Annotation of src/sys/netinet/ip_input.c, Revision 1.218.2.5
1.218.2.5! yamt 1: /* $NetBSD: ip_input.c,v 1.218.2.4 2007/09/03 14:42:58 yamt 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>
1.218.2.5! yamt 101: __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.218.2.4 2007/09/03 14:42:58 yamt 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>
1.218.2.2 yamt 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:
1.218.2.1 yamt 283: static inline int ipq_lock_try(void);
284: static inline void ipq_unlock(void);
1.75 thorpej 285:
1.218.2.1 yamt 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:
1.218.2.1 yamt 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:
1.218.2.4 yamt 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
1.218.2.2 yamt 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:
1.218.2.1 yamt 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
1.218.2.4 yamt 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:
1.218.2.2 yamt 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:
1.218.2.2 yamt 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;
1.164 matt 472: MCLAIM(m, &ip_rx_mowner);
1.89 itojun 473: ip_input(m);
474: }
475: }
476:
477: /*
1.1 cgd 478: * Ip input routine. Checksum and byte swap header. If fragmented
479: * try to reassemble. Process options. Pass to next level.
480: */
1.8 mycroft 481: void
1.89 itojun 482: ip_input(struct mbuf *m)
1.1 cgd 483: {
1.109 augustss 484: struct ip *ip = NULL;
485: struct ipq *fp;
486: struct in_ifaddr *ia;
487: struct ifaddr *ifa;
1.25 cgd 488: struct ipqent *ipqe;
1.89 itojun 489: int hlen = 0, mff, len;
1.100 itojun 490: int downmatch;
1.165 christos 491: int checkif;
1.169 itojun 492: int srcrt = 0;
1.218.2.2 yamt 493: int s;
1.190 jonathan 494: u_int hash;
1.173 jonathan 495: #ifdef FAST_IPSEC
496: struct m_tag *mtag;
497: struct tdb_ident *tdbi;
498: struct secpolicy *sp;
1.218.2.2 yamt 499: int error;
1.173 jonathan 500: #endif /* FAST_IPSEC */
1.1 cgd 501:
1.164 matt 502: MCLAIM(m, &ip_rx_mowner);
1.1 cgd 503: #ifdef DIAGNOSTIC
504: if ((m->m_flags & M_PKTHDR) == 0)
505: panic("ipintr no HDR");
1.89 itojun 506: #endif
1.164 matt 507:
1.1 cgd 508: /*
509: * If no IP addresses have been set yet but the interfaces
510: * are receiving, can't do anything with incoming packets yet.
511: */
1.181 jonathan 512: if (TAILQ_FIRST(&in_ifaddrhead) == 0)
1.1 cgd 513: goto bad;
514: ipstat.ips_total++;
1.154 thorpej 515: /*
516: * If the IP header is not aligned, slurp it up into a new
517: * mbuf with space for link headers, in the event we forward
518: * it. Otherwise, if it is aligned, make sure the entire
519: * base IP header is in the first mbuf of the chain.
520: */
1.218.2.4 yamt 521: if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
1.154 thorpej 522: if ((m = m_copyup(m, sizeof(struct ip),
523: (max_linkhdr + 3) & ~3)) == NULL) {
524: /* XXXJRT new stat, please */
525: ipstat.ips_toosmall++;
526: return;
527: }
528: } else if (__predict_false(m->m_len < sizeof (struct ip))) {
529: if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
530: ipstat.ips_toosmall++;
531: return;
532: }
1.1 cgd 533: }
534: ip = mtod(m, struct ip *);
1.13 mycroft 535: if (ip->ip_v != IPVERSION) {
536: ipstat.ips_badvers++;
537: goto bad;
538: }
1.1 cgd 539: hlen = ip->ip_hl << 2;
540: if (hlen < sizeof(struct ip)) { /* minimum header length */
541: ipstat.ips_badhlen++;
542: goto bad;
543: }
544: if (hlen > m->m_len) {
545: if ((m = m_pullup(m, hlen)) == 0) {
546: ipstat.ips_badhlen++;
1.89 itojun 547: return;
1.1 cgd 548: }
549: ip = mtod(m, struct ip *);
550: }
1.98 thorpej 551:
1.85 hwr 552: /*
1.99 thorpej 553: * RFC1122: packets with a multicast source address are
1.98 thorpej 554: * not allowed.
1.85 hwr 555: */
556: if (IN_MULTICAST(ip->ip_src.s_addr)) {
1.130 itojun 557: ipstat.ips_badaddr++;
1.85 hwr 558: goto bad;
1.129 itojun 559: }
560:
561: /* 127/8 must not appear on wire - RFC1122 */
562: if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
563: (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
1.130 itojun 564: if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
565: ipstat.ips_badaddr++;
1.129 itojun 566: goto bad;
1.130 itojun 567: }
1.85 hwr 568: }
569:
1.135 thorpej 570: switch (m->m_pkthdr.csum_flags &
1.137 thorpej 571: ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) |
1.135 thorpej 572: M_CSUM_IPv4_BAD)) {
573: case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
574: INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
575: goto badcsum;
576:
577: case M_CSUM_IPv4:
578: /* Checksum was okay. */
579: INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
580: break;
581:
582: default:
1.206 thorpej 583: /*
584: * Must compute it ourselves. Maybe skip checksum on
585: * loopback interfaces.
586: */
587: if (__predict_true(!(m->m_pkthdr.rcvif->if_flags &
588: IFF_LOOPBACK) || ip_do_loopback_cksum)) {
589: INET_CSUM_COUNTER_INCR(&ip_swcsum);
590: if (in_cksum(m, hlen) != 0)
591: goto badcsum;
592: }
1.135 thorpej 593: break;
1.1 cgd 594: }
595:
1.121 thorpej 596: /* Retrieve the packet length. */
597: len = ntohs(ip->ip_len);
1.81 proff 598:
599: /*
600: * Check for additional length bogosity
601: */
1.84 proff 602: if (len < hlen) {
1.81 proff 603: ipstat.ips_badlen++;
604: goto bad;
605: }
1.1 cgd 606:
607: /*
608: * Check that the amount of data in the buffers
609: * is as at least much as the IP header would have us expect.
610: * Trim mbufs if longer than we expect.
611: * Drop packet if shorter than we expect.
612: */
1.35 mycroft 613: if (m->m_pkthdr.len < len) {
1.1 cgd 614: ipstat.ips_tooshort++;
615: goto bad;
616: }
1.35 mycroft 617: if (m->m_pkthdr.len > len) {
1.1 cgd 618: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 619: m->m_len = len;
620: m->m_pkthdr.len = len;
1.1 cgd 621: } else
1.35 mycroft 622: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 623: }
624:
1.193 scw 625: #if defined(IPSEC)
1.149 wiz 626: /* ipflow (IP fast forwarding) is not compatible with IPsec. */
1.94 itojun 627: m->m_flags &= ~M_CANFASTFWD;
628: #else
1.64 thorpej 629: /*
630: * Assume that we can create a fast-forward IP flow entry
631: * based on this packet.
632: */
633: m->m_flags |= M_CANFASTFWD;
1.94 itojun 634: #endif
1.64 thorpej 635:
1.36 mrg 636: #ifdef PFIL_HOOKS
1.33 mrg 637: /*
1.64 thorpej 638: * Run through list of hooks for input packets. If there are any
639: * filters which require that additional packets in the flow are
640: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
641: * Note that filters must _never_ set this flag, as another filter
642: * in the list may have previously cleared it.
1.33 mrg 643: */
1.127 itojun 644: /*
645: * let ipfilter look at packet on the wire,
646: * not the decapsulated packet.
647: */
648: #ifdef IPSEC
1.136 itojun 649: if (!ipsec_getnhist(m))
1.186 scw 650: #elif defined(FAST_IPSEC)
651: if (!ipsec_indone(m))
1.127 itojun 652: #else
653: if (1)
654: #endif
655: {
1.169 itojun 656: struct in_addr odst;
657:
658: odst = ip->ip_dst;
1.127 itojun 659: if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif,
1.168 itojun 660: PFIL_IN) != 0)
661: return;
1.127 itojun 662: if (m == NULL)
663: return;
664: ip = mtod(m, struct ip *);
1.142 darrenr 665: hlen = ip->ip_hl << 2;
1.205 darrenr 666: /*
667: * XXX The setting of "srcrt" here is to prevent ip_forward()
668: * from generating ICMP redirects for packets that have
669: * been redirected by a hook back out on to the same LAN that
670: * they came from and is not an indication that the packet
671: * is being inffluenced by source routing options. This
672: * allows things like
673: * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp"
674: * where tlp0 is both on the 1.1.1.0/24 network and is the
675: * default route for hosts on 1.1.1.0/24. Of course this
676: * also requires a "map tlp0 ..." to complete the story.
677: * One might argue whether or not this kind of network config.
1.212 perry 678: * should be supported in this manner...
1.205 darrenr 679: */
1.169 itojun 680: srcrt = (odst.s_addr != ip->ip_dst.s_addr);
1.127 itojun 681: }
1.36 mrg 682: #endif /* PFIL_HOOKS */
1.123 thorpej 683:
684: #ifdef ALTQ
685: /* XXX Temporary until ALTQ is changed to use a pfil hook */
686: if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) {
687: /* packet dropped by traffic conditioner */
688: return;
689: }
690: #endif
1.121 thorpej 691:
692: /*
1.1 cgd 693: * Process options and, if not destined for us,
694: * ship it on. ip_dooptions returns 1 when an
695: * error was detected (causing an icmp message
696: * to be sent and the original packet to be freed).
