Annotation of src/sys/netinet/ip_input.c, Revision 1.74.2.1
1.74.2.1! kenh 1: /* $NetBSD: ip_input.c,v 1.74 1998/11/13 03:24:22 thorpej Exp $ */
1.14 cgd 2:
1.1 cgd 3: /*
1.13 mycroft 4: * Copyright (c) 1982, 1986, 1988, 1993
5: * The Regents of the University of California. All rights reserved.
1.1 cgd 6: *
7: * Redistribution and use in source and binary forms, with or without
8: * modification, are permitted provided that the following conditions
9: * are met:
10: * 1. Redistributions of source code must retain the above copyright
11: * notice, this list of conditions and the following disclaimer.
12: * 2. Redistributions in binary form must reproduce the above copyright
13: * notice, this list of conditions and the following disclaimer in the
14: * documentation and/or other materials provided with the distribution.
15: * 3. All advertising materials mentioning features or use of this software
16: * must display the following acknowledgement:
17: * This product includes software developed by the University of
18: * California, Berkeley and its contributors.
19: * 4. Neither the name of the University nor the names of its contributors
20: * may be used to endorse or promote products derived from this software
21: * without specific prior written permission.
22: *
23: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33: * SUCH DAMAGE.
34: *
1.14 cgd 35: * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
1.71 tls 36: */
37:
38: /*-
39: * Copyright (c) 1998 The NetBSD Foundation, Inc.
40: * All rights reserved.
41: *
42: * This code is derived from software contributed to The NetBSD Foundation
43: * by Public Access Networks Corporation ("Panix"). It was developed under
44: * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon.
45: *
46: * Redistribution and use in source and binary forms, with or without
47: * modification, are permitted provided that the following conditions
48: * are met:
49: * 1. Redistributions of source code must retain the above copyright
50: * notice, this list of conditions and the following disclaimer.
51: * 2. Redistributions in binary form must reproduce the above copyright
52: * notice, this list of conditions and the following disclaimer in the
53: * documentation and/or other materials provided with the distribution.
54: * 3. All advertising materials mentioning features or use of this software
55: * must display the following acknowledgement:
56: * This product includes software developed by the NetBSD
57: * Foundation, Inc. and its contributors.
58: * 4. Neither the name of The NetBSD Foundation nor the names of its
59: * contributors may be used to endorse or promote products derived
60: * from this software without specific prior written permission.
61: *
62: * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
63: * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
64: * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
65: * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
66: * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
67: * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
68: * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
69: * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
70: * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
71: * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
72: * POSSIBILITY OF SUCH DAMAGE.
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;
167: struct ipqhead ipq;
1.1 cgd 168:
1.72 thorpej 169: struct pool ipqent_pool;
170:
1.1 cgd 171: /*
172: * We need to save the IP options in case a protocol wants to respond
173: * to an incoming packet over the same route if the packet got here
174: * using IP source routing. This allows connection establishment and
175: * maintenance when the remote end is on a network that is not known
176: * to us.
177: */
178: int ip_nhops = 0;
179: static struct ip_srcrt {
180: struct in_addr dst; /* final destination */
181: char nop; /* one NOP to align */
182: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
183: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
184: } ip_srcrt;
185:
1.13 mycroft 186: static void save_rte __P((u_char *, struct in_addr));
1.35 mycroft 187:
1.1 cgd 188: /*
189: * IP initialization: fill in IP protocol switch table.
190: * All protocols not implemented in kernel go to raw IP protocol handler.
191: */
1.8 mycroft 192: void
1.1 cgd 193: ip_init()
194: {
195: register struct protosw *pr;
196: register int i;
197:
1.72 thorpej 198: pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl",
199: 0, NULL, NULL, M_IPQ);
200:
1.1 cgd 201: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
202: if (pr == 0)
203: panic("ip_init");
204: for (i = 0; i < IPPROTO_MAX; i++)
205: ip_protox[i] = pr - inetsw;
206: for (pr = inetdomain.dom_protosw;
207: pr < inetdomain.dom_protoswNPROTOSW; pr++)
208: if (pr->pr_domain->dom_family == PF_INET &&
209: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
210: ip_protox[pr->pr_protocol] = pr - inetsw;
1.25 cgd 211: LIST_INIT(&ipq);
1.1 cgd 212: ip_id = time.tv_sec & 0xffff;
213: ipintrq.ifq_maxlen = ipqmaxlen;
1.22 mycroft 214: TAILQ_INIT(&in_ifaddr);
1.57 tls 215: in_ifaddrhashtbl =
216: hashinit(IN_IFADDR_HASH_SIZE, M_IFADDR, M_WAITOK, &in_ifaddrhash);
1.60 kml 217: if (ip_mtudisc != 0)
218: ip_mtudisc_timeout_q =
219: rt_timer_queue_create(ip_mtudisc_timeout);
1.73 thorpej 220: #ifdef GATEWAY
221: ipflow_init();
222: #endif
1.1 cgd 223: }
224:
225: struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
226: struct route ipforward_rt;
227:
228: /*
229: * Ip input routine. Checksum and byte swap header. If fragmented
230: * try to reassemble. Process options. Pass to next level.
231: */
1.8 mycroft 232: void
1.1 cgd 233: ipintr()
234: {
1.33 mrg 235: register struct ip *ip = NULL;
1.1 cgd 236: register struct mbuf *m;
237: register struct ipq *fp;
238: register struct in_ifaddr *ia;
1.57 tls 239: register struct ifaddr *ifa;
1.25 cgd 240: struct ipqent *ipqe;
1.35 mycroft 241: int hlen = 0, mff, len, s;
1.36 mrg 242: #ifdef PFIL_HOOKS
1.33 mrg 243: struct packet_filter_hook *pfh;
244: struct mbuf *m0;
1.43 mrg 245: int rv;
1.36 mrg 246: #endif /* PFIL_HOOKS */
1.1 cgd 247:
248: next:
249: /*
250: * Get next datagram off input queue and get IP header
251: * in first mbuf.
