Annotation of src/sys/netinet/ip_input.c, Revision 1.211
1.211 ! perry 1: /* $NetBSD: ip_input.c,v 1.210 2005/02/02 21:41:55 perry 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>
1.211 ! perry 101: __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.210 2005/02/02 21:41:55 perry 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>
1.1 cgd 124:
1.5 mycroft 125: #include <net/if.h>
1.44 thorpej 126: #include <net/if_dl.h>
1.5 mycroft 127: #include <net/route.h>
1.45 mrg 128: #include <net/pfil.h>
1.1 cgd 129:
1.5 mycroft 130: #include <netinet/in.h>
131: #include <netinet/in_systm.h>
132: #include <netinet/ip.h>
133: #include <netinet/in_pcb.h>
134: #include <netinet/in_var.h>
135: #include <netinet/ip_var.h>
136: #include <netinet/ip_icmp.h>
1.89 itojun 137: /* just for gif_ttl */
138: #include <netinet/in_gif.h>
139: #include "gif.h"
1.144 martin 140: #include <net/if_gre.h>
141: #include "gre.h"
1.111 jdolecek 142:
143: #ifdef MROUTING
144: #include <netinet/ip_mroute.h>
145: #endif
1.89 itojun 146:
147: #ifdef IPSEC
148: #include <netinet6/ipsec.h>
149: #include <netkey/key.h>
150: #endif
1.173 jonathan 151: #ifdef FAST_IPSEC
152: #include <netipsec/ipsec.h>
153: #include <netipsec/key.h>
154: #endif /* FAST_IPSEC*/
1.44 thorpej 155:
1.1 cgd 156: #ifndef IPFORWARDING
157: #ifdef GATEWAY
158: #define IPFORWARDING 1 /* forward IP packets not for us */
159: #else /* GATEWAY */
160: #define IPFORWARDING 0 /* don't forward IP packets not for us */
161: #endif /* GATEWAY */
162: #endif /* IPFORWARDING */
163: #ifndef IPSENDREDIRECTS
164: #define IPSENDREDIRECTS 1
165: #endif
1.26 thorpej 166: #ifndef IPFORWSRCRT
1.47 cjs 167: #define IPFORWSRCRT 1 /* forward source-routed packets */
168: #endif
169: #ifndef IPALLOWSRCRT
1.48 mrg 170: #define IPALLOWSRCRT 1 /* allow source-routed packets */
1.26 thorpej 171: #endif
1.53 kml 172: #ifndef IPMTUDISC
1.153 itojun 173: #define IPMTUDISC 1
1.53 kml 174: #endif
1.60 kml 175: #ifndef IPMTUDISCTIMEOUT
1.61 kml 176: #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */
1.60 kml 177: #endif
1.53 kml 178:
1.27 thorpej 179: /*
180: * Note: DIRECTED_BROADCAST is handled this way so that previous
181: * configuration using this option will Just Work.
182: */
183: #ifndef IPDIRECTEDBCAST
184: #ifdef DIRECTED_BROADCAST
185: #define IPDIRECTEDBCAST 1
186: #else
187: #define IPDIRECTEDBCAST 0
188: #endif /* DIRECTED_BROADCAST */
189: #endif /* IPDIRECTEDBCAST */
1.1 cgd 190: int ipforwarding = IPFORWARDING;
191: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 192: int ip_defttl = IPDEFTTL;
1.26 thorpej 193: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 194: int ip_directedbcast = IPDIRECTEDBCAST;
1.47 cjs 195: int ip_allowsrcrt = IPALLOWSRCRT;
1.53 kml 196: int ip_mtudisc = IPMTUDISC;
1.156 itojun 197: int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
1.1 cgd 198: #ifdef DIAGNOSTIC
199: int ipprintfs = 0;
200: #endif
1.184 jonathan 201:
202: int ip_do_randomid = 0;
1.208 christos 203: int ip_do_loopback_cksum = 0;
1.184 jonathan 204:
1.165 christos 205: /*
206: * XXX - Setting ip_checkinterface mostly implements the receive side of
207: * the Strong ES model described in RFC 1122, but since the routing table
208: * and transmit implementation do not implement the Strong ES model,
209: * setting this to 1 results in an odd hybrid.
210: *
211: * XXX - ip_checkinterface currently must be disabled if you use ipnat
212: * to translate the destination address to another local interface.
213: *
214: * XXX - ip_checkinterface must be disabled if you add IP aliases
215: * to the loopback interface instead of the interface where the
216: * packets for those addresses are received.
217: */
218: int ip_checkinterface = 0;
219:
1.1 cgd 220:
1.60 kml 221: struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
222:
1.1 cgd 223: int ipqmaxlen = IFQ_MAXLEN;
1.150 matt 224: u_long in_ifaddrhash; /* size of hash table - 1 */
225: int in_ifaddrentries; /* total number of addrs */
1.181 jonathan 226: struct in_ifaddrhead in_ifaddrhead;
1.57 tls 227: struct in_ifaddrhashhead *in_ifaddrhashtbl;
1.166 matt 228: u_long in_multihash; /* size of hash table - 1 */
229: int in_multientries; /* total number of addrs */
230: struct in_multihashhead *in_multihashtbl;
1.13 mycroft 231: struct ifqueue ipintrq;
1.63 matt 232: struct ipstat ipstat;
1.183 jonathan 233: uint16_t ip_id;
1.75 thorpej 234:
1.121 thorpej 235: #ifdef PFIL_HOOKS
236: struct pfil_head inet_pfil_hook;
237: #endif
238:
1.194 jonathan 239: /*
240: * Cached copy of nmbclusters. If nbclusters is different,
241: * recalculate IP parameters derived from nmbclusters.
242: */
243: static int ip_nmbclusters; /* copy of nmbclusters */
1.210 perry 244: static void ip_nmbclusters_changed(void); /* recalc limits */
1.194 jonathan 245:
1.195 thorpej 246: #define CHECK_NMBCLUSTER_PARAMS() \
247: do { \
248: if (__predict_false(ip_nmbclusters != nmbclusters)) \
249: ip_nmbclusters_changed(); \
250: } while (/*CONSTCOND*/0)
1.194 jonathan 251:
1.190 jonathan 252: /* IP datagram reassembly queues (hashed) */
253: #define IPREASS_NHASH_LOG2 6
254: #define IPREASS_NHASH (1 << IPREASS_NHASH_LOG2)
255: #define IPREASS_HMASK (IPREASS_NHASH - 1)
256: #define IPREASS_HASH(x,y) \
257: (((((x) & 0xF) | ((((x) >> 8) & 0xF) << 4)) ^ (y)) & IPREASS_HMASK)
258: struct ipqhead ipq[IPREASS_NHASH];
1.75 thorpej 259: int ipq_locked;
1.194 jonathan 260: static int ip_nfragpackets; /* packets in reass queue */
261: static int ip_nfrags; /* total fragments in reass queues */
262:
263: int ip_maxfragpackets = 200; /* limit on packets. XXX sysctl */
264: int ip_maxfrags; /* limit on fragments. XXX sysctl */
265:
266:
267: /*
268: * Additive-Increase/Multiplicative-Decrease (AIMD) strategy for
269: * IP reassembly queue buffer managment.
270: *
271: * We keep a count of total IP fragments (NB: not fragmented packets!)
272: * awaiting reassembly (ip_nfrags) and a limit (ip_maxfrags) on fragments.
273: * If ip_nfrags exceeds ip_maxfrags the limit, we drop half the
274: * total fragments in reassembly queues.This AIMD policy avoids
275: * repeatedly deleting single packets under heavy fragmentation load
276: * (e.g., from lossy NFS peers).
277: */
1.210 perry 278: static u_int ip_reass_ttl_decr(u_int ticks);
279: static void ip_reass_drophalf(void);
1.194 jonathan 280:
1.75 thorpej 281:
1.210 perry 282: static __inline int ipq_lock_try(void);
283: static __inline void ipq_unlock(void);
1.75 thorpej 284:
285: static __inline int
1.211 ! perry 286: ipq_lock_try(void)
1.75 thorpej 287: {
288: int s;
289:
1.132 thorpej 290: /*
1.149 wiz 291: * Use splvm() -- we're blocking things that would cause
1.132 thorpej 292: * mbuf allocation.
293: */
294: s = splvm();
1.75 thorpej 295: if (ipq_locked) {
296: splx(s);
297: return (0);
298: }
299: ipq_locked = 1;
300: splx(s);
301: return (1);
302: }
303:
304: static __inline void
1.211 ! perry 305: ipq_unlock(void)
1.75 thorpej 306: {
307: int s;
308:
1.132 thorpej 309: s = splvm();
1.75 thorpej 310: ipq_locked = 0;
311: splx(s);
312: }
313:
314: #ifdef DIAGNOSTIC
315: #define IPQ_LOCK() \
316: do { \
317: if (ipq_lock_try() == 0) { \
318: printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \
319: panic("ipq_lock"); \
320: } \
1.159 perry 321: } while (/*CONSTCOND*/ 0)
1.75 thorpej 322: #define IPQ_LOCK_CHECK() \
323: do { \
324: if (ipq_locked == 0) { \
325: printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \
326: panic("ipq lock check"); \
327: } \
1.159 perry 328: } while (/*CONSTCOND*/ 0)
1.75 thorpej 329: #else
330: #define IPQ_LOCK() (void) ipq_lock_try()
331: #define IPQ_LOCK_CHECK() /* nothing */
332: #endif
333:
334: #define IPQ_UNLOCK() ipq_unlock()
1.1 cgd 335:
1.200 simonb 336: POOL_INIT(inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl", NULL);
337: POOL_INIT(ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", NULL);
1.72 thorpej 338:
1.135 thorpej 339: #ifdef INET_CSUM_COUNTERS
340: #include <sys/device.h>
341:
342: struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
343: NULL, "inet", "hwcsum bad");
344: struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
345: NULL, "inet", "hwcsum ok");
346: struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
347: NULL, "inet", "swcsum");
348:
349: #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++
350:
1.201 matt 351: EVCNT_ATTACH_STATIC(ip_hwcsum_bad);
352: EVCNT_ATTACH_STATIC(ip_hwcsum_ok);
353: EVCNT_ATTACH_STATIC(ip_swcsum);
354:
1.135 thorpej 355: #else
356:
357: #define INET_CSUM_COUNTER_INCR(ev) /* nothing */
358:
359: #endif /* INET_CSUM_COUNTERS */
360:
1.1 cgd 361: /*
362: * We need to save the IP options in case a protocol wants to respond
363: * to an incoming packet over the same route if the packet got here
364: * using IP source routing. This allows connection establishment and
365: * maintenance when the remote end is on a network that is not known
366: * to us.
367: */
368: int ip_nhops = 0;
369: static struct ip_srcrt {
370: struct in_addr dst; /* final destination */
371: char nop; /* one NOP to align */
372: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
373: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
374: } ip_srcrt;
375:
1.210 perry 376: static void save_rte(u_char *, struct in_addr);
1.35 mycroft 377:
1.164 matt 378: #ifdef MBUFTRACE
379: struct mowner ip_rx_mowner = { "internet", "rx" };
380: struct mowner ip_tx_mowner = { "internet", "tx" };
381: #endif
382:
1.1 cgd 383: /*
1.194 jonathan 384: * Compute IP limits derived from the value of nmbclusters.
