Annotation of src/sys/netinet/ip_input.c, Revision 1.84
1.84 ! proff 1: /* $NetBSD: ip_input.c,v 1.83 1999/04/07 02:31:05 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: }
1.78 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;
1.81 proff 354:
355: /*
356: * Check for additional length bogosity
357: */
1.84 ! proff 358: if (len < hlen) {
1.81 proff 359: ipstat.ips_badlen++;
360: goto bad;
361: }
1.1 cgd 362:
363: /*
364: * Check that the amount of data in the buffers
365: * is as at least much as the IP header would have us expect.
366: * Trim mbufs if longer than we expect.
367: * Drop packet if shorter than we expect.
368: */
1.35 mycroft 369: if (m->m_pkthdr.len < len) {
1.1 cgd 370: ipstat.ips_tooshort++;
371: goto bad;
372: }
1.35 mycroft 373: if (m->m_pkthdr.len > len) {
1.1 cgd 374: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 375: m->m_len = len;
376: m->m_pkthdr.len = len;
1.1 cgd 377: } else
1.35 mycroft 378: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 379: }
380:
1.64 thorpej 381: /*
382: * Assume that we can create a fast-forward IP flow entry
383: * based on this packet.
384: */
385: m->m_flags |= M_CANFASTFWD;
386:
1.36 mrg 387: #ifdef PFIL_HOOKS
1.33 mrg 388: /*
1.64 thorpej 389: * Run through list of hooks for input packets. If there are any
390: * filters which require that additional packets in the flow are
391: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
392: * Note that filters must _never_ set this flag, as another filter
393: * in the list may have previously cleared it.
1.33 mrg 394: */
395: m0 = m;
1.59 mrg 396: for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next)
1.33 mrg 397: if (pfh->pfil_func) {
1.43 mrg 398: rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0);
399: if (rv)
1.40 veego 400: goto next;
1.68 sommerfe 401: m = m0;
402: if (m == NULL)
403: goto next;
404: ip = mtod(m, struct ip *);
1.33 mrg 405: }
1.36 mrg 406: #endif /* PFIL_HOOKS */
1.33 mrg 407:
1.1 cgd 408: /*
409: * Process options and, if not destined for us,
410: * ship it on. ip_dooptions returns 1 when an
411: * error was detected (causing an icmp message
412: * to be sent and the original packet to be freed).
413: */
414: ip_nhops = 0; /* for source routed packets */
415: if (hlen > sizeof (struct ip) && ip_dooptions(m))
416: goto next;
417:
418: /*
419: * Check our list of addresses, to see if the packet is for us.
420: */
1.57 tls 421: INADDR_TO_IA(ip->ip_dst, ia);
1.83 proff 422: if (ia != NULL) {
423: if (ia->ia_ifp->if_flags & IFF_UP)
424: goto ours;
425: }
1.57 tls 426: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
427: for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
428: ifa != NULL; ifa = ifa->ifa_list.tqe_next) {
429: if (ifa->ifa_addr->sa_family != AF_INET) continue;
430: ia = ifatoia(ifa);
1.35 mycroft 431: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
432: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 433: /*
434: * Look for all-0's host part (old broadcast addr),
435: * either for subnet or net.
436: */
437: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 438: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 439: goto ours;
1.57 tls 440: /*
441: * An interface with IP address zero accepts
442: * all packets that arrive on that interface.
443: */
444: if (in_nullhost(ia->ia_addr.sin_addr))
445: goto ours;
1.1 cgd 446: }
447: }
1.18 mycroft 448: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 449: struct in_multi *inm;
450: #ifdef MROUTING
451: extern struct socket *ip_mrouter;
1.10 brezak 452:
453: if (m->m_flags & M_EXT) {
454: if ((m = m_pullup(m, hlen)) == 0) {
455: ipstat.ips_toosmall++;
456: goto next;
457: }
458: ip = mtod(m, struct ip *);
459: }
1.4 hpeyerl 460:
461: if (ip_mrouter) {
462: /*
463: * If we are acting as a multicast router, all
464: * incoming multicast packets are passed to the
465: * kernel-level multicast forwarding function.
466: * The packet is returned (relatively) intact; if
467: * ip_mforward() returns a non-zero value, the packet
468: * must be discarded, else it may be accepted below.
469: *
470: * (The IP ident field is put in the same byte order
471: * as expected when ip_mforward() is called from
472: * ip_output().)
473: */
1.13 mycroft 474: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
475: ipstat.ips_cantforward++;
1.4 hpeyerl 476: m_freem(m);
477: goto next;
478: }
479:
480: /*
481: * The process-level routing demon needs to receive
482: * all multicast IGMP packets, whether or not this
483: * host belongs to their destination groups.
484: */
485: if (ip->ip_p == IPPROTO_IGMP)
486: goto ours;
1.13 mycroft 487: ipstat.ips_forward++;
1.4 hpeyerl 488: }
489: #endif
490: /*
491: * See if we belong to the destination multicast group on the
492: * arrival interface.