252: */
253: s = splimp();
254: IF_DEQUEUE(&ipintrq, m);
255: splx(s);
1.13 mycroft 256: if (m == 0)
1.1 cgd 257: return;
258: #ifdef DIAGNOSTIC
259: if ((m->m_flags & M_PKTHDR) == 0)
260: panic("ipintr no HDR");
261: #endif
262: /*
263: * If no IP addresses have been set yet but the interfaces
264: * are receiving, can't do anything with incoming packets yet.
265: */
1.22 mycroft 266: if (in_ifaddr.tqh_first == 0)
1.1 cgd 267: goto bad;
268: ipstat.ips_total++;
269: if (m->m_len < sizeof (struct ip) &&
270: (m = m_pullup(m, sizeof (struct ip))) == 0) {
271: ipstat.ips_toosmall++;
272: goto next;
273: }
274: ip = mtod(m, struct ip *);
1.13 mycroft 275: if (ip->ip_v != IPVERSION) {
276: ipstat.ips_badvers++;
277: goto bad;
278: }
1.1 cgd 279: hlen = ip->ip_hl << 2;
280: if (hlen < sizeof(struct ip)) { /* minimum header length */
281: ipstat.ips_badhlen++;
282: goto bad;
283: }
284: if (hlen > m->m_len) {
285: if ((m = m_pullup(m, hlen)) == 0) {
286: ipstat.ips_badhlen++;
287: goto next;
288: }
289: ip = mtod(m, struct ip *);
290: }
1.28 christos 291: if ((ip->ip_sum = in_cksum(m, hlen)) != 0) {
1.1 cgd 292: ipstat.ips_badsum++;
293: goto bad;
294: }
295:
296: /*
297: * Convert fields to host representation.
298: */
299: NTOHS(ip->ip_len);
300: NTOHS(ip->ip_id);
301: NTOHS(ip->ip_off);
1.35 mycroft 302: len = ip->ip_len;
1.1 cgd 303:
304: /*
305: * Check that the amount of data in the buffers
306: * is as at least much as the IP header would have us expect.
307: * Trim mbufs if longer than we expect.
308: * Drop packet if shorter than we expect.
309: */
1.35 mycroft 310: if (m->m_pkthdr.len < len) {
1.1 cgd 311: ipstat.ips_tooshort++;
312: goto bad;
313: }
1.35 mycroft 314: if (m->m_pkthdr.len > len) {
1.1 cgd 315: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 316: m->m_len = len;
317: m->m_pkthdr.len = len;
1.1 cgd 318: } else
1.35 mycroft 319: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 320: }
321:
1.64 thorpej 322: /*
323: * Assume that we can create a fast-forward IP flow entry
324: * based on this packet.
325: */
326: m->m_flags |= M_CANFASTFWD;
327:
1.36 mrg 328: #ifdef PFIL_HOOKS
1.33 mrg 329: /*
1.64 thorpej 330: * Run through list of hooks for input packets. If there are any
331: * filters which require that additional packets in the flow are
332: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
333: * Note that filters must _never_ set this flag, as another filter
334: * in the list may have previously cleared it.
1.33 mrg 335: */
336: m0 = m;
1.59 mrg 337: for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next)
1.33 mrg 338: if (pfh->pfil_func) {
1.43 mrg 339: rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0);
340: if (rv)
1.40 veego 341: goto next;
1.68 sommerfe 342: m = m0;
343: if (m == NULL)
344: goto next;
345: ip = mtod(m, struct ip *);
1.33 mrg 346: }
1.36 mrg 347: #endif /* PFIL_HOOKS */
1.33 mrg 348:
1.1 cgd 349: /*
350: * Process options and, if not destined for us,
351: * ship it on. ip_dooptions returns 1 when an
352: * error was detected (causing an icmp message
353: * to be sent and the original packet to be freed).
354: */
355: ip_nhops = 0; /* for source routed packets */
356: if (hlen > sizeof (struct ip) && ip_dooptions(m))
357: goto next;
358:
359: /*
360: * Check our list of addresses, to see if the packet is for us.
361: */
1.74.2.1! kenh 362: s = splimp();
1.57 tls 363: INADDR_TO_IA(ip->ip_dst, ia);
1.74.2.1! kenh 364: if (ia != NULL) {
! 365: ifa_delref(&ia->ia_ifa);
! 366: splx(s);
! 367: goto ours;
! 368: }
1.57 tls 369: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
370: for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
371: ifa != NULL; ifa = ifa->ifa_list.tqe_next) {
372: if (ifa->ifa_addr->sa_family != AF_INET) continue;
373: ia = ifatoia(ifa);
1.74.2.1! kenh 374: /*
! 375: * Don't ifa_addref ia as we don't use it after
! 376: * we splx(s) below
! 377: */
1.35 mycroft 378: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
379: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 380: /*
381: * Look for all-0's host part (old broadcast addr),
382: * either for subnet or net.
383: */
384: ip->ip_dst.s_addr == ia->ia_subnet ||
1.74.2.1! kenh 385: ip->ip_dst.s_addr == ia->ia_net) {
! 386: splx(s);
1.1 cgd 387: goto ours;
1.74.2.1! kenh 388: }
1.57 tls 389: /*
390: * An interface with IP address zero accepts
391: * all packets that arrive on that interface.
392: */
1.74.2.1! kenh 393: if (in_nullhost(ia->ia_addr.sin_addr)) {
! 394: splx(s);
1.57 tls 395: goto ours;
1.74.2.1! kenh 396: }
1.1 cgd 397: }
398: }
1.74.2.1! kenh 399: splx(s);
1.18 mycroft 400: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 401: struct in_multi *inm;
402: #ifdef MROUTING
403: extern struct socket *ip_mrouter;
1.10 brezak 404:
405: if (m->m_flags & M_EXT) {
406: if ((m = m_pullup(m, hlen)) == 0) {
407: ipstat.ips_toosmall++;
408: goto next;
409: }
410: ip = mtod(m, struct ip *);
411: }
1.4 hpeyerl 412:
413: if (ip_mrouter) {
414: /*
415: * If we are acting as a multicast router, all
416: * incoming multicast packets are passed to the
417: * kernel-level multicast forwarding function.