385: */
386: static void
387: ip_nmbclusters_changed(void)
388: {
389: ip_maxfrags = nmbclusters / 4;
390: ip_nmbclusters = nmbclusters;
391: }
392:
393: /*
1.1 cgd 394: * IP initialization: fill in IP protocol switch table.
395: * All protocols not implemented in kernel go to raw IP protocol handler.
396: */
1.8 mycroft 397: void
1.211 ! perry 398: ip_init(void)
1.1 cgd 399: {
1.199 matt 400: const struct protosw *pr;
1.109 augustss 401: int i;
1.1 cgd 402:
403: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
404: if (pr == 0)
405: panic("ip_init");
406: for (i = 0; i < IPPROTO_MAX; i++)
407: ip_protox[i] = pr - inetsw;
408: for (pr = inetdomain.dom_protosw;
409: pr < inetdomain.dom_protoswNPROTOSW; pr++)
410: if (pr->pr_domain->dom_family == PF_INET &&
411: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
412: ip_protox[pr->pr_protocol] = pr - inetsw;
1.192 jonathan 413:
1.190 jonathan 414: for (i = 0; i < IPREASS_NHASH; i++)
415: LIST_INIT(&ipq[i]);
416:
1.183 jonathan 417: ip_id = time.tv_sec & 0xfffff;
1.194 jonathan 418:
1.1 cgd 419: ipintrq.ifq_maxlen = ipqmaxlen;
1.194 jonathan 420: ip_nmbclusters_changed();
421:
1.181 jonathan 422: TAILQ_INIT(&in_ifaddrhead);
1.120 ad 423: in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IFADDR,
424: M_WAITOK, &in_ifaddrhash);
1.166 matt 425: in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IPMADDR,
426: M_WAITOK, &in_multihash);
1.160 itojun 427: ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout);
1.73 thorpej 428: #ifdef GATEWAY
429: ipflow_init();
430: #endif
1.121 thorpej 431:
432: #ifdef PFIL_HOOKS
433: /* Register our Packet Filter hook. */
1.126 thorpej 434: inet_pfil_hook.ph_type = PFIL_TYPE_AF;
435: inet_pfil_hook.ph_af = AF_INET;
1.121 thorpej 436: i = pfil_head_register(&inet_pfil_hook);
437: if (i != 0)
438: printf("ip_init: WARNING: unable to register pfil hook, "
439: "error %d\n", i);
440: #endif /* PFIL_HOOKS */
1.135 thorpej 441:
1.164 matt 442: #ifdef MBUFTRACE
443: MOWNER_ATTACH(&ip_tx_mowner);
444: MOWNER_ATTACH(&ip_rx_mowner);
445: #endif /* MBUFTRACE */
1.1 cgd 446: }
447:
448: struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
449: struct route ipforward_rt;
450:
451: /*
1.89 itojun 452: * IP software interrupt routine
453: */
454: void
1.211 ! perry 455: ipintr(void)
1.89 itojun 456: {
457: int s;
458: struct mbuf *m;
459:
460: while (1) {
1.132 thorpej 461: s = splnet();
1.89 itojun 462: IF_DEQUEUE(&ipintrq, m);
463: splx(s);
464: if (m == 0)
465: return;
1.164 matt 466: MCLAIM(m, &ip_rx_mowner);
1.89 itojun 467: ip_input(m);
468: }
469: }
470:
471: /*
1.1 cgd 472: * Ip input routine. Checksum and byte swap header. If fragmented
473: * try to reassemble. Process options. Pass to next level.
474: */
1.8 mycroft 475: void
1.89 itojun 476: ip_input(struct mbuf *m)
1.1 cgd 477: {
1.109 augustss 478: struct ip *ip = NULL;
479: struct ipq *fp;
480: struct in_ifaddr *ia;
481: struct ifaddr *ifa;
1.25 cgd 482: struct ipqent *ipqe;
1.89 itojun 483: int hlen = 0, mff, len;
1.100 itojun 484: int downmatch;
1.165 christos 485: int checkif;
1.169 itojun 486: int srcrt = 0;
1.190 jonathan 487: u_int hash;
1.173 jonathan 488: #ifdef FAST_IPSEC
489: struct m_tag *mtag;
490: struct tdb_ident *tdbi;
491: struct secpolicy *sp;
492: int s, error;
493: #endif /* FAST_IPSEC */
1.1 cgd 494:
1.164 matt 495: MCLAIM(m, &ip_rx_mowner);
1.1 cgd 496: #ifdef DIAGNOSTIC
497: if ((m->m_flags & M_PKTHDR) == 0)
498: panic("ipintr no HDR");
1.89 itojun 499: #endif
1.164 matt 500:
1.1 cgd 501: /*
502: * If no IP addresses have been set yet but the interfaces
503: * are receiving, can't do anything with incoming packets yet.
504: */
1.181 jonathan 505: if (TAILQ_FIRST(&in_ifaddrhead) == 0)
1.1 cgd 506: goto bad;
507: ipstat.ips_total++;
1.154 thorpej 508: /*
509: * If the IP header is not aligned, slurp it up into a new
510: * mbuf with space for link headers, in the event we forward
511: * it. Otherwise, if it is aligned, make sure the entire
512: * base IP header is in the first mbuf of the chain.
513: */
514: if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
515: if ((m = m_copyup(m, sizeof(struct ip),
516: (max_linkhdr + 3) & ~3)) == NULL) {
517: /* XXXJRT new stat, please */
518: ipstat.ips_toosmall++;
519: return;
520: }
521: } else if (__predict_false(m->m_len < sizeof (struct ip))) {
522: if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
523: ipstat.ips_toosmall++;
524: return;
525: }
1.1 cgd 526: }
527: ip = mtod(m, struct ip *);
1.13 mycroft 528: if (ip->ip_v != IPVERSION) {
529: ipstat.ips_badvers++;
530: goto bad;
531: }
1.1 cgd 532: hlen = ip->ip_hl << 2;
533: if (hlen < sizeof(struct ip)) { /* minimum header length */
534: ipstat.ips_badhlen++;
535: goto bad;
536: }
537: if (hlen > m->m_len) {
538: if ((m = m_pullup(m, hlen)) == 0) {
539: ipstat.ips_badhlen++;
1.89 itojun 540: return;
1.1 cgd 541: }
542: ip = mtod(m, struct ip *);
543: }
1.98 thorpej 544:
1.85 hwr 545: /*
1.99 thorpej 546: * RFC1122: packets with a multicast source address are
1.98 thorpej 547: * not allowed.
1.85 hwr 548: */
549: if (IN_MULTICAST(ip->ip_src.s_addr)) {
1.130 itojun 550: ipstat.ips_badaddr++;
1.85 hwr 551: goto bad;
1.129 itojun 552: }
553:
554: /* 127/8 must not appear on wire - RFC1122 */
555: if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
556: (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
1.130 itojun 557: if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
558: ipstat.ips_badaddr++;
1.129 itojun 559: goto bad;
1.130 itojun 560: }
1.85 hwr 561: }
562:
1.135 thorpej 563: switch (m->m_pkthdr.csum_flags &
1.137 thorpej 564: ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) |
1.135 thorpej 565: M_CSUM_IPv4_BAD)) {
566: case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
567: INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
568: goto badcsum;
569:
570: case M_CSUM_IPv4:
571: /* Checksum was okay. */
572: INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
573: break;
574:
575: default:
1.206 thorpej 576: /*
577: * Must compute it ourselves. Maybe skip checksum on
578: * loopback interfaces.
579: */
580: if (__predict_true(!(m->m_pkthdr.rcvif->if_flags &
581: IFF_LOOPBACK) || ip_do_loopback_cksum)) {
582: INET_CSUM_COUNTER_INCR(&ip_swcsum);
583: if (in_cksum(m, hlen) != 0)
584: goto badcsum;
585: }
1.135 thorpej 586: break;
1.1 cgd 587: }
588:
1.121 thorpej 589: /* Retrieve the packet length. */
590: len = ntohs(ip->ip_len);
1.81 proff 591:
592: /*
593: * Check for additional length bogosity
594: */
1.84 proff 595: if (len < hlen) {
1.81 proff 596: ipstat.ips_badlen++;
597: goto bad;
598: }
1.1 cgd 599:
600: /*
601: * Check that the amount of data in the buffers
602: * is as at least much as the IP header would have us expect.
603: * Trim mbufs if longer than we expect.
604: * Drop packet if shorter than we expect.
605: */
1.35 mycroft 606: if (m->m_pkthdr.len < len) {
1.1 cgd 607: ipstat.ips_tooshort++;
608: goto bad;
609: }
1.35 mycroft 610: if (m->m_pkthdr.len > len) {
1.1 cgd 611: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 612: m->m_len = len;
613: m->m_pkthdr.len = len;
1.1 cgd 614: } else
1.35 mycroft 615: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 616: }
617:
1.193 scw 618: #if defined(IPSEC)
1.149 wiz 619: /* ipflow (IP fast forwarding) is not compatible with IPsec. */
1.94 itojun 620: m->m_flags &= ~M_CANFASTFWD;
621: #else
1.64 thorpej 622: /*
623: * Assume that we can create a fast-forward IP flow entry
624: * based on this packet.
625: */
626: m->m_flags |= M_CANFASTFWD;
1.94 itojun 627: #endif
1.64 thorpej 628:
1.36 mrg 629: #ifdef PFIL_HOOKS
1.33 mrg 630: /*
1.64 thorpej 631: * Run through list of hooks for input packets. If there are any
632: * filters which require that additional packets in the flow are
633: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
634: * Note that filters must _never_ set this flag, as another filter
635: * in the list may have previously cleared it.
1.33 mrg 636: */
1.127 itojun 637: /*
638: * let ipfilter look at packet on the wire,
639: * not the decapsulated packet.
640: */
641: #ifdef IPSEC
1.136 itojun 642: if (!ipsec_getnhist(m))
1.186 scw 643: #elif defined(FAST_IPSEC)
644: if (!ipsec_indone(m))
1.127 itojun 645: #else
646: if (1)
647: #endif
648: {
1.169 itojun 649: struct in_addr odst;
650:
651: odst = ip->ip_dst;
1.127 itojun 652: if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif,
1.168 itojun 653: PFIL_IN) != 0)
654: return;
1.127 itojun 655: if (m == NULL)
656: return;
657: ip = mtod(m, struct ip *);
1.142 darrenr 658: hlen = ip->ip_hl << 2;
1.205 darrenr 659: /*
660: * XXX The setting of "srcrt" here is to prevent ip_forward()
661: * from generating ICMP redirects for packets that have
662: * been redirected by a hook back out on to the same LAN that
663: * they came from and is not an indication that the packet
664: * is being inffluenced by source routing options. This
665: * allows things like
666: * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp"
667: * where tlp0 is both on the 1.1.1.0/24 network and is the
668: * default route for hosts on 1.1.1.0/24. Of course this
669: * also requires a "map tlp0 ..." to complete the story.
