Annotation of src/sys/netinet/ip_input.c, Revision 1.85
1.85 ! hwr 1: /* $NetBSD: ip_input.c,v 1.84 1999/04/07 05:34:32 proff 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: }
1.85 ! hwr 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: ipstat.ips_odropped++;
! 349: goto bad;
! 350: }
! 351:
1.78 mycroft 352: if (in_cksum(m, hlen) != 0) {
1.1 cgd 353: ipstat.ips_badsum++;
354: goto bad;
355: }
356:
357: /*
358: * Convert fields to host representation.
359: */
360: NTOHS(ip->ip_len);
361: NTOHS(ip->ip_off);
1.35 mycroft 362: len = ip->ip_len;
1.81 proff 363:
364: /*
365: * Check for additional length bogosity
366: */
1.84 proff 367: if (len < hlen) {
1.81 proff 368: ipstat.ips_badlen++;
369: goto bad;
370: }
1.1 cgd 371:
372: /*
373: * Check that the amount of data in the buffers
374: * is as at least much as the IP header would have us expect.
375: * Trim mbufs if longer than we expect.
376: * Drop packet if shorter than we expect.
377: */
1.35 mycroft 378: if (m->m_pkthdr.len < len) {
1.1 cgd 379: ipstat.ips_tooshort++;
380: goto bad;
381: }
1.35 mycroft 382: if (m->m_pkthdr.len > len) {
1.1 cgd 383: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 384: m->m_len = len;
385: m->m_pkthdr.len = len;
1.1 cgd 386: } else
1.35 mycroft 387: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 388: }
389:
1.64 thorpej 390: /*
391: * Assume that we can create a fast-forward IP flow entry
392: * based on this packet.
393: */
394: m->m_flags |= M_CANFASTFWD;
395:
1.36 mrg 396: #ifdef PFIL_HOOKS
1.33 mrg 397: /*
1.64 thorpej 398: * Run through list of hooks for input packets. If there are any
399: * filters which require that additional packets in the flow are
400: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
401: * Note that filters must _never_ set this flag, as another filter
402: * in the list may have previously cleared it.
1.33 mrg 403: */
404: m0 = m;
1.59 mrg 405: for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next)
1.33 mrg 406: if (pfh->pfil_func) {
1.43 mrg 407: rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0);
408: if (rv)
1.40 veego 409: goto next;
1.68 sommerfe 410: m = m0;
411: if (m == NULL)
412: goto next;
413: ip = mtod(m, struct ip *);
1.33 mrg 414: }
1.36 mrg 415: #endif /* PFIL_HOOKS */
1.33 mrg 416:
1.1 cgd 417: /*
418: * Process options and, if not destined for us,
419: * ship it on. ip_dooptions returns 1 when an
420: * error was detected (causing an icmp message
421: * to be sent and the original packet to be freed).
422: */
423: ip_nhops = 0; /* for source routed packets */
424: if (hlen > sizeof (struct ip) && ip_dooptions(m))
425: goto next;
426:
427: /*
428: * Check our list of addresses, to see if the packet is for us.
429: */
1.57 tls 430: INADDR_TO_IA(ip->ip_dst, ia);
1.83 proff 431: if (ia != NULL) {
432: if (ia->ia_ifp->if_flags & IFF_UP)
433: goto ours;
434: }
1.57 tls 435: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
436: for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
437: ifa != NULL; ifa = ifa->ifa_list.tqe_next) {
438: if (ifa->ifa_addr->sa_family != AF_INET) continue;
439: ia = ifatoia(ifa);
1.35 mycroft 440: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
441: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 442: /*
443: * Look for all-0's host part (old broadcast addr),
444: * either for subnet or net.
445: */
446: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 447: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 448: goto ours;
1.57 tls 449: /*
450: * An interface with IP address zero accepts
451: * all packets that arrive on that interface.
452: */
453: if (in_nullhost(ia->ia_addr.sin_addr))
454: goto ours;
1.1 cgd 455: }
456: }
1.18 mycroft 457: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 458: struct in_multi *inm;
459: #ifdef MROUTING
460: extern struct socket *ip_mrouter;
1.10 brezak 461:
462: if (m->m_flags & M_EXT) {
463: if ((m = m_pullup(m, hlen)) == 0) {
464: ipstat.ips_toosmall++;
465: goto next;
466: }
467: ip = mtod(m, struct ip *);
468: }
1.4 hpeyerl 469:
470: if (ip_mrouter) {
471: /*
472: * If we are acting as a multicast router, all
473: * incoming multicast packets are passed to the
474: * kernel-level multicast forwarding function.
475: * The packet is returned (relatively) intact; if
476: * ip_mforward() returns a non-zero value, the packet
477: * must be discarded, else it may be accepted below.
478: *
479: * (The IP ident field is put in the same byte order
480: * as expected when ip_mforward() is called from
481: * ip_output().)
482: */
1.13 mycroft 483: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
484: ipstat.ips_cantforward++;
1.4 hpeyerl 485: m_freem(m);
486: goto next;
487: }
488:
489: /*
490: * The process-level routing demon needs to receive
491: * all multicast IGMP packets, whether or not this
492: * host belongs to their destination groups.
