Annotation of src/sys/netinet/ip_input.c, Revision 1.80
1.80 ! mycroft 1: /* $NetBSD: ip_input.c,v 1.79 1999/01/19 23:03:21 mycroft 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: }
1.78 mycroft 343: if (in_cksum(m, hlen) != 0) {
1.1 cgd 344: ipstat.ips_badsum++;
345: goto bad;
346: }
347:
348: /*
349: * Convert fields to host representation.
350: */
351: NTOHS(ip->ip_len);
352: NTOHS(ip->ip_off);
1.35 mycroft 353: len = ip->ip_len;
1.1 cgd 354:
355: /*
356: * Check that the amount of data in the buffers
357: * is as at least much as the IP header would have us expect.
358: * Trim mbufs if longer than we expect.
359: * Drop packet if shorter than we expect.
360: */
1.35 mycroft 361: if (m->m_pkthdr.len < len) {
1.1 cgd 362: ipstat.ips_tooshort++;
363: goto bad;
364: }
1.35 mycroft 365: if (m->m_pkthdr.len > len) {
1.1 cgd 366: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 367: m->m_len = len;
368: m->m_pkthdr.len = len;
1.1 cgd 369: } else
1.35 mycroft 370: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 371: }
372:
1.64 thorpej 373: /*
374: * Assume that we can create a fast-forward IP flow entry
375: * based on this packet.
376: */
377: m->m_flags |= M_CANFASTFWD;
378:
1.36 mrg 379: #ifdef PFIL_HOOKS
1.33 mrg 380: /*
1.64 thorpej 381: * Run through list of hooks for input packets. If there are any
382: * filters which require that additional packets in the flow are
383: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
384: * Note that filters must _never_ set this flag, as another filter
385: * in the list may have previously cleared it.
1.33 mrg 386: */
387: m0 = m;
1.59 mrg 388: for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next)
1.33 mrg 389: if (pfh->pfil_func) {
1.43 mrg 390: rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0);
391: if (rv)
1.40 veego 392: goto next;
1.68 sommerfe 393: m = m0;
394: if (m == NULL)
395: goto next;
396: ip = mtod(m, struct ip *);
1.33 mrg 397: }
1.36 mrg 398: #endif /* PFIL_HOOKS */
1.33 mrg 399:
1.1 cgd 400: /*
401: * Process options and, if not destined for us,
402: * ship it on. ip_dooptions returns 1 when an
403: * error was detected (causing an icmp message
404: * to be sent and the original packet to be freed).
405: */
406: ip_nhops = 0; /* for source routed packets */
407: if (hlen > sizeof (struct ip) && ip_dooptions(m))
408: goto next;
409:
410: /*
411: * Check our list of addresses, to see if the packet is for us.
412: */
1.57 tls 413: INADDR_TO_IA(ip->ip_dst, ia);
414: if (ia != NULL) goto ours;
415: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
416: for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
417: ifa != NULL; ifa = ifa->ifa_list.tqe_next) {
418: if (ifa->ifa_addr->sa_family != AF_INET) continue;
419: ia = ifatoia(ifa);
1.35 mycroft 420: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
421: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 422: /*
423: * Look for all-0's host part (old broadcast addr),
424: * either for subnet or net.
425: */
426: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 427: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 428: goto ours;
1.57 tls 429: /*
430: * An interface with IP address zero accepts
431: * all packets that arrive on that interface.
432: */
433: if (in_nullhost(ia->ia_addr.sin_addr))
434: goto ours;
1.1 cgd 435: }
436: }
1.18 mycroft 437: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 438: struct in_multi *inm;
439: #ifdef MROUTING
440: extern struct socket *ip_mrouter;
1.10 brezak 441:
442: if (m->m_flags & M_EXT) {
443: if ((m = m_pullup(m, hlen)) == 0) {
444: ipstat.ips_toosmall++;
445: goto next;
446: }
447: ip = mtod(m, struct ip *);
448: }
1.4 hpeyerl 449:
450: if (ip_mrouter) {
451: /*
452: * If we are acting as a multicast router, all
453: * incoming multicast packets are passed to the
454: * kernel-level multicast forwarding function.
455: * The packet is returned (relatively) intact; if
456: * ip_mforward() returns a non-zero value, the packet
457: * must be discarded, else it may be accepted below.
458: *
459: * (The IP ident field is put in the same byte order
460: * as expected when ip_mforward() is called from
461: * ip_output().)
462: */
1.13 mycroft 463: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
464: ipstat.ips_cantforward++;
1.4 hpeyerl 465: m_freem(m);
466: goto next;
467: }
468:
469: /*
470: * The process-level routing demon needs to receive
471: * all multicast IGMP packets, whether or not this
472: * host belongs to their destination groups.
473: */
474: if (ip->ip_p == IPPROTO_IGMP)
475: goto ours;
1.13 mycroft 476: ipstat.ips_forward++;
1.4 hpeyerl 477: }
478: #endif
479: /*
480: * See if we belong to the destination multicast group on the
481: * arrival interface.
