Annotation of src/sys/netinet/ip_input.c, Revision 1.82.2.2.2.1
1.82.2.2.2.1! thorpej 1: /* $NetBSD: ip_input.c,v 1.87 1999/05/04 20:13:09 hwr Exp $ */
1.76 thorpej 2:
3: /*-
4: * Copyright (c) 1998 The NetBSD Foundation, Inc.
5: * All rights reserved.
6: *
7: * This code is derived from software contributed to The NetBSD Foundation
8: * by Public Access Networks Corporation ("Panix"). It was developed under
9: * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon.
10: *
11: * Redistribution and use in source and binary forms, with or without
12: * modification, are permitted provided that the following conditions
13: * are met:
14: * 1. Redistributions of source code must retain the above copyright
15: * notice, this list of conditions and the following disclaimer.
16: * 2. Redistributions in binary form must reproduce the above copyright
17: * notice, this list of conditions and the following disclaimer in the
18: * documentation and/or other materials provided with the distribution.
19: * 3. All advertising materials mentioning features or use of this software
20: * must display the following acknowledgement:
21: * This product includes software developed by the NetBSD
22: * Foundation, Inc. and its contributors.
23: * 4. Neither the name of The NetBSD Foundation nor the names of its
24: * contributors may be used to endorse or promote products derived
25: * from this software without specific prior written permission.
26: *
27: * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
28: * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
29: * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
30: * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
31: * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
32: * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
33: * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
34: * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
35: * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
36: * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
37: * POSSIBILITY OF SUCH DAMAGE.
38: */
1.14 cgd 39:
1.1 cgd 40: /*
1.13 mycroft 41: * Copyright (c) 1982, 1986, 1988, 1993
42: * The Regents of the University of California. All rights reserved.
1.1 cgd 43: *
44: * Redistribution and use in source and binary forms, with or without
45: * modification, are permitted provided that the following conditions
46: * are met:
47: * 1. Redistributions of source code must retain the above copyright
48: * notice, this list of conditions and the following disclaimer.
49: * 2. Redistributions in binary form must reproduce the above copyright
50: * notice, this list of conditions and the following disclaimer in the
51: * documentation and/or other materials provided with the distribution.
52: * 3. All advertising materials mentioning features or use of this software
53: * must display the following acknowledgement:
54: * This product includes software developed by the University of
55: * California, Berkeley and its contributors.
56: * 4. Neither the name of the University nor the names of its contributors
57: * may be used to endorse or promote products derived from this software
58: * without specific prior written permission.
59: *
60: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
61: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
62: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
63: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
64: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
65: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
66: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
67: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
68: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
69: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
70: * SUCH DAMAGE.
71: *
1.14 cgd 72: * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
1.1 cgd 73: */
1.55 scottr 74:
1.62 matt 75: #include "opt_gateway.h"
1.69 mrg 76: #include "opt_pfil_hooks.h"
1.55 scottr 77: #include "opt_mrouting.h"
1.1 cgd 78:
1.5 mycroft 79: #include <sys/param.h>
80: #include <sys/systm.h>
81: #include <sys/malloc.h>
82: #include <sys/mbuf.h>
83: #include <sys/domain.h>
84: #include <sys/protosw.h>
85: #include <sys/socket.h>
1.44 thorpej 86: #include <sys/socketvar.h>
1.5 mycroft 87: #include <sys/errno.h>
88: #include <sys/time.h>
89: #include <sys/kernel.h>
1.28 christos 90: #include <sys/proc.h>
1.72 thorpej 91: #include <sys/pool.h>
1.28 christos 92:
93: #include <vm/vm.h>
94: #include <sys/sysctl.h>
1.1 cgd 95:
1.5 mycroft 96: #include <net/if.h>
1.44 thorpej 97: #include <net/if_dl.h>
1.5 mycroft 98: #include <net/route.h>
1.45 mrg 99: #include <net/pfil.h>
1.1 cgd 100:
1.5 mycroft 101: #include <netinet/in.h>
102: #include <netinet/in_systm.h>
103: #include <netinet/ip.h>
104: #include <netinet/in_pcb.h>
105: #include <netinet/in_var.h>
106: #include <netinet/ip_var.h>
107: #include <netinet/ip_icmp.h>
1.44 thorpej 108:
1.1 cgd 109: #ifndef IPFORWARDING
110: #ifdef GATEWAY
111: #define IPFORWARDING 1 /* forward IP packets not for us */
112: #else /* GATEWAY */
113: #define IPFORWARDING 0 /* don't forward IP packets not for us */
114: #endif /* GATEWAY */
115: #endif /* IPFORWARDING */
116: #ifndef IPSENDREDIRECTS
117: #define IPSENDREDIRECTS 1
118: #endif
1.26 thorpej 119: #ifndef IPFORWSRCRT
1.47 cjs 120: #define IPFORWSRCRT 1 /* forward source-routed packets */
121: #endif
122: #ifndef IPALLOWSRCRT
1.48 mrg 123: #define IPALLOWSRCRT 1 /* allow source-routed packets */
1.26 thorpej 124: #endif
1.53 kml 125: #ifndef IPMTUDISC
126: #define IPMTUDISC 0
127: #endif
1.60 kml 128: #ifndef IPMTUDISCTIMEOUT
1.61 kml 129: #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */
1.60 kml 130: #endif
1.53 kml 131:
1.27 thorpej 132: /*
133: * Note: DIRECTED_BROADCAST is handled this way so that previous
134: * configuration using this option will Just Work.