493: */
494: if (ip->ip_p == IPPROTO_IGMP)
495: goto ours;
1.13 mycroft 496: ipstat.ips_forward++;
1.4 hpeyerl 497: }
498: #endif
499: /*
500: * See if we belong to the destination multicast group on the
501: * arrival interface.
502: */
503: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
504: if (inm == NULL) {
1.13 mycroft 505: ipstat.ips_cantforward++;
1.4 hpeyerl 506: m_freem(m);
507: goto next;
508: }
509: goto ours;
510: }
1.19 mycroft 511: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 512: in_nullhost(ip->ip_dst))
1.1 cgd 513: goto ours;
514:
515: /*
516: * Not for us; forward if possible and desirable.
517: */
518: if (ipforwarding == 0) {
519: ipstat.ips_cantforward++;
520: m_freem(m);
521: } else
522: ip_forward(m, 0);
523: goto next;
524:
525: ours:
526: /*
527: * If offset or IP_MF are set, must reassemble.
528: * Otherwise, nothing need be done.
529: * (We could look in the reassembly queue to see
530: * if the packet was previously fragmented,
531: * but it's not worth the time; just let them time out.)
532: */
1.37 perry 533: if (ip->ip_off & ~(IP_DF|IP_RF)) {
1.1 cgd 534: /*
535: * Look for queue of fragments
536: * of this datagram.
537: */
1.75 thorpej 538: IPQ_LOCK();
1.25 cgd 539: for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next)
1.1 cgd 540: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 541: in_hosteq(ip->ip_src, fp->ipq_src) &&
542: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 543: ip->ip_p == fp->ipq_p)
544: goto found;
545: fp = 0;
546: found:
547:
548: /*
549: * Adjust ip_len to not reflect header,
1.25 cgd 550: * set ipqe_mff if more fragments are expected,
1.1 cgd 551: * convert offset of this to bytes.
552: */
553: ip->ip_len -= hlen;
1.25 cgd 554: mff = (ip->ip_off & IP_MF) != 0;
555: if (mff) {
1.16 cgd 556: /*
557: * Make sure that fragments have a data length
558: * that's a non-zero multiple of 8 bytes.
559: */
1.17 cgd 560: if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
1.16 cgd 561: ipstat.ips_badfrags++;
1.75 thorpej 562: IPQ_UNLOCK();
1.16 cgd 563: goto bad;
564: }
565: }
1.1 cgd 566: ip->ip_off <<= 3;
567:
568: /*
569: * If datagram marked as having more fragments
570: * or if this is not the first fragment,
571: * attempt reassembly; if it succeeds, proceed.
572: */
1.25 cgd 573: if (mff || ip->ip_off) {
1.1 cgd 574: ipstat.ips_fragments++;
1.72 thorpej 575: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.25 cgd 576: if (ipqe == NULL) {
577: ipstat.ips_rcvmemdrop++;
1.75 thorpej 578: IPQ_UNLOCK();
1.25 cgd 579: goto bad;
580: }
581: ipqe->ipqe_mff = mff;
1.50 thorpej 582: ipqe->ipqe_m = m;
1.25 cgd 583: ipqe->ipqe_ip = ip;
1.50 thorpej 584: m = ip_reass(ipqe, fp);
1.75 thorpej 585: if (m == 0) {
586: IPQ_UNLOCK();
1.1 cgd 587: goto next;
1.75 thorpej 588: }
1.13 mycroft 589: ipstat.ips_reassembled++;
1.50 thorpej 590: ip = mtod(m, struct ip *);
1.74 thorpej 591: hlen = ip->ip_hl << 2;
1.79 mycroft 592: ip->ip_len += hlen;
1.1 cgd 593: } else
594: if (fp)
595: ip_freef(fp);
1.75 thorpej 596: IPQ_UNLOCK();
1.79 mycroft 597: }
1.1 cgd 598:
599: /*
600: * Switch out to protocol's input routine.
601: */
1.82 aidan 602: #if IFA_STATS
603: ia->ia_ifa.ifa_data.ifad_inbytes += ip->ip_len;
604: #endif
1.1 cgd 605: ipstat.ips_delivered++;
606: (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
607: goto next;
608: bad:
609: m_freem(m);
610: goto next;
611: }
612:
613: /*
614: * Take incoming datagram fragment and try to
615: * reassemble it into whole datagram. If a chain for
616: * reassembly of this datagram already exists, then it
617: * is given as fp; otherwise have to make a chain.
618: */
1.50 thorpej 619: struct mbuf *
1.25 cgd 620: ip_reass(ipqe, fp)
621: register struct ipqent *ipqe;
1.1 cgd 622: register struct ipq *fp;
623: {
1.50 thorpej 624: register struct mbuf *m = ipqe->ipqe_m;
1.25 cgd 625: register struct ipqent *nq, *p, *q;
626: struct ip *ip;
1.1 cgd 627: struct mbuf *t;
1.25 cgd 628: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 629: int i, next;
630:
1.75 thorpej 631: IPQ_LOCK_CHECK();
632:
1.1 cgd 633: /*
634: * Presence of header sizes in mbufs
635: * would confuse code below.
