Annotation of src/sys/netinet/ip_input.c, Revision 1.82.2.2
1.82.2.2! perry 1: /* $NetBSD: ip_input.c,v 1.86 1999/05/03 22:12:44 thorpej 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: */
358: if (len < hlen)
359: {
360: ipstat.ips_badlen++;
361: goto bad;
362: }
1.1 cgd 363:
364: /*
365: * Check that the amount of data in the buffers
366: * is as at least much as the IP header would have us expect.
367: * Trim mbufs if longer than we expect.
368: * Drop packet if shorter than we expect.
369: */
1.35 mycroft 370: if (m->m_pkthdr.len < len) {
1.1 cgd 371: ipstat.ips_tooshort++;
372: goto bad;
373: }
1.35 mycroft 374: if (m->m_pkthdr.len > len) {
1.1 cgd 375: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 376: m->m_len = len;
377: m->m_pkthdr.len = len;
1.1 cgd 378: } else
1.35 mycroft 379: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 380: }
381:
1.64 thorpej 382: /*
383: * Assume that we can create a fast-forward IP flow entry
384: * based on this packet.
385: */
386: m->m_flags |= M_CANFASTFWD;
387:
1.36 mrg 388: #ifdef PFIL_HOOKS
1.33 mrg 389: /*
1.64 thorpej 390: * Run through list of hooks for input packets. If there are any
391: * filters which require that additional packets in the flow are
392: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
393: * Note that filters must _never_ set this flag, as another filter
394: * in the list may have previously cleared it.
1.33 mrg 395: */
396: m0 = m;
1.59 mrg 397: for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next)
1.33 mrg 398: if (pfh->pfil_func) {
1.43 mrg 399: rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0);
400: if (rv)
1.40 veego 401: goto next;
1.68 sommerfe 402: m = m0;
403: if (m == NULL)
404: goto next;
405: ip = mtod(m, struct ip *);
1.33 mrg 406: }
1.36 mrg 407: #endif /* PFIL_HOOKS */
1.33 mrg 408:
1.1 cgd 409: /*
410: * Process options and, if not destined for us,
411: * ship it on. ip_dooptions returns 1 when an
412: * error was detected (causing an icmp message
413: * to be sent and the original packet to be freed).
414: */
415: ip_nhops = 0; /* for source routed packets */
416: if (hlen > sizeof (struct ip) && ip_dooptions(m))
417: goto next;
418:
419: /*
420: * Check our list of addresses, to see if the packet is for us.
421: */
1.57 tls 422: INADDR_TO_IA(ip->ip_dst, ia);
1.82.2.2! perry 423: if (ia != NULL)
! 424: goto ours;
1.57 tls 425: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
426: for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
427: ifa != NULL; ifa = ifa->ifa_list.tqe_next) {
428: if (ifa->ifa_addr->sa_family != AF_INET) continue;
429: ia = ifatoia(ifa);
1.35 mycroft 430: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
431: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 432: /*
433: * Look for all-0's host part (old broadcast addr),
434: * either for subnet or net.
435: */
436: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 437: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 438: goto ours;
1.57 tls 439: /*
440: * An interface with IP address zero accepts
441: * all packets that arrive on that interface.
442: */
443: if (in_nullhost(ia->ia_addr.sin_addr))
444: goto ours;
1.1 cgd 445: }
446: }
1.18 mycroft 447: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 448: struct in_multi *inm;
449: #ifdef MROUTING
450: extern struct socket *ip_mrouter;
1.10 brezak 451:
452: if (m->m_flags & M_EXT) {
453: if ((m = m_pullup(m, hlen)) == 0) {
454: ipstat.ips_toosmall++;
455: goto next;
456: }
457: ip = mtod(m, struct ip *);
458: }
1.4 hpeyerl 459:
460: if (ip_mrouter) {
461: /*
462: * If we are acting as a multicast router, all
463: * incoming multicast packets are passed to the
464: * kernel-level multicast forwarding function.
465: * The packet is returned (relatively) intact; if
466: * ip_mforward() returns a non-zero value, the packet
467: * must be discarded, else it may be accepted below.
468: *
469: * (The IP ident field is put in the same byte order
470: * as expected when ip_mforward() is called from
471: * ip_output().)
472: */
1.13 mycroft 473: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
474: ipstat.ips_cantforward++;
1.4 hpeyerl 475: m_freem(m);
476: goto next;
477: }
478:
479: /*
480: * The process-level routing demon needs to receive
481: * all multicast IGMP packets, whether or not this
482: * host belongs to their destination groups.
483: */
484: if (ip->ip_p == IPPROTO_IGMP)
485: goto ours;
1.13 mycroft 486: ipstat.ips_forward++;
1.4 hpeyerl 487: }
488: #endif
489: /*
490: * See if we belong to the destination multicast group on the
491: * arrival interface.
492: */
493: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
494: if (inm == NULL) {
1.13 mycroft 495: ipstat.ips_cantforward++;
1.4 hpeyerl 496: m_freem(m);
497: goto next;
498: }
499: goto ours;
500: }
1.19 mycroft 501: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 502: in_nullhost(ip->ip_dst))
1.1 cgd 503: goto ours;
504:
505: /*
506: * Not for us; forward if possible and desirable.