697: */
698: ip_nhops = 0; /* for source routed packets */
699: if (hlen > sizeof (struct ip) && ip_dooptions(m))
1.89 itojun 700: return;
1.1 cgd 701:
702: /*
1.165 christos 703: * Enable a consistency check between the destination address
704: * and the arrival interface for a unicast packet (the RFC 1122
705: * strong ES model) if IP forwarding is disabled and the packet
706: * is not locally generated.
707: *
708: * XXX - Checking also should be disabled if the destination
709: * address is ipnat'ed to a different interface.
710: *
711: * XXX - Checking is incompatible with IP aliases added
712: * to the loopback interface instead of the interface where
713: * the packets are received.
714: *
715: * XXX - We need to add a per ifaddr flag for this so that
716: * we get finer grain control.
717: */
718: checkif = ip_checkinterface && (ipforwarding == 0) &&
719: (m->m_pkthdr.rcvif != NULL) &&
720: ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0);
721:
722: /*
1.1 cgd 723: * Check our list of addresses, to see if the packet is for us.
1.100 itojun 724: *
725: * Traditional 4.4BSD did not consult IFF_UP at all.
726: * The behavior here is to treat addresses on !IFF_UP interface
727: * as not mine.
1.1 cgd 728: */
1.100 itojun 729: downmatch = 0;
1.140 matt 730: LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) {
1.97 itojun 731: if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
1.165 christos 732: if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif)
733: continue;
1.97 itojun 734: if ((ia->ia_ifp->if_flags & IFF_UP) != 0)
735: break;
1.100 itojun 736: else
737: downmatch++;
1.97 itojun 738: }
739: }
1.86 thorpej 740: if (ia != NULL)
741: goto ours;
1.218.2.1 yamt 742: if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
1.209 matt 743: IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) {
1.140 matt 744: if (ifa->ifa_addr->sa_family != AF_INET)
745: continue;
1.57 tls 746: ia = ifatoia(ifa);
1.35 mycroft 747: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
748: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 749: /*
750: * Look for all-0's host part (old broadcast addr),
751: * either for subnet or net.
752: */
753: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 754: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 755: goto ours;
1.57 tls 756: /*
757: * An interface with IP address zero accepts
758: * all packets that arrive on that interface.
759: */
760: if (in_nullhost(ia->ia_addr.sin_addr))
761: goto ours;
1.1 cgd 762: }
763: }
1.18 mycroft 764: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 765: struct in_multi *inm;
766: #ifdef MROUTING
767: extern struct socket *ip_mrouter;
1.10 brezak 768:
1.4 hpeyerl 769: if (ip_mrouter) {
770: /*
771: * If we are acting as a multicast router, all
772: * incoming multicast packets are passed to the
773: * kernel-level multicast forwarding function.
774: * The packet is returned (relatively) intact; if
775: * ip_mforward() returns a non-zero value, the packet
776: * must be discarded, else it may be accepted below.
777: *
778: * (The IP ident field is put in the same byte order
779: * as expected when ip_mforward() is called from
780: * ip_output().)
781: */
1.13 mycroft 782: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
783: ipstat.ips_cantforward++;
1.4 hpeyerl 784: m_freem(m);
1.89 itojun 785: return;
1.4 hpeyerl 786: }
787:
788: /*
789: * The process-level routing demon needs to receive
790: * all multicast IGMP packets, whether or not this
791: * host belongs to their destination groups.
792: */
793: if (ip->ip_p == IPPROTO_IGMP)
794: goto ours;
1.13 mycroft 795: ipstat.ips_forward++;
1.4 hpeyerl 796: }
797: #endif
798: /*
799: * See if we belong to the destination multicast group on the
800: * arrival interface.
801: */
802: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
803: if (inm == NULL) {
1.13 mycroft 804: ipstat.ips_cantforward++;
1.4 hpeyerl 805: m_freem(m);
1.89 itojun 806: return;
1.4 hpeyerl 807: }
808: goto ours;
809: }
1.19 mycroft 810: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 811: in_nullhost(ip->ip_dst))
1.1 cgd 812: goto ours;
813:
814: /*
815: * Not for us; forward if possible and desirable.
816: */
817: if (ipforwarding == 0) {
818: ipstat.ips_cantforward++;
819: m_freem(m);
1.100 itojun 820: } else {
821: /*
822: * If ip_dst matched any of my address on !IFF_UP interface,
823: * and there's no IFF_UP interface that matches ip_dst,
824: * send icmp unreach. Forwarding it will result in in-kernel
825: * forwarding loop till TTL goes to 0.
826: */
827: if (downmatch) {
828: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
829: ipstat.ips_cantforward++;
830: return;
831: }
1.145 itojun 832: #ifdef IPSEC
833: if (ipsec4_in_reject(m, NULL)) {
834: ipsecstat.in_polvio++;
835: goto bad;
836: }
837: #endif
1.173 jonathan 838: #ifdef FAST_IPSEC
839: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
840: s = splsoftnet();
841: if (mtag != NULL) {
842: tdbi = (struct tdb_ident *)(mtag + 1);
843: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
844: } else {
845: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
1.212 perry 846: IP_FORWARDING, &error);
1.173 jonathan 847: }
848: if (sp == NULL) { /* NB: can happen if error */
849: splx(s);
850: /*XXX error stat???*/
851: DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/
852: goto bad;
853: }
854:
855: /*
856: * Check security policy against packet attributes.
857: */
858: error = ipsec_in_reject(sp, m);
859: KEY_FREESP(&sp);
860: splx(s);
861: if (error) {
862: ipstat.ips_cantforward++;
863: goto bad;
1.193 scw 864: }
865:
866: /*
867: * Peek at the outbound SP for this packet to determine if
868: * it's a Fast Forward candidate.
869: */
870: mtag = m_tag_find(m, PACKET_TAG_IPSEC_PENDING_TDB, NULL);
871: if (mtag != NULL)
872: m->m_flags &= ~M_CANFASTFWD;
873: else {
874: s = splsoftnet();
875: sp = ipsec4_checkpolicy(m, IPSEC_DIR_OUTBOUND,
876: (IP_FORWARDING |
877: (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
878: &error, NULL);
879: if (sp != NULL) {
880: m->m_flags &= ~M_CANFASTFWD;
881: KEY_FREESP(&sp);
882: }
883: splx(s);
1.173 jonathan 884: }
885: #endif /* FAST_IPSEC */
1.145 itojun 886:
1.169 itojun 887: ip_forward(m, srcrt);
1.100 itojun 888: }
1.89 itojun 889: return;
1.1 cgd 890:
891: ours:
892: /*
893: * If offset or IP_MF are set, must reassemble.
894: * Otherwise, nothing need be done.
895: * (We could look in the reassembly queue to see
896: * if the packet was previously fragmented,
897: * but it's not worth the time; just let them time out.)
898: */
1.155 itojun 899: if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
900:
1.1 cgd 901: /*
902: * Look for queue of fragments
903: * of this datagram.
904: */
1.75 thorpej 905: IPQ_LOCK();
1.190 jonathan 906: hash = IPREASS_HASH(ip->ip_src.s_addr, ip->ip_id);
1.218.2.4 yamt 907: LIST_FOREACH(fp, &ipq[hash], ipq_q) {
1.1 cgd 908: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 909: in_hosteq(ip->ip_src, fp->ipq_src) &&
910: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 911: ip->ip_p == fp->ipq_p)
912: goto found;
1.190 jonathan 913:
914: }
1.1 cgd 915: fp = 0;
916: found:
917:
918: /*
919: * Adjust ip_len to not reflect header,
1.25 cgd 920: * set ipqe_mff if more fragments are expected,
1.1 cgd 921: * convert offset of this to bytes.
922: */
1.155 itojun 923: ip->ip_len = htons(ntohs(ip->ip_len) - hlen);
924: mff = (ip->ip_off & htons(IP_MF)) != 0;
1.25 cgd 925: if (mff) {
1.16 cgd 926: /*
927: * Make sure that fragments have a data length
928: * that's a non-zero multiple of 8 bytes.
929: */
1.155 itojun 930: if (ntohs(ip->ip_len) == 0 ||
931: (ntohs(ip->ip_len) & 0x7) != 0) {
1.16 cgd 932: ipstat.ips_badfrags++;
1.75 thorpej 933: IPQ_UNLOCK();
1.16 cgd 934: goto bad;
935: }
936: }
1.155 itojun 937: ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3);
1.1 cgd 938:
939: /*
940: * If datagram marked as having more fragments
941: * or if this is not the first fragment,
942: * attempt reassembly; if it succeeds, proceed.
943: */
1.155 itojun 944: if (mff || ip->ip_off != htons(0)) {
1.1 cgd 945: ipstat.ips_fragments++;
1.218.2.2 yamt 946: s = splvm();
1.72 thorpej 947: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.218.2.2 yamt 948: splx(s);
1.25 cgd 949: if (ipqe == NULL) {
950: ipstat.ips_rcvmemdrop++;
1.75 thorpej 951: IPQ_UNLOCK();
1.25 cgd 952: goto bad;
953: }
954: ipqe->ipqe_mff = mff;
1.50 thorpej 955: ipqe->ipqe_m = m;
1.25 cgd 956: ipqe->ipqe_ip = ip;
1.190 jonathan 957: m = ip_reass(ipqe, fp, &ipq[hash]);
1.75 thorpej 958: if (m == 0) {
959: IPQ_UNLOCK();
1.89 itojun 960: return;
1.75 thorpej 961: }
1.13 mycroft 962: ipstat.ips_reassembled++;
1.50 thorpej 963: ip = mtod(m, struct ip *);
1.74 thorpej 964: hlen = ip->ip_hl << 2;
1.155 itojun 965: ip->ip_len = htons(ntohs(ip->ip_len) + hlen);
1.1 cgd 966: } else
967: if (fp)
968: ip_freef(fp);
1.75 thorpej 969: IPQ_UNLOCK();
1.79 mycroft 970: }
1.128 itojun 971:
1.173 jonathan 972: #if defined(IPSEC)
1.128 itojun 973: /*
974: * enforce IPsec policy checking if we are seeing last header.
