Annotation of src/sys/netinet/ip_input.c, Revision 1.249
1.249 ! dyoung 1: /* $NetBSD: ip_input.c,v 1.248 2007/03/25 20:12:20 liamjfoy 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>
1.249 ! dyoung 101: __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.248 2007/03/25 20:12:20 liamjfoy Exp $");
1.55 scottr 102:
1.184 jonathan 103: #include "opt_inet.h"
1.62 matt 104: #include "opt_gateway.h"
1.69 mrg 105: #include "opt_pfil_hooks.h"
1.91 thorpej 106: #include "opt_ipsec.h"
1.55 scottr 107: #include "opt_mrouting.h"
1.167 martin 108: #include "opt_mbuftrace.h"
1.135 thorpej 109: #include "opt_inet_csum.h"
1.1 cgd 110:
1.5 mycroft 111: #include <sys/param.h>
112: #include <sys/systm.h>
113: #include <sys/malloc.h>
114: #include <sys/mbuf.h>
115: #include <sys/domain.h>
116: #include <sys/protosw.h>
117: #include <sys/socket.h>
1.44 thorpej 118: #include <sys/socketvar.h>
1.5 mycroft 119: #include <sys/errno.h>
120: #include <sys/time.h>
121: #include <sys/kernel.h>
1.72 thorpej 122: #include <sys/pool.h>
1.28 christos 123: #include <sys/sysctl.h>
1.230 elad 124: #include <sys/kauth.h>
1.1 cgd 125:
1.5 mycroft 126: #include <net/if.h>
1.44 thorpej 127: #include <net/if_dl.h>
1.5 mycroft 128: #include <net/route.h>
1.45 mrg 129: #include <net/pfil.h>
1.1 cgd 130:
1.5 mycroft 131: #include <netinet/in.h>
132: #include <netinet/in_systm.h>
133: #include <netinet/ip.h>
134: #include <netinet/in_pcb.h>
1.215 yamt 135: #include <netinet/in_proto.h>
1.5 mycroft 136: #include <netinet/in_var.h>
137: #include <netinet/ip_var.h>
138: #include <netinet/ip_icmp.h>
1.89 itojun 139: /* just for gif_ttl */
140: #include <netinet/in_gif.h>
141: #include "gif.h"
1.144 martin 142: #include <net/if_gre.h>
143: #include "gre.h"
1.111 jdolecek 144:
145: #ifdef MROUTING
146: #include <netinet/ip_mroute.h>
147: #endif
1.89 itojun 148:
149: #ifdef IPSEC
150: #include <netinet6/ipsec.h>
151: #include <netkey/key.h>
152: #endif
1.173 jonathan 153: #ifdef FAST_IPSEC
154: #include <netipsec/ipsec.h>
155: #include <netipsec/key.h>
156: #endif /* FAST_IPSEC*/
1.44 thorpej 157:
1.1 cgd 158: #ifndef IPFORWARDING
159: #ifdef GATEWAY
160: #define IPFORWARDING 1 /* forward IP packets not for us */
161: #else /* GATEWAY */
162: #define IPFORWARDING 0 /* don't forward IP packets not for us */
163: #endif /* GATEWAY */
164: #endif /* IPFORWARDING */
165: #ifndef IPSENDREDIRECTS
166: #define IPSENDREDIRECTS 1
167: #endif
1.26 thorpej 168: #ifndef IPFORWSRCRT
1.47 cjs 169: #define IPFORWSRCRT 1 /* forward source-routed packets */
170: #endif
171: #ifndef IPALLOWSRCRT
1.48 mrg 172: #define IPALLOWSRCRT 1 /* allow source-routed packets */
1.26 thorpej 173: #endif
1.53 kml 174: #ifndef IPMTUDISC
1.153 itojun 175: #define IPMTUDISC 1
1.53 kml 176: #endif
1.60 kml 177: #ifndef IPMTUDISCTIMEOUT
1.61 kml 178: #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */
1.60 kml 179: #endif
1.53 kml 180:
1.27 thorpej 181: /*
182: * Note: DIRECTED_BROADCAST is handled this way so that previous
183: * configuration using this option will Just Work.
184: */
185: #ifndef IPDIRECTEDBCAST
186: #ifdef DIRECTED_BROADCAST
187: #define IPDIRECTEDBCAST 1
188: #else
189: #define IPDIRECTEDBCAST 0
190: #endif /* DIRECTED_BROADCAST */
191: #endif /* IPDIRECTEDBCAST */
1.1 cgd 192: int ipforwarding = IPFORWARDING;
193: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 194: int ip_defttl = IPDEFTTL;
1.26 thorpej 195: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 196: int ip_directedbcast = IPDIRECTEDBCAST;
1.47 cjs 197: int ip_allowsrcrt = IPALLOWSRCRT;
1.53 kml 198: int ip_mtudisc = IPMTUDISC;
1.156 itojun 199: int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
1.1 cgd 200: #ifdef DIAGNOSTIC
201: int ipprintfs = 0;
202: #endif
1.184 jonathan 203:
204: int ip_do_randomid = 0;
205:
1.165 christos 206: /*
207: * XXX - Setting ip_checkinterface mostly implements the receive side of
208: * the Strong ES model described in RFC 1122, but since the routing table
209: * and transmit implementation do not implement the Strong ES model,
210: * setting this to 1 results in an odd hybrid.
211: *
212: * XXX - ip_checkinterface currently must be disabled if you use ipnat
213: * to translate the destination address to another local interface.
214: *
215: * XXX - ip_checkinterface must be disabled if you add IP aliases
216: * to the loopback interface instead of the interface where the
217: * packets for those addresses are received.
218: */
219: int ip_checkinterface = 0;
220:
1.1 cgd 221:
1.60 kml 222: struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
223:
1.1 cgd 224: int ipqmaxlen = IFQ_MAXLEN;
1.150 matt 225: u_long in_ifaddrhash; /* size of hash table - 1 */
226: int in_ifaddrentries; /* total number of addrs */
1.212 perry 227: struct in_ifaddrhead in_ifaddrhead;
1.57 tls 228: struct in_ifaddrhashhead *in_ifaddrhashtbl;
1.166 matt 229: u_long in_multihash; /* size of hash table - 1 */
230: int in_multientries; /* total number of addrs */
231: struct in_multihashhead *in_multihashtbl;
1.13 mycroft 232: struct ifqueue ipintrq;
1.63 matt 233: struct ipstat ipstat;
1.183 jonathan 234: uint16_t ip_id;
1.75 thorpej 235:
1.121 thorpej 236: #ifdef PFIL_HOOKS
237: struct pfil_head inet_pfil_hook;
238: #endif
239:
1.194 jonathan 240: /*
241: * Cached copy of nmbclusters. If nbclusters is different,
242: * recalculate IP parameters derived from nmbclusters.
243: */
244: static int ip_nmbclusters; /* copy of nmbclusters */
1.210 perry 245: static void ip_nmbclusters_changed(void); /* recalc limits */
1.194 jonathan 246:
1.195 thorpej 247: #define CHECK_NMBCLUSTER_PARAMS() \
248: do { \
249: if (__predict_false(ip_nmbclusters != nmbclusters)) \
250: ip_nmbclusters_changed(); \
251: } while (/*CONSTCOND*/0)
1.194 jonathan 252:
1.190 jonathan 253: /* IP datagram reassembly queues (hashed) */
254: #define IPREASS_NHASH_LOG2 6
255: #define IPREASS_NHASH (1 << IPREASS_NHASH_LOG2)
256: #define IPREASS_HMASK (IPREASS_NHASH - 1)
257: #define IPREASS_HASH(x,y) \
258: (((((x) & 0xF) | ((((x) >> 8) & 0xF) << 4)) ^ (y)) & IPREASS_HMASK)
259: struct ipqhead ipq[IPREASS_NHASH];
1.75 thorpej 260: int ipq_locked;
1.212 perry 261: static int ip_nfragpackets; /* packets in reass queue */
1.194 jonathan 262: static int ip_nfrags; /* total fragments in reass queues */
263:
264: int ip_maxfragpackets = 200; /* limit on packets. XXX sysctl */
265: int ip_maxfrags; /* limit on fragments. XXX sysctl */
266:
267:
268: /*
269: * Additive-Increase/Multiplicative-Decrease (AIMD) strategy for
270: * IP reassembly queue buffer managment.
1.212 perry 271: *
1.194 jonathan 272: * We keep a count of total IP fragments (NB: not fragmented packets!)
273: * awaiting reassembly (ip_nfrags) and a limit (ip_maxfrags) on fragments.
274: * If ip_nfrags exceeds ip_maxfrags the limit, we drop half the
275: * total fragments in reassembly queues.This AIMD policy avoids
276: * repeatedly deleting single packets under heavy fragmentation load
277: * (e.g., from lossy NFS peers).
278: */
1.212 perry 279: static u_int ip_reass_ttl_decr(u_int ticks);
1.210 perry 280: static void ip_reass_drophalf(void);
1.194 jonathan 281:
1.75 thorpej 282:
1.223 perry 283: static inline int ipq_lock_try(void);
284: static inline void ipq_unlock(void);
1.75 thorpej 285:
1.223 perry 286: static inline int
1.211 perry 287: ipq_lock_try(void)
1.75 thorpej 288: {
289: int s;
290:
1.132 thorpej 291: /*
1.149 wiz 292: * Use splvm() -- we're blocking things that would cause
1.132 thorpej 293: * mbuf allocation.