507: */
508: if (ipforwarding == 0) {
509: ipstat.ips_cantforward++;
510: m_freem(m);
511: } else
512: ip_forward(m, 0);
513: goto next;
514:
515: ours:
516: /*
517: * If offset or IP_MF are set, must reassemble.
518: * Otherwise, nothing need be done.
519: * (We could look in the reassembly queue to see
520: * if the packet was previously fragmented,
521: * but it's not worth the time; just let them time out.)
522: */
1.37 perry 523: if (ip->ip_off & ~(IP_DF|IP_RF)) {
1.1 cgd 524: /*
525: * Look for queue of fragments
526: * of this datagram.
527: */
1.75 thorpej 528: IPQ_LOCK();
1.25 cgd 529: for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next)
1.1 cgd 530: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 531: in_hosteq(ip->ip_src, fp->ipq_src) &&
532: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 533: ip->ip_p == fp->ipq_p)
534: goto found;
535: fp = 0;
536: found:
537:
538: /*
539: * Adjust ip_len to not reflect header,
1.25 cgd 540: * set ipqe_mff if more fragments are expected,
1.1 cgd 541: * convert offset of this to bytes.
542: */
543: ip->ip_len -= hlen;
1.25 cgd 544: mff = (ip->ip_off & IP_MF) != 0;
545: if (mff) {
1.16 cgd 546: /*
547: * Make sure that fragments have a data length
548: * that's a non-zero multiple of 8 bytes.
549: */
1.17 cgd 550: if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
1.16 cgd 551: ipstat.ips_badfrags++;
1.75 thorpej 552: IPQ_UNLOCK();
1.16 cgd 553: goto bad;
554: }
555: }
1.1 cgd 556: ip->ip_off <<= 3;
557:
558: /*
559: * If datagram marked as having more fragments
560: * or if this is not the first fragment,
561: * attempt reassembly; if it succeeds, proceed.
562: */
1.25 cgd 563: if (mff || ip->ip_off) {
1.1 cgd 564: ipstat.ips_fragments++;
1.72 thorpej 565: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.25 cgd 566: if (ipqe == NULL) {
567: ipstat.ips_rcvmemdrop++;
1.75 thorpej 568: IPQ_UNLOCK();
1.25 cgd 569: goto bad;
570: }
571: ipqe->ipqe_mff = mff;
1.50 thorpej 572: ipqe->ipqe_m = m;
1.25 cgd 573: ipqe->ipqe_ip = ip;
1.50 thorpej 574: m = ip_reass(ipqe, fp);
1.75 thorpej 575: if (m == 0) {
576: IPQ_UNLOCK();
1.1 cgd 577: goto next;
1.75 thorpej 578: }
1.13 mycroft 579: ipstat.ips_reassembled++;
1.50 thorpej 580: ip = mtod(m, struct ip *);
1.74 thorpej 581: hlen = ip->ip_hl << 2;
1.79 mycroft 582: ip->ip_len += hlen;
1.1 cgd 583: } else
584: if (fp)
585: ip_freef(fp);
1.75 thorpej 586: IPQ_UNLOCK();
1.79 mycroft 587: }
1.1 cgd 588:
589: /*
590: * Switch out to protocol's input routine.
591: */
1.82 aidan 592: #if IFA_STATS
593: ia->ia_ifa.ifa_data.ifad_inbytes += ip->ip_len;
594: #endif
1.1 cgd 595: ipstat.ips_delivered++;
596: (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
597: goto next;
598: bad:
599: m_freem(m);
600: goto next;
601: }
602:
603: /*
604: * Take incoming datagram fragment and try to
605: * reassemble it into whole datagram. If a chain for
606: * reassembly of this datagram already exists, then it
607: * is given as fp; otherwise have to make a chain.
608: */
1.50 thorpej 609: struct mbuf *
1.25 cgd 610: ip_reass(ipqe, fp)
611: register struct ipqent *ipqe;
1.1 cgd 612: register struct ipq *fp;
613: {
1.50 thorpej 614: register struct mbuf *m = ipqe->ipqe_m;
1.25 cgd 615: register struct ipqent *nq, *p, *q;
616: struct ip *ip;
1.1 cgd 617: struct mbuf *t;
1.25 cgd 618: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 619: int i, next;
620:
1.75 thorpej 621: IPQ_LOCK_CHECK();
622:
1.1 cgd 623: /*
624: * Presence of header sizes in mbufs
625: * would confuse code below.
626: */
627: m->m_data += hlen;
628: m->m_len -= hlen;
629:
630: /*
631: * If first fragment to arrive, create a reassembly queue.
632: */
633: if (fp == 0) {
1.50 thorpej 634: MALLOC(fp, struct ipq *, sizeof (struct ipq),
635: M_FTABLE, M_NOWAIT);
636: if (fp == NULL)
1.1 cgd 637: goto dropfrag;
1.25 cgd 638: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 639: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 640: fp->ipq_p = ipqe->ipqe_ip->ip_p;
641: fp->ipq_id = ipqe->ipqe_ip->ip_id;
642: LIST_INIT(&fp->ipq_fragq);
643: fp->ipq_src = ipqe->ipqe_ip->ip_src;
644: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
645: p = NULL;
1.1 cgd 646: goto insert;
647: }
648:
649: /*
650: * Find a segment which begins after this one does.
