Annotation of src/sys/netinet/ip_input.c, Revision 1.72
1.72 ! thorpej 1: /* $NetBSD: ip_input.c,v 1.71 1998/09/30 21:52:25 tls 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);
1.1 cgd 220: }
221:
222: struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
223: struct route ipforward_rt;
224:
225: /*
226: * Ip input routine. Checksum and byte swap header. If fragmented
227: * try to reassemble. Process options. Pass to next level.
228: */
1.8 mycroft 229: void
1.1 cgd 230: ipintr()
231: {
1.33 mrg 232: register struct ip *ip = NULL;
1.1 cgd 233: register struct mbuf *m;
234: register struct ipq *fp;
235: register struct in_ifaddr *ia;
1.57 tls 236: register struct ifaddr *ifa;
1.25 cgd 237: struct ipqent *ipqe;
1.35 mycroft 238: int hlen = 0, mff, len, s;
1.36 mrg 239: #ifdef PFIL_HOOKS
1.33 mrg 240: struct packet_filter_hook *pfh;
241: struct mbuf *m0;
1.43 mrg 242: int rv;
1.36 mrg 243: #endif /* PFIL_HOOKS */
1.1 cgd 244:
245: next:
246: /*
247: * Get next datagram off input queue and get IP header
248: * in first mbuf.
249: */
250: s = splimp();
251: IF_DEQUEUE(&ipintrq, m);
252: splx(s);
1.13 mycroft 253: if (m == 0)
1.1 cgd 254: return;
255: #ifdef DIAGNOSTIC
256: if ((m->m_flags & M_PKTHDR) == 0)
257: panic("ipintr no HDR");
258: #endif
259: /*
260: * If no IP addresses have been set yet but the interfaces
261: * are receiving, can't do anything with incoming packets yet.
262: */
1.22 mycroft 263: if (in_ifaddr.tqh_first == 0)
1.1 cgd 264: goto bad;
265: ipstat.ips_total++;
266: if (m->m_len < sizeof (struct ip) &&
267: (m = m_pullup(m, sizeof (struct ip))) == 0) {
268: ipstat.ips_toosmall++;
269: goto next;
270: }
271: ip = mtod(m, struct ip *);
1.13 mycroft 272: if (ip->ip_v != IPVERSION) {
273: ipstat.ips_badvers++;
274: goto bad;
275: }
1.1 cgd 276: hlen = ip->ip_hl << 2;
277: if (hlen < sizeof(struct ip)) { /* minimum header length */
278: ipstat.ips_badhlen++;
279: goto bad;
280: }
281: if (hlen > m->m_len) {
282: if ((m = m_pullup(m, hlen)) == 0) {
283: ipstat.ips_badhlen++;
284: goto next;
285: }
286: ip = mtod(m, struct ip *);
287: }
1.28 christos 288: if ((ip->ip_sum = in_cksum(m, hlen)) != 0) {
1.1 cgd 289: ipstat.ips_badsum++;
290: goto bad;
291: }
292:
293: /*
294: * Convert fields to host representation.
295: */
296: NTOHS(ip->ip_len);
297: NTOHS(ip->ip_id);
298: NTOHS(ip->ip_off);
1.35 mycroft 299: len = ip->ip_len;
1.1 cgd 300:
301: /*
302: * Check that the amount of data in the buffers
303: * is as at least much as the IP header would have us expect.
304: * Trim mbufs if longer than we expect.
305: * Drop packet if shorter than we expect.
306: */
1.35 mycroft 307: if (m->m_pkthdr.len < len) {
1.1 cgd 308: ipstat.ips_tooshort++;
309: goto bad;
310: }
1.35 mycroft 311: if (m->m_pkthdr.len > len) {
1.1 cgd 312: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 313: m->m_len = len;
314: m->m_pkthdr.len = len;
1.1 cgd 315: } else
1.35 mycroft 316: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 317: }
318:
1.64 thorpej 319: /*
320: * Assume that we can create a fast-forward IP flow entry
321: * based on this packet.