294: */
295: s = splvm();
1.75 thorpej 296: if (ipq_locked) {
297: splx(s);
298: return (0);
299: }
300: ipq_locked = 1;
301: splx(s);
302: return (1);
303: }
304:
1.223 perry 305: static inline void
1.211 perry 306: ipq_unlock(void)
1.75 thorpej 307: {
308: int s;
309:
1.132 thorpej 310: s = splvm();
1.75 thorpej 311: ipq_locked = 0;
312: splx(s);
313: }
314:
315: #ifdef DIAGNOSTIC
316: #define IPQ_LOCK() \
317: do { \
318: if (ipq_lock_try() == 0) { \
319: printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \
320: panic("ipq_lock"); \
321: } \
1.159 perry 322: } while (/*CONSTCOND*/ 0)
1.75 thorpej 323: #define IPQ_LOCK_CHECK() \
324: do { \
325: if (ipq_locked == 0) { \
326: printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \
327: panic("ipq lock check"); \
328: } \
1.159 perry 329: } while (/*CONSTCOND*/ 0)
1.75 thorpej 330: #else
331: #define IPQ_LOCK() (void) ipq_lock_try()
332: #define IPQ_LOCK_CHECK() /* nothing */
333: #endif
334:
335: #define IPQ_UNLOCK() ipq_unlock()
1.1 cgd 336:
1.246 ad 337: POOL_INIT(inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl", NULL,
338: IPL_SOFTNET);
339: POOL_INIT(ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", NULL,
340: IPL_VM);
1.72 thorpej 341:
1.135 thorpej 342: #ifdef INET_CSUM_COUNTERS
343: #include <sys/device.h>
344:
345: struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
346: NULL, "inet", "hwcsum bad");
347: struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
348: NULL, "inet", "hwcsum ok");
349: struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
350: NULL, "inet", "swcsum");
351:
352: #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++
353:
1.201 matt 354: EVCNT_ATTACH_STATIC(ip_hwcsum_bad);
355: EVCNT_ATTACH_STATIC(ip_hwcsum_ok);
356: EVCNT_ATTACH_STATIC(ip_swcsum);
357:
1.135 thorpej 358: #else
359:
360: #define INET_CSUM_COUNTER_INCR(ev) /* nothing */
361:
362: #endif /* INET_CSUM_COUNTERS */
363:
1.1 cgd 364: /*
365: * We need to save the IP options in case a protocol wants to respond
366: * to an incoming packet over the same route if the packet got here
367: * using IP source routing. This allows connection establishment and
368: * maintenance when the remote end is on a network that is not known
369: * to us.
370: */
371: int ip_nhops = 0;
372: static struct ip_srcrt {
373: struct in_addr dst; /* final destination */
374: char nop; /* one NOP to align */
375: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
376: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
377: } ip_srcrt;
378:
1.210 perry 379: static void save_rte(u_char *, struct in_addr);
1.35 mycroft 380:
1.164 matt 381: #ifdef MBUFTRACE
1.234 dogcow 382: struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx");
383: struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx");
1.164 matt 384: #endif
385:
1.1 cgd 386: /*
1.194 jonathan 387: * Compute IP limits derived from the value of nmbclusters.
388: */
389: static void
390: ip_nmbclusters_changed(void)
391: {
392: ip_maxfrags = nmbclusters / 4;
393: ip_nmbclusters = nmbclusters;
394: }
395:
396: /*
1.1 cgd 397: * IP initialization: fill in IP protocol switch table.
398: * All protocols not implemented in kernel go to raw IP protocol handler.
399: */
1.8 mycroft 400: void
1.211 perry 401: ip_init(void)
1.1 cgd 402: {
1.199 matt 403: const struct protosw *pr;
1.109 augustss 404: int i;
1.1 cgd 405:
406: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
407: if (pr == 0)
408: panic("ip_init");
409: for (i = 0; i < IPPROTO_MAX; i++)
410: ip_protox[i] = pr - inetsw;
411: for (pr = inetdomain.dom_protosw;
412: pr < inetdomain.dom_protoswNPROTOSW; pr++)
413: if (pr->pr_domain->dom_family == PF_INET &&
414: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
415: ip_protox[pr->pr_protocol] = pr - inetsw;
1.192 jonathan 416:
1.190 jonathan 417: for (i = 0; i < IPREASS_NHASH; i++)
418: LIST_INIT(&ipq[i]);
419:
1.227 kardel 420: ip_id = time_second & 0xfffff;
1.194 jonathan 421:
1.1 cgd 422: ipintrq.ifq_maxlen = ipqmaxlen;
1.194 jonathan 423: ip_nmbclusters_changed();
424:
1.181 jonathan 425: TAILQ_INIT(&in_ifaddrhead);
1.120 ad 426: in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IFADDR,
427: M_WAITOK, &in_ifaddrhash);
1.166 matt 428: in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IPMADDR,
429: M_WAITOK, &in_multihash);
1.160 itojun 430: ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout);
1.73 thorpej 431: #ifdef GATEWAY
1.248 liamjfoy 432: ipflow_init(ip_hashsize);
1.73 thorpej 433: #endif
1.121 thorpej 434:
435: #ifdef PFIL_HOOKS
436: /* Register our Packet Filter hook. */
1.126 thorpej 437: inet_pfil_hook.ph_type = PFIL_TYPE_AF;
438: inet_pfil_hook.ph_af = AF_INET;
1.121 thorpej 439: i = pfil_head_register(&inet_pfil_hook);
440: if (i != 0)
441: printf("ip_init: WARNING: unable to register pfil hook, "
442: "error %d\n", i);
443: #endif /* PFIL_HOOKS */
1.135 thorpej 444:
1.164 matt 445: #ifdef MBUFTRACE
446: MOWNER_ATTACH(&ip_tx_mowner);
447: MOWNER_ATTACH(&ip_rx_mowner);
448: #endif /* MBUFTRACE */
1.1 cgd 449: }
450:
1.229 christos 451: struct sockaddr_in ipaddr = {
452: .sin_len = sizeof(ipaddr),
453: .sin_family = AF_INET,
454: };
1.1 cgd 455: struct route ipforward_rt;
456:
457: /*
1.89 itojun 458: * IP software interrupt routine
459: */
460: void
1.211 perry 461: ipintr(void)
1.89 itojun 462: {
463: int s;
464: struct mbuf *m;
465:
1.241 ad 466: while (!IF_IS_EMPTY(&ipintrq)) {
1.132 thorpej 467: s = splnet();
1.89 itojun 468: IF_DEQUEUE(&ipintrq, m);
469: splx(s);
470: if (m == 0)
471: return;
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;
1.233 tls 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;
1.233 tls 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: */
1.244 christos 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;
1.225 christos 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);
907: /* XXX LIST_FOREACH(fp, &ipq[hash], ipq_q) */
908: for (fp = LIST_FIRST(&ipq[hash]); fp != NULL;
909: fp = LIST_NEXT(fp, ipq_q)) {
1.1 cgd 910: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 911: in_hosteq(ip->ip_src, fp->ipq_src) &&
912: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 913: ip->ip_p == fp->ipq_p)
914: goto found;
1.190 jonathan 915:
916: }
1.1 cgd 917: fp = 0;
918: found:
919:
920: /*
921: * Adjust ip_len to not reflect header,
1.25 cgd 922: * set ipqe_mff if more fragments are expected,
1.1 cgd 923: * convert offset of this to bytes.
924: */
1.155 itojun 925: ip->ip_len = htons(ntohs(ip->ip_len) - hlen);
926: mff = (ip->ip_off & htons(IP_MF)) != 0;
1.25 cgd 927: if (mff) {
1.16 cgd 928: /*
929: * Make sure that fragments have a data length
930: * that's a non-zero multiple of 8 bytes.
931: */
1.155 itojun 932: if (ntohs(ip->ip_len) == 0 ||
933: (ntohs(ip->ip_len) & 0x7) != 0) {
1.16 cgd 934: ipstat.ips_badfrags++;
1.75 thorpej 935: IPQ_UNLOCK();
1.16 cgd 936: goto bad;
937: }
938: }
1.155 itojun 939: ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3);
1.1 cgd 940:
941: /*
942: * If datagram marked as having more fragments
943: * or if this is not the first fragment,
944: * attempt reassembly; if it succeeds, proceed.
945: */
1.155 itojun 946: if (mff || ip->ip_off != htons(0)) {
1.1 cgd 947: ipstat.ips_fragments++;
1.233 tls 948: s = splvm();
1.72 thorpej 949: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.233 tls 950: splx(s);
1.25 cgd 951: if (ipqe == NULL) {
952: ipstat.ips_rcvmemdrop++;
1.75 thorpej 953: IPQ_UNLOCK();
1.25 cgd 954: goto bad;
955: }
956: ipqe->ipqe_mff = mff;
1.50 thorpej 957: ipqe->ipqe_m = m;
1.25 cgd 958: ipqe->ipqe_ip = ip;
1.190 jonathan 959: m = ip_reass(ipqe, fp, &ipq[hash]);
1.75 thorpej 960: if (m == 0) {
961: IPQ_UNLOCK();
1.89 itojun 962: return;
1.75 thorpej 963: }
1.13 mycroft 964: ipstat.ips_reassembled++;
1.50 thorpej 965: ip = mtod(m, struct ip *);
1.74 thorpej 966: hlen = ip->ip_hl << 2;
1.155 itojun 967: ip->ip_len = htons(ntohs(ip->ip_len) + hlen);
1.1 cgd 968: } else
969: if (fp)
970: ip_freef(fp);
1.75 thorpej 971: IPQ_UNLOCK();
1.79 mycroft 972: }
1.128 itojun 973:
1.173 jonathan 974: #if defined(IPSEC)
1.128 itojun 975: /*
976: * enforce IPsec policy checking if we are seeing last header.
977: * note that we do not visit this with protocols with pcb layer
978: * code - like udp/tcp/raw ip.