670: * One might argue whether or not this kind of network config.
671: * should be supported in this manner...
672: */
1.169 itojun 673: srcrt = (odst.s_addr != ip->ip_dst.s_addr);
1.127 itojun 674: }
1.36 mrg 675: #endif /* PFIL_HOOKS */
1.123 thorpej 676:
677: #ifdef ALTQ
678: /* XXX Temporary until ALTQ is changed to use a pfil hook */
679: if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) {
680: /* packet dropped by traffic conditioner */
681: return;
682: }
683: #endif
1.121 thorpej 684:
685: /*
1.1 cgd 686: * Process options and, if not destined for us,
687: * ship it on. ip_dooptions returns 1 when an
688: * error was detected (causing an icmp message
689: * to be sent and the original packet to be freed).
690: */
691: ip_nhops = 0; /* for source routed packets */
692: if (hlen > sizeof (struct ip) && ip_dooptions(m))
1.89 itojun 693: return;
1.1 cgd 694:
695: /*
1.165 christos 696: * Enable a consistency check between the destination address
697: * and the arrival interface for a unicast packet (the RFC 1122
698: * strong ES model) if IP forwarding is disabled and the packet
699: * is not locally generated.
700: *
701: * XXX - Checking also should be disabled if the destination
702: * address is ipnat'ed to a different interface.
703: *
704: * XXX - Checking is incompatible with IP aliases added
705: * to the loopback interface instead of the interface where
706: * the packets are received.
707: *
708: * XXX - We need to add a per ifaddr flag for this so that
709: * we get finer grain control.
710: */
711: checkif = ip_checkinterface && (ipforwarding == 0) &&
712: (m->m_pkthdr.rcvif != NULL) &&
713: ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0);
714:
715: /*
1.1 cgd 716: * Check our list of addresses, to see if the packet is for us.
1.100 itojun 717: *
718: * Traditional 4.4BSD did not consult IFF_UP at all.
719: * The behavior here is to treat addresses on !IFF_UP interface
720: * as not mine.
1.1 cgd 721: */
1.100 itojun 722: downmatch = 0;
1.140 matt 723: LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) {
1.97 itojun 724: if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
1.165 christos 725: if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif)
726: continue;
1.97 itojun 727: if ((ia->ia_ifp->if_flags & IFF_UP) != 0)
728: break;
1.100 itojun 729: else
730: downmatch++;
1.97 itojun 731: }
732: }
1.86 thorpej 733: if (ia != NULL)
734: goto ours;
1.57 tls 735: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
1.209 matt 736: IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) {
1.140 matt 737: if (ifa->ifa_addr->sa_family != AF_INET)
738: continue;
1.57 tls 739: ia = ifatoia(ifa);
1.35 mycroft 740: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
741: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 742: /*
743: * Look for all-0's host part (old broadcast addr),
744: * either for subnet or net.
745: */
746: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 747: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 748: goto ours;
1.57 tls 749: /*
750: * An interface with IP address zero accepts
751: * all packets that arrive on that interface.
752: */
753: if (in_nullhost(ia->ia_addr.sin_addr))
754: goto ours;
1.1 cgd 755: }
756: }
1.18 mycroft 757: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 758: struct in_multi *inm;
759: #ifdef MROUTING
760: extern struct socket *ip_mrouter;
1.10 brezak 761:
1.4 hpeyerl 762: if (ip_mrouter) {
763: /*
764: * If we are acting as a multicast router, all
765: * incoming multicast packets are passed to the
766: * kernel-level multicast forwarding function.
767: * The packet is returned (relatively) intact; if
768: * ip_mforward() returns a non-zero value, the packet
769: * must be discarded, else it may be accepted below.
770: *
771: * (The IP ident field is put in the same byte order
772: * as expected when ip_mforward() is called from
773: * ip_output().)
774: */
1.13 mycroft 775: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
776: ipstat.ips_cantforward++;
1.4 hpeyerl 777: m_freem(m);
1.89 itojun 778: return;
1.4 hpeyerl 779: }
780:
781: /*
782: * The process-level routing demon needs to receive
783: * all multicast IGMP packets, whether or not this
784: * host belongs to their destination groups.
785: */
786: if (ip->ip_p == IPPROTO_IGMP)
787: goto ours;
1.13 mycroft 788: ipstat.ips_forward++;
1.4 hpeyerl 789: }
790: #endif
791: /*
792: * See if we belong to the destination multicast group on the
793: * arrival interface.
794: */
795: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
796: if (inm == NULL) {
1.13 mycroft 797: ipstat.ips_cantforward++;
1.4 hpeyerl 798: m_freem(m);
1.89 itojun 799: return;
1.4 hpeyerl 800: }
801: goto ours;
802: }
1.19 mycroft 803: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 804: in_nullhost(ip->ip_dst))
1.1 cgd 805: goto ours;
806:
807: /*
808: * Not for us; forward if possible and desirable.
809: */
810: if (ipforwarding == 0) {
811: ipstat.ips_cantforward++;
812: m_freem(m);
1.100 itojun 813: } else {
814: /*
815: * If ip_dst matched any of my address on !IFF_UP interface,
816: * and there's no IFF_UP interface that matches ip_dst,
817: * send icmp unreach. Forwarding it will result in in-kernel
818: * forwarding loop till TTL goes to 0.
819: */
820: if (downmatch) {
821: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
822: ipstat.ips_cantforward++;
823: return;
824: }
1.145 itojun 825: #ifdef IPSEC
826: if (ipsec4_in_reject(m, NULL)) {
827: ipsecstat.in_polvio++;
828: goto bad;
829: }
830: #endif
1.173 jonathan 831: #ifdef FAST_IPSEC
832: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
833: s = splsoftnet();
834: if (mtag != NULL) {
835: tdbi = (struct tdb_ident *)(mtag + 1);
836: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
837: } else {
838: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
839: IP_FORWARDING, &error);
840: }
841: if (sp == NULL) { /* NB: can happen if error */
842: splx(s);
843: /*XXX error stat???*/
844: DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/
845: goto bad;
846: }
847:
848: /*
849: * Check security policy against packet attributes.
850: */
851: error = ipsec_in_reject(sp, m);
852: KEY_FREESP(&sp);
853: splx(s);
854: if (error) {
855: ipstat.ips_cantforward++;
856: goto bad;
1.193 scw 857: }
858:
859: /*
860: * Peek at the outbound SP for this packet to determine if
861: * it's a Fast Forward candidate.
862: */
863: mtag = m_tag_find(m, PACKET_TAG_IPSEC_PENDING_TDB, NULL);
864: if (mtag != NULL)
865: m->m_flags &= ~M_CANFASTFWD;
866: else {
867: s = splsoftnet();
868: sp = ipsec4_checkpolicy(m, IPSEC_DIR_OUTBOUND,
869: (IP_FORWARDING |
870: (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
871: &error, NULL);
872: if (sp != NULL) {
873: m->m_flags &= ~M_CANFASTFWD;
874: KEY_FREESP(&sp);
875: }
876: splx(s);
1.173 jonathan 877: }
878: #endif /* FAST_IPSEC */
1.145 itojun 879:
1.169 itojun 880: ip_forward(m, srcrt);
1.100 itojun 881: }
1.89 itojun 882: return;
1.1 cgd 883:
884: ours:
885: /*
886: * If offset or IP_MF are set, must reassemble.
887: * Otherwise, nothing need be done.
888: * (We could look in the reassembly queue to see
889: * if the packet was previously fragmented,
890: * but it's not worth the time; just let them time out.)
891: */
1.155 itojun 892: if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
893:
1.1 cgd 894: /*
895: * Look for queue of fragments
896: * of this datagram.
897: */
1.75 thorpej 898: IPQ_LOCK();
1.190 jonathan 899: hash = IPREASS_HASH(ip->ip_src.s_addr, ip->ip_id);
900: /* XXX LIST_FOREACH(fp, &ipq[hash], ipq_q) */
901: for (fp = LIST_FIRST(&ipq[hash]); fp != NULL;
902: fp = LIST_NEXT(fp, ipq_q)) {
1.1 cgd 903: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 904: in_hosteq(ip->ip_src, fp->ipq_src) &&
905: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 906: ip->ip_p == fp->ipq_p)
907: goto found;
1.190 jonathan 908:
909: }
1.1 cgd 910: fp = 0;
911: found:
912:
913: /*
914: * Adjust ip_len to not reflect header,
1.25 cgd 915: * set ipqe_mff if more fragments are expected,
1.1 cgd 916: * convert offset of this to bytes.
917: */
1.155 itojun 918: ip->ip_len = htons(ntohs(ip->ip_len) - hlen);
919: mff = (ip->ip_off & htons(IP_MF)) != 0;
1.25 cgd 920: if (mff) {
1.16 cgd 921: /*
922: * Make sure that fragments have a data length
923: * that's a non-zero multiple of 8 bytes.
924: */
1.155 itojun 925: if (ntohs(ip->ip_len) == 0 ||
926: (ntohs(ip->ip_len) & 0x7) != 0) {
1.16 cgd 927: ipstat.ips_badfrags++;
1.75 thorpej 928: IPQ_UNLOCK();
1.16 cgd 929: goto bad;
930: }
931: }
1.155 itojun 932: ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3);
1.1 cgd 933:
934: /*
935: * If datagram marked as having more fragments
936: * or if this is not the first fragment,
937: * attempt reassembly; if it succeeds, proceed.
938: */
1.155 itojun 939: if (mff || ip->ip_off != htons(0)) {
1.1 cgd 940: ipstat.ips_fragments++;
1.72 thorpej 941: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.25 cgd 942: if (ipqe == NULL) {
943: ipstat.ips_rcvmemdrop++;
1.75 thorpej 944: IPQ_UNLOCK();
1.25 cgd 945: goto bad;
946: }
947: ipqe->ipqe_mff = mff;
1.50 thorpej 948: ipqe->ipqe_m = m;
1.25 cgd 949: ipqe->ipqe_ip = ip;
1.190 jonathan 950: m = ip_reass(ipqe, fp, &ipq[hash]);
1.75 thorpej 951: if (m == 0) {
952: IPQ_UNLOCK();
1.89 itojun 953: return;
1.75 thorpej 954: }
1.13 mycroft 955: ipstat.ips_reassembled++;
1.50 thorpej 956: ip = mtod(m, struct ip *);
1.74 thorpej 957: hlen = ip->ip_hl << 2;
1.155 itojun 958: ip->ip_len = htons(ntohs(ip->ip_len) + hlen);
1.1 cgd 959: } else
960: if (fp)
961: ip_freef(fp);
1.75 thorpej 962: IPQ_UNLOCK();
1.79 mycroft 963: }
1.128 itojun 964:
1.173 jonathan 965: #if defined(IPSEC)
1.128 itojun 966: /*
967: * enforce IPsec policy checking if we are seeing last header.