975: * note that we do not visit this with protocols with pcb layer
976: * code - like udp/tcp/raw ip.
977: */
978: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 &&
979: ipsec4_in_reject(m, NULL)) {
980: ipsecstat.in_polvio++;
981: goto bad;
982: }
983: #endif
1.218.2.1 yamt 984: #ifdef FAST_IPSEC
1.173 jonathan 985: /*
986: * enforce IPsec policy checking if we are seeing last header.
987: * note that we do not visit this with protocols with pcb layer
988: * code - like udp/tcp/raw ip.
989: */
990: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) {
991: /*
992: * Check if the packet has already had IPsec processing
993: * done. If so, then just pass it along. This tag gets
994: * set during AH, ESP, etc. input handling, before the
995: * packet is returned to the ip input queue for delivery.
1.212 perry 996: */
1.173 jonathan 997: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
998: s = splsoftnet();
999: if (mtag != NULL) {
1000: tdbi = (struct tdb_ident *)(mtag + 1);
1001: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
1002: } else {
1003: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
1.212 perry 1004: IP_FORWARDING, &error);
1.173 jonathan 1005: }
1006: if (sp != NULL) {
1007: /*
1008: * Check security policy against packet attributes.
1009: */
1010: error = ipsec_in_reject(sp, m);
1011: KEY_FREESP(&sp);
1012: } else {
1013: /* XXX error stat??? */
1014: error = EINVAL;
1015: DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/
1016: }
1017: splx(s);
1018: if (error)
1019: goto bad;
1020: }
1021: #endif /* FAST_IPSEC */
1.1 cgd 1022:
1023: /*
1024: * Switch out to protocol's input routine.
1025: */
1.82 aidan 1026: #if IFA_STATS
1.122 itojun 1027: if (ia && ip)
1.155 itojun 1028: ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
1.82 aidan 1029: #endif
1.1 cgd 1030: ipstat.ips_delivered++;
1.89 itojun 1031: {
1032: int off = hlen, nh = ip->ip_p;
1033:
1034: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
1035: return;
1036: }
1.1 cgd 1037: bad:
1038: m_freem(m);
1.135 thorpej 1039: return;
1040:
1041: badcsum:
1042: ipstat.ips_badsum++;
1043: m_freem(m);
1.1 cgd 1044: }
1045:
1046: /*
1047: * Take incoming datagram fragment and try to
1048: * reassemble it into whole datagram. If a chain for
1049: * reassembly of this datagram already exists, then it
1050: * is given as fp; otherwise have to make a chain.
1051: */
1.50 thorpej 1052: struct mbuf *
1.211 perry 1053: ip_reass(struct ipqent *ipqe, struct ipq *fp, struct ipqhead *ipqhead)
1.1 cgd 1054: {
1.109 augustss 1055: struct mbuf *m = ipqe->ipqe_m;
1056: struct ipqent *nq, *p, *q;
1.25 cgd 1057: struct ip *ip;
1.1 cgd 1058: struct mbuf *t;
1.25 cgd 1059: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.218.2.2 yamt 1060: int i, next, s;
1.1 cgd 1061:
1.75 thorpej 1062: IPQ_LOCK_CHECK();
1063:
1.1 cgd 1064: /*
1065: * Presence of header sizes in mbufs
1066: * would confuse code below.
1067: */
1068: m->m_data += hlen;
1069: m->m_len -= hlen;
1070:
1.194 jonathan 1071: #ifdef notyet
1072: /* make sure fragment limit is up-to-date */
1073: CHECK_NMBCLUSTER_PARAMS();
1074:
1075: /* If we have too many fragments, drop the older half. */
1076: if (ip_nfrags >= ip_maxfrags)
1077: ip_reass_drophalf(void);
1078: #endif
1079:
1.1 cgd 1080: /*
1.192 jonathan 1081: * We are about to add a fragment; increment frag count.
1082: */
1083: ip_nfrags++;
1.212 perry 1084:
1.192 jonathan 1085: /*
1.1 cgd 1086: * If first fragment to arrive, create a reassembly queue.
1087: */
1088: if (fp == 0) {
1.131 itojun 1089: /*
1090: * Enforce upper bound on number of fragmented packets
1091: * for which we attempt reassembly;
1092: * If maxfrag is 0, never accept fragments.
1093: * If maxfrag is -1, accept all fragments without limitation.
1094: */
1095: if (ip_maxfragpackets < 0)
1096: ;
1097: else if (ip_nfragpackets >= ip_maxfragpackets)
1098: goto dropfrag;
1099: ip_nfragpackets++;
1.50 thorpej 1100: MALLOC(fp, struct ipq *, sizeof (struct ipq),
1101: M_FTABLE, M_NOWAIT);
1102: if (fp == NULL)
1.1 cgd 1103: goto dropfrag;
1.190 jonathan 1104: LIST_INSERT_HEAD(ipqhead, fp, ipq_q);
1.192 jonathan 1105: fp->ipq_nfrags = 1;
1.1 cgd 1106: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 1107: fp->ipq_p = ipqe->ipqe_ip->ip_p;
1108: fp->ipq_id = ipqe->ipqe_ip->ip_id;
1.148 matt 1109: TAILQ_INIT(&fp->ipq_fragq);
1.25 cgd 1110: fp->ipq_src = ipqe->ipqe_ip->ip_src;
1111: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
1112: p = NULL;
1.1 cgd 1113: goto insert;
1.192 jonathan 1114: } else {
1115: fp->ipq_nfrags++;
1.1 cgd 1116: }
1117:
1118: /*
1119: * Find a segment which begins after this one does.
1120: */
1.148 matt 1121: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1122: p = q, q = TAILQ_NEXT(q, ipqe_q))
1.155 itojun 1123: if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off))
1.1 cgd 1124: break;
1125:
1126: /*
1127: * If there is a preceding segment, it may provide some of
1128: * our data already. If so, drop the data from the incoming
1129: * segment. If it provides all of our data, drop us.
1130: */
1.25 cgd 1131: if (p != NULL) {
1.155 itojun 1132: i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) -
1133: ntohs(ipqe->ipqe_ip->ip_off);
1.1 cgd 1134: if (i > 0) {
1.155 itojun 1135: if (i >= ntohs(ipqe->ipqe_ip->ip_len))
1.1 cgd 1136: goto dropfrag;
1.50 thorpej 1137: m_adj(ipqe->ipqe_m, i);
1.155 itojun 1138: ipqe->ipqe_ip->ip_off =
1139: htons(ntohs(ipqe->ipqe_ip->ip_off) + i);
1140: ipqe->ipqe_ip->ip_len =
1141: htons(ntohs(ipqe->ipqe_ip->ip_len) - i);
1.1 cgd 1142: }
1143: }
1144:
1145: /*
1146: * While we overlap succeeding segments trim them or,
1147: * if they are completely covered, dequeue them.
1148: */
1.155 itojun 1149: for (; q != NULL &&
1150: ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) >
1151: ntohs(q->ipqe_ip->ip_off); q = nq) {
1152: i = (ntohs(ipqe->ipqe_ip->ip_off) +
1153: ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off);
1154: if (i < ntohs(q->ipqe_ip->ip_len)) {
1155: q->ipqe_ip->ip_len =
1156: htons(ntohs(q->ipqe_ip->ip_len) - i);
1157: q->ipqe_ip->ip_off =
1158: htons(ntohs(q->ipqe_ip->ip_off) + i);
1.50 thorpej 1159: m_adj(q->ipqe_m, i);
1.1 cgd 1160: break;
1161: }
1.148 matt 1162: nq = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1163: m_freem(q->ipqe_m);
1.148 matt 1164: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.218.2.2 yamt 1165: s = splvm();
1.72 thorpej 1166: pool_put(&ipqent_pool, q);
1.218.2.2 yamt 1167: splx(s);
1.192 jonathan 1168: fp->ipq_nfrags--;
1169: ip_nfrags--;
1.1 cgd 1170: }
1171:
1172: insert:
1173: /*
1174: * Stick new segment in its place;
1175: * check for complete reassembly.
1176: */
1.25 cgd 1177: if (p == NULL) {
1.148 matt 1178: TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
1.25 cgd 1179: } else {
1.148 matt 1180: TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q);
1.25 cgd 1181: }
1.1 cgd 1182: next = 0;
1.148 matt 1183: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1184: p = q, q = TAILQ_NEXT(q, ipqe_q)) {
1.155 itojun 1185: if (ntohs(q->ipqe_ip->ip_off) != next)
1.1 cgd 1186: return (0);
1.155 itojun 1187: next += ntohs(q->ipqe_ip->ip_len);
1.1 cgd 1188: }
1.25 cgd 1189: if (p->ipqe_mff)
1.1 cgd 1190: return (0);
1191:
1192: /*
1.41 thorpej 1193: * Reassembly is complete. Check for a bogus message size and
1194: * concatenate fragments.
