Annotation of src/sys/netinet/ip_input.c, Revision 1.78
1.78 ! mycroft 1: /* $NetBSD: ip_input.c,v 1.77 1999/01/11 22:35:06 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_id);
353: NTOHS(ip->ip_off);
1.35 mycroft 354: len = ip->ip_len;
1.1 cgd 355:
356: /*
357: * Check that the amount of data in the buffers
358: * is as at least much as the IP header would have us expect.
359: * Trim mbufs if longer than we expect.
360: * Drop packet if shorter than we expect.
361: */
1.35 mycroft 362: if (m->m_pkthdr.len < len) {
1.1 cgd 363: ipstat.ips_tooshort++;
364: goto bad;
365: }
1.35 mycroft 366: if (m->m_pkthdr.len > len) {
1.1 cgd 367: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 368: m->m_len = len;
369: m->m_pkthdr.len = len;
1.1 cgd 370: } else
1.35 mycroft 371: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 372: }
373:
1.64 thorpej 374: /*
375: * Assume that we can create a fast-forward IP flow entry
376: * based on this packet.
377: */
378: m->m_flags |= M_CANFASTFWD;
379:
1.36 mrg 380: #ifdef PFIL_HOOKS
1.33 mrg 381: /*
1.64 thorpej 382: * Run through list of hooks for input packets. If there are any
383: * filters which require that additional packets in the flow are
384: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
385: * Note that filters must _never_ set this flag, as another filter
386: * in the list may have previously cleared it.
1.33 mrg 387: */
388: m0 = m;
1.59 mrg 389: for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next)
1.33 mrg 390: if (pfh->pfil_func) {
1.43 mrg 391: rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0);
392: if (rv)
1.40 veego 393: goto next;
1.68 sommerfe 394: m = m0;
395: if (m == NULL)
396: goto next;
397: ip = mtod(m, struct ip *);
1.33 mrg 398: }
1.36 mrg 399: #endif /* PFIL_HOOKS */
1.33 mrg 400:
1.1 cgd 401: /*
402: * Process options and, if not destined for us,
403: * ship it on. ip_dooptions returns 1 when an
404: * error was detected (causing an icmp message
405: * to be sent and the original packet to be freed).
406: */
407: ip_nhops = 0; /* for source routed packets */
408: if (hlen > sizeof (struct ip) && ip_dooptions(m))
409: goto next;
410:
411: /*
412: * Check our list of addresses, to see if the packet is for us.
413: */
1.57 tls 414: INADDR_TO_IA(ip->ip_dst, ia);
415: if (ia != NULL) goto ours;
416: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
417: for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
418: ifa != NULL; ifa = ifa->ifa_list.tqe_next) {
419: if (ifa->ifa_addr->sa_family != AF_INET) continue;
420: ia = ifatoia(ifa);
1.35 mycroft 421: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
422: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 423: /*
424: * Look for all-0's host part (old broadcast addr),
425: * either for subnet or net.
426: */
427: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 428: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 429: goto ours;
1.57 tls 430: /*
431: * An interface with IP address zero accepts
432: * all packets that arrive on that interface.
433: */
434: if (in_nullhost(ia->ia_addr.sin_addr))
435: goto ours;
1.1 cgd 436: }
437: }
1.18 mycroft 438: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 439: struct in_multi *inm;
440: #ifdef MROUTING
441: extern struct socket *ip_mrouter;
1.10 brezak 442:
443: if (m->m_flags & M_EXT) {
444: if ((m = m_pullup(m, hlen)) == 0) {
445: ipstat.ips_toosmall++;
446: goto next;
447: }
448: ip = mtod(m, struct ip *);
449: }
1.4 hpeyerl 450:
451: if (ip_mrouter) {
452: /*
453: * If we are acting as a multicast router, all
454: * incoming multicast packets are passed to the
455: * kernel-level multicast forwarding function.
456: * The packet is returned (relatively) intact; if
457: * ip_mforward() returns a non-zero value, the packet
458: * must be discarded, else it may be accepted below.
459: *
460: * (The IP ident field is put in the same byte order
461: * as expected when ip_mforward() is called from
462: * ip_output().)
463: */
464: ip->ip_id = htons(ip->ip_id);
1.13 mycroft 465: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
466: ipstat.ips_cantforward++;
1.4 hpeyerl 467: m_freem(m);
468: goto next;
469: }
470: ip->ip_id = ntohs(ip->ip_id);
471:
472: /*
473: * The process-level routing demon needs to receive
474: * all multicast IGMP packets, whether or not this
475: * host belongs to their destination groups.
476: */
477: if (ip->ip_p == IPPROTO_IGMP)
478: goto ours;
1.13 mycroft 479: ipstat.ips_forward++;
1.4 hpeyerl 480: }
481: #endif
482: /*
483: * See if we belong to the destination multicast group on the
484: * arrival interface.
