojun 1050: if (ia && ip)
1.155 itojun 1051: ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
1.82 aidan 1052: #endif
1.266 ! thorpej 1053: IP_STATINC(IP_STAT_DELIVERED);
1.89 itojun 1054: {
1055: int off = hlen, nh = ip->ip_p;
1056:
1057: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
1058: return;
1059: }
1.1 cgd 1060: bad:
1061: m_freem(m);
1.135 thorpej 1062: return;
1063:
1064: badcsum:
1.266 ! thorpej 1065: IP_STATINC(IP_STAT_BADSUM);
1.135 thorpej 1066: m_freem(m);
1.1 cgd 1067: }
1068:
1069: /*
1070: * Take incoming datagram fragment and try to
1071: * reassemble it into whole datagram. If a chain for
1072: * reassembly of this datagram already exists, then it
1073: * is given as fp; otherwise have to make a chain.
1074: */
1.50 thorpej 1075: struct mbuf *
1.211 perry 1076: ip_reass(struct ipqent *ipqe, struct ipq *fp, struct ipqhead *ipqhead)
1.1 cgd 1077: {
1.109 augustss 1078: struct mbuf *m = ipqe->ipqe_m;
1079: struct ipqent *nq, *p, *q;
1.25 cgd 1080: struct ip *ip;
1.1 cgd 1081: struct mbuf *t;
1.25 cgd 1082: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.233 tls 1083: int i, next, s;
1.1 cgd 1084:
1.75 thorpej 1085: IPQ_LOCK_CHECK();
1086:
1.1 cgd 1087: /*
1088: * Presence of header sizes in mbufs
1089: * would confuse code below.
1090: */
1091: m->m_data += hlen;
1092: m->m_len -= hlen;
1093:
1.194 jonathan 1094: #ifdef notyet
1095: /* make sure fragment limit is up-to-date */
1096: CHECK_NMBCLUSTER_PARAMS();
1097:
1098: /* If we have too many fragments, drop the older half. */
1099: if (ip_nfrags >= ip_maxfrags)
1100: ip_reass_drophalf(void);
1101: #endif
1102:
1.1 cgd 1103: /*
1.192 jonathan 1104: * We are about to add a fragment; increment frag count.
1105: */
1106: ip_nfrags++;
1.212 perry 1107:
1.192 jonathan 1108: /*
1.1 cgd 1109: * If first fragment to arrive, create a reassembly queue.
1110: */
1111: if (fp == 0) {
1.131 itojun 1112: /*
1113: * Enforce upper bound on number of fragmented packets
1114: * for which we attempt reassembly;
1115: * If maxfrag is 0, never accept fragments.
1116: * If maxfrag is -1, accept all fragments without limitation.
1117: */
1118: if (ip_maxfragpackets < 0)
1119: ;
1120: else if (ip_nfragpackets >= ip_maxfragpackets)
1121: goto dropfrag;
1122: ip_nfragpackets++;
1.50 thorpej 1123: MALLOC(fp, struct ipq *, sizeof (struct ipq),
1124: M_FTABLE, M_NOWAIT);
1125: if (fp == NULL)
1.1 cgd 1126: goto dropfrag;
1.190 jonathan 1127: LIST_INSERT_HEAD(ipqhead, fp, ipq_q);
1.192 jonathan 1128: fp->ipq_nfrags = 1;
1.1 cgd 1129: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 1130: fp->ipq_p = ipqe->ipqe_ip->ip_p;
1131: fp->ipq_id = ipqe->ipqe_ip->ip_id;
1.260 matt 1132: fp->ipq_tos = ipqe->ipqe_ip->ip_tos;
1.148 matt 1133: TAILQ_INIT(&fp->ipq_fragq);
1.25 cgd 1134: fp->ipq_src = ipqe->ipqe_ip->ip_src;
1135: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
1136: p = NULL;
1.1 cgd 1137: goto insert;
1.192 jonathan 1138: } else {
1139: fp->ipq_nfrags++;
1.1 cgd 1140: }
1141:
1142: /*
1143: * Find a segment which begins after this one does.
1144: */
1.148 matt 1145: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1146: p = q, q = TAILQ_NEXT(q, ipqe_q))
1.155 itojun 1147: if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off))
1.1 cgd 1148: break;
1149:
1150: /*
1151: * If there is a preceding segment, it may provide some of
1152: * our data already. If so, drop the data from the incoming
1153: * segment. If it provides all of our data, drop us.
1154: */
1.25 cgd 1155: if (p != NULL) {
1.155 itojun 1156: i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) -
1157: ntohs(ipqe->ipqe_ip->ip_off);
1.1 cgd 1158: if (i > 0) {
1.155 itojun 1159: if (i >= ntohs(ipqe->ipqe_ip->ip_len))
1.1 cgd 1160: goto dropfrag;
1.50 thorpej 1161: m_adj(ipqe->ipqe_m, i);
1.155 itojun 1162: ipqe->ipqe_ip->ip_off =
1163: htons(ntohs(ipqe->ipqe_ip->ip_off) + i);
1164: ipqe->ipqe_ip->ip_len =
1165: htons(ntohs(ipqe->ipqe_ip->ip_len) - i);
1.1 cgd 1166: }
1167: }
1168:
1169: /*
1170: * While we overlap succeeding segments trim them or,
1171: * if they are completely covered, dequeue them.
