Annotation of src/sys/netinet/ip_input.c, Revision 1.170
1.170 ! itojun 1: /* $NetBSD: ip_input.c,v 1.169 2003/06/30 07:54:28 itojun 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.
81: * 3. All advertising materials mentioning features or use of this software
82: * must display the following acknowledgement:
83: * This product includes software developed by the University of
84: * California, Berkeley and its contributors.
85: * 4. Neither the name of the University nor the names of its contributors
86: * may be used to endorse or promote products derived from this software
87: * without specific prior written permission.
88: *
89: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
90: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
91: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
92: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
93: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
94: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
95: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
96: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
97: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
98: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
99: * SUCH DAMAGE.
100: *
1.14 cgd 101: * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
1.1 cgd 102: */
1.141 lukem 103:
104: #include <sys/cdefs.h>
1.170 ! itojun 105: __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.169 2003/06/30 07:54:28 itojun Exp $");
1.55 scottr 106:
1.62 matt 107: #include "opt_gateway.h"
1.69 mrg 108: #include "opt_pfil_hooks.h"
1.91 thorpej 109: #include "opt_ipsec.h"
1.55 scottr 110: #include "opt_mrouting.h"
1.167 martin 111: #include "opt_mbuftrace.h"
1.135 thorpej 112: #include "opt_inet_csum.h"
1.1 cgd 113:
1.5 mycroft 114: #include <sys/param.h>
115: #include <sys/systm.h>
116: #include <sys/malloc.h>
117: #include <sys/mbuf.h>
118: #include <sys/domain.h>
119: #include <sys/protosw.h>
120: #include <sys/socket.h>
1.44 thorpej 121: #include <sys/socketvar.h>
1.5 mycroft 122: #include <sys/errno.h>
123: #include <sys/time.h>
124: #include <sys/kernel.h>
1.72 thorpej 125: #include <sys/pool.h>
1.28 christos 126: #include <sys/sysctl.h>
1.1 cgd 127:
1.5 mycroft 128: #include <net/if.h>
1.44 thorpej 129: #include <net/if_dl.h>
1.5 mycroft 130: #include <net/route.h>
1.45 mrg 131: #include <net/pfil.h>
1.1 cgd 132:
1.5 mycroft 133: #include <netinet/in.h>
134: #include <netinet/in_systm.h>
135: #include <netinet/ip.h>
136: #include <netinet/in_pcb.h>
137: #include <netinet/in_var.h>
138: #include <netinet/ip_var.h>
139: #include <netinet/ip_icmp.h>
1.89 itojun 140: /* just for gif_ttl */
141: #include <netinet/in_gif.h>
142: #include "gif.h"
1.144 martin 143: #include <net/if_gre.h>
144: #include "gre.h"
1.111 jdolecek 145:
146: #ifdef MROUTING
147: #include <netinet/ip_mroute.h>
148: #endif
1.89 itojun 149:
150: #ifdef IPSEC
151: #include <netinet6/ipsec.h>
152: #include <netkey/key.h>
153: #endif
1.44 thorpej 154:
1.1 cgd 155: #ifndef IPFORWARDING
156: #ifdef GATEWAY
157: #define IPFORWARDING 1 /* forward IP packets not for us */
158: #else /* GATEWAY */
159: #define IPFORWARDING 0 /* don't forward IP packets not for us */
160: #endif /* GATEWAY */
161: #endif /* IPFORWARDING */
162: #ifndef IPSENDREDIRECTS
163: #define IPSENDREDIRECTS 1
164: #endif
1.26 thorpej 165: #ifndef IPFORWSRCRT
1.47 cjs 166: #define IPFORWSRCRT 1 /* forward source-routed packets */
167: #endif
168: #ifndef IPALLOWSRCRT
1.48 mrg 169: #define IPALLOWSRCRT 1 /* allow source-routed packets */
1.26 thorpej 170: #endif
1.53 kml 171: #ifndef IPMTUDISC
1.153 itojun 172: #define IPMTUDISC 1
1.53 kml 173: #endif
1.60 kml 174: #ifndef IPMTUDISCTIMEOUT
1.61 kml 175: #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */
1.60 kml 176: #endif
1.53 kml 177:
1.27 thorpej 178: /*
179: * Note: DIRECTED_BROADCAST is handled this way so that previous
180: * configuration using this option will Just Work.
181: */
182: #ifndef IPDIRECTEDBCAST
183: #ifdef DIRECTED_BROADCAST
184: #define IPDIRECTEDBCAST 1
185: #else
186: #define IPDIRECTEDBCAST 0
187: #endif /* DIRECTED_BROADCAST */
188: #endif /* IPDIRECTEDBCAST */
1.1 cgd 189: int ipforwarding = IPFORWARDING;
190: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 191: int ip_defttl = IPDEFTTL;
1.26 thorpej 192: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 193: int ip_directedbcast = IPDIRECTEDBCAST;
1.47 cjs 194: int ip_allowsrcrt = IPALLOWSRCRT;
1.53 kml 195: int ip_mtudisc = IPMTUDISC;
1.156 itojun 196: int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
1.1 cgd 197: #ifdef DIAGNOSTIC
198: int ipprintfs = 0;
199: #endif
1.165 christos 200: /*
201: * XXX - Setting ip_checkinterface mostly implements the receive side of
202: * the Strong ES model described in RFC 1122, but since the routing table
203: * and transmit implementation do not implement the Strong ES model,
204: * setting this to 1 results in an odd hybrid.
205: *
206: * XXX - ip_checkinterface currently must be disabled if you use ipnat
207: * to translate the destination address to another local interface.
208: *
209: * XXX - ip_checkinterface must be disabled if you add IP aliases
210: * to the loopback interface instead of the interface where the
211: * packets for those addresses are received.
212: */
213: int ip_checkinterface = 0;
214:
1.1 cgd 215:
1.60 kml 216: struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
217:
1.1 cgd 218: extern struct domain inetdomain;
219: int ipqmaxlen = IFQ_MAXLEN;
1.150 matt 220: u_long in_ifaddrhash; /* size of hash table - 1 */
221: int in_ifaddrentries; /* total number of addrs */
1.22 mycroft 222: struct in_ifaddrhead in_ifaddr;
1.57 tls 223: struct in_ifaddrhashhead *in_ifaddrhashtbl;
1.166 matt 224: u_long in_multihash; /* size of hash table - 1 */
225: int in_multientries; /* total number of addrs */
226: struct in_multihead in_multi;
227: struct in_multihashhead *in_multihashtbl;
1.13 mycroft 228: struct ifqueue ipintrq;
1.63 matt 229: struct ipstat ipstat;
230: u_int16_t ip_id;
1.75 thorpej 231:
1.121 thorpej 232: #ifdef PFIL_HOOKS
233: struct pfil_head inet_pfil_hook;
234: #endif
235:
1.63 matt 236: struct ipqhead ipq;
1.75 thorpej 237: int ipq_locked;
1.131 itojun 238: int ip_nfragpackets = 0;
1.133 itojun 239: int ip_maxfragpackets = 200;
1.75 thorpej 240:
241: static __inline int ipq_lock_try __P((void));
242: static __inline void ipq_unlock __P((void));
243:
244: static __inline int
245: ipq_lock_try()
246: {
247: int s;
248:
1.132 thorpej 249: /*
1.149 wiz 250: * Use splvm() -- we're blocking things that would cause
1.132 thorpej 251: * mbuf allocation.
252: */
253: s = splvm();
1.75 thorpej 254: if (ipq_locked) {
255: splx(s);
256: return (0);
257: }
258: ipq_locked = 1;
259: splx(s);
260: return (1);
261: }
262:
263: static __inline void
264: ipq_unlock()
265: {
266: int s;
267:
1.132 thorpej 268: s = splvm();
1.75 thorpej 269: ipq_locked = 0;
270: splx(s);
271: }
272:
273: #ifdef DIAGNOSTIC
274: #define IPQ_LOCK() \
275: do { \
276: if (ipq_lock_try() == 0) { \
277: printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \
278: panic("ipq_lock"); \
279: } \
1.159 perry 280: } while (/*CONSTCOND*/ 0)
1.75 thorpej 281: #define IPQ_LOCK_CHECK() \
282: do { \
283: if (ipq_locked == 0) { \
284: printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \
285: panic("ipq lock check"); \
286: } \
1.159 perry 287: } while (/*CONSTCOND*/ 0)
1.75 thorpej 288: #else
289: #define IPQ_LOCK() (void) ipq_lock_try()
290: #define IPQ_LOCK_CHECK() /* nothing */
291: #endif
292:
293: #define IPQ_UNLOCK() ipq_unlock()
1.1 cgd 294:
1.166 matt 295: struct pool inmulti_pool;
1.72 thorpej 296: struct pool ipqent_pool;
297:
1.135 thorpej 298: #ifdef INET_CSUM_COUNTERS
299: #include <sys/device.h>
300:
301: struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
302: NULL, "inet", "hwcsum bad");
303: struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
304: NULL, "inet", "hwcsum ok");
305: struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
306: NULL, "inet", "swcsum");
307:
308: #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++
309:
310: #else
311:
312: #define INET_CSUM_COUNTER_INCR(ev) /* nothing */
313:
314: #endif /* INET_CSUM_COUNTERS */
315:
1.1 cgd 316: /*
317: * We need to save the IP options in case a protocol wants to respond
318: * to an incoming packet over the same route if the packet got here
319: * using IP source routing. This allows connection establishment and
320: * maintenance when the remote end is on a network that is not known
321: * to us.