482: */
483: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
484: if (inm == NULL) {
1.13 mycroft 485: ipstat.ips_cantforward++;
1.4 hpeyerl 486: m_freem(m);
487: goto next;
488: }
489: goto ours;
490: }
1.19 mycroft 491: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 492: in_nullhost(ip->ip_dst))
1.1 cgd 493: goto ours;
494:
495: /*
496: * Not for us; forward if possible and desirable.
497: */
498: if (ipforwarding == 0) {
499: ipstat.ips_cantforward++;
500: m_freem(m);
501: } else
502: ip_forward(m, 0);
503: goto next;
504:
505: ours:
506: /*
507: * If offset or IP_MF are set, must reassemble.
508: * Otherwise, nothing need be done.
509: * (We could look in the reassembly queue to see
510: * if the packet was previously fragmented,
511: * but it's not worth the time; just let them time out.)
512: */
1.37 perry 513: if (ip->ip_off & ~(IP_DF|IP_RF)) {
1.1 cgd 514: /*
515: * Look for queue of fragments
516: * of this datagram.
517: */
1.75 thorpej 518: IPQ_LOCK();
1.25 cgd 519: for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next)
1.1 cgd 520: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 521: in_hosteq(ip->ip_src, fp->ipq_src) &&
522: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 523: ip->ip_p == fp->ipq_p)
524: goto found;
525: fp = 0;
526: found:
527:
528: /*
529: * Adjust ip_len to not reflect header,
1.25 cgd 530: * set ipqe_mff if more fragments are expected,
1.1 cgd 531: * convert offset of this to bytes.
532: */
533: ip->ip_len -= hlen;
1.25 cgd 534: mff = (ip->ip_off & IP_MF) != 0;
535: if (mff) {
1.16 cgd 536: /*
537: * Make sure that fragments have a data length
538: * that's a non-zero multiple of 8 bytes.
539: */
1.17 cgd 540: if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
1.16 cgd 541: ipstat.ips_badfrags++;
1.75 thorpej 542: IPQ_UNLOCK();
1.16 cgd 543: goto bad;
544: }
545: }
1.1 cgd 546: ip->ip_off <<= 3;
547:
548: /*
549: * If datagram marked as having more fragments
550: * or if this is not the first fragment,
551: * attempt reassembly; if it succeeds, proceed.
552: */
1.25 cgd 553: if (mff || ip->ip_off) {
1.1 cgd 554: ipstat.ips_fragments++;
1.72 thorpej 555: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.25 cgd 556: if (ipqe == NULL) {
557: ipstat.ips_rcvmemdrop++;
1.75 thorpej 558: IPQ_UNLOCK();
1.25 cgd 559: goto bad;
560: }
561: ipqe->ipqe_mff = mff;
1.50 thorpej 562: ipqe->ipqe_m = m;
1.25 cgd 563: ipqe->ipqe_ip = ip;
1.50 thorpej 564: m = ip_reass(ipqe, fp);
1.75 thorpej 565: if (m == 0) {
566: IPQ_UNLOCK();
1.1 cgd 567: goto next;
1.75 thorpej 568: }
1.13 mycroft 569: ipstat.ips_reassembled++;
1.50 thorpej 570: ip = mtod(m, struct ip *);
1.74 thorpej 571: hlen = ip->ip_hl << 2;
1.79 mycroft 572: ip->ip_len += hlen;
1.1 cgd 573: } else
574: if (fp)
575: ip_freef(fp);
1.75 thorpej 576: IPQ_UNLOCK();
1.79 mycroft 577: }
1.1 cgd 578:
579: /*
580: * Switch out to protocol's input routine.
581: */
582: ipstat.ips_delivered++;
583: (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
584: goto next;
585: bad:
586: m_freem(m);
587: goto next;
588: }
589:
590: /*
591: * Take incoming datagram fragment and try to
592: * reassemble it into whole datagram. If a chain for
593: * reassembly of this datagram already exists, then it
594: * is given as fp; otherwise have to make a chain.
595: */
1.50 thorpej 596: struct mbuf *
1.25 cgd 597: ip_reass(ipqe, fp)
598: register struct ipqent *ipqe;
1.1 cgd 599: register struct ipq *fp;
600: {
1.50 thorpej 601: register struct mbuf *m = ipqe->ipqe_m;
1.25 cgd 602: register struct ipqent *nq, *p, *q;
603: struct ip *ip;
1.1 cgd 604: struct mbuf *t;
1.25 cgd 605: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 606: int i, next;
607:
1.75 thorpej 608: IPQ_LOCK_CHECK();
609:
1.1 cgd 610: /*
611: * Presence of header sizes in mbufs
612: * would confuse code below.
613: */
614: m->m_data += hlen;
615: m->m_len -= hlen;
616:
617: /*
618: * If first fragment to arrive, create a reassembly queue.
619: */
620: if (fp == 0) {
1.50 thorpej 621: MALLOC(fp, struct ipq *, sizeof (struct ipq),
622: M_FTABLE, M_NOWAIT);
623: if (fp == NULL)
1.1 cgd 624: goto dropfrag;
1.25 cgd 625: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 626: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 627: fp->ipq_p = ipqe->ipqe_ip->ip_p;
628: fp->ipq_id = ipqe->ipqe_ip->ip_id;
629: LIST_INIT(&fp->ipq_fragq);
630: fp->ipq_src = ipqe->ipqe_ip->ip_src;
631: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
632: p = NULL;
1.1 cgd 633: goto insert;
634: }
635:
636: /*
637: * Find a segment which begins after this one does.