485: */
486: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
487: if (inm == NULL) {
1.13 mycroft 488: ipstat.ips_cantforward++;
1.4 hpeyerl 489: m_freem(m);
490: goto next;
491: }
492: goto ours;
493: }
1.19 mycroft 494: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 495: in_nullhost(ip->ip_dst))
1.1 cgd 496: goto ours;
497:
498: /*
499: * Not for us; forward if possible and desirable.
500: */
501: if (ipforwarding == 0) {
502: ipstat.ips_cantforward++;
503: m_freem(m);
504: } else
505: ip_forward(m, 0);
506: goto next;
507:
508: ours:
509: /*
510: * If offset or IP_MF are set, must reassemble.
511: * Otherwise, nothing need be done.
512: * (We could look in the reassembly queue to see
513: * if the packet was previously fragmented,
514: * but it's not worth the time; just let them time out.)
515: */
1.37 perry 516: if (ip->ip_off & ~(IP_DF|IP_RF)) {
1.1 cgd 517: /*
518: * Look for queue of fragments
519: * of this datagram.
520: */
1.75 thorpej 521: IPQ_LOCK();
1.25 cgd 522: for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next)
1.1 cgd 523: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 524: in_hosteq(ip->ip_src, fp->ipq_src) &&
525: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 526: ip->ip_p == fp->ipq_p)
527: goto found;
528: fp = 0;
529: found:
530:
531: /*
532: * Adjust ip_len to not reflect header,
1.25 cgd 533: * set ipqe_mff if more fragments are expected,
1.1 cgd 534: * convert offset of this to bytes.
535: */
536: ip->ip_len -= hlen;
1.25 cgd 537: mff = (ip->ip_off & IP_MF) != 0;
538: if (mff) {
1.16 cgd 539: /*
540: * Make sure that fragments have a data length
541: * that's a non-zero multiple of 8 bytes.
542: */
1.17 cgd 543: if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
1.16 cgd 544: ipstat.ips_badfrags++;
1.75 thorpej 545: IPQ_UNLOCK();
1.16 cgd 546: goto bad;
547: }
548: }
1.1 cgd 549: ip->ip_off <<= 3;
550:
551: /*
552: * If datagram marked as having more fragments
553: * or if this is not the first fragment,
554: * attempt reassembly; if it succeeds, proceed.
555: */
1.25 cgd 556: if (mff || ip->ip_off) {
1.1 cgd 557: ipstat.ips_fragments++;
1.72 thorpej 558: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.25 cgd 559: if (ipqe == NULL) {
560: ipstat.ips_rcvmemdrop++;
1.75 thorpej 561: IPQ_UNLOCK();
1.25 cgd 562: goto bad;
563: }
564: ipqe->ipqe_mff = mff;
1.50 thorpej 565: ipqe->ipqe_m = m;
1.25 cgd 566: ipqe->ipqe_ip = ip;
1.50 thorpej 567: m = ip_reass(ipqe, fp);
1.75 thorpej 568: if (m == 0) {
569: IPQ_UNLOCK();
1.1 cgd 570: goto next;
1.75 thorpej 571: }
1.13 mycroft 572: ipstat.ips_reassembled++;
1.50 thorpej 573: ip = mtod(m, struct ip *);
1.74 thorpej 574: hlen = ip->ip_hl << 2;
1.1 cgd 575: } else
576: if (fp)
577: ip_freef(fp);
1.75 thorpej 578: IPQ_UNLOCK();
1.1 cgd 579: } else
580: ip->ip_len -= hlen;
581:
582: /*
583: * Switch out to protocol's input routine.
584: */
585: ipstat.ips_delivered++;
586: (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
587: goto next;
588: bad:
589: m_freem(m);
590: goto next;
591: }
592:
593: /*
594: * Take incoming datagram fragment and try to
595: * reassemble it into whole datagram. If a chain for
596: * reassembly of this datagram already exists, then it
597: * is given as fp; otherwise have to make a chain.
598: */
1.50 thorpej 599: struct mbuf *
1.25 cgd 600: ip_reass(ipqe, fp)
601: register struct ipqent *ipqe;
1.1 cgd 602: register struct ipq *fp;
603: {
1.50 thorpej 604: register struct mbuf *m = ipqe->ipqe_m;
1.25 cgd 605: register struct ipqent *nq, *p, *q;
606: struct ip *ip;
1.1 cgd 607: struct mbuf *t;
1.25 cgd 608: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 609: int i, next;
610:
1.75 thorpej 611: IPQ_LOCK_CHECK();
612:
1.1 cgd 613: /*
614: * Presence of header sizes in mbufs
615: * would confuse code below.
616: */
617: m->m_data += hlen;
618: m->m_len -= hlen;
619:
620: /*
621: * If first fragment to arrive, create a reassembly queue.
622: */
623: if (fp == 0) {
1.50 thorpej 624: MALLOC(fp, struct ipq *, sizeof (struct ipq),
625: M_FTABLE, M_NOWAIT);
626: if (fp == NULL)
1.1 cgd 627: goto dropfrag;
1.25 cgd 628: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 629: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 630: fp->ipq_p = ipqe->ipqe_ip->ip_p;
631: fp->ipq_id = ipqe->ipqe_ip->ip_id;
632: LIST_INIT(&fp->ipq_fragq);
633: fp->ipq_src = ipqe->ipqe_ip->ip_src;
634: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
635: p = NULL;
1.1 cgd 636: goto insert;
637: }
638:
639: /*
640: * Find a segment which begins after this one does.