418: * The packet is returned (relatively) intact; if
419: * ip_mforward() returns a non-zero value, the packet
420: * must be discarded, else it may be accepted below.
421: *
422: * (The IP ident field is put in the same byte order
423: * as expected when ip_mforward() is called from
424: * ip_output().)
425: */
426: ip->ip_id = htons(ip->ip_id);
1.13 mycroft 427: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
428: ipstat.ips_cantforward++;
1.4 hpeyerl 429: m_freem(m);
430: goto next;
431: }
432: ip->ip_id = ntohs(ip->ip_id);
433:
434: /*
435: * The process-level routing demon needs to receive
436: * all multicast IGMP packets, whether or not this
437: * host belongs to their destination groups.
438: */
439: if (ip->ip_p == IPPROTO_IGMP)
440: goto ours;
1.13 mycroft 441: ipstat.ips_forward++;
1.4 hpeyerl 442: }
443: #endif
444: /*
445: * See if we belong to the destination multicast group on the
446: * arrival interface.
447: */
448: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
449: if (inm == NULL) {
1.13 mycroft 450: ipstat.ips_cantforward++;
1.4 hpeyerl 451: m_freem(m);
452: goto next;
453: }
454: goto ours;
455: }
1.19 mycroft 456: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 457: in_nullhost(ip->ip_dst))
1.1 cgd 458: goto ours;
459:
460: /*
461: * Not for us; forward if possible and desirable.
462: */
463: if (ipforwarding == 0) {
464: ipstat.ips_cantforward++;
465: m_freem(m);
466: } else
467: ip_forward(m, 0);
468: goto next;
469:
470: ours:
471: /*
472: * If offset or IP_MF are set, must reassemble.
473: * Otherwise, nothing need be done.
474: * (We could look in the reassembly queue to see
475: * if the packet was previously fragmented,
476: * but it's not worth the time; just let them time out.)
477: */
1.37 perry 478: if (ip->ip_off & ~(IP_DF|IP_RF)) {
1.1 cgd 479: /*
480: * Look for queue of fragments
481: * of this datagram.
482: */
1.25 cgd 483: for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next)
1.1 cgd 484: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 485: in_hosteq(ip->ip_src, fp->ipq_src) &&
486: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 487: ip->ip_p == fp->ipq_p)
488: goto found;
489: fp = 0;
490: found:
491:
492: /*
493: * Adjust ip_len to not reflect header,
1.25 cgd 494: * set ipqe_mff if more fragments are expected,
1.1 cgd 495: * convert offset of this to bytes.
496: */
497: ip->ip_len -= hlen;
1.25 cgd 498: mff = (ip->ip_off & IP_MF) != 0;
499: if (mff) {
1.16 cgd 500: /*
501: * Make sure that fragments have a data length
502: * that's a non-zero multiple of 8 bytes.
503: */
1.17 cgd 504: if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
1.16 cgd 505: ipstat.ips_badfrags++;
506: goto bad;
507: }
508: }
1.1 cgd 509: ip->ip_off <<= 3;
510:
511: /*
512: * If datagram marked as having more fragments
513: * or if this is not the first fragment,
514: * attempt reassembly; if it succeeds, proceed.
515: */
1.25 cgd 516: if (mff || ip->ip_off) {
1.1 cgd 517: ipstat.ips_fragments++;
1.72 thorpej 518: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.25 cgd 519: if (ipqe == NULL) {
520: ipstat.ips_rcvmemdrop++;
521: goto bad;
522: }
523: ipqe->ipqe_mff = mff;
1.50 thorpej 524: ipqe->ipqe_m = m;
1.25 cgd 525: ipqe->ipqe_ip = ip;
1.50 thorpej 526: m = ip_reass(ipqe, fp);
527: if (m == 0)
1.1 cgd 528: goto next;
1.13 mycroft 529: ipstat.ips_reassembled++;
1.50 thorpej 530: ip = mtod(m, struct ip *);
1.74 thorpej 531: hlen = ip->ip_hl << 2;
1.1 cgd 532: } else
533: if (fp)
534: ip_freef(fp);
535: } else
536: ip->ip_len -= hlen;
537:
538: /*
539: * Switch out to protocol's input routine.
540: */
541: ipstat.ips_delivered++;
542: (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
543: goto next;
544: bad:
545: m_freem(m);
546: goto next;
547: }
548:
549: /*
550: * Take incoming datagram fragment and try to
551: * reassemble it into whole datagram. If a chain for
552: * reassembly of this datagram already exists, then it
553: * is given as fp; otherwise have to make a chain.
554: */
1.50 thorpej 555: struct mbuf *
1.25 cgd 556: ip_reass(ipqe, fp)
557: register struct ipqent *ipqe;
1.1 cgd 558: register struct ipq *fp;
559: {
1.50 thorpej 560: register struct mbuf *m = ipqe->ipqe_m;
1.25 cgd 561: register struct ipqent *nq, *p, *q;
562: struct ip *ip;
1.1 cgd 563: struct mbuf *t;
1.25 cgd 564: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 565: int i, next;
566:
567: /*
568: * Presence of header sizes in mbufs
569: * would confuse code below.
570: */
571: m->m_data += hlen;
572: m->m_len -= hlen;
573:
574: /*
575: * If first fragment to arrive, create a reassembly queue.
576: */
577: if (fp == 0) {
1.50 thorpej 578: MALLOC(fp, struct ipq *, sizeof (struct ipq),
579: M_FTABLE, M_NOWAIT);
580: if (fp == NULL)
1.1 cgd 581: goto dropfrag;
1.25 cgd 582: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 583: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 584: fp->ipq_p = ipqe->ipqe_ip->ip_p;
585: fp->ipq_id = ipqe->ipqe_ip->ip_id;
586: LIST_INIT(&fp->ipq_fragq);
587: fp->ipq_src = ipqe->ipqe_ip->ip_src;
588: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
589: p = NULL;
1.1 cgd 590: goto insert;
591: }
592:
593: /*
594: * Find a segment which begins after this one does.