493: */
494: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
495: if (inm == NULL) {
1.13 mycroft 496: ipstat.ips_cantforward++;
1.4 hpeyerl 497: m_freem(m);
498: goto next;
499: }
500: goto ours;
501: }
1.19 mycroft 502: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 503: in_nullhost(ip->ip_dst))
1.1 cgd 504: goto ours;
505:
506: /*
507: * Not for us; forward if possible and desirable.
508: */
509: if (ipforwarding == 0) {
510: ipstat.ips_cantforward++;
511: m_freem(m);
512: } else
513: ip_forward(m, 0);
514: goto next;
515:
516: ours:
517: /*
518: * If offset or IP_MF are set, must reassemble.
519: * Otherwise, nothing need be done.
520: * (We could look in the reassembly queue to see
521: * if the packet was previously fragmented,
522: * but it's not worth the time; just let them time out.)
523: */
1.37 perry 524: if (ip->ip_off & ~(IP_DF|IP_RF)) {
1.1 cgd 525: /*
526: * Look for queue of fragments
527: * of this datagram.
528: */
1.75 thorpej 529: IPQ_LOCK();
1.25 cgd 530: for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next)
1.1 cgd 531: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 532: in_hosteq(ip->ip_src, fp->ipq_src) &&
533: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 534: ip->ip_p == fp->ipq_p)
535: goto found;
536: fp = 0;
537: found:
538:
539: /*
540: * Adjust ip_len to not reflect header,
1.25 cgd 541: * set ipqe_mff if more fragments are expected,
1.1 cgd 542: * convert offset of this to bytes.
543: */
544: ip->ip_len -= hlen;
1.25 cgd 545: mff = (ip->ip_off & IP_MF) != 0;
546: if (mff) {
1.16 cgd 547: /*
548: * Make sure that fragments have a data length
549: * that's a non-zero multiple of 8 bytes.
550: */
1.17 cgd 551: if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
1.16 cgd 552: ipstat.ips_badfrags++;
1.75 thorpej 553: IPQ_UNLOCK();
1.16 cgd 554: goto bad;
555: }
556: }
1.1 cgd 557: ip->ip_off <<= 3;
558:
559: /*
560: * If datagram marked as having more fragments
561: * or if this is not the first fragment,
562: * attempt reassembly; if it succeeds, proceed.
563: */
1.25 cgd 564: if (mff || ip->ip_off) {
1.1 cgd 565: ipstat.ips_fragments++;
1.72 thorpej 566: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.25 cgd 567: if (ipqe == NULL) {
568: ipstat.ips_rcvmemdrop++;
1.75 thorpej 569: IPQ_UNLOCK();
1.25 cgd 570: goto bad;
571: }
572: ipqe->ipqe_mff = mff;
1.50 thorpej 573: ipqe->ipqe_m = m;
1.25 cgd 574: ipqe->ipqe_ip = ip;
1.50 thorpej 575: m = ip_reass(ipqe, fp);
1.75 thorpej 576: if (m == 0) {
577: IPQ_UNLOCK();
1.1 cgd 578: goto next;
1.75 thorpej 579: }
1.13 mycroft 580: ipstat.ips_reassembled++;
1.50 thorpej 581: ip = mtod(m, struct ip *);
1.74 thorpej 582: hlen = ip->ip_hl << 2;
1.79 mycroft 583: ip->ip_len += hlen;
1.1 cgd 584: } else
585: if (fp)
586: ip_freef(fp);
1.75 thorpej 587: IPQ_UNLOCK();
1.79 mycroft 588: }
1.1 cgd 589:
590: /*
591: * Switch out to protocol's input routine.
592: */
1.82 aidan 593: #if IFA_STATS
594: ia->ia_ifa.ifa_data.ifad_inbytes += ip->ip_len;
595: #endif
1.1 cgd 596: ipstat.ips_delivered++;
597: (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
598: goto next;
599: bad:
600: m_freem(m);
601: goto next;
602: }
603:
604: /*
605: * Take incoming datagram fragment and try to
606: * reassemble it into whole datagram. If a chain for
607: * reassembly of this datagram already exists, then it
608: * is given as fp; otherwise have to make a chain.
609: */
1.50 thorpej 610: struct mbuf *
1.25 cgd 611: ip_reass(ipqe, fp)
612: register struct ipqent *ipqe;
1.1 cgd 613: register struct ipq *fp;
614: {
1.50 thorpej 615: register struct mbuf *m = ipqe->ipqe_m;
1.25 cgd 616: register struct ipqent *nq, *p, *q;
617: struct ip *ip;
1.1 cgd 618: struct mbuf *t;
1.25 cgd 619: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 620: int i, next;
621:
1.75 thorpej 622: IPQ_LOCK_CHECK();
623:
1.1 cgd 624: /*
625: * Presence of header sizes in mbufs
626: * would confuse code below.
627: */
628: m->m_data += hlen;
629: m->m_len -= hlen;
630:
631: /*
632: * If first fragment to arrive, create a reassembly queue.