322: */
323: m->m_flags |= M_CANFASTFWD;
324:
1.36 mrg 325: #ifdef PFIL_HOOKS
1.33 mrg 326: /*
1.64 thorpej 327: * Run through list of hooks for input packets. If there are any
328: * filters which require that additional packets in the flow are
329: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
330: * Note that filters must _never_ set this flag, as another filter
331: * in the list may have previously cleared it.
1.33 mrg 332: */
333: m0 = m;
1.59 mrg 334: for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next)
1.33 mrg 335: if (pfh->pfil_func) {
1.43 mrg 336: rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0);
337: if (rv)
1.40 veego 338: goto next;
1.68 sommerfe 339: m = m0;
340: if (m == NULL)
341: goto next;
342: ip = mtod(m, struct ip *);
1.33 mrg 343: }
1.36 mrg 344: #endif /* PFIL_HOOKS */
1.33 mrg 345:
1.1 cgd 346: /*
347: * Process options and, if not destined for us,
348: * ship it on. ip_dooptions returns 1 when an
349: * error was detected (causing an icmp message
350: * to be sent and the original packet to be freed).
351: */
352: ip_nhops = 0; /* for source routed packets */
353: if (hlen > sizeof (struct ip) && ip_dooptions(m))
354: goto next;
355:
356: /*
357: * Check our list of addresses, to see if the packet is for us.
358: */
1.57 tls 359: INADDR_TO_IA(ip->ip_dst, ia);
360: if (ia != NULL) goto ours;
361: if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
362: for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
363: ifa != NULL; ifa = ifa->ifa_list.tqe_next) {
364: if (ifa->ifa_addr->sa_family != AF_INET) continue;
365: ia = ifatoia(ifa);
1.35 mycroft 366: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
367: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
1.20 mycroft 368: /*
369: * Look for all-0's host part (old broadcast addr),
370: * either for subnet or net.
371: */
372: ip->ip_dst.s_addr == ia->ia_subnet ||
1.18 mycroft 373: ip->ip_dst.s_addr == ia->ia_net)
1.1 cgd 374: goto ours;
1.57 tls 375: /*
376: * An interface with IP address zero accepts
377: * all packets that arrive on that interface.
378: */
379: if (in_nullhost(ia->ia_addr.sin_addr))
380: goto ours;
1.1 cgd 381: }
382: }
1.18 mycroft 383: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 384: struct in_multi *inm;
385: #ifdef MROUTING
386: extern struct socket *ip_mrouter;
1.10 brezak 387:
388: if (m->m_flags & M_EXT) {
389: if ((m = m_pullup(m, hlen)) == 0) {
390: ipstat.ips_toosmall++;
391: goto next;
392: }
393: ip = mtod(m, struct ip *);
394: }
1.4 hpeyerl 395:
396: if (ip_mrouter) {
397: /*
398: * If we are acting as a multicast router, all
399: * incoming multicast packets are passed to the
400: * kernel-level multicast forwarding function.
401: * The packet is returned (relatively) intact; if
402: * ip_mforward() returns a non-zero value, the packet
403: * must be discarded, else it may be accepted below.
404: *
405: * (The IP ident field is put in the same byte order
406: * as expected when ip_mforward() is called from
407: * ip_output().)
408: */
409: ip->ip_id = htons(ip->ip_id);
1.13 mycroft 410: if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
411: ipstat.ips_cantforward++;
1.4 hpeyerl 412: m_freem(m);
413: goto next;
414: }
415: ip->ip_id = ntohs(ip->ip_id);
416:
417: /*
418: * The process-level routing demon needs to receive
419: * all multicast IGMP packets, whether or not this
420: * host belongs to their destination groups.
421: */
422: if (ip->ip_p == IPPROTO_IGMP)
423: goto ours;
1.13 mycroft 424: ipstat.ips_forward++;
1.4 hpeyerl 425: }
426: #endif
427: /*
428: * See if we belong to the destination multicast group on the
429: * arrival interface.
430: */
431: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
432: if (inm == NULL) {
1.13 mycroft 433: ipstat.ips_cantforward++;
1.4 hpeyerl 434: m_freem(m);
435: goto next;
436: }
437: goto ours;
438: }
1.19 mycroft 439: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 440: in_nullhost(ip->ip_dst))
1.1 cgd 441: goto ours;
442:
443: /*
444: * Not for us; forward if possible and desirable.