641: */
1.25 cgd 642: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
643: p = q, q = q->ipqe_q.le_next)
644: if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off)
1.1 cgd 645: break;
646:
647: /*
648: * If there is a preceding segment, it may provide some of
649: * our data already. If so, drop the data from the incoming
650: * segment. If it provides all of our data, drop us.
651: */
1.25 cgd 652: if (p != NULL) {
653: i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len -
654: ipqe->ipqe_ip->ip_off;
1.1 cgd 655: if (i > 0) {
1.25 cgd 656: if (i >= ipqe->ipqe_ip->ip_len)
1.1 cgd 657: goto dropfrag;
1.50 thorpej 658: m_adj(ipqe->ipqe_m, i);
1.25 cgd 659: ipqe->ipqe_ip->ip_off += i;
660: ipqe->ipqe_ip->ip_len -= i;
1.1 cgd 661: }
662: }
663:
664: /*
665: * While we overlap succeeding segments trim them or,
666: * if they are completely covered, dequeue them.
667: */
1.25 cgd 668: for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len >
669: q->ipqe_ip->ip_off; q = nq) {
670: i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) -
671: q->ipqe_ip->ip_off;
672: if (i < q->ipqe_ip->ip_len) {
673: q->ipqe_ip->ip_len -= i;
674: q->ipqe_ip->ip_off += i;
1.50 thorpej 675: m_adj(q->ipqe_m, i);
1.1 cgd 676: break;
677: }
1.25 cgd 678: nq = q->ipqe_q.le_next;
1.50 thorpej 679: m_freem(q->ipqe_m);
1.25 cgd 680: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 681: pool_put(&ipqent_pool, q);
1.1 cgd 682: }
683:
684: insert:
685: /*
686: * Stick new segment in its place;
687: * check for complete reassembly.
688: */
1.25 cgd 689: if (p == NULL) {
690: LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
691: } else {
692: LIST_INSERT_AFTER(p, ipqe, ipqe_q);
693: }
1.1 cgd 694: next = 0;
1.25 cgd 695: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
696: p = q, q = q->ipqe_q.le_next) {
697: if (q->ipqe_ip->ip_off != next)
1.1 cgd 698: return (0);
1.25 cgd 699: next += q->ipqe_ip->ip_len;
1.1 cgd 700: }
1.25 cgd 701: if (p->ipqe_mff)
1.1 cgd 702: return (0);
703:
704: /*
1.41 thorpej 705: * Reassembly is complete. Check for a bogus message size and
706: * concatenate fragments.
1.1 cgd 707: */
1.25 cgd 708: q = fp->ipq_fragq.lh_first;
709: ip = q->ipqe_ip;
1.41 thorpej 710: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
711: ipstat.ips_toolong++;
712: ip_freef(fp);
713: return (0);
714: }
1.50 thorpej 715: m = q->ipqe_m;
1.1 cgd 716: t = m->m_next;
717: m->m_next = 0;
718: m_cat(m, t);
1.25 cgd 719: nq = q->ipqe_q.le_next;
1.72 thorpej 720: pool_put(&ipqent_pool, q);
1.25 cgd 721: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 722: t = q->ipqe_m;
1.25 cgd 723: nq = q->ipqe_q.le_next;
1.72 thorpej 724: pool_put(&ipqent_pool, q);
1.1 cgd 725: m_cat(m, t);
726: }
727:
728: /*
729: * Create header for new ip packet by
730: * modifying header of first packet;
731: * dequeue and discard fragment reassembly header.
732: * Make header visible.
733: */
734: ip->ip_len = next;
1.25 cgd 735: ip->ip_src = fp->ipq_src;
736: ip->ip_dst = fp->ipq_dst;
737: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 738: FREE(fp, M_FTABLE);
1.1 cgd 739: m->m_len += (ip->ip_hl << 2);
740: m->m_data -= (ip->ip_hl << 2);
741: /* some debugging cruft by sklower, below, will go away soon */
742: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
743: register int plen = 0;
1.50 thorpej 744: for (t = m; t; t = t->m_next)
745: plen += t->m_len;
746: m->m_pkthdr.len = plen;
1.1 cgd 747: }
1.50 thorpej 748: return (m);
1.1 cgd 749:
750: dropfrag:
751: ipstat.ips_fragdropped++;
752: m_freem(m);
1.72 thorpej 753: pool_put(&ipqent_pool, ipqe);
1.1 cgd 754: return (0);
755: }
756:
757: /*
758: * Free a fragment reassembly header and all
759: * associated datagrams.
