Annotation of src/sys/netinet/ip_input.c, Revision 1.256
1.256 ! yamt 1: /* $NetBSD: ip_input.c,v 1.255 2007/11/09 06:59:33 kefren 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>
1.256 ! yamt 101: __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.255 2007/11/09 06:59:33 kefren 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>
138: #include <netinet/ip_icmp.h>
1.89 itojun 139: /* just for gif_ttl */
140: #include <netinet/in_gif.h>
141: #include "gif.h"
1.144 martin 142: #include <net/if_gre.h>
143: #include "gre.h"
1.111 jdolecek 144:
145: #ifdef MROUTING
146: #include <netinet/ip_mroute.h>
147: #endif
1.89 itojun 148:
149: #ifdef IPSEC
150: #include <netinet6/ipsec.h>
151: #include <netkey/key.h>
152: #endif
1.173 jonathan 153: #ifdef FAST_IPSEC
154: #include <netipsec/ipsec.h>
155: #include <netipsec/key.h>
156: #endif /* FAST_IPSEC*/
1.44 thorpej 157:
1.1 cgd 158: #ifndef IPFORWARDING
159: #ifdef GATEWAY
160: #define IPFORWARDING 1 /* forward IP packets not for us */
161: #else /* GATEWAY */
162: #define IPFORWARDING 0 /* don't forward IP packets not for us */
163: #endif /* GATEWAY */
164: #endif /* IPFORWARDING */
165: #ifndef IPSENDREDIRECTS
166: #define IPSENDREDIRECTS 1
167: #endif
1.26 thorpej 168: #ifndef IPFORWSRCRT
1.47 cjs 169: #define IPFORWSRCRT 1 /* forward source-routed packets */
170: #endif
171: #ifndef IPALLOWSRCRT
1.48 mrg 172: #define IPALLOWSRCRT 1 /* allow source-routed packets */
1.26 thorpej 173: #endif
1.53 kml 174: #ifndef IPMTUDISC
1.153 itojun 175: #define IPMTUDISC 1
1.53 kml 176: #endif
1.60 kml 177: #ifndef IPMTUDISCTIMEOUT
1.61 kml 178: #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */
1.60 kml 179: #endif
1.53 kml 180:
1.27 thorpej 181: /*
182: * Note: DIRECTED_BROADCAST is handled this way so that previous
183: * configuration using this option will Just Work.
184: */
185: #ifndef IPDIRECTEDBCAST
186: #ifdef DIRECTED_BROADCAST
187: #define IPDIRECTEDBCAST 1
188: #else
189: #define IPDIRECTEDBCAST 0
190: #endif /* DIRECTED_BROADCAST */
191: #endif /* IPDIRECTEDBCAST */
1.1 cgd 192: int ipforwarding = IPFORWARDING;
193: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 194: int ip_defttl = IPDEFTTL;
1.26 thorpej 195: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 196: int ip_directedbcast = IPDIRECTEDBCAST;
1.47 cjs 197: int ip_allowsrcrt = IPALLOWSRCRT;
1.53 kml 198: int ip_mtudisc = IPMTUDISC;
1.156 itojun 199: int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
1.1 cgd 200: #ifdef DIAGNOSTIC
201: int ipprintfs = 0;
202: #endif
1.184 jonathan 203:
204: int ip_do_randomid = 0;
205:
1.165 christos 206: /*
207: * XXX - Setting ip_checkinterface mostly implements the receive side of
208: * the Strong ES model described in RFC 1122, but since the routing table
209: * and transmit implementation do not implement the Strong ES model,
210: * setting this to 1 results in an odd hybrid.
211: *
212: * XXX - ip_checkinterface currently must be disabled if you use ipnat
213: * to translate the destination address to another local interface.
214: *
215: * XXX - ip_checkinterface must be disabled if you add IP aliases
216: * to the loopback interface instead of the interface where the
217: * packets for those addresses are received.
218: */
219: int ip_checkinterface = 0;
220:
1.1 cgd 221:
1.60 kml 222: struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
223:
1.1 cgd 224: int ipqmaxlen = IFQ_MAXLEN;
1.150 matt 225: u_long in_ifaddrhash; /* size of hash table - 1 */
226: int in_ifaddrentries; /* total number of addrs */
1.212 perry 227: struct in_ifaddrhead in_ifaddrhead;
1.57 tls 228: struct in_ifaddrhashhead *in_ifaddrhashtbl;
1.166 matt 229: u_long in_multihash; /* size of hash table - 1 */
230: int in_multientries; /* total number of addrs */
231: struct in_multihashhead *in_multihashtbl;
1.13 mycroft 232: struct ifqueue ipintrq;
1.63 matt 233: struct ipstat ipstat;
1.183 jonathan 234: uint16_t ip_id;
1.75 thorpej 235:
1.121 thorpej 236: #ifdef PFIL_HOOKS
237: struct pfil_head inet_pfil_hook;
238: #endif
239:
1.194 jonathan 240: /*
241: * Cached copy of nmbclusters. If nbclusters is different,
242: * recalculate IP parameters derived from nmbclusters.
243: */
244: static int ip_nmbclusters; /* copy of nmbclusters */
1.210 perry 245: static void ip_nmbclusters_changed(void); /* recalc limits */
1.194 jonathan 246:
1.195 thorpej 247: #define CHECK_NMBCLUSTER_PARAMS() \
248: do { \
249: if (__predict_false(ip_nmbclusters != nmbclusters)) \
250: ip_nmbclusters_changed(); \
251: } while (/*CONSTCOND*/0)
1.194 jonathan 252:
1.190 jonathan 253: /* IP datagram reassembly queues (hashed) */
254: #define IPREASS_NHASH_LOG2 6
255: #define IPREASS_NHASH (1 << IPREASS_NHASH_LOG2)
256: #define IPREASS_HMASK (IPREASS_NHASH - 1)
257: #define IPREASS_HASH(x,y) \
258: (((((x) & 0xF) | ((((x) >> 8) & 0xF) << 4)) ^ (y)) & IPREASS_HMASK)
259: struct ipqhead ipq[IPREASS_NHASH];
1.75 thorpej 260: int ipq_locked;
1.212 perry 261: static int ip_nfragpackets; /* packets in reass queue */
1.194 jonathan 262: static int ip_nfrags; /* total fragments in reass queues */
263:
264: int ip_maxfragpackets = 200; /* limit on packets. XXX sysctl */
265: int ip_maxfrags; /* limit on fragments. XXX sysctl */
266:
267:
268: /*
269: * Additive-Increase/Multiplicative-Decrease (AIMD) strategy for
270: * IP reassembly queue buffer managment.
1.212 perry 271: *
1.194 jonathan 272: * We keep a count of total IP fragments (NB: not fragmented packets!)
273: * awaiting reassembly (ip_nfrags) and a limit (ip_maxfrags) on fragments.
274: * If ip_nfrags exceeds ip_maxfrags the limit, we drop half the
275: * total fragments in reassembly queues.This AIMD policy avoids
276: * repeatedly deleting single packets under heavy fragmentation load
277: * (e.g., from lossy NFS peers).
278: */
1.212 perry 279: static u_int ip_reass_ttl_decr(u_int ticks);
1.210 perry 280: static void ip_reass_drophalf(void);
1.194 jonathan 281:
1.75 thorpej 282:
1.223 perry 283: static inline int ipq_lock_try(void);
284: static inline void ipq_unlock(void);
1.75 thorpej 285:
1.223 perry 286: static inline int
1.211 perry 287: ipq_lock_try(void)
1.75 thorpej 288: {
289: int s;
290:
1.132 thorpej 291: /*
1.149 wiz 292: * Use splvm() -- we're blocking things that would cause
1.132 thorpej 293: * mbuf allocation.
294: */
295: s = splvm();
1.75 thorpej 296: if (ipq_locked) {
297: splx(s);
298: return (0);
299: }
300: ipq_locked = 1;
301: splx(s);
302: return (1);
303: }
304:
1.223 perry 305: static inline void
1.211 perry 306: ipq_unlock(void)
1.75 thorpej 307: {
308: int s;
309:
1.132 thorpej 310: s = splvm();
1.75 thorpej 311: ipq_locked = 0;
312: splx(s);
313: }
314:
315: #ifdef DIAGNOSTIC
316: #define IPQ_LOCK() \
317: do { \
318: if (ipq_lock_try() == 0) { \
319: printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \
320: panic("ipq_lock"); \
321: } \
1.159 perry 322: } while (/*CONSTCOND*/ 0)
1.75 thorpej 323: #define IPQ_LOCK_CHECK() \
324: do { \
325: if (ipq_locked == 0) { \
326: printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \
327: panic("ipq lock check"); \
328: } \
1.159 perry 329: } while (/*CONSTCOND*/ 0)
1.75 thorpej 330: #else
331: #define IPQ_LOCK() (void) ipq_lock_try()
332: #define IPQ_LOCK_CHECK() /* nothing */
333: #endif
334:
335: #define IPQ_UNLOCK() ipq_unlock()
1.1 cgd 336:
1.246 ad 337: POOL_INIT(inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl", NULL,
338: IPL_SOFTNET);
339: POOL_INIT(ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", NULL,
340: IPL_VM);
1.72 thorpej 341:
1.135 thorpej 342: #ifdef INET_CSUM_COUNTERS
343: #include <sys/device.h>
344:
345: struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
346: NULL, "inet", "hwcsum bad");
347: struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
348: NULL, "inet", "hwcsum ok");
349: struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
350: NULL, "inet", "swcsum");
351:
352: #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++
353:
1.201 matt 354: EVCNT_ATTACH_STATIC(ip_hwcsum_bad);
355: EVCNT_ATTACH_STATIC(ip_hwcsum_ok);
356: EVCNT_ATTACH_STATIC(ip_swcsum);
357:
1.135 thorpej 358: #else
359:
360: #define INET_CSUM_COUNTER_INCR(ev) /* nothing */
361:
362: #endif /* INET_CSUM_COUNTERS */
363:
1.1 cgd 364: /*
365: * We need to save the IP options in case a protocol wants to respond
366: * to an incoming packet over the same route if the packet got here
367: * using IP source routing. This allows connection establishment and
368: * maintenance when the remote end is on a network that is not known
369: * to us.
370: */
371: int ip_nhops = 0;
372: static struct ip_srcrt {
373: struct in_addr dst; /* final destination */
374: char nop; /* one NOP to align */
375: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
376: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
377: } ip_srcrt;
378:
1.210 perry 379: static void save_rte(u_char *, struct in_addr);
1.35 mycroft 380:
1.164 matt 381: #ifdef MBUFTRACE
1.234 dogcow 382: struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx");
383: struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx");
1.164 matt 384: #endif
385:
1.1 cgd 386: /*
1.194 jonathan 387: * Compute IP limits derived from the value of nmbclusters.
388: */
389: static void
390: ip_nmbclusters_changed(void)
391: {
392: ip_maxfrags = nmbclusters / 4;
393: ip_nmbclusters = nmbclusters;
394: }
395:
396: /*
1.1 cgd 397: * IP initialization: fill in IP protocol switch table.