135: */
136: #ifndef IPDIRECTEDBCAST
137: #ifdef DIRECTED_BROADCAST
138: #define IPDIRECTEDBCAST 1
139: #else
140: #define IPDIRECTEDBCAST 0
141: #endif /* DIRECTED_BROADCAST */
142: #endif /* IPDIRECTEDBCAST */
1.1 cgd 143: int ipforwarding = IPFORWARDING;
144: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 145: int ip_defttl = IPDEFTTL;
1.26 thorpej 146: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 147: int ip_directedbcast = IPDIRECTEDBCAST;
1.47 cjs 148: int ip_allowsrcrt = IPALLOWSRCRT;
1.53 kml 149: int ip_mtudisc = IPMTUDISC;
1.60 kml 150: u_int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
1.1 cgd 151: #ifdef DIAGNOSTIC
152: int ipprintfs = 0;
153: #endif
154:
1.60 kml 155: struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
156:
1.1 cgd 157: extern struct domain inetdomain;
158: extern struct protosw inetsw[];
159: u_char ip_protox[IPPROTO_MAX];
160: int ipqmaxlen = IFQ_MAXLEN;
1.22 mycroft 161: struct in_ifaddrhead in_ifaddr;
1.57 tls 162: struct in_ifaddrhashhead *in_ifaddrhashtbl;
1.13 mycroft 163: struct ifqueue ipintrq;
1.63 matt 164: struct ipstat ipstat;
165: u_int16_t ip_id;
166: int ip_defttl;
1.75 thorpej 167:
1.63 matt 168: struct ipqhead ipq;
1.75 thorpej 169: int ipq_locked;
170:
171: static __inline int ipq_lock_try __P((void));
172: static __inline void ipq_unlock __P((void));
173:
174: static __inline int
175: ipq_lock_try()
176: {
177: int s;
178:
179: s = splimp();
180: if (ipq_locked) {
181: splx(s);
182: return (0);
183: }
184: ipq_locked = 1;
185: splx(s);
186: return (1);
187: }
188:
189: static __inline void
190: ipq_unlock()
191: {
192: int s;
193:
194: s = splimp();
195: ipq_locked = 0;
196: splx(s);
197: }
198:
199: #ifdef DIAGNOSTIC
200: #define IPQ_LOCK() \
201: do { \
202: if (ipq_lock_try() == 0) { \
203: printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \
204: panic("ipq_lock"); \
205: } \
206: } while (0)
207: #define IPQ_LOCK_CHECK() \
208: do { \
209: if (ipq_locked == 0) { \
210: printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \
211: panic("ipq lock check"); \
212: } \
213: } while (0)
214: #else
215: #define IPQ_LOCK() (void) ipq_lock_try()
216: #define IPQ_LOCK_CHECK() /* nothing */
217: #endif
218:
219: #define IPQ_UNLOCK() ipq_unlock()
1.1 cgd 220:
1.72 thorpej 221: struct pool ipqent_pool;
222:
1.1 cgd 223: /*
224: * We need to save the IP options in case a protocol wants to respond
225: * to an incoming packet over the same route if the packet got here
226: * using IP source routing. This allows connection establishment and
227: * maintenance when the remote end is on a network that is not known
228: * to us.
229: */
230: int ip_nhops = 0;
231: static struct ip_srcrt {
232: struct in_addr dst; /* final destination */
233: char nop; /* one NOP to align */
234: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
235: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
236: } ip_srcrt;
237:
1.13 mycroft 238: static void save_rte __P((u_char *, struct in_addr));
1.35 mycroft 239:
1.1 cgd 240: /*
241: * IP initialization: fill in IP protocol switch table.
242: * All protocols not implemented in kernel go to raw IP protocol handler.
243: */
1.8 mycroft 244: void
1.1 cgd 245: ip_init()
246: {
247: register struct protosw *pr;
248: register int i;
249:
1.72 thorpej 250: pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl",
251: 0, NULL, NULL, M_IPQ);
252:
1.1 cgd 253: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
254: if (pr == 0)
255: panic("ip_init");
256: for (i = 0; i < IPPROTO_MAX; i++)
257: ip_protox[i] = pr - inetsw;
258: for (pr = inetdomain.dom_protosw;
259: pr < inetdomain.dom_protoswNPROTOSW; pr++)
260: if (pr->pr_domain->dom_family == PF_INET &&
261: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
262: ip_protox[pr->pr_protocol] = pr - inetsw;
1.25 cgd 263: LIST_INIT(&ipq);
1.1 cgd 264: ip_id = time.tv_sec & 0xffff;
265: ipintrq.ifq_maxlen = ipqmaxlen;
1.22 mycroft 266: TAILQ_INIT(&in_ifaddr);
1.57 tls 267: in_ifaddrhashtbl =
268: hashinit(IN_IFADDR_HASH_SIZE, M_IFADDR, M_WAITOK, &in_ifaddrhash);
1.60 kml 269: if (ip_mtudisc != 0)
270: ip_mtudisc_timeout_q =
271: rt_timer_queue_create(ip_mtudisc_timeout);
1.73 thorpej 272: #ifdef GATEWAY
273: ipflow_init();
274: #endif
1.1 cgd 275: }
276:
277: struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
278: struct route ipforward_rt;
279:
280: /*
281: * Ip input routine. Checksum and byte swap header. If fragmented
282: * try to reassemble. Process options. Pass to next level.
283: */
1.8 mycroft 284: void
1.1 cgd 285: ipintr()
286: {
1.33 mrg 287: register struct ip *ip = NULL;
1.1 cgd 288: register struct mbuf *m;
289: register struct ipq *fp;
290: register struct in_ifaddr *ia;
1.57 tls 291: register struct ifaddr *ifa;
1.25 cgd 292: struct ipqent *ipqe;
1.35 mycroft 293: int hlen = 0, mff, len, s;
1.36 mrg 294: #ifdef PFIL_HOOKS
1.33 mrg 295: struct packet_filter_hook *pfh;
296: struct mbuf *m0;
1.43 mrg 297: int rv;
1.36 mrg 298: #endif /* PFIL_HOOKS */
1.1 cgd 299:
300: next:
301: /*
302: * Get next datagram off input queue and get IP header
303: * in first mbuf.
304: */
305: s = splimp();
306: IF_DEQUEUE(&ipintrq, m);
307: splx(s);
1.13 mycroft 308: if (m == 0)
1.1 cgd 309: return;
310: #ifdef DIAGNOSTIC
311: if ((m->m_flags & M_PKTHDR) == 0)
312: panic("ipintr no HDR");
313: #endif
314: /*
315: * If no IP addresses have been set yet but the interfaces
316: * are receiving, can't do anything with incoming packets yet.
317: */
1.22 mycroft 318: if (in_ifaddr.tqh_first == 0)
1.1 cgd 319: goto bad;
320: ipstat.ips_total++;
321: if (m->m_len < sizeof (struct ip) &&
322: (m = m_pullup(m, sizeof (struct ip))) == 0) {
323: ipstat.ips_toosmall++;
324: goto next;
325: }
326: ip = mtod(m, struct ip *);
1.13 mycroft 327: if (ip->ip_v != IPVERSION) {
328: ipstat.ips_badvers++;
329: goto bad;
330: }
1.1 cgd 331: hlen = ip->ip_hl << 2;
332: if (hlen < sizeof(struct ip)) { /* minimum header length */
333: ipstat.ips_badhlen++;
334: goto bad;
335: }
336: if (hlen > m->m_len) {
337: if ((m = m_pullup(m, hlen)) == 0) {
338: ipstat.ips_badhlen++;
339: goto next;
340: }
341: ip = mtod(m, struct ip *);
342: }
1.82.2.2.2.1! thorpej 343: /*
! 344: * we drop packets that have a multicast address as source
! 345: * as wanted by rfc 1112
! 346: */
! 347: if (IN_MULTICAST(ip->ip_src.s_addr)) {
! 348: goto bad;
! 349: }
! 350:
1.78 mycroft 351: if (in_cksum(m, hlen) != 0) {
1.1 cgd 352: ipstat.ips_badsum++;
353: goto bad;
354: }
355:
356: /*
357: * Convert fields to host representation.