651: */
1.25 cgd 652: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
653: p = q, q = q->ipqe_q.le_next)
654: if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off)
1.1 cgd 655: break;
656:
657: /*
658: * If there is a preceding segment, it may provide some of
659: * our data already. If so, drop the data from the incoming
660: * segment. If it provides all of our data, drop us.
661: */
1.25 cgd 662: if (p != NULL) {
663: i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len -
664: ipqe->ipqe_ip->ip_off;
1.1 cgd 665: if (i > 0) {
1.25 cgd 666: if (i >= ipqe->ipqe_ip->ip_len)
1.1 cgd 667: goto dropfrag;
1.50 thorpej 668: m_adj(ipqe->ipqe_m, i);
1.25 cgd 669: ipqe->ipqe_ip->ip_off += i;
670: ipqe->ipqe_ip->ip_len -= i;
1.1 cgd 671: }
672: }
673:
674: /*
675: * While we overlap succeeding segments trim them or,
676: * if they are completely covered, dequeue them.
677: */
1.25 cgd 678: for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len >
679: q->ipqe_ip->ip_off; q = nq) {
680: i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) -
681: q->ipqe_ip->ip_off;
682: if (i < q->ipqe_ip->ip_len) {
683: q->ipqe_ip->ip_len -= i;
684: q->ipqe_ip->ip_off += i;
1.50 thorpej 685: m_adj(q->ipqe_m, i);
1.1 cgd 686: break;
687: }
1.25 cgd 688: nq = q->ipqe_q.le_next;
1.50 thorpej 689: m_freem(q->ipqe_m);
1.25 cgd 690: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 691: pool_put(&ipqent_pool, q);
1.1 cgd 692: }
693:
694: insert:
695: /*
696: * Stick new segment in its place;
697: * check for complete reassembly.
698: */
1.25 cgd 699: if (p == NULL) {
700: LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
701: } else {
702: LIST_INSERT_AFTER(p, ipqe, ipqe_q);
703: }
1.1 cgd 704: next = 0;
1.25 cgd 705: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
706: p = q, q = q->ipqe_q.le_next) {
707: if (q->ipqe_ip->ip_off != next)
1.1 cgd 708: return (0);
1.25 cgd 709: next += q->ipqe_ip->ip_len;
1.1 cgd 710: }
1.25 cgd 711: if (p->ipqe_mff)
1.1 cgd 712: return (0);
713:
714: /*
1.41 thorpej 715: * Reassembly is complete. Check for a bogus message size and
716: * concatenate fragments.
1.1 cgd 717: */
1.25 cgd 718: q = fp->ipq_fragq.lh_first;
719: ip = q->ipqe_ip;
1.41 thorpej 720: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
721: ipstat.ips_toolong++;
722: ip_freef(fp);
723: return (0);
724: }
1.50 thorpej 725: m = q->ipqe_m;
1.1 cgd 726: t = m->m_next;
727: m->m_next = 0;
728: m_cat(m, t);
1.25 cgd 729: nq = q->ipqe_q.le_next;
1.72 thorpej 730: pool_put(&ipqent_pool, q);
1.25 cgd 731: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 732: t = q->ipqe_m;
1.25 cgd 733: nq = q->ipqe_q.le_next;
1.72 thorpej 734: pool_put(&ipqent_pool, q);
1.1 cgd 735: m_cat(m, t);
736: }
737:
738: /*
739: * Create header for new ip packet by
740: * modifying header of first packet;
741: * dequeue and discard fragment reassembly header.
742: * Make header visible.
743: */
744: ip->ip_len = next;
1.25 cgd 745: ip->ip_src = fp->ipq_src;
746: ip->ip_dst = fp->ipq_dst;
747: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 748: FREE(fp, M_FTABLE);
1.1 cgd 749: m->m_len += (ip->ip_hl << 2);
750: m->m_data -= (ip->ip_hl << 2);
751: /* some debugging cruft by sklower, below, will go away soon */
752: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
753: register int plen = 0;
1.50 thorpej 754: for (t = m; t; t = t->m_next)
755: plen += t->m_len;
756: m->m_pkthdr.len = plen;
1.1 cgd 757: }
1.50 thorpej 758: return (m);
1.1 cgd 759:
760: dropfrag:
761: ipstat.ips_fragdropped++;
762: m_freem(m);
1.72 thorpej 763: pool_put(&ipqent_pool, ipqe);
1.1 cgd 764: return (0);
765: }
766:
767: /*
768: * Free a fragment reassembly header and all
769: * associated datagrams.