638: */
1.25 cgd 639: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
640: p = q, q = q->ipqe_q.le_next)
641: if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off)
1.1 cgd 642: break;
643:
644: /*
645: * If there is a preceding segment, it may provide some of
646: * our data already. If so, drop the data from the incoming
647: * segment. If it provides all of our data, drop us.
648: */
1.25 cgd 649: if (p != NULL) {
650: i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len -
651: ipqe->ipqe_ip->ip_off;
1.1 cgd 652: if (i > 0) {
1.25 cgd 653: if (i >= ipqe->ipqe_ip->ip_len)
1.1 cgd 654: goto dropfrag;
1.50 thorpej 655: m_adj(ipqe->ipqe_m, i);
1.25 cgd 656: ipqe->ipqe_ip->ip_off += i;
657: ipqe->ipqe_ip->ip_len -= i;
1.1 cgd 658: }
659: }
660:
661: /*
662: * While we overlap succeeding segments trim them or,
663: * if they are completely covered, dequeue them.
664: */
1.25 cgd 665: for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len >
666: q->ipqe_ip->ip_off; q = nq) {
667: i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) -
668: q->ipqe_ip->ip_off;
669: if (i < q->ipqe_ip->ip_len) {
670: q->ipqe_ip->ip_len -= i;
671: q->ipqe_ip->ip_off += i;
1.50 thorpej 672: m_adj(q->ipqe_m, i);
1.1 cgd 673: break;
674: }
1.25 cgd 675: nq = q->ipqe_q.le_next;
1.50 thorpej 676: m_freem(q->ipqe_m);
1.25 cgd 677: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 678: pool_put(&ipqent_pool, q);
1.1 cgd 679: }
680:
681: insert:
682: /*
683: * Stick new segment in its place;
684: * check for complete reassembly.
685: */
1.25 cgd 686: if (p == NULL) {
687: LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
688: } else {
689: LIST_INSERT_AFTER(p, ipqe, ipqe_q);
690: }
1.1 cgd 691: next = 0;
1.25 cgd 692: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
693: p = q, q = q->ipqe_q.le_next) {
694: if (q->ipqe_ip->ip_off != next)
1.1 cgd 695: return (0);
1.25 cgd 696: next += q->ipqe_ip->ip_len;
1.1 cgd 697: }
1.25 cgd 698: if (p->ipqe_mff)
1.1 cgd 699: return (0);
700:
701: /*
1.41 thorpej 702: * Reassembly is complete. Check for a bogus message size and
703: * concatenate fragments.
1.1 cgd 704: */
1.25 cgd 705: q = fp->ipq_fragq.lh_first;
706: ip = q->ipqe_ip;
1.41 thorpej 707: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
708: ipstat.ips_toolong++;
709: ip_freef(fp);
710: return (0);
711: }
1.50 thorpej 712: m = q->ipqe_m;
1.1 cgd 713: t = m->m_next;
714: m->m_next = 0;
715: m_cat(m, t);
1.25 cgd 716: nq = q->ipqe_q.le_next;
1.72 thorpej 717: pool_put(&ipqent_pool, q);
1.25 cgd 718: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 719: t = q->ipqe_m;
1.25 cgd 720: nq = q->ipqe_q.le_next;
1.72 thorpej 721: pool_put(&ipqent_pool, q);
1.1 cgd 722: m_cat(m, t);
723: }
724:
725: /*
726: * Create header for new ip packet by
727: * modifying header of first packet;
728: * dequeue and discard fragment reassembly header.
729: * Make header visible.
730: */
731: ip->ip_len = next;
1.25 cgd 732: ip->ip_src = fp->ipq_src;
733: ip->ip_dst = fp->ipq_dst;
734: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 735: FREE(fp, M_FTABLE);
1.1 cgd 736: m->m_len += (ip->ip_hl << 2);
737: m->m_data -= (ip->ip_hl << 2);
738: /* some debugging cruft by sklower, below, will go away soon */
739: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
740: register int plen = 0;
1.50 thorpej 741: for (t = m; t; t = t->m_next)
742: plen += t->m_len;
743: m->m_pkthdr.len = plen;
1.1 cgd 744: }
1.50 thorpej 745: return (m);
1.1 cgd 746:
747: dropfrag:
748: ipstat.ips_fragdropped++;
749: m_freem(m);
1.72 thorpej 750: pool_put(&ipqent_pool, ipqe);
1.1 cgd 751: return (0);
752: }
753:
754: /*
755: * Free a fragment reassembly header and all
756: * associated datagrams.