398: * All protocols not implemented in kernel go to raw IP protocol handler.
399: */
1.8 mycroft 400: void
1.211 perry 401: ip_init(void)
1.1 cgd 402: {
1.199 matt 403: const struct protosw *pr;
1.109 augustss 404: int i;
1.1 cgd 405:
406: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
407: if (pr == 0)
408: panic("ip_init");
409: for (i = 0; i < IPPROTO_MAX; i++)
410: ip_protox[i] = pr - inetsw;
411: for (pr = inetdomain.dom_protosw;
412: pr < inetdomain.dom_protoswNPROTOSW; pr++)
413: if (pr->pr_domain->dom_family == PF_INET &&
414: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
415: ip_protox[pr->pr_protocol] = pr - inetsw;
1.192 jonathan 416:
1.190 jonathan 417: for (i = 0; i < IPREASS_NHASH; i++)
418: LIST_INIT(&ipq[i]);
419:
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);
1.120 ad 426: in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IFADDR,
427: M_WAITOK, &in_ifaddrhash);
1.166 matt 428: in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IPMADDR,
429: M_WAITOK, &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 */
1.1 cgd 449: }
450:
1.229 christos 451: struct sockaddr_in ipaddr = {
452: .sin_len = sizeof(ipaddr),
453: .sin_family = AF_INET,
454: };
1.1 cgd 455: struct route ipforward_rt;
456:
457: /*
1.89 itojun 458: * IP software interrupt routine
459: */
460: void
1.211 perry 461: ipintr(void)
1.89 itojun 462: {
463: int s;
464: struct mbuf *m;
465:
1.241 ad 466: while (!IF_IS_EMPTY(&ipintrq)) {
1.132 thorpej 467: s = splnet();
1.89 itojun 468: IF_DEQUEUE(&ipintrq, m);
469: splx(s);
470: if (m == 0)
471: return;
472: ip_input(m);
473: }
474: }
475:
476: /*
1.1 cgd 477: * Ip input routine. Checksum and byte swap header. If fragmented
478: * try to reassemble. Process options. Pass to next level.
479: */
1.8 mycroft 480: void
1.89 itojun 481: ip_input(struct mbuf *m)
1.1 cgd 482: {
1.109 augustss 483: struct ip *ip = NULL;
484: struct ipq *fp;
485: struct in_ifaddr *ia;
486: struct ifaddr *ifa;
1.25 cgd 487: struct ipqent *ipqe;
1.89 itojun 488: int hlen = 0, mff, len;
1.100 itojun 489: int downmatch;
1.165 christos 490: int checkif;
1.169 itojun 491: int srcrt = 0;
1.233 tls 492: int s;
1.190 jonathan 493: u_int hash;
1.173 jonathan 494: #ifdef FAST_IPSEC
495: struct m_tag *mtag;
496: struct tdb_ident *tdbi;
497: struct secpolicy *sp;
1.233 tls 498: int error;
1.173 jonathan 499: #endif /* FAST_IPSEC */
1.1 cgd 500:
1.164 matt 501: MCLAIM(m, &ip_rx_mowner);
1.1 cgd 502: #ifdef DIAGNOSTIC
503: if ((m->m_flags & M_PKTHDR) == 0)
504: panic("ipintr no HDR");
1.89 itojun 505: #endif
1.164 matt 506:
1.1 cgd 507: /*
508: * If no IP addresses have been set yet but the interfaces
509: * are receiving, can't do anything with incoming packets yet.
510: */
1.181 jonathan 511: if (TAILQ_FIRST(&in_ifaddrhead) == 0)
1.1 cgd 512: goto bad;
513: ipstat.ips_total++;
1.154 thorpej 514: /*
515: * If the IP header is not aligned, slurp it up into a new
516: * mbuf with space for link headers, in the event we forward
517: * it. Otherwise, if it is aligned, make sure the entire
518: * base IP header is in the first mbuf of the chain.
519: */
1.244 christos 520: if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
1.154 thorpej 521: if ((m = m_copyup(m, sizeof(struct ip),
522: (max_linkhdr + 3) & ~3)) == NULL) {
523: /* XXXJRT new stat, please */
524: ipstat.ips_toosmall++;
525: return;
526: }
527: } else if (__predict_false(m->m_len < sizeof (struct ip))) {
528: if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
529: ipstat.ips_toosmall++;
530: return;
531: }
1.1 cgd 532: }
533: ip = mtod(m, struct ip *);
1.13 mycroft 534: if (ip->ip_v != IPVERSION) {
535: ipstat.ips_badvers++;
536: goto bad;
537: }
1.1 cgd 538: hlen = ip->ip_hl << 2;
539: if (hlen < sizeof(struct ip)) { /* minimum header length */
540: ipstat.ips_badhlen++;
541: goto bad;
542: }
543: if (hlen > m->m_len) {
544: if ((m = m_pullup(m, hlen)) == 0) {
545: ipstat.ips_badhlen++;
1.89 itojun 546: return;
1.1 cgd 547: }
548: ip = mtod(m, struct ip *);
549: }
1.98 thorpej 550:
1.85 hwr 551: /*
1.99 thorpej 552: * RFC1122: packets with a multicast source address are
1.98 thorpej 553: * not allowed.
1.85 hwr 554: */
555: if (IN_MULTICAST(ip->ip_src.s_addr)) {
1.130 itojun 556: ipstat.ips_badaddr++;
1.85 hwr 557: goto bad;
1.129 itojun 558: }
559:
560: /* 127/8 must not appear on wire - RFC1122 */
561: if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
562: (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
1.130 itojun 563: if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
564: ipstat.ips_badaddr++;
1.129 itojun 565: goto bad;
1.130 itojun 566: }
1.85 hwr 567: }
568:
1.135 thorpej 569: switch (m->m_pkthdr.csum_flags &
1.137 thorpej 570: ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) |
1.135 thorpej 571: M_CSUM_IPv4_BAD)) {
572: case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
573: INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
574: goto badcsum;
575:
576: case M_CSUM_IPv4:
577: /* Checksum was okay. */
578: INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
579: break;
580:
581: default:
1.206 thorpej 582: /*
583: * Must compute it ourselves. Maybe skip checksum on
584: * loopback interfaces.
585: */
586: if (__predict_true(!(m->m_pkthdr.rcvif->if_flags &
587: IFF_LOOPBACK) || ip_do_loopback_cksum)) {
588: INET_CSUM_COUNTER_INCR(&ip_swcsum);
589: if (in_cksum(m, hlen) != 0)
590: goto badcsum;
591: }
1.135 thorpej 592: break;
1.1 cgd 593: }
594:
1.121 thorpej 595: /* Retrieve the packet length. */
596: len = ntohs(ip->ip_len);
1.81 proff 597:
598: /*
599: * Check for additional length bogosity
600: */
1.84 proff 601: if (len < hlen) {
1.81 proff 602: ipstat.ips_badlen++;
603: goto bad;
604: }
1.1 cgd 605:
606: /*
607: * Check that the amount of data in the buffers
608: * is as at least much as the IP header would have us expect.
609: * Trim mbufs if longer than we expect.
610: * Drop packet if shorter than we expect.
611: */
1.35 mycroft 612: if (m->m_pkthdr.len < len) {
1.1 cgd 613: ipstat.ips_tooshort++;
614: goto bad;
615: }
1.35 mycroft 616: if (m->m_pkthdr.len > len) {
1.1 cgd 617: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 618: m->m_len = len;
619: m->m_pkthdr.len = len;
1.1 cgd 620: } else
1.35 mycroft 621: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 622: }
623:
1.193 scw 624: #if defined(IPSEC)
1.149 wiz 625: /* ipflow (IP fast forwarding) is not compatible with IPsec. */
1.94 itojun 626: m->m_flags &= ~M_CANFASTFWD;
627: #else
1.64 thorpej 628: /*
629: * Assume that we can create a fast-forward IP flow entry
630: * based on this packet.
631: */
632: m->m_flags |= M_CANFASTFWD;
1.94 itojun 633: #endif
1.64 thorpej 634:
1.36 mrg 635: #ifdef PFIL_HOOKS
1.33 mrg 636: /*
1.64 thorpej 637: * Run through list of hooks for input packets. If there are any
638: * filters which require that additional packets in the flow are
639: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
640: * Note that filters must _never_ set this flag, as another filter
641: * in the list may have previously cleared it.
1.33 mrg 642: */
1.127 itojun 643: /*
644: * let ipfilter look at packet on the wire,
645: * not the decapsulated packet.
646: */
647: #ifdef IPSEC
1.136 itojun 648: if (!ipsec_getnhist(m))
1.186 scw 649: #elif defined(FAST_IPSEC)
650: if (!ipsec_indone(m))
1.127 itojun 651: #else
652: if (1)
653: #endif
654: {
1.169 itojun 655: struct in_addr odst;
656:
657: odst = ip->ip_dst;
1.127 itojun 658: if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif,
1.168 itojun 659: PFIL_IN) != 0)
660: return;
1.127 itojun 661: if (m == NULL)
662: return;
663: ip = mtod(m, struct ip *);
1.142 darrenr 664: hlen = ip->ip_hl << 2;
1.205 darrenr 665: /*
666: * XXX The setting of "srcrt" here is to prevent ip_forward()
667: * from generating ICMP redirects for packets that have
668: * been redirected by a hook back out on to the same LAN that
669: * they came from and is not an indication that the packet
670: * is being inffluenced by source routing options. This
671: * allows things like
672: * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp"
673: * where tlp0 is both on the 1.1.1.0/24 network and is the
674: * default route for hosts on 1.1.1.0/24. Of course this
675: * also requires a "map tlp0 ..." to complete the story.
676: * One might argue whether or not this kind of network config.
1.212 perry 677: * should be supported in this manner...
1.205 darrenr 678: */
1.169 itojun 679: srcrt = (odst.s_addr != ip->ip_dst.s_addr);
1.127 itojun 680: }
1.36 mrg 681: #endif /* PFIL_HOOKS */
1.123 thorpej 682:
683: #ifdef ALTQ
684: /* XXX Temporary until ALTQ is changed to use a pfil hook */
685: if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) {
686: /* packet dropped by traffic conditioner */
687: return;
688: }
689: #endif
1.121 thorpej 690:
691: /*
1.1 cgd 692: * Process options and, if not destined for us,
693: * ship it on. ip_dooptions returns 1 when an
694: * error was detected (causing an icmp message
695: * to be sent and the original packet to be freed).
696: */
697: ip_nhops = 0; /* for source routed packets */
698: if (hlen > sizeof (struct ip) && ip_dooptions(m))
1.89 itojun 699: return;
1.1 cgd 700:
701: /*
1.165 christos 702: * Enable a consistency check between the destination address
703: * and the arrival interface for a unicast packet (the RFC 1122
704: * strong ES model) if IP forwarding is disabled and the packet
705: * is not locally generated.
706: *
707: * XXX - Checking also should be disabled if the destination
708: * address is ipnat'ed to a different interface.
709: *
710: * XXX - Checking is incompatible with IP aliases added
711: * to the loopback interface instead of the interface where
712: * the packets are received.
713: *
714: * XXX - We need to add a per ifaddr flag for this so that
715: * we get finer grain control.