358: */
359: NTOHS(ip->ip_len);
360: NTOHS(ip->ip_off);
1.35 mycroft 361: len = ip->ip_len;
1.81 proff 362:
363: /*
364: * Check for additional length bogosity
365: */
1.82.2.2.2.1! thorpej 366: if (len < hlen) {
1.81 proff 367: ipstat.ips_badlen++;
368: goto bad;
369: }
1.1 cgd 370:
371: /*
372: * Check that the amount of data in the buffers
373: * is as at least much as the IP header would have us expect.
374: * Trim mbufs if longer than we expect.
375: * Drop packet if shorter than we expect.
376: */
1.35 mycroft 377: if (m->m_pkthdr.len < len) {
1.1 cgd 378: ipstat.ips_tooshort++;
379: goto bad;
380: }
1.35 mycroft 381: if (m->m_pkthdr.len > len) {
1.1 cgd 382: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 383: m->m_len = len;
384: m->m_pkthdr.len = len;
1.1 cgd 385: } else
1.35 mycroft 386: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 387: }
388:
1.64 thorpej 389: /*
390: * Assume that we can create a fast-forward IP flow entry
391: * based on this packet.
392: */
393: m->m_flags |= M_CANFASTFWD;
394:
1.36 mrg 395: #ifdef PFIL_HOOKS
1.33 mrg 396: /*
1.64 thorpej 397: * Run through list of hooks for input packets. If there are any
398: * filters which require that additional packets in the flow are
399: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
400: * Note that filters must _never_ set this flag, as another filter
401: * in the list may have previously cleared it.
1.33 mrg 402: */
403: m0 = m;
1.59 mrg 404: for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next)
1.33 mrg 405: if (pfh->pfil_func) {
1.43 mrg 406: rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0);
407: if (rv)
1.40 veego 408: goto next;
1.68 sommerfe 409: m = m0;
410: if (m == NULL)
411: goto next;
412: ip = mtod(m, struct ip *);
1.33 mrg 413: }
1.36 mrg 414: #endif /* PFIL_HOOKS */
1.33 mrg 415:
1.1 cgd 416: /*
417: * Process options and, if not destined for us,
418: * ship it on. ip_dooptions returns 1 when an
419: * error was detected (causing an icmp message
420: * to be sent and the original packet to be freed).
421: */
422: ip_nhops = 0; /* for source routed packets */
423: if (hlen > sizeof (struct ip) && ip_dooptions(m))
424: goto next;
425:
426: /*
427: * Check our list of addresses, to see if the packet is for us.
428: */
1.57 tls 429: INADDR_TO_IA(ip->ip_dst, ia);
1.82.2.2 perry 430: if (ia != NULL)
431: goto ours;
1.57 tls 432: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
433: for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
434: ifa != NULL; ifa = ifa->ifa_list.tqe_next) {
435: if (ifa->ifa_addr->sa_family != AF_INET) continue;
436: ia = ifatoia(ifa);
1.35 mycroft 437: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
438: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 439: /*
440: * Look for all-0's host part (old broadcast addr),
441: * either for subnet or net.
442: */
443: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 444: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 445: goto ours;
1.57 tls 446: /*
447: * An interface with IP address zero accepts
448: * all packets that arrive on that interface.
449: */
450: if (in_nullhost(ia->ia_addr.sin_addr))
451: goto ours;
1.1 cgd 452: }
453: }
1.18 mycroft 454: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 455: struct in_multi *inm;
456: #ifdef MROUTING
457: extern struct socket *ip_mrouter;
1.10 brezak 458:
459: if (m->m_flags & M_EXT) {
460: if ((m = m_pullup(m, hlen)) == 0) {
461: ipstat.ips_toosmall++;
462: goto next;
463: }
464: ip = mtod(m, struct ip *);
465: }
1.4 hpeyerl 466:
467: if (ip_mrouter) {
468: /*
469: * If we are acting as a multicast router, all
470: * incoming multicast packets are passed to the
471: * kernel-level multicast forwarding function.
472: * The packet is returned (relatively) intact; if
473: * ip_mforward() returns a non-zero value, the packet
474: * must be discarded, else it may be accepted below.
475: *
476: * (The IP ident field is put in the same byte order
477: * as expected when ip_mforward() is called from
478: * ip_output().)
479: */
1.13 mycroft 480: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
481: ipstat.ips_cantforward++;
1.4 hpeyerl 482: m_freem(m);
483: goto next;
484: }
485:
486: /*
487: * The process-level routing demon needs to receive
488: * all multicast IGMP packets, whether or not this
489: * host belongs to their destination groups.
490: */
491: if (ip->ip_p == IPPROTO_IGMP)
492: goto ours;
1.13 mycroft 493: ipstat.ips_forward++;
1.4 hpeyerl 494: }
495: #endif
496: /*
497: * See if we belong to the destination multicast group on the
498: * arrival interface.
499: */
500: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
501: if (inm == NULL) {
1.13 mycroft 502: ipstat.ips_cantforward++;
1.4 hpeyerl 503: m_freem(m);
504: goto next;
505: }
506: goto ours;
507: }
1.19 mycroft 508: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 509: in_nullhost(ip->ip_dst))
1.1 cgd 510: goto ours;
511:
512: /*
513: * Not for us; forward if possible and desirable.
514: */
515: if (ipforwarding == 0) {
516: ipstat.ips_cantforward++;
517: m_freem(m);
518: } else
519: ip_forward(m, 0);
520: goto next;
521:
522: ours:
523: /*
524: * If offset or IP_MF are set, must reassemble.
525: * Otherwise, nothing need be done.
526: * (We could look in the reassembly queue to see
527: * if the packet was previously fragmented,
528: * but it's not worth the time; just let them time out.)
529: */
1.37 perry 530: if (ip->ip_off & ~(IP_DF|IP_RF)) {
1.1 cgd 531: /*
532: * Look for queue of fragments
533: * of this datagram.
534: */
1.75 thorpej 535: IPQ_LOCK();
1.25 cgd 536: for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next)
1.1 cgd 537: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 538: in_hosteq(ip->ip_src, fp->ipq_src) &&
539: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 540: ip->ip_p == fp->ipq_p)
541: goto found;
542: fp = 0;
543: found:
544:
545: /*
546: * Adjust ip_len to not reflect header,
1.25 cgd 547: * set ipqe_mff if more fragments are expected,
1.1 cgd 548: * convert offset of this to bytes.
549: */
550: ip->ip_len -= hlen;
1.25 cgd 551: mff = (ip->ip_off & IP_MF) != 0;
552: if (mff) {
1.16 cgd 553: /*
554: * Make sure that fragments have a data length
555: * that's a non-zero multiple of 8 bytes.