757: */
1.8 mycroft 758: void
1.1 cgd 759: ip_freef(fp)
760: struct ipq *fp;
761: {
1.25 cgd 762: register struct ipqent *q, *p;
1.1 cgd 763:
1.75 thorpej 764: IPQ_LOCK_CHECK();
765:
1.25 cgd 766: for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) {
767: p = q->ipqe_q.le_next;
1.50 thorpej 768: m_freem(q->ipqe_m);
1.25 cgd 769: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 770: pool_put(&ipqent_pool, q);
1.1 cgd 771: }
1.25 cgd 772: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 773: FREE(fp, M_FTABLE);
1.1 cgd 774: }
775:
776: /*
777: * IP timer processing;
778: * if a timer expires on a reassembly
779: * queue, discard it.
780: */
1.8 mycroft 781: void
1.1 cgd 782: ip_slowtimo()
783: {
1.25 cgd 784: register struct ipq *fp, *nfp;
1.24 mycroft 785: int s = splsoftnet();
1.1 cgd 786:
1.75 thorpej 787: IPQ_LOCK();
1.25 cgd 788: for (fp = ipq.lh_first; fp != NULL; fp = nfp) {
789: nfp = fp->ipq_q.le_next;
790: if (--fp->ipq_ttl == 0) {
1.1 cgd 791: ipstat.ips_fragtimeout++;
1.25 cgd 792: ip_freef(fp);
1.1 cgd 793: }
794: }
1.75 thorpej 795: IPQ_UNLOCK();
1.63 matt 796: #ifdef GATEWAY
797: ipflow_slowtimo();
798: #endif
1.1 cgd 799: splx(s);
800: }
801:
802: /*
803: * Drain off all datagram fragments.
804: */
1.8 mycroft 805: void
1.1 cgd 806: ip_drain()
807: {
808:
1.75 thorpej 809: /*
810: * We may be called from a device's interrupt context. If
811: * the ipq is already busy, just bail out now.
812: */
813: if (ipq_lock_try() == 0)
814: return;
815:
1.25 cgd 816: while (ipq.lh_first != NULL) {
1.1 cgd 817: ipstat.ips_fragdropped++;
1.25 cgd 818: ip_freef(ipq.lh_first);
1.1 cgd 819: }
1.75 thorpej 820:
821: IPQ_UNLOCK();
1.1 cgd 822: }
823:
824: /*
825: * Do option processing on a datagram,
826: * possibly discarding it if bad options are encountered,
827: * or forwarding it if source-routed.
828: * Returns 1 if packet has been forwarded/freed,
829: * 0 if the packet should be processed further.
830: */
1.8 mycroft 831: int
1.1 cgd 832: ip_dooptions(m)
833: struct mbuf *m;
834: {
835: register struct ip *ip = mtod(m, struct ip *);
836: register u_char *cp;
837: register struct ip_timestamp *ipt;
838: register struct in_ifaddr *ia;
839: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.13 mycroft 840: struct in_addr *sin, dst;
1.1 cgd 841: n_time ntime;
842:
1.13 mycroft 843: dst = ip->ip_dst;
1.1 cgd 844: cp = (u_char *)(ip + 1);
845: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
846: for (; cnt > 0; cnt -= optlen, cp += optlen) {
847: opt = cp[IPOPT_OPTVAL];
848: if (opt == IPOPT_EOL)
849: break;
850: if (opt == IPOPT_NOP)
851: optlen = 1;
852: else {
853: optlen = cp[IPOPT_OLEN];
854: if (optlen <= 0 || optlen > cnt) {
855: code = &cp[IPOPT_OLEN] - (u_char *)ip;
856: goto bad;
857: }
858: }
859: switch (opt) {
860:
861: default:
862: break;
863:
864: /*
865: * Source routing with record.
866: * Find interface with current destination address.
867: * If none on this machine then drop if strictly routed,
868: * or do nothing if loosely routed.
869: * Record interface address and bring up next address
870: * component. If strictly routed make sure next
871: * address is on directly accessible net.
872: */
873: case IPOPT_LSRR:
874: case IPOPT_SSRR:
1.47 cjs 875: if (ip_allowsrcrt == 0) {
876: type = ICMP_UNREACH;
877: code = ICMP_UNREACH_NET_PROHIB;
878: goto bad;
879: }
1.1 cgd 880: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
881: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
882: goto bad;
883: }
884: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 885: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 886: if (ia == 0) {
887: if (opt == IPOPT_SSRR) {
888: type = ICMP_UNREACH;
889: code = ICMP_UNREACH_SRCFAIL;
890: goto bad;
891: }
892: /*
893: * Loose routing, and not at next destination
894: * yet; nothing to do except forward.
895: */
896: break;
897: }
898: off--; /* 0 origin */
899: if (off > optlen - sizeof(struct in_addr)) {
900: /*
901: * End of source route. Should be for us.