595: */
1.25 cgd 596: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
597: p = q, q = q->ipqe_q.le_next)
598: if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off)
1.1 cgd 599: break;
600:
601: /*
602: * If there is a preceding segment, it may provide some of
603: * our data already. If so, drop the data from the incoming
604: * segment. If it provides all of our data, drop us.
605: */
1.25 cgd 606: if (p != NULL) {
607: i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len -
608: ipqe->ipqe_ip->ip_off;
1.1 cgd 609: if (i > 0) {
1.25 cgd 610: if (i >= ipqe->ipqe_ip->ip_len)
1.1 cgd 611: goto dropfrag;
1.50 thorpej 612: m_adj(ipqe->ipqe_m, i);
1.25 cgd 613: ipqe->ipqe_ip->ip_off += i;
614: ipqe->ipqe_ip->ip_len -= i;
1.1 cgd 615: }
616: }
617:
618: /*
619: * While we overlap succeeding segments trim them or,
620: * if they are completely covered, dequeue them.
621: */
1.25 cgd 622: for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len >
623: q->ipqe_ip->ip_off; q = nq) {
624: i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) -
625: q->ipqe_ip->ip_off;
626: if (i < q->ipqe_ip->ip_len) {
627: q->ipqe_ip->ip_len -= i;
628: q->ipqe_ip->ip_off += i;
1.50 thorpej 629: m_adj(q->ipqe_m, i);
1.1 cgd 630: break;
631: }
1.25 cgd 632: nq = q->ipqe_q.le_next;
1.50 thorpej 633: m_freem(q->ipqe_m);
1.25 cgd 634: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 635: pool_put(&ipqent_pool, q);
1.1 cgd 636: }
637:
638: insert:
639: /*
640: * Stick new segment in its place;
641: * check for complete reassembly.
642: */
1.25 cgd 643: if (p == NULL) {
644: LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
645: } else {
646: LIST_INSERT_AFTER(p, ipqe, ipqe_q);
647: }
1.1 cgd 648: next = 0;
1.25 cgd 649: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
650: p = q, q = q->ipqe_q.le_next) {
651: if (q->ipqe_ip->ip_off != next)
1.1 cgd 652: return (0);
1.25 cgd 653: next += q->ipqe_ip->ip_len;
1.1 cgd 654: }
1.25 cgd 655: if (p->ipqe_mff)
1.1 cgd 656: return (0);
657:
658: /*
1.41 thorpej 659: * Reassembly is complete. Check for a bogus message size and
660: * concatenate fragments.
1.1 cgd 661: */
1.25 cgd 662: q = fp->ipq_fragq.lh_first;
663: ip = q->ipqe_ip;
1.41 thorpej 664: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
665: ipstat.ips_toolong++;
666: ip_freef(fp);
667: return (0);
668: }
1.50 thorpej 669: m = q->ipqe_m;
1.1 cgd 670: t = m->m_next;
671: m->m_next = 0;
672: m_cat(m, t);
1.25 cgd 673: nq = q->ipqe_q.le_next;
1.72 thorpej 674: pool_put(&ipqent_pool, q);
1.25 cgd 675: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 676: t = q->ipqe_m;
1.25 cgd 677: nq = q->ipqe_q.le_next;
1.72 thorpej 678: pool_put(&ipqent_pool, q);
1.1 cgd 679: m_cat(m, t);
680: }
681:
682: /*
683: * Create header for new ip packet by
684: * modifying header of first packet;
685: * dequeue and discard fragment reassembly header.
686: * Make header visible.
687: */
688: ip->ip_len = next;
1.25 cgd 689: ip->ip_src = fp->ipq_src;
690: ip->ip_dst = fp->ipq_dst;
691: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 692: FREE(fp, M_FTABLE);
1.1 cgd 693: m->m_len += (ip->ip_hl << 2);
694: m->m_data -= (ip->ip_hl << 2);
695: /* some debugging cruft by sklower, below, will go away soon */
696: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
697: register int plen = 0;
1.50 thorpej 698: for (t = m; t; t = t->m_next)
699: plen += t->m_len;
700: m->m_pkthdr.len = plen;
1.1 cgd 701: }
1.50 thorpej 702: return (m);
1.1 cgd 703:
704: dropfrag:
705: ipstat.ips_fragdropped++;
706: m_freem(m);
1.72 thorpej 707: pool_put(&ipqent_pool, ipqe);
1.1 cgd 708: return (0);
709: }
710:
711: /*
712: * Free a fragment reassembly header and all
713: * associated datagrams.
714: */
1.8 mycroft 715: void
1.1 cgd 716: ip_freef(fp)
717: struct ipq *fp;
718: {
1.25 cgd 719: register struct ipqent *q, *p;
1.1 cgd 720:
1.25 cgd 721: for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) {
722: p = q->ipqe_q.le_next;
1.50 thorpej 723: m_freem(q->ipqe_m);
1.25 cgd 724: LIST_REMOVE(q, ipqe_q);
1.72 thorpej 725: pool_put(&ipqent_pool, q);
1.1 cgd 726: }
1.25 cgd 727: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 728: FREE(fp, M_FTABLE);
1.1 cgd 729: }
730:
731: /*
732: * IP timer processing;
733: * if a timer expires on a reassembly
734: * queue, discard it.
735: */
1.8 mycroft 736: void
1.1 cgd 737: ip_slowtimo()
738: {
1.25 cgd 739: register struct ipq *fp, *nfp;
1.24 mycroft 740: int s = splsoftnet();
1.1 cgd 741:
1.25 cgd 742: for (fp = ipq.lh_first; fp != NULL; fp = nfp) {
743: nfp = fp->ipq_q.le_next;
744: if (--fp->ipq_ttl == 0) {
1.1 cgd 745: ipstat.ips_fragtimeout++;
1.25 cgd 746: ip_freef(fp);
1.1 cgd 747: }
748: }
1.63 matt 749: #ifdef GATEWAY
750: ipflow_slowtimo();
751: #endif
1.1 cgd 752: splx(s);
753: }
754:
755: /*
756: * Drain off all datagram fragments.