968: * note that we do not visit this with protocols with pcb layer
969: * code - like udp/tcp/raw ip.
970: */
971: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 &&
972: ipsec4_in_reject(m, NULL)) {
973: ipsecstat.in_polvio++;
974: goto bad;
975: }
976: #endif
1.173 jonathan 977: #if FAST_IPSEC
978: /*
979: * enforce IPsec policy checking if we are seeing last header.
980: * note that we do not visit this with protocols with pcb layer
981: * code - like udp/tcp/raw ip.
982: */
983: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) {
984: /*
985: * Check if the packet has already had IPsec processing
986: * done. If so, then just pass it along. This tag gets
987: * set during AH, ESP, etc. input handling, before the
988: * packet is returned to the ip input queue for delivery.
989: */
990: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
991: s = splsoftnet();
992: if (mtag != NULL) {
993: tdbi = (struct tdb_ident *)(mtag + 1);
994: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
995: } else {
996: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
997: IP_FORWARDING, &error);
998: }
999: if (sp != NULL) {
1000: /*
1001: * Check security policy against packet attributes.
1002: */
1003: error = ipsec_in_reject(sp, m);
1004: KEY_FREESP(&sp);
1005: } else {
1006: /* XXX error stat??? */
1007: error = EINVAL;
1008: DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/
1009: goto bad;
1010: }
1011: splx(s);
1012: if (error)
1013: goto bad;
1014: }
1015: #endif /* FAST_IPSEC */
1.1 cgd 1016:
1017: /*
1018: * Switch out to protocol's input routine.
1019: */
1.82 aidan 1020: #if IFA_STATS
1.122 itojun 1021: if (ia && ip)
1.155 itojun 1022: ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
1.82 aidan 1023: #endif
1.1 cgd 1024: ipstat.ips_delivered++;
1.89 itojun 1025: {
1026: int off = hlen, nh = ip->ip_p;
1027:
1028: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
1029: return;
1030: }
1.1 cgd 1031: bad:
1032: m_freem(m);
1.135 thorpej 1033: return;
1034:
1035: badcsum:
1036: ipstat.ips_badsum++;
1037: m_freem(m);
1.1 cgd 1038: }
1039:
1040: /*
1041: * Take incoming datagram fragment and try to
1042: * reassemble it into whole datagram. If a chain for
1043: * reassembly of this datagram already exists, then it
1044: * is given as fp; otherwise have to make a chain.
1045: */
1.50 thorpej 1046: struct mbuf *
1.211 ! perry 1047: ip_reass(struct ipqent *ipqe, struct ipq *fp, struct ipqhead *ipqhead)
1.1 cgd 1048: {
1.109 augustss 1049: struct mbuf *m = ipqe->ipqe_m;
1050: struct ipqent *nq, *p, *q;
1.25 cgd 1051: struct ip *ip;
1.1 cgd 1052: struct mbuf *t;
1.25 cgd 1053: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 1054: int i, next;
1055:
1.75 thorpej 1056: IPQ_LOCK_CHECK();
1057:
1.1 cgd 1058: /*
1059: * Presence of header sizes in mbufs
1060: * would confuse code below.
1061: */
1062: m->m_data += hlen;
1063: m->m_len -= hlen;
1064:
1.194 jonathan 1065: #ifdef notyet
1066: /* make sure fragment limit is up-to-date */
1067: CHECK_NMBCLUSTER_PARAMS();
1068:
1069: /* If we have too many fragments, drop the older half. */
1070: if (ip_nfrags >= ip_maxfrags)
1071: ip_reass_drophalf(void);
1072: #endif
1073:
1.1 cgd 1074: /*
1.192 jonathan 1075: * We are about to add a fragment; increment frag count.
1076: */
1077: ip_nfrags++;
1078:
1079: /*
1.1 cgd 1080: * If first fragment to arrive, create a reassembly queue.
1081: */
1082: if (fp == 0) {
1.131 itojun 1083: /*
1084: * Enforce upper bound on number of fragmented packets
1085: * for which we attempt reassembly;
1086: * If maxfrag is 0, never accept fragments.
1087: * If maxfrag is -1, accept all fragments without limitation.
1088: */
1089: if (ip_maxfragpackets < 0)
1090: ;
1091: else if (ip_nfragpackets >= ip_maxfragpackets)
1092: goto dropfrag;
1093: ip_nfragpackets++;
1.50 thorpej 1094: MALLOC(fp, struct ipq *, sizeof (struct ipq),
1095: M_FTABLE, M_NOWAIT);
1096: if (fp == NULL)
1.1 cgd 1097: goto dropfrag;
1.190 jonathan 1098: LIST_INSERT_HEAD(ipqhead, fp, ipq_q);
1.192 jonathan 1099: fp->ipq_nfrags = 1;
1.1 cgd 1100: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 1101: fp->ipq_p = ipqe->ipqe_ip->ip_p;
1102: fp->ipq_id = ipqe->ipqe_ip->ip_id;
1.148 matt 1103: TAILQ_INIT(&fp->ipq_fragq);
1.25 cgd 1104: fp->ipq_src = ipqe->ipqe_ip->ip_src;
1105: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
1106: p = NULL;
1.1 cgd 1107: goto insert;
1.192 jonathan 1108: } else {
1109: fp->ipq_nfrags++;
1.1 cgd 1110: }
1111:
1112: /*
1113: * Find a segment which begins after this one does.
1114: */
1.148 matt 1115: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1116: p = q, q = TAILQ_NEXT(q, ipqe_q))
1.155 itojun 1117: if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off))
1.1 cgd 1118: break;
1119:
1120: /*
1121: * If there is a preceding segment, it may provide some of
1122: * our data already. If so, drop the data from the incoming
1123: * segment. If it provides all of our data, drop us.
1124: */
1.25 cgd 1125: if (p != NULL) {
1.155 itojun 1126: i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) -
1127: ntohs(ipqe->ipqe_ip->ip_off);
1.1 cgd 1128: if (i > 0) {
1.155 itojun 1129: if (i >= ntohs(ipqe->ipqe_ip->ip_len))
1.1 cgd 1130: goto dropfrag;
1.50 thorpej 1131: m_adj(ipqe->ipqe_m, i);
1.155 itojun 1132: ipqe->ipqe_ip->ip_off =
1133: htons(ntohs(ipqe->ipqe_ip->ip_off) + i);
1134: ipqe->ipqe_ip->ip_len =
1135: htons(ntohs(ipqe->ipqe_ip->ip_len) - i);
1.1 cgd 1136: }
1137: }
1138:
1139: /*
1140: * While we overlap succeeding segments trim them or,
1141: * if they are completely covered, dequeue them.
1142: */
1.155 itojun 1143: for (; q != NULL &&
1144: ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) >
1145: ntohs(q->ipqe_ip->ip_off); q = nq) {
1146: i = (ntohs(ipqe->ipqe_ip->ip_off) +
1147: ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off);
1148: if (i < ntohs(q->ipqe_ip->ip_len)) {
1149: q->ipqe_ip->ip_len =
1150: htons(ntohs(q->ipqe_ip->ip_len) - i);
1151: q->ipqe_ip->ip_off =
1152: htons(ntohs(q->ipqe_ip->ip_off) + i);
1.50 thorpej 1153: m_adj(q->ipqe_m, i);
1.1 cgd 1154: break;
1155: }
1.148 matt 1156: nq = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1157: m_freem(q->ipqe_m);
1.148 matt 1158: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.72 thorpej 1159: pool_put(&ipqent_pool, q);
1.192 jonathan 1160: fp->ipq_nfrags--;
1161: ip_nfrags--;
1.1 cgd 1162: }
1163:
1164: insert:
1165: /*
1166: * Stick new segment in its place;
1167: * check for complete reassembly.
1168: */
1.25 cgd 1169: if (p == NULL) {
1.148 matt 1170: TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
1.25 cgd 1171: } else {
1.148 matt 1172: TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q);
1.25 cgd 1173: }
1.1 cgd 1174: next = 0;
1.148 matt 1175: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1176: p = q, q = TAILQ_NEXT(q, ipqe_q)) {
1.155 itojun 1177: if (ntohs(q->ipqe_ip->ip_off) != next)
1.1 cgd 1178: return (0);
1.155 itojun 1179: next += ntohs(q->ipqe_ip->ip_len);
1.1 cgd 1180: }
1.25 cgd 1181: if (p->ipqe_mff)
1.1 cgd 1182: return (0);
1183:
1184: /*
1.41 thorpej 1185: * Reassembly is complete. Check for a bogus message size and
1186: * concatenate fragments.
1.1 cgd 1187: */
1.148 matt 1188: q = TAILQ_FIRST(&fp->ipq_fragq);
1.25 cgd 1189: ip = q->ipqe_ip;
1.41 thorpej 1190: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
1191: ipstat.ips_toolong++;
1192: ip_freef(fp);
1193: return (0);
1194: }
1.50 thorpej 1195: m = q->ipqe_m;
1.1 cgd 1196: t = m->m_next;
1197: m->m_next = 0;
1198: m_cat(m, t);
1.148 matt 1199: nq = TAILQ_NEXT(q, ipqe_q);
1.72 thorpej 1200: pool_put(&ipqent_pool, q);
1.25 cgd 1201: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 1202: t = q->ipqe_m;
1.148 matt 1203: nq = TAILQ_NEXT(q, ipqe_q);
1.72 thorpej 1204: pool_put(&ipqent_pool, q);
1.1 cgd 1205: m_cat(m, t);
1206: }
1.192 jonathan 1207: ip_nfrags -= fp->ipq_nfrags;
1.1 cgd 1208:
1209: /*
1210: * Create header for new ip packet by
1211: * modifying header of first packet;
1212: * dequeue and discard fragment reassembly header.
1213: * Make header visible.
1214: */
1.155 itojun 1215: ip->ip_len = htons(next);
1.25 cgd 1216: ip->ip_src = fp->ipq_src;
1217: ip->ip_dst = fp->ipq_dst;
1218: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1219: FREE(fp, M_FTABLE);
1.131 itojun 1220: ip_nfragpackets--;
1.1 cgd 1221: m->m_len += (ip->ip_hl << 2);
1222: m->m_data -= (ip->ip_hl << 2);
1223: /* some debugging cruft by sklower, below, will go away soon */
1224: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
1.109 augustss 1225: int plen = 0;
1.50 thorpej 1226: for (t = m; t; t = t->m_next)
1227: plen += t->m_len;
1228: m->m_pkthdr.len = plen;
1.1 cgd 1229: }
1.50 thorpej 1230: return (m);
1.1 cgd 1231:
1232: dropfrag:
1.192 jonathan 1233: if (fp != 0)
1234: fp->ipq_nfrags--;
1235: ip_nfrags--;
1.1 cgd 1236: ipstat.ips_fragdropped++;
1237: m_freem(m);
1.72 thorpej 1238: pool_put(&ipqent_pool, ipqe);
1.1 cgd 1239: return (0);
1240: }
1241:
1242: /*
1243: * Free a fragment reassembly header and all
1244: * associated datagrams.