902: */
903: save_rte(cp, ip->ip_src);
904: break;
905: }
906: /*
907: * locate outgoing interface
908: */
909: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
910: sizeof(ipaddr.sin_addr));
911: if (opt == IPOPT_SSRR) {
912: #define INA struct in_ifaddr *
913: #define SA struct sockaddr *
1.29 mrg 914: ia = (INA)ifa_ifwithladdr((SA)&ipaddr);
1.1 cgd 915: } else
916: ia = ip_rtaddr(ipaddr.sin_addr);
917: if (ia == 0) {
918: type = ICMP_UNREACH;
919: code = ICMP_UNREACH_SRCFAIL;
920: goto bad;
921: }
922: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 923: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 924: (caddr_t)(cp + off), sizeof(struct in_addr));
925: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 926: /*
927: * Let ip_intr's mcast routing check handle mcast pkts
928: */
1.18 mycroft 929: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 930: break;
931:
932: case IPOPT_RR:
933: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
934: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
935: goto bad;
936: }
937: /*
938: * If no space remains, ignore.
939: */
940: off--; /* 0 origin */
941: if (off > optlen - sizeof(struct in_addr))
942: break;
943: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
944: sizeof(ipaddr.sin_addr));
945: /*
946: * locate outgoing interface; if we're the destination,
947: * use the incoming interface (should be same).
948: */
949: if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
950: (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
951: type = ICMP_UNREACH;
952: code = ICMP_UNREACH_HOST;
953: goto bad;
954: }
1.20 mycroft 955: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 956: (caddr_t)(cp + off), sizeof(struct in_addr));
957: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
958: break;
959:
960: case IPOPT_TS:
961: code = cp - (u_char *)ip;
962: ipt = (struct ip_timestamp *)cp;
963: if (ipt->ipt_len < 5)
964: goto bad;
1.15 cgd 965: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.1 cgd 966: if (++ipt->ipt_oflw == 0)
967: goto bad;
968: break;
969: }
970: sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
971: switch (ipt->ipt_flg) {
972:
973: case IPOPT_TS_TSONLY:
974: break;
975:
976: case IPOPT_TS_TSANDADDR:
1.66 thorpej 977: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 978: sizeof(struct in_addr) > ipt->ipt_len)
979: goto bad;
1.13 mycroft 980: ipaddr.sin_addr = dst;
981: ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
982: m->m_pkthdr.rcvif);
983: if (ia == 0)
984: continue;
1.20 mycroft 985: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 986: (caddr_t)sin, sizeof(struct in_addr));
987: ipt->ipt_ptr += sizeof(struct in_addr);
988: break;
989:
990: case IPOPT_TS_PRESPEC:
1.66 thorpej 991: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 992: sizeof(struct in_addr) > ipt->ipt_len)
993: goto bad;
994: bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
995: sizeof(struct in_addr));
996: if (ifa_ifwithaddr((SA)&ipaddr) == 0)
997: continue;
998: ipt->ipt_ptr += sizeof(struct in_addr);
999: break;
1000:
1001: default:
1002: goto bad;
1003: }
1004: ntime = iptime();
1005: bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
1006: sizeof(n_time));
1007: ipt->ipt_ptr += sizeof(n_time);
1008: }
1009: }
1010: if (forward) {
1.26 thorpej 1011: if (ip_forwsrcrt == 0) {
1012: type = ICMP_UNREACH;
1013: code = ICMP_UNREACH_SRCFAIL;
1014: goto bad;
1015: }
1.1 cgd 1016: ip_forward(m, 1);
1017: return (1);
1.13 mycroft 1018: }
1019: return (0);
1.1 cgd 1020: bad:
1.13 mycroft 1021: icmp_error(m, type, code, 0, 0);
1022: ipstat.ips_badoptions++;
1.1 cgd 1023: return (1);
1024: }
1025:
1026: /*
1027: * Given address of next destination (final or next hop),
1028: * return internet address info of interface to be used to get there.
1029: */
1030: struct in_ifaddr *
1031: ip_rtaddr(dst)
1032: struct in_addr dst;
1033: {
1034: register struct sockaddr_in *sin;
1035:
1.19 mycroft 1036: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1037:
1.35 mycroft 1038: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 1039: if (ipforward_rt.ro_rt) {
1040: RTFREE(ipforward_rt.ro_rt);
1041: ipforward_rt.ro_rt = 0;
1042: }
1043: sin->sin_family = AF_INET;
1044: sin->sin_len = sizeof(*sin);
1045: sin->sin_addr = dst;
1046:
1047: rtalloc(&ipforward_rt);
1048: }
1049: if (ipforward_rt.ro_rt == 0)
1050: return ((struct in_ifaddr *)0);
1.19 mycroft 1051: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 1052: }
1053:
1054: /*
1055: * Save incoming source route for use in replies,
1056: * to be picked up later by ip_srcroute if the receiver is interested.
1057: */
1.13 mycroft 1058: void
1.1 cgd 1059: save_rte(option, dst)
1060: u_char *option;
1061: struct in_addr dst;
1062: {
1063: unsigned olen;
1064:
1065: olen = option[IPOPT_OLEN];
1066: #ifdef DIAGNOSTIC
1067: if (ipprintfs)
1.39 christos 1068: printf("save_rte: olen %d\n", olen);
1.1 cgd 1069: #endif
1070: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1071: return;
1072: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
1073: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1074: ip_srcrt.dst = dst;
1075: }
1076:
1077: /*
1078: * Retrieve incoming source route for use in replies,
1079: * in the same form used by setsockopt.
1080: * The first hop is placed before the options, will be removed later.