1.1 cgd 1195: */
1.148 matt 1196: q = TAILQ_FIRST(&fp->ipq_fragq);
1.25 cgd 1197: ip = q->ipqe_ip;
1.41 thorpej 1198: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
1199: ipstat.ips_toolong++;
1200: ip_freef(fp);
1201: return (0);
1202: }
1.50 thorpej 1203: m = q->ipqe_m;
1.1 cgd 1204: t = m->m_next;
1205: m->m_next = 0;
1206: m_cat(m, t);
1.148 matt 1207: nq = TAILQ_NEXT(q, ipqe_q);
1.218.2.2 yamt 1208: s = splvm();
1.72 thorpej 1209: pool_put(&ipqent_pool, q);
1.218.2.2 yamt 1210: splx(s);
1.25 cgd 1211: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 1212: t = q->ipqe_m;
1.148 matt 1213: nq = TAILQ_NEXT(q, ipqe_q);
1.218.2.2 yamt 1214: s = splvm();
1.72 thorpej 1215: pool_put(&ipqent_pool, q);
1.218.2.2 yamt 1216: splx(s);
1.1 cgd 1217: m_cat(m, t);
1218: }
1.192 jonathan 1219: ip_nfrags -= fp->ipq_nfrags;
1.1 cgd 1220:
1221: /*
1222: * Create header for new ip packet by
1223: * modifying header of first packet;
1224: * dequeue and discard fragment reassembly header.
1225: * Make header visible.
1226: */
1.155 itojun 1227: ip->ip_len = htons(next);
1.25 cgd 1228: ip->ip_src = fp->ipq_src;
1229: ip->ip_dst = fp->ipq_dst;
1230: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1231: FREE(fp, M_FTABLE);
1.131 itojun 1232: ip_nfragpackets--;
1.1 cgd 1233: m->m_len += (ip->ip_hl << 2);
1234: m->m_data -= (ip->ip_hl << 2);
1235: /* some debugging cruft by sklower, below, will go away soon */
1236: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
1.109 augustss 1237: int plen = 0;
1.50 thorpej 1238: for (t = m; t; t = t->m_next)
1239: plen += t->m_len;
1240: m->m_pkthdr.len = plen;
1.213 yamt 1241: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1242: }
1.50 thorpej 1243: return (m);
1.1 cgd 1244:
1245: dropfrag:
1.192 jonathan 1246: if (fp != 0)
1247: fp->ipq_nfrags--;
1248: ip_nfrags--;
1.1 cgd 1249: ipstat.ips_fragdropped++;
1250: m_freem(m);
1.218.2.2 yamt 1251: s = splvm();
1.72 thorpej 1252: pool_put(&ipqent_pool, ipqe);
1.218.2.2 yamt 1253: splx(s);
1.1 cgd 1254: return (0);
1255: }
1256:
1257: /*
1258: * Free a fragment reassembly header and all
1259: * associated datagrams.
1260: */
1.8 mycroft 1261: void
1.211 perry 1262: ip_freef(struct ipq *fp)
1.1 cgd 1263: {
1.109 augustss 1264: struct ipqent *q, *p;
1.192 jonathan 1265: u_int nfrags = 0;
1.218.2.2 yamt 1266: int s;
1.1 cgd 1267:
1.75 thorpej 1268: IPQ_LOCK_CHECK();
1269:
1.148 matt 1270: for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) {
1271: p = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1272: m_freem(q->ipqe_m);
1.192 jonathan 1273: nfrags++;
1.148 matt 1274: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.218.2.2 yamt 1275: s = splvm();
1.72 thorpej 1276: pool_put(&ipqent_pool, q);
1.218.2.2 yamt 1277: splx(s);
1.1 cgd 1278: }
1.192 jonathan 1279:
1280: if (nfrags != fp->ipq_nfrags)
1281: printf("ip_freef: nfrags %d != %d\n", fp->ipq_nfrags, nfrags);
1282: ip_nfrags -= nfrags;
1.25 cgd 1283: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1284: FREE(fp, M_FTABLE);
1.131 itojun 1285: ip_nfragpackets--;
1.1 cgd 1286: }
1287:
1288: /*
1.194 jonathan 1289: * IP reassembly TTL machinery for multiplicative drop.
1290: */
1291: static u_int fragttl_histo[(IPFRAGTTL+1)];
1292:
1293:
1294: /*
1295: * Decrement TTL of all reasembly queue entries by `ticks'.
1296: * Count number of distinct fragments (as opposed to partial, fragmented
1297: * datagrams) in the reassembly queue. While we traverse the entire
1298: * reassembly queue, compute and return the median TTL over all fragments.
1299: */
1300: static u_int
1301: ip_reass_ttl_decr(u_int ticks)
1302: {
1.198 matt 1303: u_int nfrags, median, dropfraction, keepfraction;
1.194 jonathan 1304: struct ipq *fp, *nfp;
1.198 matt 1305: int i;
1.212 perry 1306:
1.194 jonathan 1307: nfrags = 0;
1308: memset(fragttl_histo, 0, sizeof fragttl_histo);
1.212 perry 1309:
1.194 jonathan 1310: for (i = 0; i < IPREASS_NHASH; i++) {
1311: for (fp = LIST_FIRST(&ipq[i]); fp != NULL; fp = nfp) {
1312: fp->ipq_ttl = ((fp->ipq_ttl <= ticks) ?
1313: 0 : fp->ipq_ttl - ticks);
1314: nfp = LIST_NEXT(fp, ipq_q);
1315: if (fp->ipq_ttl == 0) {
1316: ipstat.ips_fragtimeout++;
1317: ip_freef(fp);
1318: } else {
1319: nfrags += fp->ipq_nfrags;
1320: fragttl_histo[fp->ipq_ttl] += fp->ipq_nfrags;
1321: }
1322: }
1323: }
1324:
1325: KASSERT(ip_nfrags == nfrags);
1326:
1327: /* Find median (or other drop fraction) in histogram. */
1328: dropfraction = (ip_nfrags / 2);
1329: keepfraction = ip_nfrags - dropfraction;
1330: for (i = IPFRAGTTL, median = 0; i >= 0; i--) {
1331: median += fragttl_histo[i];
1332: if (median >= keepfraction)
1333: break;
1334: }
1335:
1336: /* Return TTL of median (or other fraction). */
1337: return (u_int)i;
1338: }
1339:
1340: void
1341: ip_reass_drophalf(void)
1342: {
1343:
1344: u_int median_ticks;
1345: /*
1346: * Compute median TTL of all fragments, and count frags
1347: * with that TTL or lower (roughly half of all fragments).
1348: */
1349: median_ticks = ip_reass_ttl_decr(0);
1350:
1351: /* Drop half. */
1352: median_ticks = ip_reass_ttl_decr(median_ticks);
1353:
1354: }
1355:
1356: /*
1.1 cgd 1357: * IP timer processing;
1358: * if a timer expires on a reassembly
1359: * queue, discard it.
1360: */
1.8 mycroft 1361: void
1.211 perry 1362: ip_slowtimo(void)
1.1 cgd 1363: {
1.191 jonathan 1364: static u_int dropscanidx = 0;
1365: u_int i;
1.194 jonathan 1366: u_int median_ttl;
1.24 mycroft 1367: int s = splsoftnet();
1.1 cgd 1368:
1.75 thorpej 1369: IPQ_LOCK();
1.194 jonathan 1370:
1371: /* Age TTL of all fragments by 1 tick .*/
1372: median_ttl = ip_reass_ttl_decr(1);
1373:
1374: /* make sure fragment limit is up-to-date */
1375: CHECK_NMBCLUSTER_PARAMS();
1376:
1377: /* If we have too many fragments, drop the older half. */
1378: if (ip_nfrags > ip_maxfrags)
1379: ip_reass_ttl_decr(median_ttl);
1380:
1.131 itojun 1381: /*
1.194 jonathan 1382: * If we are over the maximum number of fragmented packets
1.131 itojun 1383: * (due to the limit being lowered), drain off
1.190 jonathan 1384: * enough to get down to the new limit. Start draining
1385: * from the reassembly hashqueue most recently drained.
1.131 itojun 1386: */
1387: if (ip_maxfragpackets < 0)
1388: ;
1389: else {
1.190 jonathan 1390: int wrapped = 0;
1391:
1392: i = dropscanidx;
1393: while (ip_nfragpackets > ip_maxfragpackets && wrapped == 0) {
1394: while (LIST_FIRST(&ipq[i]) != NULL)
1395: ip_freef(LIST_FIRST(&ipq[i]));
1396: if (++i >= IPREASS_NHASH) {
1397: i = 0;
1398: }
1399: /*
1400: * Dont scan forever even if fragment counters are
1401: * wrong: stop after scanning entire reassembly queue.
1402: */
1403: if (i == dropscanidx)
1404: wrapped = 1;
1405: }
1406: dropscanidx = i;
1.131 itojun 1407: }
1.75 thorpej 1408: IPQ_UNLOCK();
1.1 cgd 1409: splx(s);
1410: }
1411:
1412: /*
1413: * Drain off all datagram fragments.
1414: */
1.8 mycroft 1415: void
1.211 perry 1416: ip_drain(void)
1.1 cgd 1417: {
1418:
1.75 thorpej 1419: /*
1420: * We may be called from a device's interrupt context. If
1421: * the ipq is already busy, just bail out now.
1422: */
1423: if (ipq_lock_try() == 0)
1424: return;
1425:
1.194 jonathan 1426: /*
1427: * Drop half the total fragments now. If more mbufs are needed,
1428: * we will be called again soon.
1429: */
1430: ip_reass_drophalf();
1.75 thorpej 1431:
1432: IPQ_UNLOCK();
1.1 cgd 1433: }
1434:
1435: /*
1436: * Do option processing on a datagram,
1437: * possibly discarding it if bad options are encountered,
1438: * or forwarding it if source-routed.
1439: * Returns 1 if packet has been forwarded/freed,
1440: * 0 if the packet should be processed further.