445: */
446: if (ipforwarding == 0) {
447: ipstat.ips_cantforward++;
448: m_freem(m);
449: } else
450: ip_forward(m, 0);
451: goto next;
452:
453: ours:
454: /*
455: * If offset or IP_MF are set, must reassemble.
456: * Otherwise, nothing need be done.
457: * (We could look in the reassembly queue to see
458: * if the packet was previously fragmented,
459: * but it's not worth the time; just let them time out.)
460: */
1.37 perry 461: if (ip->ip_off & ~(IP_DF|IP_RF)) {
1.1 cgd 462: /*
463: * Look for queue of fragments
464: * of this datagram.
465: */
1.25 cgd 466: for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next)
1.1 cgd 467: if (ip->ip_id == fp->ipq_id &&
1.35 mycroft 468: in_hosteq(ip->ip_src, fp->ipq_src) &&
469: in_hosteq(ip->ip_dst, fp->ipq_dst) &&
1.1 cgd 470: ip->ip_p == fp->ipq_p)
471: goto found;
472: fp = 0;
473: found:
474:
475: /*
476: * Adjust ip_len to not reflect header,
1.25 cgd 477: * set ipqe_mff if more fragments are expected,
1.1 cgd 478: * convert offset of this to bytes.
479: */
480: ip->ip_len -= hlen;
1.25 cgd 481: mff = (ip->ip_off & IP_MF) != 0;
482: if (mff) {
1.16 cgd 483: /*
484: * Make sure that fragments have a data length
485: * that's a non-zero multiple of 8 bytes.
486: */
1.17 cgd 487: if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
1.16 cgd 488: ipstat.ips_badfrags++;
489: goto bad;
490: }
491: }
1.1 cgd 492: ip->ip_off <<= 3;
493:
494: /*
495: * If datagram marked as having more fragments
496: * or if this is not the first fragment,
497: * attempt reassembly; if it succeeds, proceed.
498: */
1.25 cgd 499: if (mff || ip->ip_off) {
1.1 cgd 500: ipstat.ips_fragments++;
1.72 ! thorpej 501: ipqe = pool_get(&ipqent_pool, PR_NOWAIT);
1.25 cgd 502: if (ipqe == NULL) {
503: ipstat.ips_rcvmemdrop++;
504: goto bad;
505: }
506: ipqe->ipqe_mff = mff;
1.50 thorpej 507: ipqe->ipqe_m = m;
1.25 cgd 508: ipqe->ipqe_ip = ip;
1.50 thorpej 509: m = ip_reass(ipqe, fp);
510: if (m == 0)
1.1 cgd 511: goto next;
1.13 mycroft 512: ipstat.ips_reassembled++;
1.50 thorpej 513: ip = mtod(m, struct ip *);
1.1 cgd 514: } else
515: if (fp)
516: ip_freef(fp);
517: } else
518: ip->ip_len -= hlen;
519:
520: /*
521: * Switch out to protocol's input routine.
522: */
523: ipstat.ips_delivered++;
524: (*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
525: goto next;
526: bad:
527: m_freem(m);
528: goto next;
529: }
530:
531: /*
532: * Take incoming datagram fragment and try to
533: * reassemble it into whole datagram. If a chain for
534: * reassembly of this datagram already exists, then it
535: * is given as fp; otherwise have to make a chain.
536: */
1.50 thorpej 537: struct mbuf *
1.25 cgd 538: ip_reass(ipqe, fp)
539: register struct ipqent *ipqe;
1.1 cgd 540: register struct ipq *fp;
541: {
1.50 thorpej 542: register struct mbuf *m = ipqe->ipqe_m;
1.25 cgd 543: register struct ipqent *nq, *p, *q;
544: struct ip *ip;
1.1 cgd 545: struct mbuf *t;
1.25 cgd 546: int hlen = ipqe->ipqe_ip->ip_hl << 2;
1.1 cgd 547: int i, next;
548:
549: /*
550: * Presence of header sizes in mbufs
551: * would confuse code below.
552: */
553: m->m_data += hlen;
554: m->m_len -= hlen;
555:
556: /*
557: * If first fragment to arrive, create a reassembly queue.