322: */
323: int ip_nhops = 0;
324: static struct ip_srcrt {
325: struct in_addr dst; /* final destination */
326: char nop; /* one NOP to align */
327: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
328: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
329: } ip_srcrt;
330:
1.13 mycroft 331: static void save_rte __P((u_char *, struct in_addr));
1.35 mycroft 332:
1.164 matt 333: #ifdef MBUFTRACE
334: struct mowner ip_rx_mowner = { "internet", "rx" };
335: struct mowner ip_tx_mowner = { "internet", "tx" };
336: #endif
337:
1.1 cgd 338: /*
339: * IP initialization: fill in IP protocol switch table.
340: * All protocols not implemented in kernel go to raw IP protocol handler.
341: */
1.8 mycroft 342: void
1.1 cgd 343: ip_init()
344: {
1.109 augustss 345: struct protosw *pr;
346: int i;
1.1 cgd 347:
1.166 matt 348: pool_init(&inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl",
349: NULL);
1.72 thorpej 350: pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl",
1.146 thorpej 351: NULL);
1.72 thorpej 352:
1.1 cgd 353: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
354: if (pr == 0)
355: panic("ip_init");
356: for (i = 0; i < IPPROTO_MAX; i++)
357: ip_protox[i] = pr - inetsw;
358: for (pr = inetdomain.dom_protosw;
359: pr < inetdomain.dom_protoswNPROTOSW; pr++)
360: if (pr->pr_domain->dom_family == PF_INET &&
361: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
362: ip_protox[pr->pr_protocol] = pr - inetsw;
1.25 cgd 363: LIST_INIT(&ipq);
1.1 cgd 364: ip_id = time.tv_sec & 0xffff;
365: ipintrq.ifq_maxlen = ipqmaxlen;
1.22 mycroft 366: TAILQ_INIT(&in_ifaddr);
1.120 ad 367: in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IFADDR,
368: M_WAITOK, &in_ifaddrhash);
1.166 matt 369: in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IPMADDR,
370: M_WAITOK, &in_multihash);
1.160 itojun 371: ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout);
1.73 thorpej 372: #ifdef GATEWAY
373: ipflow_init();
374: #endif
1.121 thorpej 375:
376: #ifdef PFIL_HOOKS
377: /* Register our Packet Filter hook. */
1.126 thorpej 378: inet_pfil_hook.ph_type = PFIL_TYPE_AF;
379: inet_pfil_hook.ph_af = AF_INET;
1.121 thorpej 380: i = pfil_head_register(&inet_pfil_hook);
381: if (i != 0)
382: printf("ip_init: WARNING: unable to register pfil hook, "
383: "error %d\n", i);
384: #endif /* PFIL_HOOKS */
1.135 thorpej 385:
386: #ifdef INET_CSUM_COUNTERS
387: evcnt_attach_static(&ip_hwcsum_bad);
388: evcnt_attach_static(&ip_hwcsum_ok);
389: evcnt_attach_static(&ip_swcsum);
390: #endif /* INET_CSUM_COUNTERS */
1.164 matt 391:
392: #ifdef MBUFTRACE
393: MOWNER_ATTACH(&ip_tx_mowner);
394: MOWNER_ATTACH(&ip_rx_mowner);
395: #endif /* MBUFTRACE */
1.1 cgd 396: }
397:
398: struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
399: struct route ipforward_rt;
400:
401: /*
1.89 itojun 402: * IP software interrupt routine
403: */
404: void
405: ipintr()
406: {
407: int s;
408: struct mbuf *m;
409:
410: while (1) {
1.132 thorpej 411: s = splnet();
1.89 itojun 412: IF_DEQUEUE(&ipintrq, m);
413: splx(s);
414: if (m == 0)
415: return;
1.164 matt 416: MCLAIM(m, &ip_rx_mowner);
1.89 itojun 417: ip_input(m);
418: }
419: }
420:
421: /*
1.1 cgd 422: * Ip input routine. Checksum and byte swap header. If fragmented
423: * try to reassemble. Process options. Pass to next level.
424: */
1.8 mycroft 425: void
1.89 itojun 426: ip_input(struct mbuf *m)
1.1 cgd 427: {
1.109 augustss 428: struct ip *ip = NULL;
429: struct ipq *fp;
430: struct in_ifaddr *ia;
431: struct ifaddr *ifa;
1.25 cgd 432: struct ipqent *ipqe;
1.89 itojun 433: int hlen = 0, mff, len;
1.100 itojun 434: int downmatch;
1.165 christos 435: int checkif;
1.169 itojun 436: int srcrt = 0;
1.1 cgd 437:
1.164 matt 438: MCLAIM(m, &ip_rx_mowner);
1.1 cgd 439: #ifdef DIAGNOSTIC
440: if ((m->m_flags & M_PKTHDR) == 0)
441: panic("ipintr no HDR");
442: #endif
1.89 itojun 443: #ifdef IPSEC
444: /*
445: * should the inner packet be considered authentic?
446: * see comment in ah4_input().
447: */
448: if (m) {
449: m->m_flags &= ~M_AUTHIPHDR;
450: m->m_flags &= ~M_AUTHIPDGM;
451: }
452: #endif
1.164 matt 453:
1.1 cgd 454: /*
455: * If no IP addresses have been set yet but the interfaces
456: * are receiving, can't do anything with incoming packets yet.
457: */
1.140 matt 458: if (TAILQ_FIRST(&in_ifaddr) == 0)
1.1 cgd 459: goto bad;
460: ipstat.ips_total++;
1.154 thorpej 461: /*
462: * If the IP header is not aligned, slurp it up into a new
463: * mbuf with space for link headers, in the event we forward
464: * it. Otherwise, if it is aligned, make sure the entire
465: * base IP header is in the first mbuf of the chain.
466: */
467: if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
468: if ((m = m_copyup(m, sizeof(struct ip),
469: (max_linkhdr + 3) & ~3)) == NULL) {
470: /* XXXJRT new stat, please */
471: ipstat.ips_toosmall++;
472: return;
473: }
474: } else if (__predict_false(m->m_len < sizeof (struct ip))) {
475: if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
476: ipstat.ips_toosmall++;
477: return;
478: }
1.1 cgd 479: }
480: ip = mtod(m, struct ip *);
1.13 mycroft 481: if (ip->ip_v != IPVERSION) {
482: ipstat.ips_badvers++;
483: goto bad;
484: }
1.1 cgd 485: hlen = ip->ip_hl << 2;
486: if (hlen < sizeof(struct ip)) { /* minimum header length */
487: ipstat.ips_badhlen++;
488: goto bad;
489: }
490: if (hlen > m->m_len) {
491: if ((m = m_pullup(m, hlen)) == 0) {
492: ipstat.ips_badhlen++;
1.89 itojun 493: return;
1.1 cgd 494: }
495: ip = mtod(m, struct ip *);
496: }
1.98 thorpej 497:
1.85 hwr 498: /*
1.99 thorpej 499: * RFC1122: packets with a multicast source address are
1.98 thorpej 500: * not allowed.
1.85 hwr 501: */
502: if (IN_MULTICAST(ip->ip_src.s_addr)) {
1.130 itojun 503: ipstat.ips_badaddr++;
1.85 hwr 504: goto bad;
1.129 itojun 505: }
506:
507: /* 127/8 must not appear on wire - RFC1122 */
508: if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
509: (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
1.130 itojun 510: if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
511: ipstat.ips_badaddr++;
1.129 itojun 512: goto bad;
1.130 itojun 513: }
1.85 hwr 514: }
515:
1.135 thorpej 516: switch (m->m_pkthdr.csum_flags &
1.137 thorpej 517: ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) |
1.135 thorpej 518: M_CSUM_IPv4_BAD)) {
519: case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
520: INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
521: goto badcsum;
522:
523: case M_CSUM_IPv4:
524: /* Checksum was okay. */
525: INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
526: break;
527:
528: default:
529: /* Must compute it ourselves. */
530: INET_CSUM_COUNTER_INCR(&ip_swcsum);
531: if (in_cksum(m, hlen) != 0)
532: goto bad;
533: break;
1.1 cgd 534: }
535:
1.121 thorpej 536: /* Retrieve the packet length. */
537: len = ntohs(ip->ip_len);
1.81 proff 538:
539: /*
540: * Check for additional length bogosity
541: */
1.84 proff 542: if (len < hlen) {
1.81 proff 543: ipstat.ips_badlen++;
544: goto bad;
545: }
1.1 cgd 546:
547: /*
548: * Check that the amount of data in the buffers
549: * is as at least much as the IP header would have us expect.