1441: */
1.8 mycroft 1442: int
1.211 perry 1443: ip_dooptions(struct mbuf *m)
1.1 cgd 1444: {
1.109 augustss 1445: struct ip *ip = mtod(m, struct ip *);
1446: u_char *cp, *cp0;
1447: struct ip_timestamp *ipt;
1448: struct in_ifaddr *ia;
1.1 cgd 1449: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 1450: struct in_addr dst;
1.1 cgd 1451: n_time ntime;
1452:
1.13 mycroft 1453: dst = ip->ip_dst;
1.1 cgd 1454: cp = (u_char *)(ip + 1);
1455: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
1456: for (; cnt > 0; cnt -= optlen, cp += optlen) {
1457: opt = cp[IPOPT_OPTVAL];
1458: if (opt == IPOPT_EOL)
1459: break;
1460: if (opt == IPOPT_NOP)
1461: optlen = 1;
1462: else {
1.113 itojun 1463: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1464: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1465: goto bad;
1466: }
1.1 cgd 1467: optlen = cp[IPOPT_OLEN];
1.114 itojun 1468: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 1469: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1470: goto bad;
1471: }
1472: }
1473: switch (opt) {
1474:
1475: default:
1476: break;
1477:
1478: /*
1479: * Source routing with record.
1480: * Find interface with current destination address.
1481: * If none on this machine then drop if strictly routed,
1482: * or do nothing if loosely routed.
1483: * Record interface address and bring up next address
1484: * component. If strictly routed make sure next
1485: * address is on directly accessible net.
1486: */
1487: case IPOPT_LSRR:
1488: case IPOPT_SSRR:
1.47 cjs 1489: if (ip_allowsrcrt == 0) {
1490: type = ICMP_UNREACH;
1491: code = ICMP_UNREACH_NET_PROHIB;
1492: goto bad;
1493: }
1.114 itojun 1494: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1495: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1496: goto bad;
1497: }
1.1 cgd 1498: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1499: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1500: goto bad;
1501: }
1502: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 1503: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 1504: if (ia == 0) {
1505: if (opt == IPOPT_SSRR) {
1506: type = ICMP_UNREACH;
1507: code = ICMP_UNREACH_SRCFAIL;
1508: goto bad;
1509: }
1510: /*
1511: * Loose routing, and not at next destination
1512: * yet; nothing to do except forward.
1513: */
1514: break;
1515: }
1516: off--; /* 0 origin */
1.112 sommerfe 1517: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 1518: /*
1519: * End of source route. Should be for us.
1520: */
1521: save_rte(cp, ip->ip_src);
1522: break;
1523: }
1524: /*
1525: * locate outgoing interface
1526: */
1.218.2.4 yamt 1527: bcopy((void *)(cp + off), (void *)&ipaddr.sin_addr,
1.1 cgd 1528: sizeof(ipaddr.sin_addr));
1.96 thorpej 1529: if (opt == IPOPT_SSRR)
1.196 itojun 1530: ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr)));
1.96 thorpej 1531: else
1.1 cgd 1532: ia = ip_rtaddr(ipaddr.sin_addr);
1533: if (ia == 0) {
1534: type = ICMP_UNREACH;
1535: code = ICMP_UNREACH_SRCFAIL;
1536: goto bad;
1537: }
1538: ip->ip_dst = ipaddr.sin_addr;
1.218.2.4 yamt 1539: bcopy((void *)&ia->ia_addr.sin_addr,
1540: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1541: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1542: /*
1543: * Let ip_intr's mcast routing check handle mcast pkts
1544: */
1.18 mycroft 1545: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1546: break;
1547:
1548: case IPOPT_RR:
1.114 itojun 1549: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1550: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1551: goto bad;
1552: }
1.1 cgd 1553: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1554: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1555: goto bad;
1556: }
1557: /*
1558: * If no space remains, ignore.
1559: */
1560: off--; /* 0 origin */
1.112 sommerfe 1561: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1562: break;
1.218.2.4 yamt 1563: bcopy((void *)(&ip->ip_dst), (void *)&ipaddr.sin_addr,
1.1 cgd 1564: sizeof(ipaddr.sin_addr));
1565: /*
1566: * locate outgoing interface; if we're the destination,
1567: * use the incoming interface (should be same).
1568: */
1.96 thorpej 1569: if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))))
1570: == NULL &&
1571: (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
1.1 cgd 1572: type = ICMP_UNREACH;
1573: code = ICMP_UNREACH_HOST;
1574: goto bad;
1575: }
1.218.2.4 yamt 1576: bcopy((void *)&ia->ia_addr.sin_addr,
1577: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1578: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1579: break;
1580:
1581: case IPOPT_TS:
1582: code = cp - (u_char *)ip;
1583: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1584: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1585: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1586: goto bad;
1.114 itojun 1587: }
1588: if (ipt->ipt_ptr < 5) {
1589: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1590: goto bad;
1591: }
1.15 cgd 1592: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1593: if (++ipt->ipt_oflw == 0) {
1594: code = (u_char *)&ipt->ipt_ptr -
1595: (u_char *)ip;
1.1 cgd 1596: goto bad;
1.114 itojun 1597: }
1.1 cgd 1598: break;
1599: }
1.104 thorpej 1600: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1601: switch (ipt->ipt_flg) {
1602:
1603: case IPOPT_TS_TSONLY:
1604: break;
1605:
1606: case IPOPT_TS_TSANDADDR:
1.66 thorpej 1607: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1608: sizeof(struct in_addr) > ipt->ipt_len) {
1609: code = (u_char *)&ipt->ipt_ptr -
1610: (u_char *)ip;
1.1 cgd 1611: goto bad;
1.114 itojun 1612: }
1.13 mycroft 1613: ipaddr.sin_addr = dst;
1.96 thorpej 1614: ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr),
1615: m->m_pkthdr.rcvif));
1.13 mycroft 1616: if (ia == 0)
1617: continue;
1.104 thorpej 1618: bcopy(&ia->ia_addr.sin_addr,
1619: cp0, sizeof(struct in_addr));
1.1 cgd 1620: ipt->ipt_ptr += sizeof(struct in_addr);
1621: break;
1622:
1623: case IPOPT_TS_PRESPEC:
1.66 thorpej 1624: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1625: sizeof(struct in_addr) > ipt->ipt_len) {
1626: code = (u_char *)&ipt->ipt_ptr -
1627: (u_char *)ip;
1.1 cgd 1628: goto bad;
1.114 itojun 1629: }
1.104 thorpej 1630: bcopy(cp0, &ipaddr.sin_addr,
1.1 cgd 1631: sizeof(struct in_addr));
1.96 thorpej 1632: if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))
1633: == NULL)
1.1 cgd 1634: continue;
1635: ipt->ipt_ptr += sizeof(struct in_addr);
1636: break;
1637:
1638: default:
1.114 itojun 1639: /* XXX can't take &ipt->ipt_flg */
1640: code = (u_char *)&ipt->ipt_ptr -
1641: (u_char *)ip + 1;
1.1 cgd 1642: goto bad;
1643: }
1644: ntime = iptime();
1.107 thorpej 1645: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1.218.2.4 yamt 1646: memmove((char *)cp + ipt->ipt_ptr - 1, cp0,
1.1 cgd 1647: sizeof(n_time));
1648: ipt->ipt_ptr += sizeof(n_time);
1649: }
1650: }
1651: if (forward) {
1.26 thorpej 1652: if (ip_forwsrcrt == 0) {
1653: type = ICMP_UNREACH;
1654: code = ICMP_UNREACH_SRCFAIL;
1655: goto bad;
1656: }
1.1 cgd 1657: ip_forward(m, 1);
1658: return (1);
1.13 mycroft 1659: }
1660: return (0);
1.1 cgd 1661: bad:
1.13 mycroft 1662: icmp_error(m, type, code, 0, 0);
1663: ipstat.ips_badoptions++;
1.1 cgd 1664: return (1);
1665: }
1666:
1667: /*
1668: * Given address of next destination (final or next hop),
1669: * return internet address info of interface to be used to get there.
1670: */
1671: struct in_ifaddr *
1.211 perry 1672: ip_rtaddr(struct in_addr dst)
1.1 cgd 1673: {
1.218.2.4 yamt 1674: struct rtentry *rt;
1675: union {
1676: struct sockaddr dst;
1677: struct sockaddr_in dst4;
1678: } u;
1679:
1680: sockaddr_in_init(&u.dst4, &dst, 0);
1681:
1682: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL)
1683: return NULL;
1684:
1685: return ifatoia(rt->rt_ifa);
1.1 cgd 1686: }
1687:
1688: /*
1689: * Save incoming source route for use in replies,
1690: * to be picked up later by ip_srcroute if the receiver is interested.
1691: */
1.13 mycroft 1692: void
1.211 perry 1693: save_rte(u_char *option, struct in_addr dst)
1.1 cgd 1694: {
1695: unsigned olen;
1696:
1697: olen = option[IPOPT_OLEN];
1698: #ifdef DIAGNOSTIC
1699: if (ipprintfs)
1.39 christos 1700: printf("save_rte: olen %d\n", olen);
1.89 itojun 1701: #endif /* 0 */
1.1 cgd 1702: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1703: return;
1.218.2.4 yamt 1704: bcopy((void *)option, (void *)ip_srcrt.srcopt, olen);
1.1 cgd 1705: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1706: ip_srcrt.dst = dst;
1707: }
1708:
1709: /*
1710: * Retrieve incoming source route for use in replies,
1711: * in the same form used by setsockopt.
1712: * The first hop is placed before the options, will be removed later.