556: */
1.17 cgd 557: if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
1.16 cgd 558: ipstat.ips_badfrags++;
1.75 thorpej 559: IPQ_UNLOCK();
1.16 cgd 560: goto bad;
561: }
562: }
1.1 cgd 563: ip->ip_off <<= 3;
564:
565: /*
566: * If datagram marked as having more fragments
567: * or if this is not the first fragment,
568: * attempt reassembly; if it succeeds, proceed.
569: */
1.25 cgd 570: if (mff || ip->ip_off) {
1.1 cgd 571: ipstat.ips_fragments++;
1.72 thorpej 572: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.25 cgd 573: if (ipqe == NULL) {
574: ipstat.ips_rcvmemdrop++;
1.75 thorpej 575: IPQ_UNLOCK();
1.25 cgd 576: goto bad;
577: }
578: ipqe->ipqe_mff = mff;
1.50 thorpej 579: ipqe->ipqe_m = m;
1.25 cgd 580: ipqe->ipqe_ip = ip;
1.50 thorpej 581: m = ip_reass(ipqe, fp);
1.75 thorpej 582: if (m == 0) {
583: IPQ_UNLOCK();
1.1 cgd 584: goto next;
1.75 thorpej 585: }
1.13 mycroft 586: ipstat.ips_reassembled++;
1.50 thorpej 587: ip = mtod(m, struct ip *);
1.74 thorpej 588: hlen = ip->ip_hl << 2;
1.79 mycroft 589: ip->ip_len += hlen;
1.1 cgd 590: } else
591: if (fp)
592: ip_freef(fp);
1.75 thorpej 593: IPQ_UNLOCK();
1.79 mycroft 594: }
1.1 cgd 595:
596: /*
597: * Switch out to protocol's input routine.
598: */
1.82 aidan 599: #if IFA_STATS
600: ia->ia_ifa.ifa_data.ifad_inbytes += ip->ip_len;
601: #endif
1.1 cgd 602: ipstat.ips_delivered++;
603: (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
604: goto next;
605: bad:
606: m_freem(m);
607: goto next;
608: }
609:
610: /*
611: * Take incoming datagram fragment and try to
612: * reassemble it into whole datagram. If a chain for
613: * reassembly of this datagram already exists, then it
614: * is given as fp; otherwise have to make a chain.
615: */
1.50 thorpej 616: struct mbuf *
1.25 cgd 617: ip_reass(ipqe, fp)
618: register struct ipqent *ipqe;
1.1 cgd 619: register struct ipq *fp;
620: {
1.50 thorpej 621: register struct mbuf *m = ipqe->ipqe_m;
1.25 cgd 622: register struct ipqent *nq, *p, *q;
623: struct ip *ip;
1.1 cgd 624: struct mbuf *t;
1.25 cgd 625: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 626: int i, next;
627:
1.75 thorpej 628: IPQ_LOCK_CHECK();
629:
1.1 cgd 630: /*
631: * Presence of header sizes in mbufs
632: * would confuse code below.
633: */
634: m->m_data += hlen;
635: m->m_len -= hlen;
636:
637: /*
638: * If first fragment to arrive, create a reassembly queue.
639: */
640: if (fp == 0) {
1.50 thorpej 641: MALLOC(fp, struct ipq *, sizeof (struct ipq),
642: M_FTABLE, M_NOWAIT);
643: if (fp == NULL)
1.1 cgd 644: goto dropfrag;
1.25 cgd 645: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 646: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 647: fp->ipq_p = ipqe->ipqe_ip->ip_p;
648: fp->ipq_id = ipqe->ipqe_ip->ip_id;
649: LIST_INIT(&fp->ipq_fragq);
650: fp->ipq_src = ipqe->ipqe_ip->ip_src;
651: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
652: p = NULL;
1.1 cgd 653: goto insert;
654: }
655:
656: /*
657: * Find a segment which begins after this one does.
658: */
1.25 cgd 659: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
660: p = q, q = q->ipqe_q.le_next)
661: if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off)
1.1 cgd 662: break;
663:
664: /*
665: * If there is a preceding segment, it may provide some of
666: * our data already. If so, drop the data from the incoming
667: * segment. If it provides all of our data, drop us.
668: */
1.25 cgd 669: if (p != NULL) {
670: i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len -
671: ipqe->ipqe_ip->ip_off;
1.1 cgd 672: if (i > 0) {
1.25 cgd 673: if (i >= ipqe->ipqe_ip->ip_len)
1.1 cgd 674: goto dropfrag;
1.50 thorpej 675: m_adj(ipqe->ipqe_m, i);
1.25 cgd 676: ipqe->ipqe_ip->ip_off += i;
677: ipqe->ipqe_ip->ip_len -= i;
1.1 cgd 678: }
679: }
680:
681: /*
682: * While we overlap succeeding segments trim them or,
683: * if they are completely covered, dequeue them.
684: */
1.25 cgd 685: for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len >
686: q->ipqe_ip->ip_off; q = nq) {
687: i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) -
688: q->ipqe_ip->ip_off;
689: if (i < q->ipqe_ip->ip_len) {
690: q->ipqe_ip->ip_len -= i;
691: q->ipqe_ip->ip_off += i;
1.50 thorpej 692: m_adj(q->ipqe_m, i);
1.1 cgd 693: break;
694: }
1.25 cgd 695: nq = q->ipqe_q.le_next;
1.50 thorpej 696: m_freem(q->ipqe_m);
1.25 cgd 697: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 698: pool_put(&ipqent_pool, q);
1.1 cgd 699: }
700:
701: insert:
702: /*
703: * Stick new segment in its place;
704: * check for complete reassembly.
705: */
1.25 cgd 706: if (p == NULL) {
707: LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
708: } else {
709: LIST_INSERT_AFTER(p, ipqe, ipqe_q);
710: }
1.1 cgd 711: next = 0;
1.25 cgd 712: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
713: p = q, q = q->ipqe_q.le_next) {
714: if (q->ipqe_ip->ip_off != next)
1.1 cgd 715: return (0);
1.25 cgd 716: next += q->ipqe_ip->ip_len;
1.1 cgd 717: }
1.25 cgd 718: if (p->ipqe_mff)
1.1 cgd 719: return (0);
720:
721: /*
1.41 thorpej 722: * Reassembly is complete. Check for a bogus message size and
723: * concatenate fragments.