757: */
1.8 mycroft 758: void
1.1 cgd 759: ip_drain()
760: {
761:
1.25 cgd 762: while (ipq.lh_first != NULL) {
1.1 cgd 763: ipstat.ips_fragdropped++;
1.25 cgd 764: ip_freef(ipq.lh_first);
1.1 cgd 765: }
766: }
767:
768: /*
769: * Do option processing on a datagram,
770: * possibly discarding it if bad options are encountered,
771: * or forwarding it if source-routed.
772: * Returns 1 if packet has been forwarded/freed,
773: * 0 if the packet should be processed further.
774: */
1.8 mycroft 775: int
1.1 cgd 776: ip_dooptions(m)
777: struct mbuf *m;
778: {
779: register struct ip *ip = mtod(m, struct ip *);
780: register u_char *cp;
781: register struct ip_timestamp *ipt;
782: register struct in_ifaddr *ia;
783: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.13 mycroft 784: struct in_addr *sin, dst;
1.1 cgd 785: n_time ntime;
786:
1.13 mycroft 787: dst = ip->ip_dst;
1.1 cgd 788: cp = (u_char *)(ip + 1);
789: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
790: for (; cnt > 0; cnt -= optlen, cp += optlen) {
791: opt = cp[IPOPT_OPTVAL];
792: if (opt == IPOPT_EOL)
793: break;
794: if (opt == IPOPT_NOP)
795: optlen = 1;
796: else {
797: optlen = cp[IPOPT_OLEN];
798: if (optlen <= 0 || optlen > cnt) {
799: code = &cp[IPOPT_OLEN] - (u_char *)ip;
800: goto bad;
801: }
802: }
803: switch (opt) {
804:
805: default:
806: break;
807:
808: /*
809: * Source routing with record.
810: * Find interface with current destination address.
811: * If none on this machine then drop if strictly routed,
812: * or do nothing if loosely routed.
813: * Record interface address and bring up next address
814: * component. If strictly routed make sure next
815: * address is on directly accessible net.
816: */
817: case IPOPT_LSRR:
818: case IPOPT_SSRR:
1.47 cjs 819: if (ip_allowsrcrt == 0) {
820: type = ICMP_UNREACH;
821: code = ICMP_UNREACH_NET_PROHIB;
822: goto bad;
823: }
1.1 cgd 824: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
825: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
826: goto bad;
827: }
828: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 829: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 830: if (ia == 0) {
831: if (opt == IPOPT_SSRR) {
832: type = ICMP_UNREACH;
833: code = ICMP_UNREACH_SRCFAIL;
834: goto bad;
835: }
836: /*
837: * Loose routing, and not at next destination
838: * yet; nothing to do except forward.
839: */
840: break;
841: }
842: off--; /* 0 origin */
843: if (off > optlen - sizeof(struct in_addr)) {
844: /*
845: * End of source route. Should be for us.
846: */
847: save_rte(cp, ip->ip_src);
1.74.2.1! kenh 848: ifa_delref(&ia->ia_ifa);
1.1 cgd 849: break;
850: }
851: /*
852: * locate outgoing interface
853: */
854: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
855: sizeof(ipaddr.sin_addr));
1.74.2.1! kenh 856: ifa_delref(&ia->ia_ifa);
1.1 cgd 857: if (opt == IPOPT_SSRR) {
858: #define INA struct in_ifaddr *
859: #define SA struct sockaddr *
1.29 mrg 860: ia = (INA)ifa_ifwithladdr((SA)&ipaddr);
1.1 cgd 861: } else
862: ia = ip_rtaddr(ipaddr.sin_addr);
863: if (ia == 0) {
864: type = ICMP_UNREACH;
865: code = ICMP_UNREACH_SRCFAIL;
866: goto bad;
867: }
868: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 869: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 870: (caddr_t)(cp + off), sizeof(struct in_addr));
871: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 872: /*
873: * Let ip_intr's mcast routing check handle mcast pkts
874: */
1.18 mycroft 875: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.74.2.1! kenh 876: ifa_delref(&ia->ia_ifa);
1.1 cgd 877: break;
878:
879: case IPOPT_RR:
880: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
881: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
882: goto bad;
883: }
884: /*
885: * If no space remains, ignore.
886: */
887: off--; /* 0 origin */
888: if (off > optlen - sizeof(struct in_addr))
889: break;
890: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
891: sizeof(ipaddr.sin_addr));
892: /*
893: * locate outgoing interface; if we're the destination,
894: * use the incoming interface (should be same).
895: */
896: if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
897: (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
898: type = ICMP_UNREACH;
899: code = ICMP_UNREACH_HOST;
900: goto bad;
901: }
1.20 mycroft 902: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 903: (caddr_t)(cp + off), sizeof(struct in_addr));
904: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.74.2.1! kenh 905: ifa_delref(&ia->ia_ifa);
1.1 cgd 906: break;
907:
908: case IPOPT_TS:
909: code = cp - (u_char *)ip;
910: ipt = (struct ip_timestamp *)cp;
911: if (ipt->ipt_len < 5)
912: goto bad;
1.15 cgd 913: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.1 cgd 914: if (++ipt->ipt_oflw == 0)
915: goto bad;
916: break;
917: }
918: sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
919: switch (ipt->ipt_flg) {
920:
921: case IPOPT_TS_TSONLY:
922: break;
923:
924: case IPOPT_TS_TSANDADDR:
1.66 thorpej 925: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 926: sizeof(struct in_addr) > ipt->ipt_len)
927: goto bad;
1.13 mycroft 928: ipaddr.sin_addr = dst;
929: ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
930: m->m_pkthdr.rcvif);
931: if (ia == 0)
932: continue;
1.20 mycroft 933: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 934: (caddr_t)sin, sizeof(struct in_addr));
1.74.2.1! kenh 935: ifa_delref(&ia->ia_ifa);
1.1 cgd 936: ipt->ipt_ptr += sizeof(struct in_addr);
937: break;
938:
939: case IPOPT_TS_PRESPEC:
1.66 thorpej 940: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 941: sizeof(struct in_addr) > ipt->ipt_len)
942: goto bad;
943: bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
944: sizeof(struct in_addr));
1.74.2.1! kenh 945: ia = (INA)ifa_ifwithaddr((SA)&ipaddr);
! 946: if (ia == 0)
1.1 cgd 947: continue;
1.74.2.1! kenh 948: ifa_delref(&ia->ia_ifa);
1.1 cgd 949: ipt->ipt_ptr += sizeof(struct in_addr);
950: break;
951:
952: default:
953: goto bad;
954: }
955: ntime = iptime();
956: bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
957: sizeof(n_time));
958: ipt->ipt_ptr += sizeof(n_time);
959: }
960: }
961: if (forward) {
1.26 thorpej 962: if (ip_forwsrcrt == 0) {
963: type = ICMP_UNREACH;
964: code = ICMP_UNREACH_SRCFAIL;
965: goto bad;
966: }
1.1 cgd 967: ip_forward(m, 1);
968: return (1);
1.13 mycroft 969: }
970: return (0);
1.1 cgd 971: bad:
1.13 mycroft 972: ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */
973: icmp_error(m, type, code, 0, 0);
974: ipstat.ips_badoptions++;
1.1 cgd 975: return (1);
976: }
977:
978: /*
979: * Given address of next destination (final or next hop),
980: * return internet address info of interface to be used to get there.