1081: */
1082: struct mbuf *
1083: ip_srcroute()
1084: {
1085: register struct in_addr *p, *q;
1086: register struct mbuf *m;
1087:
1088: if (ip_nhops == 0)
1089: return ((struct mbuf *)0);
1090: m = m_get(M_DONTWAIT, MT_SOOPTS);
1091: if (m == 0)
1092: return ((struct mbuf *)0);
1093:
1.13 mycroft 1094: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1095:
1096: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1097: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1098: OPTSIZ;
1099: #ifdef DIAGNOSTIC
1100: if (ipprintfs)
1.39 christos 1101: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1102: #endif
1103:
1104: /*
1105: * First save first hop for return route
1106: */
1107: p = &ip_srcrt.route[ip_nhops - 1];
1108: *(mtod(m, struct in_addr *)) = *p--;
1109: #ifdef DIAGNOSTIC
1110: if (ipprintfs)
1.39 christos 1111: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1112: #endif
1113:
1114: /*
1115: * Copy option fields and padding (nop) to mbuf.
1116: */
1117: ip_srcrt.nop = IPOPT_NOP;
1118: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1119: bcopy((caddr_t)&ip_srcrt.nop,
1120: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1121: q = (struct in_addr *)(mtod(m, caddr_t) +
1122: sizeof(struct in_addr) + OPTSIZ);
1123: #undef OPTSIZ
1124: /*
1125: * Record return path as an IP source route,
1126: * reversing the path (pointers are now aligned).
1127: */
1128: while (p >= ip_srcrt.route) {
1129: #ifdef DIAGNOSTIC
1130: if (ipprintfs)
1.39 christos 1131: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1132: #endif
1133: *q++ = *p--;
1134: }
1135: /*
1136: * Last hop goes to final destination.
1137: */
1138: *q = ip_srcrt.dst;
1139: #ifdef DIAGNOSTIC
1140: if (ipprintfs)
1.39 christos 1141: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1142: #endif
1143: return (m);
1144: }
1145:
1146: /*
1147: * Strip out IP options, at higher
1148: * level protocol in the kernel.
1149: * Second argument is buffer to which options
1150: * will be moved, and return value is their length.
1151: * XXX should be deleted; last arg currently ignored.
1152: */
1.8 mycroft 1153: void
1.1 cgd 1154: ip_stripoptions(m, mopt)
1155: register struct mbuf *m;
1156: struct mbuf *mopt;
1157: {
1158: register int i;
1159: struct ip *ip = mtod(m, struct ip *);
1160: register caddr_t opts;
1161: int olen;
1162:
1.79 mycroft 1163: olen = (ip->ip_hl << 2) - sizeof (struct ip);
1.1 cgd 1164: opts = (caddr_t)(ip + 1);
1165: i = m->m_len - (sizeof (struct ip) + olen);
1166: bcopy(opts + olen, opts, (unsigned)i);
1167: m->m_len -= olen;
1168: if (m->m_flags & M_PKTHDR)
1169: m->m_pkthdr.len -= olen;
1.79 mycroft 1170: ip->ip_len -= olen;
1171: ip->ip_hl = sizeof (struct ip) >> 2;
1.1 cgd 1172: }
1173:
1.23 mycroft 1174: int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1175: 0, 0, 0, 0,
1176: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1177: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1178: EMSGSIZE, EHOSTUNREACH, 0, 0,
1179: 0, 0, 0, 0,
1180: ENOPROTOOPT
1181: };
1182:
1183: /*
1184: * Forward a packet. If some error occurs return the sender
1185: * an icmp packet. Note we can't always generate a meaningful
1186: * icmp message because icmp doesn't have a large enough repertoire
1187: * of codes and types.
1188: *
1189: * If not forwarding, just drop the packet. This could be confusing
1190: * if ipforwarding was zero but some routing protocol was advancing
1191: * us as a gateway to somewhere. However, we must let the routing
1192: * protocol deal with that.
1193: *
1194: * The srcrt parameter indicates whether the packet is being forwarded
1195: * via a source route.
1196: */
1.13 mycroft 1197: void
1.1 cgd 1198: ip_forward(m, srcrt)
1199: struct mbuf *m;
1200: int srcrt;
1201: {
1202: register struct ip *ip = mtod(m, struct ip *);
1203: register struct sockaddr_in *sin;
1204: register struct rtentry *rt;
1.28 christos 1205: int error, type = 0, code = 0;
1.1 cgd 1206: struct mbuf *mcopy;
1.13 mycroft 1207: n_long dest;
1208: struct ifnet *destifp;
1.1 cgd 1209:
1.13 mycroft 1210: dest = 0;
1.1 cgd 1211: #ifdef DIAGNOSTIC
1212: if (ipprintfs)
1.70 thorpej 1213: printf("forward: src %2.2x dst %2.2x ttl %x\n",
1214: ntohl(ip->ip_src.s_addr),
1215: ntohl(ip->ip_dst.s_addr), ip->ip_ttl);
1.1 cgd 1216: #endif
1217: if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1218: ipstat.ips_cantforward++;
1219: m_freem(m);
1220: return;
1221: }
1222: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1223: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1224: return;
1225: }
1226: ip->ip_ttl -= IPTTLDEC;
1227:
1.19 mycroft 1228: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1229: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1230: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1231: if (ipforward_rt.ro_rt) {
1232: RTFREE(ipforward_rt.ro_rt);
1233: ipforward_rt.ro_rt = 0;
1234: }
1235: sin->sin_family = AF_INET;
1.35 mycroft 1236: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1237: sin->sin_addr = ip->ip_dst;
1238:
1239: rtalloc(&ipforward_rt);
1240: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1241: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1242: return;
1243: }
1244: rt = ipforward_rt.ro_rt;
1245: }
1246:
1247: /*
1.34 mycroft 1248: * Save at most 68 bytes of the packet in case
1.1 cgd 1249: * we need to generate an ICMP message to the src.