1.1 cgd 724: */
1.25 cgd 725: q = fp->ipq_fragq.lh_first;
726: ip = q->ipqe_ip;
1.41 thorpej 727: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
728: ipstat.ips_toolong++;
729: ip_freef(fp);
730: return (0);
731: }
1.50 thorpej 732: m = q->ipqe_m;
1.1 cgd 733: t = m->m_next;
734: m->m_next = 0;
735: m_cat(m, t);
1.25 cgd 736: nq = q->ipqe_q.le_next;
1.72 thorpej 737: pool_put(&ipqent_pool, q);
1.25 cgd 738: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 739: t = q->ipqe_m;
1.25 cgd 740: nq = q->ipqe_q.le_next;
1.72 thorpej 741: pool_put(&ipqent_pool, q);
1.1 cgd 742: m_cat(m, t);
743: }
744:
745: /*
746: * Create header for new ip packet by
747: * modifying header of first packet;
748: * dequeue and discard fragment reassembly header.
749: * Make header visible.
750: */
751: ip->ip_len = next;
1.25 cgd 752: ip->ip_src = fp->ipq_src;
753: ip->ip_dst = fp->ipq_dst;
754: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 755: FREE(fp, M_FTABLE);
1.1 cgd 756: m->m_len += (ip->ip_hl << 2);
757: m->m_data -= (ip->ip_hl << 2);
758: /* some debugging cruft by sklower, below, will go away soon */
759: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
760: register int plen = 0;
1.50 thorpej 761: for (t = m; t; t = t->m_next)
762: plen += t->m_len;
763: m->m_pkthdr.len = plen;
1.1 cgd 764: }
1.50 thorpej 765: return (m);
1.1 cgd 766:
767: dropfrag:
768: ipstat.ips_fragdropped++;
769: m_freem(m);
1.72 thorpej 770: pool_put(&ipqent_pool, ipqe);
1.1 cgd 771: return (0);
772: }
773:
774: /*
775: * Free a fragment reassembly header and all
776: * associated datagrams.
777: */
1.8 mycroft 778: void
1.1 cgd 779: ip_freef(fp)
780: struct ipq *fp;
781: {
1.25 cgd 782: register struct ipqent *q, *p;
1.1 cgd 783:
1.75 thorpej 784: IPQ_LOCK_CHECK();
785:
1.25 cgd 786: for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) {
787: p = q->ipqe_q.le_next;
1.50 thorpej 788: m_freem(q->ipqe_m);
1.25 cgd 789: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 790: pool_put(&ipqent_pool, q);
1.1 cgd 791: }
1.25 cgd 792: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 793: FREE(fp, M_FTABLE);
1.1 cgd 794: }
795:
796: /*
797: * IP timer processing;
798: * if a timer expires on a reassembly
799: * queue, discard it.
800: */
1.8 mycroft 801: void
1.1 cgd 802: ip_slowtimo()
803: {
1.25 cgd 804: register struct ipq *fp, *nfp;
1.24 mycroft 805: int s = splsoftnet();
1.1 cgd 806:
1.75 thorpej 807: IPQ_LOCK();
1.25 cgd 808: for (fp = ipq.lh_first; fp != NULL; fp = nfp) {
809: nfp = fp->ipq_q.le_next;
810: if (--fp->ipq_ttl == 0) {
1.1 cgd 811: ipstat.ips_fragtimeout++;
1.25 cgd 812: ip_freef(fp);
1.1 cgd 813: }
814: }
1.75 thorpej 815: IPQ_UNLOCK();
1.63 matt 816: #ifdef GATEWAY
817: ipflow_slowtimo();
818: #endif
1.1 cgd 819: splx(s);
820: }
821:
822: /*
823: * Drain off all datagram fragments.
824: */
1.8 mycroft 825: void
1.1 cgd 826: ip_drain()
827: {
828:
1.75 thorpej 829: /*
830: * We may be called from a device's interrupt context. If
831: * the ipq is already busy, just bail out now.
832: */
833: if (ipq_lock_try() == 0)
834: return;
835:
1.25 cgd 836: while (ipq.lh_first != NULL) {
1.1 cgd 837: ipstat.ips_fragdropped++;
1.25 cgd 838: ip_freef(ipq.lh_first);
1.1 cgd 839: }
1.75 thorpej 840:
841: IPQ_UNLOCK();
1.1 cgd 842: }
843:
844: /*
845: * Do option processing on a datagram,
846: * possibly discarding it if bad options are encountered,
847: * or forwarding it if source-routed.
848: * Returns 1 if packet has been forwarded/freed,
849: * 0 if the packet should be processed further.
850: */
1.8 mycroft 851: int
1.1 cgd 852: ip_dooptions(m)
853: struct mbuf *m;
854: {
855: register struct ip *ip = mtod(m, struct ip *);
856: register u_char *cp;
857: register struct ip_timestamp *ipt;
858: register struct in_ifaddr *ia;
859: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.13 mycroft 860: struct in_addr *sin, dst;
1.1 cgd 861: n_time ntime;
862:
1.13 mycroft 863: dst = ip->ip_dst;
1.1 cgd 864: cp = (u_char *)(ip + 1);
865: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
866: for (; cnt > 0; cnt -= optlen, cp += optlen) {
867: opt = cp[IPOPT_OPTVAL];
868: if (opt == IPOPT_EOL)
869: break;
870: if (opt == IPOPT_NOP)
871: optlen = 1;
872: else {
873: optlen = cp[IPOPT_OLEN];
874: if (optlen <= 0 || optlen > cnt) {
875: code = &cp[IPOPT_OLEN] - (u_char *)ip;
876: goto bad;
877: }
878: }
879: switch (opt) {
880:
881: default:
882: break;
883:
884: /*
885: * Source routing with record.
886: * Find interface with current destination address.
887: * If none on this machine then drop if strictly routed,
888: * or do nothing if loosely routed.
889: * Record interface address and bring up next address
890: * component. If strictly routed make sure next
891: * address is on directly accessible net.
892: */
893: case IPOPT_LSRR:
894: case IPOPT_SSRR:
1.47 cjs 895: if (ip_allowsrcrt == 0) {
896: type = ICMP_UNREACH;
897: code = ICMP_UNREACH_NET_PROHIB;
898: goto bad;
899: }
1.1 cgd 900: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
901: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
902: goto bad;
903: }
904: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 905: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 906: if (ia == 0) {
907: if (opt == IPOPT_SSRR) {
908: type = ICMP_UNREACH;
909: code = ICMP_UNREACH_SRCFAIL;
910: goto bad;
911: }
912: /*
913: * Loose routing, and not at next destination
914: * yet; nothing to do except forward.
915: */
916: break;
917: }
918: off--; /* 0 origin */
919: if (off > optlen - sizeof(struct in_addr)) {
920: /*
921: * End of source route. Should be for us.