636: */
637: m->m_data += hlen;
638: m->m_len -= hlen;
639:
640: /*
641: * If first fragment to arrive, create a reassembly queue.
642: */
643: if (fp == 0) {
1.50 thorpej 644: MALLOC(fp, struct ipq *, sizeof (struct ipq),
645: M_FTABLE, M_NOWAIT);
646: if (fp == NULL)
1.1 cgd 647: goto dropfrag;
1.25 cgd 648: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 649: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 650: fp->ipq_p = ipqe->ipqe_ip->ip_p;
651: fp->ipq_id = ipqe->ipqe_ip->ip_id;
652: LIST_INIT(&fp->ipq_fragq);
653: fp->ipq_src = ipqe->ipqe_ip->ip_src;
654: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
655: p = NULL;
1.1 cgd 656: goto insert;
657: }
658:
659: /*
660: * Find a segment which begins after this one does.
661: */
1.25 cgd 662: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
663: p = q, q = q->ipqe_q.le_next)
664: if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off)
1.1 cgd 665: break;
666:
667: /*
668: * If there is a preceding segment, it may provide some of
669: * our data already. If so, drop the data from the incoming
670: * segment. If it provides all of our data, drop us.
671: */
1.25 cgd 672: if (p != NULL) {
673: i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len -
674: ipqe->ipqe_ip->ip_off;
1.1 cgd 675: if (i > 0) {
1.25 cgd 676: if (i >= ipqe->ipqe_ip->ip_len)
1.1 cgd 677: goto dropfrag;
1.50 thorpej 678: m_adj(ipqe->ipqe_m, i);
1.25 cgd 679: ipqe->ipqe_ip->ip_off += i;
680: ipqe->ipqe_ip->ip_len -= i;
1.1 cgd 681: }
682: }
683:
684: /*
685: * While we overlap succeeding segments trim them or,
686: * if they are completely covered, dequeue them.
687: */
1.25 cgd 688: for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len >
689: q->ipqe_ip->ip_off; q = nq) {
690: i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) -
691: q->ipqe_ip->ip_off;
692: if (i < q->ipqe_ip->ip_len) {
693: q->ipqe_ip->ip_len -= i;
694: q->ipqe_ip->ip_off += i;
1.50 thorpej 695: m_adj(q->ipqe_m, i);
1.1 cgd 696: break;
697: }
1.25 cgd 698: nq = q->ipqe_q.le_next;
1.50 thorpej 699: m_freem(q->ipqe_m);
1.25 cgd 700: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 701: pool_put(&ipqent_pool, q);
1.1 cgd 702: }
703:
704: insert:
705: /*
706: * Stick new segment in its place;
707: * check for complete reassembly.
708: */
1.25 cgd 709: if (p == NULL) {
710: LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
711: } else {
712: LIST_INSERT_AFTER(p, ipqe, ipqe_q);
713: }
1.1 cgd 714: next = 0;
1.25 cgd 715: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
716: p = q, q = q->ipqe_q.le_next) {
717: if (q->ipqe_ip->ip_off != next)
1.1 cgd 718: return (0);
1.25 cgd 719: next += q->ipqe_ip->ip_len;
1.1 cgd 720: }
1.25 cgd 721: if (p->ipqe_mff)
1.1 cgd 722: return (0);
723:
724: /*
1.41 thorpej 725: * Reassembly is complete. Check for a bogus message size and
726: * concatenate fragments.
1.1 cgd 727: */
1.25 cgd 728: q = fp->ipq_fragq.lh_first;
729: ip = q->ipqe_ip;
1.41 thorpej 730: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
731: ipstat.ips_toolong++;
732: ip_freef(fp);
733: return (0);
734: }
1.50 thorpej 735: m = q->ipqe_m;
1.1 cgd 736: t = m->m_next;
737: m->m_next = 0;
738: m_cat(m, t);
1.25 cgd 739: nq = q->ipqe_q.le_next;
1.72 thorpej 740: pool_put(&ipqent_pool, q);
1.25 cgd 741: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 742: t = q->ipqe_m;
1.25 cgd 743: nq = q->ipqe_q.le_next;
1.72 thorpej 744: pool_put(&ipqent_pool, q);
1.1 cgd 745: m_cat(m, t);
746: }
747:
748: /*
749: * Create header for new ip packet by
750: * modifying header of first packet;
751: * dequeue and discard fragment reassembly header.
752: * Make header visible.
753: */
754: ip->ip_len = next;
1.25 cgd 755: ip->ip_src = fp->ipq_src;
756: ip->ip_dst = fp->ipq_dst;
757: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 758: FREE(fp, M_FTABLE);
1.1 cgd 759: m->m_len += (ip->ip_hl << 2);
760: m->m_data -= (ip->ip_hl << 2);
761: /* some debugging cruft by sklower, below, will go away soon */
762: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
763: register int plen = 0;
1.50 thorpej 764: for (t = m; t; t = t->m_next)
765: plen += t->m_len;
766: m->m_pkthdr.len = plen;
1.1 cgd 767: }
1.50 thorpej 768: return (m);
1.1 cgd 769:
770: dropfrag:
771: ipstat.ips_fragdropped++;
772: m_freem(m);
1.72 thorpej 773: pool_put(&ipqent_pool, ipqe);
1.1 cgd 774: return (0);
775: }
776:
777: /*
778: * Free a fragment reassembly header and all
779: * associated datagrams.