550: * Trim mbufs if longer than we expect.
551: * Drop packet if shorter than we expect.
552: */
1.35 mycroft 553: if (m->m_pkthdr.len < len) {
1.1 cgd 554: ipstat.ips_tooshort++;
555: goto bad;
556: }
1.35 mycroft 557: if (m->m_pkthdr.len > len) {
1.1 cgd 558: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 559: m->m_len = len;
560: m->m_pkthdr.len = len;
1.1 cgd 561: } else
1.35 mycroft 562: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 563: }
564:
1.94 itojun 565: #ifdef IPSEC
1.149 wiz 566: /* ipflow (IP fast forwarding) is not compatible with IPsec. */
1.94 itojun 567: m->m_flags &= ~M_CANFASTFWD;
568: #else
1.64 thorpej 569: /*
570: * Assume that we can create a fast-forward IP flow entry
571: * based on this packet.
572: */
573: m->m_flags |= M_CANFASTFWD;
1.94 itojun 574: #endif
1.64 thorpej 575:
1.36 mrg 576: #ifdef PFIL_HOOKS
1.33 mrg 577: /*
1.64 thorpej 578: * Run through list of hooks for input packets. If there are any
579: * filters which require that additional packets in the flow are
580: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
581: * Note that filters must _never_ set this flag, as another filter
582: * in the list may have previously cleared it.
1.33 mrg 583: */
1.127 itojun 584: /*
585: * let ipfilter look at packet on the wire,
586: * not the decapsulated packet.
587: */
588: #ifdef IPSEC
1.136 itojun 589: if (!ipsec_getnhist(m))
1.127 itojun 590: #else
591: if (1)
592: #endif
593: {
1.169 itojun 594: struct in_addr odst;
595:
596: odst = ip->ip_dst;
1.127 itojun 597: if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif,
1.168 itojun 598: PFIL_IN) != 0)
599: return;
1.127 itojun 600: if (m == NULL)
601: return;
602: ip = mtod(m, struct ip *);
1.142 darrenr 603: hlen = ip->ip_hl << 2;
1.169 itojun 604: srcrt = (odst.s_addr != ip->ip_dst.s_addr);
1.127 itojun 605: }
1.36 mrg 606: #endif /* PFIL_HOOKS */
1.123 thorpej 607:
608: #ifdef ALTQ
609: /* XXX Temporary until ALTQ is changed to use a pfil hook */
610: if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) {
611: /* packet dropped by traffic conditioner */
612: return;
613: }
614: #endif
1.121 thorpej 615:
616: /*
1.1 cgd 617: * Process options and, if not destined for us,
618: * ship it on. ip_dooptions returns 1 when an
619: * error was detected (causing an icmp message
620: * to be sent and the original packet to be freed).
621: */
622: ip_nhops = 0; /* for source routed packets */
623: if (hlen > sizeof (struct ip) && ip_dooptions(m))
1.89 itojun 624: return;
1.1 cgd 625:
626: /*
1.165 christos 627: * Enable a consistency check between the destination address
628: * and the arrival interface for a unicast packet (the RFC 1122
629: * strong ES model) if IP forwarding is disabled and the packet
630: * is not locally generated.
631: *
632: * XXX - Checking also should be disabled if the destination
633: * address is ipnat'ed to a different interface.
634: *
635: * XXX - Checking is incompatible with IP aliases added
636: * to the loopback interface instead of the interface where
637: * the packets are received.
638: *
639: * XXX - We need to add a per ifaddr flag for this so that
640: * we get finer grain control.
641: */
642: checkif = ip_checkinterface && (ipforwarding == 0) &&
643: (m->m_pkthdr.rcvif != NULL) &&
644: ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0);
645:
646: /*
1.1 cgd 647: * Check our list of addresses, to see if the packet is for us.
1.100 itojun 648: *
649: * Traditional 4.4BSD did not consult IFF_UP at all.
650: * The behavior here is to treat addresses on !IFF_UP interface
651: * as not mine.
1.1 cgd 652: */
1.100 itojun 653: downmatch = 0;
1.140 matt 654: LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) {
1.97 itojun 655: if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
1.165 christos 656: if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif)
657: continue;
1.97 itojun 658: if ((ia->ia_ifp->if_flags & IFF_UP) != 0)
659: break;
1.100 itojun 660: else
661: downmatch++;
1.97 itojun 662: }
663: }
1.86 thorpej 664: if (ia != NULL)
665: goto ours;
1.57 tls 666: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
1.140 matt 667: TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) {
668: if (ifa->ifa_addr->sa_family != AF_INET)
669: continue;
1.57 tls 670: ia = ifatoia(ifa);
1.35 mycroft 671: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
672: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 673: /*
674: * Look for all-0's host part (old broadcast addr),
675: * either for subnet or net.
676: */
677: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 678: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 679: goto ours;
1.57 tls 680: /*
681: * An interface with IP address zero accepts
682: * all packets that arrive on that interface.
683: */
684: if (in_nullhost(ia->ia_addr.sin_addr))
685: goto ours;
1.1 cgd 686: }
687: }
1.18 mycroft 688: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 689: struct in_multi *inm;
690: #ifdef MROUTING
691: extern struct socket *ip_mrouter;
1.10 brezak 692:
1.147 matt 693: if (M_READONLY(m)) {
1.10 brezak 694: if ((m = m_pullup(m, hlen)) == 0) {
695: ipstat.ips_toosmall++;
1.89 itojun 696: return;
1.10 brezak 697: }
698: ip = mtod(m, struct ip *);
699: }
1.4 hpeyerl 700:
701: if (ip_mrouter) {
702: /*
703: * If we are acting as a multicast router, all
704: * incoming multicast packets are passed to the
705: * kernel-level multicast forwarding function.
706: * The packet is returned (relatively) intact; if
707: * ip_mforward() returns a non-zero value, the packet
708: * must be discarded, else it may be accepted below.
709: *
710: * (The IP ident field is put in the same byte order
711: * as expected when ip_mforward() is called from
712: * ip_output().)
713: */
1.13 mycroft 714: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
715: ipstat.ips_cantforward++;
1.4 hpeyerl 716: m_freem(m);
1.89 itojun 717: return;
1.4 hpeyerl 718: }
719:
720: /*
721: * The process-level routing demon needs to receive
722: * all multicast IGMP packets, whether or not this
723: * host belongs to their destination groups.
724: */
725: if (ip->ip_p == IPPROTO_IGMP)
726: goto ours;
1.13 mycroft 727: ipstat.ips_forward++;
1.4 hpeyerl 728: }
729: #endif
730: /*
731: * See if we belong to the destination multicast group on the
732: * arrival interface.
733: */
734: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
735: if (inm == NULL) {
1.13 mycroft 736: ipstat.ips_cantforward++;
1.4 hpeyerl 737: m_freem(m);
1.89 itojun 738: return;
1.4 hpeyerl 739: }
740: goto ours;
741: }
1.19 mycroft 742: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 743: in_nullhost(ip->ip_dst))
1.1 cgd 744: goto ours;
745:
746: /*
747: * Not for us; forward if possible and desirable.
748: */
749: if (ipforwarding == 0) {
750: ipstat.ips_cantforward++;
751: m_freem(m);
1.100 itojun 752: } else {
753: /*
754: * If ip_dst matched any of my address on !IFF_UP interface,
755: * and there's no IFF_UP interface that matches ip_dst,
756: * send icmp unreach. Forwarding it will result in in-kernel
757: * forwarding loop till TTL goes to 0.
758: */
759: if (downmatch) {
760: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
761: ipstat.ips_cantforward++;
762: return;
763: }
1.145 itojun 764: #ifdef IPSEC
765: if (ipsec4_in_reject(m, NULL)) {
766: ipsecstat.in_polvio++;
767: goto bad;
768: }
769: #endif
770:
1.169 itojun 771: ip_forward(m, srcrt);
1.100 itojun 772: }
1.89 itojun 773: return;
1.1 cgd 774:
775: ours:
776: /*
777: * If offset or IP_MF are set, must reassemble.
778: * Otherwise, nothing need be done.
779: * (We could look in the reassembly queue to see
780: * if the packet was previously fragmented,
781: * but it's not worth the time; just let them time out.)
782: */
1.155 itojun 783: if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
784: if (M_READONLY(m)) {
785: if ((m = m_pullup(m, hlen)) == NULL) {
786: ipstat.ips_toosmall++;
787: goto bad;
788: }
789: ip = mtod(m, struct ip *);
790: }
791:
1.1 cgd 792: /*
793: * Look for queue of fragments
794: * of this datagram.
795: */
1.75 thorpej 796: IPQ_LOCK();
1.140 matt 797: LIST_FOREACH(fp, &ipq, ipq_q)
1.1 cgd 798: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 799: in_hosteq(ip->ip_src, fp->ipq_src) &&
800: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 801: ip->ip_p == fp->ipq_p)
802: goto found;
803: fp = 0;
804: found:
805:
806: /*
807: * Adjust ip_len to not reflect header,
1.25 cgd 808: * set ipqe_mff if more fragments are expected,
1.1 cgd 809: * convert offset of this to bytes.