1245: */
1.8 mycroft 1246: void
1.211 ! perry 1247: ip_freef(struct ipq *fp)
1.1 cgd 1248: {
1.109 augustss 1249: struct ipqent *q, *p;
1.192 jonathan 1250: u_int nfrags = 0;
1.1 cgd 1251:
1.75 thorpej 1252: IPQ_LOCK_CHECK();
1253:
1.148 matt 1254: for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) {
1255: p = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1256: m_freem(q->ipqe_m);
1.192 jonathan 1257: nfrags++;
1.148 matt 1258: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.72 thorpej 1259: pool_put(&ipqent_pool, q);
1.1 cgd 1260: }
1.192 jonathan 1261:
1262: if (nfrags != fp->ipq_nfrags)
1263: printf("ip_freef: nfrags %d != %d\n", fp->ipq_nfrags, nfrags);
1264: ip_nfrags -= nfrags;
1.25 cgd 1265: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1266: FREE(fp, M_FTABLE);
1.131 itojun 1267: ip_nfragpackets--;
1.1 cgd 1268: }
1269:
1270: /*
1.194 jonathan 1271: * IP reassembly TTL machinery for multiplicative drop.
1272: */
1273: static u_int fragttl_histo[(IPFRAGTTL+1)];
1274:
1275:
1276: /*
1277: * Decrement TTL of all reasembly queue entries by `ticks'.
1278: * Count number of distinct fragments (as opposed to partial, fragmented
1279: * datagrams) in the reassembly queue. While we traverse the entire
1280: * reassembly queue, compute and return the median TTL over all fragments.
1281: */
1282: static u_int
1283: ip_reass_ttl_decr(u_int ticks)
1284: {
1.198 matt 1285: u_int nfrags, median, dropfraction, keepfraction;
1.194 jonathan 1286: struct ipq *fp, *nfp;
1.198 matt 1287: int i;
1.194 jonathan 1288:
1289: nfrags = 0;
1290: memset(fragttl_histo, 0, sizeof fragttl_histo);
1291:
1292: for (i = 0; i < IPREASS_NHASH; i++) {
1293: for (fp = LIST_FIRST(&ipq[i]); fp != NULL; fp = nfp) {
1294: fp->ipq_ttl = ((fp->ipq_ttl <= ticks) ?
1295: 0 : fp->ipq_ttl - ticks);
1296: nfp = LIST_NEXT(fp, ipq_q);
1297: if (fp->ipq_ttl == 0) {
1298: ipstat.ips_fragtimeout++;
1299: ip_freef(fp);
1300: } else {
1301: nfrags += fp->ipq_nfrags;
1302: fragttl_histo[fp->ipq_ttl] += fp->ipq_nfrags;
1303: }
1304: }
1305: }
1306:
1307: KASSERT(ip_nfrags == nfrags);
1308:
1309: /* Find median (or other drop fraction) in histogram. */
1310: dropfraction = (ip_nfrags / 2);
1311: keepfraction = ip_nfrags - dropfraction;
1312: for (i = IPFRAGTTL, median = 0; i >= 0; i--) {
1313: median += fragttl_histo[i];
1314: if (median >= keepfraction)
1315: break;
1316: }
1317:
1318: /* Return TTL of median (or other fraction). */
1319: return (u_int)i;
1320: }
1321:
1322: void
1323: ip_reass_drophalf(void)
1324: {
1325:
1326: u_int median_ticks;
1327: /*
1328: * Compute median TTL of all fragments, and count frags
1329: * with that TTL or lower (roughly half of all fragments).
1330: */
1331: median_ticks = ip_reass_ttl_decr(0);
1332:
1333: /* Drop half. */
1334: median_ticks = ip_reass_ttl_decr(median_ticks);
1335:
1336: }
1337:
1338: /*
1.1 cgd 1339: * IP timer processing;
1340: * if a timer expires on a reassembly
1341: * queue, discard it.
1342: */
1.8 mycroft 1343: void
1.211 ! perry 1344: ip_slowtimo(void)
1.1 cgd 1345: {
1.191 jonathan 1346: static u_int dropscanidx = 0;
1347: u_int i;
1.194 jonathan 1348: u_int median_ttl;
1.24 mycroft 1349: int s = splsoftnet();
1.1 cgd 1350:
1.75 thorpej 1351: IPQ_LOCK();
1.194 jonathan 1352:
1353: /* Age TTL of all fragments by 1 tick .*/
1354: median_ttl = ip_reass_ttl_decr(1);
1355:
1356: /* make sure fragment limit is up-to-date */
1357: CHECK_NMBCLUSTER_PARAMS();
1358:
1359: /* If we have too many fragments, drop the older half. */
1360: if (ip_nfrags > ip_maxfrags)
1361: ip_reass_ttl_decr(median_ttl);
1362:
1.131 itojun 1363: /*
1.194 jonathan 1364: * If we are over the maximum number of fragmented packets
1.131 itojun 1365: * (due to the limit being lowered), drain off
1.190 jonathan 1366: * enough to get down to the new limit. Start draining
1367: * from the reassembly hashqueue most recently drained.
1.131 itojun 1368: */
1369: if (ip_maxfragpackets < 0)
1370: ;
1371: else {
1.190 jonathan 1372: int wrapped = 0;
1373:
1374: i = dropscanidx;
1375: while (ip_nfragpackets > ip_maxfragpackets && wrapped == 0) {
1376: while (LIST_FIRST(&ipq[i]) != NULL)
1377: ip_freef(LIST_FIRST(&ipq[i]));
1378: if (++i >= IPREASS_NHASH) {
1379: i = 0;
1380: }
1381: /*
1382: * Dont scan forever even if fragment counters are
1383: * wrong: stop after scanning entire reassembly queue.
1384: */
1385: if (i == dropscanidx)
1386: wrapped = 1;
1387: }
1388: dropscanidx = i;
1.131 itojun 1389: }
1.75 thorpej 1390: IPQ_UNLOCK();
1.63 matt 1391: #ifdef GATEWAY
1392: ipflow_slowtimo();
1393: #endif
1.1 cgd 1394: splx(s);
1395: }
1396:
1397: /*
1398: * Drain off all datagram fragments.
1399: */
1.8 mycroft 1400: void
1.211 ! perry 1401: ip_drain(void)
1.1 cgd 1402: {
1403:
1.75 thorpej 1404: /*
1405: * We may be called from a device's interrupt context. If
1406: * the ipq is already busy, just bail out now.
1407: */
1408: if (ipq_lock_try() == 0)
1409: return;
1410:
1.194 jonathan 1411: /*
1412: * Drop half the total fragments now. If more mbufs are needed,
1413: * we will be called again soon.
1414: */
1415: ip_reass_drophalf();
1.75 thorpej 1416:
1417: IPQ_UNLOCK();
1.1 cgd 1418: }
1419:
1420: /*
1421: * Do option processing on a datagram,
1422: * possibly discarding it if bad options are encountered,
1423: * or forwarding it if source-routed.
1424: * Returns 1 if packet has been forwarded/freed,
1425: * 0 if the packet should be processed further.
1426: */
1.8 mycroft 1427: int
1.211 ! perry 1428: ip_dooptions(struct mbuf *m)
1.1 cgd 1429: {
1.109 augustss 1430: struct ip *ip = mtod(m, struct ip *);
1431: u_char *cp, *cp0;
1432: struct ip_timestamp *ipt;
1433: struct in_ifaddr *ia;
1.1 cgd 1434: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 1435: struct in_addr dst;
1.1 cgd 1436: n_time ntime;
1437:
1.13 mycroft 1438: dst = ip->ip_dst;
1.1 cgd 1439: cp = (u_char *)(ip + 1);
1440: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
1441: for (; cnt > 0; cnt -= optlen, cp += optlen) {
1442: opt = cp[IPOPT_OPTVAL];
1443: if (opt == IPOPT_EOL)
1444: break;
1445: if (opt == IPOPT_NOP)
1446: optlen = 1;
1447: else {
1.113 itojun 1448: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1449: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1450: goto bad;
1451: }
1.1 cgd 1452: optlen = cp[IPOPT_OLEN];
1.114 itojun 1453: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 1454: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1455: goto bad;
1456: }
1457: }
1458: switch (opt) {
1459:
1460: default:
1461: break;
1462:
1463: /*
1464: * Source routing with record.
1465: * Find interface with current destination address.
1466: * If none on this machine then drop if strictly routed,
1467: * or do nothing if loosely routed.
1468: * Record interface address and bring up next address
1469: * component. If strictly routed make sure next
1470: * address is on directly accessible net.
1471: */
1472: case IPOPT_LSRR:
1473: case IPOPT_SSRR:
1.47 cjs 1474: if (ip_allowsrcrt == 0) {
1475: type = ICMP_UNREACH;
1476: code = ICMP_UNREACH_NET_PROHIB;
1477: goto bad;
1478: }
1.114 itojun 1479: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1480: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1481: goto bad;
1482: }
1.1 cgd 1483: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1484: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1485: goto bad;
1486: }
1487: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 1488: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 1489: if (ia == 0) {
1490: if (opt == IPOPT_SSRR) {
1491: type = ICMP_UNREACH;
1492: code = ICMP_UNREACH_SRCFAIL;
1493: goto bad;
1494: }
1495: /*
1496: * Loose routing, and not at next destination
1497: * yet; nothing to do except forward.
1498: */
1499: break;
1500: }
1501: off--; /* 0 origin */
1.112 sommerfe 1502: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 1503: /*
1504: * End of source route. Should be for us.
1505: */
1506: save_rte(cp, ip->ip_src);
1507: break;
1508: }
1509: /*
1510: * locate outgoing interface
1511: */
1512: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
1513: sizeof(ipaddr.sin_addr));
1.96 thorpej 1514: if (opt == IPOPT_SSRR)
1.196 itojun 1515: ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr)));
1.96 thorpej 1516: else
1.1 cgd 1517: ia = ip_rtaddr(ipaddr.sin_addr);
1518: if (ia == 0) {
1519: type = ICMP_UNREACH;
1520: code = ICMP_UNREACH_SRCFAIL;
1521: goto bad;
1522: }
1523: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 1524: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 1525: (caddr_t)(cp + off), sizeof(struct in_addr));
1526: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1527: /*
1528: * Let ip_intr's mcast routing check handle mcast pkts
1529: */
1.18 mycroft 1530: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1531: break;
1532:
1533: case IPOPT_RR:
1.114 itojun 1534: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1535: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1536: goto bad;
1537: }
1.1 cgd 1538: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1539: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1540: goto bad;
1541: }
1542: /*
1543: * If no space remains, ignore.
1544: */
1545: off--; /* 0 origin */
1.112 sommerfe 1546: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1547: break;
1548: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
1549: sizeof(ipaddr.sin_addr));
1550: /*
1551: * locate outgoing interface; if we're the destination,
1552: * use the incoming interface (should be same).