716: */
717: checkif = ip_checkinterface && (ipforwarding == 0) &&
718: (m->m_pkthdr.rcvif != NULL) &&
719: ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0);
720:
721: /*
1.1 cgd 722: * Check our list of addresses, to see if the packet is for us.
1.100 itojun 723: *
724: * Traditional 4.4BSD did not consult IFF_UP at all.
725: * The behavior here is to treat addresses on !IFF_UP interface
726: * as not mine.
1.1 cgd 727: */
1.100 itojun 728: downmatch = 0;
1.140 matt 729: LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) {
1.97 itojun 730: if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
1.165 christos 731: if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif)
732: continue;
1.97 itojun 733: if ((ia->ia_ifp->if_flags & IFF_UP) != 0)
734: break;
1.100 itojun 735: else
736: downmatch++;
1.97 itojun 737: }
738: }
1.86 thorpej 739: if (ia != NULL)
740: goto ours;
1.225 christos 741: if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
1.209 matt 742: IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) {
1.140 matt 743: if (ifa->ifa_addr->sa_family != AF_INET)
744: continue;
1.57 tls 745: ia = ifatoia(ifa);
1.35 mycroft 746: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
747: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 748: /*
749: * Look for all-0's host part (old broadcast addr),
750: * either for subnet or net.
751: */
752: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 753: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 754: goto ours;
1.57 tls 755: /*
756: * An interface with IP address zero accepts
757: * all packets that arrive on that interface.
758: */
759: if (in_nullhost(ia->ia_addr.sin_addr))
760: goto ours;
1.1 cgd 761: }
762: }
1.18 mycroft 763: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 764: struct in_multi *inm;
765: #ifdef MROUTING
766: extern struct socket *ip_mrouter;
1.10 brezak 767:
1.4 hpeyerl 768: if (ip_mrouter) {
769: /*
770: * If we are acting as a multicast router, all
771: * incoming multicast packets are passed to the
772: * kernel-level multicast forwarding function.
773: * The packet is returned (relatively) intact; if
774: * ip_mforward() returns a non-zero value, the packet
775: * must be discarded, else it may be accepted below.
776: *
777: * (The IP ident field is put in the same byte order
778: * as expected when ip_mforward() is called from
779: * ip_output().)
780: */
1.13 mycroft 781: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
782: ipstat.ips_cantforward++;
1.4 hpeyerl 783: m_freem(m);
1.89 itojun 784: return;
1.4 hpeyerl 785: }
786:
787: /*
788: * The process-level routing demon needs to receive
789: * all multicast IGMP packets, whether or not this
790: * host belongs to their destination groups.
791: */
792: if (ip->ip_p == IPPROTO_IGMP)
793: goto ours;
1.13 mycroft 794: ipstat.ips_forward++;
1.4 hpeyerl 795: }
796: #endif
797: /*
798: * See if we belong to the destination multicast group on the
799: * arrival interface.
800: */
801: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
802: if (inm == NULL) {
1.13 mycroft 803: ipstat.ips_cantforward++;
1.4 hpeyerl 804: m_freem(m);
1.89 itojun 805: return;
1.4 hpeyerl 806: }
807: goto ours;
808: }
1.19 mycroft 809: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 810: in_nullhost(ip->ip_dst))
1.1 cgd 811: goto ours;
812:
813: /*
814: * Not for us; forward if possible and desirable.
815: */
816: if (ipforwarding == 0) {
817: ipstat.ips_cantforward++;
818: m_freem(m);
1.100 itojun 819: } else {
820: /*
821: * If ip_dst matched any of my address on !IFF_UP interface,
822: * and there's no IFF_UP interface that matches ip_dst,
823: * send icmp unreach. Forwarding it will result in in-kernel
824: * forwarding loop till TTL goes to 0.
825: */
826: if (downmatch) {
827: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
828: ipstat.ips_cantforward++;
829: return;
830: }
1.145 itojun 831: #ifdef IPSEC
832: if (ipsec4_in_reject(m, NULL)) {
833: ipsecstat.in_polvio++;
834: goto bad;
835: }
836: #endif
1.173 jonathan 837: #ifdef FAST_IPSEC
838: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
839: s = splsoftnet();
840: if (mtag != NULL) {
841: tdbi = (struct tdb_ident *)(mtag + 1);
842: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
843: } else {
844: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
1.212 perry 845: IP_FORWARDING, &error);
1.173 jonathan 846: }
847: if (sp == NULL) { /* NB: can happen if error */
848: splx(s);
849: /*XXX error stat???*/
850: DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/
851: goto bad;
852: }
853:
854: /*
855: * Check security policy against packet attributes.
856: */
857: error = ipsec_in_reject(sp, m);
858: KEY_FREESP(&sp);
859: splx(s);
860: if (error) {
861: ipstat.ips_cantforward++;
862: goto bad;
1.193 scw 863: }
864:
865: /*
866: * Peek at the outbound SP for this packet to determine if
867: * it's a Fast Forward candidate.
868: */
869: mtag = m_tag_find(m, PACKET_TAG_IPSEC_PENDING_TDB, NULL);
870: if (mtag != NULL)
871: m->m_flags &= ~M_CANFASTFWD;
872: else {
873: s = splsoftnet();
874: sp = ipsec4_checkpolicy(m, IPSEC_DIR_OUTBOUND,
875: (IP_FORWARDING |
876: (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
877: &error, NULL);
878: if (sp != NULL) {
879: m->m_flags &= ~M_CANFASTFWD;
880: KEY_FREESP(&sp);
881: }
882: splx(s);
1.173 jonathan 883: }
884: #endif /* FAST_IPSEC */
1.145 itojun 885:
1.169 itojun 886: ip_forward(m, srcrt);
1.100 itojun 887: }
1.89 itojun 888: return;
1.1 cgd 889:
890: ours:
891: /*
892: * If offset or IP_MF are set, must reassemble.
893: * Otherwise, nothing need be done.
894: * (We could look in the reassembly queue to see
895: * if the packet was previously fragmented,
896: * but it's not worth the time; just let them time out.)
897: */
1.155 itojun 898: if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
899:
1.1 cgd 900: /*
901: * Look for queue of fragments
902: * of this datagram.
903: */
1.75 thorpej 904: IPQ_LOCK();
1.190 jonathan 905: hash = IPREASS_HASH(ip->ip_src.s_addr, ip->ip_id);
1.250 dyoung 906: LIST_FOREACH(fp, &ipq[hash], ipq_q) {
1.1 cgd 907: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 908: in_hosteq(ip->ip_src, fp->ipq_src) &&
909: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 910: ip->ip_p == fp->ipq_p)
911: goto found;
1.190 jonathan 912:
913: }
1.1 cgd 914: fp = 0;
915: found:
916:
917: /*
918: * Adjust ip_len to not reflect header,
1.25 cgd 919: * set ipqe_mff if more fragments are expected,
1.1 cgd 920: * convert offset of this to bytes.
921: */
1.155 itojun 922: ip->ip_len = htons(ntohs(ip->ip_len) - hlen);
923: mff = (ip->ip_off & htons(IP_MF)) != 0;
1.25 cgd 924: if (mff) {
1.16 cgd 925: /*
926: * Make sure that fragments have a data length
927: * that's a non-zero multiple of 8 bytes.
928: */
1.155 itojun 929: if (ntohs(ip->ip_len) == 0 ||
930: (ntohs(ip->ip_len) & 0x7) != 0) {
1.16 cgd 931: ipstat.ips_badfrags++;
1.75 thorpej 932: IPQ_UNLOCK();
1.16 cgd 933: goto bad;
934: }
935: }
1.155 itojun 936: ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3);
1.1 cgd 937:
938: /*
939: * If datagram marked as having more fragments
940: * or if this is not the first fragment,
941: * attempt reassembly; if it succeeds, proceed.
942: */
1.155 itojun 943: if (mff || ip->ip_off != htons(0)) {
1.1 cgd 944: ipstat.ips_fragments++;
1.233 tls 945: s = splvm();
1.72 thorpej 946: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.233 tls 947: splx(s);
1.25 cgd 948: if (ipqe == NULL) {
949: ipstat.ips_rcvmemdrop++;
1.75 thorpej 950: IPQ_UNLOCK();
1.25 cgd 951: goto bad;
952: }
953: ipqe->ipqe_mff = mff;
1.50 thorpej 954: ipqe->ipqe_m = m;
1.25 cgd 955: ipqe->ipqe_ip = ip;
1.190 jonathan 956: m = ip_reass(ipqe, fp, &ipq[hash]);
1.75 thorpej 957: if (m == 0) {
958: IPQ_UNLOCK();
1.89 itojun 959: return;
1.75 thorpej 960: }
1.13 mycroft 961: ipstat.ips_reassembled++;
1.50 thorpej 962: ip = mtod(m, struct ip *);
1.74 thorpej 963: hlen = ip->ip_hl << 2;
1.155 itojun 964: ip->ip_len = htons(ntohs(ip->ip_len) + hlen);
1.1 cgd 965: } else
966: if (fp)
967: ip_freef(fp);
1.75 thorpej 968: IPQ_UNLOCK();
1.79 mycroft 969: }
1.128 itojun 970:
1.173 jonathan 971: #if defined(IPSEC)
1.128 itojun 972: /*
973: * enforce IPsec policy checking if we are seeing last header.
974: * note that we do not visit this with protocols with pcb layer
975: * code - like udp/tcp/raw ip.
976: */
977: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 &&
978: ipsec4_in_reject(m, NULL)) {
979: ipsecstat.in_polvio++;
980: goto bad;
981: }
982: #endif
1.226 liamjfoy 983: #ifdef FAST_IPSEC
1.173 jonathan 984: /*
985: * enforce IPsec policy checking if we are seeing last header.
986: * note that we do not visit this with protocols with pcb layer
987: * code - like udp/tcp/raw ip.
988: */
989: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) {
990: /*
991: * Check if the packet has already had IPsec processing
992: * done. If so, then just pass it along. This tag gets
993: * set during AH, ESP, etc. input handling, before the
994: * packet is returned to the ip input queue for delivery.
1.212 perry 995: */
1.173 jonathan 996: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
997: s = splsoftnet();
998: if (mtag != NULL) {
999: tdbi = (struct tdb_ident *)(mtag + 1);
1000: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
1001: } else {
1002: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
1.212 perry 1003: IP_FORWARDING, &error);
1.173 jonathan 1004: }
1005: if (sp != NULL) {
1006: /*
1007: * Check security policy against packet attributes.
1008: */
1009: error = ipsec_in_reject(sp, m);
1010: KEY_FREESP(&sp);
1011: } else {
1012: /* XXX error stat??? */
1013: error = EINVAL;
1014: DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/
1015: }
1016: splx(s);
1017: if (error)
1018: goto bad;
1019: }
1020: #endif /* FAST_IPSEC */
1.1 cgd 1021:
1022: /*
1023: * Switch out to protocol's input routine.