1172: */
1.155 itojun 1173: for (; q != NULL &&
1174: ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) >
1175: ntohs(q->ipqe_ip->ip_off); q = nq) {
1176: i = (ntohs(ipqe->ipqe_ip->ip_off) +
1177: ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off);
1178: if (i < ntohs(q->ipqe_ip->ip_len)) {
1179: q->ipqe_ip->ip_len =
1180: htons(ntohs(q->ipqe_ip->ip_len) - i);
1181: q->ipqe_ip->ip_off =
1182: htons(ntohs(q->ipqe_ip->ip_off) + i);
1.50 thorpej 1183: m_adj(q->ipqe_m, i);
1.1 cgd 1184: break;
1185: }
1.148 matt 1186: nq = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1187: m_freem(q->ipqe_m);
1.148 matt 1188: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.233 tls 1189: s = splvm();
1.72 thorpej 1190: pool_put(&ipqent_pool, q);
1.233 tls 1191: splx(s);
1.192 jonathan 1192: fp->ipq_nfrags--;
1193: ip_nfrags--;
1.1 cgd 1194: }
1195:
1196: insert:
1197: /*
1198: * Stick new segment in its place;
1199: * check for complete reassembly.
1200: */
1.25 cgd 1201: if (p == NULL) {
1.148 matt 1202: TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
1.25 cgd 1203: } else {
1.148 matt 1204: TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q);
1.25 cgd 1205: }
1.1 cgd 1206: next = 0;
1.148 matt 1207: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1208: p = q, q = TAILQ_NEXT(q, ipqe_q)) {
1.155 itojun 1209: if (ntohs(q->ipqe_ip->ip_off) != next)
1.1 cgd 1210: return (0);
1.155 itojun 1211: next += ntohs(q->ipqe_ip->ip_len);
1.1 cgd 1212: }
1.25 cgd 1213: if (p->ipqe_mff)
1.1 cgd 1214: return (0);
1215:
1216: /*
1.41 thorpej 1217: * Reassembly is complete. Check for a bogus message size and
1218: * concatenate fragments.
1.1 cgd 1219: */
1.148 matt 1220: q = TAILQ_FIRST(&fp->ipq_fragq);
1.25 cgd 1221: ip = q->ipqe_ip;
1.41 thorpej 1222: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
1.266 ! thorpej 1223: IP_STATINC(IP_STAT_TOOLONG);
1.41 thorpej 1224: ip_freef(fp);
1225: return (0);
1226: }
1.50 thorpej 1227: m = q->ipqe_m;
1.1 cgd 1228: t = m->m_next;
1229: m->m_next = 0;
1230: m_cat(m, t);
1.148 matt 1231: nq = TAILQ_NEXT(q, ipqe_q);
1.233 tls 1232: s = splvm();
1.72 thorpej 1233: pool_put(&ipqent_pool, q);
1.233 tls 1234: splx(s);
1.25 cgd 1235: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 1236: t = q->ipqe_m;
1.148 matt 1237: nq = TAILQ_NEXT(q, ipqe_q);
1.233 tls 1238: s = splvm();
1.72 thorpej 1239: pool_put(&ipqent_pool, q);
1.233 tls 1240: splx(s);
1.1 cgd 1241: m_cat(m, t);
1242: }
1.192 jonathan 1243: ip_nfrags -= fp->ipq_nfrags;
1.1 cgd 1244:
1245: /*
1246: * Create header for new ip packet by
1247: * modifying header of first packet;
1248: * dequeue and discard fragment reassembly header.
1249: * Make header visible.
1250: */
1.155 itojun 1251: ip->ip_len = htons(next);
1.25 cgd 1252: ip->ip_src = fp->ipq_src;
1253: ip->ip_dst = fp->ipq_dst;
1254: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1255: FREE(fp, M_FTABLE);
1.131 itojun 1256: ip_nfragpackets--;
1.1 cgd 1257: m->m_len += (ip->ip_hl << 2);
1258: m->m_data -= (ip->ip_hl << 2);
1259: /* some debugging cruft by sklower, below, will go away soon */
1260: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
1.109 augustss 1261: int plen = 0;
1.50 thorpej 1262: for (t = m; t; t = t->m_next)
1263: plen += t->m_len;
1264: m->m_pkthdr.len = plen;
1.213 yamt 1265: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1266: }
1.50 thorpej 1267: return (m);
1.1 cgd 1268:
1269: dropfrag:
1.192 jonathan 1270: if (fp != 0)
1271: fp->ipq_nfrags--;
1272: ip_nfrags--;
1.266 ! thorpej 1273: IP_STATINC(IP_STAT_FRAGDROPPED);
1.1 cgd 1274: m_freem(m);
1.233 tls 1275: s = splvm();
1.72 thorpej 1276: pool_put(&ipqent_pool, ipqe);
1.233 tls 1277: splx(s);
1.1 cgd 1278: return (0);
1279: }
1280:
1281: /*
1282: * Free a fragment reassembly header and all
1283: * associated datagrams.
1284: */
1.8 mycroft 1285: void
1.211 perry 1286: ip_freef(struct ipq *fp)
1.1 cgd 1287: {
1.109 augustss 1288: struct ipqent *q, *p;
1.192 jonathan 1289: u_int nfrags = 0;
1.233 tls 1290: int s;
1.1 cgd 1291:
1.75 thorpej 1292: IPQ_LOCK_CHECK();
1293:
1.148 matt 1294: for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) {
1295: p = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1296: m_freem(q->ipqe_m);
1.192 jonathan 1297: nfrags++;
1.148 matt 1298: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.233 tls 1299: s = splvm();
1.72 thorpej 1300: pool_put(&ipqent_pool, q);
1.233 tls 1301: splx(s);
1.1 cgd 1302: }
1.192 jonathan 1303:
1304: if (nfrags != fp->ipq_nfrags)
1305: printf("ip_freef: nfrags %d != %d\n", fp->ipq_nfrags, nfrags);
1306: ip_nfrags -= nfrags;
1.25 cgd 1307: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1308: FREE(fp, M_FTABLE);
1.131 itojun 1309: ip_nfragpackets--;
1.1 cgd 1310: }
1311:
1312: /*
1.194 jonathan 1313: * IP reassembly TTL machinery for multiplicative drop.