1553: */
1.96 thorpej 1554: if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))))
1555: == NULL &&
1556: (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
1.1 cgd 1557: type = ICMP_UNREACH;
1558: code = ICMP_UNREACH_HOST;
1559: goto bad;
1560: }
1.20 mycroft 1561: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 1562: (caddr_t)(cp + off), sizeof(struct in_addr));
1563: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1564: break;
1565:
1566: case IPOPT_TS:
1567: code = cp - (u_char *)ip;
1568: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1569: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1570: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1571: goto bad;
1.114 itojun 1572: }
1573: if (ipt->ipt_ptr < 5) {
1574: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1575: goto bad;
1576: }
1.15 cgd 1577: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1578: if (++ipt->ipt_oflw == 0) {
1579: code = (u_char *)&ipt->ipt_ptr -
1580: (u_char *)ip;
1.1 cgd 1581: goto bad;
1.114 itojun 1582: }
1.1 cgd 1583: break;
1584: }
1.104 thorpej 1585: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1586: switch (ipt->ipt_flg) {
1587:
1588: case IPOPT_TS_TSONLY:
1589: break;
1590:
1591: case IPOPT_TS_TSANDADDR:
1.66 thorpej 1592: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1593: sizeof(struct in_addr) > ipt->ipt_len) {
1594: code = (u_char *)&ipt->ipt_ptr -
1595: (u_char *)ip;
1.1 cgd 1596: goto bad;
1.114 itojun 1597: }
1.13 mycroft 1598: ipaddr.sin_addr = dst;
1.96 thorpej 1599: ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr),
1600: m->m_pkthdr.rcvif));
1.13 mycroft 1601: if (ia == 0)
1602: continue;
1.104 thorpej 1603: bcopy(&ia->ia_addr.sin_addr,
1604: cp0, sizeof(struct in_addr));
1.1 cgd 1605: ipt->ipt_ptr += sizeof(struct in_addr);
1606: break;
1607:
1608: case IPOPT_TS_PRESPEC:
1.66 thorpej 1609: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1610: sizeof(struct in_addr) > ipt->ipt_len) {
1611: code = (u_char *)&ipt->ipt_ptr -
1612: (u_char *)ip;
1.1 cgd 1613: goto bad;
1.114 itojun 1614: }
1.104 thorpej 1615: bcopy(cp0, &ipaddr.sin_addr,
1.1 cgd 1616: sizeof(struct in_addr));
1.96 thorpej 1617: if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))
1618: == NULL)
1.1 cgd 1619: continue;
1620: ipt->ipt_ptr += sizeof(struct in_addr);
1621: break;
1622:
1623: default:
1.114 itojun 1624: /* XXX can't take &ipt->ipt_flg */
1625: code = (u_char *)&ipt->ipt_ptr -
1626: (u_char *)ip + 1;
1.1 cgd 1627: goto bad;
1628: }
1629: ntime = iptime();
1.107 thorpej 1630: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1631: bcopy(cp0, (caddr_t)cp + ipt->ipt_ptr - 1,
1.1 cgd 1632: sizeof(n_time));
1633: ipt->ipt_ptr += sizeof(n_time);
1634: }
1635: }
1636: if (forward) {
1.26 thorpej 1637: if (ip_forwsrcrt == 0) {
1638: type = ICMP_UNREACH;
1639: code = ICMP_UNREACH_SRCFAIL;
1640: goto bad;
1641: }
1.1 cgd 1642: ip_forward(m, 1);
1643: return (1);
1.13 mycroft 1644: }
1645: return (0);
1.1 cgd 1646: bad:
1.13 mycroft 1647: icmp_error(m, type, code, 0, 0);
1648: ipstat.ips_badoptions++;
1.1 cgd 1649: return (1);
1650: }
1651:
1652: /*
1653: * Given address of next destination (final or next hop),
1654: * return internet address info of interface to be used to get there.
1655: */
1656: struct in_ifaddr *
1.211 ! perry 1657: ip_rtaddr(struct in_addr dst)
1.1 cgd 1658: {
1.109 augustss 1659: struct sockaddr_in *sin;
1.1 cgd 1660:
1.19 mycroft 1661: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1662:
1.35 mycroft 1663: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 1664: if (ipforward_rt.ro_rt) {
1665: RTFREE(ipforward_rt.ro_rt);
1666: ipforward_rt.ro_rt = 0;
1667: }
1668: sin->sin_family = AF_INET;
1669: sin->sin_len = sizeof(*sin);
1670: sin->sin_addr = dst;
1671:
1672: rtalloc(&ipforward_rt);
1673: }
1674: if (ipforward_rt.ro_rt == 0)
1675: return ((struct in_ifaddr *)0);
1.19 mycroft 1676: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 1677: }
1678:
1679: /*
1680: * Save incoming source route for use in replies,
1681: * to be picked up later by ip_srcroute if the receiver is interested.
1682: */
1.13 mycroft 1683: void
1.211 ! perry 1684: save_rte(u_char *option, struct in_addr dst)
1.1 cgd 1685: {
1686: unsigned olen;
1687:
1688: olen = option[IPOPT_OLEN];
1689: #ifdef DIAGNOSTIC
1690: if (ipprintfs)
1.39 christos 1691: printf("save_rte: olen %d\n", olen);
1.89 itojun 1692: #endif /* 0 */
1.1 cgd 1693: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1694: return;
1695: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
1696: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1697: ip_srcrt.dst = dst;
1698: }
1699:
1700: /*
1701: * Retrieve incoming source route for use in replies,
1702: * in the same form used by setsockopt.
1703: * The first hop is placed before the options, will be removed later.
1704: */
1705: struct mbuf *
1.211 ! perry 1706: ip_srcroute(void)
1.1 cgd 1707: {
1.109 augustss 1708: struct in_addr *p, *q;
1709: struct mbuf *m;
1.1 cgd 1710:
1711: if (ip_nhops == 0)
1712: return ((struct mbuf *)0);
1713: m = m_get(M_DONTWAIT, MT_SOOPTS);
1714: if (m == 0)
1715: return ((struct mbuf *)0);
1716:
1.164 matt 1717: MCLAIM(m, &inetdomain.dom_mowner);
1.13 mycroft 1718: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1719:
1720: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1721: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1722: OPTSIZ;
1723: #ifdef DIAGNOSTIC
1724: if (ipprintfs)
1.39 christos 1725: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1726: #endif
1727:
1728: /*
1729: * First save first hop for return route
1730: */
1731: p = &ip_srcrt.route[ip_nhops - 1];
1732: *(mtod(m, struct in_addr *)) = *p--;
1733: #ifdef DIAGNOSTIC
1734: if (ipprintfs)
1.39 christos 1735: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1736: #endif
1737:
1738: /*
1739: * Copy option fields and padding (nop) to mbuf.
1740: */
1741: ip_srcrt.nop = IPOPT_NOP;
1742: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1743: bcopy((caddr_t)&ip_srcrt.nop,
1744: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1745: q = (struct in_addr *)(mtod(m, caddr_t) +
1746: sizeof(struct in_addr) + OPTSIZ);
1747: #undef OPTSIZ
1748: /*
1749: * Record return path as an IP source route,
1750: * reversing the path (pointers are now aligned).
1751: */
1752: while (p >= ip_srcrt.route) {
1753: #ifdef DIAGNOSTIC
1754: if (ipprintfs)
1.39 christos 1755: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1756: #endif
1757: *q++ = *p--;
1758: }
1759: /*
1760: * Last hop goes to final destination.
1761: */
1762: *q = ip_srcrt.dst;
1763: #ifdef DIAGNOSTIC
1764: if (ipprintfs)
1.39 christos 1765: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1766: #endif
1767: return (m);
1768: }
1769:
1770: /*
1771: * Strip out IP options, at higher
1772: * level protocol in the kernel.
1773: * Second argument is buffer to which options
1774: * will be moved, and return value is their length.
1775: * XXX should be deleted; last arg currently ignored.
1776: */
1.8 mycroft 1777: void
1.211 ! perry 1778: ip_stripoptions(struct mbuf *m, struct mbuf *mopt)
1.1 cgd 1779: {
1.109 augustss 1780: int i;
1.1 cgd 1781: struct ip *ip = mtod(m, struct ip *);
1.109 augustss 1782: caddr_t opts;
1.1 cgd 1783: int olen;
1784:
1.79 mycroft 1785: olen = (ip->ip_hl << 2) - sizeof (struct ip);
1.1 cgd 1786: opts = (caddr_t)(ip + 1);
1787: i = m->m_len - (sizeof (struct ip) + olen);
1788: bcopy(opts + olen, opts, (unsigned)i);
1789: m->m_len -= olen;
1790: if (m->m_flags & M_PKTHDR)
1791: m->m_pkthdr.len -= olen;
1.155 itojun 1792: ip->ip_len = htons(ntohs(ip->ip_len) - olen);
1.79 mycroft 1793: ip->ip_hl = sizeof (struct ip) >> 2;
1.1 cgd 1794: }
1795:
1.139 matt 1796: const int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1797: 0, 0, 0, 0,
1798: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1799: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1800: EMSGSIZE, EHOSTUNREACH, 0, 0,
1801: 0, 0, 0, 0,
1802: ENOPROTOOPT
1803: };
1804:
1805: /*
1806: * Forward a packet. If some error occurs return the sender
1807: * an icmp packet. Note we can't always generate a meaningful
1808: * icmp message because icmp doesn't have a large enough repertoire
1809: * of codes and types.
1810: *
1811: * If not forwarding, just drop the packet. This could be confusing
1812: * if ipforwarding was zero but some routing protocol was advancing
1813: * us as a gateway to somewhere. However, we must let the routing
1814: * protocol deal with that.
1815: *
1816: * The srcrt parameter indicates whether the packet is being forwarded
1817: * via a source route.
1818: */
1.13 mycroft 1819: void
1.211 ! perry 1820: ip_forward(struct mbuf *m, int srcrt)
1.1 cgd 1821: {
1.109 augustss 1822: struct ip *ip = mtod(m, struct ip *);
1823: struct sockaddr_in *sin;
1824: struct rtentry *rt;
1.28 christos 1825: int error, type = 0, code = 0;
1.1 cgd 1826: struct mbuf *mcopy;
1.13 mycroft 1827: n_long dest;
1828: struct ifnet *destifp;
1.173 jonathan 1829: #if defined(IPSEC) || defined(FAST_IPSEC)
1.89 itojun 1830: struct ifnet dummyifp;
1831: #endif
1.164 matt 1832:
1833: /*
1834: * We are now in the output path.
1835: */
1836: MCLAIM(m, &ip_tx_mowner);
1.135 thorpej 1837:
1838: /*
1839: * Clear any in-bound checksum flags for this packet.
