Annotation of src/sys/netinet/ip_input.c, Revision 1.144
1.144 ! martin 1: /* $NetBSD: ip_input.c,v 1.143 2002/02/21 08:39:33 itojun Exp $ */
1.89 itojun 2:
3: /*
4: * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5: * All rights reserved.
6: *
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.
18: *
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>
1.144 ! martin 105: __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.143 2002/02/21 08:39:33 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"
1.135 thorpej 111: #include "opt_inet_csum.h"
1.1 cgd 112:
1.5 mycroft 113: #include <sys/param.h>
114: #include <sys/systm.h>
115: #include <sys/malloc.h>
116: #include <sys/mbuf.h>
117: #include <sys/domain.h>
118: #include <sys/protosw.h>
119: #include <sys/socket.h>
1.44 thorpej 120: #include <sys/socketvar.h>
1.5 mycroft 121: #include <sys/errno.h>
122: #include <sys/time.h>
123: #include <sys/kernel.h>
1.72 thorpej 124: #include <sys/pool.h>
1.28 christos 125: #include <sys/sysctl.h>
1.1 cgd 126:
1.5 mycroft 127: #include <net/if.h>
1.44 thorpej 128: #include <net/if_dl.h>
1.5 mycroft 129: #include <net/route.h>
1.45 mrg 130: #include <net/pfil.h>
1.1 cgd 131:
1.5 mycroft 132: #include <netinet/in.h>
133: #include <netinet/in_systm.h>
134: #include <netinet/ip.h>
135: #include <netinet/in_pcb.h>
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
1.44 thorpej 153:
1.1 cgd 154: #ifndef IPFORWARDING
155: #ifdef GATEWAY
156: #define IPFORWARDING 1 /* forward IP packets not for us */
157: #else /* GATEWAY */
158: #define IPFORWARDING 0 /* don't forward IP packets not for us */
159: #endif /* GATEWAY */
160: #endif /* IPFORWARDING */
161: #ifndef IPSENDREDIRECTS
162: #define IPSENDREDIRECTS 1
163: #endif
1.26 thorpej 164: #ifndef IPFORWSRCRT
1.47 cjs 165: #define IPFORWSRCRT 1 /* forward source-routed packets */
166: #endif
167: #ifndef IPALLOWSRCRT
1.48 mrg 168: #define IPALLOWSRCRT 1 /* allow source-routed packets */
1.26 thorpej 169: #endif
1.53 kml 170: #ifndef IPMTUDISC
171: #define IPMTUDISC 0
172: #endif
1.60 kml 173: #ifndef IPMTUDISCTIMEOUT
1.61 kml 174: #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */
1.60 kml 175: #endif
1.53 kml 176:
1.27 thorpej 177: /*
178: * Note: DIRECTED_BROADCAST is handled this way so that previous
179: * configuration using this option will Just Work.
180: */
181: #ifndef IPDIRECTEDBCAST
182: #ifdef DIRECTED_BROADCAST
183: #define IPDIRECTEDBCAST 1
184: #else
185: #define IPDIRECTEDBCAST 0
186: #endif /* DIRECTED_BROADCAST */
187: #endif /* IPDIRECTEDBCAST */
1.1 cgd 188: int ipforwarding = IPFORWARDING;
189: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 190: int ip_defttl = IPDEFTTL;
1.26 thorpej 191: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 192: int ip_directedbcast = IPDIRECTEDBCAST;
1.47 cjs 193: int ip_allowsrcrt = IPALLOWSRCRT;
1.53 kml 194: int ip_mtudisc = IPMTUDISC;
1.60 kml 195: u_int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
1.1 cgd 196: #ifdef DIAGNOSTIC
197: int ipprintfs = 0;
198: #endif
199:
1.60 kml 200: struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
201:
1.1 cgd 202: extern struct domain inetdomain;
203: int ipqmaxlen = IFQ_MAXLEN;
1.22 mycroft 204: struct in_ifaddrhead in_ifaddr;
1.57 tls 205: struct in_ifaddrhashhead *in_ifaddrhashtbl;
1.13 mycroft 206: struct ifqueue ipintrq;
1.63 matt 207: struct ipstat ipstat;
208: u_int16_t ip_id;
1.75 thorpej 209:
1.121 thorpej 210: #ifdef PFIL_HOOKS
211: struct pfil_head inet_pfil_hook;
212: #endif
213:
1.63 matt 214: struct ipqhead ipq;
1.75 thorpej 215: int ipq_locked;
1.131 itojun 216: int ip_nfragpackets = 0;
1.133 itojun 217: int ip_maxfragpackets = 200;
1.75 thorpej 218:
219: static __inline int ipq_lock_try __P((void));
220: static __inline void ipq_unlock __P((void));
221:
222: static __inline int
223: ipq_lock_try()
224: {
225: int s;
226:
1.132 thorpej 227: /*
228: * Use splvm() -- we're bloking things that would cause
229: * mbuf allocation.
230: */
231: s = splvm();
1.75 thorpej 232: if (ipq_locked) {
233: splx(s);
234: return (0);
235: }
236: ipq_locked = 1;
237: splx(s);
238: return (1);
239: }
240:
241: static __inline void
242: ipq_unlock()
243: {
244: int s;
245:
1.132 thorpej 246: s = splvm();
1.75 thorpej 247: ipq_locked = 0;
248: splx(s);
249: }
250:
251: #ifdef DIAGNOSTIC
252: #define IPQ_LOCK() \
253: do { \
254: if (ipq_lock_try() == 0) { \
255: printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \
256: panic("ipq_lock"); \
257: } \
258: } while (0)
259: #define IPQ_LOCK_CHECK() \
260: do { \
261: if (ipq_locked == 0) { \
262: printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \
263: panic("ipq lock check"); \
264: } \
265: } while (0)
266: #else
267: #define IPQ_LOCK() (void) ipq_lock_try()
268: #define IPQ_LOCK_CHECK() /* nothing */
269: #endif
270:
271: #define IPQ_UNLOCK() ipq_unlock()
1.1 cgd 272:
1.72 thorpej 273: struct pool ipqent_pool;
274:
1.135 thorpej 275: #ifdef INET_CSUM_COUNTERS
276: #include <sys/device.h>
277:
278: struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
279: NULL, "inet", "hwcsum bad");
280: struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
281: NULL, "inet", "hwcsum ok");
282: struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
283: NULL, "inet", "swcsum");
284:
285: #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++
286:
287: #else
288:
289: #define INET_CSUM_COUNTER_INCR(ev) /* nothing */
290:
291: #endif /* INET_CSUM_COUNTERS */
292:
1.1 cgd 293: /*
294: * We need to save the IP options in case a protocol wants to respond
295: * to an incoming packet over the same route if the packet got here
296: * using IP source routing. This allows connection establishment and
297: * maintenance when the remote end is on a network that is not known
298: * to us.
299: */
300: int ip_nhops = 0;
301: static struct ip_srcrt {
302: struct in_addr dst; /* final destination */
303: char nop; /* one NOP to align */
304: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
305: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
306: } ip_srcrt;
307:
1.13 mycroft 308: static void save_rte __P((u_char *, struct in_addr));
1.35 mycroft 309:
1.1 cgd 310: /*
311: * IP initialization: fill in IP protocol switch table.
312: * All protocols not implemented in kernel go to raw IP protocol handler.