1314: */
1315: static u_int fragttl_histo[(IPFRAGTTL+1)];
1316:
1317:
1318: /*
1319: * Decrement TTL of all reasembly queue entries by `ticks'.
1320: * Count number of distinct fragments (as opposed to partial, fragmented
1321: * datagrams) in the reassembly queue. While we traverse the entire
1322: * reassembly queue, compute and return the median TTL over all fragments.
1323: */
1324: static u_int
1325: ip_reass_ttl_decr(u_int ticks)
1326: {
1.198 matt 1327: u_int nfrags, median, dropfraction, keepfraction;
1.194 jonathan 1328: struct ipq *fp, *nfp;
1.198 matt 1329: int i;
1.212 perry 1330:
1.194 jonathan 1331: nfrags = 0;
1332: memset(fragttl_histo, 0, sizeof fragttl_histo);
1.212 perry 1333:
1.194 jonathan 1334: for (i = 0; i < IPREASS_NHASH; i++) {
1335: for (fp = LIST_FIRST(&ipq[i]); fp != NULL; fp = nfp) {
1336: fp->ipq_ttl = ((fp->ipq_ttl <= ticks) ?
1337: 0 : fp->ipq_ttl - ticks);
1338: nfp = LIST_NEXT(fp, ipq_q);
1339: if (fp->ipq_ttl == 0) {
1.266 ! thorpej 1340: IP_STATINC(IP_STAT_FRAGTIMEOUT);
1.194 jonathan 1341: ip_freef(fp);
1342: } else {
1343: nfrags += fp->ipq_nfrags;
1344: fragttl_histo[fp->ipq_ttl] += fp->ipq_nfrags;
1345: }
1346: }
1347: }
1348:
1349: KASSERT(ip_nfrags == nfrags);
1350:
1351: /* Find median (or other drop fraction) in histogram. */
1352: dropfraction = (ip_nfrags / 2);
1353: keepfraction = ip_nfrags - dropfraction;
1354: for (i = IPFRAGTTL, median = 0; i >= 0; i--) {
1355: median += fragttl_histo[i];
1356: if (median >= keepfraction)
1357: break;
1358: }
1359:
1360: /* Return TTL of median (or other fraction). */
1361: return (u_int)i;
1362: }
1363:
1364: void
1365: ip_reass_drophalf(void)
1366: {
1367:
1368: u_int median_ticks;
1369: /*
1370: * Compute median TTL of all fragments, and count frags
1371: * with that TTL or lower (roughly half of all fragments).
1372: */
1373: median_ticks = ip_reass_ttl_decr(0);
1374:
1375: /* Drop half. */
1376: median_ticks = ip_reass_ttl_decr(median_ticks);
1377:
1378: }
1379:
1380: /*
1.1 cgd 1381: * IP timer processing;
1382: * if a timer expires on a reassembly
1383: * queue, discard it.
1384: */
1.8 mycroft 1385: void
1.211 perry 1386: ip_slowtimo(void)
1.1 cgd 1387: {
1.191 jonathan 1388: static u_int dropscanidx = 0;
1389: u_int i;
1.194 jonathan 1390: u_int median_ttl;
1.24 mycroft 1391: int s = splsoftnet();
1.1 cgd 1392:
1.75 thorpej 1393: IPQ_LOCK();
1.194 jonathan 1394:
1395: /* Age TTL of all fragments by 1 tick .*/
1396: median_ttl = ip_reass_ttl_decr(1);
1397:
1398: /* make sure fragment limit is up-to-date */
1399: CHECK_NMBCLUSTER_PARAMS();
1400:
1401: /* If we have too many fragments, drop the older half. */
1402: if (ip_nfrags > ip_maxfrags)
1403: ip_reass_ttl_decr(median_ttl);
1404:
1.131 itojun 1405: /*
1.194 jonathan 1406: * If we are over the maximum number of fragmented packets
1.131 itojun 1407: * (due to the limit being lowered), drain off
1.190 jonathan 1408: * enough to get down to the new limit. Start draining
1409: * from the reassembly hashqueue most recently drained.
1.131 itojun 1410: */
1411: if (ip_maxfragpackets < 0)
1412: ;
1413: else {
1.190 jonathan 1414: int wrapped = 0;
1415:
1416: i = dropscanidx;
1417: while (ip_nfragpackets > ip_maxfragpackets && wrapped == 0) {
1418: while (LIST_FIRST(&ipq[i]) != NULL)
1419: ip_freef(LIST_FIRST(&ipq[i]));
1420: if (++i >= IPREASS_NHASH) {
1421: i = 0;
1422: }
1423: /*
1424: * Dont scan forever even if fragment counters are
1425: * wrong: stop after scanning entire reassembly queue.
1426: */
1427: if (i == dropscanidx)
1428: wrapped = 1;
1429: }
1430: dropscanidx = i;
1.131 itojun 1431: }
1.75 thorpej 1432: IPQ_UNLOCK();
1.1 cgd 1433: splx(s);
1434: }
1435:
1436: /*
1437: * Drain off all datagram fragments.
1438: */
1.8 mycroft 1439: void
1.211 perry 1440: ip_drain(void)
1.1 cgd 1441: {
1442:
1.75 thorpej 1443: /*
1444: * We may be called from a device's interrupt context. If
1445: * the ipq is already busy, just bail out now.
1446: */
1447: if (ipq_lock_try() == 0)
1448: return;
1449:
1.194 jonathan 1450: /*
1451: * Drop half the total fragments now. If more mbufs are needed,
1452: * we will be called again soon.