780: */
1.8 mycroft 781: void
1.1 cgd 782: ip_freef(fp)
783: struct ipq *fp;
784: {
1.25 cgd 785: register struct ipqent *q, *p;
1.1 cgd 786:
1.75 thorpej 787: IPQ_LOCK_CHECK();
788:
1.25 cgd 789: for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) {
790: p = q->ipqe_q.le_next;
1.50 thorpej 791: m_freem(q->ipqe_m);
1.25 cgd 792: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 793: pool_put(&ipqent_pool, q);
1.1 cgd 794: }
1.25 cgd 795: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 796: FREE(fp, M_FTABLE);
1.1 cgd 797: }
798:
799: /*
800: * IP timer processing;
801: * if a timer expires on a reassembly
802: * queue, discard it.
803: */
1.8 mycroft 804: void
1.1 cgd 805: ip_slowtimo()
806: {
1.25 cgd 807: register struct ipq *fp, *nfp;
1.24 mycroft 808: int s = splsoftnet();
1.1 cgd 809:
1.75 thorpej 810: IPQ_LOCK();
1.25 cgd 811: for (fp = ipq.lh_first; fp != NULL; fp = nfp) {
812: nfp = fp->ipq_q.le_next;
813: if (--fp->ipq_ttl == 0) {
1.1 cgd 814: ipstat.ips_fragtimeout++;
1.25 cgd 815: ip_freef(fp);
1.1 cgd 816: }
817: }
1.75 thorpej 818: IPQ_UNLOCK();
1.63 matt 819: #ifdef GATEWAY
820: ipflow_slowtimo();
821: #endif
1.1 cgd 822: splx(s);
823: }
824:
825: /*
826: * Drain off all datagram fragments.
827: */
1.8 mycroft 828: void
1.1 cgd 829: ip_drain()
830: {
831:
1.75 thorpej 832: /*
833: * We may be called from a device's interrupt context. If
834: * the ipq is already busy, just bail out now.
835: */
836: if (ipq_lock_try() == 0)
837: return;
838:
1.25 cgd 839: while (ipq.lh_first != NULL) {
1.1 cgd 840: ipstat.ips_fragdropped++;
1.25 cgd 841: ip_freef(ipq.lh_first);
1.1 cgd 842: }
1.75 thorpej 843:
844: IPQ_UNLOCK();
1.1 cgd 845: }
846:
847: /*
848: * Do option processing on a datagram,
849: * possibly discarding it if bad options are encountered,
850: * or forwarding it if source-routed.
851: * Returns 1 if packet has been forwarded/freed,
852: * 0 if the packet should be processed further.
853: */
1.8 mycroft 854: int
1.1 cgd 855: ip_dooptions(m)
856: struct mbuf *m;
857: {
858: register struct ip *ip = mtod(m, struct ip *);
859: register u_char *cp;
860: register struct ip_timestamp *ipt;
861: register struct in_ifaddr *ia;
862: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.13 mycroft 863: struct in_addr *sin, dst;
1.1 cgd 864: n_time ntime;
865:
1.13 mycroft 866: dst = ip->ip_dst;
1.1 cgd 867: cp = (u_char *)(ip + 1);
868: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
869: for (; cnt > 0; cnt -= optlen, cp += optlen) {
870: opt = cp[IPOPT_OPTVAL];
871: if (opt == IPOPT_EOL)
872: break;
873: if (opt == IPOPT_NOP)
874: optlen = 1;
875: else {
876: optlen = cp[IPOPT_OLEN];
877: if (optlen <= 0 || optlen > cnt) {
878: code = &cp[IPOPT_OLEN] - (u_char *)ip;
879: goto bad;
880: }
881: }
882: switch (opt) {
883:
884: default:
885: break;
886:
887: /*
888: * Source routing with record.
889: * Find interface with current destination address.
890: * If none on this machine then drop if strictly routed,
891: * or do nothing if loosely routed.
892: * Record interface address and bring up next address
893: * component. If strictly routed make sure next
894: * address is on directly accessible net.
895: */
896: case IPOPT_LSRR:
897: case IPOPT_SSRR:
1.47 cjs 898: if (ip_allowsrcrt == 0) {
899: type = ICMP_UNREACH;
900: code = ICMP_UNREACH_NET_PROHIB;
901: goto bad;
902: }
1.1 cgd 903: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
904: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
905: goto bad;
906: }
907: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 908: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 909: if (ia == 0) {
910: if (opt == IPOPT_SSRR) {
911: type = ICMP_UNREACH;
912: code = ICMP_UNREACH_SRCFAIL;
913: goto bad;
914: }
915: /*
916: * Loose routing, and not at next destination
917: * yet; nothing to do except forward.
918: */
919: break;
920: }
921: off--; /* 0 origin */
922: if (off > optlen - sizeof(struct in_addr)) {
923: /*
924: * End of source route. Should be for us.