633: */
634: if (fp == 0) {
1.50 thorpej 635: MALLOC(fp, struct ipq *, sizeof (struct ipq),
636: M_FTABLE, M_NOWAIT);
637: if (fp == NULL)
1.1 cgd 638: goto dropfrag;
1.25 cgd 639: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 640: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 641: fp->ipq_p = ipqe->ipqe_ip->ip_p;
642: fp->ipq_id = ipqe->ipqe_ip->ip_id;
643: LIST_INIT(&fp->ipq_fragq);
644: fp->ipq_src = ipqe->ipqe_ip->ip_src;
645: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
646: p = NULL;
1.1 cgd 647: goto insert;
648: }
649:
650: /*
651: * Find a segment which begins after this one does.
652: */
1.25 cgd 653: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
654: p = q, q = q->ipqe_q.le_next)
655: if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off)
1.1 cgd 656: break;
657:
658: /*
659: * If there is a preceding segment, it may provide some of
660: * our data already. If so, drop the data from the incoming
661: * segment. If it provides all of our data, drop us.
662: */
1.25 cgd 663: if (p != NULL) {
664: i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len -
665: ipqe->ipqe_ip->ip_off;
1.1 cgd 666: if (i > 0) {
1.25 cgd 667: if (i >= ipqe->ipqe_ip->ip_len)
1.1 cgd 668: goto dropfrag;
1.50 thorpej 669: m_adj(ipqe->ipqe_m, i);
1.25 cgd 670: ipqe->ipqe_ip->ip_off += i;
671: ipqe->ipqe_ip->ip_len -= i;
1.1 cgd 672: }
673: }
674:
675: /*
676: * While we overlap succeeding segments trim them or,
677: * if they are completely covered, dequeue them.
678: */
1.25 cgd 679: for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len >
680: q->ipqe_ip->ip_off; q = nq) {
681: i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) -
682: q->ipqe_ip->ip_off;
683: if (i < q->ipqe_ip->ip_len) {
684: q->ipqe_ip->ip_len -= i;
685: q->ipqe_ip->ip_off += i;
1.50 thorpej 686: m_adj(q->ipqe_m, i);
1.1 cgd 687: break;
688: }
1.25 cgd 689: nq = q->ipqe_q.le_next;
1.50 thorpej 690: m_freem(q->ipqe_m);
1.25 cgd 691: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 692: pool_put(&ipqent_pool, q);
1.1 cgd 693: }
694:
695: insert:
696: /*
697: * Stick new segment in its place;
698: * check for complete reassembly.
699: */
1.25 cgd 700: if (p == NULL) {
701: LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
702: } else {
703: LIST_INSERT_AFTER(p, ipqe, ipqe_q);
704: }
1.1 cgd 705: next = 0;
1.25 cgd 706: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
707: p = q, q = q->ipqe_q.le_next) {
708: if (q->ipqe_ip->ip_off != next)
1.1 cgd 709: return (0);
1.25 cgd 710: next += q->ipqe_ip->ip_len;
1.1 cgd 711: }
1.25 cgd 712: if (p->ipqe_mff)
1.1 cgd 713: return (0);
714:
715: /*
1.41 thorpej 716: * Reassembly is complete. Check for a bogus message size and
717: * concatenate fragments.
1.1 cgd 718: */
1.25 cgd 719: q = fp->ipq_fragq.lh_first;
720: ip = q->ipqe_ip;
1.41 thorpej 721: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
722: ipstat.ips_toolong++;
723: ip_freef(fp);
724: return (0);
725: }
1.50 thorpej 726: m = q->ipqe_m;
1.1 cgd 727: t = m->m_next;
728: m->m_next = 0;
729: m_cat(m, t);
1.25 cgd 730: nq = q->ipqe_q.le_next;
1.72 thorpej 731: pool_put(&ipqent_pool, q);
1.25 cgd 732: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 733: t = q->ipqe_m;
1.25 cgd 734: nq = q->ipqe_q.le_next;
1.72 thorpej 735: pool_put(&ipqent_pool, q);
1.1 cgd 736: m_cat(m, t);
737: }
738:
739: /*
740: * Create header for new ip packet by
741: * modifying header of first packet;
742: * dequeue and discard fragment reassembly header.
743: * Make header visible.
744: */
745: ip->ip_len = next;
1.25 cgd 746: ip->ip_src = fp->ipq_src;
747: ip->ip_dst = fp->ipq_dst;
748: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 749: FREE(fp, M_FTABLE);
1.1 cgd 750: m->m_len += (ip->ip_hl << 2);
751: m->m_data -= (ip->ip_hl << 2);
752: /* some debugging cruft by sklower, below, will go away soon */
753: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
754: register int plen = 0;
1.50 thorpej 755: for (t = m; t; t = t->m_next)
756: plen += t->m_len;
757: m->m_pkthdr.len = plen;
1.1 cgd 758: }
1.50 thorpej 759: return (m);
1.1 cgd 760:
761: dropfrag:
762: ipstat.ips_fragdropped++;
763: m_freem(m);
1.72 thorpej 764: pool_put(&ipqent_pool, ipqe);
1.1 cgd 765: return (0);
766: }
767:
768: /*
769: * Free a fragment reassembly header and all
770: * associated datagrams.