979: */
980: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 &&
981: ipsec4_in_reject(m, NULL)) {
982: ipsecstat.in_polvio++;
983: goto bad;
984: }
985: #endif
1.226 liamjfoy 986: #ifdef FAST_IPSEC
1.173 jonathan 987: /*
988: * enforce IPsec policy checking if we are seeing last header.
989: * note that we do not visit this with protocols with pcb layer
990: * code - like udp/tcp/raw ip.
991: */
992: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) {
993: /*
994: * Check if the packet has already had IPsec processing
995: * done. If so, then just pass it along. This tag gets
996: * set during AH, ESP, etc. input handling, before the
997: * packet is returned to the ip input queue for delivery.
1.212 perry 998: */
1.173 jonathan 999: mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
1000: s = splsoftnet();
1001: if (mtag != NULL) {
1002: tdbi = (struct tdb_ident *)(mtag + 1);
1003: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
1004: } else {
1005: sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
1.212 perry 1006: IP_FORWARDING, &error);
1.173 jonathan 1007: }
1008: if (sp != NULL) {
1009: /*
1010: * Check security policy against packet attributes.
1011: */
1012: error = ipsec_in_reject(sp, m);
1013: KEY_FREESP(&sp);
1014: } else {
1015: /* XXX error stat??? */
1016: error = EINVAL;
1017: DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/
1018: goto bad;
1019: }
1020: splx(s);
1021: if (error)
1022: goto bad;
1023: }
1024: #endif /* FAST_IPSEC */
1.1 cgd 1025:
1026: /*
1027: * Switch out to protocol's input routine.
1028: */
1.82 aidan 1029: #if IFA_STATS
1.122 itojun 1030: if (ia && ip)
1.155 itojun 1031: ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
1.82 aidan 1032: #endif
1.1 cgd 1033: ipstat.ips_delivered++;
1.89 itojun 1034: {
1035: int off = hlen, nh = ip->ip_p;
1036:
1037: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
1038: return;
1039: }
1.1 cgd 1040: bad:
1041: m_freem(m);
1.135 thorpej 1042: return;
1043:
1044: badcsum:
1045: ipstat.ips_badsum++;
1046: m_freem(m);
1.1 cgd 1047: }
1048:
1049: /*
1050: * Take incoming datagram fragment and try to
1051: * reassemble it into whole datagram. If a chain for
1052: * reassembly of this datagram already exists, then it
1053: * is given as fp; otherwise have to make a chain.
1054: */
1.50 thorpej 1055: struct mbuf *
1.211 perry 1056: ip_reass(struct ipqent *ipqe, struct ipq *fp, struct ipqhead *ipqhead)
1.1 cgd 1057: {
1.109 augustss 1058: struct mbuf *m = ipqe->ipqe_m;
1059: struct ipqent *nq, *p, *q;
1.25 cgd 1060: struct ip *ip;
1.1 cgd 1061: struct mbuf *t;
1.25 cgd 1062: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.233 tls 1063: int i, next, s;
1.1 cgd 1064:
1.75 thorpej 1065: IPQ_LOCK_CHECK();
1066:
1.1 cgd 1067: /*
1068: * Presence of header sizes in mbufs
1069: * would confuse code below.
1070: */
1071: m->m_data += hlen;
1072: m->m_len -= hlen;
1073:
1.194 jonathan 1074: #ifdef notyet
1075: /* make sure fragment limit is up-to-date */
1076: CHECK_NMBCLUSTER_PARAMS();
1077:
1078: /* If we have too many fragments, drop the older half. */
1079: if (ip_nfrags >= ip_maxfrags)
1080: ip_reass_drophalf(void);
1081: #endif
1082:
1.1 cgd 1083: /*
1.192 jonathan 1084: * We are about to add a fragment; increment frag count.
1085: */
1086: ip_nfrags++;
1.212 perry 1087:
1.192 jonathan 1088: /*
1.1 cgd 1089: * If first fragment to arrive, create a reassembly queue.
1090: */
1091: if (fp == 0) {
1.131 itojun 1092: /*
1093: * Enforce upper bound on number of fragmented packets
1094: * for which we attempt reassembly;
1095: * If maxfrag is 0, never accept fragments.
1096: * If maxfrag is -1, accept all fragments without limitation.
1097: */
1098: if (ip_maxfragpackets < 0)
1099: ;
1100: else if (ip_nfragpackets >= ip_maxfragpackets)
1101: goto dropfrag;
1102: ip_nfragpackets++;
1.50 thorpej 1103: MALLOC(fp, struct ipq *, sizeof (struct ipq),
1104: M_FTABLE, M_NOWAIT);
1105: if (fp == NULL)
1.1 cgd 1106: goto dropfrag;
1.190 jonathan 1107: LIST_INSERT_HEAD(ipqhead, fp, ipq_q);
1.192 jonathan 1108: fp->ipq_nfrags = 1;
1.1 cgd 1109: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 1110: fp->ipq_p = ipqe->ipqe_ip->ip_p;
1111: fp->ipq_id = ipqe->ipqe_ip->ip_id;
1.148 matt 1112: TAILQ_INIT(&fp->ipq_fragq);
1.25 cgd 1113: fp->ipq_src = ipqe->ipqe_ip->ip_src;
1114: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
1115: p = NULL;
1.1 cgd 1116: goto insert;
1.192 jonathan 1117: } else {
1118: fp->ipq_nfrags++;
1.1 cgd 1119: }
1120:
1121: /*
1122: * Find a segment which begins after this one does.
1123: */
1.148 matt 1124: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1125: p = q, q = TAILQ_NEXT(q, ipqe_q))
1.155 itojun 1126: if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off))
1.1 cgd 1127: break;
1128:
1129: /*
1130: * If there is a preceding segment, it may provide some of
1131: * our data already. If so, drop the data from the incoming
1132: * segment. If it provides all of our data, drop us.
1133: */
1.25 cgd 1134: if (p != NULL) {
1.155 itojun 1135: i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) -
1136: ntohs(ipqe->ipqe_ip->ip_off);
1.1 cgd 1137: if (i > 0) {
1.155 itojun 1138: if (i >= ntohs(ipqe->ipqe_ip->ip_len))
1.1 cgd 1139: goto dropfrag;
1.50 thorpej 1140: m_adj(ipqe->ipqe_m, i);
1.155 itojun 1141: ipqe->ipqe_ip->ip_off =
1142: htons(ntohs(ipqe->ipqe_ip->ip_off) + i);
1143: ipqe->ipqe_ip->ip_len =
1144: htons(ntohs(ipqe->ipqe_ip->ip_len) - i);
1.1 cgd 1145: }
1146: }
1147:
1148: /*
1149: * While we overlap succeeding segments trim them or,
1150: * if they are completely covered, dequeue them.
1151: */
1.155 itojun 1152: for (; q != NULL &&
1153: ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) >
1154: ntohs(q->ipqe_ip->ip_off); q = nq) {
1155: i = (ntohs(ipqe->ipqe_ip->ip_off) +
1156: ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off);
1157: if (i < ntohs(q->ipqe_ip->ip_len)) {
1158: q->ipqe_ip->ip_len =
1159: htons(ntohs(q->ipqe_ip->ip_len) - i);
1160: q->ipqe_ip->ip_off =
1161: htons(ntohs(q->ipqe_ip->ip_off) + i);
1.50 thorpej 1162: m_adj(q->ipqe_m, i);
1.1 cgd 1163: break;
1164: }
1.148 matt 1165: nq = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1166: m_freem(q->ipqe_m);
1.148 matt 1167: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.233 tls 1168: s = splvm();
1.72 thorpej 1169: pool_put(&ipqent_pool, q);
1.233 tls 1170: splx(s);
1.192 jonathan 1171: fp->ipq_nfrags--;
1172: ip_nfrags--;
1.1 cgd 1173: }
1174:
1175: insert:
1176: /*
1177: * Stick new segment in its place;
1178: * check for complete reassembly.
1179: */
1.25 cgd 1180: if (p == NULL) {
1.148 matt 1181: TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
1.25 cgd 1182: } else {
1.148 matt 1183: TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q);
1.25 cgd 1184: }
1.1 cgd 1185: next = 0;
1.148 matt 1186: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1187: p = q, q = TAILQ_NEXT(q, ipqe_q)) {
1.155 itojun 1188: if (ntohs(q->ipqe_ip->ip_off) != next)
1.1 cgd 1189: return (0);
1.155 itojun 1190: next += ntohs(q->ipqe_ip->ip_len);
1.1 cgd 1191: }
1.25 cgd 1192: if (p->ipqe_mff)
1.1 cgd 1193: return (0);
1194:
1195: /*
1.41 thorpej 1196: * Reassembly is complete. Check for a bogus message size and
1197: * concatenate fragments.
1.1 cgd 1198: */
1.148 matt 1199: q = TAILQ_FIRST(&fp->ipq_fragq);
1.25 cgd 1200: ip = q->ipqe_ip;
1.41 thorpej 1201: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
1202: ipstat.ips_toolong++;
1203: ip_freef(fp);
1204: return (0);
1205: }
1.50 thorpej 1206: m = q->ipqe_m;
1.1 cgd 1207: t = m->m_next;
1208: m->m_next = 0;
1209: m_cat(m, t);
1.148 matt 1210: nq = TAILQ_NEXT(q, ipqe_q);
1.233 tls 1211: s = splvm();
1.72 thorpej 1212: pool_put(&ipqent_pool, q);
1.233 tls 1213: splx(s);
1.25 cgd 1214: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 1215: t = q->ipqe_m;
1.148 matt 1216: nq = TAILQ_NEXT(q, ipqe_q);
1.233 tls 1217: s = splvm();
1.72 thorpej 1218: pool_put(&ipqent_pool, q);
1.233 tls 1219: splx(s);
1.1 cgd 1220: m_cat(m, t);
1221: }
1.192 jonathan 1222: ip_nfrags -= fp->ipq_nfrags;
1.1 cgd 1223:
1224: /*
1225: * Create header for new ip packet by
1226: * modifying header of first packet;
1227: * dequeue and discard fragment reassembly header.