1840: */
1841: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1842:
1.13 mycroft 1843: dest = 0;
1.1 cgd 1844: #ifdef DIAGNOSTIC
1845: if (ipprintfs)
1.70 thorpej 1846: printf("forward: src %2.2x dst %2.2x ttl %x\n",
1847: ntohl(ip->ip_src.s_addr),
1848: ntohl(ip->ip_dst.s_addr), ip->ip_ttl);
1.1 cgd 1849: #endif
1.93 sommerfe 1850: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.1 cgd 1851: ipstat.ips_cantforward++;
1852: m_freem(m);
1853: return;
1854: }
1855: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1856: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1857: return;
1858: }
1859: ip->ip_ttl -= IPTTLDEC;
1860:
1.19 mycroft 1861: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1862: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1863: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1864: if (ipforward_rt.ro_rt) {
1865: RTFREE(ipforward_rt.ro_rt);
1866: ipforward_rt.ro_rt = 0;
1867: }
1868: sin->sin_family = AF_INET;
1.35 mycroft 1869: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1870: sin->sin_addr = ip->ip_dst;
1871:
1872: rtalloc(&ipforward_rt);
1873: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1874: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1875: return;
1876: }
1877: rt = ipforward_rt.ro_rt;
1878: }
1879:
1880: /*
1.34 mycroft 1881: * Save at most 68 bytes of the packet in case
1.1 cgd 1882: * we need to generate an ICMP message to the src.
1.119 itojun 1883: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1884: */
1.155 itojun 1885: mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1.119 itojun 1886: if (mcopy)
1887: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1888:
1889: /*
1890: * If forwarding packet using same interface that it came in on,
1891: * perhaps should send a redirect to sender to shortcut a hop.
1892: * Only send redirect if source is sending directly to us,
1893: * and if packet was not source routed (or has any options).
1894: * Also, don't send redirect if forwarding using a default route
1895: * or a route modified by a redirect.
1896: */
1897: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1898: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1899: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1900: ipsendredirects && !srcrt) {
1.19 mycroft 1901: if (rt->rt_ifa &&
1902: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1903: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1904: if (rt->rt_flags & RTF_GATEWAY)
1905: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1906: else
1907: dest = ip->ip_dst.s_addr;
1908: /*
1909: * Router requirements says to only send host
1910: * redirects.
1911: */
1912: type = ICMP_REDIRECT;
1913: code = ICMP_REDIRECT_HOST;
1.1 cgd 1914: #ifdef DIAGNOSTIC
1.77 thorpej 1915: if (ipprintfs)
1916: printf("redirect (%d) to %x\n", code,
1917: (u_int32_t)dest);
1.1 cgd 1918: #endif
1919: }
1920: }
1921:
1.27 thorpej 1922: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1.173 jonathan 1923: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
1.174 itojun 1924: (struct ip_moptions *)NULL, (struct socket *)NULL);
1.173 jonathan 1925:
1.1 cgd 1926: if (error)
1927: ipstat.ips_cantforward++;
1928: else {
1929: ipstat.ips_forward++;
1930: if (type)
1931: ipstat.ips_redirectsent++;
1932: else {
1.63 matt 1933: if (mcopy) {
1934: #ifdef GATEWAY
1.64 thorpej 1935: if (mcopy->m_flags & M_CANFASTFWD)
1936: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1937: #endif
1.1 cgd 1938: m_freem(mcopy);
1.63 matt 1939: }
1.1 cgd 1940: return;
1941: }
1942: }
1943: if (mcopy == NULL)
1944: return;
1.13 mycroft 1945: destifp = NULL;
1946:
1.1 cgd 1947: switch (error) {
1948:
1949: case 0: /* forwarded, but need redirect */
1950: /* type, code set above */
1951: break;
1952:
1953: case ENETUNREACH: /* shouldn't happen, checked above */
1954: case EHOSTUNREACH:
1955: case ENETDOWN:
1956: case EHOSTDOWN:
1957: default:
1958: type = ICMP_UNREACH;
1959: code = ICMP_UNREACH_HOST;
1960: break;
1961:
1962: case EMSGSIZE:
1963: type = ICMP_UNREACH;
1964: code = ICMP_UNREACH_NEEDFRAG;
1.173 jonathan 1965: #if !defined(IPSEC) && !defined(FAST_IPSEC)
1.13 mycroft 1966: if (ipforward_rt.ro_rt)
1967: destifp = ipforward_rt.ro_rt->rt_ifp;
1.89 itojun 1968: #else
1969: /*
1970: * If the packet is routed over IPsec tunnel, tell the
1971: * originator the tunnel MTU.
1972: * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1973: * XXX quickhack!!!
1974: */
1975: if (ipforward_rt.ro_rt) {
1976: struct secpolicy *sp;
1977: int ipsecerror;
1.95 itojun 1978: size_t ipsechdr;
1.89 itojun 1979: struct route *ro;
1980:
1981: sp = ipsec4_getpolicybyaddr(mcopy,
1.170 itojun 1982: IPSEC_DIR_OUTBOUND, IP_FORWARDING,
1983: &ipsecerror);
1.89 itojun 1984:
1985: if (sp == NULL)
1986: destifp = ipforward_rt.ro_rt->rt_ifp;
1987: else {
1988: /* count IPsec header size */
1.95 itojun 1989: ipsechdr = ipsec4_hdrsiz(mcopy,
1.170 itojun 1990: IPSEC_DIR_OUTBOUND, NULL);
1.89 itojun 1991:
1992: /*
1993: * find the correct route for outer IPv4
1994: * header, compute tunnel MTU.
1995: *
1996: * XXX BUG ALERT
1997: * The "dummyifp" code relies upon the fact
1998: * that icmp_error() touches only ifp->if_mtu.
1999: */
2000: /*XXX*/
2001: destifp = NULL;
2002: if (sp->req != NULL
1.95 itojun 2003: && sp->req->sav != NULL
2004: && sp->req->sav->sah != NULL) {
2005: ro = &sp->req->sav->sah->sa_route;
1.89 itojun 2006: if (ro->ro_rt && ro->ro_rt->rt_ifp) {
2007: dummyifp.if_mtu =
1.151 itojun 2008: ro->ro_rt->rt_rmx.rmx_mtu ?
2009: ro->ro_rt->rt_rmx.rmx_mtu :
1.89 itojun 2010: ro->ro_rt->rt_ifp->if_mtu;
2011: dummyifp.if_mtu -= ipsechdr;
2012: destifp = &dummyifp;
2013: }
2014: }
2015:
1.173 jonathan 2016: #ifdef IPSEC
1.89 itojun 2017: key_freesp(sp);
1.173 jonathan 2018: #else
2019: KEY_FREESP(&sp);
2020: #endif
1.89 itojun 2021: }
2022: }
2023: #endif /*IPSEC*/
1.1 cgd 2024: ipstat.ips_cantfrag++;
2025: break;
2026:
2027: case ENOBUFS:
1.143 itojun 2028: #if 1
2029: /*
2030: * a router should not generate ICMP_SOURCEQUENCH as
2031: * required in RFC1812 Requirements for IP Version 4 Routers.
2032: * source quench could be a big problem under DoS attacks,
1.149 wiz 2033: * or if the underlying interface is rate-limited.
1.143 itojun 2034: */
2035: if (mcopy)
2036: m_freem(mcopy);
2037: return;
2038: #else
1.1 cgd 2039: type = ICMP_SOURCEQUENCH;
2040: code = 0;
2041: break;
1.143 itojun 2042: #endif
1.1 cgd 2043: }
1.13 mycroft 2044: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 2045: }
2046:
2047: void
1.211 ! perry 2048: ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
! 2049: struct mbuf *m)
1.44 thorpej 2050: {
2051:
2052: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
2053: struct timeval tv;
2054:
2055: microtime(&tv);
2056: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
2057: SCM_TIMESTAMP, SOL_SOCKET);
2058: if (*mp)
2059: mp = &(*mp)->m_next;
2060: }
2061: if (inp->inp_flags & INP_RECVDSTADDR) {
2062: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
2063: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
2064: if (*mp)
2065: mp = &(*mp)->m_next;
2066: }
2067: #ifdef notyet
2068: /*
2069: * XXX
2070: * Moving these out of udp_input() made them even more broken
2071: * than they already were.
2072: * - fenner@parc.xerox.com
2073: */
2074: /* options were tossed already */
2075: if (inp->inp_flags & INP_RECVOPTS) {
2076: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
2077: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
2078: if (*mp)
2079: mp = &(*mp)->m_next;
2080: }
2081: /* ip_srcroute doesn't do what we want here, need to fix */
2082: if (inp->inp_flags & INP_RECVRETOPTS) {
2083: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
2084: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
2085: if (*mp)
2086: mp = &(*mp)->m_next;
2087: }
2088: #endif
2089: if (inp->inp_flags & INP_RECVIF) {
2090: struct sockaddr_dl sdl;
2091:
2092: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
2093: sdl.sdl_family = AF_LINK;
2094: sdl.sdl_index = m->m_pkthdr.rcvif ?
2095: m->m_pkthdr.rcvif->if_index : 0;
2096: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
2097: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
2098: IP_RECVIF, IPPROTO_IP);
2099: if (*mp)
2100: mp = &(*mp)->m_next;
2101: }
1.13 mycroft 2102: }
2103:
1.189 atatat 2104: /*
2105: * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the
2106: * range of the new value and tweaks timers if it changes.
2107: */
2108: static int
2109: sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS)
1.13 mycroft 2110: {
1.189 atatat 2111: int error, tmp;
2112: struct sysctlnode node;
2113:
2114: node = *rnode;
2115: tmp = ip_mtudisc_timeout;
2116: node.sysctl_data = &tmp;
2117: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2118: if (error || newp == NULL)
2119: return (error);
2120: if (tmp < 0)
2121: return (EINVAL);
1.52 thorpej 2122:
1.189 atatat 2123: ip_mtudisc_timeout = tmp;
2124: rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout);
2125:
2126: return (0);
2127: }
1.54 lukem 2128:
1.65 matt 2129: #ifdef GATEWAY
1.189 atatat 2130: /*
2131: * sysctl helper routine for net.inet.ip.maxflows. apparently if
2132: * maxflows is even looked up, we "reap flows".