1453: */
1454: ip_reass_drophalf();
1.75 thorpej 1455:
1456: IPQ_UNLOCK();
1.1 cgd 1457: }
1458:
1459: /*
1460: * Do option processing on a datagram,
1461: * possibly discarding it if bad options are encountered,
1462: * or forwarding it if source-routed.
1463: * Returns 1 if packet has been forwarded/freed,
1464: * 0 if the packet should be processed further.
1465: */
1.8 mycroft 1466: int
1.211 perry 1467: ip_dooptions(struct mbuf *m)
1.1 cgd 1468: {
1.109 augustss 1469: struct ip *ip = mtod(m, struct ip *);
1470: u_char *cp, *cp0;
1471: struct ip_timestamp *ipt;
1472: struct in_ifaddr *ia;
1.1 cgd 1473: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 1474: struct in_addr dst;
1.1 cgd 1475: n_time ntime;
1476:
1.13 mycroft 1477: dst = ip->ip_dst;
1.1 cgd 1478: cp = (u_char *)(ip + 1);
1479: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
1480: for (; cnt > 0; cnt -= optlen, cp += optlen) {
1481: opt = cp[IPOPT_OPTVAL];
1482: if (opt == IPOPT_EOL)
1483: break;
1484: if (opt == IPOPT_NOP)
1485: optlen = 1;
1486: else {
1.113 itojun 1487: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1488: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1489: goto bad;
1490: }
1.1 cgd 1491: optlen = cp[IPOPT_OLEN];
1.114 itojun 1492: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 1493: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1494: goto bad;
1495: }
1496: }
1497: switch (opt) {
1498:
1499: default:
1500: break;
1501:
1502: /*
1503: * Source routing with record.
1504: * Find interface with current destination address.
1505: * If none on this machine then drop if strictly routed,
1506: * or do nothing if loosely routed.
1507: * Record interface address and bring up next address
1508: * component. If strictly routed make sure next
1509: * address is on directly accessible net.
1510: */
1511: case IPOPT_LSRR:
1512: case IPOPT_SSRR:
1.47 cjs 1513: if (ip_allowsrcrt == 0) {
1514: type = ICMP_UNREACH;
1515: code = ICMP_UNREACH_NET_PROHIB;
1516: goto bad;
1517: }
1.114 itojun 1518: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1519: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1520: goto bad;
1521: }
1.1 cgd 1522: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1523: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1524: goto bad;
1525: }
1526: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 1527: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 1528: if (ia == 0) {
1529: if (opt == IPOPT_SSRR) {
1530: type = ICMP_UNREACH;
1531: code = ICMP_UNREACH_SRCFAIL;
1532: goto bad;
1533: }
1534: /*
1535: * Loose routing, and not at next destination
1536: * yet; nothing to do except forward.
1537: */
1538: break;
1539: }
1540: off--; /* 0 origin */
1.112 sommerfe 1541: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 1542: /*
1543: * End of source route. Should be for us.
1544: */
1545: save_rte(cp, ip->ip_src);
1546: break;
1547: }
1548: /*
1549: * locate outgoing interface
1550: */
1.244 christos 1551: bcopy((void *)(cp + off), (void *)&ipaddr.sin_addr,
1.1 cgd 1552: sizeof(ipaddr.sin_addr));
1.96 thorpej 1553: if (opt == IPOPT_SSRR)
1.196 itojun 1554: ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr)));
1.96 thorpej 1555: else
1.1 cgd 1556: ia = ip_rtaddr(ipaddr.sin_addr);
1557: if (ia == 0) {
1558: type = ICMP_UNREACH;
1559: code = ICMP_UNREACH_SRCFAIL;
1560: goto bad;
1561: }
1562: ip->ip_dst = ipaddr.sin_addr;
1.244 christos 1563: bcopy((void *)&ia->ia_addr.sin_addr,
1564: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1565: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1566: /*
1567: * Let ip_intr's mcast routing check handle mcast pkts
1568: */
1.18 mycroft 1569: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1570: break;
1571:
1572: case IPOPT_RR:
1.114 itojun 1573: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1574: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1575: goto bad;
1576: }
1.1 cgd 1577: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1578: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1579: goto bad;
1580: }
1581: /*
1582: * If no space remains, ignore.
1583: */
1584: off--; /* 0 origin */
1.112 sommerfe 1585: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1586: break;
1.244 christos 1587: bcopy((void *)(&ip->ip_dst), (void *)&ipaddr.sin_addr,
1.1 cgd 1588: sizeof(ipaddr.sin_addr));
1589: /*
1590: * locate outgoing interface; if we're the destination,
1591: * use the incoming interface (should be same).