981: */
982: struct in_ifaddr *
983: ip_rtaddr(dst)
984: struct in_addr dst;
985: {
986: register struct sockaddr_in *sin;
1.74.2.1! kenh 987: struct in_ifaddr *ia;
! 988: int s;
1.1 cgd 989:
1.19 mycroft 990: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 991:
1.35 mycroft 992: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 993: if (ipforward_rt.ro_rt) {
994: RTFREE(ipforward_rt.ro_rt);
995: ipforward_rt.ro_rt = 0;
996: }
997: sin->sin_family = AF_INET;
998: sin->sin_len = sizeof(*sin);
999: sin->sin_addr = dst;
1000:
1001: rtalloc(&ipforward_rt);
1002: }
1003: if (ipforward_rt.ro_rt == 0)
1004: return ((struct in_ifaddr *)0);
1.74.2.1! kenh 1005: s = splimp();
! 1006: ia = (ifatoia(ipforward_rt.ro_rt->rt_ifa));
! 1007: ifa_addref(&ia->ia_ifa);
! 1008: splx(s);
! 1009: return ia;
1.1 cgd 1010: }
1011:
1012: /*
1013: * Save incoming source route for use in replies,
1014: * to be picked up later by ip_srcroute if the receiver is interested.
1015: */
1.13 mycroft 1016: void
1.1 cgd 1017: save_rte(option, dst)
1018: u_char *option;
1019: struct in_addr dst;
1020: {
1021: unsigned olen;
1022:
1023: olen = option[IPOPT_OLEN];
1024: #ifdef DIAGNOSTIC
1025: if (ipprintfs)
1.39 christos 1026: printf("save_rte: olen %d\n", olen);
1.1 cgd 1027: #endif
1028: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1029: return;
1030: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
1031: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1032: ip_srcrt.dst = dst;
1033: }
1034:
1035: /*
1036: * Retrieve incoming source route for use in replies,
1037: * in the same form used by setsockopt.
1038: * The first hop is placed before the options, will be removed later.
1039: */
1040: struct mbuf *
1041: ip_srcroute()
1042: {
1043: register struct in_addr *p, *q;
1044: register struct mbuf *m;
1045:
1046: if (ip_nhops == 0)
1047: return ((struct mbuf *)0);
1048: m = m_get(M_DONTWAIT, MT_SOOPTS);
1049: if (m == 0)
1050: return ((struct mbuf *)0);
1051:
1.13 mycroft 1052: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1053:
1054: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1055: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1056: OPTSIZ;
1057: #ifdef DIAGNOSTIC
1058: if (ipprintfs)
1.39 christos 1059: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1060: #endif
1061:
1062: /*
1063: * First save first hop for return route
1064: */
1065: p = &ip_srcrt.route[ip_nhops - 1];
1066: *(mtod(m, struct in_addr *)) = *p--;
1067: #ifdef DIAGNOSTIC
1068: if (ipprintfs)
1.39 christos 1069: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1070: #endif
1071:
1072: /*
1073: * Copy option fields and padding (nop) to mbuf.
1074: */
1075: ip_srcrt.nop = IPOPT_NOP;
1076: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1077: bcopy((caddr_t)&ip_srcrt.nop,
1078: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1079: q = (struct in_addr *)(mtod(m, caddr_t) +
1080: sizeof(struct in_addr) + OPTSIZ);
1081: #undef OPTSIZ
1082: /*
1083: * Record return path as an IP source route,
1084: * reversing the path (pointers are now aligned).
1085: */
1086: while (p >= ip_srcrt.route) {
1087: #ifdef DIAGNOSTIC
1088: if (ipprintfs)
1.39 christos 1089: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1090: #endif
1091: *q++ = *p--;
1092: }
1093: /*
1094: * Last hop goes to final destination.
1095: */
1096: *q = ip_srcrt.dst;
1097: #ifdef DIAGNOSTIC
1098: if (ipprintfs)
1.39 christos 1099: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1100: #endif
1101: return (m);
1102: }
1103:
1104: /*
1105: * Strip out IP options, at higher
1106: * level protocol in the kernel.
1107: * Second argument is buffer to which options
1108: * will be moved, and return value is their length.
1109: * XXX should be deleted; last arg currently ignored.
1110: */
1.8 mycroft 1111: void
1.1 cgd 1112: ip_stripoptions(m, mopt)
1113: register struct mbuf *m;
1114: struct mbuf *mopt;
1115: {
1116: register int i;
1117: struct ip *ip = mtod(m, struct ip *);
1118: register caddr_t opts;
1119: int olen;
1120:
1121: olen = (ip->ip_hl<<2) - sizeof (struct ip);
1122: opts = (caddr_t)(ip + 1);
1123: i = m->m_len - (sizeof (struct ip) + olen);
1124: bcopy(opts + olen, opts, (unsigned)i);
1125: m->m_len -= olen;
1126: if (m->m_flags & M_PKTHDR)
1127: m->m_pkthdr.len -= olen;
1128: ip->ip_hl = sizeof(struct ip) >> 2;
1129: }
1130:
1.23 mycroft 1131: int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1132: 0, 0, 0, 0,
1133: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1134: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1135: EMSGSIZE, EHOSTUNREACH, 0, 0,
1136: 0, 0, 0, 0,
1137: ENOPROTOOPT
1138: };
1139:
1140: /*
1141: * Forward a packet. If some error occurs return the sender
1142: * an icmp packet. Note we can't always generate a meaningful
1143: * icmp message because icmp doesn't have a large enough repertoire
1144: * of codes and types.