313: */
1.8 mycroft 314: void
1.1 cgd 315: ip_init()
316: {
1.109 augustss 317: struct protosw *pr;
318: int i;
1.1 cgd 319:
1.72 thorpej 320: pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl",
321: 0, NULL, NULL, M_IPQ);
322:
1.1 cgd 323: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
324: if (pr == 0)
325: panic("ip_init");
326: for (i = 0; i < IPPROTO_MAX; i++)
327: ip_protox[i] = pr - inetsw;
328: for (pr = inetdomain.dom_protosw;
329: pr < inetdomain.dom_protoswNPROTOSW; pr++)
330: if (pr->pr_domain->dom_family == PF_INET &&
331: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
332: ip_protox[pr->pr_protocol] = pr - inetsw;
1.25 cgd 333: LIST_INIT(&ipq);
1.1 cgd 334: ip_id = time.tv_sec & 0xffff;
335: ipintrq.ifq_maxlen = ipqmaxlen;
1.22 mycroft 336: TAILQ_INIT(&in_ifaddr);
1.120 ad 337: in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IFADDR,
338: M_WAITOK, &in_ifaddrhash);
1.60 kml 339: if (ip_mtudisc != 0)
340: ip_mtudisc_timeout_q =
341: rt_timer_queue_create(ip_mtudisc_timeout);
1.73 thorpej 342: #ifdef GATEWAY
343: ipflow_init();
344: #endif
1.121 thorpej 345:
346: #ifdef PFIL_HOOKS
347: /* Register our Packet Filter hook. */
1.126 thorpej 348: inet_pfil_hook.ph_type = PFIL_TYPE_AF;
349: inet_pfil_hook.ph_af = AF_INET;
1.121 thorpej 350: i = pfil_head_register(&inet_pfil_hook);
351: if (i != 0)
352: printf("ip_init: WARNING: unable to register pfil hook, "
353: "error %d\n", i);
354: #endif /* PFIL_HOOKS */
1.135 thorpej 355:
356: #ifdef INET_CSUM_COUNTERS
357: evcnt_attach_static(&ip_hwcsum_bad);
358: evcnt_attach_static(&ip_hwcsum_ok);
359: evcnt_attach_static(&ip_swcsum);
360: #endif /* INET_CSUM_COUNTERS */
1.1 cgd 361: }
362:
363: struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
364: struct route ipforward_rt;
365:
366: /*
1.89 itojun 367: * IP software interrupt routine
368: */
369: void
370: ipintr()
371: {
372: int s;
373: struct mbuf *m;
374:
375: while (1) {
1.132 thorpej 376: s = splnet();
1.89 itojun 377: IF_DEQUEUE(&ipintrq, m);
378: splx(s);
379: if (m == 0)
380: return;
381: ip_input(m);
382: }
383: }
384:
385: /*
1.1 cgd 386: * Ip input routine. Checksum and byte swap header. If fragmented
387: * try to reassemble. Process options. Pass to next level.
388: */
1.8 mycroft 389: void
1.89 itojun 390: ip_input(struct mbuf *m)
1.1 cgd 391: {
1.109 augustss 392: struct ip *ip = NULL;
393: struct ipq *fp;
394: struct in_ifaddr *ia;
395: struct ifaddr *ifa;
1.25 cgd 396: struct ipqent *ipqe;
1.89 itojun 397: int hlen = 0, mff, len;
1.100 itojun 398: int downmatch;
1.1 cgd 399:
400: #ifdef DIAGNOSTIC
401: if ((m->m_flags & M_PKTHDR) == 0)
402: panic("ipintr no HDR");
403: #endif
1.89 itojun 404: #ifdef IPSEC
405: /*
406: * should the inner packet be considered authentic?
407: * see comment in ah4_input().
408: */
409: if (m) {
410: m->m_flags &= ~M_AUTHIPHDR;
411: m->m_flags &= ~M_AUTHIPDGM;
412: }
413: #endif
1.1 cgd 414: /*
415: * If no IP addresses have been set yet but the interfaces
416: * are receiving, can't do anything with incoming packets yet.
417: */
1.140 matt 418: if (TAILQ_FIRST(&in_ifaddr) == 0)
1.1 cgd 419: goto bad;
420: ipstat.ips_total++;
421: if (m->m_len < sizeof (struct ip) &&
422: (m = m_pullup(m, sizeof (struct ip))) == 0) {
423: ipstat.ips_toosmall++;
1.89 itojun 424: return;
1.1 cgd 425: }
426: ip = mtod(m, struct ip *);
1.13 mycroft 427: if (ip->ip_v != IPVERSION) {
428: ipstat.ips_badvers++;
429: goto bad;
430: }
1.1 cgd 431: hlen = ip->ip_hl << 2;
432: if (hlen < sizeof(struct ip)) { /* minimum header length */
433: ipstat.ips_badhlen++;
434: goto bad;
435: }
436: if (hlen > m->m_len) {
437: if ((m = m_pullup(m, hlen)) == 0) {
438: ipstat.ips_badhlen++;
1.89 itojun 439: return;
1.1 cgd 440: }
441: ip = mtod(m, struct ip *);
442: }
1.98 thorpej 443:
1.85 hwr 444: /*
1.99 thorpej 445: * RFC1122: packets with a multicast source address are
1.98 thorpej 446: * not allowed.
1.85 hwr 447: */
448: if (IN_MULTICAST(ip->ip_src.s_addr)) {
1.130 itojun 449: ipstat.ips_badaddr++;
1.85 hwr 450: goto bad;
1.129 itojun 451: }
452:
453: /* 127/8 must not appear on wire - RFC1122 */
454: if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
455: (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
1.130 itojun 456: if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
457: ipstat.ips_badaddr++;
1.129 itojun 458: goto bad;
1.130 itojun 459: }
1.85 hwr 460: }
461:
1.135 thorpej 462: switch (m->m_pkthdr.csum_flags &
1.137 thorpej 463: ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) |
1.135 thorpej 464: M_CSUM_IPv4_BAD)) {
465: case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
466: INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
467: goto badcsum;
468:
469: case M_CSUM_IPv4:
470: /* Checksum was okay. */
471: INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
472: break;
473:
474: default:
475: /* Must compute it ourselves. */
476: INET_CSUM_COUNTER_INCR(&ip_swcsum);
477: if (in_cksum(m, hlen) != 0)
478: goto bad;
479: break;
1.1 cgd 480: }
481:
1.121 thorpej 482: /* Retrieve the packet length. */
483: len = ntohs(ip->ip_len);
1.81 proff 484:
485: /*
486: * Check for additional length bogosity
487: */
1.84 proff 488: if (len < hlen) {
1.81 proff 489: ipstat.ips_badlen++;
490: goto bad;
491: }
1.1 cgd 492:
493: /*
494: * Check that the amount of data in the buffers
495: * is as at least much as the IP header would have us expect.
496: * Trim mbufs if longer than we expect.
497: * Drop packet if shorter than we expect.
498: */
1.35 mycroft 499: if (m->m_pkthdr.len < len) {
1.1 cgd 500: ipstat.ips_tooshort++;
501: goto bad;
502: }
1.35 mycroft 503: if (m->m_pkthdr.len > len) {
1.1 cgd 504: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 505: m->m_len = len;
506: m->m_pkthdr.len = len;
1.1 cgd 507: } else
1.35 mycroft 508: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 509: }
510:
1.94 itojun 511: #ifdef IPSEC
512: /* ipflow (IP fast fowarding) is not compatible with IPsec. */
513: m->m_flags &= ~M_CANFASTFWD;
514: #else
1.64 thorpej 515: /*
516: * Assume that we can create a fast-forward IP flow entry
517: * based on this packet.
518: */
519: m->m_flags |= M_CANFASTFWD;
1.94 itojun 520: #endif
1.64 thorpej 521:
1.36 mrg 522: #ifdef PFIL_HOOKS
1.33 mrg 523: /*
1.64 thorpej 524: * Run through list of hooks for input packets. If there are any
525: * filters which require that additional packets in the flow are
526: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
527: * Note that filters must _never_ set this flag, as another filter
528: * in the list may have previously cleared it.
1.33 mrg 529: */
1.127 itojun 530: /*
531: * let ipfilter look at packet on the wire,
532: * not the decapsulated packet.
533: */
534: #ifdef IPSEC
1.136 itojun 535: if (!ipsec_getnhist(m))
1.127 itojun 536: #else
537: if (1)
538: #endif
539: {
540: if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif,
541: PFIL_IN) != 0)
1.121 thorpej 542: return;
1.127 itojun 543: if (m == NULL)
544: return;
545: ip = mtod(m, struct ip *);
1.142 darrenr 546: hlen = ip->ip_hl << 2;
1.127 itojun 547: }
1.36 mrg 548: #endif /* PFIL_HOOKS */
1.123 thorpej 549:
550: #ifdef ALTQ
551: /* XXX Temporary until ALTQ is changed to use a pfil hook */
552: if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) {
553: /* packet dropped by traffic conditioner */
554: return;
555: }
556: #endif
1.121 thorpej 557:
558: /*
559: * Convert fields to host representation.