760: */
1.8 mycroft 761: void
1.1 cgd 762: ip_freef(fp)
763: struct ipq *fp;
764: {
1.25 cgd 765: register struct ipqent *q, *p;
1.1 cgd 766:
1.75 thorpej 767: IPQ_LOCK_CHECK();
768:
1.25 cgd 769: for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) {
770: p = q->ipqe_q.le_next;
1.50 thorpej 771: m_freem(q->ipqe_m);
1.25 cgd 772: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 773: pool_put(&ipqent_pool, q);
1.1 cgd 774: }
1.25 cgd 775: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 776: FREE(fp, M_FTABLE);
1.1 cgd 777: }
778:
779: /*
780: * IP timer processing;
781: * if a timer expires on a reassembly
782: * queue, discard it.
783: */
1.8 mycroft 784: void
1.1 cgd 785: ip_slowtimo()
786: {
1.25 cgd 787: register struct ipq *fp, *nfp;
1.24 mycroft 788: int s = splsoftnet();
1.1 cgd 789:
1.75 thorpej 790: IPQ_LOCK();
1.25 cgd 791: for (fp = ipq.lh_first; fp != NULL; fp = nfp) {
792: nfp = fp->ipq_q.le_next;
793: if (--fp->ipq_ttl == 0) {
1.1 cgd 794: ipstat.ips_fragtimeout++;
1.25 cgd 795: ip_freef(fp);
1.1 cgd 796: }
797: }
1.75 thorpej 798: IPQ_UNLOCK();
1.63 matt 799: #ifdef GATEWAY
800: ipflow_slowtimo();
801: #endif
1.1 cgd 802: splx(s);
803: }
804:
805: /*
806: * Drain off all datagram fragments.
807: */
1.8 mycroft 808: void
1.1 cgd 809: ip_drain()
810: {
811:
1.75 thorpej 812: /*
813: * We may be called from a device's interrupt context. If
814: * the ipq is already busy, just bail out now.
815: */
816: if (ipq_lock_try() == 0)
817: return;
818:
1.25 cgd 819: while (ipq.lh_first != NULL) {
1.1 cgd 820: ipstat.ips_fragdropped++;
1.25 cgd 821: ip_freef(ipq.lh_first);
1.1 cgd 822: }
1.75 thorpej 823:
824: IPQ_UNLOCK();
1.1 cgd 825: }
826:
827: /*
828: * Do option processing on a datagram,
829: * possibly discarding it if bad options are encountered,
830: * or forwarding it if source-routed.
831: * Returns 1 if packet has been forwarded/freed,
832: * 0 if the packet should be processed further.
833: */
1.8 mycroft 834: int
1.1 cgd 835: ip_dooptions(m)
836: struct mbuf *m;
837: {
838: register struct ip *ip = mtod(m, struct ip *);
839: register u_char *cp;
840: register struct ip_timestamp *ipt;
841: register struct in_ifaddr *ia;
842: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.13 mycroft 843: struct in_addr *sin, dst;
1.1 cgd 844: n_time ntime;
845:
1.13 mycroft 846: dst = ip->ip_dst;
1.1 cgd 847: cp = (u_char *)(ip + 1);
848: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
849: for (; cnt > 0; cnt -= optlen, cp += optlen) {
850: opt = cp[IPOPT_OPTVAL];
851: if (opt == IPOPT_EOL)
852: break;
853: if (opt == IPOPT_NOP)
854: optlen = 1;
855: else {
856: optlen = cp[IPOPT_OLEN];
857: if (optlen <= 0 || optlen > cnt) {
858: code = &cp[IPOPT_OLEN] - (u_char *)ip;
859: goto bad;
860: }
861: }
862: switch (opt) {
863:
864: default:
865: break;
866:
867: /*
868: * Source routing with record.
869: * Find interface with current destination address.
870: * If none on this machine then drop if strictly routed,
871: * or do nothing if loosely routed.
872: * Record interface address and bring up next address
873: * component. If strictly routed make sure next
874: * address is on directly accessible net.
875: */
876: case IPOPT_LSRR:
877: case IPOPT_SSRR:
1.47 cjs 878: if (ip_allowsrcrt == 0) {
879: type = ICMP_UNREACH;
880: code = ICMP_UNREACH_NET_PROHIB;
881: goto bad;
882: }
1.1 cgd 883: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
884: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
885: goto bad;
886: }
887: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 888: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 889: if (ia == 0) {
890: if (opt == IPOPT_SSRR) {
891: type = ICMP_UNREACH;
892: code = ICMP_UNREACH_SRCFAIL;
893: goto bad;
894: }
895: /*
896: * Loose routing, and not at next destination
897: * yet; nothing to do except forward.
898: */
899: break;
900: }
901: off--; /* 0 origin */
902: if (off > optlen - sizeof(struct in_addr)) {
903: /*
904: * End of source route. Should be for us.