1145: *
1146: * If not forwarding, just drop the packet. This could be confusing
1147: * if ipforwarding was zero but some routing protocol was advancing
1148: * us as a gateway to somewhere. However, we must let the routing
1149: * protocol deal with that.
1150: *
1151: * The srcrt parameter indicates whether the packet is being forwarded
1152: * via a source route.
1153: */
1.13 mycroft 1154: void
1.1 cgd 1155: ip_forward(m, srcrt)
1156: struct mbuf *m;
1157: int srcrt;
1158: {
1159: register struct ip *ip = mtod(m, struct ip *);
1160: register struct sockaddr_in *sin;
1161: register struct rtentry *rt;
1.28 christos 1162: int error, type = 0, code = 0;
1.1 cgd 1163: struct mbuf *mcopy;
1.13 mycroft 1164: n_long dest;
1165: struct ifnet *destifp;
1.1 cgd 1166:
1.13 mycroft 1167: dest = 0;
1.1 cgd 1168: #ifdef DIAGNOSTIC
1169: if (ipprintfs)
1.70 thorpej 1170: printf("forward: src %2.2x dst %2.2x ttl %x\n",
1171: ntohl(ip->ip_src.s_addr),
1172: ntohl(ip->ip_dst.s_addr), ip->ip_ttl);
1.1 cgd 1173: #endif
1174: if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1175: ipstat.ips_cantforward++;
1176: m_freem(m);
1177: return;
1178: }
1179: HTONS(ip->ip_id);
1180: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1181: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1182: return;
1183: }
1184: ip->ip_ttl -= IPTTLDEC;
1185:
1.19 mycroft 1186: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1187: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1188: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1189: if (ipforward_rt.ro_rt) {
1190: RTFREE(ipforward_rt.ro_rt);
1191: ipforward_rt.ro_rt = 0;
1192: }
1193: sin->sin_family = AF_INET;
1.35 mycroft 1194: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1195: sin->sin_addr = ip->ip_dst;
1196:
1197: rtalloc(&ipforward_rt);
1198: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1199: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1200: return;
1201: }
1202: rt = ipforward_rt.ro_rt;
1203: }
1204:
1205: /*
1.34 mycroft 1206: * Save at most 68 bytes of the packet in case
1.1 cgd 1207: * we need to generate an ICMP message to the src.
1208: */
1.34 mycroft 1209: mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68));
1.1 cgd 1210:
1211: /*
1212: * If forwarding packet using same interface that it came in on,
1213: * perhaps should send a redirect to sender to shortcut a hop.
1214: * Only send redirect if source is sending directly to us,
1215: * and if packet was not source routed (or has any options).
1216: * Also, don't send redirect if forwarding using a default route
1217: * or a route modified by a redirect.
1218: */
1219: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1220: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1221: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1222: ipsendredirects && !srcrt) {
1.19 mycroft 1223: if (rt->rt_ifa &&
1224: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1225: ifatoia(rt->rt_ifa)->ia_subnet) {
1.1 cgd 1226: if (rt->rt_flags & RTF_GATEWAY)
1.13 mycroft 1227: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1.1 cgd 1228: else
1.13 mycroft 1229: dest = ip->ip_dst.s_addr;
1230: /* Router requirements says to only send host redirects */
1.1 cgd 1231: type = ICMP_REDIRECT;
1.13 mycroft 1232: code = ICMP_REDIRECT_HOST;
1.1 cgd 1233: #ifdef DIAGNOSTIC
1234: if (ipprintfs)
1.39 christos 1235: printf("redirect (%d) to %x\n", code, (u_int32_t)dest);
1.1 cgd 1236: #endif
1237: }
1238: }
1239:
1.27 thorpej 1240: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1241: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1.1 cgd 1242: if (error)
1243: ipstat.ips_cantforward++;
1244: else {
1245: ipstat.ips_forward++;
1246: if (type)
1247: ipstat.ips_redirectsent++;
1248: else {
1.63 matt 1249: if (mcopy) {
1250: #ifdef GATEWAY
1.64 thorpej 1251: if (mcopy->m_flags & M_CANFASTFWD)
1252: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1253: #endif
1.1 cgd 1254: m_freem(mcopy);
1.63 matt 1255: }
1.1 cgd 1256: return;
1257: }
1258: }
1259: if (mcopy == NULL)
1260: return;
1.13 mycroft 1261: destifp = NULL;
1262:
1.1 cgd 1263: switch (error) {
1264:
1265: case 0: /* forwarded, but need redirect */
1266: /* type, code set above */
1267: break;
1268:
1269: case ENETUNREACH: /* shouldn't happen, checked above */
1270: case EHOSTUNREACH:
1271: case ENETDOWN:
1272: case EHOSTDOWN:
1273: default:
1274: type = ICMP_UNREACH;
1275: code = ICMP_UNREACH_HOST;
1276: break;
1277:
1278: case EMSGSIZE:
1279: type = ICMP_UNREACH;
1280: code = ICMP_UNREACH_NEEDFRAG;
1.13 mycroft 1281: if (ipforward_rt.ro_rt)
1282: destifp = ipforward_rt.ro_rt->rt_ifp;
1.1 cgd 1283: ipstat.ips_cantfrag++;
1284: break;
1285:
1286: case ENOBUFS:
1287: type = ICMP_SOURCEQUENCH;
1288: code = 0;
1289: break;
1290: }
1.13 mycroft 1291: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 1292: }
1293:
1294: void
1295: ip_savecontrol(inp, mp, ip, m)
1296: register struct inpcb *inp;
1297: register struct mbuf **mp;
1298: register struct ip *ip;
1299: register struct mbuf *m;
1300: {
1301:
1302: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1303: struct timeval tv;
1304:
1305: microtime(&tv);
1306: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1307: SCM_TIMESTAMP, SOL_SOCKET);
1308: if (*mp)
1309: mp = &(*mp)->m_next;
1310: }
1311: if (inp->inp_flags & INP_RECVDSTADDR) {
1312: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1313: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1314: if (*mp)
1315: mp = &(*mp)->m_next;
1316: }
1317: #ifdef notyet
1318: /*
1319: * XXX
1320: * Moving these out of udp_input() made them even more broken
1321: * than they already were.