925: */
926: save_rte(cp, ip->ip_src);
927: break;
928: }
929: /*
930: * locate outgoing interface
931: */
932: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
933: sizeof(ipaddr.sin_addr));
934: if (opt == IPOPT_SSRR) {
935: #define INA struct in_ifaddr *
936: #define SA struct sockaddr *
1.29 mrg 937: ia = (INA)ifa_ifwithladdr((SA)&ipaddr);
1.1 cgd 938: } else
939: ia = ip_rtaddr(ipaddr.sin_addr);
940: if (ia == 0) {
941: type = ICMP_UNREACH;
942: code = ICMP_UNREACH_SRCFAIL;
943: goto bad;
944: }
945: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 946: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 947: (caddr_t)(cp + off), sizeof(struct in_addr));
948: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 949: /*
950: * Let ip_intr's mcast routing check handle mcast pkts
951: */
1.18 mycroft 952: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 953: break;
954:
955: case IPOPT_RR:
956: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
957: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
958: goto bad;
959: }
960: /*
961: * If no space remains, ignore.
962: */
963: off--; /* 0 origin */
964: if (off > optlen - sizeof(struct in_addr))
965: break;
966: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
967: sizeof(ipaddr.sin_addr));
968: /*
969: * locate outgoing interface; if we're the destination,
970: * use the incoming interface (should be same).
971: */
972: if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
973: (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
974: type = ICMP_UNREACH;
975: code = ICMP_UNREACH_HOST;
976: goto bad;
977: }
1.20 mycroft 978: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 979: (caddr_t)(cp + off), sizeof(struct in_addr));
980: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
981: break;
982:
983: case IPOPT_TS:
984: code = cp - (u_char *)ip;
985: ipt = (struct ip_timestamp *)cp;
986: if (ipt->ipt_len < 5)
987: goto bad;
1.15 cgd 988: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.1 cgd 989: if (++ipt->ipt_oflw == 0)
990: goto bad;
991: break;
992: }
993: sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
994: switch (ipt->ipt_flg) {
995:
996: case IPOPT_TS_TSONLY:
997: break;
998:
999: case IPOPT_TS_TSANDADDR:
1.66 thorpej 1000: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 1001: sizeof(struct in_addr) > ipt->ipt_len)
1002: goto bad;
1.13 mycroft 1003: ipaddr.sin_addr = dst;
1004: ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
1005: m->m_pkthdr.rcvif);
1006: if (ia == 0)
1007: continue;
1.20 mycroft 1008: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 1009: (caddr_t)sin, sizeof(struct in_addr));
1010: ipt->ipt_ptr += sizeof(struct in_addr);
1011: break;
1012:
1013: case IPOPT_TS_PRESPEC:
1.66 thorpej 1014: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 1015: sizeof(struct in_addr) > ipt->ipt_len)
1016: goto bad;
1017: bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
1018: sizeof(struct in_addr));
1019: if (ifa_ifwithaddr((SA)&ipaddr) == 0)
1020: continue;
1021: ipt->ipt_ptr += sizeof(struct in_addr);
1022: break;
1023:
1024: default:
1025: goto bad;
1026: }
1027: ntime = iptime();
1028: bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
1029: sizeof(n_time));
1030: ipt->ipt_ptr += sizeof(n_time);
1031: }
1032: }
1033: if (forward) {
1.26 thorpej 1034: if (ip_forwsrcrt == 0) {
1035: type = ICMP_UNREACH;
1036: code = ICMP_UNREACH_SRCFAIL;
1037: goto bad;
1038: }
1.1 cgd 1039: ip_forward(m, 1);
1040: return (1);
1.13 mycroft 1041: }
1042: return (0);
1.1 cgd 1043: bad:
1.13 mycroft 1044: icmp_error(m, type, code, 0, 0);
1045: ipstat.ips_badoptions++;
1.1 cgd 1046: return (1);
1047: }
1048:
1049: /*
1050: * Given address of next destination (final or next hop),
1051: * return internet address info of interface to be used to get there.
1052: */
1053: struct in_ifaddr *
1054: ip_rtaddr(dst)
1055: struct in_addr dst;
1056: {
1057: register struct sockaddr_in *sin;
1058:
1.19 mycroft 1059: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1060:
1.35 mycroft 1061: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 1062: if (ipforward_rt.ro_rt) {
1063: RTFREE(ipforward_rt.ro_rt);
1064: ipforward_rt.ro_rt = 0;
1065: }
1066: sin->sin_family = AF_INET;
1067: sin->sin_len = sizeof(*sin);
1068: sin->sin_addr = dst;
1069:
1070: rtalloc(&ipforward_rt);
1071: }
1072: if (ipforward_rt.ro_rt == 0)
1073: return ((struct in_ifaddr *)0);
1.19 mycroft 1074: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 1075: }
1076:
1077: /*
1078: * Save incoming source route for use in replies,
1079: * to be picked up later by ip_srcroute if the receiver is interested.
1080: */
1.13 mycroft 1081: void
1.1 cgd 1082: save_rte(option, dst)
1083: u_char *option;
1084: struct in_addr dst;
1085: {
1086: unsigned olen;
1087:
1088: olen = option[IPOPT_OLEN];
1089: #ifdef DIAGNOSTIC
1090: if (ipprintfs)
1.39 christos 1091: printf("save_rte: olen %d\n", olen);
1.1 cgd 1092: #endif
1093: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1094: return;
1095: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
1096: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1097: ip_srcrt.dst = dst;
1098: }
1099:
1100: /*
1101: * Retrieve incoming source route for use in replies,
1102: * in the same form used by setsockopt.