1228: * Make header visible.
1229: */
1.155 itojun 1230: ip->ip_len = htons(next);
1.25 cgd 1231: ip->ip_src = fp->ipq_src;
1232: ip->ip_dst = fp->ipq_dst;
1233: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1234: FREE(fp, M_FTABLE);
1.131 itojun 1235: ip_nfragpackets--;
1.1 cgd 1236: m->m_len += (ip->ip_hl << 2);
1237: m->m_data -= (ip->ip_hl << 2);
1238: /* some debugging cruft by sklower, below, will go away soon */
1239: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
1.109 augustss 1240: int plen = 0;
1.50 thorpej 1241: for (t = m; t; t = t->m_next)
1242: plen += t->m_len;
1243: m->m_pkthdr.len = plen;
1.213 yamt 1244: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1245: }
1.50 thorpej 1246: return (m);
1.1 cgd 1247:
1248: dropfrag:
1.192 jonathan 1249: if (fp != 0)
1250: fp->ipq_nfrags--;
1251: ip_nfrags--;
1.1 cgd 1252: ipstat.ips_fragdropped++;
1253: m_freem(m);
1.233 tls 1254: s = splvm();
1.72 thorpej 1255: pool_put(&ipqent_pool, ipqe);
1.233 tls 1256: splx(s);
1.1 cgd 1257: return (0);
1258: }
1259:
1260: /*
1261: * Free a fragment reassembly header and all
1262: * associated datagrams.
1263: */
1.8 mycroft 1264: void
1.211 perry 1265: ip_freef(struct ipq *fp)
1.1 cgd 1266: {
1.109 augustss 1267: struct ipqent *q, *p;
1.192 jonathan 1268: u_int nfrags = 0;
1.233 tls 1269: int s;
1.1 cgd 1270:
1.75 thorpej 1271: IPQ_LOCK_CHECK();
1272:
1.148 matt 1273: for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) {
1274: p = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1275: m_freem(q->ipqe_m);
1.192 jonathan 1276: nfrags++;
1.148 matt 1277: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.233 tls 1278: s = splvm();
1.72 thorpej 1279: pool_put(&ipqent_pool, q);
1.233 tls 1280: splx(s);
1.1 cgd 1281: }
1.192 jonathan 1282:
1283: if (nfrags != fp->ipq_nfrags)
1284: printf("ip_freef: nfrags %d != %d\n", fp->ipq_nfrags, nfrags);
1285: ip_nfrags -= nfrags;
1.25 cgd 1286: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1287: FREE(fp, M_FTABLE);
1.131 itojun 1288: ip_nfragpackets--;
1.1 cgd 1289: }
1290:
1291: /*
1.194 jonathan 1292: * IP reassembly TTL machinery for multiplicative drop.
1293: */
1294: static u_int fragttl_histo[(IPFRAGTTL+1)];
1295:
1296:
1297: /*
1298: * Decrement TTL of all reasembly queue entries by `ticks'.
1299: * Count number of distinct fragments (as opposed to partial, fragmented
1300: * datagrams) in the reassembly queue. While we traverse the entire
1301: * reassembly queue, compute and return the median TTL over all fragments.
1302: */
1303: static u_int
1304: ip_reass_ttl_decr(u_int ticks)
1305: {
1.198 matt 1306: u_int nfrags, median, dropfraction, keepfraction;
1.194 jonathan 1307: struct ipq *fp, *nfp;
1.198 matt 1308: int i;
1.212 perry 1309:
1.194 jonathan 1310: nfrags = 0;
1311: memset(fragttl_histo, 0, sizeof fragttl_histo);
1.212 perry 1312:
1.194 jonathan 1313: for (i = 0; i < IPREASS_NHASH; i++) {
1314: for (fp = LIST_FIRST(&ipq[i]); fp != NULL; fp = nfp) {
1315: fp->ipq_ttl = ((fp->ipq_ttl <= ticks) ?
1316: 0 : fp->ipq_ttl - ticks);
1317: nfp = LIST_NEXT(fp, ipq_q);
1318: if (fp->ipq_ttl == 0) {
1319: ipstat.ips_fragtimeout++;
1320: ip_freef(fp);
1321: } else {
1322: nfrags += fp->ipq_nfrags;
1323: fragttl_histo[fp->ipq_ttl] += fp->ipq_nfrags;
1324: }
1325: }
1326: }
1327:
1328: KASSERT(ip_nfrags == nfrags);
1329:
1330: /* Find median (or other drop fraction) in histogram. */
1331: dropfraction = (ip_nfrags / 2);
1332: keepfraction = ip_nfrags - dropfraction;
1333: for (i = IPFRAGTTL, median = 0; i >= 0; i--) {
1334: median += fragttl_histo[i];
1335: if (median >= keepfraction)
1336: break;
1337: }
1338:
1339: /* Return TTL of median (or other fraction). */
1340: return (u_int)i;
1341: }
1342:
1343: void
1344: ip_reass_drophalf(void)
1345: {
1346:
1347: u_int median_ticks;
1348: /*
1349: * Compute median TTL of all fragments, and count frags
1350: * with that TTL or lower (roughly half of all fragments).
1351: */
1352: median_ticks = ip_reass_ttl_decr(0);
1353:
1354: /* Drop half. */
1355: median_ticks = ip_reass_ttl_decr(median_ticks);
1356:
1357: }
1358:
1359: /*
1.1 cgd 1360: * IP timer processing;
1361: * if a timer expires on a reassembly
1362: * queue, discard it.
1363: */
1.8 mycroft 1364: void
1.211 perry 1365: ip_slowtimo(void)
1.1 cgd 1366: {
1.191 jonathan 1367: static u_int dropscanidx = 0;
1368: u_int i;
1.194 jonathan 1369: u_int median_ttl;
1.24 mycroft 1370: int s = splsoftnet();
1.1 cgd 1371:
1.75 thorpej 1372: IPQ_LOCK();
1.194 jonathan 1373:
1374: /* Age TTL of all fragments by 1 tick .*/
1375: median_ttl = ip_reass_ttl_decr(1);
1376:
1377: /* make sure fragment limit is up-to-date */
1378: CHECK_NMBCLUSTER_PARAMS();
1379:
1380: /* If we have too many fragments, drop the older half. */
1381: if (ip_nfrags > ip_maxfrags)
1382: ip_reass_ttl_decr(median_ttl);
1383:
1.131 itojun 1384: /*
1.194 jonathan 1385: * If we are over the maximum number of fragmented packets
1.131 itojun 1386: * (due to the limit being lowered), drain off
1.190 jonathan 1387: * enough to get down to the new limit. Start draining
1388: * from the reassembly hashqueue most recently drained.
1.131 itojun 1389: */
1390: if (ip_maxfragpackets < 0)
1391: ;
1392: else {
1.190 jonathan 1393: int wrapped = 0;
1394:
1395: i = dropscanidx;
1396: while (ip_nfragpackets > ip_maxfragpackets && wrapped == 0) {
1397: while (LIST_FIRST(&ipq[i]) != NULL)
1398: ip_freef(LIST_FIRST(&ipq[i]));
1399: if (++i >= IPREASS_NHASH) {
1400: i = 0;
1401: }
1402: /*
1403: * Dont scan forever even if fragment counters are
1404: * wrong: stop after scanning entire reassembly queue.
1405: */
1406: if (i == dropscanidx)
1407: wrapped = 1;
1408: }
1409: dropscanidx = i;
1.131 itojun 1410: }
1.75 thorpej 1411: IPQ_UNLOCK();
1.1 cgd 1412: splx(s);
1413: }
1414:
1415: /*
1416: * Drain off all datagram fragments.
1417: */
1.8 mycroft 1418: void
1.211 perry 1419: ip_drain(void)
1.1 cgd 1420: {
1421:
1.75 thorpej 1422: /*
1423: * We may be called from a device's interrupt context. If
1424: * the ipq is already busy, just bail out now.
1425: */
1426: if (ipq_lock_try() == 0)
1427: return;
1428:
1.194 jonathan 1429: /*
1430: * Drop half the total fragments now. If more mbufs are needed,
1431: * we will be called again soon.
1432: */
1433: ip_reass_drophalf();
1.75 thorpej 1434:
1435: IPQ_UNLOCK();
1.1 cgd 1436: }
1437:
1438: /*
1439: * Do option processing on a datagram,
1440: * possibly discarding it if bad options are encountered,
1441: * or forwarding it if source-routed.
1442: * Returns 1 if packet has been forwarded/freed,
1443: * 0 if the packet should be processed further.
1444: */
1.8 mycroft 1445: int
1.211 perry 1446: ip_dooptions(struct mbuf *m)
1.1 cgd 1447: {
1.109 augustss 1448: struct ip *ip = mtod(m, struct ip *);
1449: u_char *cp, *cp0;
1450: struct ip_timestamp *ipt;
1451: struct in_ifaddr *ia;
1.1 cgd 1452: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 1453: struct in_addr dst;
1.1 cgd 1454: n_time ntime;
1455:
1.13 mycroft 1456: dst = ip->ip_dst;
1.1 cgd 1457: cp = (u_char *)(ip + 1);
1458: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
1459: for (; cnt > 0; cnt -= optlen, cp += optlen) {
1460: opt = cp[IPOPT_OPTVAL];
1461: if (opt == IPOPT_EOL)
1462: break;
1463: if (opt == IPOPT_NOP)
1464: optlen = 1;
1465: else {
1.113 itojun 1466: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1467: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1468: goto bad;
1469: }
1.1 cgd 1470: optlen = cp[IPOPT_OLEN];
1.114 itojun 1471: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 1472: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1473: goto bad;
1474: }
1475: }
1476: switch (opt) {
1477:
1478: default:
1479: break;
1480:
1481: /*
1482: * Source routing with record.