905: */
906: save_rte(cp, ip->ip_src);
907: break;
908: }
909: /*
910: * locate outgoing interface
911: */
912: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
913: sizeof(ipaddr.sin_addr));
914: if (opt == IPOPT_SSRR) {
915: #define INA struct in_ifaddr *
916: #define SA struct sockaddr *
1.29 mrg 917: ia = (INA)ifa_ifwithladdr((SA)&ipaddr);
1.1 cgd 918: } else
919: ia = ip_rtaddr(ipaddr.sin_addr);
920: if (ia == 0) {
921: type = ICMP_UNREACH;
922: code = ICMP_UNREACH_SRCFAIL;
923: goto bad;
924: }
925: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 926: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 927: (caddr_t)(cp + off), sizeof(struct in_addr));
928: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 929: /*
930: * Let ip_intr's mcast routing check handle mcast pkts
931: */
1.18 mycroft 932: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 933: break;
934:
935: case IPOPT_RR:
936: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
937: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
938: goto bad;
939: }
940: /*
941: * If no space remains, ignore.
942: */
943: off--; /* 0 origin */
944: if (off > optlen - sizeof(struct in_addr))
945: break;
946: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
947: sizeof(ipaddr.sin_addr));
948: /*
949: * locate outgoing interface; if we're the destination,
950: * use the incoming interface (should be same).
951: */
952: if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
953: (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
954: type = ICMP_UNREACH;
955: code = ICMP_UNREACH_HOST;
956: goto bad;
957: }
1.20 mycroft 958: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 959: (caddr_t)(cp + off), sizeof(struct in_addr));
960: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
961: break;
962:
963: case IPOPT_TS:
964: code = cp - (u_char *)ip;
965: ipt = (struct ip_timestamp *)cp;
966: if (ipt->ipt_len < 5)
967: goto bad;
1.15 cgd 968: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.1 cgd 969: if (++ipt->ipt_oflw == 0)
970: goto bad;
971: break;
972: }
973: sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
974: switch (ipt->ipt_flg) {
975:
976: case IPOPT_TS_TSONLY:
977: break;
978:
979: case IPOPT_TS_TSANDADDR:
1.66 thorpej 980: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 981: sizeof(struct in_addr) > ipt->ipt_len)
982: goto bad;
1.13 mycroft 983: ipaddr.sin_addr = dst;
984: ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
985: m->m_pkthdr.rcvif);
986: if (ia == 0)
987: continue;
1.20 mycroft 988: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 989: (caddr_t)sin, sizeof(struct in_addr));
990: ipt->ipt_ptr += sizeof(struct in_addr);
991: break;
992:
993: case IPOPT_TS_PRESPEC:
1.66 thorpej 994: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 995: sizeof(struct in_addr) > ipt->ipt_len)
996: goto bad;
997: bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
998: sizeof(struct in_addr));
999: if (ifa_ifwithaddr((SA)&ipaddr) == 0)
1000: continue;
1001: ipt->ipt_ptr += sizeof(struct in_addr);
1002: break;
1003:
1004: default:
1005: goto bad;
1006: }
1007: ntime = iptime();
1008: bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
1009: sizeof(n_time));
1010: ipt->ipt_ptr += sizeof(n_time);
1011: }
1012: }
1013: if (forward) {
1.26 thorpej 1014: if (ip_forwsrcrt == 0) {
1015: type = ICMP_UNREACH;
1016: code = ICMP_UNREACH_SRCFAIL;
1017: goto bad;
1018: }
1.1 cgd 1019: ip_forward(m, 1);
1020: return (1);
1.13 mycroft 1021: }
1022: return (0);
1.1 cgd 1023: bad:
1.13 mycroft 1024: ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */
1025: icmp_error(m, type, code, 0, 0);
1026: ipstat.ips_badoptions++;
1.1 cgd 1027: return (1);
1028: }
1029:
1030: /*
1031: * Given address of next destination (final or next hop),
1032: * return internet address info of interface to be used to get there.
1033: */
1034: struct in_ifaddr *
1035: ip_rtaddr(dst)
1036: struct in_addr dst;
1037: {
1038: register struct sockaddr_in *sin;
1039:
1.19 mycroft 1040: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1041:
1.35 mycroft 1042: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 1043: if (ipforward_rt.ro_rt) {
1044: RTFREE(ipforward_rt.ro_rt);
1045: ipforward_rt.ro_rt = 0;
1046: }
1047: sin->sin_family = AF_INET;
1048: sin->sin_len = sizeof(*sin);
1049: sin->sin_addr = dst;
1050:
1051: rtalloc(&ipforward_rt);
1052: }
1053: if (ipforward_rt.ro_rt == 0)
1054: return ((struct in_ifaddr *)0);
1.19 mycroft 1055: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 1056: }
1057:
1058: /*
1059: * Save incoming source route for use in replies,
1060: * to be picked up later by ip_srcroute if the receiver is interested.
1061: */
1.13 mycroft 1062: void
1.1 cgd 1063: save_rte(option, dst)
1064: u_char *option;
1065: struct in_addr dst;
1066: {
1067: unsigned olen;
1068:
1069: olen = option[IPOPT_OLEN];
1070: #ifdef DIAGNOSTIC
1071: if (ipprintfs)
1.39 christos 1072: printf("save_rte: olen %d\n", olen);
1.1 cgd 1073: #endif
1074: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1075: return;
1076: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
1077: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1078: ip_srcrt.dst = dst;
1079: }
1080:
1081: /*
1082: * Retrieve incoming source route for use in replies,
1083: * in the same form used by setsockopt.