810: */
1.155 itojun 811: ip->ip_len = htons(ntohs(ip->ip_len) - hlen);
812: mff = (ip->ip_off & htons(IP_MF)) != 0;
1.25 cgd 813: if (mff) {
1.16 cgd 814: /*
815: * Make sure that fragments have a data length
816: * that's a non-zero multiple of 8 bytes.
817: */
1.155 itojun 818: if (ntohs(ip->ip_len) == 0 ||
819: (ntohs(ip->ip_len) & 0x7) != 0) {
1.16 cgd 820: ipstat.ips_badfrags++;
1.75 thorpej 821: IPQ_UNLOCK();
1.16 cgd 822: goto bad;
823: }
824: }
1.155 itojun 825: ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3);
1.1 cgd 826:
827: /*
828: * If datagram marked as having more fragments
829: * or if this is not the first fragment,
830: * attempt reassembly; if it succeeds, proceed.
831: */
1.155 itojun 832: if (mff || ip->ip_off != htons(0)) {
1.1 cgd 833: ipstat.ips_fragments++;
1.72 thorpej 834: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.25 cgd 835: if (ipqe == NULL) {
836: ipstat.ips_rcvmemdrop++;
1.75 thorpej 837: IPQ_UNLOCK();
1.25 cgd 838: goto bad;
839: }
840: ipqe->ipqe_mff = mff;
1.50 thorpej 841: ipqe->ipqe_m = m;
1.25 cgd 842: ipqe->ipqe_ip = ip;
1.50 thorpej 843: m = ip_reass(ipqe, fp);
1.75 thorpej 844: if (m == 0) {
845: IPQ_UNLOCK();
1.89 itojun 846: return;
1.75 thorpej 847: }
1.13 mycroft 848: ipstat.ips_reassembled++;
1.50 thorpej 849: ip = mtod(m, struct ip *);
1.74 thorpej 850: hlen = ip->ip_hl << 2;
1.155 itojun 851: ip->ip_len = htons(ntohs(ip->ip_len) + hlen);
1.1 cgd 852: } else
853: if (fp)
854: ip_freef(fp);
1.75 thorpej 855: IPQ_UNLOCK();
1.79 mycroft 856: }
1.128 itojun 857:
858: #ifdef IPSEC
859: /*
860: * enforce IPsec policy checking if we are seeing last header.
861: * note that we do not visit this with protocols with pcb layer
862: * code - like udp/tcp/raw ip.
863: */
864: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 &&
865: ipsec4_in_reject(m, NULL)) {
866: ipsecstat.in_polvio++;
867: goto bad;
868: }
869: #endif
1.1 cgd 870:
871: /*
872: * Switch out to protocol's input routine.
873: */
1.82 aidan 874: #if IFA_STATS
1.122 itojun 875: if (ia && ip)
1.155 itojun 876: ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
1.82 aidan 877: #endif
1.1 cgd 878: ipstat.ips_delivered++;
1.89 itojun 879: {
880: int off = hlen, nh = ip->ip_p;
881:
882: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
883: return;
884: }
1.1 cgd 885: bad:
886: m_freem(m);
1.135 thorpej 887: return;
888:
889: badcsum:
890: ipstat.ips_badsum++;
891: m_freem(m);
1.1 cgd 892: }
893:
894: /*
895: * Take incoming datagram fragment and try to
896: * reassemble it into whole datagram. If a chain for
897: * reassembly of this datagram already exists, then it
898: * is given as fp; otherwise have to make a chain.
899: */
1.50 thorpej 900: struct mbuf *
1.25 cgd 901: ip_reass(ipqe, fp)
1.109 augustss 902: struct ipqent *ipqe;
903: struct ipq *fp;
1.1 cgd 904: {
1.109 augustss 905: struct mbuf *m = ipqe->ipqe_m;
906: struct ipqent *nq, *p, *q;
1.25 cgd 907: struct ip *ip;
1.1 cgd 908: struct mbuf *t;
1.25 cgd 909: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 910: int i, next;
911:
1.75 thorpej 912: IPQ_LOCK_CHECK();
913:
1.1 cgd 914: /*
915: * Presence of header sizes in mbufs
916: * would confuse code below.
917: */
918: m->m_data += hlen;
919: m->m_len -= hlen;
920:
921: /*
922: * If first fragment to arrive, create a reassembly queue.
923: */
924: if (fp == 0) {
1.131 itojun 925: /*
926: * Enforce upper bound on number of fragmented packets
927: * for which we attempt reassembly;
928: * If maxfrag is 0, never accept fragments.
929: * If maxfrag is -1, accept all fragments without limitation.
930: */
931: if (ip_maxfragpackets < 0)
932: ;
933: else if (ip_nfragpackets >= ip_maxfragpackets)
934: goto dropfrag;
935: ip_nfragpackets++;
1.50 thorpej 936: MALLOC(fp, struct ipq *, sizeof (struct ipq),
937: M_FTABLE, M_NOWAIT);
938: if (fp == NULL)
1.1 cgd 939: goto dropfrag;
1.25 cgd 940: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 941: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 942: fp->ipq_p = ipqe->ipqe_ip->ip_p;
943: fp->ipq_id = ipqe->ipqe_ip->ip_id;
1.148 matt 944: TAILQ_INIT(&fp->ipq_fragq);
1.25 cgd 945: fp->ipq_src = ipqe->ipqe_ip->ip_src;
946: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
947: p = NULL;
1.1 cgd 948: goto insert;
949: }
950:
951: /*
952: * Find a segment which begins after this one does.
953: */
1.148 matt 954: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
955: p = q, q = TAILQ_NEXT(q, ipqe_q))
1.155 itojun 956: if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off))
1.1 cgd 957: break;
958:
959: /*
960: * If there is a preceding segment, it may provide some of
961: * our data already. If so, drop the data from the incoming
962: * segment. If it provides all of our data, drop us.
963: */
1.25 cgd 964: if (p != NULL) {
1.155 itojun 965: i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) -
966: ntohs(ipqe->ipqe_ip->ip_off);
1.1 cgd 967: if (i > 0) {
1.155 itojun 968: if (i >= ntohs(ipqe->ipqe_ip->ip_len))
1.1 cgd 969: goto dropfrag;
1.50 thorpej 970: m_adj(ipqe->ipqe_m, i);
1.155 itojun 971: ipqe->ipqe_ip->ip_off =
972: htons(ntohs(ipqe->ipqe_ip->ip_off) + i);
973: ipqe->ipqe_ip->ip_len =
974: htons(ntohs(ipqe->ipqe_ip->ip_len) - i);
1.1 cgd 975: }
976: }
977:
978: /*
979: * While we overlap succeeding segments trim them or,
980: * if they are completely covered, dequeue them.
981: */
1.155 itojun 982: for (; q != NULL &&
983: ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) >
984: ntohs(q->ipqe_ip->ip_off); q = nq) {
985: i = (ntohs(ipqe->ipqe_ip->ip_off) +
986: ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off);
987: if (i < ntohs(q->ipqe_ip->ip_len)) {
988: q->ipqe_ip->ip_len =
989: htons(ntohs(q->ipqe_ip->ip_len) - i);
990: q->ipqe_ip->ip_off =
991: htons(ntohs(q->ipqe_ip->ip_off) + i);
1.50 thorpej 992: m_adj(q->ipqe_m, i);
1.1 cgd 993: break;
994: }
1.148 matt 995: nq = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 996: m_freem(q->ipqe_m);
1.148 matt 997: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.72 thorpej 998: pool_put(&ipqent_pool, q);
1.1 cgd 999: }
1000:
1001: insert:
1002: /*
1003: * Stick new segment in its place;
1004: * check for complete reassembly.
1005: */
1.25 cgd 1006: if (p == NULL) {
1.148 matt 1007: TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
1.25 cgd 1008: } else {
1.148 matt 1009: TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q);
1.25 cgd 1010: }
1.1 cgd 1011: next = 0;
1.148 matt 1012: for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL;
1013: p = q, q = TAILQ_NEXT(q, ipqe_q)) {
1.155 itojun 1014: if (ntohs(q->ipqe_ip->ip_off) != next)
1.1 cgd 1015: return (0);
1.155 itojun 1016: next += ntohs(q->ipqe_ip->ip_len);
1.1 cgd 1017: }
1.25 cgd 1018: if (p->ipqe_mff)
1.1 cgd 1019: return (0);
1020:
1021: /*
1.41 thorpej 1022: * Reassembly is complete. Check for a bogus message size and
1023: * concatenate fragments.