1483: * Find interface with current destination address.
1484: * If none on this machine then drop if strictly routed,
1485: * or do nothing if loosely routed.
1486: * Record interface address and bring up next address
1487: * component. If strictly routed make sure next
1488: * address is on directly accessible net.
1489: */
1490: case IPOPT_LSRR:
1491: case IPOPT_SSRR:
1.47 cjs 1492: if (ip_allowsrcrt == 0) {
1493: type = ICMP_UNREACH;
1494: code = ICMP_UNREACH_NET_PROHIB;
1495: goto bad;
1496: }
1.114 itojun 1497: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1498: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1499: goto bad;
1500: }
1.1 cgd 1501: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1502: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1503: goto bad;
1504: }
1505: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 1506: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 1507: if (ia == 0) {
1508: if (opt == IPOPT_SSRR) {
1509: type = ICMP_UNREACH;
1510: code = ICMP_UNREACH_SRCFAIL;
1511: goto bad;
1512: }
1513: /*
1514: * Loose routing, and not at next destination
1515: * yet; nothing to do except forward.
1516: */
1517: break;
1518: }
1519: off--; /* 0 origin */
1.112 sommerfe 1520: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 1521: /*
1522: * End of source route. Should be for us.
1523: */
1524: save_rte(cp, ip->ip_src);
1525: break;
1526: }
1527: /*
1528: * locate outgoing interface
1529: */
1.244 christos 1530: bcopy((void *)(cp + off), (void *)&ipaddr.sin_addr,
1.1 cgd 1531: sizeof(ipaddr.sin_addr));
1.96 thorpej 1532: if (opt == IPOPT_SSRR)
1.196 itojun 1533: ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr)));
1.96 thorpej 1534: else
1.1 cgd 1535: ia = ip_rtaddr(ipaddr.sin_addr);
1536: if (ia == 0) {
1537: type = ICMP_UNREACH;
1538: code = ICMP_UNREACH_SRCFAIL;
1539: goto bad;
1540: }
1541: ip->ip_dst = ipaddr.sin_addr;
1.244 christos 1542: bcopy((void *)&ia->ia_addr.sin_addr,
1543: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1544: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1545: /*
1546: * Let ip_intr's mcast routing check handle mcast pkts
1547: */
1.18 mycroft 1548: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1549: break;
1550:
1551: case IPOPT_RR:
1.114 itojun 1552: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1553: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1554: goto bad;
1555: }
1.1 cgd 1556: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1557: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1558: goto bad;
1559: }
1560: /*
1561: * If no space remains, ignore.
1562: */
1563: off--; /* 0 origin */
1.112 sommerfe 1564: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1565: break;
1.244 christos 1566: bcopy((void *)(&ip->ip_dst), (void *)&ipaddr.sin_addr,
1.1 cgd 1567: sizeof(ipaddr.sin_addr));
1568: /*
1569: * locate outgoing interface; if we're the destination,
1570: * use the incoming interface (should be same).
1571: */
1.96 thorpej 1572: if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))))
1573: == NULL &&
1574: (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
1.1 cgd 1575: type = ICMP_UNREACH;
1576: code = ICMP_UNREACH_HOST;
1577: goto bad;
1578: }
1.244 christos 1579: bcopy((void *)&ia->ia_addr.sin_addr,
1580: (void *)(cp + off), sizeof(struct in_addr));
1.1 cgd 1581: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1582: break;
1583:
1584: case IPOPT_TS:
1585: code = cp - (u_char *)ip;
1586: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1587: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1588: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1589: goto bad;
1.114 itojun 1590: }
1591: if (ipt->ipt_ptr < 5) {
1592: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1593: goto bad;
1594: }
1.15 cgd 1595: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1596: if (++ipt->ipt_oflw == 0) {
1597: code = (u_char *)&ipt->ipt_ptr -
1598: (u_char *)ip;
1.1 cgd 1599: goto bad;
1.114 itojun 1600: }
1.1 cgd 1601: break;
1602: }
1.104 thorpej 1603: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1604: switch (ipt->ipt_flg) {
1605:
1606: case IPOPT_TS_TSONLY:
1607: break;
1608:
1609: case IPOPT_TS_TSANDADDR:
1.66 thorpej 1610: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1611: sizeof(struct in_addr) > ipt->ipt_len) {
1612: code = (u_char *)&ipt->ipt_ptr -
1613: (u_char *)ip;
1.1 cgd 1614: goto bad;
1.114 itojun 1615: }
1.13 mycroft 1616: ipaddr.sin_addr = dst;
1.96 thorpej 1617: ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr),
1618: m->m_pkthdr.rcvif));
1.13 mycroft 1619: if (ia == 0)
1620: continue;
1.104 thorpej 1621: bcopy(&ia->ia_addr.sin_addr,
1622: cp0, sizeof(struct in_addr));
1.1 cgd 1623: ipt->ipt_ptr += sizeof(struct in_addr);
1624: break;
1625:
1626: case IPOPT_TS_PRESPEC:
1.66 thorpej 1627: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1628: sizeof(struct in_addr) > ipt->ipt_len) {
1629: code = (u_char *)&ipt->ipt_ptr -
1630: (u_char *)ip;
1.1 cgd 1631: goto bad;
1.114 itojun 1632: }
1.104 thorpej 1633: bcopy(cp0, &ipaddr.sin_addr,
1.1 cgd 1634: sizeof(struct in_addr));
1.96 thorpej 1635: if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))
1636: == NULL)
1.1 cgd 1637: continue;
1638: ipt->ipt_ptr += sizeof(struct in_addr);
1639: break;
1640:
1641: default:
1.114 itojun 1642: /* XXX can't take &ipt->ipt_flg */
1643: code = (u_char *)&ipt->ipt_ptr -
1644: (u_char *)ip + 1;
1.1 cgd 1645: goto bad;
1646: }
1647: ntime = iptime();
1.107 thorpej 1648: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1.244 christos 1649: memmove((char *)cp + ipt->ipt_ptr - 1, cp0,
1.1 cgd 1650: sizeof(n_time));
1651: ipt->ipt_ptr += sizeof(n_time);
1652: }
1653: }
1654: if (forward) {
1.26 thorpej 1655: if (ip_forwsrcrt == 0) {
1656: type = ICMP_UNREACH;
1657: code = ICMP_UNREACH_SRCFAIL;
1658: goto bad;
1659: }
1.1 cgd 1660: ip_forward(m, 1);
1661: return (1);
1.13 mycroft 1662: }
1663: return (0);
1.1 cgd 1664: bad:
1.13 mycroft 1665: icmp_error(m, type, code, 0, 0);
1666: ipstat.ips_badoptions++;
1.1 cgd 1667: return (1);
1668: }
1669:
1670: /*
1671: * Given address of next destination (final or next hop),
1672: * return internet address info of interface to be used to get there.
1673: */
1674: struct in_ifaddr *
1.211 perry 1675: ip_rtaddr(struct in_addr dst)
1.1 cgd 1676: {
1.249 ! dyoung 1677: struct rtentry *rt;
! 1678: union {
! 1679: struct sockaddr dst;
! 1680: struct sockaddr_in dst4;
! 1681: } u;
! 1682:
! 1683: sockaddr_in_init(&u.dst4, &dst, 0);
! 1684:
! 1685: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL)
! 1686: return NULL;
! 1687:
! 1688: return ifatoia(rt->rt_ifa);
1.1 cgd 1689: }
1690:
1691: /*
1692: * Save incoming source route for use in replies,
1693: * to be picked up later by ip_srcroute if the receiver is interested.
1694: */
1.13 mycroft 1695: void
1.211 perry 1696: save_rte(u_char *option, struct in_addr dst)
1.1 cgd 1697: {
1698: unsigned olen;
1699:
1700: olen = option[IPOPT_OLEN];
1701: #ifdef DIAGNOSTIC
1702: if (ipprintfs)
1.39 christos 1703: printf("save_rte: olen %d\n", olen);
1.89 itojun 1704: #endif /* 0 */
1.1 cgd 1705: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1706: return;
1.244 christos 1707: bcopy((void *)option, (void *)ip_srcrt.srcopt, olen);
1.1 cgd 1708: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1709: ip_srcrt.dst = dst;
1710: }
1711:
1712: /*
1713: * Retrieve incoming source route for use in replies,
1714: * in the same form used by setsockopt.
1715: * The first hop is placed before the options, will be removed later.
1716: */
1717: struct mbuf *
1.211 perry 1718: ip_srcroute(void)
1.1 cgd 1719: {
1.109 augustss 1720: struct in_addr *p, *q;
1721: struct mbuf *m;
1.1 cgd 1722:
1723: if (ip_nhops == 0)
1.237 dyoung 1724: return NULL;
1.1 cgd 1725: m = m_get(M_DONTWAIT, MT_SOOPTS);
1726: if (m == 0)
1.237 dyoung 1727: return NULL;
1.1 cgd 1728:
1.164 matt 1729: MCLAIM(m, &inetdomain.dom_mowner);
1.13 mycroft 1730: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1731:
1732: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1733: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1734: OPTSIZ;
1735: #ifdef DIAGNOSTIC
1736: if (ipprintfs)
1.39 christos 1737: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1738: #endif
1739:
1740: /*
1741: * First save first hop for return route
1742: */
1743: p = &ip_srcrt.route[ip_nhops - 1];
1744: *(mtod(m, struct in_addr *)) = *p--;
1745: #ifdef DIAGNOSTIC
1746: if (ipprintfs)
1.39 christos 1747: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1748: #endif
1749:
1750: /*
1751: * Copy option fields and padding (nop) to mbuf.