1084: * The first hop is placed before the options, will be removed later.
1085: */
1086: struct mbuf *
1087: ip_srcroute()
1088: {
1089: register struct in_addr *p, *q;
1090: register struct mbuf *m;
1091:
1092: if (ip_nhops == 0)
1093: return ((struct mbuf *)0);
1094: m = m_get(M_DONTWAIT, MT_SOOPTS);
1095: if (m == 0)
1096: return ((struct mbuf *)0);
1097:
1.13 mycroft 1098: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1099:
1100: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1101: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1102: OPTSIZ;
1103: #ifdef DIAGNOSTIC
1104: if (ipprintfs)
1.39 christos 1105: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1106: #endif
1107:
1108: /*
1109: * First save first hop for return route
1110: */
1111: p = &ip_srcrt.route[ip_nhops - 1];
1112: *(mtod(m, struct in_addr *)) = *p--;
1113: #ifdef DIAGNOSTIC
1114: if (ipprintfs)
1.39 christos 1115: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1116: #endif
1117:
1118: /*
1119: * Copy option fields and padding (nop) to mbuf.
1120: */
1121: ip_srcrt.nop = IPOPT_NOP;
1122: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1123: bcopy((caddr_t)&ip_srcrt.nop,
1124: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1125: q = (struct in_addr *)(mtod(m, caddr_t) +
1126: sizeof(struct in_addr) + OPTSIZ);
1127: #undef OPTSIZ
1128: /*
1129: * Record return path as an IP source route,
1130: * reversing the path (pointers are now aligned).
1131: */
1132: while (p >= ip_srcrt.route) {
1133: #ifdef DIAGNOSTIC
1134: if (ipprintfs)
1.39 christos 1135: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1136: #endif
1137: *q++ = *p--;
1138: }
1139: /*
1140: * Last hop goes to final destination.
1141: */
1142: *q = ip_srcrt.dst;
1143: #ifdef DIAGNOSTIC
1144: if (ipprintfs)
1.39 christos 1145: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1146: #endif
1147: return (m);
1148: }
1149:
1150: /*
1151: * Strip out IP options, at higher
1152: * level protocol in the kernel.
1153: * Second argument is buffer to which options
1154: * will be moved, and return value is their length.
1155: * XXX should be deleted; last arg currently ignored.
1156: */
1.8 mycroft 1157: void
1.1 cgd 1158: ip_stripoptions(m, mopt)
1159: register struct mbuf *m;
1160: struct mbuf *mopt;
1161: {
1162: register int i;
1163: struct ip *ip = mtod(m, struct ip *);
1164: register caddr_t opts;
1165: int olen;
1166:
1167: olen = (ip->ip_hl<<2) - sizeof (struct ip);
1168: opts = (caddr_t)(ip + 1);
1169: i = m->m_len - (sizeof (struct ip) + olen);
1170: bcopy(opts + olen, opts, (unsigned)i);
1171: m->m_len -= olen;
1172: if (m->m_flags & M_PKTHDR)
1173: m->m_pkthdr.len -= olen;
1174: ip->ip_hl = sizeof(struct ip) >> 2;
1175: }
1176:
1.23 mycroft 1177: int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1178: 0, 0, 0, 0,
1179: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1180: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1181: EMSGSIZE, EHOSTUNREACH, 0, 0,
1182: 0, 0, 0, 0,
1183: ENOPROTOOPT
1184: };
1185:
1186: /*
1187: * Forward a packet. If some error occurs return the sender
1188: * an icmp packet. Note we can't always generate a meaningful
1189: * icmp message because icmp doesn't have a large enough repertoire
1190: * of codes and types.
1191: *
1192: * If not forwarding, just drop the packet. This could be confusing
1193: * if ipforwarding was zero but some routing protocol was advancing
1194: * us as a gateway to somewhere. However, we must let the routing
1195: * protocol deal with that.
1196: *
1197: * The srcrt parameter indicates whether the packet is being forwarded
1198: * via a source route.
1199: */
1.13 mycroft 1200: void
1.1 cgd 1201: ip_forward(m, srcrt)
1202: struct mbuf *m;
1203: int srcrt;
1204: {
1205: register struct ip *ip = mtod(m, struct ip *);
1206: register struct sockaddr_in *sin;
1207: register struct rtentry *rt;
1.28 christos 1208: int error, type = 0, code = 0;
1.1 cgd 1209: struct mbuf *mcopy;
1.13 mycroft 1210: n_long dest;
1211: struct ifnet *destifp;
1.1 cgd 1212:
1.13 mycroft 1213: dest = 0;
1.1 cgd 1214: #ifdef DIAGNOSTIC
1215: if (ipprintfs)
1.70 thorpej 1216: printf("forward: src %2.2x dst %2.2x ttl %x\n",
1217: ntohl(ip->ip_src.s_addr),
1218: ntohl(ip->ip_dst.s_addr), ip->ip_ttl);
1.1 cgd 1219: #endif
1220: if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1221: ipstat.ips_cantforward++;
1222: m_freem(m);
1223: return;
1224: }
1225: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1226: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1227: return;
1228: }
1229: ip->ip_ttl -= IPTTLDEC;
1230:
1.19 mycroft 1231: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1232: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1233: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1234: if (ipforward_rt.ro_rt) {
1235: RTFREE(ipforward_rt.ro_rt);
1236: ipforward_rt.ro_rt = 0;
1237: }
1238: sin->sin_family = AF_INET;
1.35 mycroft 1239: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1240: sin->sin_addr = ip->ip_dst;
1241:
1242: rtalloc(&ipforward_rt);
1243: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1244: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1245: return;
1246: }
1247: rt = ipforward_rt.ro_rt;
1248: }
1249:
1250: /*
1.34 mycroft 1251: * Save at most 68 bytes of the packet in case
1.1 cgd 1252: * we need to generate an ICMP message to the src.