558: */
559: if (fp == 0) {
1.50 thorpej 560: MALLOC(fp, struct ipq *, sizeof (struct ipq),
561: M_FTABLE, M_NOWAIT);
562: if (fp == NULL)
1.1 cgd 563: goto dropfrag;
1.25 cgd 564: LIST_INSERT_HEAD(&ipq, fp, ipq_q);
1.1 cgd 565: fp->ipq_ttl = IPFRAGTTL;
1.25 cgd 566: fp->ipq_p = ipqe->ipqe_ip->ip_p;
567: fp->ipq_id = ipqe->ipqe_ip->ip_id;
568: LIST_INIT(&fp->ipq_fragq);
569: fp->ipq_src = ipqe->ipqe_ip->ip_src;
570: fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
571: p = NULL;
1.1 cgd 572: goto insert;
573: }
574:
575: /*
576: * Find a segment which begins after this one does.
577: */
1.25 cgd 578: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
579: p = q, q = q->ipqe_q.le_next)
580: if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off)
1.1 cgd 581: break;
582:
583: /*
584: * If there is a preceding segment, it may provide some of
585: * our data already. If so, drop the data from the incoming
586: * segment. If it provides all of our data, drop us.
587: */
1.25 cgd 588: if (p != NULL) {
589: i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len -
590: ipqe->ipqe_ip->ip_off;
1.1 cgd 591: if (i > 0) {
1.25 cgd 592: if (i >= ipqe->ipqe_ip->ip_len)
1.1 cgd 593: goto dropfrag;
1.50 thorpej 594: m_adj(ipqe->ipqe_m, i);
1.25 cgd 595: ipqe->ipqe_ip->ip_off += i;
596: ipqe->ipqe_ip->ip_len -= i;
1.1 cgd 597: }
598: }
599:
600: /*
601: * While we overlap succeeding segments trim them or,
602: * if they are completely covered, dequeue them.
603: */
1.25 cgd 604: for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len >
605: q->ipqe_ip->ip_off; q = nq) {
606: i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) -
607: q->ipqe_ip->ip_off;
608: if (i < q->ipqe_ip->ip_len) {
609: q->ipqe_ip->ip_len -= i;
610: q->ipqe_ip->ip_off += i;
1.50 thorpej 611: m_adj(q->ipqe_m, i);
1.1 cgd 612: break;
613: }
1.25 cgd 614: nq = q->ipqe_q.le_next;
1.50 thorpej 615: m_freem(q->ipqe_m);
1.25 cgd 616: LIST_REMOVE(q, ipqe_q);
1.72 ! thorpej 617: pool_put(&ipqent_pool, q);
1.1 cgd 618: }
619:
620: insert:
621: /*
622: * Stick new segment in its place;
623: * check for complete reassembly.
624: */
1.25 cgd 625: if (p == NULL) {
626: LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
627: } else {
628: LIST_INSERT_AFTER(p, ipqe, ipqe_q);
629: }
1.1 cgd 630: next = 0;
1.25 cgd 631: for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
632: p = q, q = q->ipqe_q.le_next) {
633: if (q->ipqe_ip->ip_off != next)
1.1 cgd 634: return (0);
1.25 cgd 635: next += q->ipqe_ip->ip_len;
1.1 cgd 636: }
1.25 cgd 637: if (p->ipqe_mff)
1.1 cgd 638: return (0);
639:
640: /*
1.41 thorpej 641: * Reassembly is complete. Check for a bogus message size and
642: * concatenate fragments.
1.1 cgd 643: */
1.25 cgd 644: q = fp->ipq_fragq.lh_first;
645: ip = q->ipqe_ip;
1.41 thorpej 646: if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
647: ipstat.ips_toolong++;
648: ip_freef(fp);
649: return (0);
650: }
1.50 thorpej 651: m = q->ipqe_m;
1.1 cgd 652: t = m->m_next;
653: m->m_next = 0;
654: m_cat(m, t);
1.25 cgd 655: nq = q->ipqe_q.le_next;
1.72 ! thorpej 656: pool_put(&ipqent_pool, q);
1.25 cgd 657: for (q = nq; q != NULL; q = nq) {
1.50 thorpej 658: t = q->ipqe_m;
1.25 cgd 659: nq = q->ipqe_q.le_next;
1.72 ! thorpej 660: pool_put(&ipqent_pool, q);
1.1 cgd 661: m_cat(m, t);
662: }
663:
664: /*
665: * Create header for new ip packet by
666: * modifying header of first packet;
667: * dequeue and discard fragment reassembly header.