1024: */
1.82 aidan 1025: #if IFA_STATS
1.122 itojun 1026: if (ia && ip)
1.155 itojun 1027: ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
1.82 aidan 1028: #endif
1.1 cgd 1029: ipstat.ips_delivered++;
1.89 itojun 1030: {
1031: int off = hlen, nh = ip->ip_p;
1032:
1033: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
1034: return;
1035: }
1.1 cgd 1036: bad:
1037: m_freem(m);
1.135 thorpej 1038: return;
1039:
1040: badcsum:
1041: ipstat.ips_badsum++;
1042: m_freem(m);
1.1 cgd 1043: }
1044:
1045: /*
1046: * Take incoming datagram fragment and try to
1047: * reassemble it into whole datagram. If a chain for
1048: * reassembly of this datagram already exists, then it
1049: * is given as fp; otherwise have to make a chain.
1050: */
1.50 thorpej 1051: struct mbuf *
1.211 perry 1052: ip_reass(struct ipqent *ipqe, struct ipq *fp, struct ipqhead *ipqhead)
1.1 cgd 1053: {
1.109 augustss 1054: struct mbuf *m = ipqe->ipqe_m;
1055: struct ipqent *nq, *p, *q;
1.25 cgd 1056: struct ip *ip;
1.1 cgd 1057: struct mbuf *t;
1.25 cgd 1058: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.233 tls 1059: int i, next, s;
1.1 cgd 1060:
1.75 thorpej 1061: IPQ_LOCK_CHECK();
1062:
1.1 cgd 1063: /*
1064: * Presence of header sizes in mbufs
1065: * would confuse code below.
1066: */
1067: m->m_data += hlen;
1068: m->m_len -= hlen;
1069:
1.194 jonathan 1070: #ifdef notyet
1071: /* make sure fragment limit is up-to-date */
1072: CHECK_NMBCLUSTER_PARAMS();
1073:
1074: /* If we have too many fragments, drop the older half. */
1075: if (ip_nfrags >= ip_maxfrags)
1076: ip_reass_drophalf(void);
1077: #endif
1078:
1.1 cgd 1079: /*
1.192 jonathan 1080: * We are about to add a fragment; increment frag count.
1081: */
1082: ip_nfrags++;
1.212 perry 1083:
1.192 jonathan 1084: /*
1.1 cgd 1085: * If first fragment to arrive, create a reassembly queue.
1086: */
1087: if (fp == 0) {
1.131 itojun 1088: /*
1089: * Enforce upper bound on number of fragmented packets
1090: * for which we attempt reassembly;
1091: * If maxfrag is 0, never accept fragments.
1092: * If maxfrag is -1, accept all fragments without limitation.
1093: */
1094: if (ip_maxfragpackets < 0)
1095: ;
1096: else if (ip_nfragpackets >= ip_maxfragpackets)
1097: goto dropfrag;
1098: ip_nfragpackets++;
1.50 thorpej 1099: MALLOC(fp, struct ipq *, sizeof (struct ipq),
1100: M_FTABLE, M_NOWAIT);
1101: if (fp == NULL)
1.1 cgd 1102: goto dropfrag;
1.190 jonathan 1103: LIST_INSERT_HEAD(ipqhead, fp, ipq_q);
1.192 jonathan 1104: fp->ipq_nfrags = 1;
1.1 cgd 1105: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 1106: fp->ipq_p = ipqe->ipqe_ip->ip_p;
1107: fp->ipq_id = ipqe->ipqe_ip->ip_id;
1.148 matt 1108: TAILQ_INIT(&fp->ipq_fragq);
1.25 cgd 1109: fp->ipq_src = ipqe->ipqe_ip->ip_src;
1110: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
1111: p = NULL;
1.1 cgd 1112: goto insert;
1.192 jonathan 1113: } else {
1114: fp->ipq_nfrags++;
1.1 cgd 1115: }
1116:
1117: /*
1118: * Find a segment which begins after this one does.
1119: */
1.148 matt 1120: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1121: p = q, q = TAILQ_NEXT(q, ipqe_q))
1.155 itojun 1122: if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off))
1.1 cgd 1123: break;
1124:
1125: /*
1126: * If there is a preceding segment, it may provide some of
1127: * our data already. If so, drop the data from the incoming
1128: * segment. If it provides all of our data, drop us.
1129: */
1.25 cgd 1130: if (p != NULL) {
1.155 itojun 1131: i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) -
1132: ntohs(ipqe->ipqe_ip->ip_off);
1.1 cgd 1133: if (i > 0) {
1.155 itojun 1134: if (i >= ntohs(ipqe->ipqe_ip->ip_len))
1.1 cgd 1135: goto dropfrag;
1.50 thorpej 1136: m_adj(ipqe->ipqe_m, i);
1.155 itojun 1137: ipqe->ipqe_ip->ip_off =
1138: htons(ntohs(ipqe->ipqe_ip->ip_off) + i);
1139: ipqe->ipqe_ip->ip_len =
1140: htons(ntohs(ipqe->ipqe_ip->ip_len) - i);
1.1 cgd 1141: }
1142: }
1143:
1144: /*
1145: * While we overlap succeeding segments trim them or,
1146: * if they are completely covered, dequeue them.
1147: */
1.155 itojun 1148: for (; q != NULL &&
1149: ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) >
1150: ntohs(q->ipqe_ip->ip_off); q = nq) {
1151: i = (ntohs(ipqe->ipqe_ip->ip_off) +
1152: ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off);
1153: if (i < ntohs(q->ipqe_ip->ip_len)) {
1154: q->ipqe_ip->ip_len =
1155: htons(ntohs(q->ipqe_ip->ip_len) - i);
1156: q->ipqe_ip->ip_off =
1157: htons(ntohs(q->ipqe_ip->ip_off) + i);
1.50 thorpej 1158: m_adj(q->ipqe_m, i);
1.1 cgd 1159: break;
1160: }
1.148 matt 1161: nq = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1162: m_freem(q->ipqe_m);
1.148 matt 1163: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.233 tls 1164: s = splvm();
1.72 thorpej 1165: pool_put(&ipqent_pool, q);
1.233 tls 1166: splx(s);
1.192 jonathan 1167: fp->ipq_nfrags--;
1168: ip_nfrags--;
1.1 cgd 1169: }
1170:
1171: insert:
1172: /*
1173: * Stick new segment in its place;
1174: * check for complete reassembly.
1175: */
1.25 cgd 1176: if (p == NULL) {
1.148 matt 1177: TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
1.25 cgd 1178: } else {
1.148 matt 1179: TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q);
1.25 cgd 1180: }
1.1 cgd 1181: next = 0;
1.148 matt 1182: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1183: p = q, q = TAILQ_NEXT(q, ipqe_q)) {
1.155 itojun 1184: if (ntohs(q->ipqe_ip->ip_off) != next)
1.1 cgd 1185: return (0);
1.155 itojun 1186: next += ntohs(q->ipqe_ip->ip_len);
1.1 cgd 1187: }
1.25 cgd 1188: if (p->ipqe_mff)
1.1 cgd 1189: return (0);
1190:
1191: /*
1.41 thorpej 1192: * Reassembly is complete. Check for a bogus message size and
1193: * concatenate fragments.
1.1 cgd 1194: */
1.148 matt 1195: q = TAILQ_FIRST(&fp->ipq_fragq);
1.25 cgd 1196: ip = q->ipqe_ip;
1.41 thorpej 1197: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
1198: ipstat.ips_toolong++;
1199: ip_freef(fp);
1200: return (0);
1201: }
1.50 thorpej 1202: m = q->ipqe_m;
1.1 cgd 1203: t = m->m_next;
1204: m->m_next = 0;
1205: m_cat(m, t);
1.148 matt 1206: nq = TAILQ_NEXT(q, ipqe_q);
1.233 tls 1207: s = splvm();
1.72 thorpej 1208: pool_put(&ipqent_pool, q);
1.233 tls 1209: splx(s);
1.25 cgd 1210: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 1211: t = q->ipqe_m;
1.148 matt 1212: nq = TAILQ_NEXT(q, ipqe_q);
1.233 tls 1213: s = splvm();
1.72 thorpej 1214: pool_put(&ipqent_pool, q);
1.233 tls 1215: splx(s);
1.1 cgd 1216: m_cat(m, t);
1217: }
1.192 jonathan 1218: ip_nfrags -= fp->ipq_nfrags;
1.1 cgd 1219:
1220: /*
1221: * Create header for new ip packet by
1222: * modifying header of first packet;
1223: * dequeue and discard fragment reassembly header.
1224: * Make header visible.
1225: */
1.155 itojun 1226: ip->ip_len = htons(next);
1.25 cgd 1227: ip->ip_src = fp->ipq_src;
1228: ip->ip_dst = fp->ipq_dst;
1229: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1230: FREE(fp, M_FTABLE);
1.131 itojun 1231: ip_nfragpackets--;
1.1 cgd 1232: m->m_len += (ip->ip_hl << 2);
1233: m->m_data -= (ip->ip_hl << 2);
1234: /* some debugging cruft by sklower, below, will go away soon */
1235: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
1.109 augustss 1236: int plen = 0;
1.50 thorpej 1237: for (t = m; t; t = t->m_next)
1238: plen += t->m_len;
1239: m->m_pkthdr.len = plen;
1.213 yamt 1240: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1241: }
1.50 thorpej 1242: return (m);
1.1 cgd 1243:
1244: dropfrag:
1.192 jonathan 1245: if (fp != 0)
1246: fp->ipq_nfrags--;
1247: ip_nfrags--;
1.1 cgd 1248: ipstat.ips_fragdropped++;
1249: m_freem(m);
1.233 tls 1250: s = splvm();
1.72 thorpej 1251: pool_put(&ipqent_pool, ipqe);
1.233 tls 1252: splx(s);
1.1 cgd 1253: return (0);
1254: }
1255:
1256: /*
1257: * Free a fragment reassembly header and all
1258: * associated datagrams.
1259: */
1.8 mycroft 1260: void
1.211 perry 1261: ip_freef(struct ipq *fp)
1.1 cgd 1262: {
1.109 augustss 1263: struct ipqent *q, *p;
1.192 jonathan 1264: u_int nfrags = 0;
1.233 tls 1265: int s;
1.1 cgd 1266:
1.75 thorpej 1267: IPQ_LOCK_CHECK();
1268:
1.148 matt 1269: for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) {
1270: p = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1271: m_freem(q->ipqe_m);
1.192 jonathan 1272: nfrags++;
1.148 matt 1273: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.233 tls 1274: s = splvm();
1.72 thorpej 1275: pool_put(&ipqent_pool, q);
1.233 tls 1276: splx(s);
1.1 cgd 1277: }
1.192 jonathan 1278:
1279: if (nfrags != fp->ipq_nfrags)
1280: printf("ip_freef: nfrags %d != %d\n", fp->ipq_nfrags, nfrags);
1281: ip_nfrags -= nfrags;
1.25 cgd 1282: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1283: FREE(fp, M_FTABLE);
1.131 itojun 1284: ip_nfragpackets--;
1.1 cgd 1285: }
1286:
1287: /*
1.194 jonathan 1288: * IP reassembly TTL machinery for multiplicative drop.
1289: */
1290: static u_int fragttl_histo[(IPFRAGTTL+1)];
1291:
1292:
1293: /*
1294: * Decrement TTL of all reasembly queue entries by `ticks'.