560: */
561: NTOHS(ip->ip_len);
562: NTOHS(ip->ip_off);
1.33 mrg 563:
1.1 cgd 564: /*
565: * Process options and, if not destined for us,
566: * ship it on. ip_dooptions returns 1 when an
567: * error was detected (causing an icmp message
568: * to be sent and the original packet to be freed).
569: */
570: ip_nhops = 0; /* for source routed packets */
571: if (hlen > sizeof (struct ip) && ip_dooptions(m))
1.89 itojun 572: return;
1.1 cgd 573:
574: /*
575: * Check our list of addresses, to see if the packet is for us.
1.100 itojun 576: *
577: * Traditional 4.4BSD did not consult IFF_UP at all.
578: * The behavior here is to treat addresses on !IFF_UP interface
579: * as not mine.
1.1 cgd 580: */
1.100 itojun 581: downmatch = 0;
1.140 matt 582: LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) {
1.97 itojun 583: if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
584: if ((ia->ia_ifp->if_flags & IFF_UP) != 0)
585: break;
1.100 itojun 586: else
587: downmatch++;
1.97 itojun 588: }
589: }
1.86 thorpej 590: if (ia != NULL)
591: goto ours;
1.57 tls 592: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
1.140 matt 593: TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) {
594: if (ifa->ifa_addr->sa_family != AF_INET)
595: continue;
1.57 tls 596: ia = ifatoia(ifa);
1.35 mycroft 597: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
598: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 599: /*
600: * Look for all-0's host part (old broadcast addr),
601: * either for subnet or net.
602: */
603: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 604: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 605: goto ours;
1.57 tls 606: /*
607: * An interface with IP address zero accepts
608: * all packets that arrive on that interface.
609: */
610: if (in_nullhost(ia->ia_addr.sin_addr))
611: goto ours;
1.1 cgd 612: }
613: }
1.18 mycroft 614: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 615: struct in_multi *inm;
616: #ifdef MROUTING
617: extern struct socket *ip_mrouter;
1.10 brezak 618:
619: if (m->m_flags & M_EXT) {
620: if ((m = m_pullup(m, hlen)) == 0) {
621: ipstat.ips_toosmall++;
1.89 itojun 622: return;
1.10 brezak 623: }
624: ip = mtod(m, struct ip *);
625: }
1.4 hpeyerl 626:
627: if (ip_mrouter) {
628: /*
629: * If we are acting as a multicast router, all
630: * incoming multicast packets are passed to the
631: * kernel-level multicast forwarding function.
632: * The packet is returned (relatively) intact; if
633: * ip_mforward() returns a non-zero value, the packet
634: * must be discarded, else it may be accepted below.
635: *
636: * (The IP ident field is put in the same byte order
637: * as expected when ip_mforward() is called from
638: * ip_output().)
639: */
1.13 mycroft 640: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
641: ipstat.ips_cantforward++;
1.4 hpeyerl 642: m_freem(m);
1.89 itojun 643: return;
1.4 hpeyerl 644: }
645:
646: /*
647: * The process-level routing demon needs to receive
648: * all multicast IGMP packets, whether or not this
649: * host belongs to their destination groups.
650: */
651: if (ip->ip_p == IPPROTO_IGMP)
652: goto ours;
1.13 mycroft 653: ipstat.ips_forward++;
1.4 hpeyerl 654: }
655: #endif
656: /*
657: * See if we belong to the destination multicast group on the
658: * arrival interface.
659: */
660: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
661: if (inm == NULL) {
1.13 mycroft 662: ipstat.ips_cantforward++;
1.4 hpeyerl 663: m_freem(m);
1.89 itojun 664: return;
1.4 hpeyerl 665: }
666: goto ours;
667: }
1.19 mycroft 668: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 669: in_nullhost(ip->ip_dst))
1.1 cgd 670: goto ours;
671:
672: /*
673: * Not for us; forward if possible and desirable.
674: */
675: if (ipforwarding == 0) {
676: ipstat.ips_cantforward++;
677: m_freem(m);
1.100 itojun 678: } else {
679: /*
680: * If ip_dst matched any of my address on !IFF_UP interface,
681: * and there's no IFF_UP interface that matches ip_dst,
682: * send icmp unreach. Forwarding it will result in in-kernel
683: * forwarding loop till TTL goes to 0.
684: */
685: if (downmatch) {
686: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
687: ipstat.ips_cantforward++;
688: return;
689: }
1.1 cgd 690: ip_forward(m, 0);
1.100 itojun 691: }
1.89 itojun 692: return;
1.1 cgd 693:
694: ours:
695: /*
696: * If offset or IP_MF are set, must reassemble.
697: * Otherwise, nothing need be done.
698: * (We could look in the reassembly queue to see
699: * if the packet was previously fragmented,
700: * but it's not worth the time; just let them time out.)
701: */
1.37 perry 702: if (ip->ip_off & ~(IP_DF|IP_RF)) {
1.1 cgd 703: /*
704: * Look for queue of fragments
705: * of this datagram.
706: */
1.75 thorpej 707: IPQ_LOCK();
1.140 matt 708: LIST_FOREACH(fp, &ipq, ipq_q)
1.1 cgd 709: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 710: in_hosteq(ip->ip_src, fp->ipq_src) &&
711: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 712: ip->ip_p == fp->ipq_p)
713: goto found;
714: fp = 0;
715: found:
716:
717: /*
718: * Adjust ip_len to not reflect header,
1.25 cgd 719: * set ipqe_mff if more fragments are expected,
1.1 cgd 720: * convert offset of this to bytes.
721: */
722: ip->ip_len -= hlen;
1.25 cgd 723: mff = (ip->ip_off & IP_MF) != 0;
724: if (mff) {
1.16 cgd 725: /*
726: * Make sure that fragments have a data length
727: * that's a non-zero multiple of 8 bytes.
728: */
1.17 cgd 729: if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
1.16 cgd 730: ipstat.ips_badfrags++;
1.75 thorpej 731: IPQ_UNLOCK();
1.16 cgd 732: goto bad;
733: }
734: }
1.1 cgd 735: ip->ip_off <<= 3;
736:
737: /*
738: * If datagram marked as having more fragments
739: * or if this is not the first fragment,
740: * attempt reassembly; if it succeeds, proceed.
741: */
1.25 cgd 742: if (mff || ip->ip_off) {
1.1 cgd 743: ipstat.ips_fragments++;
1.72 thorpej 744: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.25 cgd 745: if (ipqe == NULL) {
746: ipstat.ips_rcvmemdrop++;
1.75 thorpej 747: IPQ_UNLOCK();
1.25 cgd 748: goto bad;
749: }
750: ipqe->ipqe_mff = mff;
1.50 thorpej 751: ipqe->ipqe_m = m;
1.25 cgd 752: ipqe->ipqe_ip = ip;
1.50 thorpej 753: m = ip_reass(ipqe, fp);
1.75 thorpej 754: if (m == 0) {
755: IPQ_UNLOCK();
1.89 itojun 756: return;
1.75 thorpej 757: }
1.13 mycroft 758: ipstat.ips_reassembled++;
1.50 thorpej 759: ip = mtod(m, struct ip *);
1.74 thorpej 760: hlen = ip->ip_hl << 2;
1.79 mycroft 761: ip->ip_len += hlen;
1.1 cgd 762: } else
763: if (fp)
764: ip_freef(fp);
1.75 thorpej 765: IPQ_UNLOCK();
1.79 mycroft 766: }
1.128 itojun 767:
768: #ifdef IPSEC
769: /*
770: * enforce IPsec policy checking if we are seeing last header.
771: * note that we do not visit this with protocols with pcb layer
772: * code - like udp/tcp/raw ip.
773: */
774: if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 &&
775: ipsec4_in_reject(m, NULL)) {
776: ipsecstat.in_polvio++;
777: goto bad;
778: }
779: #endif
1.1 cgd 780:
781: /*
782: * Switch out to protocol's input routine.