668: * Make header visible.
669: */
670: ip->ip_len = next;
1.25 cgd 671: ip->ip_src = fp->ipq_src;
672: ip->ip_dst = fp->ipq_dst;
673: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 674: FREE(fp, M_FTABLE);
1.1 cgd 675: m->m_len += (ip->ip_hl << 2);
676: m->m_data -= (ip->ip_hl << 2);
677: /* some debugging cruft by sklower, below, will go away soon */
678: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
679: register int plen = 0;
1.50 thorpej 680: for (t = m; t; t = t->m_next)
681: plen += t->m_len;
682: m->m_pkthdr.len = plen;
1.1 cgd 683: }
1.50 thorpej 684: return (m);
1.1 cgd 685:
686: dropfrag:
687: ipstat.ips_fragdropped++;
688: m_freem(m);
1.72 ! thorpej 689: pool_put(&ipqent_pool, ipqe);
1.1 cgd 690: return (0);
691: }
692:
693: /*
694: * Free a fragment reassembly header and all
695: * associated datagrams.
696: */
1.8 mycroft 697: void
1.1 cgd 698: ip_freef(fp)
699: struct ipq *fp;
700: {
1.25 cgd 701: register struct ipqent *q, *p;
1.1 cgd 702:
1.25 cgd 703: for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) {
704: p = q->ipqe_q.le_next;
1.50 thorpej 705: m_freem(q->ipqe_m);
1.25 cgd 706: LIST_REMOVE(q, ipqe_q);
1.72 ! thorpej 707: pool_put(&ipqent_pool, q);
1.1 cgd 708: }
1.25 cgd 709: LIST_REMOVE(fp, ipq_q);
1.50 thorpej 710: FREE(fp, M_FTABLE);
1.1 cgd 711: }
712:
713: /*
714: * IP timer processing;
715: * if a timer expires on a reassembly
716: * queue, discard it.
717: */
1.8 mycroft 718: void
1.1 cgd 719: ip_slowtimo()
720: {
1.25 cgd 721: register struct ipq *fp, *nfp;
1.24 mycroft 722: int s = splsoftnet();
1.1 cgd 723:
1.25 cgd 724: for (fp = ipq.lh_first; fp != NULL; fp = nfp) {
725: nfp = fp->ipq_q.le_next;
726: if (--fp->ipq_ttl == 0) {
1.1 cgd 727: ipstat.ips_fragtimeout++;
1.25 cgd 728: ip_freef(fp);
1.1 cgd 729: }
730: }
1.63 matt 731: #ifdef GATEWAY
732: ipflow_slowtimo();
733: #endif
1.1 cgd 734: splx(s);
735: }
736:
737: /*
738: * Drain off all datagram fragments.
739: */
1.8 mycroft 740: void
1.1 cgd 741: ip_drain()
742: {
743:
1.25 cgd 744: while (ipq.lh_first != NULL) {
1.1 cgd 745: ipstat.ips_fragdropped++;
1.25 cgd 746: ip_freef(ipq.lh_first);
1.1 cgd 747: }
748: }
749:
750: /*
751: * Do option processing on a datagram,
752: * possibly discarding it if bad options are encountered,
753: * or forwarding it if source-routed.
754: * Returns 1 if packet has been forwarded/freed,
755: * 0 if the packet should be processed further.