770: */
1.8 mycroft 771: void
1.1 cgd 772: ip_freef(fp)
773: struct ipq *fp;
774: {
1.25 cgd 775: register struct ipqent *q, *p;
1.1 cgd 776:
1.75 thorpej 777: IPQ_LOCK_CHECK();
778:
1.25 cgd 779: for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) {
780: p = q->ipqe_q.le_next;
1.50 thorpej 781: m_freem(q->ipqe_m);
1.25 cgd 782: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 783: pool_put(&ipqent_pool, q);
1.1 cgd 784: }
1.25 cgd 785: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 786: FREE(fp, M_FTABLE);
1.1 cgd 787: }
788:
789: /*
790: * IP timer processing;
791: * if a timer expires on a reassembly
792: * queue, discard it.
793: */
1.8 mycroft 794: void
1.1 cgd 795: ip_slowtimo()
796: {
1.25 cgd 797: register struct ipq *fp, *nfp;
1.24 mycroft 798: int s = splsoftnet();
1.1 cgd 799:
1.75 thorpej 800: IPQ_LOCK();
1.25 cgd 801: for (fp = ipq.lh_first; fp != NULL; fp = nfp) {
802: nfp = fp->ipq_q.le_next;
803: if (--fp->ipq_ttl == 0) {
1.1 cgd 804: ipstat.ips_fragtimeout++;
1.25 cgd 805: ip_freef(fp);
1.1 cgd 806: }
807: }
1.75 thorpej 808: IPQ_UNLOCK();
1.63 matt 809: #ifdef GATEWAY
810: ipflow_slowtimo();
811: #endif
1.1 cgd 812: splx(s);
813: }
814:
815: /*
816: * Drain off all datagram fragments.
817: */
1.8 mycroft 818: void
1.1 cgd 819: ip_drain()
820: {
821:
1.75 thorpej 822: /*
823: * We may be called from a device's interrupt context. If
824: * the ipq is already busy, just bail out now.
825: */
826: if (ipq_lock_try() == 0)
827: return;
828:
1.25 cgd 829: while (ipq.lh_first != NULL) {
1.1 cgd 830: ipstat.ips_fragdropped++;
1.25 cgd 831: ip_freef(ipq.lh_first);
1.1 cgd 832: }
1.75 thorpej 833:
834: IPQ_UNLOCK();
1.1 cgd 835: }
836:
837: /*
838: * Do option processing on a datagram,
839: * possibly discarding it if bad options are encountered,
840: * or forwarding it if source-routed.
841: * Returns 1 if packet has been forwarded/freed,
842: * 0 if the packet should be processed further.
843: */
1.8 mycroft 844: int
1.1 cgd 845: ip_dooptions(m)
846: struct mbuf *m;
847: {
848: register struct ip *ip = mtod(m, struct ip *);
849: register u_char *cp;
850: register struct ip_timestamp *ipt;
851: register struct in_ifaddr *ia;
852: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.13 mycroft 853: struct in_addr *sin, dst;
1.1 cgd 854: n_time ntime;
855:
1.13 mycroft 856: dst = ip->ip_dst;
1.1 cgd 857: cp = (u_char *)(ip + 1);
858: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
859: for (; cnt > 0; cnt -= optlen, cp += optlen) {
860: opt = cp[IPOPT_OPTVAL];
861: if (opt == IPOPT_EOL)
862: break;
863: if (opt == IPOPT_NOP)
864: optlen = 1;
865: else {
866: optlen = cp[IPOPT_OLEN];
867: if (optlen <= 0 || optlen > cnt) {
868: code = &cp[IPOPT_OLEN] - (u_char *)ip;
869: goto bad;
870: }
871: }
872: switch (opt) {
873:
874: default:
875: break;
876:
877: /*
878: * Source routing with record.
879: * Find interface with current destination address.
880: * If none on this machine then drop if strictly routed,
881: * or do nothing if loosely routed.
882: * Record interface address and bring up next address
883: * component. If strictly routed make sure next
884: * address is on directly accessible net.
885: */
886: case IPOPT_LSRR:
887: case IPOPT_SSRR:
1.47 cjs 888: if (ip_allowsrcrt == 0) {
889: type = ICMP_UNREACH;
890: code = ICMP_UNREACH_NET_PROHIB;
891: goto bad;
892: }
1.1 cgd 893: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
894: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
895: goto bad;
896: }
897: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 898: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 899: if (ia == 0) {
900: if (opt == IPOPT_SSRR) {
901: type = ICMP_UNREACH;
902: code = ICMP_UNREACH_SRCFAIL;
903: goto bad;
904: }
905: /*
906: * Loose routing, and not at next destination
907: * yet; nothing to do except forward.
908: */
909: break;
910: }
911: off--; /* 0 origin */
912: if (off > optlen - sizeof(struct in_addr)) {
913: /*
914: * End of source route. Should be for us.