783: */
1.82 aidan 784: #if IFA_STATS
1.122 itojun 785: if (ia && ip)
786: ia->ia_ifa.ifa_data.ifad_inbytes += ip->ip_len;
1.82 aidan 787: #endif
1.1 cgd 788: ipstat.ips_delivered++;
1.89 itojun 789: {
790: int off = hlen, nh = ip->ip_p;
791:
792: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
793: return;
794: }
1.1 cgd 795: bad:
796: m_freem(m);
1.135 thorpej 797: return;
798:
799: badcsum:
800: ipstat.ips_badsum++;
801: m_freem(m);
1.1 cgd 802: }
803:
804: /*
805: * Take incoming datagram fragment and try to
806: * reassemble it into whole datagram. If a chain for
807: * reassembly of this datagram already exists, then it
808: * is given as fp; otherwise have to make a chain.
809: */
1.50 thorpej 810: struct mbuf *
1.25 cgd 811: ip_reass(ipqe, fp)
1.109 augustss 812: struct ipqent *ipqe;
813: struct ipq *fp;
1.1 cgd 814: {
1.109 augustss 815: struct mbuf *m = ipqe->ipqe_m;
816: struct ipqent *nq, *p, *q;
1.25 cgd 817: struct ip *ip;
1.1 cgd 818: struct mbuf *t;
1.25 cgd 819: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 820: int i, next;
821:
1.75 thorpej 822: IPQ_LOCK_CHECK();
823:
1.1 cgd 824: /*
825: * Presence of header sizes in mbufs
826: * would confuse code below.
827: */
828: m->m_data += hlen;
829: m->m_len -= hlen;
830:
831: /*
832: * If first fragment to arrive, create a reassembly queue.
833: */
834: if (fp == 0) {
1.131 itojun 835: /*
836: * Enforce upper bound on number of fragmented packets
837: * for which we attempt reassembly;
838: * If maxfrag is 0, never accept fragments.
839: * If maxfrag is -1, accept all fragments without limitation.
840: */
841: if (ip_maxfragpackets < 0)
842: ;
843: else if (ip_nfragpackets >= ip_maxfragpackets)
844: goto dropfrag;
845: ip_nfragpackets++;
1.50 thorpej 846: MALLOC(fp, struct ipq *, sizeof (struct ipq),
847: M_FTABLE, M_NOWAIT);
848: if (fp == NULL)
1.1 cgd 849: goto dropfrag;
1.25 cgd 850: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 851: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 852: fp->ipq_p = ipqe->ipqe_ip->ip_p;
853: fp->ipq_id = ipqe->ipqe_ip->ip_id;
854: LIST_INIT(&fp->ipq_fragq);
855: fp->ipq_src = ipqe->ipqe_ip->ip_src;
856: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
857: p = NULL;
1.1 cgd 858: goto insert;
859: }
860:
861: /*
862: * Find a segment which begins after this one does.
863: */
1.140 matt 864: for (p = NULL, q = LIST_FIRST(&fp->ipq_fragq); q != NULL;
865: p = q, q = LIST_NEXT(q, ipqe_q))
1.25 cgd 866: if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off)
1.1 cgd 867: break;
868:
869: /*
870: * If there is a preceding segment, it may provide some of
871: * our data already. If so, drop the data from the incoming
872: * segment. If it provides all of our data, drop us.
873: */
1.25 cgd 874: if (p != NULL) {
875: i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len -
876: ipqe->ipqe_ip->ip_off;
1.1 cgd 877: if (i > 0) {
1.25 cgd 878: if (i >= ipqe->ipqe_ip->ip_len)
1.1 cgd 879: goto dropfrag;
1.50 thorpej 880: m_adj(ipqe->ipqe_m, i);
1.25 cgd 881: ipqe->ipqe_ip->ip_off += i;
882: ipqe->ipqe_ip->ip_len -= i;
1.1 cgd 883: }
884: }
885:
886: /*
887: * While we overlap succeeding segments trim them or,
888: * if they are completely covered, dequeue them.
889: */
1.25 cgd 890: for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len >
891: q->ipqe_ip->ip_off; q = nq) {
892: i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) -
893: q->ipqe_ip->ip_off;
894: if (i < q->ipqe_ip->ip_len) {
895: q->ipqe_ip->ip_len -= i;
896: q->ipqe_ip->ip_off += i;
1.50 thorpej 897: m_adj(q->ipqe_m, i);
1.1 cgd 898: break;
899: }
1.140 matt 900: nq = LIST_NEXT(q, ipqe_q);
1.50 thorpej 901: m_freem(q->ipqe_m);
1.25 cgd 902: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 903: pool_put(&ipqent_pool, q);
1.1 cgd 904: }
905:
906: insert:
907: /*
908: * Stick new segment in its place;
909: * check for complete reassembly.
910: */
1.25 cgd 911: if (p == NULL) {
912: LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
913: } else {
914: LIST_INSERT_AFTER(p, ipqe, ipqe_q);
915: }
1.1 cgd 916: next = 0;
1.140 matt 917: for (p = NULL, q = LIST_FIRST(&fp->ipq_fragq); q != NULL;
918: p = q, q = LIST_NEXT(q, ipqe_q)) {
1.25 cgd 919: if (q->ipqe_ip->ip_off != next)
1.1 cgd 920: return (0);
1.25 cgd 921: next += q->ipqe_ip->ip_len;
1.1 cgd 922: }
1.25 cgd 923: if (p->ipqe_mff)
1.1 cgd 924: return (0);
925:
926: /*
1.41 thorpej 927: * Reassembly is complete. Check for a bogus message size and
928: * concatenate fragments.
1.1 cgd 929: */
1.140 matt 930: q = LIST_FIRST(&fp->ipq_fragq);
1.25 cgd 931: ip = q->ipqe_ip;
1.41 thorpej 932: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
933: ipstat.ips_toolong++;
934: ip_freef(fp);
935: return (0);
936: }
1.50 thorpej 937: m = q->ipqe_m;
1.1 cgd 938: t = m->m_next;
939: m->m_next = 0;
940: m_cat(m, t);
1.140 matt 941: nq = LIST_NEXT(q, ipqe_q);
1.72 thorpej 942: pool_put(&ipqent_pool, q);
1.25 cgd 943: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 944: t = q->ipqe_m;
1.140 matt 945: nq = LIST_NEXT(q, ipqe_q);
1.72 thorpej 946: pool_put(&ipqent_pool, q);
1.1 cgd 947: m_cat(m, t);
948: }
949:
950: /*
951: * Create header for new ip packet by
952: * modifying header of first packet;
953: * dequeue and discard fragment reassembly header.
954: * Make header visible.
955: */
956: ip->ip_len = next;
1.25 cgd 957: ip->ip_src = fp->ipq_src;
958: ip->ip_dst = fp->ipq_dst;
959: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 960: FREE(fp, M_FTABLE);
1.131 itojun 961: ip_nfragpackets--;
1.1 cgd 962: m->m_len += (ip->ip_hl << 2);
963: m->m_data -= (ip->ip_hl << 2);
964: /* some debugging cruft by sklower, below, will go away soon */
965: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
1.109 augustss 966: int plen = 0;
1.50 thorpej 967: for (t = m; t; t = t->m_next)
968: plen += t->m_len;
969: m->m_pkthdr.len = plen;
1.1 cgd 970: }
1.50 thorpej 971: return (m);
1.1 cgd 972:
973: dropfrag:
974: ipstat.ips_fragdropped++;
975: m_freem(m);
1.72 thorpej 976: pool_put(&ipqent_pool, ipqe);
1.1 cgd 977: return (0);
978: }
979:
980: /*
981: * Free a fragment reassembly header and all
982: * associated datagrams.
983: */
1.8 mycroft 984: void
1.1 cgd 985: ip_freef(fp)
986: struct ipq *fp;
987: {
1.109 augustss 988: struct ipqent *q, *p;
1.1 cgd 989:
1.75 thorpej 990: IPQ_LOCK_CHECK();
991:
1.140 matt 992: for (q = LIST_FIRST(&fp->ipq_fragq); q != NULL; q = p) {
993: p = LIST_NEXT(q, ipqe_q);
1.50 thorpej 994: m_freem(q->ipqe_m);
1.25 cgd 995: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 996: pool_put(&ipqent_pool, q);
1.1 cgd 997: }
1.25 cgd 998: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 999: FREE(fp, M_FTABLE);
1.131 itojun 1000: ip_nfragpackets--;
1.1 cgd 1001: }
1002:
1003: /*
1004: * IP timer processing;
1005: * if a timer expires on a reassembly
1006: * queue, discard it.