1713: */
1714: struct mbuf *
1.211 perry 1715: ip_srcroute(void)
1.1 cgd 1716: {
1.109 augustss 1717: struct in_addr *p, *q;
1718: struct mbuf *m;
1.1 cgd 1719:
1720: if (ip_nhops == 0)
1.218.2.2 yamt 1721: return NULL;
1.1 cgd 1722: m = m_get(M_DONTWAIT, MT_SOOPTS);
1723: if (m == 0)
1.218.2.2 yamt 1724: return NULL;
1.1 cgd 1725:
1.164 matt 1726: MCLAIM(m, &inetdomain.dom_mowner);
1.13 mycroft 1727: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1728:
1729: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1730: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1731: OPTSIZ;
1732: #ifdef DIAGNOSTIC
1733: if (ipprintfs)
1.39 christos 1734: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1735: #endif
1736:
1737: /*
1738: * First save first hop for return route
1739: */
1740: p = &ip_srcrt.route[ip_nhops - 1];
1741: *(mtod(m, struct in_addr *)) = *p--;
1742: #ifdef DIAGNOSTIC
1743: if (ipprintfs)
1.39 christos 1744: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1745: #endif
1746:
1747: /*
1748: * Copy option fields and padding (nop) to mbuf.
1749: */
1750: ip_srcrt.nop = IPOPT_NOP;
1751: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1.218.2.4 yamt 1752: memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop,
1753: OPTSIZ);
1754: q = (struct in_addr *)(mtod(m, char *) +
1.1 cgd 1755: sizeof(struct in_addr) + OPTSIZ);
1756: #undef OPTSIZ
1757: /*
1758: * Record return path as an IP source route,
1759: * reversing the path (pointers are now aligned).
1760: */
1761: while (p >= ip_srcrt.route) {
1762: #ifdef DIAGNOSTIC
1763: if (ipprintfs)
1.39 christos 1764: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1765: #endif
1766: *q++ = *p--;
1767: }
1768: /*
1769: * Last hop goes to final destination.
1770: */
1771: *q = ip_srcrt.dst;
1772: #ifdef DIAGNOSTIC
1773: if (ipprintfs)
1.39 christos 1774: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1775: #endif
1776: return (m);
1777: }
1778:
1.139 matt 1779: const int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1780: 0, 0, 0, 0,
1781: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1782: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1783: EMSGSIZE, EHOSTUNREACH, 0, 0,
1784: 0, 0, 0, 0,
1785: ENOPROTOOPT
1786: };
1787:
1788: /*
1789: * Forward a packet. If some error occurs return the sender
1790: * an icmp packet. Note we can't always generate a meaningful
1791: * icmp message because icmp doesn't have a large enough repertoire
1792: * of codes and types.
1793: *
1794: * If not forwarding, just drop the packet. This could be confusing
1795: * if ipforwarding was zero but some routing protocol was advancing
1796: * us as a gateway to somewhere. However, we must let the routing
1797: * protocol deal with that.
1798: *
1799: * The srcrt parameter indicates whether the packet is being forwarded
1800: * via a source route.
1801: */
1.13 mycroft 1802: void
1.211 perry 1803: ip_forward(struct mbuf *m, int srcrt)
1.1 cgd 1804: {
1.109 augustss 1805: struct ip *ip = mtod(m, struct ip *);
1806: struct rtentry *rt;
1.218.2.1 yamt 1807: int error, type = 0, code = 0, destmtu = 0;
1.1 cgd 1808: struct mbuf *mcopy;
1.13 mycroft 1809: n_long dest;
1.218.2.4 yamt 1810: union {
1811: struct sockaddr dst;
1812: struct sockaddr_in dst4;
1813: } u;
1.164 matt 1814:
1815: /*
1816: * We are now in the output path.
1817: */
1818: MCLAIM(m, &ip_tx_mowner);
1.135 thorpej 1819:
1820: /*
1821: * Clear any in-bound checksum flags for this packet.
1822: */
1823: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1824:
1.13 mycroft 1825: dest = 0;
1.1 cgd 1826: #ifdef DIAGNOSTIC
1.218.2.1 yamt 1827: if (ipprintfs) {
1828: printf("forward: src %s ", inet_ntoa(ip->ip_src));
1829: printf("dst %s ttl %x\n", inet_ntoa(ip->ip_dst), ip->ip_ttl);
1830: }
1.1 cgd 1831: #endif
1.93 sommerfe 1832: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.1 cgd 1833: ipstat.ips_cantforward++;
1834: m_freem(m);
1835: return;
1836: }
1837: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1838: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1839: return;
1840: }
1841:
1.218.2.4 yamt 1842: sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
1843: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) {
1844: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0);
1845: return;
1.1 cgd 1846: }
1847:
1848: /*
1.34 mycroft 1849: * Save at most 68 bytes of the packet in case
1.1 cgd 1850: * we need to generate an ICMP message to the src.
1.119 itojun 1851: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1852: */
1.155 itojun 1853: mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1.119 itojun 1854: if (mcopy)
1855: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1856:
1.218.2.1 yamt 1857: ip->ip_ttl -= IPTTLDEC;
1858:
1.1 cgd 1859: /*
1860: * If forwarding packet using same interface that it came in on,
1861: * perhaps should send a redirect to sender to shortcut a hop.
1862: * Only send redirect if source is sending directly to us,
1863: * and if packet was not source routed (or has any options).
1864: * Also, don't send redirect if forwarding using a default route
1865: * or a route modified by a redirect.
1866: */
1867: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1868: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.218.2.4 yamt 1869: !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) &&
1.1 cgd 1870: ipsendredirects && !srcrt) {
1.19 mycroft 1871: if (rt->rt_ifa &&
1872: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1873: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1874: if (rt->rt_flags & RTF_GATEWAY)
1875: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1876: else
1877: dest = ip->ip_dst.s_addr;
1878: /*
1879: * Router requirements says to only send host
1880: * redirects.
1881: */
1882: type = ICMP_REDIRECT;
1883: code = ICMP_REDIRECT_HOST;
1.1 cgd 1884: #ifdef DIAGNOSTIC
1.77 thorpej 1885: if (ipprintfs)
1886: printf("redirect (%d) to %x\n", code,
1887: (u_int32_t)dest);
1.1 cgd 1888: #endif
1889: }
1890: }
1891:
1.218.2.2 yamt 1892: error = ip_output(m, NULL, &ipforward_rt,
1.173 jonathan 1893: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
1.174 itojun 1894: (struct ip_moptions *)NULL, (struct socket *)NULL);
1.173 jonathan 1895:
1.1 cgd 1896: if (error)
1897: ipstat.ips_cantforward++;
1898: else {
1899: ipstat.ips_forward++;
1900: if (type)
1901: ipstat.ips_redirectsent++;
1902: else {
1.63 matt 1903: if (mcopy) {
1904: #ifdef GATEWAY
1.64 thorpej 1905: if (mcopy->m_flags & M_CANFASTFWD)
1906: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1907: #endif
1.1 cgd 1908: m_freem(mcopy);
1.63 matt 1909: }
1.1 cgd 1910: return;
1911: }
1912: }
1913: if (mcopy == NULL)
1914: return;
1.13 mycroft 1915:
1.1 cgd 1916: switch (error) {
1917:
1918: case 0: /* forwarded, but need redirect */
1919: /* type, code set above */
1920: break;
1921:
1922: case ENETUNREACH: /* shouldn't happen, checked above */
1923: case EHOSTUNREACH:
1924: case ENETDOWN:
1925: case EHOSTDOWN:
1926: default:
1927: type = ICMP_UNREACH;
1928: code = ICMP_UNREACH_HOST;
1929: break;
1930:
1931: case EMSGSIZE:
1932: type = ICMP_UNREACH;
1933: code = ICMP_UNREACH_NEEDFRAG;
1.173 jonathan 1934: #if !defined(IPSEC) && !defined(FAST_IPSEC)
1.218.2.2 yamt 1935: if (ipforward_rt.ro_rt != NULL)
1.218.2.1 yamt 1936: destmtu = ipforward_rt.ro_rt->rt_ifp->if_mtu;
1.89 itojun 1937: #else
1938: /*
1939: * If the packet is routed over IPsec tunnel, tell the
1940: * originator the tunnel MTU.
1941: * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1942: * XXX quickhack!!!
1943: */
1.218.2.2 yamt 1944: if (ipforward_rt.ro_rt != NULL) {
1.89 itojun 1945: struct secpolicy *sp;
1946: int ipsecerror;
1.95 itojun 1947: size_t ipsechdr;
1.89 itojun 1948: struct route *ro;
1949:
1950: sp = ipsec4_getpolicybyaddr(mcopy,
1.170 itojun 1951: IPSEC_DIR_OUTBOUND, IP_FORWARDING,
1952: &ipsecerror);
1.89 itojun 1953:
1954: if (sp == NULL)
1.218.2.1 yamt 1955: destmtu = ipforward_rt.ro_rt->rt_ifp->if_mtu;
1.89 itojun 1956: else {
1957: /* count IPsec header size */
1.95 itojun 1958: ipsechdr = ipsec4_hdrsiz(mcopy,
1.170 itojun 1959: IPSEC_DIR_OUTBOUND, NULL);
1.89 itojun 1960:
1961: /*
1962: * find the correct route for outer IPv4
1963: * header, compute tunnel MTU.
1964: */
1.218.2.1 yamt 1965:
1.89 itojun 1966: if (sp->req != NULL
1.95 itojun 1967: && sp->req->sav != NULL
1968: && sp->req->sav->sah != NULL) {
1969: ro = &sp->req->sav->sah->sa_route;
1.89 itojun 1970: if (ro->ro_rt && ro->ro_rt->rt_ifp) {
1.218.2.1 yamt 1971: destmtu =
1.151 itojun 1972: ro->ro_rt->rt_rmx.rmx_mtu ?