1295: * Count number of distinct fragments (as opposed to partial, fragmented
1296: * datagrams) in the reassembly queue. While we traverse the entire
1297: * reassembly queue, compute and return the median TTL over all fragments.
1298: */
1299: static u_int
1300: ip_reass_ttl_decr(u_int ticks)
1301: {
1.198 matt 1302: u_int nfrags, median, dropfraction, keepfraction;
1.194 jonathan 1303: struct ipq *fp, *nfp;
1.198 matt 1304: int i;
1.212 perry 1305:
1.194 jonathan 1306: nfrags = 0;
1307: memset(fragttl_histo, 0, sizeof fragttl_histo);
1.212 perry 1308:
1.194 jonathan 1309: for (i = 0; i < IPREASS_NHASH; i++) {
1310: for (fp = LIST_FIRST(&ipq[i]); fp != NULL; fp = nfp) {
1311: fp->ipq_ttl = ((fp->ipq_ttl <= ticks) ?
1312: 0 : fp->ipq_ttl - ticks);
1313: nfp = LIST_NEXT(fp, ipq_q);
1314: if (fp->ipq_ttl == 0) {
1315: ipstat.ips_fragtimeout++;
1316: ip_freef(fp);
1317: } else {
1318: nfrags += fp->ipq_nfrags;
1319: fragttl_histo[fp->ipq_ttl] += fp->ipq_nfrags;
1320: }
1321: }
1322: }
1323:
1324: KASSERT(ip_nfrags == nfrags);
1325:
1326: /* Find median (or other drop fraction) in histogram. */
1327: dropfraction = (ip_nfrags / 2);
1328: keepfraction = ip_nfrags - dropfraction;
1329: for (i = IPFRAGTTL, median = 0; i >= 0; i--) {
1330: median += fragttl_histo[i];
1331: if (median >= keepfraction)
1332: break;
1333: }
1334:
1335: /* Return TTL of median (or other fraction). */
1336: return (u_int)i;
1337: }
1338:
1339: void
1340: ip_reass_drophalf(void)
1341: {
1342:
1343: u_int median_ticks;
1344: /*
1345: * Compute median TTL of all fragments, and count frags
1346: * with that TTL or lower (roughly half of all fragments).
1347: */
1348: median_ticks = ip_reass_ttl_decr(0);
1349:
1350: /* Drop half. */
1351: median_ticks = ip_reass_ttl_decr(median_ticks);
1352:
1353: }
1354:
1355: /*
1.1 cgd 1356: * IP timer processing;
1357: * if a timer expires on a reassembly
1358: * queue, discard it.
1359: */
1.8 mycroft 1360: void
1.211 perry 1361: ip_slowtimo(void)
1.1 cgd 1362: {
1.191 jonathan 1363: static u_int dropscanidx = 0;
1364: u_int i;
1.194 jonathan 1365: u_int median_ttl;
1.24 mycroft 1366: int s = splsoftnet();
1.1 cgd 1367:
1.75 thorpej 1368: IPQ_LOCK();
1.194 jonathan 1369:
1370: /* Age TTL of all fragments by 1 tick .*/
1371: median_ttl = ip_reass_ttl_decr(1);
1372:
1373: /* make sure fragment limit is up-to-date */
1374: CHECK_NMBCLUSTER_PARAMS();
1375:
1376: /* If we have too many fragments, drop the older half. */
1377: if (ip_nfrags > ip_maxfrags)
1378: ip_reass_ttl_decr(median_ttl);
1379:
1.131 itojun 1380: /*
1.194 jonathan 1381: * If we are over the maximum number of fragmented packets
1.131 itojun 1382: * (due to the limit being lowered), drain off
1.190 jonathan 1383: * enough to get down to the new limit. Start draining
1384: * from the reassembly hashqueue most recently drained.
1.131 itojun 1385: */
1386: if (ip_maxfragpackets < 0)
1387: ;
1388: else {
1.190 jonathan 1389: int wrapped = 0;
1390:
1391: i = dropscanidx;
1392: while (ip_nfragpackets > ip_maxfragpackets && wrapped == 0) {
1393: while (LIST_FIRST(&ipq[i]) != NULL)
1394: ip_freef(LIST_FIRST(&ipq[i]));
1395: if (++i >= IPREASS_NHASH) {
1396: i = 0;
1397: }
1398: /*
1399: * Dont scan forever even if fragment counters are
1400: * wrong: stop after scanning entire reassembly queue.
1401: */
1402: if (i == dropscanidx)
1403: wrapped = 1;
1404: }
1405: dropscanidx = i;
1.131 itojun 1406: }
1.75 thorpej 1407: IPQ_UNLOCK();
1.1 cgd 1408: splx(s);
1409: }
1410:
1411: /*
1412: * Drain off all datagram fragments.
1413: */
1.8 mycroft 1414: void
1.211 perry 1415: ip_drain(void)
1.1 cgd 1416: {
1417:
1.75 thorpej 1418: /*
1419: * We may be called from a device's interrupt context. If
1420: * the ipq is already busy, just bail out now.
1421: */
1422: if (ipq_lock_try() == 0)
1423: return;
1424:
1.194 jonathan 1425: /*
1426: * Drop half the total fragments now. If more mbufs are needed,
1427: * we will be called again soon.
1428: */
1429: ip_reass_drophalf();
1.75 thorpej 1430:
1431: IPQ_UNLOCK();
1.1 cgd 1432: }
1433:
1434: /*
1435: * Do option processing on a datagram,
1436: * possibly discarding it if bad options are encountered,
1437: * or forwarding it if source-routed.
1438: * Returns 1 if packet has been forwarded/freed,
1439: * 0 if the packet should be processed further.
1440: */
1.8 mycroft 1441: int
1.211 perry 1442: ip_dooptions(struct mbuf *m)
1.1 cgd 1443: {
1.109 augustss 1444: struct ip *ip = mtod(m, struct ip *);
1445: u_char *cp, *cp0;
1446: struct ip_timestamp *ipt;
1447: struct in_ifaddr *ia;
1.1 cgd 1448: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 1449: struct in_addr dst;
1.1 cgd 1450: n_time ntime;
1451:
1.13 mycroft 1452: dst = ip->ip_dst;
1.1 cgd 1453: cp = (u_char *)(ip + 1);
1454: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
1455: for (; cnt > 0; cnt -= optlen, cp += optlen) {
1456: opt = cp[IPOPT_OPTVAL];
1457: if (opt == IPOPT_EOL)
1458: break;
1459: if (opt == IPOPT_NOP)
1460: optlen = 1;
1461: else {
1.113 itojun 1462: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1463: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1464: goto bad;
1465: }
1.1 cgd 1466: optlen = cp[IPOPT_OLEN];
1.114 itojun 1467: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 1468: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1469: goto bad;
1470: }
1471: }
1472: switch (opt) {
1473:
1474: default:
1475: break;
1476:
1477: /*
1478: * Source routing with record.
1479: * Find interface with current destination address.
1480: * If none on this machine then drop if strictly routed,
1481: * or do nothing if loosely routed.
1482: * Record interface address and bring up next address
1483: * component. If strictly routed make sure next
1484: * address is on directly accessible net.
1485: */
1486: case IPOPT_LSRR:
1487: case IPOPT_SSRR:
1.47 cjs 1488: if (ip_allowsrcrt == 0) {
1489: type = ICMP_UNREACH;
1490: code = ICMP_UNREACH_NET_PROHIB;
1491: goto bad;
1492: }
1.114 itojun 1493: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1494: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1495: goto bad;
1496: }
1.1 cgd 1497: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1498: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1499: goto bad;
1500: }
1501: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 1502: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 1503: if (ia == 0) {
1504: if (opt == IPOPT_SSRR) {
1505: type = ICMP_UNREACH;
1506: code = ICMP_UNREACH_SRCFAIL;
1507: goto bad;
1508: }
1509: /*
1510: * Loose routing, and not at next destination
1511: * yet; nothing to do except forward.
1512: */
1513: break;
1514: }
1515: off--; /* 0 origin */
1.112 sommerfe 1516: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 1517: /*
1518: * End of source route. Should be for us.
1519: */
1520: save_rte(cp, ip->ip_src);
1521: break;
1522: }
1523: /*
1524: * locate outgoing interface
1525: */
1.244 christos 1526: bcopy((void *)(cp + off), (void *)&ipaddr.sin_addr,
1.1 cgd 1527: sizeof(ipaddr.sin_addr));
1.96 thorpej 1528: if (opt == IPOPT_SSRR)
1.196 itojun 1529: ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr)));
1.96 thorpej 1530: else
1.1 cgd 1531: ia = ip_rtaddr(ipaddr.sin_addr);
1532: if (ia == 0) {
1533: type = ICMP_UNREACH;
1534: code = ICMP_UNREACH_SRCFAIL;
1535: goto bad;
1536: }
1537: ip->ip_dst = ipaddr.sin_addr;
1.244 christos 1538: bcopy((void *)&ia->ia_addr.sin_addr,
1539: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1540: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1541: /*
1542: * Let ip_intr's mcast routing check handle mcast pkts
1543: */
1.18 mycroft 1544: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1545: break;
1546:
1547: case IPOPT_RR:
1.114 itojun 1548: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1549: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1550: goto bad;
1551: }
1.1 cgd 1552: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1553: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1554: goto bad;
1555: }
1556: /*
1557: * If no space remains, ignore.
1558: */
1559: off--; /* 0 origin */
1.112 sommerfe 1560: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1561: break;
1.244 christos 1562: bcopy((void *)(&ip->ip_dst), (void *)&ipaddr.sin_addr,
1.1 cgd 1563: sizeof(ipaddr.sin_addr));
1564: /*
1565: * locate outgoing interface; if we're the destination,
1566: * use the incoming interface (should be same).