1592: */
1.96 thorpej 1593: if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))))
1594: == NULL &&
1595: (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
1.1 cgd 1596: type = ICMP_UNREACH;
1597: code = ICMP_UNREACH_HOST;
1598: goto bad;
1599: }
1.244 christos 1600: bcopy((void *)&ia->ia_addr.sin_addr,
1601: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1602: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1603: break;
1604:
1605: case IPOPT_TS:
1606: code = cp - (u_char *)ip;
1607: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1608: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1609: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1610: goto bad;
1.114 itojun 1611: }
1612: if (ipt->ipt_ptr < 5) {
1613: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1614: goto bad;
1615: }
1.15 cgd 1616: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1617: if (++ipt->ipt_oflw == 0) {
1618: code = (u_char *)&ipt->ipt_ptr -
1619: (u_char *)ip;
1.1 cgd 1620: goto bad;
1.114 itojun 1621: }
1.1 cgd 1622: break;
1623: }
1.104 thorpej 1624: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1625: switch (ipt->ipt_flg) {
1626:
1627: case IPOPT_TS_TSONLY:
1628: break;
1629:
1630: case IPOPT_TS_TSANDADDR:
1.66 thorpej 1631: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1632: sizeof(struct in_addr) > ipt->ipt_len) {
1633: code = (u_char *)&ipt->ipt_ptr -
1634: (u_char *)ip;
1.1 cgd 1635: goto bad;
1.114 itojun 1636: }
1.13 mycroft 1637: ipaddr.sin_addr = dst;
1.96 thorpej 1638: ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr),
1639: m->m_pkthdr.rcvif));
1.13 mycroft 1640: if (ia == 0)
1641: continue;
1.104 thorpej 1642: bcopy(&ia->ia_addr.sin_addr,
1643: cp0, sizeof(struct in_addr));
1.1 cgd 1644: ipt->ipt_ptr += sizeof(struct in_addr);
1645: break;
1646:
1647: case IPOPT_TS_PRESPEC:
1.66 thorpej 1648: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1649: sizeof(struct in_addr) > ipt->ipt_len) {
1650: code = (u_char *)&ipt->ipt_ptr -
1651: (u_char *)ip;
1.1 cgd 1652: goto bad;
1.114 itojun 1653: }
1.104 thorpej 1654: bcopy(cp0, &ipaddr.sin_addr,
1.1 cgd 1655: sizeof(struct in_addr));
1.96 thorpej 1656: if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))
1657: == NULL)
1.1 cgd 1658: continue;
1659: ipt->ipt_ptr += sizeof(struct in_addr);
1660: break;
1661:
1662: default:
1.114 itojun 1663: /* XXX can't take &ipt->ipt_flg */
1664: code = (u_char *)&ipt->ipt_ptr -
1665: (u_char *)ip + 1;
1.1 cgd 1666: goto bad;
1667: }
1668: ntime = iptime();
1.107 thorpej 1669: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1.244 christos 1670: memmove((char *)cp + ipt->ipt_ptr - 1, cp0,
1.1 cgd 1671: sizeof(n_time));
1672: ipt->ipt_ptr += sizeof(n_time);
1673: }
1674: }
1675: if (forward) {
1.26 thorpej 1676: if (ip_forwsrcrt == 0) {
1677: type = ICMP_UNREACH;
1678: code = ICMP_UNREACH_SRCFAIL;
1679: goto bad;
1680: }
1.1 cgd 1681: ip_forward(m, 1);
1682: return (1);
1.13 mycroft 1683: }
1684: return (0);
1.1 cgd 1685: bad:
1.13 mycroft 1686: icmp_error(m, type, code, 0, 0);
1.266 ! thorpej 1687: IP_STATINC(IP_STAT_BADOPTIONS);
1.1 cgd 1688: return (1);
1689: }
1690:
1691: /*
1692: * Given address of next destination (final or next hop),
1693: * return internet address info of interface to be used to get there.
1694: */
1695: struct in_ifaddr *
1.211 perry 1696: ip_rtaddr(struct in_addr dst)
1.1 cgd 1697: {
1.249 dyoung 1698: struct rtentry *rt;
1699: union {
1700: struct sockaddr dst;
1701: struct sockaddr_in dst4;
1702: } u;
1703:
1704: sockaddr_in_init(&u.dst4, &dst, 0);
1705:
1706: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL)
1707: return NULL;
1708:
1709: return ifatoia(rt->rt_ifa);
1.1 cgd 1710: }
1711:
1712: /*
1713: * Save incoming source route for use in replies,
1714: * to be picked up later by ip_srcroute if the receiver is interested.
1715: */
1.13 mycroft 1716: void
1.211 perry 1717: save_rte(u_char *option, struct in_addr dst)
1.1 cgd 1718: {
1719: unsigned olen;
1720:
1721: olen = option[IPOPT_OLEN];
1722: #ifdef DIAGNOSTIC
1723: if (ipprintfs)
1.39 christos 1724: printf("save_rte: olen %d\n", olen);
1.89 itojun 1725: #endif /* 0 */
1.1 cgd 1726: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1727: return;
1.244 christos 1728: bcopy((void *)option, (void *)ip_srcrt.srcopt, olen);
1.1 cgd 1729: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1730: ip_srcrt.dst = dst;
1731: }
1732:
1733: /*
1734: * Retrieve incoming source route for use in replies,
1735: * in the same form used by setsockopt.
1736: * The first hop is placed before the options, will be removed later.
1737: */
1738: struct mbuf *
1.211 perry 1739: ip_srcroute(void)
1.1 cgd 1740: {
1.109 augustss 1741: struct in_addr *p, *q;
1742: struct mbuf *m;
1.1 cgd 1743:
1744: if (ip_nhops == 0)
1.237 dyoung 1745: return NULL;
1.1 cgd 1746: m = m_get(M_DONTWAIT, MT_SOOPTS);
1747: if (m == 0)
1.237 dyoung 1748: return NULL;
1.1 cgd 1749:
1.164 matt 1750: MCLAIM(m, &inetdomain.dom_mowner);
1.13 mycroft 1751: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1752:
1753: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1754: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1755: OPTSIZ;
1756: #ifdef DIAGNOSTIC
1757: if (ipprintfs)
1.39 christos 1758: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1759: #endif
1760:
1761: /*
1762: * First save first hop for return route
1763: */
1764: p = &ip_srcrt.route[ip_nhops - 1];
1765: *(mtod(m, struct in_addr *)) = *p--;
1766: #ifdef DIAGNOSTIC
1767: if (ipprintfs)
1.39 christos 1768: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1769: #endif
1770:
1771: /*
1772: * Copy option fields and padding (nop) to mbuf.