1250: */
1.34 mycroft 1251: mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68));
1.1 cgd 1252:
1253: /*
1254: * If forwarding packet using same interface that it came in on,
1255: * perhaps should send a redirect to sender to shortcut a hop.
1256: * Only send redirect if source is sending directly to us,
1257: * and if packet was not source routed (or has any options).
1258: * Also, don't send redirect if forwarding using a default route
1259: * or a route modified by a redirect.
1260: */
1261: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1262: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1263: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1264: ipsendredirects && !srcrt) {
1.19 mycroft 1265: if (rt->rt_ifa &&
1266: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1267: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1268: if (rt->rt_flags & RTF_GATEWAY)
1269: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1270: else
1271: dest = ip->ip_dst.s_addr;
1272: /*
1273: * Router requirements says to only send host
1274: * redirects.
1275: */
1276: type = ICMP_REDIRECT;
1277: code = ICMP_REDIRECT_HOST;
1.1 cgd 1278: #ifdef DIAGNOSTIC
1.77 thorpej 1279: if (ipprintfs)
1280: printf("redirect (%d) to %x\n", code,
1281: (u_int32_t)dest);
1.1 cgd 1282: #endif
1283: }
1284: }
1285:
1.27 thorpej 1286: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1287: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1.1 cgd 1288: if (error)
1289: ipstat.ips_cantforward++;
1290: else {
1291: ipstat.ips_forward++;
1292: if (type)
1293: ipstat.ips_redirectsent++;
1294: else {
1.63 matt 1295: if (mcopy) {
1296: #ifdef GATEWAY
1.64 thorpej 1297: if (mcopy->m_flags & M_CANFASTFWD)
1298: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1299: #endif
1.1 cgd 1300: m_freem(mcopy);
1.63 matt 1301: }
1.1 cgd 1302: return;
1303: }
1304: }
1305: if (mcopy == NULL)
1306: return;
1.13 mycroft 1307: destifp = NULL;
1308:
1.1 cgd 1309: switch (error) {
1310:
1311: case 0: /* forwarded, but need redirect */
1312: /* type, code set above */
1313: break;
1314:
1315: case ENETUNREACH: /* shouldn't happen, checked above */
1316: case EHOSTUNREACH:
1317: case ENETDOWN:
1318: case EHOSTDOWN:
1319: default:
1320: type = ICMP_UNREACH;
1321: code = ICMP_UNREACH_HOST;
1322: break;
1323:
1324: case EMSGSIZE:
1325: type = ICMP_UNREACH;
1326: code = ICMP_UNREACH_NEEDFRAG;
1.13 mycroft 1327: if (ipforward_rt.ro_rt)
1328: destifp = ipforward_rt.ro_rt->rt_ifp;
1.1 cgd 1329: ipstat.ips_cantfrag++;
1330: break;
1331:
1332: case ENOBUFS:
1333: type = ICMP_SOURCEQUENCH;
1334: code = 0;
1335: break;
1336: }
1.13 mycroft 1337: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 1338: }
1339:
1340: void
1341: ip_savecontrol(inp, mp, ip, m)
1342: register struct inpcb *inp;
1343: register struct mbuf **mp;
1344: register struct ip *ip;
1345: register struct mbuf *m;
1346: {
1347:
1348: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1349: struct timeval tv;
1350:
1351: microtime(&tv);
1352: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1353: SCM_TIMESTAMP, SOL_SOCKET);
1354: if (*mp)
1355: mp = &(*mp)->m_next;
1356: }
1357: if (inp->inp_flags & INP_RECVDSTADDR) {
1358: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1359: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1360: if (*mp)
1361: mp = &(*mp)->m_next;
1362: }
1363: #ifdef notyet
1364: /*
1365: * XXX
1366: * Moving these out of udp_input() made them even more broken
1367: * than they already were.
1368: * - fenner@parc.xerox.com
1369: */
1370: /* options were tossed already */
1371: if (inp->inp_flags & INP_RECVOPTS) {
1372: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1373: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1374: if (*mp)
1375: mp = &(*mp)->m_next;
1376: }
1377: /* ip_srcroute doesn't do what we want here, need to fix */
1378: if (inp->inp_flags & INP_RECVRETOPTS) {
1379: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1380: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1381: if (*mp)
1382: mp = &(*mp)->m_next;
1383: }
1384: #endif
1385: if (inp->inp_flags & INP_RECVIF) {
1386: struct sockaddr_dl sdl;
1387:
1388: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1389: sdl.sdl_family = AF_LINK;
1390: sdl.sdl_index = m->m_pkthdr.rcvif ?