1567: */
1.96 thorpej 1568: if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))))
1569: == NULL &&
1570: (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
1.1 cgd 1571: type = ICMP_UNREACH;
1572: code = ICMP_UNREACH_HOST;
1573: goto bad;
1574: }
1.244 christos 1575: bcopy((void *)&ia->ia_addr.sin_addr,
1576: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1577: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1578: break;
1579:
1580: case IPOPT_TS:
1581: code = cp - (u_char *)ip;
1582: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1583: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1584: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1585: goto bad;
1.114 itojun 1586: }
1587: if (ipt->ipt_ptr < 5) {
1588: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1589: goto bad;
1590: }
1.15 cgd 1591: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1592: if (++ipt->ipt_oflw == 0) {
1593: code = (u_char *)&ipt->ipt_ptr -
1594: (u_char *)ip;
1.1 cgd 1595: goto bad;
1.114 itojun 1596: }
1.1 cgd 1597: break;
1598: }
1.104 thorpej 1599: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1600: switch (ipt->ipt_flg) {
1601:
1602: case IPOPT_TS_TSONLY:
1603: break;
1604:
1605: case IPOPT_TS_TSANDADDR:
1.66 thorpej 1606: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1607: sizeof(struct in_addr) > ipt->ipt_len) {
1608: code = (u_char *)&ipt->ipt_ptr -
1609: (u_char *)ip;
1.1 cgd 1610: goto bad;
1.114 itojun 1611: }
1.13 mycroft 1612: ipaddr.sin_addr = dst;
1.96 thorpej 1613: ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr),
1614: m->m_pkthdr.rcvif));
1.13 mycroft 1615: if (ia == 0)
1616: continue;
1.104 thorpej 1617: bcopy(&ia->ia_addr.sin_addr,
1618: cp0, sizeof(struct in_addr));
1.1 cgd 1619: ipt->ipt_ptr += sizeof(struct in_addr);
1620: break;
1621:
1622: case IPOPT_TS_PRESPEC:
1.66 thorpej 1623: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1624: sizeof(struct in_addr) > ipt->ipt_len) {
1625: code = (u_char *)&ipt->ipt_ptr -
1626: (u_char *)ip;
1.1 cgd 1627: goto bad;
1.114 itojun 1628: }
1.104 thorpej 1629: bcopy(cp0, &ipaddr.sin_addr,
1.1 cgd 1630: sizeof(struct in_addr));
1.96 thorpej 1631: if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))
1632: == NULL)
1.1 cgd 1633: continue;
1634: ipt->ipt_ptr += sizeof(struct in_addr);
1635: break;
1636:
1637: default:
1.114 itojun 1638: /* XXX can't take &ipt->ipt_flg */
1639: code = (u_char *)&ipt->ipt_ptr -
1640: (u_char *)ip + 1;
1.1 cgd 1641: goto bad;
1642: }
1643: ntime = iptime();
1.107 thorpej 1644: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1.244 christos 1645: memmove((char *)cp + ipt->ipt_ptr - 1, cp0,
1.1 cgd 1646: sizeof(n_time));
1647: ipt->ipt_ptr += sizeof(n_time);
1648: }
1649: }
1650: if (forward) {
1.26 thorpej 1651: if (ip_forwsrcrt == 0) {
1652: type = ICMP_UNREACH;
1653: code = ICMP_UNREACH_SRCFAIL;
1654: goto bad;
1655: }
1.1 cgd 1656: ip_forward(m, 1);
1657: return (1);
1.13 mycroft 1658: }
1659: return (0);
1.1 cgd 1660: bad:
1.13 mycroft 1661: icmp_error(m, type, code, 0, 0);
1662: ipstat.ips_badoptions++;
1.1 cgd 1663: return (1);
1664: }
1665:
1666: /*
1667: * Given address of next destination (final or next hop),
1668: * return internet address info of interface to be used to get there.
1669: */
1670: struct in_ifaddr *
1.211 perry 1671: ip_rtaddr(struct in_addr dst)
1.1 cgd 1672: {
1.249 dyoung 1673: struct rtentry *rt;
1674: union {
1675: struct sockaddr dst;
1676: struct sockaddr_in dst4;
1677: } u;
1678:
1679: sockaddr_in_init(&u.dst4, &dst, 0);
1680:
1681: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL)
1682: return NULL;
1683:
1684: return ifatoia(rt->rt_ifa);
1.1 cgd 1685: }
1686:
1687: /*
1688: * Save incoming source route for use in replies,
1689: * to be picked up later by ip_srcroute if the receiver is interested.
1690: */
1.13 mycroft 1691: void
1.211 perry 1692: save_rte(u_char *option, struct in_addr dst)
1.1 cgd 1693: {
1694: unsigned olen;
1695:
1696: olen = option[IPOPT_OLEN];
1697: #ifdef DIAGNOSTIC
1698: if (ipprintfs)
1.39 christos 1699: printf("save_rte: olen %d\n", olen);
1.89 itojun 1700: #endif /* 0 */
1.1 cgd 1701: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1702: return;
1.244 christos 1703: bcopy((void *)option, (void *)ip_srcrt.srcopt, olen);
1.1 cgd 1704: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1705: ip_srcrt.dst = dst;
1706: }
1707:
1708: /*
1709: * Retrieve incoming source route for use in replies,
1710: * in the same form used by setsockopt.
1711: * The first hop is placed before the options, will be removed later.
1712: */
1713: struct mbuf *
1.211 perry 1714: ip_srcroute(void)
1.1 cgd 1715: {
1.109 augustss 1716: struct in_addr *p, *q;
1717: struct mbuf *m;
1.1 cgd 1718:
1719: if (ip_nhops == 0)
1.237 dyoung 1720: return NULL;
1.1 cgd 1721: m = m_get(M_DONTWAIT, MT_SOOPTS);
1722: if (m == 0)
1.237 dyoung 1723: return NULL;
1.1 cgd 1724:
1.164 matt 1725: MCLAIM(m, &inetdomain.dom_mowner);
1.13 mycroft 1726: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1727:
1728: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1729: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1730: OPTSIZ;
1731: #ifdef DIAGNOSTIC
1732: if (ipprintfs)
1.39 christos 1733: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1734: #endif
1735:
1736: /*
1737: * First save first hop for return route
1738: */
1739: p = &ip_srcrt.route[ip_nhops - 1];
1740: *(mtod(m, struct in_addr *)) = *p--;
1741: #ifdef DIAGNOSTIC
1742: if (ipprintfs)
1.39 christos 1743: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1744: #endif
1745:
1746: /*
1747: * Copy option fields and padding (nop) to mbuf.
1748: */
1749: ip_srcrt.nop = IPOPT_NOP;
1750: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1.244 christos 1751: memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop,
1752: OPTSIZ);
1753: q = (struct in_addr *)(mtod(m, char *) +
1.1 cgd 1754: sizeof(struct in_addr) + OPTSIZ);
1755: #undef OPTSIZ
1756: /*
1757: * Record return path as an IP source route,
1758: * reversing the path (pointers are now aligned).
1759: */
1760: while (p >= ip_srcrt.route) {
1761: #ifdef DIAGNOSTIC
1762: if (ipprintfs)
1.39 christos 1763: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1764: #endif
1765: *q++ = *p--;
1766: }
1767: /*
1768: * Last hop goes to final destination.
1769: */
1770: *q = ip_srcrt.dst;
1771: #ifdef DIAGNOSTIC
1772: if (ipprintfs)
1.39 christos 1773: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1774: #endif
1775: return (m);
1776: }
1777:
1.139 matt 1778: const int inetctlerrmap[PRC_NCMDS] = {
1.256 ! yamt 1779: [PRC_MSGSIZE] = EMSGSIZE,
! 1780: [PRC_HOSTDEAD] = EHOSTDOWN,
! 1781: [PRC_HOSTUNREACH] = EHOSTUNREACH,
! 1782: [PRC_UNREACH_NET] = EHOSTUNREACH,
! 1783: [PRC_UNREACH_HOST] = EHOSTUNREACH,
! 1784: [PRC_UNREACH_PROTOCOL] = ECONNREFUSED,
! 1785: [PRC_UNREACH_PORT] = ECONNREFUSED,
! 1786: [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH,
! 1787: [PRC_PARAMPROB] = ENOPROTOOPT,
1.1 cgd 1788: };
1789:
1790: /*
1791: * Forward a packet. If some error occurs return the sender
1792: * an icmp packet. Note we can't always generate a meaningful
1793: * icmp message because icmp doesn't have a large enough repertoire
1794: * of codes and types.
1795: *
1796: * If not forwarding, just drop the packet. This could be confusing
1797: * if ipforwarding was zero but some routing protocol was advancing
1798: * us as a gateway to somewhere. However, we must let the routing
1799: * protocol deal with that.
1800: *
1801: * The srcrt parameter indicates whether the packet is being forwarded
1802: * via a source route.
1803: */
1.13 mycroft 1804: void
1.211 perry 1805: ip_forward(struct mbuf *m, int srcrt)
1.1 cgd 1806: {
1.109 augustss 1807: struct ip *ip = mtod(m, struct ip *);
1808: struct rtentry *rt;
1.220 christos 1809: int error, type = 0, code = 0, destmtu = 0;
1.1 cgd 1810: struct mbuf *mcopy;
1.13 mycroft 1811: n_long dest;
1.249 dyoung 1812: union {
1813: struct sockaddr dst;
1814: struct sockaddr_in dst4;
1815: } u;
1.164 matt 1816:
1817: /*
1818: * We are now in the output path.
1819: */
1820: MCLAIM(m, &ip_tx_mowner);
1.135 thorpej 1821:
1822: /*
1823: * Clear any in-bound checksum flags for this packet.
1824: */
1825: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1826:
1.13 mycroft 1827: dest = 0;
1.1 cgd 1828: #ifdef DIAGNOSTIC
1.224 joerg 1829: if (ipprintfs) {
1830: printf("forward: src %s ", inet_ntoa(ip->ip_src));
1831: printf("dst %s ttl %x\n", inet_ntoa(ip->ip_dst), ip->ip_ttl);
1832: }
1.1 cgd 1833: #endif
1.93 sommerfe 1834: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.1 cgd 1835: ipstat.ips_cantforward++;
1836: m_freem(m);
1837: return;
1838: }
1839: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1840: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1841: return;
1842: }
1843:
1.249 dyoung 1844: sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
1845: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) {
1846: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0);
1847: return;
1.1 cgd 1848: }
1849:
1850: /*
1.34 mycroft 1851: * Save at most 68 bytes of the packet in case
1.1 cgd 1852: * we need to generate an ICMP message to the src.
1.119 itojun 1853: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1854: */
1.155 itojun 1855: mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1.119 itojun 1856: if (mcopy)
1857: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1858:
1.221 christos 1859: ip->ip_ttl -= IPTTLDEC;
1860:
1.1 cgd 1861: /*
1862: * If forwarding packet using same interface that it came in on,
1863: * perhaps should send a redirect to sender to shortcut a hop.
1864: * Only send redirect if source is sending directly to us,
1865: * and if packet was not source routed (or has any options).
1866: * Also, don't send redirect if forwarding using a default route
1867: * or a route modified by a redirect.
1868: */
1869: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1870: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.250 dyoung 1871: !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) &&
1.1 cgd 1872: ipsendredirects && !srcrt) {
1.19 mycroft 1873: if (rt->rt_ifa &&
1874: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1875: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1876: if (rt->rt_flags & RTF_GATEWAY)
1877: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1878: else
1879: dest = ip->ip_dst.s_addr;
1880: /*
1881: * Router requirements says to only send host
1882: * redirects.