1322: * - fenner@parc.xerox.com
1323: */
1324: /* options were tossed already */
1325: if (inp->inp_flags & INP_RECVOPTS) {
1326: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1327: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1328: if (*mp)
1329: mp = &(*mp)->m_next;
1330: }
1331: /* ip_srcroute doesn't do what we want here, need to fix */
1332: if (inp->inp_flags & INP_RECVRETOPTS) {
1333: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1334: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1335: if (*mp)
1336: mp = &(*mp)->m_next;
1337: }
1338: #endif
1339: if (inp->inp_flags & INP_RECVIF) {
1340: struct sockaddr_dl sdl;
1341:
1342: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1343: sdl.sdl_family = AF_LINK;
1344: sdl.sdl_index = m->m_pkthdr.rcvif ?
1345: m->m_pkthdr.rcvif->if_index : 0;
1346: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1347: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1348: IP_RECVIF, IPPROTO_IP);
1349: if (*mp)
1350: mp = &(*mp)->m_next;
1351: }
1.13 mycroft 1352: }
1353:
1354: int
1355: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1356: int *name;
1357: u_int namelen;
1358: void *oldp;
1359: size_t *oldlenp;
1360: void *newp;
1361: size_t newlen;
1362: {
1.52 thorpej 1363: extern int subnetsarelocal;
1364:
1.54 lukem 1365: int error, old;
1366:
1.13 mycroft 1367: /* All sysctl names at this level are terminal. */
1368: if (namelen != 1)
1369: return (ENOTDIR);
1370:
1371: switch (name[0]) {
1372: case IPCTL_FORWARDING:
1373: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1374: case IPCTL_SENDREDIRECTS:
1375: return (sysctl_int(oldp, oldlenp, newp, newlen,
1376: &ipsendredirects));
1377: case IPCTL_DEFTTL:
1378: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1379: #ifdef notyet
1380: case IPCTL_DEFMTU:
1381: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1382: #endif
1.26 thorpej 1383: case IPCTL_FORWSRCRT:
1.47 cjs 1384: /* Don't allow this to change in a secure environment. */
1.26 thorpej 1385: if (securelevel > 0)
1.46 cjs 1386: return (sysctl_rdint(oldp, oldlenp, newp,
1387: ip_forwsrcrt));
1388: else
1389: return (sysctl_int(oldp, oldlenp, newp, newlen,
1390: &ip_forwsrcrt));
1.27 thorpej 1391: case IPCTL_DIRECTEDBCAST:
1392: return (sysctl_int(oldp, oldlenp, newp, newlen,
1393: &ip_directedbcast));
1.47 cjs 1394: case IPCTL_ALLOWSRCRT:
1395: return (sysctl_int(oldp, oldlenp, newp, newlen,
1396: &ip_allowsrcrt));
1.52 thorpej 1397: case IPCTL_SUBNETSARELOCAL:
1398: return (sysctl_int(oldp, oldlenp, newp, newlen,
1399: &subnetsarelocal));
1.53 kml 1400: case IPCTL_MTUDISC:
1.60 kml 1401: error = sysctl_int(oldp, oldlenp, newp, newlen,
1402: &ip_mtudisc);
1403: if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) {
1404: ip_mtudisc_timeout_q =
1405: rt_timer_queue_create(ip_mtudisc_timeout);
1406: } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) {
1407: rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE);
1408: ip_mtudisc_timeout_q = NULL;
1409: }
1410: return error;
1.54 lukem 1411: case IPCTL_ANONPORTMIN:
1412: old = anonportmin;
1413: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1414: if (anonportmin >= anonportmax || anonportmin > 65535
1415: #ifndef IPNOPRIVPORTS
1416: || anonportmin < IPPORT_RESERVED
1417: #endif
1418: ) {
1419: anonportmin = old;
1420: return (EINVAL);
1421: }
1422: return (error);
1423: case IPCTL_ANONPORTMAX:
1424: old = anonportmax;
1425: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1426: if (anonportmin >= anonportmax || anonportmax > 65535
1427: #ifndef IPNOPRIVPORTS
1428: || anonportmax < IPPORT_RESERVED
1429: #endif
1430: ) {
1431: anonportmax = old;
1432: return (EINVAL);
1433: }
1.60 kml 1434: return (error);
1435: case IPCTL_MTUDISCTIMEOUT:
1436: error = sysctl_int(oldp, oldlenp, newp, newlen,
1437: &ip_mtudisc_timeout);
1438: if (ip_mtudisc_timeout_q != NULL)
1439: rt_timer_queue_change(ip_mtudisc_timeout_q,
1440: ip_mtudisc_timeout);
1.54 lukem 1441: return (error);
1.65 matt 1442: #ifdef GATEWAY
1443: case IPCTL_MAXFLOWS:
1.67 thorpej 1444: {
1445: int s;
1446:
1.65 matt 1447: error = sysctl_int(oldp, oldlenp, newp, newlen,
1448: &ip_maxflows);
1.67 thorpej 1449: s = splsoftnet();
1.65 matt 1450: ipflow_reap(0);
1.67 thorpej 1451: splx(s);
1.65 matt 1452: return (error);
1.67 thorpej 1453: }
1.65 matt 1454: #endif
1.13 mycroft 1455: default:
1456: return (EOPNOTSUPP);
1457: }
1458: /* NOTREACHED */
1.1 cgd 1459: }
CVSweb <webmaster@jp.NetBSD.org>