756: */
1.8 mycroft 757: int
1.1 cgd 758: ip_dooptions(m)
759: struct mbuf *m;
760: {
761: register struct ip *ip = mtod(m, struct ip *);
762: register u_char *cp;
763: register struct ip_timestamp *ipt;
764: register struct in_ifaddr *ia;
765: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.13 mycroft 766: struct in_addr *sin, dst;
1.1 cgd 767: n_time ntime;
768:
1.13 mycroft 769: dst = ip->ip_dst;
1.1 cgd 770: cp = (u_char *)(ip + 1);
771: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
772: for (; cnt > 0; cnt -= optlen, cp += optlen) {
773: opt = cp[IPOPT_OPTVAL];
774: if (opt == IPOPT_EOL)
775: break;
776: if (opt == IPOPT_NOP)
777: optlen = 1;
778: else {
779: optlen = cp[IPOPT_OLEN];
780: if (optlen <= 0 || optlen > cnt) {
781: code = &cp[IPOPT_OLEN] - (u_char *)ip;
782: goto bad;
783: }
784: }
785: switch (opt) {
786:
787: default:
788: break;
789:
790: /*
791: * Source routing with record.
792: * Find interface with current destination address.
793: * If none on this machine then drop if strictly routed,
794: * or do nothing if loosely routed.
795: * Record interface address and bring up next address
796: * component. If strictly routed make sure next
797: * address is on directly accessible net.
798: */
799: case IPOPT_LSRR:
800: case IPOPT_SSRR:
1.47 cjs 801: if (ip_allowsrcrt == 0) {
802: type = ICMP_UNREACH;
803: code = ICMP_UNREACH_NET_PROHIB;
804: goto bad;
805: }
1.1 cgd 806: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
807: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
808: goto bad;
809: }
810: ipaddr.sin_addr = ip->ip_dst;
1.19 mycroft 811: ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
1.1 cgd 812: if (ia == 0) {
813: if (opt == IPOPT_SSRR) {
814: type = ICMP_UNREACH;
815: code = ICMP_UNREACH_SRCFAIL;
816: goto bad;
817: }
818: /*
819: * Loose routing, and not at next destination
820: * yet; nothing to do except forward.
821: */
822: break;
823: }
824: off--; /* 0 origin */
825: if (off > optlen - sizeof(struct in_addr)) {
826: /*
827: * End of source route. Should be for us.
828: */
829: save_rte(cp, ip->ip_src);
830: break;
831: }
832: /*
833: * locate outgoing interface
834: */
835: bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
836: sizeof(ipaddr.sin_addr));
837: if (opt == IPOPT_SSRR) {
838: #define INA struct in_ifaddr *
839: #define SA struct sockaddr *
1.29 mrg 840: ia = (INA)ifa_ifwithladdr((SA)&ipaddr);
1.1 cgd 841: } else
842: ia = ip_rtaddr(ipaddr.sin_addr);
843: if (ia == 0) {
844: type = ICMP_UNREACH;
845: code = ICMP_UNREACH_SRCFAIL;
846: goto bad;
847: }
848: ip->ip_dst = ipaddr.sin_addr;
1.20 mycroft 849: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 850: (caddr_t)(cp + off), sizeof(struct in_addr));
851: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 852: /*
853: * Let ip_intr's mcast routing check handle mcast pkts
854: */
1.18 mycroft 855: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 856: break;
857:
858: case IPOPT_RR:
859: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
860: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
861: goto bad;
862: }
863: /*
864: * If no space remains, ignore.
865: */
866: off--; /* 0 origin */
867: if (off > optlen - sizeof(struct in_addr))
868: break;
869: bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
870: sizeof(ipaddr.sin_addr));
871: /*
872: * locate outgoing interface; if we're the destination,
873: * use the incoming interface (should be same).