1883: */
1884: type = ICMP_REDIRECT;
1885: code = ICMP_REDIRECT_HOST;
1.1 cgd 1886: #ifdef DIAGNOSTIC
1.77 thorpej 1887: if (ipprintfs)
1888: printf("redirect (%d) to %x\n", code,
1889: (u_int32_t)dest);
1.1 cgd 1890: #endif
1891: }
1892: }
1893:
1.238 dyoung 1894: error = ip_output(m, NULL, &ipforward_rt,
1.173 jonathan 1895: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
1.174 itojun 1896: (struct ip_moptions *)NULL, (struct socket *)NULL);
1.173 jonathan 1897:
1.1 cgd 1898: if (error)
1899: ipstat.ips_cantforward++;
1900: else {
1901: ipstat.ips_forward++;
1902: if (type)
1903: ipstat.ips_redirectsent++;
1904: else {
1.63 matt 1905: if (mcopy) {
1906: #ifdef GATEWAY
1.64 thorpej 1907: if (mcopy->m_flags & M_CANFASTFWD)
1908: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1909: #endif
1.1 cgd 1910: m_freem(mcopy);
1.63 matt 1911: }
1.1 cgd 1912: return;
1913: }
1914: }
1915: if (mcopy == NULL)
1916: return;
1.13 mycroft 1917:
1.1 cgd 1918: switch (error) {
1919:
1920: case 0: /* forwarded, but need redirect */
1921: /* type, code set above */
1922: break;
1923:
1924: case ENETUNREACH: /* shouldn't happen, checked above */
1925: case EHOSTUNREACH:
1926: case ENETDOWN:
1927: case EHOSTDOWN:
1928: default:
1929: type = ICMP_UNREACH;
1930: code = ICMP_UNREACH_HOST;
1931: break;
1932:
1933: case EMSGSIZE:
1934: type = ICMP_UNREACH;
1935: code = ICMP_UNREACH_NEEDFRAG;
1.173 jonathan 1936: #if !defined(IPSEC) && !defined(FAST_IPSEC)
1.238 dyoung 1937: if (ipforward_rt.ro_rt != NULL)
1.220 christos 1938: destmtu = ipforward_rt.ro_rt->rt_ifp->if_mtu;
1.89 itojun 1939: #else
1940: /*
1941: * If the packet is routed over IPsec tunnel, tell the
1942: * originator the tunnel MTU.
1943: * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1944: * XXX quickhack!!!
1945: */
1.238 dyoung 1946: if (ipforward_rt.ro_rt != NULL) {
1.89 itojun 1947: struct secpolicy *sp;
1948: int ipsecerror;
1.95 itojun 1949: size_t ipsechdr;
1.89 itojun 1950: struct route *ro;
1951:
1952: sp = ipsec4_getpolicybyaddr(mcopy,
1.170 itojun 1953: IPSEC_DIR_OUTBOUND, IP_FORWARDING,
1954: &ipsecerror);
1.89 itojun 1955:
1956: if (sp == NULL)
1.220 christos 1957: destmtu = ipforward_rt.ro_rt->rt_ifp->if_mtu;
1.89 itojun 1958: else {
1959: /* count IPsec header size */
1.95 itojun 1960: ipsechdr = ipsec4_hdrsiz(mcopy,
1.170 itojun 1961: IPSEC_DIR_OUTBOUND, NULL);
1.89 itojun 1962:
1963: /*
1964: * find the correct route for outer IPv4
1965: * header, compute tunnel MTU.
1966: */
1.220 christos 1967:
1.89 itojun 1968: if (sp->req != NULL
1.95 itojun 1969: && sp->req->sav != NULL
1970: && sp->req->sav->sah != NULL) {
1971: ro = &sp->req->sav->sah->sa_route;
1.89 itojun 1972: if (ro->ro_rt && ro->ro_rt->rt_ifp) {
1.220 christos 1973: destmtu =
1.151 itojun 1974: ro->ro_rt->rt_rmx.rmx_mtu ?
1975: ro->ro_rt->rt_rmx.rmx_mtu :
1.89 itojun 1976: ro->ro_rt->rt_ifp->if_mtu;
1.220 christos 1977: destmtu -= ipsechdr;
1.89 itojun 1978: }
1979: }
1980:
1.173 jonathan 1981: #ifdef IPSEC
1.89 itojun 1982: key_freesp(sp);
1.173 jonathan 1983: #else
1984: KEY_FREESP(&sp);
1985: #endif
1.89 itojun 1986: }
1987: }
1988: #endif /*IPSEC*/
1.1 cgd 1989: ipstat.ips_cantfrag++;
1990: break;
1991:
1992: case ENOBUFS:
1.143 itojun 1993: #if 1
1994: /*
1995: * a router should not generate ICMP_SOURCEQUENCH as
1996: * required in RFC1812 Requirements for IP Version 4 Routers.
1997: * source quench could be a big problem under DoS attacks,
1.149 wiz 1998: * or if the underlying interface is rate-limited.
1.143 itojun 1999: */
2000: if (mcopy)
2001: m_freem(mcopy);
2002: return;
2003: #else
1.1 cgd 2004: type = ICMP_SOURCEQUENCH;
2005: code = 0;
2006: break;
1.143 itojun 2007: #endif
1.1 cgd 2008: }
1.220 christos 2009: icmp_error(mcopy, type, code, dest, destmtu);
1.44 thorpej 2010: }
2011:
2012: void
1.211 perry 2013: ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
2014: struct mbuf *m)
1.44 thorpej 2015: {
2016:
2017: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
2018: struct timeval tv;
2019:
2020: microtime(&tv);
1.244 christos 2021: *mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1.44 thorpej 2022: SCM_TIMESTAMP, SOL_SOCKET);
2023: if (*mp)
2024: mp = &(*mp)->m_next;
2025: }
2026: if (inp->inp_flags & INP_RECVDSTADDR) {
1.244 christos 2027: *mp = sbcreatecontrol((void *) &ip->ip_dst,
1.44 thorpej 2028: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
2029: if (*mp)
2030: mp = &(*mp)->m_next;
2031: }
2032: #ifdef notyet
2033: /*
2034: * XXX
2035: * Moving these out of udp_input() made them even more broken
2036: * than they already were.
2037: * - fenner@parc.xerox.com
2038: */
2039: /* options were tossed already */
2040: if (inp->inp_flags & INP_RECVOPTS) {
1.244 christos 2041: *mp = sbcreatecontrol((void *) opts_deleted_above,
1.44 thorpej 2042: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
2043: if (*mp)
2044: mp = &(*mp)->m_next;
2045: }
2046: /* ip_srcroute doesn't do what we want here, need to fix */
2047: if (inp->inp_flags & INP_RECVRETOPTS) {
1.244 christos 2048: *mp = sbcreatecontrol((void *) ip_srcroute(),
1.44 thorpej 2049: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
2050: if (*mp)
2051: mp = &(*mp)->m_next;
2052: }
2053: #endif
2054: if (inp->inp_flags & INP_RECVIF) {
2055: struct sockaddr_dl sdl;
2056:
1.252 dyoung 2057: sockaddr_dl_init(&sdl, sizeof(sdl),
2058: (m->m_pkthdr.rcvif != NULL)
2059: ? m->m_pkthdr.rcvif->if_index
2060: : 0,
2061: 0, NULL, 0, NULL, 0);
1.251 dyoung 2062: *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP);
1.44 thorpej 2063: if (*mp)
2064: mp = &(*mp)->m_next;
2065: }
1.13 mycroft 2066: }
2067:
1.189 atatat 2068: /*
1.228 elad 2069: * sysctl helper routine for net.inet.ip.forwsrcrt.
2070: */
2071: static int
2072: sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS)
2073: {
2074: int error, tmp;
2075: struct sysctlnode node;
2076:
2077: node = *rnode;
2078: tmp = ip_forwsrcrt;
2079: node.sysctl_data = &tmp;
2080: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2081: if (error || newp == NULL)
2082: return (error);
2083:
1.230 elad 2084: if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT,
1.232 elad 2085: 0, NULL, NULL, NULL))
1.228 elad 2086: return (EPERM);
2087:
2088: ip_forwsrcrt = tmp;
2089:
2090: return (0);
2091: }
2092:
2093: /*
1.189 atatat 2094: * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the
2095: * range of the new value and tweaks timers if it changes.
2096: */
2097: static int
2098: sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS)
1.13 mycroft 2099: {
1.189 atatat 2100: int error, tmp;
2101: struct sysctlnode node;
2102:
2103: node = *rnode;
2104: tmp = ip_mtudisc_timeout;
2105: node.sysctl_data = &tmp;
2106: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2107: if (error || newp == NULL)
2108: return (error);
2109: if (tmp < 0)
2110: return (EINVAL);
1.52 thorpej 2111:
1.189 atatat 2112: ip_mtudisc_timeout = tmp;
2113: rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout);
2114:
2115: return (0);
2116: }
1.54 lukem 2117:
1.65 matt 2118: #ifdef GATEWAY
1.189 atatat 2119: /*
1.247 liamjfoy 2120: * sysctl helper routine for net.inet.ip.maxflows.