922: */
923: save_rte(cp, ip->ip_src);
924: break;
925: }
926: /*
927: * locate outgoing interface
928: */
929: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
930: sizeof(ipaddr.sin_addr));
931: if (opt == IPOPT_SSRR) {
932: #define INA struct in_ifaddr *
933: #define SA struct sockaddr *
1.29 mrg 934: ia = (INA)ifa_ifwithladdr((SA)&ipaddr);
1.1 cgd 935: } else
936: ia = ip_rtaddr(ipaddr.sin_addr);
937: if (ia == 0) {
938: type = ICMP_UNREACH;
939: code = ICMP_UNREACH_SRCFAIL;
940: goto bad;
941: }
942: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 943: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 944: (caddr_t)(cp + off), sizeof(struct in_addr));
945: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 946: /*
947: * Let ip_intr's mcast routing check handle mcast pkts
948: */
1.18 mycroft 949: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 950: break;
951:
952: case IPOPT_RR:
953: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
954: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
955: goto bad;
956: }
957: /*
958: * If no space remains, ignore.
959: */
960: off--; /* 0 origin */
961: if (off > optlen - sizeof(struct in_addr))
962: break;
963: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
964: sizeof(ipaddr.sin_addr));
965: /*
966: * locate outgoing interface; if we're the destination,
967: * use the incoming interface (should be same).
968: */
969: if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
970: (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
971: type = ICMP_UNREACH;
972: code = ICMP_UNREACH_HOST;
973: goto bad;
974: }
1.20 mycroft 975: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 976: (caddr_t)(cp + off), sizeof(struct in_addr));
977: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
978: break;
979:
980: case IPOPT_TS:
981: code = cp - (u_char *)ip;
982: ipt = (struct ip_timestamp *)cp;
983: if (ipt->ipt_len < 5)
984: goto bad;
1.15 cgd 985: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.1 cgd 986: if (++ipt->ipt_oflw == 0)
987: goto bad;
988: break;
989: }
990: sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
991: switch (ipt->ipt_flg) {
992:
993: case IPOPT_TS_TSONLY:
994: break;
995:
996: case IPOPT_TS_TSANDADDR:
1.66 thorpej 997: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 998: sizeof(struct in_addr) > ipt->ipt_len)
999: goto bad;
1.13 mycroft 1000: ipaddr.sin_addr = dst;
1001: ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
1002: m->m_pkthdr.rcvif);
1003: if (ia == 0)
1004: continue;
1.20 mycroft 1005: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 1006: (caddr_t)sin, sizeof(struct in_addr));
1007: ipt->ipt_ptr += sizeof(struct in_addr);
1008: break;
1009:
1010: case IPOPT_TS_PRESPEC:
1.66 thorpej 1011: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 1012: sizeof(struct in_addr) > ipt->ipt_len)
1013: goto bad;
1014: bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
1015: sizeof(struct in_addr));
1016: if (ifa_ifwithaddr((SA)&ipaddr) == 0)
1017: continue;
1018: ipt->ipt_ptr += sizeof(struct in_addr);
1019: break;
1020:
1021: default:
1022: goto bad;
1023: }
1024: ntime = iptime();
1025: bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
1026: sizeof(n_time));
1027: ipt->ipt_ptr += sizeof(n_time);
1028: }
1029: }
1030: if (forward) {
1.26 thorpej 1031: if (ip_forwsrcrt == 0) {
1032: type = ICMP_UNREACH;
1033: code = ICMP_UNREACH_SRCFAIL;
1034: goto bad;
1035: }
1.1 cgd 1036: ip_forward(m, 1);
1037: return (1);
1.13 mycroft 1038: }
1039: return (0);
1.1 cgd 1040: bad:
1.13 mycroft 1041: icmp_error(m, type, code, 0, 0);
1042: ipstat.ips_badoptions++;
1.1 cgd 1043: return (1);
1044: }
1045:
1046: /*
1047: * Given address of next destination (final or next hop),
1048: * return internet address info of interface to be used to get there.
1049: */
1050: struct in_ifaddr *
1051: ip_rtaddr(dst)
1052: struct in_addr dst;
1053: {
1054: register struct sockaddr_in *sin;
1055:
1.19 mycroft 1056: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1057:
1.35 mycroft 1058: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 1059: if (ipforward_rt.ro_rt) {
1060: RTFREE(ipforward_rt.ro_rt);
1061: ipforward_rt.ro_rt = 0;
1062: }
1063: sin->sin_family = AF_INET;
1064: sin->sin_len = sizeof(*sin);
1065: sin->sin_addr = dst;
1066:
1067: rtalloc(&ipforward_rt);
1068: }
1069: if (ipforward_rt.ro_rt == 0)
1070: return ((struct in_ifaddr *)0);
1.19 mycroft 1071: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 1072: }
1073:
1074: /*
1075: * Save incoming source route for use in replies,
1076: * to be picked up later by ip_srcroute if the receiver is interested.
1077: */
1.13 mycroft 1078: void
1.1 cgd 1079: save_rte(option, dst)
1080: u_char *option;
1081: struct in_addr dst;
1082: {
1083: unsigned olen;
1084:
1085: olen = option[IPOPT_OLEN];
1086: #ifdef DIAGNOSTIC
1087: if (ipprintfs)
1.39 christos 1088: printf("save_rte: olen %d\n", olen);
1.1 cgd 1089: #endif
1090: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1091: return;
1092: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
1093: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1094: ip_srcrt.dst = dst;
1095: }
1096:
1097: /*
1098: * Retrieve incoming source route for use in replies,
1099: * in the same form used by setsockopt.
1100: * The first hop is placed before the options, will be removed later.
1101: */
1102: struct mbuf *
1103: ip_srcroute()
1104: {
1105: register struct in_addr *p, *q;
1106: register struct mbuf *m;
1107:
1108: if (ip_nhops == 0)
1109: return ((struct mbuf *)0);
1110: m = m_get(M_DONTWAIT, MT_SOOPTS);
1111: if (m == 0)
1112: return ((struct mbuf *)0);
1113:
1.13 mycroft 1114: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1115:
1116: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1117: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1118: OPTSIZ;
1119: #ifdef DIAGNOSTIC
1120: if (ipprintfs)
1.39 christos 1121: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1122: #endif
1123:
1124: /*
1125: * First save first hop for return route
1126: */
1127: p = &ip_srcrt.route[ip_nhops - 1];
1128: *(mtod(m, struct in_addr *)) = *p--;
1129: #ifdef DIAGNOSTIC
1130: if (ipprintfs)
1.39 christos 1131: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1132: #endif
1133:
1134: /*
1135: * Copy option fields and padding (nop) to mbuf.
1136: */
1137: ip_srcrt.nop = IPOPT_NOP;
1138: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1139: bcopy((caddr_t)&ip_srcrt.nop,
1140: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1141: q = (struct in_addr *)(mtod(m, caddr_t) +
1142: sizeof(struct in_addr) + OPTSIZ);
1143: #undef OPTSIZ
1144: /*
1145: * Record return path as an IP source route,
1146: * reversing the path (pointers are now aligned).