771: */
1.8 mycroft 772: void
1.1 cgd 773: ip_freef(fp)
774: struct ipq *fp;
775: {
1.25 cgd 776: register struct ipqent *q, *p;
1.1 cgd 777:
1.75 thorpej 778: IPQ_LOCK_CHECK();
779:
1.25 cgd 780: for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) {
781: p = q->ipqe_q.le_next;
1.50 thorpej 782: m_freem(q->ipqe_m);
1.25 cgd 783: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 784: pool_put(&ipqent_pool, q);
1.1 cgd 785: }
1.25 cgd 786: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 787: FREE(fp, M_FTABLE);
1.1 cgd 788: }
789:
790: /*
791: * IP timer processing;
792: * if a timer expires on a reassembly
793: * queue, discard it.
794: */
1.8 mycroft 795: void
1.1 cgd 796: ip_slowtimo()
797: {
1.25 cgd 798: register struct ipq *fp, *nfp;
1.24 mycroft 799: int s = splsoftnet();
1.1 cgd 800:
1.75 thorpej 801: IPQ_LOCK();
1.25 cgd 802: for (fp = ipq.lh_first; fp != NULL; fp = nfp) {
803: nfp = fp->ipq_q.le_next;
804: if (--fp->ipq_ttl == 0) {
1.1 cgd 805: ipstat.ips_fragtimeout++;
1.25 cgd 806: ip_freef(fp);
1.1 cgd 807: }
808: }
1.75 thorpej 809: IPQ_UNLOCK();
1.63 matt 810: #ifdef GATEWAY
811: ipflow_slowtimo();
812: #endif
1.1 cgd 813: splx(s);
814: }
815:
816: /*
817: * Drain off all datagram fragments.
818: */
1.8 mycroft 819: void
1.1 cgd 820: ip_drain()
821: {
822:
1.75 thorpej 823: /*
824: * We may be called from a device's interrupt context. If
825: * the ipq is already busy, just bail out now.
826: */
827: if (ipq_lock_try() == 0)
828: return;
829:
1.25 cgd 830: while (ipq.lh_first != NULL) {
1.1 cgd 831: ipstat.ips_fragdropped++;
1.25 cgd 832: ip_freef(ipq.lh_first);
1.1 cgd 833: }
1.75 thorpej 834:
835: IPQ_UNLOCK();
1.1 cgd 836: }
837:
838: /*
839: * Do option processing on a datagram,
840: * possibly discarding it if bad options are encountered,
841: * or forwarding it if source-routed.
842: * Returns 1 if packet has been forwarded/freed,
843: * 0 if the packet should be processed further.
844: */
1.8 mycroft 845: int
1.1 cgd 846: ip_dooptions(m)
847: struct mbuf *m;
848: {
849: register struct ip *ip = mtod(m, struct ip *);
850: register u_char *cp;
851: register struct ip_timestamp *ipt;
852: register struct in_ifaddr *ia;
853: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.13 mycroft 854: struct in_addr *sin, dst;
1.1 cgd 855: n_time ntime;
856:
1.13 mycroft 857: dst = ip->ip_dst;
1.1 cgd 858: cp = (u_char *)(ip + 1);
859: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
860: for (; cnt > 0; cnt -= optlen, cp += optlen) {
861: opt = cp[IPOPT_OPTVAL];
862: if (opt == IPOPT_EOL)
863: break;
864: if (opt == IPOPT_NOP)
865: optlen = 1;
866: else {
867: optlen = cp[IPOPT_OLEN];
868: if (optlen <= 0 || optlen > cnt) {
869: code = &cp[IPOPT_OLEN] - (u_char *)ip;
870: goto bad;
871: }
872: }
873: switch (opt) {
874:
875: default:
876: break;
877:
878: /*
879: * Source routing with record.
880: * Find interface with current destination address.
881: * If none on this machine then drop if strictly routed,
882: * or do nothing if loosely routed.
883: * Record interface address and bring up next address
884: * component. If strictly routed make sure next
885: * address is on directly accessible net.
886: */
887: case IPOPT_LSRR:
888: case IPOPT_SSRR:
1.47 cjs 889: if (ip_allowsrcrt == 0) {
890: type = ICMP_UNREACH;
891: code = ICMP_UNREACH_NET_PROHIB;
892: goto bad;
893: }
1.1 cgd 894: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
895: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
896: goto bad;
897: }
898: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 899: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 900: if (ia == 0) {
901: if (opt == IPOPT_SSRR) {
902: type = ICMP_UNREACH;
903: code = ICMP_UNREACH_SRCFAIL;
904: goto bad;
905: }
906: /*
907: * Loose routing, and not at next destination
908: * yet; nothing to do except forward.
909: */
910: break;
911: }
912: off--; /* 0 origin */
913: if (off > optlen - sizeof(struct in_addr)) {
914: /*
915: * End of source route. Should be for us.