915: */
916: save_rte(cp, ip->ip_src);
917: break;
918: }
919: /*
920: * locate outgoing interface
921: */
922: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
923: sizeof(ipaddr.sin_addr));
924: if (opt == IPOPT_SSRR) {
925: #define INA struct in_ifaddr *
926: #define SA struct sockaddr *
1.29 mrg 927: ia = (INA)ifa_ifwithladdr((SA)&ipaddr);
1.1 cgd 928: } else
929: ia = ip_rtaddr(ipaddr.sin_addr);
930: if (ia == 0) {
931: type = ICMP_UNREACH;
932: code = ICMP_UNREACH_SRCFAIL;
933: goto bad;
934: }
935: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 936: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 937: (caddr_t)(cp + off), sizeof(struct in_addr));
938: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 939: /*
940: * Let ip_intr's mcast routing check handle mcast pkts
941: */
1.18 mycroft 942: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 943: break;
944:
945: case IPOPT_RR:
946: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
947: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
948: goto bad;
949: }
950: /*
951: * If no space remains, ignore.
952: */
953: off--; /* 0 origin */
954: if (off > optlen - sizeof(struct in_addr))
955: break;
956: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
957: sizeof(ipaddr.sin_addr));
958: /*
959: * locate outgoing interface; if we're the destination,
960: * use the incoming interface (should be same).
961: */
962: if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
963: (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
964: type = ICMP_UNREACH;
965: code = ICMP_UNREACH_HOST;
966: goto bad;
967: }
1.20 mycroft 968: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 969: (caddr_t)(cp + off), sizeof(struct in_addr));
970: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
971: break;
972:
973: case IPOPT_TS:
974: code = cp - (u_char *)ip;
975: ipt = (struct ip_timestamp *)cp;
976: if (ipt->ipt_len < 5)
977: goto bad;
1.15 cgd 978: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.1 cgd 979: if (++ipt->ipt_oflw == 0)
980: goto bad;
981: break;
982: }
983: sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
984: switch (ipt->ipt_flg) {
985:
986: case IPOPT_TS_TSONLY:
987: break;
988:
989: case IPOPT_TS_TSANDADDR:
1.66 thorpej 990: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 991: sizeof(struct in_addr) > ipt->ipt_len)
992: goto bad;
1.13 mycroft 993: ipaddr.sin_addr = dst;
994: ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
995: m->m_pkthdr.rcvif);
996: if (ia == 0)
997: continue;
1.20 mycroft 998: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 999: (caddr_t)sin, sizeof(struct in_addr));
1000: ipt->ipt_ptr += sizeof(struct in_addr);
1001: break;
1002:
1003: case IPOPT_TS_PRESPEC:
1.66 thorpej 1004: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 1005: sizeof(struct in_addr) > ipt->ipt_len)
1006: goto bad;
1007: bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
1008: sizeof(struct in_addr));
1009: if (ifa_ifwithaddr((SA)&ipaddr) == 0)
1010: continue;
1011: ipt->ipt_ptr += sizeof(struct in_addr);
1012: break;
1013:
1014: default:
1015: goto bad;
1016: }
1017: ntime = iptime();
1018: bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
1019: sizeof(n_time));
1020: ipt->ipt_ptr += sizeof(n_time);
1021: }
1022: }
1023: if (forward) {
1.26 thorpej 1024: if (ip_forwsrcrt == 0) {
1025: type = ICMP_UNREACH;
1026: code = ICMP_UNREACH_SRCFAIL;
1027: goto bad;
1028: }
1.1 cgd 1029: ip_forward(m, 1);
1030: return (1);
1.13 mycroft 1031: }
1032: return (0);
1.1 cgd 1033: bad:
1.13 mycroft 1034: icmp_error(m, type, code, 0, 0);
1035: ipstat.ips_badoptions++;
1.1 cgd 1036: return (1);
1037: }
1038:
1039: /*
1040: * Given address of next destination (final or next hop),
1041: * return internet address info of interface to be used to get there.
1042: */
1043: struct in_ifaddr *
1044: ip_rtaddr(dst)
1045: struct in_addr dst;
1046: {
1047: register struct sockaddr_in *sin;
1048:
1.19 mycroft 1049: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1050:
1.35 mycroft 1051: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 1052: if (ipforward_rt.ro_rt) {
1053: RTFREE(ipforward_rt.ro_rt);
1054: ipforward_rt.ro_rt = 0;
1055: }
1056: sin->sin_family = AF_INET;
1057: sin->sin_len = sizeof(*sin);
1058: sin->sin_addr = dst;
1059:
1060: rtalloc(&ipforward_rt);
1061: }
1062: if (ipforward_rt.ro_rt == 0)
1063: return ((struct in_ifaddr *)0);
1.19 mycroft 1064: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 1065: }
1066:
1067: /*
1068: * Save incoming source route for use in replies,
1069: * to be picked up later by ip_srcroute if the receiver is interested.
1070: */
1.13 mycroft 1071: void
1.1 cgd 1072: save_rte(option, dst)
1073: u_char *option;
1074: struct in_addr dst;
1075: {
1076: unsigned olen;
1077:
1078: olen = option[IPOPT_OLEN];
1079: #ifdef DIAGNOSTIC
1080: if (ipprintfs)
1.39 christos 1081: printf("save_rte: olen %d\n", olen);
1.1 cgd 1082: #endif
1083: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1084: return;
1085: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
1086: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1087: ip_srcrt.dst = dst;
1088: }
1089:
1090: /*
1091: * Retrieve incoming source route for use in replies,
1092: * in the same form used by setsockopt.