1007: */
1.8 mycroft 1008: void
1.1 cgd 1009: ip_slowtimo()
1010: {
1.109 augustss 1011: struct ipq *fp, *nfp;
1.24 mycroft 1012: int s = splsoftnet();
1.1 cgd 1013:
1.75 thorpej 1014: IPQ_LOCK();
1.140 matt 1015: for (fp = LIST_FIRST(&ipq); fp != NULL; fp = nfp) {
1016: nfp = LIST_NEXT(fp, ipq_q);
1.25 cgd 1017: if (--fp->ipq_ttl == 0) {
1.1 cgd 1018: ipstat.ips_fragtimeout++;
1.25 cgd 1019: ip_freef(fp);
1.1 cgd 1020: }
1021: }
1.131 itojun 1022: /*
1023: * If we are over the maximum number of fragments
1024: * (due to the limit being lowered), drain off
1025: * enough to get down to the new limit.
1026: */
1027: if (ip_maxfragpackets < 0)
1028: ;
1029: else {
1.140 matt 1030: while (ip_nfragpackets > ip_maxfragpackets && LIST_FIRST(&ipq))
1031: ip_freef(LIST_FIRST(&ipq));
1.131 itojun 1032: }
1.75 thorpej 1033: IPQ_UNLOCK();
1.63 matt 1034: #ifdef GATEWAY
1035: ipflow_slowtimo();
1036: #endif
1.1 cgd 1037: splx(s);
1038: }
1039:
1040: /*
1041: * Drain off all datagram fragments.
1042: */
1.8 mycroft 1043: void
1.1 cgd 1044: ip_drain()
1045: {
1046:
1.75 thorpej 1047: /*
1048: * We may be called from a device's interrupt context. If
1049: * the ipq is already busy, just bail out now.
1050: */
1051: if (ipq_lock_try() == 0)
1052: return;
1053:
1.140 matt 1054: while (LIST_FIRST(&ipq) != NULL) {
1.1 cgd 1055: ipstat.ips_fragdropped++;
1.140 matt 1056: ip_freef(LIST_FIRST(&ipq));
1.1 cgd 1057: }
1.75 thorpej 1058:
1059: IPQ_UNLOCK();
1.1 cgd 1060: }
1061:
1062: /*
1063: * Do option processing on a datagram,
1064: * possibly discarding it if bad options are encountered,
1065: * or forwarding it if source-routed.
1066: * Returns 1 if packet has been forwarded/freed,
1067: * 0 if the packet should be processed further.
1068: */
1.8 mycroft 1069: int
1.1 cgd 1070: ip_dooptions(m)
1071: struct mbuf *m;
1072: {
1.109 augustss 1073: struct ip *ip = mtod(m, struct ip *);
1074: u_char *cp, *cp0;
1075: struct ip_timestamp *ipt;
1076: struct in_ifaddr *ia;
1.1 cgd 1077: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 1078: struct in_addr dst;
1.1 cgd 1079: n_time ntime;
1080:
1.13 mycroft 1081: dst = ip->ip_dst;
1.1 cgd 1082: cp = (u_char *)(ip + 1);
1083: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
1084: for (; cnt > 0; cnt -= optlen, cp += optlen) {
1085: opt = cp[IPOPT_OPTVAL];
1086: if (opt == IPOPT_EOL)
1087: break;
1088: if (opt == IPOPT_NOP)
1089: optlen = 1;
1090: else {
1.113 itojun 1091: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1092: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1093: goto bad;
1094: }
1.1 cgd 1095: optlen = cp[IPOPT_OLEN];
1.114 itojun 1096: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 1097: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1098: goto bad;
1099: }
1100: }
1101: switch (opt) {
1102:
1103: default:
1104: break;
1105:
1106: /*
1107: * Source routing with record.
1108: * Find interface with current destination address.
1109: * If none on this machine then drop if strictly routed,
1110: * or do nothing if loosely routed.
1111: * Record interface address and bring up next address
1112: * component. If strictly routed make sure next
1113: * address is on directly accessible net.
1114: */
1115: case IPOPT_LSRR:
1116: case IPOPT_SSRR:
1.47 cjs 1117: if (ip_allowsrcrt == 0) {
1118: type = ICMP_UNREACH;
1119: code = ICMP_UNREACH_NET_PROHIB;
1120: goto bad;
1121: }
1.114 itojun 1122: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1123: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1124: goto bad;
1125: }
1.1 cgd 1126: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1127: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1128: goto bad;
1129: }
1130: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 1131: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 1132: if (ia == 0) {
1133: if (opt == IPOPT_SSRR) {
1134: type = ICMP_UNREACH;
1135: code = ICMP_UNREACH_SRCFAIL;
1136: goto bad;
1137: }
1138: /*
1139: * Loose routing, and not at next destination
1140: * yet; nothing to do except forward.
1141: */
1142: break;
1143: }
1144: off--; /* 0 origin */
1.112 sommerfe 1145: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 1146: /*
1147: * End of source route. Should be for us.
1148: */
1149: save_rte(cp, ip->ip_src);
1150: break;
1151: }
1152: /*
1153: * locate outgoing interface
1154: */
1155: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
1156: sizeof(ipaddr.sin_addr));
1.96 thorpej 1157: if (opt == IPOPT_SSRR)
1158: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1159: else
1.1 cgd 1160: ia = ip_rtaddr(ipaddr.sin_addr);
1161: if (ia == 0) {
1162: type = ICMP_UNREACH;
1163: code = ICMP_UNREACH_SRCFAIL;
1164: goto bad;
1165: }
1166: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 1167: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 1168: (caddr_t)(cp + off), sizeof(struct in_addr));
1169: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1170: /*
1171: * Let ip_intr's mcast routing check handle mcast pkts
1172: */
1.18 mycroft 1173: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1174: break;
1175:
1176: case IPOPT_RR:
1.114 itojun 1177: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1178: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1179: goto bad;
1180: }
1.1 cgd 1181: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1182: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1183: goto bad;
1184: }
1185: /*
1186: * If no space remains, ignore.
1187: */
1188: off--; /* 0 origin */
1.112 sommerfe 1189: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1190: break;
1191: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
1192: sizeof(ipaddr.sin_addr));
1193: /*
1194: * locate outgoing interface; if we're the destination,
1195: * use the incoming interface (should be same).