874: */
875: if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
876: (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
877: type = ICMP_UNREACH;
878: code = ICMP_UNREACH_HOST;
879: goto bad;
880: }
1.20 mycroft 881: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 882: (caddr_t)(cp + off), sizeof(struct in_addr));
883: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
884: break;
885:
886: case IPOPT_TS:
887: code = cp - (u_char *)ip;
888: ipt = (struct ip_timestamp *)cp;
889: if (ipt->ipt_len < 5)
890: goto bad;
1.15 cgd 891: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.1 cgd 892: if (++ipt->ipt_oflw == 0)
893: goto bad;
894: break;
895: }
896: sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
897: switch (ipt->ipt_flg) {
898:
899: case IPOPT_TS_TSONLY:
900: break;
901:
902: case IPOPT_TS_TSANDADDR:
1.66 thorpej 903: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 904: sizeof(struct in_addr) > ipt->ipt_len)
905: goto bad;
1.13 mycroft 906: ipaddr.sin_addr = dst;
907: ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
908: m->m_pkthdr.rcvif);
909: if (ia == 0)
910: continue;
1.20 mycroft 911: bcopy((caddr_t)&ia->ia_addr.sin_addr,
1.1 cgd 912: (caddr_t)sin, sizeof(struct in_addr));
913: ipt->ipt_ptr += sizeof(struct in_addr);
914: break;
915:
916: case IPOPT_TS_PRESPEC:
1.66 thorpej 917: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.1 cgd 918: sizeof(struct in_addr) > ipt->ipt_len)
919: goto bad;
920: bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
921: sizeof(struct in_addr));
922: if (ifa_ifwithaddr((SA)&ipaddr) == 0)
923: continue;
924: ipt->ipt_ptr += sizeof(struct in_addr);
925: break;
926:
927: default:
928: goto bad;
929: }
930: ntime = iptime();
931: bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
932: sizeof(n_time));
933: ipt->ipt_ptr += sizeof(n_time);
934: }
935: }
936: if (forward) {
1.26 thorpej 937: if (ip_forwsrcrt == 0) {
938: type = ICMP_UNREACH;
939: code = ICMP_UNREACH_SRCFAIL;
940: goto bad;
941: }
1.1 cgd 942: ip_forward(m, 1);
943: return (1);
1.13 mycroft 944: }
945: return (0);
1.1 cgd 946: bad:
1.13 mycroft 947: ip->ip_len -= ip->ip_hl << 2; /* XXX icmp_error adds in hdr length */
948: icmp_error(m, type, code, 0, 0);
949: ipstat.ips_badoptions++;
1.1 cgd 950: return (1);
951: }
952:
953: /*
954: * Given address of next destination (final or next hop),
955: * return internet address info of interface to be used to get there.
956: */
957: struct in_ifaddr *
958: ip_rtaddr(dst)
959: struct in_addr dst;
960: {
961: register struct sockaddr_in *sin;
962:
1.19 mycroft 963: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 964:
1.35 mycroft 965: if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
1.1 cgd 966: if (ipforward_rt.ro_rt) {
967: RTFREE(ipforward_rt.ro_rt);
968: ipforward_rt.ro_rt = 0;
969: }
970: sin->sin_family = AF_INET;
971: sin->sin_len = sizeof(*sin);
972: sin->sin_addr = dst;
973:
974: rtalloc(&ipforward_rt);
975: }
976: if (ipforward_rt.ro_rt == 0)
977: return ((struct in_ifaddr *)0);
1.19 mycroft 978: return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
1.1 cgd 979: }
980:
981: /*
982: * Save incoming source route for use in replies,
983: * to be picked up later by ip_srcroute if the receiver is interested.
984: */
1.13 mycroft 985: void
1.1 cgd 986: save_rte(option, dst)
987: u_char *option;
988: struct in_addr dst;
989: {
990: unsigned olen;
991:
992: olen = option[IPOPT_OLEN];
993: #ifdef DIAGNOSTIC
994: if (ipprintfs)
1.39 christos 995: printf("save_rte: olen %d\n", olen);
1.1 cgd 996: #endif
997: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
998: return;
999: bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
1000: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1001: ip_srcrt.dst = dst;
1002: }
1003:
1004: /*
1005: * Retrieve incoming source route for use in replies,
1006: * in the same form used by setsockopt.
1007: * The first hop is placed before the options, will be removed later.
1008: */
1009: struct mbuf *
1010: ip_srcroute()
1011: {
1012: register struct in_addr *p, *q;
1013: register struct mbuf *m;
1014:
1015: if (ip_nhops == 0)
1016: return ((struct mbuf *)0);
1017: m = m_get(M_DONTWAIT, MT_SOOPTS);
1018: if (m == 0)
1019: return ((struct mbuf *)0);
1020:
1.13 mycroft 1021: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1022:
1023: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1024: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1025: OPTSIZ;
1026: #ifdef DIAGNOSTIC
1027: if (ipprintfs)
1.39 christos 1028: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1.1 cgd 1029: #endif
1030:
1031: /*
1032: * First save first hop for return route
1033: */
1034: p = &ip_srcrt.route[ip_nhops - 1];
1035: *(mtod(m, struct in_addr *)) = *p--;
1036: #ifdef DIAGNOSTIC
1037: if (ipprintfs)
1.39 christos 1038: printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1.1 cgd 1039: #endif
1040:
1041: /*
1042: * Copy option fields and padding (nop) to mbuf.