1773: */
1774: ip_srcrt.nop = IPOPT_NOP;
1775: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1.244 christos 1776: memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop,
1777: OPTSIZ);
1778: q = (struct in_addr *)(mtod(m, char *) +
1.1 cgd 1779: sizeof(struct in_addr) + OPTSIZ);
1780: #undef OPTSIZ
1781: /*
1782: * Record return path as an IP source route,
1783: * reversing the path (pointers are now aligned).
1784: */
1785: while (p >= ip_srcrt.route) {
1786: #ifdef DIAGNOSTIC
1787: if (ipprintfs)
1.39 christos 1788: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1789: #endif
1790: *q++ = *p--;
1791: }
1792: /*
1793: * Last hop goes to final destination.
1794: */
1795: *q = ip_srcrt.dst;
1796: #ifdef DIAGNOSTIC
1797: if (ipprintfs)
1.39 christos 1798: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1799: #endif
1800: return (m);
1801: }
1802:
1.139 matt 1803: const int inetctlerrmap[PRC_NCMDS] = {
1.256 yamt 1804: [PRC_MSGSIZE] = EMSGSIZE,
1805: [PRC_HOSTDEAD] = EHOSTDOWN,
1806: [PRC_HOSTUNREACH] = EHOSTUNREACH,
1807: [PRC_UNREACH_NET] = EHOSTUNREACH,
1808: [PRC_UNREACH_HOST] = EHOSTUNREACH,
1809: [PRC_UNREACH_PROTOCOL] = ECONNREFUSED,
1810: [PRC_UNREACH_PORT] = ECONNREFUSED,
1811: [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH,
1812: [PRC_PARAMPROB] = ENOPROTOOPT,
1.1 cgd 1813: };
1814:
1815: /*
1816: * Forward a packet. If some error occurs return the sender
1817: * an icmp packet. Note we can't always generate a meaningful
1818: * icmp message because icmp doesn't have a large enough repertoire
1819: * of codes and types.
1820: *
1821: * If not forwarding, just drop the packet. This could be confusing
1822: * if ipforwarding was zero but some routing protocol was advancing
1823: * us as a gateway to somewhere. However, we must let the routing
1824: * protocol deal with that.
1825: *
1826: * The srcrt parameter indicates whether the packet is being forwarded
1827: * via a source route.
1828: */
1.13 mycroft 1829: void
1.211 perry 1830: ip_forward(struct mbuf *m, int srcrt)
1.1 cgd 1831: {
1.109 augustss 1832: struct ip *ip = mtod(m, struct ip *);
1833: struct rtentry *rt;
1.220 christos 1834: int error, type = 0, code = 0, destmtu = 0;
1.1 cgd 1835: struct mbuf *mcopy;
1.13 mycroft 1836: n_long dest;
1.249 dyoung 1837: union {
1838: struct sockaddr dst;
1839: struct sockaddr_in dst4;
1840: } u;
1.164 matt 1841:
1842: /*
1843: * We are now in the output path.
1844: */
1845: MCLAIM(m, &ip_tx_mowner);
1.135 thorpej 1846:
1847: /*
1848: * Clear any in-bound checksum flags for this packet.
1849: */
1850: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1851:
1.13 mycroft 1852: dest = 0;
1.1 cgd 1853: #ifdef DIAGNOSTIC
1.224 joerg 1854: if (ipprintfs) {
1855: printf("forward: src %s ", inet_ntoa(ip->ip_src));
1856: printf("dst %s ttl %x\n", inet_ntoa(ip->ip_dst), ip->ip_ttl);
1857: }
1.1 cgd 1858: #endif
1.93 sommerfe 1859: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.266 ! thorpej 1860: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 1861: m_freem(m);
1862: return;
1863: }
1864: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1865: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1866: return;
1867: }
1868:
1.249 dyoung 1869: sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
1870: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) {
1871: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0);
1872: return;
1.1 cgd 1873: }
1874:
1875: /*
1.34 mycroft 1876: * Save at most 68 bytes of the packet in case
1.1 cgd 1877: * we need to generate an ICMP message to the src.
1.119 itojun 1878: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1879: */
1.155 itojun 1880: mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1.119 itojun 1881: if (mcopy)
1882: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1883:
1.221 christos 1884: ip->ip_ttl -= IPTTLDEC;
1885:
1.1 cgd 1886: /*
1887: * If forwarding packet using same interface that it came in on,
1888: * perhaps should send a redirect to sender to shortcut a hop.
1889: * Only send redirect if source is sending directly to us,
1890: * and if packet was not source routed (or has any options).
1891: * Also, don't send redirect if forwarding using a default route
1892: * or a route modified by a redirect.
1893: */
1894: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1895: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.250 dyoung 1896: !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) &&
1.1 cgd 1897: ipsendredirects && !srcrt) {
1.19 mycroft 1898: if (rt->rt_ifa &&
1899: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1900: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1901: if (rt->rt_flags & RTF_GATEWAY)
1902: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1903: else
1904: dest = ip->ip_dst.s_addr;
1905: /*
1906: * Router requirements says to only send host
1907: * redirects.