1147: */
1148: while (p >= ip_srcrt.route) {
1149: #ifdef DIAGNOSTIC
1150: if (ipprintfs)
1.39 christos 1151: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1152: #endif
1153: *q++ = *p--;
1154: }
1155: /*
1156: * Last hop goes to final destination.
1157: */
1158: *q = ip_srcrt.dst;
1159: #ifdef DIAGNOSTIC
1160: if (ipprintfs)
1.39 christos 1161: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1162: #endif
1163: return (m);
1164: }
1165:
1166: /*
1167: * Strip out IP options, at higher
1168: * level protocol in the kernel.
1169: * Second argument is buffer to which options
1170: * will be moved, and return value is their length.
1171: * XXX should be deleted; last arg currently ignored.
1172: */
1.8 mycroft 1173: void
1.1 cgd 1174: ip_stripoptions(m, mopt)
1175: register struct mbuf *m;
1176: struct mbuf *mopt;
1177: {
1178: register int i;
1179: struct ip *ip = mtod(m, struct ip *);
1180: register caddr_t opts;
1181: int olen;
1182:
1.79 mycroft 1183: olen = (ip->ip_hl << 2) - sizeof (struct ip);
1.1 cgd 1184: opts = (caddr_t)(ip + 1);
1185: i = m->m_len - (sizeof (struct ip) + olen);
1186: bcopy(opts + olen, opts, (unsigned)i);
1187: m->m_len -= olen;
1188: if (m->m_flags & M_PKTHDR)
1189: m->m_pkthdr.len -= olen;
1.79 mycroft 1190: ip->ip_len -= olen;
1191: ip->ip_hl = sizeof (struct ip) >> 2;
1.1 cgd 1192: }
1193:
1.23 mycroft 1194: int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1195: 0, 0, 0, 0,
1196: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1197: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1198: EMSGSIZE, EHOSTUNREACH, 0, 0,
1199: 0, 0, 0, 0,
1200: ENOPROTOOPT
1201: };
1202:
1203: /*
1204: * Forward a packet. If some error occurs return the sender
1205: * an icmp packet. Note we can't always generate a meaningful
1206: * icmp message because icmp doesn't have a large enough repertoire
1207: * of codes and types.
1208: *
1209: * If not forwarding, just drop the packet. This could be confusing
1210: * if ipforwarding was zero but some routing protocol was advancing
1211: * us as a gateway to somewhere. However, we must let the routing
1212: * protocol deal with that.
1213: *
1214: * The srcrt parameter indicates whether the packet is being forwarded
1215: * via a source route.
1216: */
1.13 mycroft 1217: void
1.1 cgd 1218: ip_forward(m, srcrt)
1219: struct mbuf *m;
1220: int srcrt;
1221: {
1222: register struct ip *ip = mtod(m, struct ip *);
1223: register struct sockaddr_in *sin;
1224: register struct rtentry *rt;
1.28 christos 1225: int error, type = 0, code = 0;
1.1 cgd 1226: struct mbuf *mcopy;
1.13 mycroft 1227: n_long dest;
1228: struct ifnet *destifp;
1.1 cgd 1229:
1.13 mycroft 1230: dest = 0;
1.1 cgd 1231: #ifdef DIAGNOSTIC
1232: if (ipprintfs)
1.70 thorpej 1233: printf("forward: src %2.2x dst %2.2x ttl %x\n",
1234: ntohl(ip->ip_src.s_addr),
1235: ntohl(ip->ip_dst.s_addr), ip->ip_ttl);
1.1 cgd 1236: #endif
1237: if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1238: ipstat.ips_cantforward++;
1239: m_freem(m);
1240: return;
1241: }
1242: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1243: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1244: return;
1245: }
1246: ip->ip_ttl -= IPTTLDEC;
1247:
1.19 mycroft 1248: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1249: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1250: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1251: if (ipforward_rt.ro_rt) {
1252: RTFREE(ipforward_rt.ro_rt);
1253: ipforward_rt.ro_rt = 0;
1254: }
1255: sin->sin_family = AF_INET;
1.35 mycroft 1256: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1257: sin->sin_addr = ip->ip_dst;
1258:
1259: rtalloc(&ipforward_rt);
1260: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1261: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1262: return;
1263: }
1264: rt = ipforward_rt.ro_rt;
1265: }
1266:
1267: /*
1.34 mycroft 1268: * Save at most 68 bytes of the packet in case
1.1 cgd 1269: * we need to generate an ICMP message to the src.
1270: */
1.34 mycroft 1271: mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68));
1.1 cgd 1272:
1273: /*
1274: * If forwarding packet using same interface that it came in on,
1275: * perhaps should send a redirect to sender to shortcut a hop.
1276: * Only send redirect if source is sending directly to us,
1277: * and if packet was not source routed (or has any options).
1278: * Also, don't send redirect if forwarding using a default route
1279: * or a route modified by a redirect.
1280: */
1281: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1282: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1283: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1284: ipsendredirects && !srcrt) {
1.19 mycroft 1285: if (rt->rt_ifa &&
1286: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1287: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1288: if (rt->rt_flags & RTF_GATEWAY)
1289: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1290: else
1291: dest = ip->ip_dst.s_addr;
1292: /*
1293: * Router requirements says to only send host
1294: * redirects.
1295: */
1296: type = ICMP_REDIRECT;
1297: code = ICMP_REDIRECT_HOST;
1.1 cgd 1298: #ifdef DIAGNOSTIC
1.77 thorpej 1299: if (ipprintfs)
1300: printf("redirect (%d) to %x\n", code,
1301: (u_int32_t)dest);
1.1 cgd 1302: #endif
1303: }
1304: }
1305:
1.27 thorpej 1306: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1307: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1.1 cgd 1308: if (error)
1309: ipstat.ips_cantforward++;
1310: else {
1311: ipstat.ips_forward++;
1312: if (type)
1313: ipstat.ips_redirectsent++;
1314: else {
1.63 matt 1315: if (mcopy) {
1316: #ifdef GATEWAY
1.64 thorpej 1317: if (mcopy->m_flags & M_CANFASTFWD)
1318: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1319: #endif
1.1 cgd 1320: m_freem(mcopy);
1.63 matt 1321: }
1.1 cgd 1322: return;
1323: }
1324: }
1325: if (mcopy == NULL)
1326: return;
1.13 mycroft 1327: destifp = NULL;
1328:
1.1 cgd 1329: switch (error) {
1330:
1331: case 0: /* forwarded, but need redirect */
1332: /* type, code set above */
1333: break;
1334:
1335: case ENETUNREACH: /* shouldn't happen, checked above */
1336: case EHOSTUNREACH:
1337: case ENETDOWN:
1338: case EHOSTDOWN:
1339: default:
1340: type = ICMP_UNREACH;
1341: code = ICMP_UNREACH_HOST;
1342: break;
1343:
1344: case EMSGSIZE:
1345: type = ICMP_UNREACH;
1346: code = ICMP_UNREACH_NEEDFRAG;
1.13 mycroft 1347: if (ipforward_rt.ro_rt)
1348: destifp = ipforward_rt.ro_rt->rt_ifp;
1.1 cgd 1349: ipstat.ips_cantfrag++;
1350: break;
1351:
1352: case ENOBUFS:
1353: type = ICMP_SOURCEQUENCH;
1354: code = 0;
1355: break;
1356: }
1.13 mycroft 1357: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 1358: }
1359:
1360: void
1361: ip_savecontrol(inp, mp, ip, m)
1362: register struct inpcb *inp;
1363: register struct mbuf **mp;
1364: register struct ip *ip;
1365: register struct mbuf *m;
1366: {
1367:
1368: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1369: struct timeval tv;
1370:
1371: microtime(&tv);
1372: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1373: SCM_TIMESTAMP, SOL_SOCKET);
1374: if (*mp)
1375: mp = &(*mp)->m_next;
1376: }
1377: if (inp->inp_flags & INP_RECVDSTADDR) {
1378: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1379: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1380: if (*mp)
1381: mp = &(*mp)->m_next;
1382: }
1383: #ifdef notyet
1384: /*
1385: * XXX
1386: * Moving these out of udp_input() made them even more broken
1387: * than they already were.