1103: * The first hop is placed before the options, will be removed later.
1104: */
1105: struct mbuf *
1106: ip_srcroute()
1107: {
1108: register struct in_addr *p, *q;
1109: register struct mbuf *m;
1110:
1111: if (ip_nhops == 0)
1112: return ((struct mbuf *)0);
1113: m = m_get(M_DONTWAIT, MT_SOOPTS);
1114: if (m == 0)
1115: return ((struct mbuf *)0);
1116:
1.13 mycroft 1117: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1118:
1119: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1120: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1121: OPTSIZ;
1122: #ifdef DIAGNOSTIC
1123: if (ipprintfs)
1.39 christos 1124: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1125: #endif
1126:
1127: /*
1128: * First save first hop for return route
1129: */
1130: p = &ip_srcrt.route[ip_nhops - 1];
1131: *(mtod(m, struct in_addr *)) = *p--;
1132: #ifdef DIAGNOSTIC
1133: if (ipprintfs)
1.39 christos 1134: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1135: #endif
1136:
1137: /*
1138: * Copy option fields and padding (nop) to mbuf.
1139: */
1140: ip_srcrt.nop = IPOPT_NOP;
1141: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1142: bcopy((caddr_t)&ip_srcrt.nop,
1143: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1144: q = (struct in_addr *)(mtod(m, caddr_t) +
1145: sizeof(struct in_addr) + OPTSIZ);
1146: #undef OPTSIZ
1147: /*
1148: * Record return path as an IP source route,
1149: * reversing the path (pointers are now aligned).
1150: */
1151: while (p >= ip_srcrt.route) {
1152: #ifdef DIAGNOSTIC
1153: if (ipprintfs)
1.39 christos 1154: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1155: #endif
1156: *q++ = *p--;
1157: }
1158: /*
1159: * Last hop goes to final destination.
1160: */
1161: *q = ip_srcrt.dst;
1162: #ifdef DIAGNOSTIC
1163: if (ipprintfs)
1.39 christos 1164: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1165: #endif
1166: return (m);
1167: }
1168:
1169: /*
1170: * Strip out IP options, at higher
1171: * level protocol in the kernel.
1172: * Second argument is buffer to which options
1173: * will be moved, and return value is their length.
1174: * XXX should be deleted; last arg currently ignored.
1175: */
1.8 mycroft 1176: void
1.1 cgd 1177: ip_stripoptions(m, mopt)
1178: register struct mbuf *m;
1179: struct mbuf *mopt;
1180: {
1181: register int i;
1182: struct ip *ip = mtod(m, struct ip *);
1183: register caddr_t opts;
1184: int olen;
1185:
1.79 mycroft 1186: olen = (ip->ip_hl << 2) - sizeof (struct ip);
1.1 cgd 1187: opts = (caddr_t)(ip + 1);
1188: i = m->m_len - (sizeof (struct ip) + olen);
1189: bcopy(opts + olen, opts, (unsigned)i);
1190: m->m_len -= olen;
1191: if (m->m_flags & M_PKTHDR)
1192: m->m_pkthdr.len -= olen;
1.79 mycroft 1193: ip->ip_len -= olen;
1194: ip->ip_hl = sizeof (struct ip) >> 2;
1.1 cgd 1195: }
1196:
1.23 mycroft 1197: int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1198: 0, 0, 0, 0,
1199: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1200: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1201: EMSGSIZE, EHOSTUNREACH, 0, 0,
1202: 0, 0, 0, 0,
1203: ENOPROTOOPT
1204: };
1205:
1206: /*
1207: * Forward a packet. If some error occurs return the sender
1208: * an icmp packet. Note we can't always generate a meaningful
1209: * icmp message because icmp doesn't have a large enough repertoire
1210: * of codes and types.
1211: *
1212: * If not forwarding, just drop the packet. This could be confusing
1213: * if ipforwarding was zero but some routing protocol was advancing
1214: * us as a gateway to somewhere. However, we must let the routing
1215: * protocol deal with that.
1216: *
1217: * The srcrt parameter indicates whether the packet is being forwarded
1218: * via a source route.