1196: */
1.96 thorpej 1197: if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))))
1198: == NULL &&
1199: (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
1.1 cgd 1200: type = ICMP_UNREACH;
1201: code = ICMP_UNREACH_HOST;
1202: goto bad;
1203: }
1.20 mycroft 1204: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 1205: (caddr_t)(cp + off), sizeof(struct in_addr));
1206: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1207: break;
1208:
1209: case IPOPT_TS:
1210: code = cp - (u_char *)ip;
1211: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1212: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1213: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1214: goto bad;
1.114 itojun 1215: }
1216: if (ipt->ipt_ptr < 5) {
1217: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1218: goto bad;
1219: }
1.15 cgd 1220: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1221: if (++ipt->ipt_oflw == 0) {
1222: code = (u_char *)&ipt->ipt_ptr -
1223: (u_char *)ip;
1.1 cgd 1224: goto bad;
1.114 itojun 1225: }
1.1 cgd 1226: break;
1227: }
1.104 thorpej 1228: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1229: switch (ipt->ipt_flg) {
1230:
1231: case IPOPT_TS_TSONLY:
1232: break;
1233:
1234: case IPOPT_TS_TSANDADDR:
1.66 thorpej 1235: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1236: sizeof(struct in_addr) > ipt->ipt_len) {
1237: code = (u_char *)&ipt->ipt_ptr -
1238: (u_char *)ip;
1.1 cgd 1239: goto bad;
1.114 itojun 1240: }
1.13 mycroft 1241: ipaddr.sin_addr = dst;
1.96 thorpej 1242: ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr),
1243: m->m_pkthdr.rcvif));
1.13 mycroft 1244: if (ia == 0)
1245: continue;
1.104 thorpej 1246: bcopy(&ia->ia_addr.sin_addr,
1247: cp0, sizeof(struct in_addr));
1.1 cgd 1248: ipt->ipt_ptr += sizeof(struct in_addr);
1249: break;
1250:
1251: case IPOPT_TS_PRESPEC:
1.66 thorpej 1252: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1253: sizeof(struct in_addr) > ipt->ipt_len) {
1254: code = (u_char *)&ipt->ipt_ptr -
1255: (u_char *)ip;
1.1 cgd 1256: goto bad;
1.114 itojun 1257: }
1.104 thorpej 1258: bcopy(cp0, &ipaddr.sin_addr,
1.1 cgd 1259: sizeof(struct in_addr));
1.96 thorpej 1260: if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))
1261: == NULL)
1.1 cgd 1262: continue;
1263: ipt->ipt_ptr += sizeof(struct in_addr);
1264: break;
1265:
1266: default:
1.114 itojun 1267: /* XXX can't take &ipt->ipt_flg */
1268: code = (u_char *)&ipt->ipt_ptr -
1269: (u_char *)ip + 1;
1.1 cgd 1270: goto bad;
1271: }
1272: ntime = iptime();
1.107 thorpej 1273: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1274: bcopy(cp0, (caddr_t)cp + ipt->ipt_ptr - 1,
1.1 cgd 1275: sizeof(n_time));
1276: ipt->ipt_ptr += sizeof(n_time);
1277: }
1278: }
1279: if (forward) {
1.26 thorpej 1280: if (ip_forwsrcrt == 0) {
1281: type = ICMP_UNREACH;
1282: code = ICMP_UNREACH_SRCFAIL;
1283: goto bad;
1284: }
1.1 cgd 1285: ip_forward(m, 1);
1286: return (1);
1.13 mycroft 1287: }
1288: return (0);
1.1 cgd 1289: bad:
1.13 mycroft 1290: icmp_error(m, type, code, 0, 0);
1291: ipstat.ips_badoptions++;
1.1 cgd 1292: return (1);
1293: }
1294:
1295: /*
1296: * Given address of next destination (final or next hop),
1297: * return internet address info of interface to be used to get there.
1298: */
1299: struct in_ifaddr *
1300: ip_rtaddr(dst)
1301: struct in_addr dst;
1302: {
1.109 augustss 1303: struct sockaddr_in *sin;
1.1 cgd 1304:
1.19 mycroft 1305: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1306:
1.35 mycroft 1307: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 1308: if (ipforward_rt.ro_rt) {
1309: RTFREE(ipforward_rt.ro_rt);
1310: ipforward_rt.ro_rt = 0;
1311: }
1312: sin->sin_family = AF_INET;
1313: sin->sin_len = sizeof(*sin);
1314: sin->sin_addr = dst;
1315:
1316: rtalloc(&ipforward_rt);
1317: }
1318: if (ipforward_rt.ro_rt == 0)
1319: return ((struct in_ifaddr *)0);
1.19 mycroft 1320: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 1321: }
1322:
1323: /*
1324: * Save incoming source route for use in replies,
1325: * to be picked up later by ip_srcroute if the receiver is interested.
1326: */
1.13 mycroft 1327: void
1.1 cgd 1328: save_rte(option, dst)
1329: u_char *option;
1330: struct in_addr dst;
1331: {
1332: unsigned olen;
1333:
1334: olen = option[IPOPT_OLEN];
1335: #ifdef DIAGNOSTIC
1336: if (ipprintfs)
1.39 christos 1337: printf("save_rte: olen %d\n", olen);
1.89 itojun 1338: #endif /* 0 */
1.1 cgd 1339: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1340: return;
1341: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
1342: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1343: ip_srcrt.dst = dst;
1344: }
1345:
1346: /*
1347: * Retrieve incoming source route for use in replies,
1348: * in the same form used by setsockopt.
1349: * The first hop is placed before the options, will be removed later.
1350: */
1351: struct mbuf *
1352: ip_srcroute()
1353: {
1.109 augustss 1354: struct in_addr *p, *q;
1355: struct mbuf *m;
1.1 cgd 1356:
1357: if (ip_nhops == 0)
1358: return ((struct mbuf *)0);
1359: m = m_get(M_DONTWAIT, MT_SOOPTS);
1360: if (m == 0)
1361: return ((struct mbuf *)0);
1362:
1.13 mycroft 1363: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1364:
1365: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1366: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1367: OPTSIZ;
1368: #ifdef DIAGNOSTIC
1369: if (ipprintfs)
1.39 christos 1370: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1371: #endif
1372:
1373: /*
1374: * First save first hop for return route
1375: */
1376: p = &ip_srcrt.route[ip_nhops - 1];
1377: *(mtod(m, struct in_addr *)) = *p--;
1378: #ifdef DIAGNOSTIC
1379: if (ipprintfs)
1.39 christos 1380: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1381: #endif
1382:
1383: /*
1384: * Copy option fields and padding (nop) to mbuf.
1385: */
1386: ip_srcrt.nop = IPOPT_NOP;
1387: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1388: bcopy((caddr_t)&ip_srcrt.nop,
1389: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1390: q = (struct in_addr *)(mtod(m, caddr_t) +
1391: sizeof(struct in_addr) + OPTSIZ);
1392: #undef OPTSIZ
1393: /*
1394: * Record return path as an IP source route,
1395: * reversing the path (pointers are now aligned).
1396: */
1397: while (p >= ip_srcrt.route) {
1398: #ifdef DIAGNOSTIC
1399: if (ipprintfs)
1.39 christos 1400: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1401: #endif
1402: *q++ = *p--;
1403: }
1404: /*
1405: * Last hop goes to final destination.
1406: */
1407: *q = ip_srcrt.dst;
1408: #ifdef DIAGNOSTIC
1409: if (ipprintfs)
1.39 christos 1410: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1411: #endif
1412: return (m);
1413: }
1414:
1415: /*
1416: * Strip out IP options, at higher
1417: * level protocol in the kernel.
1418: * Second argument is buffer to which options
1419: * will be moved, and return value is their length.
1420: * XXX should be deleted; last arg currently ignored.
1421: */
1.8 mycroft 1422: void
1.1 cgd 1423: ip_stripoptions(m, mopt)
1.109 augustss 1424: struct mbuf *m;
1.1 cgd 1425: struct mbuf *mopt;
1426: {
1.109 augustss 1427: int i;
1.1 cgd 1428: struct ip *ip = mtod(m, struct ip *);
1.109 augustss 1429: caddr_t opts;
1.1 cgd 1430: int olen;
1431:
1.79 mycroft 1432: olen = (ip->ip_hl << 2) - sizeof (struct ip);
1.1 cgd 1433: opts = (caddr_t)(ip + 1);
1434: i = m->m_len - (sizeof (struct ip) + olen);
1435: bcopy(opts + olen, opts, (unsigned)i);
1436: m->m_len -= olen;
1437: if (m->m_flags & M_PKTHDR)
1438: m->m_pkthdr.len -= olen;
1.79 mycroft 1439: ip->ip_len -= olen;
1440: ip->ip_hl = sizeof (struct ip) >> 2;
1.1 cgd 1441: }
1442:
1.139 matt 1443: const int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1444: 0, 0, 0, 0,
1445: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1446: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1447: EMSGSIZE, EHOSTUNREACH, 0, 0,
1448: 0, 0, 0, 0,
1449: ENOPROTOOPT
1450: };
1451:
1452: /*
1453: * Forward a packet. If some error occurs return the sender
1454: * an icmp packet. Note we can't always generate a meaningful
1455: * icmp message because icmp doesn't have a large enough repertoire
1456: * of codes and types.
1457: *
1458: * If not forwarding, just drop the packet. This could be confusing
1459: * if ipforwarding was zero but some routing protocol was advancing
1460: * us as a gateway to somewhere. However, we must let the routing
1461: * protocol deal with that.