1.1 cgd 1024: */
1.148 matt 1025: q = TAILQ_FIRST(&fp->ipq_fragq);
1.25 cgd 1026: ip = q->ipqe_ip;
1.41 thorpej 1027: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
1028: ipstat.ips_toolong++;
1029: ip_freef(fp);
1030: return (0);
1031: }
1.50 thorpej 1032: m = q->ipqe_m;
1.1 cgd 1033: t = m->m_next;
1034: m->m_next = 0;
1035: m_cat(m, t);
1.148 matt 1036: nq = TAILQ_NEXT(q, ipqe_q);
1.72 thorpej 1037: pool_put(&ipqent_pool, q);
1.25 cgd 1038: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 1039: t = q->ipqe_m;
1.148 matt 1040: nq = TAILQ_NEXT(q, ipqe_q);
1.72 thorpej 1041: pool_put(&ipqent_pool, q);
1.1 cgd 1042: m_cat(m, t);
1043: }
1044:
1045: /*
1046: * Create header for new ip packet by
1047: * modifying header of first packet;
1048: * dequeue and discard fragment reassembly header.
1049: * Make header visible.
1050: */
1.155 itojun 1051: ip->ip_len = htons(next);
1.25 cgd 1052: ip->ip_src = fp->ipq_src;
1053: ip->ip_dst = fp->ipq_dst;
1054: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1055: FREE(fp, M_FTABLE);
1.131 itojun 1056: ip_nfragpackets--;
1.1 cgd 1057: m->m_len += (ip->ip_hl << 2);
1058: m->m_data -= (ip->ip_hl << 2);
1059: /* some debugging cruft by sklower, below, will go away soon */
1060: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
1.109 augustss 1061: int plen = 0;
1.50 thorpej 1062: for (t = m; t; t = t->m_next)
1063: plen += t->m_len;
1064: m->m_pkthdr.len = plen;
1.1 cgd 1065: }
1.50 thorpej 1066: return (m);
1.1 cgd 1067:
1068: dropfrag:
1069: ipstat.ips_fragdropped++;
1070: m_freem(m);
1.72 thorpej 1071: pool_put(&ipqent_pool, ipqe);
1.1 cgd 1072: return (0);
1073: }
1074:
1075: /*
1076: * Free a fragment reassembly header and all
1077: * associated datagrams.
1078: */
1.8 mycroft 1079: void
1.1 cgd 1080: ip_freef(fp)
1081: struct ipq *fp;
1082: {
1.109 augustss 1083: struct ipqent *q, *p;
1.1 cgd 1084:
1.75 thorpej 1085: IPQ_LOCK_CHECK();
1086:
1.148 matt 1087: for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) {
1088: p = TAILQ_NEXT(q, ipqe_q);
1.50 thorpej 1089: m_freem(q->ipqe_m);
1.148 matt 1090: TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q);
1.72 thorpej 1091: pool_put(&ipqent_pool, q);
1.1 cgd 1092: }
1.25 cgd 1093: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 1094: FREE(fp, M_FTABLE);
1.131 itojun 1095: ip_nfragpackets--;
1.1 cgd 1096: }
1097:
1098: /*
1099: * IP timer processing;
1100: * if a timer expires on a reassembly
1101: * queue, discard it.
1102: */
1.8 mycroft 1103: void
1.1 cgd 1104: ip_slowtimo()
1105: {
1.109 augustss 1106: struct ipq *fp, *nfp;
1.24 mycroft 1107: int s = splsoftnet();
1.1 cgd 1108:
1.75 thorpej 1109: IPQ_LOCK();
1.140 matt 1110: for (fp = LIST_FIRST(&ipq); fp != NULL; fp = nfp) {
1111: nfp = LIST_NEXT(fp, ipq_q);
1.25 cgd 1112: if (--fp->ipq_ttl == 0) {
1.1 cgd 1113: ipstat.ips_fragtimeout++;
1.25 cgd 1114: ip_freef(fp);
1.1 cgd 1115: }
1116: }
1.131 itojun 1117: /*
1118: * If we are over the maximum number of fragments
1119: * (due to the limit being lowered), drain off
1120: * enough to get down to the new limit.
1121: */
1122: if (ip_maxfragpackets < 0)
1123: ;
1124: else {
1.140 matt 1125: while (ip_nfragpackets > ip_maxfragpackets && LIST_FIRST(&ipq))
1126: ip_freef(LIST_FIRST(&ipq));
1.131 itojun 1127: }
1.75 thorpej 1128: IPQ_UNLOCK();
1.63 matt 1129: #ifdef GATEWAY
1130: ipflow_slowtimo();
1131: #endif
1.1 cgd 1132: splx(s);
1133: }
1134:
1135: /*
1136: * Drain off all datagram fragments.
1137: */
1.8 mycroft 1138: void
1.1 cgd 1139: ip_drain()
1140: {
1141:
1.75 thorpej 1142: /*
1143: * We may be called from a device's interrupt context. If
1144: * the ipq is already busy, just bail out now.
1145: */
1146: if (ipq_lock_try() == 0)
1147: return;
1148:
1.140 matt 1149: while (LIST_FIRST(&ipq) != NULL) {
1.1 cgd 1150: ipstat.ips_fragdropped++;
1.140 matt 1151: ip_freef(LIST_FIRST(&ipq));
1.1 cgd 1152: }
1.75 thorpej 1153:
1154: IPQ_UNLOCK();
1.1 cgd 1155: }
1156:
1157: /*
1158: * Do option processing on a datagram,
1159: * possibly discarding it if bad options are encountered,
1160: * or forwarding it if source-routed.
1161: * Returns 1 if packet has been forwarded/freed,
1162: * 0 if the packet should be processed further.
1163: */
1.8 mycroft 1164: int
1.1 cgd 1165: ip_dooptions(m)
1166: struct mbuf *m;
1167: {
1.109 augustss 1168: struct ip *ip = mtod(m, struct ip *);
1169: u_char *cp, *cp0;
1170: struct ip_timestamp *ipt;
1171: struct in_ifaddr *ia;
1.1 cgd 1172: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 1173: struct in_addr dst;
1.1 cgd 1174: n_time ntime;
1175:
1.13 mycroft 1176: dst = ip->ip_dst;
1.1 cgd 1177: cp = (u_char *)(ip + 1);
1178: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
1179: for (; cnt > 0; cnt -= optlen, cp += optlen) {
1180: opt = cp[IPOPT_OPTVAL];
1181: if (opt == IPOPT_EOL)
1182: break;
1183: if (opt == IPOPT_NOP)
1184: optlen = 1;
1185: else {
1.113 itojun 1186: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1187: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1188: goto bad;
1189: }
1.1 cgd 1190: optlen = cp[IPOPT_OLEN];
1.114 itojun 1191: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 1192: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1193: goto bad;
1194: }
1195: }
1196: switch (opt) {
1197:
1198: default:
1199: break;
1200:
1201: /*
1202: * Source routing with record.
1203: * Find interface with current destination address.
1204: * If none on this machine then drop if strictly routed,
1205: * or do nothing if loosely routed.
1206: * Record interface address and bring up next address
1207: * component. If strictly routed make sure next
1208: * address is on directly accessible net.
1209: */
1210: case IPOPT_LSRR:
1211: case IPOPT_SSRR:
1.47 cjs 1212: if (ip_allowsrcrt == 0) {
1213: type = ICMP_UNREACH;
1214: code = ICMP_UNREACH_NET_PROHIB;
1215: goto bad;
1216: }
1.114 itojun 1217: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1218: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1219: goto bad;
1220: }
1.1 cgd 1221: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1222: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1223: goto bad;
1224: }
1225: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 1226: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 1227: if (ia == 0) {
1228: if (opt == IPOPT_SSRR) {
1229: type = ICMP_UNREACH;
1230: code = ICMP_UNREACH_SRCFAIL;
1231: goto bad;
1232: }
1233: /*
1234: * Loose routing, and not at next destination
1235: * yet; nothing to do except forward.
1236: */
1237: break;
1238: }
1239: off--; /* 0 origin */
1.112 sommerfe 1240: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 1241: /*
1242: * End of source route. Should be for us.
1243: */
1244: save_rte(cp, ip->ip_src);
1245: break;
1246: }
1247: /*
1248: * locate outgoing interface
1249: */
1250: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
1251: sizeof(ipaddr.sin_addr));
1.96 thorpej 1252: if (opt == IPOPT_SSRR)
1253: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1254: else
1.1 cgd 1255: ia = ip_rtaddr(ipaddr.sin_addr);
1256: if (ia == 0) {
1257: type = ICMP_UNREACH;
1258: code = ICMP_UNREACH_SRCFAIL;
1259: goto bad;
1260: }
1261: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 1262: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 1263: (caddr_t)(cp + off), sizeof(struct in_addr));
1264: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1265: /*
1266: * Let ip_intr's mcast routing check handle mcast pkts
1267: */
1.18 mycroft 1268: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1269: break;
1270:
1271: case IPOPT_RR:
1.114 itojun 1272: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1273: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1274: goto bad;
1275: }
1.1 cgd 1276: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1277: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1278: goto bad;
1279: }
1280: /*
1281: * If no space remains, ignore.
1282: */
1283: off--; /* 0 origin */
1.112 sommerfe 1284: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1285: break;
1286: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
1287: sizeof(ipaddr.sin_addr));
1288: /*
1289: * locate outgoing interface; if we're the destination,
1290: * use the incoming interface (should be same).