1219: */
1.13 mycroft 1220: void
1.1 cgd 1221: ip_forward(m, srcrt)
1222: struct mbuf *m;
1223: int srcrt;
1224: {
1225: register struct ip *ip = mtod(m, struct ip *);
1226: register struct sockaddr_in *sin;
1227: register struct rtentry *rt;
1.28 christos 1228: int error, type = 0, code = 0;
1.1 cgd 1229: struct mbuf *mcopy;
1.13 mycroft 1230: n_long dest;
1231: struct ifnet *destifp;
1.1 cgd 1232:
1.13 mycroft 1233: dest = 0;
1.1 cgd 1234: #ifdef DIAGNOSTIC
1235: if (ipprintfs)
1.70 thorpej 1236: printf("forward: src %2.2x dst %2.2x ttl %x\n",
1237: ntohl(ip->ip_src.s_addr),
1238: ntohl(ip->ip_dst.s_addr), ip->ip_ttl);
1.1 cgd 1239: #endif
1240: if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1241: ipstat.ips_cantforward++;
1242: m_freem(m);
1243: return;
1244: }
1245: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1246: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1247: return;
1248: }
1249: ip->ip_ttl -= IPTTLDEC;
1250:
1.19 mycroft 1251: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1252: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1253: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1254: if (ipforward_rt.ro_rt) {
1255: RTFREE(ipforward_rt.ro_rt);
1256: ipforward_rt.ro_rt = 0;
1257: }
1258: sin->sin_family = AF_INET;
1.35 mycroft 1259: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1260: sin->sin_addr = ip->ip_dst;
1261:
1262: rtalloc(&ipforward_rt);
1263: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1264: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1265: return;
1266: }
1267: rt = ipforward_rt.ro_rt;
1268: }
1269:
1270: /*
1.34 mycroft 1271: * Save at most 68 bytes of the packet in case
1.1 cgd 1272: * we need to generate an ICMP message to the src.
1273: */
1.34 mycroft 1274: mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68));
1.1 cgd 1275:
1276: /*
1277: * If forwarding packet using same interface that it came in on,
1278: * perhaps should send a redirect to sender to shortcut a hop.
1279: * Only send redirect if source is sending directly to us,
1280: * and if packet was not source routed (or has any options).
1281: * Also, don't send redirect if forwarding using a default route
1282: * or a route modified by a redirect.
1283: */
1284: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1285: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1286: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1287: ipsendredirects && !srcrt) {
1.19 mycroft 1288: if (rt->rt_ifa &&
1289: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1290: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1291: if (rt->rt_flags & RTF_GATEWAY)
1292: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1293: else
1294: dest = ip->ip_dst.s_addr;
1295: /*
1296: * Router requirements says to only send host
1297: * redirects.
1298: */
1299: type = ICMP_REDIRECT;
1300: code = ICMP_REDIRECT_HOST;
1.1 cgd 1301: #ifdef DIAGNOSTIC
1.77 thorpej 1302: if (ipprintfs)
1303: printf("redirect (%d) to %x\n", code,
1304: (u_int32_t)dest);
1.1 cgd 1305: #endif
1306: }
1307: }
1308:
1.27 thorpej 1309: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1310: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1.1 cgd 1311: if (error)
1312: ipstat.ips_cantforward++;
1313: else {
1314: ipstat.ips_forward++;
1315: if (type)
1316: ipstat.ips_redirectsent++;
1317: else {
1.63 matt 1318: if (mcopy) {
1319: #ifdef GATEWAY
1.64 thorpej 1320: if (mcopy->m_flags & M_CANFASTFWD)
1321: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1322: #endif
1.1 cgd 1323: m_freem(mcopy);
1.63 matt 1324: }
1.1 cgd 1325: return;
1326: }
1327: }
1328: if (mcopy == NULL)
1329: return;
1.13 mycroft 1330: destifp = NULL;
1331:
1.1 cgd 1332: switch (error) {
1333:
1334: case 0: /* forwarded, but need redirect */
1335: /* type, code set above */
1336: break;
1337:
1338: case ENETUNREACH: /* shouldn't happen, checked above */
1339: case EHOSTUNREACH:
1340: case ENETDOWN:
1341: case EHOSTDOWN:
1342: default:
1343: type = ICMP_UNREACH;
1344: code = ICMP_UNREACH_HOST;
1345: break;
1346:
1347: case EMSGSIZE:
1348: type = ICMP_UNREACH;
1349: code = ICMP_UNREACH_NEEDFRAG;
1.13 mycroft 1350: if (ipforward_rt.ro_rt)
1351: destifp = ipforward_rt.ro_rt->rt_ifp;
1.1 cgd 1352: ipstat.ips_cantfrag++;
1353: break;
1354:
1355: case ENOBUFS:
1356: type = ICMP_SOURCEQUENCH;
1357: code = 0;
1358: break;
1359: }
1.13 mycroft 1360: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 1361: }
1362:
1363: void
1364: ip_savecontrol(inp, mp, ip, m)
1365: register struct inpcb *inp;
1366: register struct mbuf **mp;
1367: register struct ip *ip;
1368: register struct mbuf *m;
1369: {
1370:
1371: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1372: struct timeval tv;
1373:
1374: microtime(&tv);
1375: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1376: SCM_TIMESTAMP, SOL_SOCKET);
1377: if (*mp)
1378: mp = &(*mp)->m_next;
1379: }
1380: if (inp->inp_flags & INP_RECVDSTADDR) {
1381: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1382: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1383: if (*mp)
1384: mp = &(*mp)->m_next;
1385: }
1386: #ifdef notyet
1387: /*
1388: * XXX
1389: * Moving these out of udp_input() made them even more broken
1390: * than they already were.