916: */
917: save_rte(cp, ip->ip_src);
918: break;
919: }
920: /*
921: * locate outgoing interface
922: */
923: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
924: sizeof(ipaddr.sin_addr));
925: if (opt == IPOPT_SSRR) {
926: #define INA struct in_ifaddr *
927: #define SA struct sockaddr *
1.29 mrg 928: ia = (INA)ifa_ifwithladdr((SA)&ipaddr);
1.1 cgd 929: } else
930: ia = ip_rtaddr(ipaddr.sin_addr);
931: if (ia == 0) {
932: type = ICMP_UNREACH;
933: code = ICMP_UNREACH_SRCFAIL;
934: goto bad;
935: }
936: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 937: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 938: (caddr_t)(cp + off), sizeof(struct in_addr));
939: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 940: /*
941: * Let ip_intr's mcast routing check handle mcast pkts
942: */
1.18 mycroft 943: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 944: break;
945:
946: case IPOPT_RR:
947: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
948: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
949: goto bad;
950: }
951: /*
952: * If no space remains, ignore.
953: */
954: off--; /* 0 origin */
955: if (off > optlen - sizeof(struct in_addr))
956: break;
957: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
958: sizeof(ipaddr.sin_addr));
959: /*
960: * locate outgoing interface; if we're the destination,
961: * use the incoming interface (should be same).
962: */
963: if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
964: (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
965: type = ICMP_UNREACH;
966: code = ICMP_UNREACH_HOST;
967: goto bad;
968: }
1.20 mycroft 969: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 970: (caddr_t)(cp + off), sizeof(struct in_addr));
971: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
972: break;
973:
974: case IPOPT_TS:
975: code = cp - (u_char *)ip;
976: ipt = (struct ip_timestamp *)cp;
977: if (ipt->ipt_len < 5)
978: goto bad;
1.15 cgd 979: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.1 cgd 980: if (++ipt->ipt_oflw == 0)
981: goto bad;
982: break;
983: }
984: sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
985: switch (ipt->ipt_flg) {
986:
987: case IPOPT_TS_TSONLY:
988: break;
989:
990: case IPOPT_TS_TSANDADDR:
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;
1.13 mycroft 994: ipaddr.sin_addr = dst;
995: ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
996: m->m_pkthdr.rcvif);
997: if (ia == 0)
998: continue;
1.20 mycroft 999: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 1000: (caddr_t)sin, sizeof(struct in_addr));
1001: ipt->ipt_ptr += sizeof(struct in_addr);
1002: break;
1003:
1004: case IPOPT_TS_PRESPEC:
1.66 thorpej 1005: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 1006: sizeof(struct in_addr) > ipt->ipt_len)
1007: goto bad;
1008: bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
1009: sizeof(struct in_addr));
1010: if (ifa_ifwithaddr((SA)&ipaddr) == 0)
1011: continue;
1012: ipt->ipt_ptr += sizeof(struct in_addr);
1013: break;
1014:
1015: default:
1016: goto bad;
1017: }
1018: ntime = iptime();
1019: bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
1020: sizeof(n_time));
1021: ipt->ipt_ptr += sizeof(n_time);
1022: }
1023: }
1024: if (forward) {
1.26 thorpej 1025: if (ip_forwsrcrt == 0) {
1026: type = ICMP_UNREACH;
1027: code = ICMP_UNREACH_SRCFAIL;
1028: goto bad;
1029: }
1.1 cgd 1030: ip_forward(m, 1);
1031: return (1);
1.13 mycroft 1032: }
1033: return (0);
1.1 cgd 1034: bad:
1.13 mycroft 1035: icmp_error(m, type, code, 0, 0);
1036: ipstat.ips_badoptions++;
1.1 cgd 1037: return (1);
1038: }
1039:
1040: /*
1041: * Given address of next destination (final or next hop),
1042: * return internet address info of interface to be used to get there.
1043: */
1044: struct in_ifaddr *
1045: ip_rtaddr(dst)
1046: struct in_addr dst;
1047: {
1048: register struct sockaddr_in *sin;
1049:
1.19 mycroft 1050: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1051:
1.35 mycroft 1052: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 1053: if (ipforward_rt.ro_rt) {
1054: RTFREE(ipforward_rt.ro_rt);
1055: ipforward_rt.ro_rt = 0;
1056: }
1057: sin->sin_family = AF_INET;
1058: sin->sin_len = sizeof(*sin);
1059: sin->sin_addr = dst;
1060:
1061: rtalloc(&ipforward_rt);
1062: }
1063: if (ipforward_rt.ro_rt == 0)
1064: return ((struct in_ifaddr *)0);
1.19 mycroft 1065: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 1066: }
1067:
1068: /*
1069: * Save incoming source route for use in replies,
1070: * to be picked up later by ip_srcroute if the receiver is interested.
1071: */
1.13 mycroft 1072: void
1.1 cgd 1073: save_rte(option, dst)
1074: u_char *option;
1075: struct in_addr dst;
1076: {
1077: unsigned olen;
1078:
1079: olen = option[IPOPT_OLEN];
1080: #ifdef DIAGNOSTIC
1081: if (ipprintfs)
1.39 christos 1082: printf("save_rte: olen %d\n", olen);
1.1 cgd 1083: #endif
1084: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1085: return;
1086: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
1087: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1088: ip_srcrt.dst = dst;
1089: }
1090:
1091: /*
1092: * Retrieve incoming source route for use in replies,
1093: * in the same form used by setsockopt.