1253: */
1.34 mycroft 1254: mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68));
1.1 cgd 1255:
1256: /*
1257: * If forwarding packet using same interface that it came in on,
1258: * perhaps should send a redirect to sender to shortcut a hop.
1259: * Only send redirect if source is sending directly to us,
1260: * and if packet was not source routed (or has any options).
1261: * Also, don't send redirect if forwarding using a default route
1262: * or a route modified by a redirect.
1263: */
1264: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1265: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1266: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1267: ipsendredirects && !srcrt) {
1.19 mycroft 1268: if (rt->rt_ifa &&
1269: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1270: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1271: if (rt->rt_flags & RTF_GATEWAY)
1272: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1273: else
1274: dest = ip->ip_dst.s_addr;
1275: /*
1276: * Router requirements says to only send host
1277: * redirects.
1278: */
1279: type = ICMP_REDIRECT;
1280: code = ICMP_REDIRECT_HOST;
1.1 cgd 1281: #ifdef DIAGNOSTIC
1.77 thorpej 1282: if (ipprintfs)
1283: printf("redirect (%d) to %x\n", code,
1284: (u_int32_t)dest);
1.1 cgd 1285: #endif
1286: }
1287: }
1288:
1.27 thorpej 1289: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1290: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1.1 cgd 1291: if (error)
1292: ipstat.ips_cantforward++;
1293: else {
1294: ipstat.ips_forward++;
1295: if (type)
1296: ipstat.ips_redirectsent++;
1297: else {
1.63 matt 1298: if (mcopy) {
1299: #ifdef GATEWAY
1.64 thorpej 1300: if (mcopy->m_flags & M_CANFASTFWD)
1301: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1302: #endif
1.1 cgd 1303: m_freem(mcopy);
1.63 matt 1304: }
1.1 cgd 1305: return;
1306: }
1307: }
1308: if (mcopy == NULL)
1309: return;
1.13 mycroft 1310: destifp = NULL;
1311:
1.1 cgd 1312: switch (error) {
1313:
1314: case 0: /* forwarded, but need redirect */
1315: /* type, code set above */
1316: break;
1317:
1318: case ENETUNREACH: /* shouldn't happen, checked above */
1319: case EHOSTUNREACH:
1320: case ENETDOWN:
1321: case EHOSTDOWN:
1322: default:
1323: type = ICMP_UNREACH;
1324: code = ICMP_UNREACH_HOST;
1325: break;
1326:
1327: case EMSGSIZE:
1328: type = ICMP_UNREACH;
1329: code = ICMP_UNREACH_NEEDFRAG;
1.13 mycroft 1330: if (ipforward_rt.ro_rt)
1331: destifp = ipforward_rt.ro_rt->rt_ifp;
1.1 cgd 1332: ipstat.ips_cantfrag++;
1333: break;
1334:
1335: case ENOBUFS:
1336: type = ICMP_SOURCEQUENCH;
1337: code = 0;
1338: break;
1339: }
1.13 mycroft 1340: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 1341: }
1342:
1343: void
1344: ip_savecontrol(inp, mp, ip, m)
1345: register struct inpcb *inp;
1346: register struct mbuf **mp;
1347: register struct ip *ip;
1348: register struct mbuf *m;
1349: {
1350:
1351: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1352: struct timeval tv;
1353:
1354: microtime(&tv);
1355: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1356: SCM_TIMESTAMP, SOL_SOCKET);
1357: if (*mp)
1358: mp = &(*mp)->m_next;
1359: }
1360: if (inp->inp_flags & INP_RECVDSTADDR) {
1361: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1362: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1363: if (*mp)
1364: mp = &(*mp)->m_next;
1365: }
1366: #ifdef notyet
1367: /*
1368: * XXX
1369: * Moving these out of udp_input() made them even more broken
1370: * than they already were.