1093: * The first hop is placed before the options, will be removed later.
1094: */
1095: struct mbuf *
1096: ip_srcroute()
1097: {
1098: register struct in_addr *p, *q;
1099: register struct mbuf *m;
1100:
1101: if (ip_nhops == 0)
1102: return ((struct mbuf *)0);
1103: m = m_get(M_DONTWAIT, MT_SOOPTS);
1104: if (m == 0)
1105: return ((struct mbuf *)0);
1106:
1.13 mycroft 1107: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1108:
1109: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1110: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1111: OPTSIZ;
1112: #ifdef DIAGNOSTIC
1113: if (ipprintfs)
1.39 christos 1114: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1115: #endif
1116:
1117: /*
1118: * First save first hop for return route
1119: */
1120: p = &ip_srcrt.route[ip_nhops - 1];
1121: *(mtod(m, struct in_addr *)) = *p--;
1122: #ifdef DIAGNOSTIC
1123: if (ipprintfs)
1.39 christos 1124: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1125: #endif
1126:
1127: /*
1128: * Copy option fields and padding (nop) to mbuf.
1129: */
1130: ip_srcrt.nop = IPOPT_NOP;
1131: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1132: bcopy((caddr_t)&ip_srcrt.nop,
1133: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1134: q = (struct in_addr *)(mtod(m, caddr_t) +
1135: sizeof(struct in_addr) + OPTSIZ);
1136: #undef OPTSIZ
1137: /*
1138: * Record return path as an IP source route,
1139: * reversing the path (pointers are now aligned).
1140: */
1141: while (p >= ip_srcrt.route) {
1142: #ifdef DIAGNOSTIC
1143: if (ipprintfs)
1.39 christos 1144: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1145: #endif
1146: *q++ = *p--;
1147: }
1148: /*
1149: * Last hop goes to final destination.
1150: */
1151: *q = ip_srcrt.dst;
1152: #ifdef DIAGNOSTIC
1153: if (ipprintfs)
1.39 christos 1154: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1155: #endif
1156: return (m);
1157: }
1158:
1159: /*
1160: * Strip out IP options, at higher
1161: * level protocol in the kernel.
1162: * Second argument is buffer to which options
1163: * will be moved, and return value is their length.
1164: * XXX should be deleted; last arg currently ignored.
1165: */
1.8 mycroft 1166: void
1.1 cgd 1167: ip_stripoptions(m, mopt)
1168: register struct mbuf *m;
1169: struct mbuf *mopt;
1170: {
1171: register int i;
1172: struct ip *ip = mtod(m, struct ip *);
1173: register caddr_t opts;
1174: int olen;
1175:
1.79 mycroft 1176: olen = (ip->ip_hl << 2) - sizeof (struct ip);
1.1 cgd 1177: opts = (caddr_t)(ip + 1);
1178: i = m->m_len - (sizeof (struct ip) + olen);
1179: bcopy(opts + olen, opts, (unsigned)i);
1180: m->m_len -= olen;
1181: if (m->m_flags & M_PKTHDR)
1182: m->m_pkthdr.len -= olen;
1.79 mycroft 1183: ip->ip_len -= olen;
1184: ip->ip_hl = sizeof (struct ip) >> 2;
1.1 cgd 1185: }
1186:
1.23 mycroft 1187: int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1188: 0, 0, 0, 0,
1189: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1190: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1191: EMSGSIZE, EHOSTUNREACH, 0, 0,
1192: 0, 0, 0, 0,
1193: ENOPROTOOPT
1194: };
1195:
1196: /*
1197: * Forward a packet. If some error occurs return the sender
1198: * an icmp packet. Note we can't always generate a meaningful
1199: * icmp message because icmp doesn't have a large enough repertoire
1200: * of codes and types.
1201: *
1202: * If not forwarding, just drop the packet. This could be confusing
1203: * if ipforwarding was zero but some routing protocol was advancing
1204: * us as a gateway to somewhere. However, we must let the routing
1205: * protocol deal with that.
1206: *
1207: * The srcrt parameter indicates whether the packet is being forwarded
1208: * via a source route.