1462: *
1463: * The srcrt parameter indicates whether the packet is being forwarded
1464: * via a source route.
1465: */
1.13 mycroft 1466: void
1.1 cgd 1467: ip_forward(m, srcrt)
1468: struct mbuf *m;
1469: int srcrt;
1470: {
1.109 augustss 1471: struct ip *ip = mtod(m, struct ip *);
1472: struct sockaddr_in *sin;
1473: struct rtentry *rt;
1.28 christos 1474: int error, type = 0, code = 0;
1.1 cgd 1475: struct mbuf *mcopy;
1.13 mycroft 1476: n_long dest;
1477: struct ifnet *destifp;
1.89 itojun 1478: #ifdef IPSEC
1479: struct ifnet dummyifp;
1480: #endif
1.135 thorpej 1481:
1482: /*
1483: * Clear any in-bound checksum flags for this packet.
1484: */
1485: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1486:
1.13 mycroft 1487: dest = 0;
1.1 cgd 1488: #ifdef DIAGNOSTIC
1489: if (ipprintfs)
1.70 thorpej 1490: printf("forward: src %2.2x dst %2.2x ttl %x\n",
1491: ntohl(ip->ip_src.s_addr),
1492: ntohl(ip->ip_dst.s_addr), ip->ip_ttl);
1.1 cgd 1493: #endif
1.93 sommerfe 1494: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.1 cgd 1495: ipstat.ips_cantforward++;
1496: m_freem(m);
1497: return;
1498: }
1499: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1500: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1501: return;
1502: }
1503: ip->ip_ttl -= IPTTLDEC;
1504:
1.19 mycroft 1505: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1506: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1507: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1508: if (ipforward_rt.ro_rt) {
1509: RTFREE(ipforward_rt.ro_rt);
1510: ipforward_rt.ro_rt = 0;
1511: }
1512: sin->sin_family = AF_INET;
1.35 mycroft 1513: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1514: sin->sin_addr = ip->ip_dst;
1515:
1516: rtalloc(&ipforward_rt);
1517: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1518: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1519: return;
1520: }
1521: rt = ipforward_rt.ro_rt;
1522: }
1523:
1524: /*
1.34 mycroft 1525: * Save at most 68 bytes of the packet in case
1.1 cgd 1526: * we need to generate an ICMP message to the src.
1.119 itojun 1527: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1528: */
1.119 itojun 1529: mcopy = m_copym(m, 0, imin((int)ip->ip_len, 68), M_DONTWAIT);
1530: if (mcopy)
1531: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1532:
1533: /*
1534: * If forwarding packet using same interface that it came in on,
1535: * perhaps should send a redirect to sender to shortcut a hop.
1536: * Only send redirect if source is sending directly to us,
1537: * and if packet was not source routed (or has any options).
1538: * Also, don't send redirect if forwarding using a default route
1539: * or a route modified by a redirect.
1540: */
1541: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1542: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1543: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1544: ipsendredirects && !srcrt) {
1.19 mycroft 1545: if (rt->rt_ifa &&
1546: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1547: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1548: if (rt->rt_flags & RTF_GATEWAY)
1549: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1550: else
1551: dest = ip->ip_dst.s_addr;
1552: /*
1553: * Router requirements says to only send host
1554: * redirects.
1555: */
1556: type = ICMP_REDIRECT;
1557: code = ICMP_REDIRECT_HOST;
1.1 cgd 1558: #ifdef DIAGNOSTIC
1.77 thorpej 1559: if (ipprintfs)
1560: printf("redirect (%d) to %x\n", code,
1561: (u_int32_t)dest);
1.1 cgd 1562: #endif
1563: }
1564: }
1565:
1.89 itojun 1566: #ifdef IPSEC
1.134 lukem 1567: /* Don't lookup socket in forwarding case */
1.127 itojun 1568: (void)ipsec_setsocket(m, NULL);
1.103 itojun 1569: #endif
1.27 thorpej 1570: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1571: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1.1 cgd 1572: if (error)
1573: ipstat.ips_cantforward++;
1574: else {
1575: ipstat.ips_forward++;
1576: if (type)
1577: ipstat.ips_redirectsent++;
1578: else {
1.63 matt 1579: if (mcopy) {
1580: #ifdef GATEWAY
1.64 thorpej 1581: if (mcopy->m_flags & M_CANFASTFWD)
1582: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1583: #endif
1.1 cgd 1584: m_freem(mcopy);
1.63 matt 1585: }
1.1 cgd 1586: return;
1587: }
1588: }
1589: if (mcopy == NULL)
1590: return;
1.13 mycroft 1591: destifp = NULL;
1592:
1.1 cgd 1593: switch (error) {
1594:
1595: case 0: /* forwarded, but need redirect */
1596: /* type, code set above */
1597: break;
1598:
1599: case ENETUNREACH: /* shouldn't happen, checked above */
1600: case EHOSTUNREACH:
1601: case ENETDOWN:
1602: case EHOSTDOWN:
1603: default:
1604: type = ICMP_UNREACH;
1605: code = ICMP_UNREACH_HOST;
1606: break;
1607:
1608: case EMSGSIZE:
1609: type = ICMP_UNREACH;
1610: code = ICMP_UNREACH_NEEDFRAG;
1.89 itojun 1611: #ifndef IPSEC
1.13 mycroft 1612: if (ipforward_rt.ro_rt)
1613: destifp = ipforward_rt.ro_rt->rt_ifp;
1.89 itojun 1614: #else
1615: /*
1616: * If the packet is routed over IPsec tunnel, tell the
1617: * originator the tunnel MTU.
1618: * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1619: * XXX quickhack!!!
1620: */
1621: if (ipforward_rt.ro_rt) {
1622: struct secpolicy *sp;
1623: int ipsecerror;
1.95 itojun 1624: size_t ipsechdr;
1.89 itojun 1625: struct route *ro;
1626:
1627: sp = ipsec4_getpolicybyaddr(mcopy,
1.95 itojun 1628: IPSEC_DIR_OUTBOUND,
1629: IP_FORWARDING,
1630: &ipsecerror);
1.89 itojun 1631:
1632: if (sp == NULL)
1633: destifp = ipforward_rt.ro_rt->rt_ifp;
1634: else {
1635: /* count IPsec header size */
1.95 itojun 1636: ipsechdr = ipsec4_hdrsiz(mcopy,
1637: IPSEC_DIR_OUTBOUND,
1638: NULL);
1.89 itojun 1639:
1640: /*
1641: * find the correct route for outer IPv4
1642: * header, compute tunnel MTU.
1643: *
1644: * XXX BUG ALERT
1645: * The "dummyifp" code relies upon the fact
1646: * that icmp_error() touches only ifp->if_mtu.
1647: */
1648: /*XXX*/
1649: destifp = NULL;
1650: if (sp->req != NULL
1.95 itojun 1651: && sp->req->sav != NULL
1652: && sp->req->sav->sah != NULL) {
1653: ro = &sp->req->sav->sah->sa_route;
1.89 itojun 1654: if (ro->ro_rt && ro->ro_rt->rt_ifp) {
1655: dummyifp.if_mtu =
1656: ro->ro_rt->rt_ifp->if_mtu;
1657: dummyifp.if_mtu -= ipsechdr;
1658: destifp = &dummyifp;
1659: }
1660: }
1661:
1662: key_freesp(sp);
1663: }
1664: }
1665: #endif /*IPSEC*/
1.1 cgd 1666: ipstat.ips_cantfrag++;
1667: break;
1668:
1669: case ENOBUFS:
1.143 itojun 1670: #if 1
1671: /*
1672: * a router should not generate ICMP_SOURCEQUENCH as
1673: * required in RFC1812 Requirements for IP Version 4 Routers.
1674: * source quench could be a big problem under DoS attacks,
1675: * or the underlying interface is rate-limited.