1371: * - fenner@parc.xerox.com
1372: */
1373: /* options were tossed already */
1374: if (inp->inp_flags & INP_RECVOPTS) {
1375: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1376: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1377: if (*mp)
1378: mp = &(*mp)->m_next;
1379: }
1380: /* ip_srcroute doesn't do what we want here, need to fix */
1381: if (inp->inp_flags & INP_RECVRETOPTS) {
1382: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1383: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1384: if (*mp)
1385: mp = &(*mp)->m_next;
1386: }
1387: #endif
1388: if (inp->inp_flags & INP_RECVIF) {
1389: struct sockaddr_dl sdl;
1390:
1391: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1392: sdl.sdl_family = AF_LINK;
1393: sdl.sdl_index = m->m_pkthdr.rcvif ?
1394: m->m_pkthdr.rcvif->if_index : 0;
1395: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1396: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1397: IP_RECVIF, IPPROTO_IP);
1398: if (*mp)
1399: mp = &(*mp)->m_next;
1400: }
1.13 mycroft 1401: }
1402:
1403: int
1404: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1405: int *name;
1406: u_int namelen;
1407: void *oldp;
1408: size_t *oldlenp;
1409: void *newp;
1410: size_t newlen;
1411: {
1.52 thorpej 1412: extern int subnetsarelocal;
1413:
1.54 lukem 1414: int error, old;
1415:
1.13 mycroft 1416: /* All sysctl names at this level are terminal. */
1417: if (namelen != 1)
1418: return (ENOTDIR);
1419:
1420: switch (name[0]) {
1421: case IPCTL_FORWARDING:
1422: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1423: case IPCTL_SENDREDIRECTS:
1424: return (sysctl_int(oldp, oldlenp, newp, newlen,
1425: &ipsendredirects));
1426: case IPCTL_DEFTTL:
1427: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1428: #ifdef notyet
1429: case IPCTL_DEFMTU:
1430: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1431: #endif
1.26 thorpej 1432: case IPCTL_FORWSRCRT:
1.47 cjs 1433: /* Don't allow this to change in a secure environment. */
1.26 thorpej 1434: if (securelevel > 0)
1.46 cjs 1435: return (sysctl_rdint(oldp, oldlenp, newp,
1436: ip_forwsrcrt));
1437: else
1438: return (sysctl_int(oldp, oldlenp, newp, newlen,
1439: &ip_forwsrcrt));
1.27 thorpej 1440: case IPCTL_DIRECTEDBCAST:
1441: return (sysctl_int(oldp, oldlenp, newp, newlen,
1442: &ip_directedbcast));
1.47 cjs 1443: case IPCTL_ALLOWSRCRT:
1444: return (sysctl_int(oldp, oldlenp, newp, newlen,
1445: &ip_allowsrcrt));
1.52 thorpej 1446: case IPCTL_SUBNETSARELOCAL:
1447: return (sysctl_int(oldp, oldlenp, newp, newlen,
1448: &subnetsarelocal));
1.53 kml 1449: case IPCTL_MTUDISC:
1.60 kml 1450: error = sysctl_int(oldp, oldlenp, newp, newlen,
1451: &ip_mtudisc);
1452: if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) {
1453: ip_mtudisc_timeout_q =
1454: rt_timer_queue_create(ip_mtudisc_timeout);
1455: } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) {
1456: rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE);
1457: ip_mtudisc_timeout_q = NULL;
1458: }
1459: return error;
1.54 lukem 1460: case IPCTL_ANONPORTMIN:
1461: old = anonportmin;
1462: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1463: if (anonportmin >= anonportmax || anonportmin > 65535
1464: #ifndef IPNOPRIVPORTS
1465: || anonportmin < IPPORT_RESERVED
1466: #endif
1467: ) {
1468: anonportmin = old;
1469: return (EINVAL);
1470: }
1471: return (error);
1472: case IPCTL_ANONPORTMAX:
1473: old = anonportmax;
1474: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1475: if (anonportmin >= anonportmax || anonportmax > 65535
1476: #ifndef IPNOPRIVPORTS
1477: || anonportmax < IPPORT_RESERVED
1478: #endif
1479: ) {
1480: anonportmax = old;
1481: return (EINVAL);
1482: }
1.60 kml 1483: return (error);
1484: case IPCTL_MTUDISCTIMEOUT:
1485: error = sysctl_int(oldp, oldlenp, newp, newlen,
1486: &ip_mtudisc_timeout);
1487: if (ip_mtudisc_timeout_q != NULL)
1488: rt_timer_queue_change(ip_mtudisc_timeout_q,
1489: ip_mtudisc_timeout);
1.54 lukem 1490: return (error);
1.65 matt 1491: #ifdef GATEWAY
1492: case IPCTL_MAXFLOWS:
1.67 thorpej 1493: {
1494: int s;
1495:
1.65 matt 1496: error = sysctl_int(oldp, oldlenp, newp, newlen,
1497: &ip_maxflows);
1.67 thorpej 1498: s = splsoftnet();
1.65 matt 1499: ipflow_reap(0);
1.67 thorpej 1500: splx(s);
1.65 matt 1501: return (error);
1.67 thorpej 1502: }
1.65 matt 1503: #endif
1.13 mycroft 1504: default:
1505: return (EOPNOTSUPP);
1506: }
1507: /* NOTREACHED */
1.1 cgd 1508: }
CVSweb <webmaster@jp.NetBSD.org>