1094: * The first hop is placed before the options, will be removed later.
1095: */
1096: struct mbuf *
1097: ip_srcroute()
1098: {
1099: register struct in_addr *p, *q;
1100: register struct mbuf *m;
1101:
1102: if (ip_nhops == 0)
1103: return ((struct mbuf *)0);
1104: m = m_get(M_DONTWAIT, MT_SOOPTS);
1105: if (m == 0)
1106: return ((struct mbuf *)0);
1107:
1.13 mycroft 1108: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1109:
1110: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1111: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1112: OPTSIZ;
1113: #ifdef DIAGNOSTIC
1114: if (ipprintfs)
1.39 christos 1115: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1116: #endif
1117:
1118: /*
1119: * First save first hop for return route
1120: */
1121: p = &ip_srcrt.route[ip_nhops - 1];
1122: *(mtod(m, struct in_addr *)) = *p--;
1123: #ifdef DIAGNOSTIC
1124: if (ipprintfs)
1.39 christos 1125: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1126: #endif
1127:
1128: /*
1129: * Copy option fields and padding (nop) to mbuf.
1130: */
1131: ip_srcrt.nop = IPOPT_NOP;
1132: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1133: bcopy((caddr_t)&ip_srcrt.nop,
1134: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1135: q = (struct in_addr *)(mtod(m, caddr_t) +
1136: sizeof(struct in_addr) + OPTSIZ);
1137: #undef OPTSIZ
1138: /*
1139: * Record return path as an IP source route,
1140: * reversing the path (pointers are now aligned).
1141: */
1142: while (p >= ip_srcrt.route) {
1143: #ifdef DIAGNOSTIC
1144: if (ipprintfs)
1.39 christos 1145: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1146: #endif
1147: *q++ = *p--;
1148: }
1149: /*
1150: * Last hop goes to final destination.
1151: */
1152: *q = ip_srcrt.dst;
1153: #ifdef DIAGNOSTIC
1154: if (ipprintfs)
1.39 christos 1155: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1156: #endif
1157: return (m);
1158: }
1159:
1160: /*
1161: * Strip out IP options, at higher
1162: * level protocol in the kernel.
1163: * Second argument is buffer to which options
1164: * will be moved, and return value is their length.
1165: * XXX should be deleted; last arg currently ignored.
1166: */
1.8 mycroft 1167: void
1.1 cgd 1168: ip_stripoptions(m, mopt)
1169: register struct mbuf *m;
1170: struct mbuf *mopt;
1171: {
1172: register int i;
1173: struct ip *ip = mtod(m, struct ip *);
1174: register caddr_t opts;
1175: int olen;
1176:
1.79 mycroft 1177: olen = (ip->ip_hl << 2) - sizeof (struct ip);
1.1 cgd 1178: opts = (caddr_t)(ip + 1);
1179: i = m->m_len - (sizeof (struct ip) + olen);
1180: bcopy(opts + olen, opts, (unsigned)i);
1181: m->m_len -= olen;
1182: if (m->m_flags & M_PKTHDR)
1183: m->m_pkthdr.len -= olen;
1.79 mycroft 1184: ip->ip_len -= olen;
1185: ip->ip_hl = sizeof (struct ip) >> 2;
1.1 cgd 1186: }
1187:
1.23 mycroft 1188: int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1189: 0, 0, 0, 0,
1190: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1191: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1192: EMSGSIZE, EHOSTUNREACH, 0, 0,
1193: 0, 0, 0, 0,
1194: ENOPROTOOPT
1195: };
1196:
1197: /*
1198: * Forward a packet. If some error occurs return the sender
1199: * an icmp packet. Note we can't always generate a meaningful
1200: * icmp message because icmp doesn't have a large enough repertoire
1201: * of codes and types.
1202: *
1203: * If not forwarding, just drop the packet. This could be confusing
1204: * if ipforwarding was zero but some routing protocol was advancing
1205: * us as a gateway to somewhere. However, we must let the routing
1206: * protocol deal with that.
1207: *
1208: * The srcrt parameter indicates whether the packet is being forwarded
1209: * via a source route.