1908: */
1909: type = ICMP_REDIRECT;
1910: code = ICMP_REDIRECT_HOST;
1.1 cgd 1911: #ifdef DIAGNOSTIC
1.77 thorpej 1912: if (ipprintfs)
1913: printf("redirect (%d) to %x\n", code,
1914: (u_int32_t)dest);
1.1 cgd 1915: #endif
1916: }
1917: }
1918:
1.238 dyoung 1919: error = ip_output(m, NULL, &ipforward_rt,
1.173 jonathan 1920: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
1.174 itojun 1921: (struct ip_moptions *)NULL, (struct socket *)NULL);
1.173 jonathan 1922:
1.1 cgd 1923: if (error)
1.266 ! thorpej 1924: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 1925: else {
1.266 ! thorpej 1926: uint64_t *ips = IP_STAT_GETREF();
! 1927: ips[IP_STAT_FORWARD]++;
! 1928: if (type) {
! 1929: ips[IP_STAT_REDIRECTSENT]++;
! 1930: IP_STAT_PUTREF();
! 1931: } else {
! 1932: IP_STAT_PUTREF();
1.63 matt 1933: if (mcopy) {
1934: #ifdef GATEWAY
1.64 thorpej 1935: if (mcopy->m_flags & M_CANFASTFWD)
1936: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1937: #endif
1.1 cgd 1938: m_freem(mcopy);
1.63 matt 1939: }
1.1 cgd 1940: return;
1941: }
1942: }
1943: if (mcopy == NULL)
1944: return;
1.13 mycroft 1945:
1.1 cgd 1946: switch (error) {
1947:
1948: case 0: /* forwarded, but need redirect */
1949: /* type, code set above */
1950: break;
1951:
1952: case ENETUNREACH: /* shouldn't happen, checked above */
1953: case EHOSTUNREACH:
1954: case ENETDOWN:
1955: case EHOSTDOWN:
1956: default:
1957: type = ICMP_UNREACH;
1958: code = ICMP_UNREACH_HOST;
1959: break;
1960:
1961: case EMSGSIZE:
1962: type = ICMP_UNREACH;
1963: code = ICMP_UNREACH_NEEDFRAG;
1.263 cube 1964:
1.261 dyoung 1965: if ((rt = rtcache_validate(&ipforward_rt)) != NULL) {
1.263 cube 1966:
1967: #if defined(IPSEC) || defined(FAST_IPSEC)
1968: /*
1969: * If the packet is routed over IPsec tunnel, tell the
1970: * originator the tunnel MTU.
1971: * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1972: * XXX quickhack!!!
1973: */
1974:
1.89 itojun 1975: struct secpolicy *sp;
1976: int ipsecerror;
1.95 itojun 1977: size_t ipsechdr;
1.89 itojun 1978: struct route *ro;
1979:
1980: sp = ipsec4_getpolicybyaddr(mcopy,
1.170 itojun 1981: IPSEC_DIR_OUTBOUND, IP_FORWARDING,
1982: &ipsecerror);
1.263 cube 1983: #endif
1.89 itojun 1984:
1.263 cube 1985: destmtu = rt->rt_ifp->if_mtu;
1986: #if defined(IPSEC) || defined(FAST_IPSEC)
1987: if (sp != NULL) {
1.89 itojun 1988: /* count IPsec header size */
1.95 itojun 1989: ipsechdr = ipsec4_hdrsiz(mcopy,
1.170 itojun 1990: IPSEC_DIR_OUTBOUND, NULL);
1.89 itojun 1991:
1992: /*
1993: * find the correct route for outer IPv4
1994: * header, compute tunnel MTU.
1995: */
1.220 christos 1996:
1.89 itojun 1997: if (sp->req != NULL
1.95 itojun 1998: && sp->req->sav != NULL
1999: && sp->req->sav->sah != NULL) {
2000: ro = &sp->req->sav->sah->sa_route;
1.257 dyoung 2001: if (rt && rt->rt_ifp) {
1.220 christos 2002: destmtu =
1.257 dyoung 2003: rt->rt_rmx.rmx_mtu ?
2004: rt->rt_rmx.rmx_mtu :
2005: rt->rt_ifp->if_mtu;
1.220 christos 2006: destmtu -= ipsechdr;
1.89 itojun 2007: }
2008: }
2009:
1.173 jonathan 2010: #ifdef IPSEC
1.89 itojun 2011: key_freesp(sp);
1.173 jonathan 2012: #else
2013: KEY_FREESP(&sp);
2014: #endif
1.89 itojun 2015: }
1.263 cube 2016: #endif /*defined(IPSEC) || defined(FAST_IPSEC)*/
1.89 itojun 2017: }
1.266 ! thorpej 2018: IP_STATINC(IP_STAT_CANTFRAG);
1.1 cgd 2019: break;
2020:
2021: case ENOBUFS:
1.143 itojun 2022: #if 1
2023: /*
2024: * a router should not generate ICMP_SOURCEQUENCH as
2025: * required in RFC1812 Requirements for IP Version 4 Routers.
2026: * source quench could be a big problem under DoS attacks,
1.149 wiz 2027: * or if the underlying interface is rate-limited.
1.143 itojun 2028: */
2029: if (mcopy)
2030: m_freem(mcopy);
2031: return;
2032: #else
1.1 cgd 2033: type = ICMP_SOURCEQUENCH;
2034: code = 0;
2035: break;
1.143 itojun 2036: #endif
1.1 cgd 2037: }
1.220 christos 2038: icmp_error(mcopy, type, code, dest, destmtu);
1.44 thorpej 2039: }
2040:
2041: void
1.211 perry 2042: ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
2043: struct mbuf *m)
1.44 thorpej 2044: {
2045:
2046: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
2047: struct timeval tv;
2048:
2049: microtime(&tv);
1.244 christos 2050: *mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1.44 thorpej 2051: SCM_TIMESTAMP, SOL_SOCKET);
2052: if (*mp)
2053: mp = &(*mp)->m_next;
2054: }
2055: if (inp->inp_flags & INP_RECVDSTADDR) {
1.244 christos 2056: *mp = sbcreatecontrol((void *) &ip->ip_dst,
1.44 thorpej 2057: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
2058: if (*mp)
2059: mp = &(*mp)->m_next;
2060: }
2061: #ifdef notyet
2062: /*
2063: * XXX
2064: * Moving these out of udp_input() made them even more broken
2065: * than they already were.