1752: */
1753: ip_srcrt.nop = IPOPT_NOP;
1754: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1.244 christos 1755: memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop,
1756: OPTSIZ);
1757: q = (struct in_addr *)(mtod(m, char *) +
1.1 cgd 1758: sizeof(struct in_addr) + OPTSIZ);
1759: #undef OPTSIZ
1760: /*
1761: * Record return path as an IP source route,
1762: * reversing the path (pointers are now aligned).
1763: */
1764: while (p >= ip_srcrt.route) {
1765: #ifdef DIAGNOSTIC
1766: if (ipprintfs)
1.39 christos 1767: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1768: #endif
1769: *q++ = *p--;
1770: }
1771: /*
1772: * Last hop goes to final destination.
1773: */
1774: *q = ip_srcrt.dst;
1775: #ifdef DIAGNOSTIC
1776: if (ipprintfs)
1.39 christos 1777: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1778: #endif
1779: return (m);
1780: }
1781:
1782: /*
1783: * Strip out IP options, at higher
1784: * level protocol in the kernel.
1785: * Second argument is buffer to which options
1786: * will be moved, and return value is their length.
1787: * XXX should be deleted; last arg currently ignored.
1788: */
1.8 mycroft 1789: void
1.236 christos 1790: ip_stripoptions(struct mbuf *m, struct mbuf *mopt)
1.1 cgd 1791: {
1.109 augustss 1792: int i;
1.1 cgd 1793: struct ip *ip = mtod(m, struct ip *);
1.244 christos 1794: void *opts;
1.1 cgd 1795: int olen;
1796:
1.79 mycroft 1797: olen = (ip->ip_hl << 2) - sizeof (struct ip);
1.244 christos 1798: opts = (void *)(ip + 1);
1.1 cgd 1799: i = m->m_len - (sizeof (struct ip) + olen);
1.244 christos 1800: memmove(opts, (char *)opts + olen, (unsigned)i);
1.1 cgd 1801: m->m_len -= olen;
1802: if (m->m_flags & M_PKTHDR)
1803: m->m_pkthdr.len -= olen;
1.155 itojun 1804: ip->ip_len = htons(ntohs(ip->ip_len) - olen);
1.79 mycroft 1805: ip->ip_hl = sizeof (struct ip) >> 2;
1.1 cgd 1806: }
1807:
1.139 matt 1808: const int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1809: 0, 0, 0, 0,
1810: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1811: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1812: EMSGSIZE, EHOSTUNREACH, 0, 0,
1813: 0, 0, 0, 0,
1814: ENOPROTOOPT
1815: };
1816:
1817: /*
1818: * Forward a packet. If some error occurs return the sender
1819: * an icmp packet. Note we can't always generate a meaningful
1820: * icmp message because icmp doesn't have a large enough repertoire
1821: * of codes and types.
1822: *
1823: * If not forwarding, just drop the packet. This could be confusing
1824: * if ipforwarding was zero but some routing protocol was advancing
1825: * us as a gateway to somewhere. However, we must let the routing
1826: * protocol deal with that.
1827: *
1828: * The srcrt parameter indicates whether the packet is being forwarded
1829: * via a source route.
1830: */
1.13 mycroft 1831: void
1.211 perry 1832: ip_forward(struct mbuf *m, int srcrt)
1.1 cgd 1833: {
1.109 augustss 1834: struct ip *ip = mtod(m, struct ip *);
1835: struct rtentry *rt;
1.220 christos 1836: int error, type = 0, code = 0, destmtu = 0;
1.1 cgd 1837: struct mbuf *mcopy;
1.13 mycroft 1838: n_long dest;
1.249 ! dyoung 1839: union {
! 1840: struct sockaddr dst;
! 1841: struct sockaddr_in dst4;
! 1842: } u;
1.164 matt 1843:
1844: /*
1845: * We are now in the output path.
1846: */
1847: MCLAIM(m, &ip_tx_mowner);
1.135 thorpej 1848:
1849: /*
1850: * Clear any in-bound checksum flags for this packet.
1851: */
1852: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1853:
1.13 mycroft 1854: dest = 0;
1.1 cgd 1855: #ifdef DIAGNOSTIC
1.224 joerg 1856: if (ipprintfs) {
1857: printf("forward: src %s ", inet_ntoa(ip->ip_src));
1858: printf("dst %s ttl %x\n", inet_ntoa(ip->ip_dst), ip->ip_ttl);
1859: }
1.1 cgd 1860: #endif
1.93 sommerfe 1861: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.1 cgd 1862: ipstat.ips_cantforward++;
1863: m_freem(m);
1864: return;
1865: }
1866: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1867: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1868: return;
1869: }
1870:
1.249 ! dyoung 1871: sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
! 1872: if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) {
! 1873: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0);
! 1874: return;
1.1 cgd 1875: }
1876:
1877: /*
1.34 mycroft 1878: * Save at most 68 bytes of the packet in case
1.1 cgd 1879: * we need to generate an ICMP message to the src.
1.119 itojun 1880: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1881: */
1.155 itojun 1882: mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1.119 itojun 1883: if (mcopy)
1884: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1885:
1.221 christos 1886: ip->ip_ttl -= IPTTLDEC;
1887:
1.1 cgd 1888: /*
1889: * If forwarding packet using same interface that it came in on,
1890: * perhaps should send a redirect to sender to shortcut a hop.
1891: * Only send redirect if source is sending directly to us,
1892: * and if packet was not source routed (or has any options).
1893: * Also, don't send redirect if forwarding using a default route
1894: * or a route modified by a redirect.
1895: */
1896: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1897: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1898: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1899: ipsendredirects && !srcrt) {
1.19 mycroft 1900: if (rt->rt_ifa &&
1901: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1902: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1903: if (rt->rt_flags & RTF_GATEWAY)
1904: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1905: else
1906: dest = ip->ip_dst.s_addr;
1907: /*
1908: * Router requirements says to only send host
1909: * redirects.
1910: */
1911: type = ICMP_REDIRECT;
1912: code = ICMP_REDIRECT_HOST;
1.1 cgd 1913: #ifdef DIAGNOSTIC
1.77 thorpej 1914: if (ipprintfs)
1915: printf("redirect (%d) to %x\n", code,
1916: (u_int32_t)dest);
1.1 cgd 1917: #endif
1918: }
1919: }
1920:
1.238 dyoung 1921: error = ip_output(m, NULL, &ipforward_rt,
1.173 jonathan 1922: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
1.174 itojun 1923: (struct ip_moptions *)NULL, (struct socket *)NULL);
1.173 jonathan 1924:
1.1 cgd 1925: if (error)
1926: ipstat.ips_cantforward++;
1927: else {
1928: ipstat.ips_forward++;
1929: if (type)
1930: ipstat.ips_redirectsent++;
1931: else {
1.63 matt 1932: if (mcopy) {
1933: #ifdef GATEWAY
1.64 thorpej 1934: if (mcopy->m_flags & M_CANFASTFWD)
1935: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1936: #endif
1.1 cgd 1937: m_freem(mcopy);
1.63 matt 1938: }
1.1 cgd 1939: return;
1940: }
1941: }
1942: if (mcopy == NULL)
1943: return;
1.13 mycroft 1944:
1.1 cgd 1945: switch (error) {
1946:
1947: case 0: /* forwarded, but need redirect */
1948: /* type, code set above */
1949: break;
1950:
1951: case ENETUNREACH: /* shouldn't happen, checked above */
1952: case EHOSTUNREACH:
1953: case ENETDOWN:
1954: case EHOSTDOWN:
1955: default:
1956: type = ICMP_UNREACH;
1957: code = ICMP_UNREACH_HOST;
1958: break;
1959:
1960: case EMSGSIZE:
1961: type = ICMP_UNREACH;
1962: code = ICMP_UNREACH_NEEDFRAG;
1.173 jonathan 1963: #if !defined(IPSEC) && !defined(FAST_IPSEC)
1.238 dyoung 1964: if (ipforward_rt.ro_rt != NULL)
1.220 christos 1965: destmtu = ipforward_rt.ro_rt->rt_ifp->if_mtu;
1.89 itojun 1966: #else
1967: /*
1968: * If the packet is routed over IPsec tunnel, tell the
1969: * originator the tunnel MTU.
1970: * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1971: * XXX quickhack!!!
1972: */
1.238 dyoung 1973: if (ipforward_rt.ro_rt != NULL) {
1.89 itojun 1974: struct secpolicy *sp;
1975: int ipsecerror;
1.95 itojun 1976: size_t ipsechdr;
1.89 itojun 1977: struct route *ro;
1978:
1979: sp = ipsec4_getpolicybyaddr(mcopy,
1.170 itojun 1980: IPSEC_DIR_OUTBOUND, IP_FORWARDING,
1981: &ipsecerror);
1.89 itojun 1982:
1983: if (sp == NULL)
1.220 christos 1984: destmtu = ipforward_rt.ro_rt->rt_ifp->if_mtu;
1.89 itojun 1985: else {
1986: /* count IPsec header size */
1.95 itojun 1987: ipsechdr = ipsec4_hdrsiz(mcopy,
1.170 itojun 1988: IPSEC_DIR_OUTBOUND, NULL);
1.89 itojun 1989:
1990: /*
1991: * find the correct route for outer IPv4
1992: * header, compute tunnel MTU.
1993: */
1.220 christos 1994:
1.89 itojun 1995: if (sp->req != NULL
1.95 itojun 1996: && sp->req->sav != NULL
1997: && sp->req->sav->sah != NULL) {
1998: ro = &sp->req->sav->sah->sa_route;
1.89 itojun 1999: if (ro->ro_rt && ro->ro_rt->rt_ifp) {
1.220 christos 2000: destmtu =
1.151 itojun 2001: ro->ro_rt->rt_rmx.rmx_mtu ?