1209: */
1.13 mycroft 1210: void
1.1 cgd 1211: ip_forward(m, srcrt)
1212: struct mbuf *m;
1213: int srcrt;
1214: {
1215: register struct ip *ip = mtod(m, struct ip *);
1216: register struct sockaddr_in *sin;
1217: register struct rtentry *rt;
1.28 christos 1218: int error, type = 0, code = 0;
1.1 cgd 1219: struct mbuf *mcopy;
1.13 mycroft 1220: n_long dest;
1221: struct ifnet *destifp;
1.1 cgd 1222:
1.13 mycroft 1223: dest = 0;
1.1 cgd 1224: #ifdef DIAGNOSTIC
1225: if (ipprintfs)
1.70 thorpej 1226: printf("forward: src %2.2x dst %2.2x ttl %x\n",
1227: ntohl(ip->ip_src.s_addr),
1228: ntohl(ip->ip_dst.s_addr), ip->ip_ttl);
1.1 cgd 1229: #endif
1230: if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1231: ipstat.ips_cantforward++;
1232: m_freem(m);
1233: return;
1234: }
1235: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1236: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1237: return;
1238: }
1239: ip->ip_ttl -= IPTTLDEC;
1240:
1.19 mycroft 1241: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1242: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1243: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1244: if (ipforward_rt.ro_rt) {
1245: RTFREE(ipforward_rt.ro_rt);
1246: ipforward_rt.ro_rt = 0;
1247: }
1248: sin->sin_family = AF_INET;
1.35 mycroft 1249: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1250: sin->sin_addr = ip->ip_dst;
1251:
1252: rtalloc(&ipforward_rt);
1253: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1254: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1255: return;
1256: }
1257: rt = ipforward_rt.ro_rt;
1258: }
1259:
1260: /*
1.34 mycroft 1261: * Save at most 68 bytes of the packet in case
1.1 cgd 1262: * we need to generate an ICMP message to the src.
1263: */
1.34 mycroft 1264: mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68));
1.1 cgd 1265:
1266: /*
1267: * If forwarding packet using same interface that it came in on,
1268: * perhaps should send a redirect to sender to shortcut a hop.
1269: * Only send redirect if source is sending directly to us,
1270: * and if packet was not source routed (or has any options).
1271: * Also, don't send redirect if forwarding using a default route
1272: * or a route modified by a redirect.
1273: */
1274: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1275: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1276: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1277: ipsendredirects && !srcrt) {
1.19 mycroft 1278: if (rt->rt_ifa &&
1279: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1280: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1281: if (rt->rt_flags & RTF_GATEWAY)
1282: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1283: else
1284: dest = ip->ip_dst.s_addr;
1285: /*
1286: * Router requirements says to only send host
1287: * redirects.
1288: */
1289: type = ICMP_REDIRECT;
1290: code = ICMP_REDIRECT_HOST;
1.1 cgd 1291: #ifdef DIAGNOSTIC
1.77 thorpej 1292: if (ipprintfs)
1293: printf("redirect (%d) to %x\n", code,
1294: (u_int32_t)dest);
1.1 cgd 1295: #endif
1296: }
1297: }
1298:
1.27 thorpej 1299: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1300: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1.1 cgd 1301: if (error)
1302: ipstat.ips_cantforward++;
1303: else {
1304: ipstat.ips_forward++;
1305: if (type)
1306: ipstat.ips_redirectsent++;
1307: else {
1.63 matt 1308: if (mcopy) {
1309: #ifdef GATEWAY
1.64 thorpej 1310: if (mcopy->m_flags & M_CANFASTFWD)
1311: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1312: #endif
1.1 cgd 1313: m_freem(mcopy);
1.63 matt 1314: }
1.1 cgd 1315: return;
1316: }
1317: }
1318: if (mcopy == NULL)
1319: return;
1.13 mycroft 1320: destifp = NULL;
1321:
1.1 cgd 1322: switch (error) {
1323:
1324: case 0: /* forwarded, but need redirect */
1325: /* type, code set above */
1326: break;
1327:
1328: case ENETUNREACH: /* shouldn't happen, checked above */
1329: case EHOSTUNREACH:
1330: case ENETDOWN:
1331: case EHOSTDOWN:
1332: default:
1333: type = ICMP_UNREACH;
1334: code = ICMP_UNREACH_HOST;
1335: break;
1336:
1337: case EMSGSIZE:
1338: type = ICMP_UNREACH;
1339: code = ICMP_UNREACH_NEEDFRAG;
1.13 mycroft 1340: if (ipforward_rt.ro_rt)
1341: destifp = ipforward_rt.ro_rt->rt_ifp;
1.1 cgd 1342: ipstat.ips_cantfrag++;
1343: break;
1344:
1345: case ENOBUFS:
1346: type = ICMP_SOURCEQUENCH;
1347: code = 0;
1348: break;
1349: }
1.13 mycroft 1350: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 1351: }
1352:
1353: void
1354: ip_savecontrol(inp, mp, ip, m)
1355: register struct inpcb *inp;
1356: register struct mbuf **mp;
1357: register struct ip *ip;
1358: register struct mbuf *m;
1359: {
1360:
1361: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1362: struct timeval tv;
1363:
1364: microtime(&tv);
1365: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1366: SCM_TIMESTAMP, SOL_SOCKET);
1367: if (*mp)
1368: mp = &(*mp)->m_next;
1369: }
1370: if (inp->inp_flags & INP_RECVDSTADDR) {
1371: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1372: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1373: if (*mp)
1374: mp = &(*mp)->m_next;
1375: }
1376: #ifdef notyet
1377: /*
1378: * XXX
1379: * Moving these out of udp_input() made them even more broken
1380: * than they already were.