1210: */
1.13 mycroft 1211: void
1.1 cgd 1212: ip_forward(m, srcrt)
1213: struct mbuf *m;
1214: int srcrt;
1215: {
1216: register struct ip *ip = mtod(m, struct ip *);
1217: register struct sockaddr_in *sin;
1218: register struct rtentry *rt;
1.28 christos 1219: int error, type = 0, code = 0;
1.1 cgd 1220: struct mbuf *mcopy;
1.13 mycroft 1221: n_long dest;
1222: struct ifnet *destifp;
1.1 cgd 1223:
1.13 mycroft 1224: dest = 0;
1.1 cgd 1225: #ifdef DIAGNOSTIC
1226: if (ipprintfs)
1.70 thorpej 1227: printf("forward: src %2.2x dst %2.2x ttl %x\n",
1228: ntohl(ip->ip_src.s_addr),
1229: ntohl(ip->ip_dst.s_addr), ip->ip_ttl);
1.1 cgd 1230: #endif
1231: if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1232: ipstat.ips_cantforward++;
1233: m_freem(m);
1234: return;
1235: }
1236: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1237: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1238: return;
1239: }
1240: ip->ip_ttl -= IPTTLDEC;
1241:
1.19 mycroft 1242: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1243: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1244: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1245: if (ipforward_rt.ro_rt) {
1246: RTFREE(ipforward_rt.ro_rt);
1247: ipforward_rt.ro_rt = 0;
1248: }
1249: sin->sin_family = AF_INET;
1.35 mycroft 1250: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1251: sin->sin_addr = ip->ip_dst;
1252:
1253: rtalloc(&ipforward_rt);
1254: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1255: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1256: return;
1257: }
1258: rt = ipforward_rt.ro_rt;
1259: }
1260:
1261: /*
1.34 mycroft 1262: * Save at most 68 bytes of the packet in case
1.1 cgd 1263: * we need to generate an ICMP message to the src.
1264: */
1.34 mycroft 1265: mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68));
1.1 cgd 1266:
1267: /*
1268: * If forwarding packet using same interface that it came in on,
1269: * perhaps should send a redirect to sender to shortcut a hop.
1270: * Only send redirect if source is sending directly to us,
1271: * and if packet was not source routed (or has any options).
1272: * Also, don't send redirect if forwarding using a default route
1273: * or a route modified by a redirect.
1274: */
1275: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1276: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1277: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1278: ipsendredirects && !srcrt) {
1.19 mycroft 1279: if (rt->rt_ifa &&
1280: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1281: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1282: if (rt->rt_flags & RTF_GATEWAY)
1283: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1284: else
1285: dest = ip->ip_dst.s_addr;
1286: /*
1287: * Router requirements says to only send host
1288: * redirects.
1289: */
1290: type = ICMP_REDIRECT;
1291: code = ICMP_REDIRECT_HOST;
1.1 cgd 1292: #ifdef DIAGNOSTIC
1.77 thorpej 1293: if (ipprintfs)
1294: printf("redirect (%d) to %x\n", code,
1295: (u_int32_t)dest);
1.1 cgd 1296: #endif
1297: }
1298: }
1299:
1.27 thorpej 1300: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1301: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1.1 cgd 1302: if (error)
1303: ipstat.ips_cantforward++;
1304: else {
1305: ipstat.ips_forward++;
1306: if (type)
1307: ipstat.ips_redirectsent++;
1308: else {
1.63 matt 1309: if (mcopy) {
1310: #ifdef GATEWAY
1.64 thorpej 1311: if (mcopy->m_flags & M_CANFASTFWD)
1312: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1313: #endif
1.1 cgd 1314: m_freem(mcopy);
1.63 matt 1315: }
1.1 cgd 1316: return;
1317: }
1318: }
1319: if (mcopy == NULL)
1320: return;
1.13 mycroft 1321: destifp = NULL;
1322:
1.1 cgd 1323: switch (error) {
1324:
1325: case 0: /* forwarded, but need redirect */
1326: /* type, code set above */
1327: break;
1328:
1329: case ENETUNREACH: /* shouldn't happen, checked above */
1330: case EHOSTUNREACH:
1331: case ENETDOWN:
1332: case EHOSTDOWN:
1333: default:
1334: type = ICMP_UNREACH;
1335: code = ICMP_UNREACH_HOST;
1336: break;
1337:
1338: case EMSGSIZE:
1339: type = ICMP_UNREACH;
1340: code = ICMP_UNREACH_NEEDFRAG;
1.13 mycroft 1341: if (ipforward_rt.ro_rt)
1342: destifp = ipforward_rt.ro_rt->rt_ifp;
1.1 cgd 1343: ipstat.ips_cantfrag++;
1344: break;
1345:
1346: case ENOBUFS:
1347: type = ICMP_SOURCEQUENCH;
1348: code = 0;
1349: break;
1350: }
1.13 mycroft 1351: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 1352: }
1353:
1354: void
1355: ip_savecontrol(inp, mp, ip, m)
1356: register struct inpcb *inp;
1357: register struct mbuf **mp;
1358: register struct ip *ip;
1359: register struct mbuf *m;
1360: {
1361:
1362: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1363: struct timeval tv;
1364:
1365: microtime(&tv);
1366: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1367: SCM_TIMESTAMP, SOL_SOCKET);
1368: if (*mp)
1369: mp = &(*mp)->m_next;
1370: }
1371: if (inp->inp_flags & INP_RECVDSTADDR) {
1372: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1373: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1374: if (*mp)
1375: mp = &(*mp)->m_next;
1376: }
1377: #ifdef notyet
1378: /*
1379: * XXX
1380: * Moving these out of udp_input() made them even more broken
1381: * than they already were.