2002: ro->ro_rt->rt_rmx.rmx_mtu :
1.89 itojun 2003: ro->ro_rt->rt_ifp->if_mtu;
1.220 christos 2004: destmtu -= ipsechdr;
1.89 itojun 2005: }
2006: }
2007:
1.173 jonathan 2008: #ifdef IPSEC
1.89 itojun 2009: key_freesp(sp);
1.173 jonathan 2010: #else
2011: KEY_FREESP(&sp);
2012: #endif
1.89 itojun 2013: }
2014: }
2015: #endif /*IPSEC*/
1.1 cgd 2016: ipstat.ips_cantfrag++;
2017: break;
2018:
2019: case ENOBUFS:
1.143 itojun 2020: #if 1
2021: /*
2022: * a router should not generate ICMP_SOURCEQUENCH as
2023: * required in RFC1812 Requirements for IP Version 4 Routers.
2024: * source quench could be a big problem under DoS attacks,
1.149 wiz 2025: * or if the underlying interface is rate-limited.
1.143 itojun 2026: */
2027: if (mcopy)
2028: m_freem(mcopy);
2029: return;
2030: #else
1.1 cgd 2031: type = ICMP_SOURCEQUENCH;
2032: code = 0;
2033: break;
1.143 itojun 2034: #endif
1.1 cgd 2035: }
1.220 christos 2036: icmp_error(mcopy, type, code, dest, destmtu);
1.44 thorpej 2037: }
2038:
2039: void
1.211 perry 2040: ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
2041: struct mbuf *m)
1.44 thorpej 2042: {
2043:
2044: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
2045: struct timeval tv;
2046:
2047: microtime(&tv);
1.244 christos 2048: *mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1.44 thorpej 2049: SCM_TIMESTAMP, SOL_SOCKET);
2050: if (*mp)
2051: mp = &(*mp)->m_next;
2052: }
2053: if (inp->inp_flags & INP_RECVDSTADDR) {
1.244 christos 2054: *mp = sbcreatecontrol((void *) &ip->ip_dst,
1.44 thorpej 2055: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
2056: if (*mp)
2057: mp = &(*mp)->m_next;
2058: }
2059: #ifdef notyet
2060: /*
2061: * XXX
2062: * Moving these out of udp_input() made them even more broken
2063: * than they already were.
2064: * - fenner@parc.xerox.com
2065: */
2066: /* options were tossed already */
2067: if (inp->inp_flags & INP_RECVOPTS) {
1.244 christos 2068: *mp = sbcreatecontrol((void *) opts_deleted_above,
1.44 thorpej 2069: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
2070: if (*mp)
2071: mp = &(*mp)->m_next;
2072: }
2073: /* ip_srcroute doesn't do what we want here, need to fix */
2074: if (inp->inp_flags & INP_RECVRETOPTS) {
1.244 christos 2075: *mp = sbcreatecontrol((void *) ip_srcroute(),
1.44 thorpej 2076: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
2077: if (*mp)
2078: mp = &(*mp)->m_next;
2079: }
2080: #endif
2081: if (inp->inp_flags & INP_RECVIF) {
2082: struct sockaddr_dl sdl;
2083:
2084: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
2085: sdl.sdl_family = AF_LINK;
2086: sdl.sdl_index = m->m_pkthdr.rcvif ?
2087: m->m_pkthdr.rcvif->if_index : 0;
2088: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1.244 christos 2089: *mp = sbcreatecontrol((void *) &sdl, sdl.sdl_len,
1.44 thorpej 2090: IP_RECVIF, IPPROTO_IP);
2091: if (*mp)
2092: mp = &(*mp)->m_next;
2093: }
1.13 mycroft 2094: }
2095:
1.189 atatat 2096: /*
1.228 elad 2097: * sysctl helper routine for net.inet.ip.forwsrcrt.
2098: */
2099: static int
2100: sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS)
2101: {
2102: int error, tmp;
2103: struct sysctlnode node;
2104:
2105: node = *rnode;
2106: tmp = ip_forwsrcrt;
2107: node.sysctl_data = &tmp;
2108: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2109: if (error || newp == NULL)
2110: return (error);
2111:
1.230 elad 2112: if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT,
1.232 elad 2113: 0, NULL, NULL, NULL))
1.228 elad 2114: return (EPERM);
2115:
2116: ip_forwsrcrt = tmp;
2117:
2118: return (0);
2119: }
2120:
2121: /*
1.189 atatat 2122: * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the
2123: * range of the new value and tweaks timers if it changes.
2124: */
2125: static int
2126: sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS)
1.13 mycroft 2127: {
1.189 atatat 2128: int error, tmp;
2129: struct sysctlnode node;
2130:
2131: node = *rnode;
2132: tmp = ip_mtudisc_timeout;
2133: node.sysctl_data = &tmp;
2134: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2135: if (error || newp == NULL)
2136: return (error);
2137: if (tmp < 0)
2138: return (EINVAL);
1.52 thorpej 2139:
1.189 atatat 2140: ip_mtudisc_timeout = tmp;
2141: rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout);
2142:
2143: return (0);
2144: }
1.54 lukem 2145:
1.65 matt 2146: #ifdef GATEWAY
1.189 atatat 2147: /*
1.247 liamjfoy 2148: * sysctl helper routine for net.inet.ip.maxflows.
1.189 atatat 2149: */
2150: static int
2151: sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS)
2152: {
2153: int s;
1.67 thorpej 2154:
1.217 atatat 2155: s = sysctl_lookup(SYSCTLFN_CALL(rnode));
1.247 liamjfoy 2156: if (s || newp == NULL)
1.189 atatat 2157: return (s);
1.212 perry 2158:
1.189 atatat 2159: s = splsoftnet();
2160: ipflow_reap(0);
2161: splx(s);
1.144 martin 2162:
1.189 atatat 2163: return (0);
2164: }
1.248 liamjfoy 2165:
2166: static int
2167: sysctl_net_inet_ip_hashsize(SYSCTLFN_ARGS)
2168: {
2169: int error, tmp;
2170: struct sysctlnode node;
2171:
2172: node = *rnode;
2173: tmp = ip_hashsize;
2174: node.sysctl_data = &tmp;
2175: error = sysctl_lookup(SYSCTLFN_CALL(&node));
2176: if (error || newp == NULL)
2177: return (error);
2178:
2179: if ((tmp & (tmp - 1)) == 0 && tmp != 0) {
2180: /*
2181: * Can only fail due to malloc()
2182: */
2183: if (ipflow_invalidate_all(tmp))
2184: return ENOMEM;
2185: } else {
2186: /*
2187: * EINVAL if not a power of 2
2188: */
2189: return EINVAL;
2190: }
2191:
2192: return (0);
2193: }
1.189 atatat 2194: #endif /* GATEWAY */
1.117 tron 2195:
1.131 itojun 2196:
1.189 atatat 2197: SYSCTL_SETUP(sysctl_net_inet_ip_setup, "sysctl net.inet.ip subtree setup")
2198: {
2199: extern int subnetsarelocal, hostzeroisbroadcast;
1.180 jonathan 2200:
1.197 atatat 2201: sysctl_createv(clog, 0, NULL, NULL,
2202: CTLFLAG_PERMANENT,
1.189 atatat 2203: CTLTYPE_NODE, "net", NULL,
2204: NULL, 0, NULL, 0,
2205: CTL_NET, CTL_EOL);
1.197 atatat 2206: sysctl_createv(clog, 0, NULL, NULL,
2207: CTLFLAG_PERMANENT,
1.203 atatat 2208: CTLTYPE_NODE, "inet",
2209: SYSCTL_DESCR("PF_INET related settings"),
1.189 atatat 2210: NULL, 0, NULL, 0,
2211: CTL_NET, PF_INET, CTL_EOL);
1.197 atatat 2212: sysctl_createv(clog, 0, NULL, NULL,
2213: CTLFLAG_PERMANENT,
1.203 atatat 2214: CTLTYPE_NODE, "ip",
2215: SYSCTL_DESCR("IPv4 related settings"),
1.189 atatat 2216: NULL, 0, NULL, 0,
2217: CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL);
1.212 perry 2218:
1.197 atatat 2219: sysctl_createv(clog, 0, NULL, NULL,
2220: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2221: CTLTYPE_INT, "forwarding",
2222: SYSCTL_DESCR("Enable forwarding of INET datagrams"),
1.189 atatat 2223: NULL, 0, &ipforwarding, 0,
2224: CTL_NET, PF_INET, IPPROTO_IP,
2225: IPCTL_FORWARDING, CTL_EOL);
1.197 atatat 2226: sysctl_createv(clog, 0, NULL, NULL,
2227: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2228: CTLTYPE_INT, "redirect",
2229: SYSCTL_DESCR("Enable sending of ICMP redirect messages"),
1.189 atatat 2230: NULL, 0, &ipsendredirects, 0,
2231: CTL_NET, PF_INET, IPPROTO_IP,
2232: IPCTL_SENDREDIRECTS, CTL_EOL);
1.197 atatat 2233: sysctl_createv(clog, 0, NULL, NULL,
2234: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2235: CTLTYPE_INT, "ttl",
2236: SYSCTL_DESCR("Default TTL for an INET datagram"),
1.189 atatat 2237: NULL, 0, &ip_defttl, 0,
2238: CTL_NET, PF_INET, IPPROTO_IP,
2239: IPCTL_DEFTTL, CTL_EOL);