1973: ro->ro_rt->rt_rmx.rmx_mtu :
1.89 itojun 1974: ro->ro_rt->rt_ifp->if_mtu;
1.218.2.1 yamt 1975: destmtu -= ipsechdr;
1.89 itojun 1976: }
1977: }
1978:
1.173 jonathan 1979: #ifdef IPSEC
1.89 itojun 1980: key_freesp(sp);
1.173 jonathan 1981: #else
1982: KEY_FREESP(&sp);
1983: #endif
1.89 itojun 1984: }
1985: }
1986: #endif /*IPSEC*/
1.1 cgd 1987: ipstat.ips_cantfrag++;
1988: break;
1989:
1990: case ENOBUFS:
1.143 itojun 1991: #if 1
1992: /*
1993: * a router should not generate ICMP_SOURCEQUENCH as
1994: * required in RFC1812 Requirements for IP Version 4 Routers.
1995: * source quench could be a big problem under DoS attacks,
1.149 wiz 1996: * or if the underlying interface is rate-limited.
1.143 itojun 1997: */
1998: if (mcopy)
1999: m_freem(mcopy);
2000: return;
2001: #else
1.1 cgd 2002: type = ICMP_SOURCEQUENCH;
2003: code = 0;
2004: break;
1.143 itojun 2005: #endif
1.1 cgd 2006: }
1.218.2.1 yamt 2007: icmp_error(mcopy, type, code, dest, destmtu);
1.44 thorpej 2008: }
2009:
2010: void
1.211 perry 2011: ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
2012: struct mbuf *m)
1.44 thorpej 2013: {
2014:
2015: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
2016: struct timeval tv;
2017:
2018: microtime(&tv);
1.218.2.4 yamt 2019: *mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1.44 thorpej 2020: SCM_TIMESTAMP, SOL_SOCKET);
2021: if (*mp)
2022: mp = &(*mp)->m_next;
2023: }
2024: if (inp->inp_flags & INP_RECVDSTADDR) {
1.218.2.4 yamt 2025: *mp = sbcreatecontrol((void *) &ip->ip_dst,
1.44 thorpej 2026: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
2027: if (*mp)
2028: mp = &(*mp)->m_next;
2029: }
2030: #ifdef notyet
2031: /*
2032: * XXX
2033: * Moving these out of udp_input() made them even more broken
2034: * than they already were.
2035: * - fenner@parc.xerox.com
2036: */
2037: /* options were tossed already */
2038: if (inp->inp_flags & INP_RECVOPTS) {
1.218.2.4 yamt 2039: *mp = sbcreatecontrol((void *) opts_deleted_above,
1.44 thorpej 2040: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
2041: if (*mp)
2042: mp = &(*mp)->m_next;
2043: }
2044: /* ip_srcroute doesn't do what we want here, need to fix */
2045: if (inp->inp_flags & INP_RECVRETOPTS) {
1.218.2.4 yamt 2046: *mp = sbcreatecontrol((void *) ip_srcroute(),
1.44 thorpej 2047: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
2048: if (*mp)
2049: mp = &(*mp)->m_next;
2050: }
2051: #endif
2052: if (inp->inp_flags & INP_RECVIF) {
2053: struct sockaddr_dl sdl;
2054:
1.218.2.4 yamt 2055: sockaddr_dl_init(&sdl, sizeof(sdl),
2056: (m->m_pkthdr.rcvif != NULL)
2057: ? m->m_pkthdr.rcvif->if_index
2058: : 0,
2059: 0, NULL, 0, NULL, 0);
2060: *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP);
1.44 thorpej 2061: if (*mp)
2062: mp = &(*mp)->m_next;
2063: }
1.13 mycroft 2064: }
2065:
1.189 atatat 2066: /*
1.218.2.2 yamt 2067: * sysctl helper routine for net.inet.ip.forwsrcrt.
2068: */
2069: static int
2070: sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS)
2071: {
2072: int error, tmp;
2073: struct sysctlnode node;
2074:
2075: node = *rnode;
2076: tmp = ip_forwsrcrt;
2077: node.sysctl_data = &tmp;
2078: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2079: if (error || newp == NULL)
2080: return (error);
2081:
2082: if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT,
2083: 0, NULL, NULL, NULL))
2084: return (EPERM);
2085:
2086: ip_forwsrcrt = tmp;
2087:
2088: return (0);
2089: }
2090:
2091: /*
1.189 atatat 2092: * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the
2093: * range of the new value and tweaks timers if it changes.
2094: */
2095: static int
2096: sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS)
1.13 mycroft 2097: {
1.189 atatat 2098: int error, tmp;
2099: struct sysctlnode node;
2100:
2101: node = *rnode;
2102: tmp = ip_mtudisc_timeout;
2103: node.sysctl_data = &tmp;
2104: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2105: if (error || newp == NULL)
2106: return (error);
2107: if (tmp < 0)
2108: return (EINVAL);
1.52 thorpej 2109:
1.189 atatat 2110: ip_mtudisc_timeout = tmp;
2111: rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout);
2112:
2113: return (0);
2114: }
1.54 lukem 2115:
1.65 matt 2116: #ifdef GATEWAY
1.189 atatat 2117: /*
1.218.2.4 yamt 2118: * sysctl helper routine for net.inet.ip.maxflows.
1.189 atatat 2119: */
2120: static int
2121: sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS)
2122: {
2123: int s;
1.67 thorpej 2124:
1.217 atatat 2125: s = sysctl_lookup(SYSCTLFN_CALL(rnode));
1.218.2.4 yamt 2126: if (s || newp == NULL)
1.189 atatat 2127: return (s);
1.212 perry 2128:
1.189 atatat 2129: s = splsoftnet();
2130: ipflow_reap(0);
2131: splx(s);
1.144 martin 2132:
1.189 atatat 2133: return (0);
2134: }
1.218.2.4 yamt 2135:
2136: static int
2137: sysctl_net_inet_ip_hashsize(SYSCTLFN_ARGS)
2138: {
2139: int error, tmp;
2140: struct sysctlnode node;
2141:
2142: node = *rnode;
2143: tmp = ip_hashsize;
2144: node.sysctl_data = &tmp;
2145: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2146: if (error || newp == NULL)
2147: return (error);
2148:
2149: if ((tmp & (tmp - 1)) == 0 && tmp != 0) {
2150: /*
2151: * Can only fail due to malloc()
2152: */
2153: if (ipflow_invalidate_all(tmp))
2154: return ENOMEM;
2155: } else {
2156: /*
2157: * EINVAL if not a power of 2
2158: */
2159: return EINVAL;
2160: }
2161:
2162: return (0);
2163: }
1.189 atatat 2164: #endif /* GATEWAY */
1.117 tron 2165:
1.131 itojun 2166:
1.189 atatat 2167: SYSCTL_SETUP(sysctl_net_inet_ip_setup, "sysctl net.inet.ip subtree setup")
2168: {
2169: extern int subnetsarelocal, hostzeroisbroadcast;
1.180 jonathan 2170:
1.197 atatat 2171: sysctl_createv(clog, 0, NULL, NULL,
2172: CTLFLAG_PERMANENT,
1.189 atatat 2173: CTLTYPE_NODE, "net", NULL,
2174: NULL, 0, NULL, 0,
2175: CTL_NET, CTL_EOL);
1.197 atatat 2176: sysctl_createv(clog, 0, NULL, NULL,
2177: CTLFLAG_PERMANENT,
1.203 atatat 2178: CTLTYPE_NODE, "inet",
2179: SYSCTL_DESCR("PF_INET related settings"),
1.189 atatat 2180: NULL, 0, NULL, 0,
2181: CTL_NET, PF_INET, CTL_EOL);
1.197 atatat 2182: sysctl_createv(clog, 0, NULL, NULL,
2183: CTLFLAG_PERMANENT,
1.203 atatat 2184: CTLTYPE_NODE, "ip",
2185: SYSCTL_DESCR("IPv4 related settings"),
1.189 atatat 2186: NULL, 0, NULL, 0,
2187: CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL);
1.212 perry 2188:
1.197 atatat 2189: sysctl_createv(clog, 0, NULL, NULL,
2190: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2191: CTLTYPE_INT, "forwarding",
2192: SYSCTL_DESCR("Enable forwarding of INET datagrams"),
1.189 atatat 2193: NULL, 0, &ipforwarding, 0,
2194: CTL_NET, PF_INET, IPPROTO_IP,
2195: IPCTL_FORWARDING, CTL_EOL);
1.197 atatat 2196: sysctl_createv(clog, 0, NULL, NULL,
2197: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2198: CTLTYPE_INT, "redirect",
2199: SYSCTL_DESCR("Enable sending of ICMP redirect messages"),
1.189 atatat 2200: NULL, 0, &ipsendredirects, 0,
2201: CTL_NET, PF_INET, IPPROTO_IP,
2202: IPCTL_SENDREDIRECTS, CTL_EOL);
1.197 atatat 2203: sysctl_createv(clog, 0, NULL, NULL,
2204: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2205: CTLTYPE_INT, "ttl",
2206: SYSCTL_DESCR("Default TTL for an INET datagram"),
1.189 atatat 2207: NULL, 0, &ip_defttl, 0,
2208: CTL_NET, PF_INET, IPPROTO_IP,
2209: IPCTL_DEFTTL, CTL_EOL);
2210: #ifdef IPCTL_DEFMTU
1.197 atatat 2211: sysctl_createv(clog, 0, NULL, NULL,