1391: * - fenner@parc.xerox.com
1392: */
1393: /* options were tossed already */
1394: if (inp->inp_flags & INP_RECVOPTS) {
1395: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1396: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1397: if (*mp)
1398: mp = &(*mp)->m_next;
1399: }
1400: /* ip_srcroute doesn't do what we want here, need to fix */
1401: if (inp->inp_flags & INP_RECVRETOPTS) {
1402: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1403: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1404: if (*mp)
1405: mp = &(*mp)->m_next;
1406: }
1407: #endif
1408: if (inp->inp_flags & INP_RECVIF) {
1409: struct sockaddr_dl sdl;
1410:
1411: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1412: sdl.sdl_family = AF_LINK;
1413: sdl.sdl_index = m->m_pkthdr.rcvif ?
1414: m->m_pkthdr.rcvif->if_index : 0;
1415: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1416: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1417: IP_RECVIF, IPPROTO_IP);
1418: if (*mp)
1419: mp = &(*mp)->m_next;
1420: }
1.13 mycroft 1421: }
1422:
1423: int
1424: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1425: int *name;
1426: u_int namelen;
1427: void *oldp;
1428: size_t *oldlenp;
1429: void *newp;
1430: size_t newlen;
1431: {
1.52 thorpej 1432: extern int subnetsarelocal;
1433:
1.54 lukem 1434: int error, old;
1435:
1.13 mycroft 1436: /* All sysctl names at this level are terminal. */
1437: if (namelen != 1)
1438: return (ENOTDIR);
1439:
1440: switch (name[0]) {
1441: case IPCTL_FORWARDING:
1442: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1443: case IPCTL_SENDREDIRECTS:
1444: return (sysctl_int(oldp, oldlenp, newp, newlen,
1445: &ipsendredirects));
1446: case IPCTL_DEFTTL:
1447: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1448: #ifdef notyet
1449: case IPCTL_DEFMTU:
1450: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1451: #endif
1.26 thorpej 1452: case IPCTL_FORWSRCRT:
1.47 cjs 1453: /* Don't allow this to change in a secure environment. */
1.26 thorpej 1454: if (securelevel > 0)
1.46 cjs 1455: return (sysctl_rdint(oldp, oldlenp, newp,
1456: ip_forwsrcrt));
1457: else
1458: return (sysctl_int(oldp, oldlenp, newp, newlen,
1459: &ip_forwsrcrt));
1.27 thorpej 1460: case IPCTL_DIRECTEDBCAST:
1461: return (sysctl_int(oldp, oldlenp, newp, newlen,
1462: &ip_directedbcast));
1.47 cjs 1463: case IPCTL_ALLOWSRCRT:
1464: return (sysctl_int(oldp, oldlenp, newp, newlen,
1465: &ip_allowsrcrt));
1.52 thorpej 1466: case IPCTL_SUBNETSARELOCAL:
1467: return (sysctl_int(oldp, oldlenp, newp, newlen,
1468: &subnetsarelocal));
1.53 kml 1469: case IPCTL_MTUDISC:
1.60 kml 1470: error = sysctl_int(oldp, oldlenp, newp, newlen,
1471: &ip_mtudisc);
1472: if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) {
1473: ip_mtudisc_timeout_q =
1474: rt_timer_queue_create(ip_mtudisc_timeout);
1475: } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) {
1476: rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE);
1477: ip_mtudisc_timeout_q = NULL;
1478: }
1479: return error;
1.54 lukem 1480: case IPCTL_ANONPORTMIN:
1481: old = anonportmin;
1482: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1483: if (anonportmin >= anonportmax || anonportmin > 65535
1484: #ifndef IPNOPRIVPORTS
1485: || anonportmin < IPPORT_RESERVED
1486: #endif
1487: ) {
1488: anonportmin = old;
1489: return (EINVAL);
1490: }
1491: return (error);
1492: case IPCTL_ANONPORTMAX:
1493: old = anonportmax;
1494: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1495: if (anonportmin >= anonportmax || anonportmax > 65535
1496: #ifndef IPNOPRIVPORTS
1497: || anonportmax < IPPORT_RESERVED
1498: #endif
1499: ) {
1500: anonportmax = old;
1501: return (EINVAL);
1502: }
1.60 kml 1503: return (error);
1504: case IPCTL_MTUDISCTIMEOUT:
1505: error = sysctl_int(oldp, oldlenp, newp, newlen,
1506: &ip_mtudisc_timeout);
1507: if (ip_mtudisc_timeout_q != NULL)
1508: rt_timer_queue_change(ip_mtudisc_timeout_q,
1509: ip_mtudisc_timeout);
1.54 lukem 1510: return (error);
1.65 matt 1511: #ifdef GATEWAY
1512: case IPCTL_MAXFLOWS:
1.67 thorpej 1513: {
1514: int s;
1515:
1.65 matt 1516: error = sysctl_int(oldp, oldlenp, newp, newlen,
1517: &ip_maxflows);
1.67 thorpej 1518: s = splsoftnet();
1.65 matt 1519: ipflow_reap(0);
1.67 thorpej 1520: splx(s);
1.65 matt 1521: return (error);
1.67 thorpej 1522: }
1.65 matt 1523: #endif
1.13 mycroft 1524: default:
1525: return (EOPNOTSUPP);
1526: }
1527: /* NOTREACHED */
1.1 cgd 1528: }
CVSweb <webmaster@jp.NetBSD.org>