2240: #ifdef IPCTL_DEFMTU
1.197 atatat 2241: sysctl_createv(clog, 0, NULL, NULL,
2242: CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */,
1.203 atatat 2243: CTLTYPE_INT, "mtu",
2244: SYSCTL_DESCR("Default MTA for an INET route"),
1.189 atatat 2245: NULL, 0, &ip_mtu, 0,
2246: CTL_NET, PF_INET, IPPROTO_IP,
2247: IPCTL_DEFMTU, CTL_EOL);
2248: #endif /* IPCTL_DEFMTU */
1.197 atatat 2249: sysctl_createv(clog, 0, NULL, NULL,
1.228 elad 2250: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2251: CTLTYPE_INT, "forwsrcrt",
2252: SYSCTL_DESCR("Enable forwarding of source-routed "
2253: "datagrams"),
1.228 elad 2254: sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0,
1.189 atatat 2255: CTL_NET, PF_INET, IPPROTO_IP,
2256: IPCTL_FORWSRCRT, CTL_EOL);
1.197 atatat 2257: sysctl_createv(clog, 0, NULL, NULL,
2258: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2259: CTLTYPE_INT, "directed-broadcast",
2260: SYSCTL_DESCR("Enable forwarding of broadcast datagrams"),
1.189 atatat 2261: NULL, 0, &ip_directedbcast, 0,
2262: CTL_NET, PF_INET, IPPROTO_IP,
2263: IPCTL_DIRECTEDBCAST, CTL_EOL);
1.197 atatat 2264: sysctl_createv(clog, 0, NULL, NULL,
2265: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2266: CTLTYPE_INT, "allowsrcrt",
2267: SYSCTL_DESCR("Accept source-routed datagrams"),
1.189 atatat 2268: NULL, 0, &ip_allowsrcrt, 0,
2269: CTL_NET, PF_INET, IPPROTO_IP,
2270: IPCTL_ALLOWSRCRT, CTL_EOL);
1.197 atatat 2271: sysctl_createv(clog, 0, NULL, NULL,
2272: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2273: CTLTYPE_INT, "subnetsarelocal",
2274: SYSCTL_DESCR("Whether logical subnets are considered "
2275: "local"),
1.189 atatat 2276: NULL, 0, &subnetsarelocal, 0,
2277: CTL_NET, PF_INET, IPPROTO_IP,
2278: IPCTL_SUBNETSARELOCAL, CTL_EOL);
1.197 atatat 2279: sysctl_createv(clog, 0, NULL, NULL,
2280: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2281: CTLTYPE_INT, "mtudisc",
2282: SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"),
1.189 atatat 2283: NULL, 0, &ip_mtudisc, 0,
2284: CTL_NET, PF_INET, IPPROTO_IP,
2285: IPCTL_MTUDISC, CTL_EOL);
1.197 atatat 2286: sysctl_createv(clog, 0, NULL, NULL,
2287: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2288: CTLTYPE_INT, "anonportmin",
2289: SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1.189 atatat 2290: sysctl_net_inet_ip_ports, 0, &anonportmin, 0,
2291: CTL_NET, PF_INET, IPPROTO_IP,
2292: IPCTL_ANONPORTMIN, CTL_EOL);
1.197 atatat 2293: sysctl_createv(clog, 0, NULL, NULL,
2294: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2295: CTLTYPE_INT, "anonportmax",
2296: SYSCTL_DESCR("Highest ephemeral port number to assign"),
1.189 atatat 2297: sysctl_net_inet_ip_ports, 0, &anonportmax, 0,
2298: CTL_NET, PF_INET, IPPROTO_IP,
2299: IPCTL_ANONPORTMAX, CTL_EOL);
1.197 atatat 2300: sysctl_createv(clog, 0, NULL, NULL,
2301: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2302: CTLTYPE_INT, "mtudisctimeout",
2303: SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"),
1.189 atatat 2304: sysctl_net_inet_ip_pmtudto, 0, &ip_mtudisc_timeout, 0,
2305: CTL_NET, PF_INET, IPPROTO_IP,
2306: IPCTL_MTUDISCTIMEOUT, CTL_EOL);
2307: #ifdef GATEWAY
1.197 atatat 2308: sysctl_createv(clog, 0, NULL, NULL,
2309: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2310: CTLTYPE_INT, "maxflows",
2311: SYSCTL_DESCR("Number of flows for fast forwarding"),
1.189 atatat 2312: sysctl_net_inet_ip_maxflows, 0, &ip_maxflows, 0,
2313: CTL_NET, PF_INET, IPPROTO_IP,
2314: IPCTL_MAXFLOWS, CTL_EOL);
1.248 liamjfoy 2315: sysctl_createv(clog, 0, NULL, NULL,
2316: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2317: CTLTYPE_INT, "hashsize",
2318: SYSCTL_DESCR("Size of hash table for fast forwarding (IPv4)"),
2319: sysctl_net_inet_ip_hashsize, 0, &ip_hashsize, 0,
2320: CTL_NET, PF_INET, IPPROTO_IP,
2321: CTL_CREATE, CTL_EOL);
1.189 atatat 2322: #endif /* GATEWAY */
1.197 atatat 2323: sysctl_createv(clog, 0, NULL, NULL,
2324: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2325: CTLTYPE_INT, "hostzerobroadcast",
2326: SYSCTL_DESCR("All zeroes address is broadcast address"),
1.189 atatat 2327: NULL, 0, &hostzeroisbroadcast, 0,
2328: CTL_NET, PF_INET, IPPROTO_IP,
2329: IPCTL_HOSTZEROBROADCAST, CTL_EOL);
2330: #if NGIF > 0
1.197 atatat 2331: sysctl_createv(clog, 0, NULL, NULL,
2332: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2333: CTLTYPE_INT, "gifttl",
2334: SYSCTL_DESCR("Default TTL for a gif tunnel datagram"),
1.189 atatat 2335: NULL, 0, &ip_gif_ttl, 0,
2336: CTL_NET, PF_INET, IPPROTO_IP,
2337: IPCTL_GIF_TTL, CTL_EOL);
2338: #endif /* NGIF */
2339: #ifndef IPNOPRIVPORTS
1.197 atatat 2340: sysctl_createv(clog, 0, NULL, NULL,
2341: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2342: CTLTYPE_INT, "lowportmin",
2343: SYSCTL_DESCR("Lowest privileged ephemeral port number "
2344: "to assign"),
1.189 atatat 2345: sysctl_net_inet_ip_ports, 0, &lowportmin, 0,
2346: CTL_NET, PF_INET, IPPROTO_IP,
2347: IPCTL_LOWPORTMIN, CTL_EOL);
1.197 atatat 2348: sysctl_createv(clog, 0, NULL, NULL,
2349: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2350: CTLTYPE_INT, "lowportmax",
2351: SYSCTL_DESCR("Highest privileged ephemeral port number "
2352: "to assign"),
1.189 atatat 2353: sysctl_net_inet_ip_ports, 0, &lowportmax, 0,
2354: CTL_NET, PF_INET, IPPROTO_IP,
2355: IPCTL_LOWPORTMAX, CTL_EOL);
2356: #endif /* IPNOPRIVPORTS */
1.197 atatat 2357: sysctl_createv(clog, 0, NULL, NULL,
2358: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2359: CTLTYPE_INT, "maxfragpackets",
2360: SYSCTL_DESCR("Maximum number of fragments to retain for "
2361: "possible reassembly"),
1.189 atatat 2362: NULL, 0, &ip_maxfragpackets, 0,
2363: CTL_NET, PF_INET, IPPROTO_IP,
2364: IPCTL_MAXFRAGPACKETS, CTL_EOL);
2365: #if NGRE > 0
1.197 atatat 2366: sysctl_createv(clog, 0, NULL, NULL,
2367: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2368: CTLTYPE_INT, "grettl",
2369: SYSCTL_DESCR("Default TTL for a gre tunnel datagram"),
1.189 atatat 2370: NULL, 0, &ip_gre_ttl, 0,
2371: CTL_NET, PF_INET, IPPROTO_IP,
2372: IPCTL_GRE_TTL, CTL_EOL);
2373: #endif /* NGRE */
1.197 atatat 2374: sysctl_createv(clog, 0, NULL, NULL,
2375: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2376: CTLTYPE_INT, "checkinterface",
2377: SYSCTL_DESCR("Enable receive side of Strong ES model "
2378: "from RFC1122"),
1.189 atatat 2379: NULL, 0, &ip_checkinterface, 0,
2380: CTL_NET, PF_INET, IPPROTO_IP,
2381: IPCTL_CHECKINTERFACE, CTL_EOL);
1.197 atatat 2382: sysctl_createv(clog, 0, NULL, NULL,
2383: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 2384: CTLTYPE_INT, "random_id",
2385: SYSCTL_DESCR("Assign random ip_id values"),
1.189 atatat 2386: NULL, 0, &ip_do_randomid, 0,
2387: CTL_NET, PF_INET, IPPROTO_IP,
2388: IPCTL_RANDOMID, CTL_EOL);
1.206 thorpej 2389: sysctl_createv(clog, 0, NULL, NULL,
2390: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2391: CTLTYPE_INT, "do_loopback_cksum",
2392: SYSCTL_DESCR("Perform IP checksum on loopback"),
2393: NULL, 0, &ip_do_loopback_cksum, 0,
2394: CTL_NET, PF_INET, IPPROTO_IP,
2395: IPCTL_LOOPBACKCKSUM, CTL_EOL);
1.219 elad 2396: sysctl_createv(clog, 0, NULL, NULL,
2397: CTLFLAG_PERMANENT,
2398: CTLTYPE_STRUCT, "stats",
2399: SYSCTL_DESCR("IP statistics"),
2400: NULL, 0, &ipstat, sizeof(ipstat),
2401: CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS,
2402: CTL_EOL);
1.1 cgd 2403: }
CVSweb <webmaster@jp.NetBSD.org>