2133: */
2134: static int
2135: sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS)
2136: {
2137: int s;
1.67 thorpej 2138:
1.189 atatat 2139: s = sysctl_lookup(SYSCTLFN_CALL(rnode));
2140: if (s)
2141: return (s);
2142:
2143: s = splsoftnet();
2144: ipflow_reap(0);
2145: splx(s);
1.144 martin 2146:
1.189 atatat 2147: return (0);
2148: }
2149: #endif /* GATEWAY */
1.117 tron 2150:
1.131 itojun 2151:
1.189 atatat 2152: SYSCTL_SETUP(sysctl_net_inet_ip_setup, "sysctl net.inet.ip subtree setup")
2153: {
2154: extern int subnetsarelocal, hostzeroisbroadcast;
1.180 jonathan 2155:
1.197 atatat 2156: sysctl_createv(clog, 0, NULL, NULL,
2157: CTLFLAG_PERMANENT,
1.189 atatat 2158: CTLTYPE_NODE, "net", NULL,
2159: NULL, 0, NULL, 0,
2160: CTL_NET, CTL_EOL);
1.197 atatat 2161: sysctl_createv(clog, 0, NULL, NULL,
2162: CTLFLAG_PERMANENT,
1.203 atatat 2163: CTLTYPE_NODE, "inet",
2164: SYSCTL_DESCR("PF_INET related settings"),
1.189 atatat 2165: NULL, 0, NULL, 0,
2166: CTL_NET, PF_INET, CTL_EOL);
1.197 atatat 2167: sysctl_createv(clog, 0, NULL, NULL,
2168: CTLFLAG_PERMANENT,
1.203 atatat 2169: CTLTYPE_NODE, "ip",
2170: SYSCTL_DESCR("IPv4 related settings"),
1.189 atatat 2171: NULL, 0, NULL, 0,
2172: CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL);
2173:
1.197 atatat 2174: sysctl_createv(clog, 0, NULL, NULL,
2175: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2176: CTLTYPE_INT, "forwarding",
2177: SYSCTL_DESCR("Enable forwarding of INET datagrams"),
1.189 atatat 2178: NULL, 0, &ipforwarding, 0,
2179: CTL_NET, PF_INET, IPPROTO_IP,
2180: IPCTL_FORWARDING, CTL_EOL);
1.197 atatat 2181: sysctl_createv(clog, 0, NULL, NULL,
2182: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2183: CTLTYPE_INT, "redirect",
2184: SYSCTL_DESCR("Enable sending of ICMP redirect messages"),
1.189 atatat 2185: NULL, 0, &ipsendredirects, 0,
2186: CTL_NET, PF_INET, IPPROTO_IP,
2187: IPCTL_SENDREDIRECTS, CTL_EOL);
1.197 atatat 2188: sysctl_createv(clog, 0, NULL, NULL,
2189: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2190: CTLTYPE_INT, "ttl",
2191: SYSCTL_DESCR("Default TTL for an INET datagram"),
1.189 atatat 2192: NULL, 0, &ip_defttl, 0,
2193: CTL_NET, PF_INET, IPPROTO_IP,
2194: IPCTL_DEFTTL, CTL_EOL);
2195: #ifdef IPCTL_DEFMTU
1.197 atatat 2196: sysctl_createv(clog, 0, NULL, NULL,
2197: CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */,
1.203 atatat 2198: CTLTYPE_INT, "mtu",
2199: SYSCTL_DESCR("Default MTA for an INET route"),
1.189 atatat 2200: NULL, 0, &ip_mtu, 0,
2201: CTL_NET, PF_INET, IPPROTO_IP,
2202: IPCTL_DEFMTU, CTL_EOL);
2203: #endif /* IPCTL_DEFMTU */
1.197 atatat 2204: sysctl_createv(clog, 0, NULL, NULL,
2205: CTLFLAG_PERMANENT|CTLFLAG_READONLY1,
1.203 atatat 2206: CTLTYPE_INT, "forwsrcrt",
2207: SYSCTL_DESCR("Enable forwarding of source-routed "
2208: "datagrams"),
1.189 atatat 2209: NULL, 0, &ip_forwsrcrt, 0,
2210: CTL_NET, PF_INET, IPPROTO_IP,
2211: IPCTL_FORWSRCRT, CTL_EOL);
1.197 atatat 2212: sysctl_createv(clog, 0, NULL, NULL,
2213: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2214: CTLTYPE_INT, "directed-broadcast",
2215: SYSCTL_DESCR("Enable forwarding of broadcast datagrams"),
1.189 atatat 2216: NULL, 0, &ip_directedbcast, 0,
2217: CTL_NET, PF_INET, IPPROTO_IP,
2218: IPCTL_DIRECTEDBCAST, CTL_EOL);
1.197 atatat 2219: sysctl_createv(clog, 0, NULL, NULL,
2220: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2221: CTLTYPE_INT, "allowsrcrt",
2222: SYSCTL_DESCR("Accept source-routed datagrams"),
1.189 atatat 2223: NULL, 0, &ip_allowsrcrt, 0,
2224: CTL_NET, PF_INET, IPPROTO_IP,
2225: IPCTL_ALLOWSRCRT, CTL_EOL);
1.197 atatat 2226: sysctl_createv(clog, 0, NULL, NULL,
2227: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2228: CTLTYPE_INT, "subnetsarelocal",
2229: SYSCTL_DESCR("Whether logical subnets are considered "
2230: "local"),
1.189 atatat 2231: NULL, 0, &subnetsarelocal, 0,
2232: CTL_NET, PF_INET, IPPROTO_IP,
2233: IPCTL_SUBNETSARELOCAL, CTL_EOL);
1.197 atatat 2234: sysctl_createv(clog, 0, NULL, NULL,
2235: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2236: CTLTYPE_INT, "mtudisc",
2237: SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"),
1.189 atatat 2238: NULL, 0, &ip_mtudisc, 0,
2239: CTL_NET, PF_INET, IPPROTO_IP,
2240: IPCTL_MTUDISC, CTL_EOL);
1.197 atatat 2241: sysctl_createv(clog, 0, NULL, NULL,
2242: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2243: CTLTYPE_INT, "anonportmin",
2244: SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1.189 atatat 2245: sysctl_net_inet_ip_ports, 0, &anonportmin, 0,
2246: CTL_NET, PF_INET, IPPROTO_IP,
2247: IPCTL_ANONPORTMIN, CTL_EOL);
1.197 atatat 2248: sysctl_createv(clog, 0, NULL, NULL,
2249: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2250: CTLTYPE_INT, "anonportmax",
2251: SYSCTL_DESCR("Highest ephemeral port number to assign"),
1.189 atatat 2252: sysctl_net_inet_ip_ports, 0, &anonportmax, 0,
2253: CTL_NET, PF_INET, IPPROTO_IP,
2254: IPCTL_ANONPORTMAX, CTL_EOL);
1.197 atatat 2255: sysctl_createv(clog, 0, NULL, NULL,
2256: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2257: CTLTYPE_INT, "mtudisctimeout",
2258: SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"),
1.189 atatat 2259: sysctl_net_inet_ip_pmtudto, 0, &ip_mtudisc_timeout, 0,
2260: CTL_NET, PF_INET, IPPROTO_IP,
2261: IPCTL_MTUDISCTIMEOUT, CTL_EOL);
2262: #ifdef GATEWAY
1.197 atatat 2263: sysctl_createv(clog, 0, NULL, NULL,
2264: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2265: CTLTYPE_INT, "maxflows",
2266: SYSCTL_DESCR("Number of flows for fast forwarding"),
1.189 atatat 2267: sysctl_net_inet_ip_maxflows, 0, &ip_maxflows, 0,
2268: CTL_NET, PF_INET, IPPROTO_IP,
2269: IPCTL_MAXFLOWS, CTL_EOL);
2270: #endif /* GATEWAY */
1.197 atatat 2271: sysctl_createv(clog, 0, NULL, NULL,
2272: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2273: CTLTYPE_INT, "hostzerobroadcast",
2274: SYSCTL_DESCR("All zeroes address is broadcast address"),
1.189 atatat 2275: NULL, 0, &hostzeroisbroadcast, 0,
2276: CTL_NET, PF_INET, IPPROTO_IP,
2277: IPCTL_HOSTZEROBROADCAST, CTL_EOL);
2278: #if NGIF > 0
1.197 atatat 2279: sysctl_createv(clog, 0, NULL, NULL,
2280: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2281: CTLTYPE_INT, "gifttl",
2282: SYSCTL_DESCR("Default TTL for a gif tunnel datagram"),
1.189 atatat 2283: NULL, 0, &ip_gif_ttl, 0,
2284: CTL_NET, PF_INET, IPPROTO_IP,
2285: IPCTL_GIF_TTL, CTL_EOL);
2286: #endif /* NGIF */
2287: #ifndef IPNOPRIVPORTS
1.197 atatat 2288: sysctl_createv(clog, 0, NULL, NULL,
2289: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2290: CTLTYPE_INT, "lowportmin",
2291: SYSCTL_DESCR("Lowest privileged ephemeral port number "
2292: "to assign"),
1.189 atatat 2293: sysctl_net_inet_ip_ports, 0, &lowportmin, 0,
2294: CTL_NET, PF_INET, IPPROTO_IP,
2295: IPCTL_LOWPORTMIN, CTL_EOL);
1.197 atatat 2296: sysctl_createv(clog, 0, NULL, NULL,
2297: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2298: CTLTYPE_INT, "lowportmax",
2299: SYSCTL_DESCR("Highest privileged ephemeral port number "
2300: "to assign"),
1.189 atatat 2301: sysctl_net_inet_ip_ports, 0, &lowportmax, 0,
2302: CTL_NET, PF_INET, IPPROTO_IP,
2303: IPCTL_LOWPORTMAX, CTL_EOL);
2304: #endif /* IPNOPRIVPORTS */
1.197 atatat 2305: sysctl_createv(clog, 0, NULL, NULL,
2306: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2307: CTLTYPE_INT, "maxfragpackets",
2308: SYSCTL_DESCR("Maximum number of fragments to retain for "
2309: "possible reassembly"),
1.189 atatat 2310: NULL, 0, &ip_maxfragpackets, 0,
2311: CTL_NET, PF_INET, IPPROTO_IP,
2312: IPCTL_MAXFRAGPACKETS, CTL_EOL);
2313: #if NGRE > 0
1.197 atatat 2314: sysctl_createv(clog, 0, NULL, NULL,
2315: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2316: CTLTYPE_INT, "grettl",
2317: SYSCTL_DESCR("Default TTL for a gre tunnel datagram"),
1.189 atatat 2318: NULL, 0, &ip_gre_ttl, 0,
2319: CTL_NET, PF_INET, IPPROTO_IP,
2320: IPCTL_GRE_TTL, CTL_EOL);
2321: #endif /* NGRE */
1.197 atatat 2322: sysctl_createv(clog, 0, NULL, NULL,
2323: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2324: CTLTYPE_INT, "checkinterface",
2325: SYSCTL_DESCR("Enable receive side of Strong ES model "
2326: "from RFC1122"),
1.189 atatat 2327: NULL, 0, &ip_checkinterface, 0,
2328: CTL_NET, PF_INET, IPPROTO_IP,
2329: IPCTL_CHECKINTERFACE, CTL_EOL);
1.197 atatat 2330: sysctl_createv(clog, 0, NULL, NULL,
2331: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2332: CTLTYPE_INT, "random_id",
2333: SYSCTL_DESCR("Assign random ip_id values"),
1.189 atatat 2334: NULL, 0, &ip_do_randomid, 0,
2335: CTL_NET, PF_INET, IPPROTO_IP,
2336: IPCTL_RANDOMID, CTL_EOL);
1.206 thorpej 2337: sysctl_createv(clog, 0, NULL, NULL,
2338: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2339: CTLTYPE_INT, "do_loopback_cksum",
2340: SYSCTL_DESCR("Perform IP checksum on loopback"),
2341: NULL, 0, &ip_do_loopback_cksum, 0,
2342: CTL_NET, PF_INET, IPPROTO_IP,
2343: IPCTL_LOOPBACKCKSUM, CTL_EOL);
1.1 cgd 2344: }
CVSweb <webmaster@jp.NetBSD.org>