1388: * - fenner@parc.xerox.com
1389: */
1390: /* options were tossed already */
1391: if (inp->inp_flags & INP_RECVOPTS) {
1392: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1393: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1394: if (*mp)
1395: mp = &(*mp)->m_next;
1396: }
1397: /* ip_srcroute doesn't do what we want here, need to fix */
1398: if (inp->inp_flags & INP_RECVRETOPTS) {
1399: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1400: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1401: if (*mp)
1402: mp = &(*mp)->m_next;
1403: }
1404: #endif
1405: if (inp->inp_flags & INP_RECVIF) {
1406: struct sockaddr_dl sdl;
1407:
1408: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1409: sdl.sdl_family = AF_LINK;
1410: sdl.sdl_index = m->m_pkthdr.rcvif ?
1411: m->m_pkthdr.rcvif->if_index : 0;
1412: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1413: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1414: IP_RECVIF, IPPROTO_IP);
1415: if (*mp)
1416: mp = &(*mp)->m_next;
1417: }
1.13 mycroft 1418: }
1419:
1420: int
1421: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1422: int *name;
1423: u_int namelen;
1424: void *oldp;
1425: size_t *oldlenp;
1426: void *newp;
1427: size_t newlen;
1428: {
1.52 thorpej 1429: extern int subnetsarelocal;
1430:
1.54 lukem 1431: int error, old;
1432:
1.13 mycroft 1433: /* All sysctl names at this level are terminal. */
1434: if (namelen != 1)
1435: return (ENOTDIR);
1436:
1437: switch (name[0]) {
1438: case IPCTL_FORWARDING:
1439: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1440: case IPCTL_SENDREDIRECTS:
1441: return (sysctl_int(oldp, oldlenp, newp, newlen,
1442: &ipsendredirects));
1443: case IPCTL_DEFTTL:
1444: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1445: #ifdef notyet
1446: case IPCTL_DEFMTU:
1447: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1448: #endif
1.26 thorpej 1449: case IPCTL_FORWSRCRT:
1.47 cjs 1450: /* Don't allow this to change in a secure environment. */
1.26 thorpej 1451: if (securelevel > 0)
1.46 cjs 1452: return (sysctl_rdint(oldp, oldlenp, newp,
1453: ip_forwsrcrt));
1454: else
1455: return (sysctl_int(oldp, oldlenp, newp, newlen,
1456: &ip_forwsrcrt));
1.27 thorpej 1457: case IPCTL_DIRECTEDBCAST:
1458: return (sysctl_int(oldp, oldlenp, newp, newlen,
1459: &ip_directedbcast));
1.47 cjs 1460: case IPCTL_ALLOWSRCRT:
1461: return (sysctl_int(oldp, oldlenp, newp, newlen,
1462: &ip_allowsrcrt));
1.52 thorpej 1463: case IPCTL_SUBNETSARELOCAL:
1464: return (sysctl_int(oldp, oldlenp, newp, newlen,
1465: &subnetsarelocal));
1.53 kml 1466: case IPCTL_MTUDISC:
1.60 kml 1467: error = sysctl_int(oldp, oldlenp, newp, newlen,
1468: &ip_mtudisc);
1469: if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) {
1470: ip_mtudisc_timeout_q =
1471: rt_timer_queue_create(ip_mtudisc_timeout);
1472: } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) {
1473: rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE);
1474: ip_mtudisc_timeout_q = NULL;
1475: }
1476: return error;
1.54 lukem 1477: case IPCTL_ANONPORTMIN:
1478: old = anonportmin;
1479: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1480: if (anonportmin >= anonportmax || anonportmin > 65535
1481: #ifndef IPNOPRIVPORTS
1482: || anonportmin < IPPORT_RESERVED
1483: #endif
1484: ) {
1485: anonportmin = old;
1486: return (EINVAL);
1487: }
1488: return (error);
1489: case IPCTL_ANONPORTMAX:
1490: old = anonportmax;
1491: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1492: if (anonportmin >= anonportmax || anonportmax > 65535
1493: #ifndef IPNOPRIVPORTS
1494: || anonportmax < IPPORT_RESERVED
1495: #endif
1496: ) {
1497: anonportmax = old;
1498: return (EINVAL);
1499: }
1.60 kml 1500: return (error);
1501: case IPCTL_MTUDISCTIMEOUT:
1502: error = sysctl_int(oldp, oldlenp, newp, newlen,
1503: &ip_mtudisc_timeout);
1504: if (ip_mtudisc_timeout_q != NULL)
1505: rt_timer_queue_change(ip_mtudisc_timeout_q,
1506: ip_mtudisc_timeout);
1.54 lukem 1507: return (error);
1.65 matt 1508: #ifdef GATEWAY
1509: case IPCTL_MAXFLOWS:
1.67 thorpej 1510: {
1511: int s;
1512:
1.65 matt 1513: error = sysctl_int(oldp, oldlenp, newp, newlen,
1514: &ip_maxflows);
1.67 thorpej 1515: s = splsoftnet();
1.65 matt 1516: ipflow_reap(0);
1.67 thorpej 1517: splx(s);
1.65 matt 1518: return (error);
1.67 thorpej 1519: }
1.65 matt 1520: #endif
1.13 mycroft 1521: default:
1522: return (EOPNOTSUPP);
1523: }
1524: /* NOTREACHED */
1.1 cgd 1525: }
CVSweb <webmaster@jp.NetBSD.org>