1291: */
1.96 thorpej 1292: if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))))
1293: == NULL &&
1294: (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
1.1 cgd 1295: type = ICMP_UNREACH;
1296: code = ICMP_UNREACH_HOST;
1297: goto bad;
1298: }
1.20 mycroft 1299: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 1300: (caddr_t)(cp + off), sizeof(struct in_addr));
1301: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1302: break;
1303:
1304: case IPOPT_TS:
1305: code = cp - (u_char *)ip;
1306: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1307: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1308: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1309: goto bad;
1.114 itojun 1310: }
1311: if (ipt->ipt_ptr < 5) {
1312: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1313: goto bad;
1314: }
1.15 cgd 1315: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1316: if (++ipt->ipt_oflw == 0) {
1317: code = (u_char *)&ipt->ipt_ptr -
1318: (u_char *)ip;
1.1 cgd 1319: goto bad;
1.114 itojun 1320: }
1.1 cgd 1321: break;
1322: }
1.104 thorpej 1323: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1324: switch (ipt->ipt_flg) {
1325:
1326: case IPOPT_TS_TSONLY:
1327: break;
1328:
1329: case IPOPT_TS_TSANDADDR:
1.66 thorpej 1330: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1331: sizeof(struct in_addr) > ipt->ipt_len) {
1332: code = (u_char *)&ipt->ipt_ptr -
1333: (u_char *)ip;
1.1 cgd 1334: goto bad;
1.114 itojun 1335: }
1.13 mycroft 1336: ipaddr.sin_addr = dst;
1.96 thorpej 1337: ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr),
1338: m->m_pkthdr.rcvif));
1.13 mycroft 1339: if (ia == 0)
1340: continue;
1.104 thorpej 1341: bcopy(&ia->ia_addr.sin_addr,
1342: cp0, sizeof(struct in_addr));
1.1 cgd 1343: ipt->ipt_ptr += sizeof(struct in_addr);
1344: break;
1345:
1346: case IPOPT_TS_PRESPEC:
1.66 thorpej 1347: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1348: sizeof(struct in_addr) > ipt->ipt_len) {
1349: code = (u_char *)&ipt->ipt_ptr -
1350: (u_char *)ip;
1.1 cgd 1351: goto bad;
1.114 itojun 1352: }
1.104 thorpej 1353: bcopy(cp0, &ipaddr.sin_addr,
1.1 cgd 1354: sizeof(struct in_addr));
1.96 thorpej 1355: if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))
1356: == NULL)
1.1 cgd 1357: continue;
1358: ipt->ipt_ptr += sizeof(struct in_addr);
1359: break;
1360:
1361: default:
1.114 itojun 1362: /* XXX can't take &ipt->ipt_flg */
1363: code = (u_char *)&ipt->ipt_ptr -
1364: (u_char *)ip + 1;
1.1 cgd 1365: goto bad;
1366: }
1367: ntime = iptime();
1.107 thorpej 1368: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1369: bcopy(cp0, (caddr_t)cp + ipt->ipt_ptr - 1,
1.1 cgd 1370: sizeof(n_time));
1371: ipt->ipt_ptr += sizeof(n_time);
1372: }
1373: }
1374: if (forward) {
1.26 thorpej 1375: if (ip_forwsrcrt == 0) {
1376: type = ICMP_UNREACH;
1377: code = ICMP_UNREACH_SRCFAIL;
1378: goto bad;
1379: }
1.1 cgd 1380: ip_forward(m, 1);
1381: return (1);
1.13 mycroft 1382: }
1383: return (0);
1.1 cgd 1384: bad:
1.13 mycroft 1385: icmp_error(m, type, code, 0, 0);
1386: ipstat.ips_badoptions++;
1.1 cgd 1387: return (1);
1388: }
1389:
1390: /*
1391: * Given address of next destination (final or next hop),
1392: * return internet address info of interface to be used to get there.
1393: */
1394: struct in_ifaddr *
1395: ip_rtaddr(dst)
1396: struct in_addr dst;
1397: {
1.109 augustss 1398: struct sockaddr_in *sin;
1.1 cgd 1399:
1.19 mycroft 1400: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1401:
1.35 mycroft 1402: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 1403: if (ipforward_rt.ro_rt) {
1404: RTFREE(ipforward_rt.ro_rt);
1405: ipforward_rt.ro_rt = 0;
1406: }
1407: sin->sin_family = AF_INET;
1408: sin->sin_len = sizeof(*sin);
1409: sin->sin_addr = dst;
1410:
1411: rtalloc(&ipforward_rt);
1412: }
1413: if (ipforward_rt.ro_rt == 0)
1414: return ((struct in_ifaddr *)0);
1.19 mycroft 1415: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 1416: }
1417:
1418: /*
1419: * Save incoming source route for use in replies,
1420: * to be picked up later by ip_srcroute if the receiver is interested.
1421: */
1.13 mycroft 1422: void
1.1 cgd 1423: save_rte(option, dst)
1424: u_char *option;
1425: struct in_addr dst;
1426: {
1427: unsigned olen;
1428:
1429: olen = option[IPOPT_OLEN];
1430: #ifdef DIAGNOSTIC
1431: if (ipprintfs)
1.39 christos 1432: printf("save_rte: olen %d\n", olen);
1.89 itojun 1433: #endif /* 0 */
1.1 cgd 1434: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1435: return;
1436: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
1437: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1438: ip_srcrt.dst = dst;
1439: }
1440:
1441: /*
1442: * Retrieve incoming source route for use in replies,
1443: * in the same form used by setsockopt.
1444: * The first hop is placed before the options, will be removed later.
1445: */
1446: struct mbuf *
1447: ip_srcroute()
1448: {
1.109 augustss 1449: struct in_addr *p, *q;
1450: struct mbuf *m;
1.1 cgd 1451:
1452: if (ip_nhops == 0)
1453: return ((struct mbuf *)0);
1454: m = m_get(M_DONTWAIT, MT_SOOPTS);
1455: if (m == 0)
1456: return ((struct mbuf *)0);
1457:
1.164 matt 1458: MCLAIM(m, &inetdomain.dom_mowner);
1.13 mycroft 1459: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1460:
1461: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1462: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1463: OPTSIZ;
1464: #ifdef DIAGNOSTIC
1465: if (ipprintfs)
1.39 christos 1466: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1467: #endif
1468:
1469: /*
1470: * First save first hop for return route
1471: */
1472: p = &ip_srcrt.route[ip_nhops - 1];
1473: *(mtod(m, struct in_addr *)) = *p--;
1474: #ifdef DIAGNOSTIC
1475: if (ipprintfs)
1.39 christos 1476: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1477: #endif
1478:
1479: /*
1480: * Copy option fields and padding (nop) to mbuf.
1481: */
1482: ip_srcrt.nop = IPOPT_NOP;
1483: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1484: bcopy((caddr_t)&ip_srcrt.nop,
1485: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1486: q = (struct in_addr *)(mtod(m, caddr_t) +
1487: sizeof(struct in_addr) + OPTSIZ);
1488: #undef OPTSIZ
1489: /*
1490: * Record return path as an IP source route,
1491: * reversing the path (pointers are now aligned).
1492: */
1493: while (p >= ip_srcrt.route) {
1494: #ifdef DIAGNOSTIC
1495: if (ipprintfs)
1.39 christos 1496: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1497: #endif
1498: *q++ = *p--;
1499: }
1500: /*
1501: * Last hop goes to final destination.
1502: */
1503: *q = ip_srcrt.dst;
1504: #ifdef DIAGNOSTIC
1505: if (ipprintfs)
1.39 christos 1506: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1507: #endif
1508: return (m);
1509: }
1510:
1511: /*
1512: * Strip out IP options, at higher
1513: * level protocol in the kernel.
1514: * Second argument is buffer to which options
1515: * will be moved, and return value is their length.
1516: * XXX should be deleted; last arg currently ignored.
1517: */
1.8 mycroft 1518: void
1.1 cgd 1519: ip_stripoptions(m, mopt)
1.109 augustss 1520: struct mbuf *m;
1.1 cgd 1521: struct mbuf *mopt;
1522: {
1.109 augustss 1523: int i;
1.1 cgd 1524: struct ip *ip = mtod(m, struct ip *);
1.109 augustss 1525: caddr_t opts;
1.1 cgd 1526: int olen;
1527:
1.79 mycroft 1528: olen = (ip->ip_hl << 2) - sizeof (struct ip);
1.1 cgd 1529: opts = (caddr_t)(ip + 1);
1530: i = m->m_len - (sizeof (struct ip) + olen);
1531: bcopy(opts + olen, opts, (unsigned)i);
1532: m->m_len -= olen;
1533: if (m->m_flags & M_PKTHDR)
1534: m->m_pkthdr.len -= olen;
1.155 itojun 1535: ip->ip_len = htons(ntohs(ip->ip_len) - olen);
1.79 mycroft 1536: ip->ip_hl = sizeof (struct ip) >> 2;
1.1 cgd 1537: }
1538:
1.139 matt 1539: const int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1540: 0, 0, 0, 0,
1541: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1542: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1543: EMSGSIZE, EHOSTUNREACH, 0, 0,
1544: 0, 0, 0, 0,
1545: ENOPROTOOPT
1546: };
1547:
1548: /*
1549: * Forward a packet. If some error occurs return the sender
1550: * an icmp packet. Note we can't always generate a meaningful
1551: * icmp message because icmp doesn't have a large enough repertoire
1552: * of codes and types.