2212: CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */,
1.203 atatat 2213: CTLTYPE_INT, "mtu",
2214: SYSCTL_DESCR("Default MTA for an INET route"),
1.189 atatat 2215: NULL, 0, &ip_mtu, 0,
2216: CTL_NET, PF_INET, IPPROTO_IP,
2217: IPCTL_DEFMTU, CTL_EOL);
2218: #endif /* IPCTL_DEFMTU */
1.197 atatat 2219: sysctl_createv(clog, 0, NULL, NULL,
1.218.2.2 yamt 2220: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2221: CTLTYPE_INT, "forwsrcrt",
2222: SYSCTL_DESCR("Enable forwarding of source-routed "
2223: "datagrams"),
1.218.2.2 yamt 2224: sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0,
1.189 atatat 2225: CTL_NET, PF_INET, IPPROTO_IP,
2226: IPCTL_FORWSRCRT, CTL_EOL);
1.197 atatat 2227: sysctl_createv(clog, 0, NULL, NULL,
2228: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2229: CTLTYPE_INT, "directed-broadcast",
2230: SYSCTL_DESCR("Enable forwarding of broadcast datagrams"),
1.189 atatat 2231: NULL, 0, &ip_directedbcast, 0,
2232: CTL_NET, PF_INET, IPPROTO_IP,
2233: IPCTL_DIRECTEDBCAST, CTL_EOL);
1.197 atatat 2234: sysctl_createv(clog, 0, NULL, NULL,
2235: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2236: CTLTYPE_INT, "allowsrcrt",
2237: SYSCTL_DESCR("Accept source-routed datagrams"),
1.189 atatat 2238: NULL, 0, &ip_allowsrcrt, 0,
2239: CTL_NET, PF_INET, IPPROTO_IP,
2240: IPCTL_ALLOWSRCRT, CTL_EOL);
1.197 atatat 2241: sysctl_createv(clog, 0, NULL, NULL,
2242: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2243: CTLTYPE_INT, "subnetsarelocal",
2244: SYSCTL_DESCR("Whether logical subnets are considered "
2245: "local"),
1.189 atatat 2246: NULL, 0, &subnetsarelocal, 0,
2247: CTL_NET, PF_INET, IPPROTO_IP,
2248: IPCTL_SUBNETSARELOCAL, CTL_EOL);
1.197 atatat 2249: sysctl_createv(clog, 0, NULL, NULL,
2250: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2251: CTLTYPE_INT, "mtudisc",
2252: SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"),
1.189 atatat 2253: NULL, 0, &ip_mtudisc, 0,
2254: CTL_NET, PF_INET, IPPROTO_IP,
2255: IPCTL_MTUDISC, CTL_EOL);
1.197 atatat 2256: sysctl_createv(clog, 0, NULL, NULL,
2257: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2258: CTLTYPE_INT, "anonportmin",
2259: SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1.189 atatat 2260: sysctl_net_inet_ip_ports, 0, &anonportmin, 0,
2261: CTL_NET, PF_INET, IPPROTO_IP,
2262: IPCTL_ANONPORTMIN, CTL_EOL);
1.197 atatat 2263: sysctl_createv(clog, 0, NULL, NULL,
2264: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2265: CTLTYPE_INT, "anonportmax",
2266: SYSCTL_DESCR("Highest ephemeral port number to assign"),
1.189 atatat 2267: sysctl_net_inet_ip_ports, 0, &anonportmax, 0,
2268: CTL_NET, PF_INET, IPPROTO_IP,
2269: IPCTL_ANONPORTMAX, CTL_EOL);
1.197 atatat 2270: sysctl_createv(clog, 0, NULL, NULL,
2271: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2272: CTLTYPE_INT, "mtudisctimeout",
2273: SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"),
1.189 atatat 2274: sysctl_net_inet_ip_pmtudto, 0, &ip_mtudisc_timeout, 0,
2275: CTL_NET, PF_INET, IPPROTO_IP,
2276: IPCTL_MTUDISCTIMEOUT, CTL_EOL);
2277: #ifdef GATEWAY
1.197 atatat 2278: sysctl_createv(clog, 0, NULL, NULL,
2279: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2280: CTLTYPE_INT, "maxflows",
2281: SYSCTL_DESCR("Number of flows for fast forwarding"),
1.189 atatat 2282: sysctl_net_inet_ip_maxflows, 0, &ip_maxflows, 0,
2283: CTL_NET, PF_INET, IPPROTO_IP,
2284: IPCTL_MAXFLOWS, CTL_EOL);
1.218.2.4 yamt 2285: sysctl_createv(clog, 0, NULL, NULL,
2286: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2287: CTLTYPE_INT, "hashsize",
2288: SYSCTL_DESCR("Size of hash table for fast forwarding (IPv4)"),
2289: sysctl_net_inet_ip_hashsize, 0, &ip_hashsize, 0,
2290: CTL_NET, PF_INET, IPPROTO_IP,
2291: CTL_CREATE, CTL_EOL);
1.189 atatat 2292: #endif /* GATEWAY */
1.197 atatat 2293: sysctl_createv(clog, 0, NULL, NULL,
2294: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2295: CTLTYPE_INT, "hostzerobroadcast",
2296: SYSCTL_DESCR("All zeroes address is broadcast address"),
1.189 atatat 2297: NULL, 0, &hostzeroisbroadcast, 0,
2298: CTL_NET, PF_INET, IPPROTO_IP,
2299: IPCTL_HOSTZEROBROADCAST, CTL_EOL);
2300: #if NGIF > 0
1.197 atatat 2301: sysctl_createv(clog, 0, NULL, NULL,
2302: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2303: CTLTYPE_INT, "gifttl",
2304: SYSCTL_DESCR("Default TTL for a gif tunnel datagram"),
1.189 atatat 2305: NULL, 0, &ip_gif_ttl, 0,
2306: CTL_NET, PF_INET, IPPROTO_IP,
2307: IPCTL_GIF_TTL, CTL_EOL);
2308: #endif /* NGIF */
2309: #ifndef IPNOPRIVPORTS
1.197 atatat 2310: sysctl_createv(clog, 0, NULL, NULL,
2311: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2312: CTLTYPE_INT, "lowportmin",
2313: SYSCTL_DESCR("Lowest privileged ephemeral port number "
2314: "to assign"),
1.189 atatat 2315: sysctl_net_inet_ip_ports, 0, &lowportmin, 0,
2316: CTL_NET, PF_INET, IPPROTO_IP,
2317: IPCTL_LOWPORTMIN, CTL_EOL);
1.197 atatat 2318: sysctl_createv(clog, 0, NULL, NULL,
2319: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2320: CTLTYPE_INT, "lowportmax",
2321: SYSCTL_DESCR("Highest privileged ephemeral port number "
2322: "to assign"),
1.189 atatat 2323: sysctl_net_inet_ip_ports, 0, &lowportmax, 0,
2324: CTL_NET, PF_INET, IPPROTO_IP,
2325: IPCTL_LOWPORTMAX, CTL_EOL);
2326: #endif /* IPNOPRIVPORTS */
1.197 atatat 2327: sysctl_createv(clog, 0, NULL, NULL,
2328: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2329: CTLTYPE_INT, "maxfragpackets",
2330: SYSCTL_DESCR("Maximum number of fragments to retain for "
2331: "possible reassembly"),
1.189 atatat 2332: NULL, 0, &ip_maxfragpackets, 0,
2333: CTL_NET, PF_INET, IPPROTO_IP,
2334: IPCTL_MAXFRAGPACKETS, CTL_EOL);
2335: #if NGRE > 0
1.197 atatat 2336: sysctl_createv(clog, 0, NULL, NULL,
2337: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2338: CTLTYPE_INT, "grettl",
2339: SYSCTL_DESCR("Default TTL for a gre tunnel datagram"),
1.189 atatat 2340: NULL, 0, &ip_gre_ttl, 0,
2341: CTL_NET, PF_INET, IPPROTO_IP,
2342: IPCTL_GRE_TTL, CTL_EOL);
2343: #endif /* NGRE */
1.197 atatat 2344: sysctl_createv(clog, 0, NULL, NULL,
2345: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2346: CTLTYPE_INT, "checkinterface",
2347: SYSCTL_DESCR("Enable receive side of Strong ES model "
2348: "from RFC1122"),
1.189 atatat 2349: NULL, 0, &ip_checkinterface, 0,
2350: CTL_NET, PF_INET, IPPROTO_IP,
2351: IPCTL_CHECKINTERFACE, CTL_EOL);
1.197 atatat 2352: sysctl_createv(clog, 0, NULL, NULL,
2353: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2354: CTLTYPE_INT, "random_id",
2355: SYSCTL_DESCR("Assign random ip_id values"),
1.189 atatat 2356: NULL, 0, &ip_do_randomid, 0,
2357: CTL_NET, PF_INET, IPPROTO_IP,
2358: IPCTL_RANDOMID, CTL_EOL);
1.206 thorpej 2359: sysctl_createv(clog, 0, NULL, NULL,
2360: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2361: CTLTYPE_INT, "do_loopback_cksum",
2362: SYSCTL_DESCR("Perform IP checksum on loopback"),
2363: NULL, 0, &ip_do_loopback_cksum, 0,
2364: CTL_NET, PF_INET, IPPROTO_IP,
2365: IPCTL_LOOPBACKCKSUM, CTL_EOL);
1.218.2.1 yamt 2366: sysctl_createv(clog, 0, NULL, NULL,
2367: CTLFLAG_PERMANENT,
2368: CTLTYPE_STRUCT, "stats",
2369: SYSCTL_DESCR("IP statistics"),
2370: NULL, 0, &ipstat, sizeof(ipstat),
2371: CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS,
2372: CTL_EOL);
1.1 cgd 2373: }
CVSweb <webmaster@jp.NetBSD.org>