1.189 atatat 2121: */
2122: static int
2123: sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS)
2124: {
2125: int s;
1.67 thorpej 2126:
1.217 atatat 2127: s = sysctl_lookup(SYSCTLFN_CALL(rnode));
1.247 liamjfoy 2128: if (s || newp == NULL)
1.189 atatat 2129: return (s);
1.212 perry 2130:
1.189 atatat 2131: s = splsoftnet();
2132: ipflow_reap(0);
2133: splx(s);
1.144 martin 2134:
1.189 atatat 2135: return (0);
2136: }
1.248 liamjfoy 2137:
2138: static int
2139: sysctl_net_inet_ip_hashsize(SYSCTLFN_ARGS)
2140: {
2141: int error, tmp;
2142: struct sysctlnode node;
2143:
2144: node = *rnode;
2145: tmp = ip_hashsize;
2146: node.sysctl_data = &tmp;
2147: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2148: if (error || newp == NULL)
2149: return (error);
2150:
2151: if ((tmp & (tmp - 1)) == 0 && tmp != 0) {
2152: /*
2153: * Can only fail due to malloc()
2154: */
2155: if (ipflow_invalidate_all(tmp))
2156: return ENOMEM;
2157: } else {
2158: /*
2159: * EINVAL if not a power of 2
2160: */
2161: return EINVAL;
2162: }
2163:
2164: return (0);
2165: }
1.189 atatat 2166: #endif /* GATEWAY */
1.117 tron 2167:
1.131 itojun 2168:
1.189 atatat 2169: SYSCTL_SETUP(sysctl_net_inet_ip_setup, "sysctl net.inet.ip subtree setup")
2170: {
2171: extern int subnetsarelocal, hostzeroisbroadcast;
1.180 jonathan 2172:
1.197 atatat 2173: sysctl_createv(clog, 0, NULL, NULL,
2174: CTLFLAG_PERMANENT,
1.189 atatat 2175: CTLTYPE_NODE, "net", NULL,
2176: NULL, 0, NULL, 0,
2177: CTL_NET, CTL_EOL);
1.197 atatat 2178: sysctl_createv(clog, 0, NULL, NULL,
2179: CTLFLAG_PERMANENT,
1.203 atatat 2180: CTLTYPE_NODE, "inet",
2181: SYSCTL_DESCR("PF_INET related settings"),
1.189 atatat 2182: NULL, 0, NULL, 0,
2183: CTL_NET, PF_INET, CTL_EOL);
1.197 atatat 2184: sysctl_createv(clog, 0, NULL, NULL,
2185: CTLFLAG_PERMANENT,
1.203 atatat 2186: CTLTYPE_NODE, "ip",
2187: SYSCTL_DESCR("IPv4 related settings"),
1.189 atatat 2188: NULL, 0, NULL, 0,
2189: CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL);
1.212 perry 2190:
1.197 atatat 2191: sysctl_createv(clog, 0, NULL, NULL,
2192: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2193: CTLTYPE_INT, "forwarding",
2194: SYSCTL_DESCR("Enable forwarding of INET datagrams"),
1.189 atatat 2195: NULL, 0, &ipforwarding, 0,
2196: CTL_NET, PF_INET, IPPROTO_IP,
2197: IPCTL_FORWARDING, CTL_EOL);
1.197 atatat 2198: sysctl_createv(clog, 0, NULL, NULL,
2199: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2200: CTLTYPE_INT, "redirect",
2201: SYSCTL_DESCR("Enable sending of ICMP redirect messages"),
1.189 atatat 2202: NULL, 0, &ipsendredirects, 0,
2203: CTL_NET, PF_INET, IPPROTO_IP,
2204: IPCTL_SENDREDIRECTS, CTL_EOL);
1.197 atatat 2205: sysctl_createv(clog, 0, NULL, NULL,
2206: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2207: CTLTYPE_INT, "ttl",
2208: SYSCTL_DESCR("Default TTL for an INET datagram"),
1.189 atatat 2209: NULL, 0, &ip_defttl, 0,
2210: CTL_NET, PF_INET, IPPROTO_IP,
2211: IPCTL_DEFTTL, CTL_EOL);
2212: #ifdef IPCTL_DEFMTU
1.197 atatat 2213: sysctl_createv(clog, 0, NULL, NULL,
2214: CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */,
1.203 atatat 2215: CTLTYPE_INT, "mtu",
2216: SYSCTL_DESCR("Default MTA for an INET route"),
1.189 atatat 2217: NULL, 0, &ip_mtu, 0,
2218: CTL_NET, PF_INET, IPPROTO_IP,
2219: IPCTL_DEFMTU, CTL_EOL);
2220: #endif /* IPCTL_DEFMTU */
1.197 atatat 2221: sysctl_createv(clog, 0, NULL, NULL,
1.228 elad 2222: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2223: CTLTYPE_INT, "forwsrcrt",
2224: SYSCTL_DESCR("Enable forwarding of source-routed "
2225: "datagrams"),
1.228 elad 2226: sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0,
1.189 atatat 2227: CTL_NET, PF_INET, IPPROTO_IP,
2228: IPCTL_FORWSRCRT, CTL_EOL);
1.197 atatat 2229: sysctl_createv(clog, 0, NULL, NULL,
2230: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2231: CTLTYPE_INT, "directed-broadcast",
2232: SYSCTL_DESCR("Enable forwarding of broadcast datagrams"),
1.189 atatat 2233: NULL, 0, &ip_directedbcast, 0,
2234: CTL_NET, PF_INET, IPPROTO_IP,
2235: IPCTL_DIRECTEDBCAST, CTL_EOL);
1.197 atatat 2236: sysctl_createv(clog, 0, NULL, NULL,
2237: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2238: CTLTYPE_INT, "allowsrcrt",
2239: SYSCTL_DESCR("Accept source-routed datagrams"),
1.189 atatat 2240: NULL, 0, &ip_allowsrcrt, 0,
2241: CTL_NET, PF_INET, IPPROTO_IP,
2242: IPCTL_ALLOWSRCRT, CTL_EOL);
1.197 atatat 2243: sysctl_createv(clog, 0, NULL, NULL,
2244: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2245: CTLTYPE_INT, "subnetsarelocal",
2246: SYSCTL_DESCR("Whether logical subnets are considered "
2247: "local"),
1.189 atatat 2248: NULL, 0, &subnetsarelocal, 0,
2249: CTL_NET, PF_INET, IPPROTO_IP,
2250: IPCTL_SUBNETSARELOCAL, CTL_EOL);
1.197 atatat 2251: sysctl_createv(clog, 0, NULL, NULL,
2252: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2253: CTLTYPE_INT, "mtudisc",
2254: SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"),
1.189 atatat 2255: NULL, 0, &ip_mtudisc, 0,
2256: CTL_NET, PF_INET, IPPROTO_IP,
2257: IPCTL_MTUDISC, CTL_EOL);
1.197 atatat 2258: sysctl_createv(clog, 0, NULL, NULL,
2259: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2260: CTLTYPE_INT, "anonportmin",
2261: SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1.189 atatat 2262: sysctl_net_inet_ip_ports, 0, &anonportmin, 0,
2263: CTL_NET, PF_INET, IPPROTO_IP,
2264: IPCTL_ANONPORTMIN, CTL_EOL);
1.197 atatat 2265: sysctl_createv(clog, 0, NULL, NULL,
2266: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2267: CTLTYPE_INT, "anonportmax",
2268: SYSCTL_DESCR("Highest ephemeral port number to assign"),
1.189 atatat 2269: sysctl_net_inet_ip_ports, 0, &anonportmax, 0,
2270: CTL_NET, PF_INET, IPPROTO_IP,
2271: IPCTL_ANONPORTMAX, CTL_EOL);
1.197 atatat 2272: sysctl_createv(clog, 0, NULL, NULL,
2273: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2274: CTLTYPE_INT, "mtudisctimeout",
2275: SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"),
1.189 atatat 2276: sysctl_net_inet_ip_pmtudto, 0, &ip_mtudisc_timeout, 0,
2277: CTL_NET, PF_INET, IPPROTO_IP,
2278: IPCTL_MTUDISCTIMEOUT, CTL_EOL);
2279: #ifdef GATEWAY
1.197 atatat 2280: sysctl_createv(clog, 0, NULL, NULL,
2281: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2282: CTLTYPE_INT, "maxflows",
2283: SYSCTL_DESCR("Number of flows for fast forwarding"),
1.189 atatat 2284: sysctl_net_inet_ip_maxflows, 0, &ip_maxflows, 0,
2285: CTL_NET, PF_INET, IPPROTO_IP,
2286: IPCTL_MAXFLOWS, CTL_EOL);
1.248 liamjfoy 2287: sysctl_createv(clog, 0, NULL, NULL,
2288: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2289: CTLTYPE_INT, "hashsize",
2290: SYSCTL_DESCR("Size of hash table for fast forwarding (IPv4)"),
2291: sysctl_net_inet_ip_hashsize, 0, &ip_hashsize, 0,
2292: CTL_NET, PF_INET, IPPROTO_IP,
2293: CTL_CREATE, CTL_EOL);
1.189 atatat 2294: #endif /* GATEWAY */
1.197 atatat 2295: sysctl_createv(clog, 0, NULL, NULL,
2296: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2297: CTLTYPE_INT, "hostzerobroadcast",
2298: SYSCTL_DESCR("All zeroes address is broadcast address"),
1.189 atatat 2299: NULL, 0, &hostzeroisbroadcast, 0,
2300: CTL_NET, PF_INET, IPPROTO_IP,
2301: IPCTL_HOSTZEROBROADCAST, CTL_EOL);
2302: #if NGIF > 0
1.197 atatat 2303: sysctl_createv(clog, 0, NULL, NULL,
2304: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2305: CTLTYPE_INT, "gifttl",
2306: SYSCTL_DESCR("Default TTL for a gif tunnel datagram"),
1.189 atatat 2307: NULL, 0, &ip_gif_ttl, 0,
2308: CTL_NET, PF_INET, IPPROTO_IP,
2309: IPCTL_GIF_TTL, CTL_EOL);
2310: #endif /* NGIF */
2311: #ifndef IPNOPRIVPORTS
1.197 atatat 2312: sysctl_createv(clog, 0, NULL, NULL,
2313: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2314: CTLTYPE_INT, "lowportmin",
2315: SYSCTL_DESCR("Lowest privileged ephemeral port number "
2316: "to assign"),
1.189 atatat 2317: sysctl_net_inet_ip_ports, 0, &lowportmin, 0,
2318: CTL_NET, PF_INET, IPPROTO_IP,
2319: IPCTL_LOWPORTMIN, CTL_EOL);
1.197 atatat 2320: sysctl_createv(clog, 0, NULL, NULL,
2321: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2322: CTLTYPE_INT, "lowportmax",
2323: SYSCTL_DESCR("Highest privileged ephemeral port number "
2324: "to assign"),
1.189 atatat 2325: sysctl_net_inet_ip_ports, 0, &lowportmax, 0,
2326: CTL_NET, PF_INET, IPPROTO_IP,
2327: IPCTL_LOWPORTMAX, CTL_EOL);
2328: #endif /* IPNOPRIVPORTS */
1.197 atatat 2329: sysctl_createv(clog, 0, NULL, NULL,
2330: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2331: CTLTYPE_INT, "maxfragpackets",
2332: SYSCTL_DESCR("Maximum number of fragments to retain for "
2333: "possible reassembly"),
1.189 atatat 2334: NULL, 0, &ip_maxfragpackets, 0,
2335: CTL_NET, PF_INET, IPPROTO_IP,
2336: IPCTL_MAXFRAGPACKETS, CTL_EOL);
2337: #if NGRE > 0
1.197 atatat 2338: sysctl_createv(clog, 0, NULL, NULL,
2339: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2340: CTLTYPE_INT, "grettl",
2341: SYSCTL_DESCR("Default TTL for a gre tunnel datagram"),
1.189 atatat 2342: NULL, 0, &ip_gre_ttl, 0,
2343: CTL_NET, PF_INET, IPPROTO_IP,
2344: IPCTL_GRE_TTL, CTL_EOL);
2345: #endif /* NGRE */
1.197 atatat 2346: sysctl_createv(clog, 0, NULL, NULL,
2347: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2348: CTLTYPE_INT, "checkinterface",
2349: SYSCTL_DESCR("Enable receive side of Strong ES model "
2350: "from RFC1122"),
1.189 atatat 2351: NULL, 0, &ip_checkinterface, 0,
2352: CTL_NET, PF_INET, IPPROTO_IP,
2353: IPCTL_CHECKINTERFACE, CTL_EOL);
1.197 atatat 2354: sysctl_createv(clog, 0, NULL, NULL,
2355: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2356: CTLTYPE_INT, "random_id",
2357: SYSCTL_DESCR("Assign random ip_id values"),
1.189 atatat 2358: NULL, 0, &ip_do_randomid, 0,
2359: CTL_NET, PF_INET, IPPROTO_IP,
2360: IPCTL_RANDOMID, CTL_EOL);
1.206 thorpej 2361: sysctl_createv(clog, 0, NULL, NULL,
2362: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2363: CTLTYPE_INT, "do_loopback_cksum",
2364: SYSCTL_DESCR("Perform IP checksum on loopback"),
2365: NULL, 0, &ip_do_loopback_cksum, 0,
2366: CTL_NET, PF_INET, IPPROTO_IP,
2367: IPCTL_LOOPBACKCKSUM, CTL_EOL);
1.219 elad 2368: sysctl_createv(clog, 0, NULL, NULL,
2369: CTLFLAG_PERMANENT,
2370: CTLTYPE_STRUCT, "stats",
2371: SYSCTL_DESCR("IP statistics"),
2372: NULL, 0, &ipstat, sizeof(ipstat),
2373: CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS,
2374: CTL_EOL);
1.1 cgd 2375: }
CVSweb <webmaster@jp.NetBSD.org>