1553: *
1554: * If not forwarding, just drop the packet. This could be confusing
1555: * if ipforwarding was zero but some routing protocol was advancing
1556: * us as a gateway to somewhere. However, we must let the routing
1557: * protocol deal with that.
1558: *
1559: * The srcrt parameter indicates whether the packet is being forwarded
1560: * via a source route.
1561: */
1.13 mycroft 1562: void
1.1 cgd 1563: ip_forward(m, srcrt)
1564: struct mbuf *m;
1565: int srcrt;
1566: {
1.109 augustss 1567: struct ip *ip = mtod(m, struct ip *);
1568: struct sockaddr_in *sin;
1569: struct rtentry *rt;
1.28 christos 1570: int error, type = 0, code = 0;
1.1 cgd 1571: struct mbuf *mcopy;
1.13 mycroft 1572: n_long dest;
1573: struct ifnet *destifp;
1.89 itojun 1574: #ifdef IPSEC
1575: struct ifnet dummyifp;
1576: #endif
1.164 matt 1577:
1578: /*
1579: * We are now in the output path.
1580: */
1581: MCLAIM(m, &ip_tx_mowner);
1.135 thorpej 1582:
1583: /*
1584: * Clear any in-bound checksum flags for this packet.
1585: */
1586: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1587:
1.13 mycroft 1588: dest = 0;
1.1 cgd 1589: #ifdef DIAGNOSTIC
1590: if (ipprintfs)
1.70 thorpej 1591: printf("forward: src %2.2x dst %2.2x ttl %x\n",
1592: ntohl(ip->ip_src.s_addr),
1593: ntohl(ip->ip_dst.s_addr), ip->ip_ttl);
1.1 cgd 1594: #endif
1.93 sommerfe 1595: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.1 cgd 1596: ipstat.ips_cantforward++;
1597: m_freem(m);
1598: return;
1599: }
1600: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1601: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1602: return;
1603: }
1604: ip->ip_ttl -= IPTTLDEC;
1605:
1.19 mycroft 1606: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1607: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1608: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1609: if (ipforward_rt.ro_rt) {
1610: RTFREE(ipforward_rt.ro_rt);
1611: ipforward_rt.ro_rt = 0;
1612: }
1613: sin->sin_family = AF_INET;
1.35 mycroft 1614: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1615: sin->sin_addr = ip->ip_dst;
1616:
1617: rtalloc(&ipforward_rt);
1618: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1619: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1620: return;
1621: }
1622: rt = ipforward_rt.ro_rt;
1623: }
1624:
1625: /*
1.34 mycroft 1626: * Save at most 68 bytes of the packet in case
1.1 cgd 1627: * we need to generate an ICMP message to the src.
1.119 itojun 1628: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1629: */
1.155 itojun 1630: mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1.119 itojun 1631: if (mcopy)
1632: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1633:
1634: /*
1635: * If forwarding packet using same interface that it came in on,
1636: * perhaps should send a redirect to sender to shortcut a hop.
1637: * Only send redirect if source is sending directly to us,
1638: * and if packet was not source routed (or has any options).
1639: * Also, don't send redirect if forwarding using a default route
1640: * or a route modified by a redirect.
1641: */
1642: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1643: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1644: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1645: ipsendredirects && !srcrt) {
1.19 mycroft 1646: if (rt->rt_ifa &&
1647: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1648: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1649: if (rt->rt_flags & RTF_GATEWAY)
1650: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1651: else
1652: dest = ip->ip_dst.s_addr;
1653: /*
1654: * Router requirements says to only send host
1655: * redirects.
1656: */
1657: type = ICMP_REDIRECT;
1658: code = ICMP_REDIRECT_HOST;
1.1 cgd 1659: #ifdef DIAGNOSTIC
1.77 thorpej 1660: if (ipprintfs)
1661: printf("redirect (%d) to %x\n", code,
1662: (u_int32_t)dest);
1.1 cgd 1663: #endif
1664: }
1665: }
1666:
1.89 itojun 1667: #ifdef IPSEC
1.134 lukem 1668: /* Don't lookup socket in forwarding case */
1.127 itojun 1669: (void)ipsec_setsocket(m, NULL);
1.103 itojun 1670: #endif
1.27 thorpej 1671: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1672: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1.1 cgd 1673: if (error)
1674: ipstat.ips_cantforward++;
1675: else {
1676: ipstat.ips_forward++;
1677: if (type)
1678: ipstat.ips_redirectsent++;
1679: else {
1.63 matt 1680: if (mcopy) {
1681: #ifdef GATEWAY
1.64 thorpej 1682: if (mcopy->m_flags & M_CANFASTFWD)
1683: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1684: #endif
1.1 cgd 1685: m_freem(mcopy);
1.63 matt 1686: }
1.1 cgd 1687: return;
1688: }
1689: }
1690: if (mcopy == NULL)
1691: return;
1.13 mycroft 1692: destifp = NULL;
1693:
1.1 cgd 1694: switch (error) {
1695:
1696: case 0: /* forwarded, but need redirect */
1697: /* type, code set above */
1698: break;
1699:
1700: case ENETUNREACH: /* shouldn't happen, checked above */
1701: case EHOSTUNREACH:
1702: case ENETDOWN:
1703: case EHOSTDOWN:
1704: default:
1705: type = ICMP_UNREACH;
1706: code = ICMP_UNREACH_HOST;
1707: break;
1708:
1709: case EMSGSIZE:
1710: type = ICMP_UNREACH;
1711: code = ICMP_UNREACH_NEEDFRAG;
1.89 itojun 1712: #ifndef IPSEC
1.13 mycroft 1713: if (ipforward_rt.ro_rt)
1714: destifp = ipforward_rt.ro_rt->rt_ifp;
1.89 itojun 1715: #else
1716: /*
1717: * If the packet is routed over IPsec tunnel, tell the
1718: * originator the tunnel MTU.
1719: * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1720: * XXX quickhack!!!
1721: */
1722: if (ipforward_rt.ro_rt) {
1723: struct secpolicy *sp;
1724: int ipsecerror;
1.95 itojun 1725: size_t ipsechdr;
1.89 itojun 1726: struct route *ro;
1727:
1728: sp = ipsec4_getpolicybyaddr(mcopy,
1.170 ! itojun 1729: IPSEC_DIR_OUTBOUND, IP_FORWARDING,
! 1730: &ipsecerror);
1.89 itojun 1731:
1732: if (sp == NULL)
1733: destifp = ipforward_rt.ro_rt->rt_ifp;
1734: else {
1735: /* count IPsec header size */
1.95 itojun 1736: ipsechdr = ipsec4_hdrsiz(mcopy,
1.170 ! itojun 1737: IPSEC_DIR_OUTBOUND, NULL);
1.89 itojun 1738:
1739: /*
1740: * find the correct route for outer IPv4
1741: * header, compute tunnel MTU.
1742: *
1743: * XXX BUG ALERT
1744: * The "dummyifp" code relies upon the fact
1745: * that icmp_error() touches only ifp->if_mtu.
1746: */
1747: /*XXX*/
1748: destifp = NULL;
1749: if (sp->req != NULL
1.95 itojun 1750: && sp->req->sav != NULL
1751: && sp->req->sav->sah != NULL) {
1752: ro = &sp->req->sav->sah->sa_route;
1.89 itojun 1753: if (ro->ro_rt && ro->ro_rt->rt_ifp) {
1754: dummyifp.if_mtu =
1.151 itojun 1755: ro->ro_rt->rt_rmx.rmx_mtu ?
1756: ro->ro_rt->rt_rmx.rmx_mtu :
1.89 itojun 1757: ro->ro_rt->rt_ifp->if_mtu;
1758: dummyifp.if_mtu -= ipsechdr;
1759: destifp = &dummyifp;
1760: }
1761: }
1762:
1763: key_freesp(sp);
1764: }
1765: }
1766: #endif /*IPSEC*/
1.1 cgd 1767: ipstat.ips_cantfrag++;
1768: break;
1769:
1770: case ENOBUFS:
1.143 itojun 1771: #if 1
1772: /*
1773: * a router should not generate ICMP_SOURCEQUENCH as
1774: * required in RFC1812 Requirements for IP Version 4 Routers.
1775: * source quench could be a big problem under DoS attacks,
1.149 wiz 1776: * or if the underlying interface is rate-limited.