1381: * - fenner@parc.xerox.com
1382: */
1383: /* options were tossed already */
1384: if (inp->inp_flags & INP_RECVOPTS) {
1385: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1386: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1387: if (*mp)
1388: mp = &(*mp)->m_next;
1389: }
1390: /* ip_srcroute doesn't do what we want here, need to fix */
1391: if (inp->inp_flags & INP_RECVRETOPTS) {
1392: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1393: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1394: if (*mp)
1395: mp = &(*mp)->m_next;
1396: }
1397: #endif
1398: if (inp->inp_flags & INP_RECVIF) {
1399: struct sockaddr_dl sdl;
1400:
1401: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1402: sdl.sdl_family = AF_LINK;
1403: sdl.sdl_index = m->m_pkthdr.rcvif ?
1404: m->m_pkthdr.rcvif->if_index : 0;
1405: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1406: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1407: IP_RECVIF, IPPROTO_IP);
1408: if (*mp)
1409: mp = &(*mp)->m_next;
1410: }
1.13 mycroft 1411: }
1412:
1413: int
1414: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1415: int *name;
1416: u_int namelen;
1417: void *oldp;
1418: size_t *oldlenp;
1419: void *newp;
1420: size_t newlen;
1421: {
1.52 thorpej 1422: extern int subnetsarelocal;
1423:
1.54 lukem 1424: int error, old;
1425:
1.13 mycroft 1426: /* All sysctl names at this level are terminal. */
1427: if (namelen != 1)
1428: return (ENOTDIR);
1429:
1430: switch (name[0]) {
1431: case IPCTL_FORWARDING:
1432: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1433: case IPCTL_SENDREDIRECTS:
1434: return (sysctl_int(oldp, oldlenp, newp, newlen,
1435: &ipsendredirects));
1436: case IPCTL_DEFTTL:
1437: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1438: #ifdef notyet
1439: case IPCTL_DEFMTU:
1440: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1441: #endif
1.26 thorpej 1442: case IPCTL_FORWSRCRT:
1.47 cjs 1443: /* Don't allow this to change in a secure environment. */
1.26 thorpej 1444: if (securelevel > 0)
1.46 cjs 1445: return (sysctl_rdint(oldp, oldlenp, newp,
1446: ip_forwsrcrt));
1447: else
1448: return (sysctl_int(oldp, oldlenp, newp, newlen,
1449: &ip_forwsrcrt));
1.27 thorpej 1450: case IPCTL_DIRECTEDBCAST:
1451: return (sysctl_int(oldp, oldlenp, newp, newlen,
1452: &ip_directedbcast));
1.47 cjs 1453: case IPCTL_ALLOWSRCRT:
1454: return (sysctl_int(oldp, oldlenp, newp, newlen,
1455: &ip_allowsrcrt));
1.52 thorpej 1456: case IPCTL_SUBNETSARELOCAL:
1457: return (sysctl_int(oldp, oldlenp, newp, newlen,
1458: &subnetsarelocal));
1.53 kml 1459: case IPCTL_MTUDISC:
1.60 kml 1460: error = sysctl_int(oldp, oldlenp, newp, newlen,
1461: &ip_mtudisc);
1462: if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) {
1463: ip_mtudisc_timeout_q =
1464: rt_timer_queue_create(ip_mtudisc_timeout);
1465: } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) {
1466: rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE);
1467: ip_mtudisc_timeout_q = NULL;
1468: }
1469: return error;
1.54 lukem 1470: case IPCTL_ANONPORTMIN:
1471: old = anonportmin;
1472: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1473: if (anonportmin >= anonportmax || anonportmin > 65535
1474: #ifndef IPNOPRIVPORTS
1475: || anonportmin < IPPORT_RESERVED
1476: #endif
1477: ) {
1478: anonportmin = old;
1479: return (EINVAL);
1480: }
1481: return (error);
1482: case IPCTL_ANONPORTMAX:
1483: old = anonportmax;
1484: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1485: if (anonportmin >= anonportmax || anonportmax > 65535
1486: #ifndef IPNOPRIVPORTS
1487: || anonportmax < IPPORT_RESERVED
1488: #endif
1489: ) {
1490: anonportmax = old;
1491: return (EINVAL);
1492: }
1.60 kml 1493: return (error);
1494: case IPCTL_MTUDISCTIMEOUT:
1495: error = sysctl_int(oldp, oldlenp, newp, newlen,
1496: &ip_mtudisc_timeout);
1497: if (ip_mtudisc_timeout_q != NULL)
1498: rt_timer_queue_change(ip_mtudisc_timeout_q,
1499: ip_mtudisc_timeout);
1.54 lukem 1500: return (error);
1.65 matt 1501: #ifdef GATEWAY
1502: case IPCTL_MAXFLOWS:
1.67 thorpej 1503: {
1504: int s;
1505:
1.65 matt 1506: error = sysctl_int(oldp, oldlenp, newp, newlen,
1507: &ip_maxflows);
1.67 thorpej 1508: s = splsoftnet();
1.65 matt 1509: ipflow_reap(0);
1.67 thorpej 1510: splx(s);
1.65 matt 1511: return (error);
1.67 thorpej 1512: }
1.65 matt 1513: #endif
1.13 mycroft 1514: default:
1515: return (EOPNOTSUPP);
1516: }
1517: /* NOTREACHED */
1.1 cgd 1518: }
CVSweb <webmaster@jp.NetBSD.org>