1676: */
1677: if (mcopy)
1678: m_freem(mcopy);
1679: return;
1680: #else
1.1 cgd 1681: type = ICMP_SOURCEQUENCH;
1682: code = 0;
1683: break;
1.143 itojun 1684: #endif
1.1 cgd 1685: }
1.13 mycroft 1686: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 1687: }
1688:
1689: void
1690: ip_savecontrol(inp, mp, ip, m)
1.109 augustss 1691: struct inpcb *inp;
1692: struct mbuf **mp;
1693: struct ip *ip;
1694: struct mbuf *m;
1.44 thorpej 1695: {
1696:
1697: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1698: struct timeval tv;
1699:
1700: microtime(&tv);
1701: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1702: SCM_TIMESTAMP, SOL_SOCKET);
1703: if (*mp)
1704: mp = &(*mp)->m_next;
1705: }
1706: if (inp->inp_flags & INP_RECVDSTADDR) {
1707: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1708: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1709: if (*mp)
1710: mp = &(*mp)->m_next;
1711: }
1712: #ifdef notyet
1713: /*
1714: * XXX
1715: * Moving these out of udp_input() made them even more broken
1716: * than they already were.
1717: * - fenner@parc.xerox.com
1718: */
1719: /* options were tossed already */
1720: if (inp->inp_flags & INP_RECVOPTS) {
1721: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1722: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1723: if (*mp)
1724: mp = &(*mp)->m_next;
1725: }
1726: /* ip_srcroute doesn't do what we want here, need to fix */
1727: if (inp->inp_flags & INP_RECVRETOPTS) {
1728: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1729: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1730: if (*mp)
1731: mp = &(*mp)->m_next;
1732: }
1733: #endif
1734: if (inp->inp_flags & INP_RECVIF) {
1735: struct sockaddr_dl sdl;
1736:
1737: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1738: sdl.sdl_family = AF_LINK;
1739: sdl.sdl_index = m->m_pkthdr.rcvif ?
1740: m->m_pkthdr.rcvif->if_index : 0;
1741: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1742: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1743: IP_RECVIF, IPPROTO_IP);
1744: if (*mp)
1745: mp = &(*mp)->m_next;
1746: }
1.13 mycroft 1747: }
1748:
1749: int
1750: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1751: int *name;
1752: u_int namelen;
1753: void *oldp;
1754: size_t *oldlenp;
1755: void *newp;
1756: size_t newlen;
1757: {
1.88 sommerfe 1758: extern int subnetsarelocal, hostzeroisbroadcast;
1.52 thorpej 1759:
1.54 lukem 1760: int error, old;
1761:
1.13 mycroft 1762: /* All sysctl names at this level are terminal. */
1763: if (namelen != 1)
1764: return (ENOTDIR);
1765:
1766: switch (name[0]) {
1767: case IPCTL_FORWARDING:
1768: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1769: case IPCTL_SENDREDIRECTS:
1770: return (sysctl_int(oldp, oldlenp, newp, newlen,
1771: &ipsendredirects));
1772: case IPCTL_DEFTTL:
1773: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1774: #ifdef notyet
1775: case IPCTL_DEFMTU:
1776: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1777: #endif
1.26 thorpej 1778: case IPCTL_FORWSRCRT:
1.47 cjs 1779: /* Don't allow this to change in a secure environment. */
1.26 thorpej 1780: if (securelevel > 0)
1.46 cjs 1781: return (sysctl_rdint(oldp, oldlenp, newp,
1782: ip_forwsrcrt));
1783: else
1784: return (sysctl_int(oldp, oldlenp, newp, newlen,
1785: &ip_forwsrcrt));
1.27 thorpej 1786: case IPCTL_DIRECTEDBCAST:
1787: return (sysctl_int(oldp, oldlenp, newp, newlen,
1788: &ip_directedbcast));
1.47 cjs 1789: case IPCTL_ALLOWSRCRT:
1790: return (sysctl_int(oldp, oldlenp, newp, newlen,
1791: &ip_allowsrcrt));
1.52 thorpej 1792: case IPCTL_SUBNETSARELOCAL:
1793: return (sysctl_int(oldp, oldlenp, newp, newlen,
1794: &subnetsarelocal));
1.53 kml 1795: case IPCTL_MTUDISC:
1.60 kml 1796: error = sysctl_int(oldp, oldlenp, newp, newlen,
1797: &ip_mtudisc);
1798: if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) {
1799: ip_mtudisc_timeout_q =
1800: rt_timer_queue_create(ip_mtudisc_timeout);
1801: } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) {
1802: rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE);
1803: ip_mtudisc_timeout_q = NULL;
1804: }
1805: return error;
1.54 lukem 1806: case IPCTL_ANONPORTMIN:
1807: old = anonportmin;
1808: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1.118 itojun 1809: if (anonportmin >= anonportmax || anonportmin < 0
1810: || anonportmin > 65535
1.54 lukem 1811: #ifndef IPNOPRIVPORTS
1812: || anonportmin < IPPORT_RESERVED
1813: #endif
1814: ) {
1815: anonportmin = old;
1816: return (EINVAL);
1817: }
1818: return (error);
1819: case IPCTL_ANONPORTMAX:
1820: old = anonportmax;
1821: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1.118 itojun 1822: if (anonportmin >= anonportmax || anonportmax < 0
1823: || anonportmax > 65535
1.54 lukem 1824: #ifndef IPNOPRIVPORTS
1825: || anonportmax < IPPORT_RESERVED
1826: #endif
1827: ) {
1828: anonportmax = old;
1829: return (EINVAL);
1830: }
1.60 kml 1831: return (error);
1832: case IPCTL_MTUDISCTIMEOUT:
1833: error = sysctl_int(oldp, oldlenp, newp, newlen,
1834: &ip_mtudisc_timeout);
1835: if (ip_mtudisc_timeout_q != NULL)
1836: rt_timer_queue_change(ip_mtudisc_timeout_q,
1837: ip_mtudisc_timeout);
1.54 lukem 1838: return (error);
1.65 matt 1839: #ifdef GATEWAY
1840: case IPCTL_MAXFLOWS:
1.67 thorpej 1841: {
1842: int s;
1843:
1.65 matt 1844: error = sysctl_int(oldp, oldlenp, newp, newlen,
1845: &ip_maxflows);
1.67 thorpej 1846: s = splsoftnet();
1.65 matt 1847: ipflow_reap(0);
1.67 thorpej 1848: splx(s);
1.65 matt 1849: return (error);
1.67 thorpej 1850: }
1.89 itojun 1851: #endif
1.90 itojun 1852: case IPCTL_HOSTZEROBROADCAST:
1853: return (sysctl_int(oldp, oldlenp, newp, newlen,
1854: &hostzeroisbroadcast));
1.89 itojun 1855: #if NGIF > 0
1856: case IPCTL_GIF_TTL:
1857: return(sysctl_int(oldp, oldlenp, newp, newlen,
1.90 itojun 1858: &ip_gif_ttl));
1.144 ! martin 1859: #endif
! 1860:
! 1861: #if NGRE > 0
! 1862: case IPCTL_GRE_TTL:
! 1863: return(sysctl_int(oldp, oldlenp, newp, newlen,
! 1864: &ip_gre_ttl));
1.117 tron 1865: #endif
1866:
1867: #ifndef IPNOPRIVPORTS
1868: case IPCTL_LOWPORTMIN:
1869: old = lowportmin;
1870: error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmin);
1871: if (lowportmin >= lowportmax
1872: || lowportmin > IPPORT_RESERVEDMAX
1873: || lowportmin < IPPORT_RESERVEDMIN
1874: ) {
1875: lowportmin = old;
1876: return (EINVAL);
1877: }
1878: return (error);
1879: case IPCTL_LOWPORTMAX:
1880: old = lowportmax;
1881: error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmax);
1882: if (lowportmin >= lowportmax
1883: || lowportmax > IPPORT_RESERVEDMAX
1884: || lowportmax < IPPORT_RESERVEDMIN
1885: ) {
1886: lowportmax = old;
1887: return (EINVAL);
1888: }
1889: return (error);
1.65 matt 1890: #endif
1.131 itojun 1891:
1892: case IPCTL_MAXFRAGPACKETS:
1893: return (sysctl_int(oldp, oldlenp, newp, newlen,
1894: &ip_maxfragpackets));
1.88 sommerfe 1895:
1.13 mycroft 1896: default:
1897: return (EOPNOTSUPP);
1898: }
1899: /* NOTREACHED */
1.1 cgd 1900: }
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