2066: * - fenner@parc.xerox.com
2067: */
2068: /* options were tossed already */
2069: if (inp->inp_flags & INP_RECVOPTS) {
1.244 christos 2070: *mp = sbcreatecontrol((void *) opts_deleted_above,
1.44 thorpej 2071: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
2072: if (*mp)
2073: mp = &(*mp)->m_next;
2074: }
2075: /* ip_srcroute doesn't do what we want here, need to fix */
2076: if (inp->inp_flags & INP_RECVRETOPTS) {
1.244 christos 2077: *mp = sbcreatecontrol((void *) ip_srcroute(),
1.44 thorpej 2078: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
2079: if (*mp)
2080: mp = &(*mp)->m_next;
2081: }
2082: #endif
2083: if (inp->inp_flags & INP_RECVIF) {
2084: struct sockaddr_dl sdl;
2085:
1.252 dyoung 2086: sockaddr_dl_init(&sdl, sizeof(sdl),
2087: (m->m_pkthdr.rcvif != NULL)
2088: ? m->m_pkthdr.rcvif->if_index
2089: : 0,
2090: 0, NULL, 0, NULL, 0);
1.251 dyoung 2091: *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP);
1.44 thorpej 2092: if (*mp)
2093: mp = &(*mp)->m_next;
2094: }
1.13 mycroft 2095: }
2096:
1.189 atatat 2097: /*
1.228 elad 2098: * sysctl helper routine for net.inet.ip.forwsrcrt.
2099: */
2100: static int
2101: sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS)
2102: {
2103: int error, tmp;
2104: struct sysctlnode node;
2105:
2106: node = *rnode;
2107: tmp = ip_forwsrcrt;
2108: node.sysctl_data = &tmp;
2109: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2110: if (error || newp == NULL)
2111: return (error);
2112:
1.230 elad 2113: if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT,
1.232 elad 2114: 0, NULL, NULL, NULL))
1.228 elad 2115: return (EPERM);
2116:
2117: ip_forwsrcrt = tmp;
2118:
2119: return (0);
2120: }
2121:
2122: /*
1.189 atatat 2123: * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the
2124: * range of the new value and tweaks timers if it changes.
2125: */
2126: static int
2127: sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS)
1.13 mycroft 2128: {
1.189 atatat 2129: int error, tmp;
2130: struct sysctlnode node;
2131:
2132: node = *rnode;
2133: tmp = ip_mtudisc_timeout;
2134: node.sysctl_data = &tmp;
2135: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2136: if (error || newp == NULL)
2137: return (error);
2138: if (tmp < 0)
2139: return (EINVAL);
1.52 thorpej 2140:
1.189 atatat 2141: ip_mtudisc_timeout = tmp;
2142: rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout);
2143:
2144: return (0);
2145: }
1.54 lukem 2146:
1.65 matt 2147: #ifdef GATEWAY
1.189 atatat 2148: /*
1.247 liamjfoy 2149: * sysctl helper routine for net.inet.ip.maxflows.
1.189 atatat 2150: */
2151: static int
2152: sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS)
2153: {
2154: int s;
1.67 thorpej 2155:
1.217 atatat 2156: s = sysctl_lookup(SYSCTLFN_CALL(rnode));
1.247 liamjfoy 2157: if (s || newp == NULL)
1.189 atatat 2158: return (s);
1.212 perry 2159:
1.189 atatat 2160: s = splsoftnet();
1.265 thorpej 2161: ipflow_prune();
1.189 atatat 2162: splx(s);
1.144 martin 2163:
1.189 atatat 2164: return (0);
2165: }
1.248 liamjfoy 2166:
2167: static int
2168: sysctl_net_inet_ip_hashsize(SYSCTLFN_ARGS)
2169: {
2170: int error, tmp;
2171: struct sysctlnode node;
2172:
2173: node = *rnode;
2174: tmp = ip_hashsize;
2175: node.sysctl_data = &tmp;
2176: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2177: if (error || newp == NULL)
2178: return (error);
2179:
2180: if ((tmp & (tmp - 1)) == 0 && tmp != 0) {
2181: /*
2182: * Can only fail due to malloc()
2183: */
2184: if (ipflow_invalidate_all(tmp))
2185: return ENOMEM;
2186: } else {
2187: /*
2188: * EINVAL if not a power of 2
2189: */
2190: return EINVAL;
2191: }
2192:
2193: return (0);
2194: }
1.189 atatat 2195: #endif /* GATEWAY */
1.117 tron 2196:
1.266 ! thorpej 2197: static void
! 2198: ipstat_convert_to_user_cb(void *v1, void *v2, struct cpu_info *ci)
! 2199: {
! 2200: uint64_t *ipsc = v1;
! 2201: uint64_t *ips = v2;
! 2202: u_int i;
! 2203:
! 2204: for (i = 0; i < IP_NSTATS; i++)
! 2205: ips[i] += ipsc[i];
! 2206: }
! 2207:
! 2208: static void
! 2209: ipstat_convert_to_user(uint64_t *ips)
! 2210: {
! 2211:
! 2212: memset(ips, 0, sizeof(uint64_t) * IP_NSTATS);
! 2213: percpu_foreach(ipstat_percpu, ipstat_convert_to_user_cb, ips);
! 2214: }
! 2215:
! 2216: static int
! 2217: sysctl_net_inet_ip_stats(SYSCTLFN_ARGS)
! 2218: {
! 2219: struct sysctlnode node;
! 2220: uint64_t ips[IP_NSTATS];
! 2221:
! 2222: ipstat_convert_to_user(ips);
! 2223: node = *rnode;
! 2224: node.sysctl_data = ips;
! 2225: node.sysctl_size = sizeof(ips);
! 2226: return (sysctl_lookup(SYSCTLFN_CALL(&node)));
! 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