1382: * - fenner@parc.xerox.com
1383: */
1384: /* options were tossed already */
1385: if (inp->inp_flags & INP_RECVOPTS) {
1386: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1387: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1388: if (*mp)
1389: mp = &(*mp)->m_next;
1390: }
1391: /* ip_srcroute doesn't do what we want here, need to fix */
1392: if (inp->inp_flags & INP_RECVRETOPTS) {
1393: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1394: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1395: if (*mp)
1396: mp = &(*mp)->m_next;
1397: }
1398: #endif
1399: if (inp->inp_flags & INP_RECVIF) {
1400: struct sockaddr_dl sdl;
1401:
1402: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1403: sdl.sdl_family = AF_LINK;
1404: sdl.sdl_index = m->m_pkthdr.rcvif ?
1405: m->m_pkthdr.rcvif->if_index : 0;
1406: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1407: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1408: IP_RECVIF, IPPROTO_IP);
1409: if (*mp)
1410: mp = &(*mp)->m_next;
1411: }
1.13 mycroft 1412: }
1413:
1414: int
1415: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1416: int *name;
1417: u_int namelen;
1418: void *oldp;
1419: size_t *oldlenp;
1420: void *newp;
1421: size_t newlen;
1422: {
1.52 thorpej 1423: extern int subnetsarelocal;
1424:
1.54 lukem 1425: int error, old;
1426:
1.13 mycroft 1427: /* All sysctl names at this level are terminal. */
1428: if (namelen != 1)
1429: return (ENOTDIR);
1430:
1431: switch (name[0]) {
1432: case IPCTL_FORWARDING:
1433: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1434: case IPCTL_SENDREDIRECTS:
1435: return (sysctl_int(oldp, oldlenp, newp, newlen,
1436: &ipsendredirects));
1437: case IPCTL_DEFTTL:
1438: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1439: #ifdef notyet
1440: case IPCTL_DEFMTU:
1441: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1442: #endif
1.26 thorpej 1443: case IPCTL_FORWSRCRT:
1.47 cjs 1444: /* Don't allow this to change in a secure environment. */
1.26 thorpej 1445: if (securelevel > 0)
1.46 cjs 1446: return (sysctl_rdint(oldp, oldlenp, newp,
1447: ip_forwsrcrt));
1448: else
1449: return (sysctl_int(oldp, oldlenp, newp, newlen,
1450: &ip_forwsrcrt));
1.27 thorpej 1451: case IPCTL_DIRECTEDBCAST:
1452: return (sysctl_int(oldp, oldlenp, newp, newlen,
1453: &ip_directedbcast));
1.47 cjs 1454: case IPCTL_ALLOWSRCRT:
1455: return (sysctl_int(oldp, oldlenp, newp, newlen,
1456: &ip_allowsrcrt));
1.52 thorpej 1457: case IPCTL_SUBNETSARELOCAL:
1458: return (sysctl_int(oldp, oldlenp, newp, newlen,
1459: &subnetsarelocal));
1.53 kml 1460: case IPCTL_MTUDISC:
1.60 kml 1461: error = sysctl_int(oldp, oldlenp, newp, newlen,
1462: &ip_mtudisc);
1463: if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) {
1464: ip_mtudisc_timeout_q =
1465: rt_timer_queue_create(ip_mtudisc_timeout);
1466: } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) {
1467: rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE);
1468: ip_mtudisc_timeout_q = NULL;
1469: }
1470: return error;
1.54 lukem 1471: case IPCTL_ANONPORTMIN:
1472: old = anonportmin;
1473: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1474: if (anonportmin >= anonportmax || anonportmin > 65535
1475: #ifndef IPNOPRIVPORTS
1476: || anonportmin < IPPORT_RESERVED
1477: #endif
1478: ) {
1479: anonportmin = old;
1480: return (EINVAL);
1481: }
1482: return (error);
1483: case IPCTL_ANONPORTMAX:
1484: old = anonportmax;
1485: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1486: if (anonportmin >= anonportmax || anonportmax > 65535
1487: #ifndef IPNOPRIVPORTS
1488: || anonportmax < IPPORT_RESERVED
1489: #endif
1490: ) {
1491: anonportmax = old;
1492: return (EINVAL);
1493: }
1.60 kml 1494: return (error);
1495: case IPCTL_MTUDISCTIMEOUT:
1496: error = sysctl_int(oldp, oldlenp, newp, newlen,
1497: &ip_mtudisc_timeout);
1498: if (ip_mtudisc_timeout_q != NULL)
1499: rt_timer_queue_change(ip_mtudisc_timeout_q,
1500: ip_mtudisc_timeout);
1.54 lukem 1501: return (error);
1.65 matt 1502: #ifdef GATEWAY
1503: case IPCTL_MAXFLOWS:
1.67 thorpej 1504: {
1505: int s;
1506:
1.65 matt 1507: error = sysctl_int(oldp, oldlenp, newp, newlen,
1508: &ip_maxflows);
1.67 thorpej 1509: s = splsoftnet();
1.65 matt 1510: ipflow_reap(0);
1.67 thorpej 1511: splx(s);
1.65 matt 1512: return (error);
1.67 thorpej 1513: }
1.65 matt 1514: #endif
1.13 mycroft 1515: default:
1516: return (EOPNOTSUPP);
1517: }
1518: /* NOTREACHED */
1.1 cgd 1519: }
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