1.143 itojun 1777: */
1778: if (mcopy)
1779: m_freem(mcopy);
1780: return;
1781: #else
1.1 cgd 1782: type = ICMP_SOURCEQUENCH;
1783: code = 0;
1784: break;
1.143 itojun 1785: #endif
1.1 cgd 1786: }
1.13 mycroft 1787: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 1788: }
1789:
1790: void
1791: ip_savecontrol(inp, mp, ip, m)
1.109 augustss 1792: struct inpcb *inp;
1793: struct mbuf **mp;
1794: struct ip *ip;
1795: struct mbuf *m;
1.44 thorpej 1796: {
1797:
1798: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1799: struct timeval tv;
1800:
1801: microtime(&tv);
1802: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1803: SCM_TIMESTAMP, SOL_SOCKET);
1804: if (*mp)
1805: mp = &(*mp)->m_next;
1806: }
1807: if (inp->inp_flags & INP_RECVDSTADDR) {
1808: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1809: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1810: if (*mp)
1811: mp = &(*mp)->m_next;
1812: }
1813: #ifdef notyet
1814: /*
1815: * XXX
1816: * Moving these out of udp_input() made them even more broken
1817: * than they already were.
1818: * - fenner@parc.xerox.com
1819: */
1820: /* options were tossed already */
1821: if (inp->inp_flags & INP_RECVOPTS) {
1822: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1823: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1824: if (*mp)
1825: mp = &(*mp)->m_next;
1826: }
1827: /* ip_srcroute doesn't do what we want here, need to fix */
1828: if (inp->inp_flags & INP_RECVRETOPTS) {
1829: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1830: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1831: if (*mp)
1832: mp = &(*mp)->m_next;
1833: }
1834: #endif
1835: if (inp->inp_flags & INP_RECVIF) {
1836: struct sockaddr_dl sdl;
1837:
1838: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1839: sdl.sdl_family = AF_LINK;
1840: sdl.sdl_index = m->m_pkthdr.rcvif ?
1841: m->m_pkthdr.rcvif->if_index : 0;
1842: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1843: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1844: IP_RECVIF, IPPROTO_IP);
1845: if (*mp)
1846: mp = &(*mp)->m_next;
1847: }
1.13 mycroft 1848: }
1849:
1850: int
1851: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1852: int *name;
1853: u_int namelen;
1854: void *oldp;
1855: size_t *oldlenp;
1856: void *newp;
1857: size_t newlen;
1858: {
1.88 sommerfe 1859: extern int subnetsarelocal, hostzeroisbroadcast;
1.52 thorpej 1860:
1.54 lukem 1861: int error, old;
1862:
1.13 mycroft 1863: /* All sysctl names at this level are terminal. */
1864: if (namelen != 1)
1865: return (ENOTDIR);
1866:
1867: switch (name[0]) {
1868: case IPCTL_FORWARDING:
1869: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1870: case IPCTL_SENDREDIRECTS:
1871: return (sysctl_int(oldp, oldlenp, newp, newlen,
1872: &ipsendredirects));
1873: case IPCTL_DEFTTL:
1874: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1875: #ifdef notyet
1876: case IPCTL_DEFMTU:
1877: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1878: #endif
1.26 thorpej 1879: case IPCTL_FORWSRCRT:
1.47 cjs 1880: /* Don't allow this to change in a secure environment. */
1.26 thorpej 1881: if (securelevel > 0)
1.46 cjs 1882: return (sysctl_rdint(oldp, oldlenp, newp,
1883: ip_forwsrcrt));
1884: else
1885: return (sysctl_int(oldp, oldlenp, newp, newlen,
1886: &ip_forwsrcrt));
1.27 thorpej 1887: case IPCTL_DIRECTEDBCAST:
1888: return (sysctl_int(oldp, oldlenp, newp, newlen,
1889: &ip_directedbcast));
1.47 cjs 1890: case IPCTL_ALLOWSRCRT:
1891: return (sysctl_int(oldp, oldlenp, newp, newlen,
1892: &ip_allowsrcrt));
1.52 thorpej 1893: case IPCTL_SUBNETSARELOCAL:
1894: return (sysctl_int(oldp, oldlenp, newp, newlen,
1895: &subnetsarelocal));
1.53 kml 1896: case IPCTL_MTUDISC:
1.60 kml 1897: error = sysctl_int(oldp, oldlenp, newp, newlen,
1898: &ip_mtudisc);
1.163 itojun 1899: if (error == 0 && ip_mtudisc == 0)
1900: rt_timer_queue_remove_all(ip_mtudisc_timeout_q, TRUE);
1.60 kml 1901: return error;
1.54 lukem 1902: case IPCTL_ANONPORTMIN:
1903: old = anonportmin;
1904: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1.118 itojun 1905: if (anonportmin >= anonportmax || anonportmin < 0
1906: || anonportmin > 65535
1.54 lukem 1907: #ifndef IPNOPRIVPORTS
1908: || anonportmin < IPPORT_RESERVED
1909: #endif
1910: ) {
1911: anonportmin = old;
1912: return (EINVAL);
1913: }
1914: return (error);
1915: case IPCTL_ANONPORTMAX:
1916: old = anonportmax;
1917: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1.118 itojun 1918: if (anonportmin >= anonportmax || anonportmax < 0
1919: || anonportmax > 65535
1.54 lukem 1920: #ifndef IPNOPRIVPORTS
1921: || anonportmax < IPPORT_RESERVED
1922: #endif
1923: ) {
1924: anonportmax = old;
1925: return (EINVAL);
1926: }
1.60 kml 1927: return (error);
1928: case IPCTL_MTUDISCTIMEOUT:
1.158 itojun 1929: old = ip_mtudisc_timeout;
1.60 kml 1930: error = sysctl_int(oldp, oldlenp, newp, newlen,
1931: &ip_mtudisc_timeout);
1.158 itojun 1932: if (ip_mtudisc_timeout < 0) {
1933: ip_mtudisc_timeout = old;
1.156 itojun 1934: return (EINVAL);
1.158 itojun 1935: }
1.163 itojun 1936: if (error == 0)
1.162 itojun 1937: rt_timer_queue_change(ip_mtudisc_timeout_q,
1938: ip_mtudisc_timeout);
1.54 lukem 1939: return (error);
1.65 matt 1940: #ifdef GATEWAY
1941: case IPCTL_MAXFLOWS:
1.67 thorpej 1942: {
1943: int s;
1944:
1.65 matt 1945: error = sysctl_int(oldp, oldlenp, newp, newlen,
1946: &ip_maxflows);
1.67 thorpej 1947: s = splsoftnet();
1.65 matt 1948: ipflow_reap(0);
1.67 thorpej 1949: splx(s);
1.65 matt 1950: return (error);
1.67 thorpej 1951: }
1.89 itojun 1952: #endif
1.90 itojun 1953: case IPCTL_HOSTZEROBROADCAST:
1954: return (sysctl_int(oldp, oldlenp, newp, newlen,
1955: &hostzeroisbroadcast));
1.89 itojun 1956: #if NGIF > 0
1957: case IPCTL_GIF_TTL:
1.157 itojun 1958: return (sysctl_int(oldp, oldlenp, newp, newlen,
1.90 itojun 1959: &ip_gif_ttl));
1.144 martin 1960: #endif
1961:
1962: #if NGRE > 0
1963: case IPCTL_GRE_TTL:
1.157 itojun 1964: return (sysctl_int(oldp, oldlenp, newp, newlen,
1.144 martin 1965: &ip_gre_ttl));
1.117 tron 1966: #endif
1967:
1968: #ifndef IPNOPRIVPORTS
1969: case IPCTL_LOWPORTMIN:
1970: old = lowportmin;
1971: error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmin);
1972: if (lowportmin >= lowportmax
1973: || lowportmin > IPPORT_RESERVEDMAX
1974: || lowportmin < IPPORT_RESERVEDMIN
1975: ) {
1976: lowportmin = old;
1977: return (EINVAL);
1978: }
1979: return (error);
1980: case IPCTL_LOWPORTMAX:
1981: old = lowportmax;
1982: error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmax);
1983: if (lowportmin >= lowportmax
1984: || lowportmax > IPPORT_RESERVEDMAX
1985: || lowportmax < IPPORT_RESERVEDMIN
1986: ) {
1987: lowportmax = old;
1988: return (EINVAL);
1989: }
1990: return (error);
1.65 matt 1991: #endif
1.131 itojun 1992:
1993: case IPCTL_MAXFRAGPACKETS:
1994: return (sysctl_int(oldp, oldlenp, newp, newlen,
1995: &ip_maxfragpackets));
1.88 sommerfe 1996:
1.165 christos 1997: case IPCTL_CHECKINTERFACE:
1998: return (sysctl_int(oldp, oldlenp, newp, newlen,
1999: &ip_checkinterface));
1.13 mycroft 2000: default:
2001: return (EOPNOTSUPP);
2002: }
2003: /* NOTREACHED */
1.1 cgd 2004: }
CVSweb <webmaster@jp.NetBSD.org>
![[BACK]](/icons/back.gif){kind=icon}
Return to ip_input.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [cvs.NetBSD.org] / src / sys / netinet