1391: m->m_pkthdr.rcvif->if_index : 0;
1392: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1393: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1394: IP_RECVIF, IPPROTO_IP);
1395: if (*mp)
1396: mp = &(*mp)->m_next;
1397: }
1.13 mycroft 1398: }
1399:
1400: int
1401: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1402: int *name;
1403: u_int namelen;
1404: void *oldp;
1405: size_t *oldlenp;
1406: void *newp;
1407: size_t newlen;
1408: {
1.52 thorpej 1409: extern int subnetsarelocal;
1410:
1.54 lukem 1411: int error, old;
1412:
1.13 mycroft 1413: /* All sysctl names at this level are terminal. */
1414: if (namelen != 1)
1415: return (ENOTDIR);
1416:
1417: switch (name[0]) {
1418: case IPCTL_FORWARDING:
1419: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1420: case IPCTL_SENDREDIRECTS:
1421: return (sysctl_int(oldp, oldlenp, newp, newlen,
1422: &ipsendredirects));
1423: case IPCTL_DEFTTL:
1424: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1425: #ifdef notyet
1426: case IPCTL_DEFMTU:
1427: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1428: #endif
1.26 thorpej 1429: case IPCTL_FORWSRCRT:
1.47 cjs 1430: /* Don't allow this to change in a secure environment. */
1.26 thorpej 1431: if (securelevel > 0)
1.46 cjs 1432: return (sysctl_rdint(oldp, oldlenp, newp,
1433: ip_forwsrcrt));
1434: else
1435: return (sysctl_int(oldp, oldlenp, newp, newlen,
1436: &ip_forwsrcrt));
1.27 thorpej 1437: case IPCTL_DIRECTEDBCAST:
1438: return (sysctl_int(oldp, oldlenp, newp, newlen,
1439: &ip_directedbcast));
1.47 cjs 1440: case IPCTL_ALLOWSRCRT:
1441: return (sysctl_int(oldp, oldlenp, newp, newlen,
1442: &ip_allowsrcrt));
1.52 thorpej 1443: case IPCTL_SUBNETSARELOCAL:
1444: return (sysctl_int(oldp, oldlenp, newp, newlen,
1445: &subnetsarelocal));
1.53 kml 1446: case IPCTL_MTUDISC:
1.60 kml 1447: error = sysctl_int(oldp, oldlenp, newp, newlen,
1448: &ip_mtudisc);
1449: if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) {
1450: ip_mtudisc_timeout_q =
1451: rt_timer_queue_create(ip_mtudisc_timeout);
1452: } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) {
1453: rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE);
1454: ip_mtudisc_timeout_q = NULL;
1455: }
1456: return error;
1.54 lukem 1457: case IPCTL_ANONPORTMIN:
1458: old = anonportmin;
1459: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1460: if (anonportmin >= anonportmax || anonportmin > 65535
1461: #ifndef IPNOPRIVPORTS
1462: || anonportmin < IPPORT_RESERVED
1463: #endif
1464: ) {
1465: anonportmin = old;
1466: return (EINVAL);
1467: }
1468: return (error);
1469: case IPCTL_ANONPORTMAX:
1470: old = anonportmax;
1471: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1472: if (anonportmin >= anonportmax || anonportmax > 65535
1473: #ifndef IPNOPRIVPORTS
1474: || anonportmax < IPPORT_RESERVED
1475: #endif
1476: ) {
1477: anonportmax = old;
1478: return (EINVAL);
1479: }
1.60 kml 1480: return (error);
1481: case IPCTL_MTUDISCTIMEOUT:
1482: error = sysctl_int(oldp, oldlenp, newp, newlen,
1483: &ip_mtudisc_timeout);
1484: if (ip_mtudisc_timeout_q != NULL)
1485: rt_timer_queue_change(ip_mtudisc_timeout_q,
1486: ip_mtudisc_timeout);
1.54 lukem 1487: return (error);
1.65 matt 1488: #ifdef GATEWAY
1489: case IPCTL_MAXFLOWS:
1.67 thorpej 1490: {
1491: int s;
1492:
1.65 matt 1493: error = sysctl_int(oldp, oldlenp, newp, newlen,
1494: &ip_maxflows);
1.67 thorpej 1495: s = splsoftnet();
1.65 matt 1496: ipflow_reap(0);
1.67 thorpej 1497: splx(s);
1.65 matt 1498: return (error);
1.67 thorpej 1499: }
1.65 matt 1500: #endif
1.13 mycroft 1501: default:
1502: return (EOPNOTSUPP);
1503: }
1504: /* NOTREACHED */
1.1 cgd 1505: }
CVSweb <webmaster@jp.NetBSD.org>
![[BACK]](/icons/back.gif){kind=icon}
Return to ip_input.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [cvs.NetBSD.org] / src / sys / netinet