1043: */
1044: ip_srcrt.nop = IPOPT_NOP;
1045: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1046: bcopy((caddr_t)&ip_srcrt.nop,
1047: mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1048: q = (struct in_addr *)(mtod(m, caddr_t) +
1049: sizeof(struct in_addr) + OPTSIZ);
1050: #undef OPTSIZ
1051: /*
1052: * Record return path as an IP source route,
1053: * reversing the path (pointers are now aligned).
1054: */
1055: while (p >= ip_srcrt.route) {
1056: #ifdef DIAGNOSTIC
1057: if (ipprintfs)
1.39 christos 1058: printf(" %x", ntohl(q->s_addr));
1.1 cgd 1059: #endif
1060: *q++ = *p--;
1061: }
1062: /*
1063: * Last hop goes to final destination.
1064: */
1065: *q = ip_srcrt.dst;
1066: #ifdef DIAGNOSTIC
1067: if (ipprintfs)
1.39 christos 1068: printf(" %x\n", ntohl(q->s_addr));
1.1 cgd 1069: #endif
1070: return (m);
1071: }
1072:
1073: /*
1074: * Strip out IP options, at higher
1075: * level protocol in the kernel.
1076: * Second argument is buffer to which options
1077: * will be moved, and return value is their length.
1078: * XXX should be deleted; last arg currently ignored.
1079: */
1.8 mycroft 1080: void
1.1 cgd 1081: ip_stripoptions(m, mopt)
1082: register struct mbuf *m;
1083: struct mbuf *mopt;
1084: {
1085: register int i;
1086: struct ip *ip = mtod(m, struct ip *);
1087: register caddr_t opts;
1088: int olen;
1089:
1090: olen = (ip->ip_hl<<2) - sizeof (struct ip);
1091: opts = (caddr_t)(ip + 1);
1092: i = m->m_len - (sizeof (struct ip) + olen);
1093: bcopy(opts + olen, opts, (unsigned)i);
1094: m->m_len -= olen;
1095: if (m->m_flags & M_PKTHDR)
1096: m->m_pkthdr.len -= olen;
1097: ip->ip_hl = sizeof(struct ip) >> 2;
1098: }
1099:
1.23 mycroft 1100: int inetctlerrmap[PRC_NCMDS] = {
1.1 cgd 1101: 0, 0, 0, 0,
1102: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1103: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1104: EMSGSIZE, EHOSTUNREACH, 0, 0,
1105: 0, 0, 0, 0,
1106: ENOPROTOOPT
1107: };
1108:
1109: /*
1110: * Forward a packet. If some error occurs return the sender
1111: * an icmp packet. Note we can't always generate a meaningful
1112: * icmp message because icmp doesn't have a large enough repertoire
1113: * of codes and types.
1114: *
1115: * If not forwarding, just drop the packet. This could be confusing
1116: * if ipforwarding was zero but some routing protocol was advancing
1117: * us as a gateway to somewhere. However, we must let the routing
1118: * protocol deal with that.
1119: *
1120: * The srcrt parameter indicates whether the packet is being forwarded
1121: * via a source route.
1122: */
1.13 mycroft 1123: void
1.1 cgd 1124: ip_forward(m, srcrt)
1125: struct mbuf *m;
1126: int srcrt;
1127: {
1128: register struct ip *ip = mtod(m, struct ip *);
1129: register struct sockaddr_in *sin;
1130: register struct rtentry *rt;
1.28 christos 1131: int error, type = 0, code = 0;
1.1 cgd 1132: struct mbuf *mcopy;
1.13 mycroft 1133: n_long dest;
1134: struct ifnet *destifp;
1.1 cgd 1135:
1.13 mycroft 1136: dest = 0;
1.1 cgd 1137: #ifdef DIAGNOSTIC
1138: if (ipprintfs)
1.70 thorpej 1139: printf("forward: src %2.2x dst %2.2x ttl %x\n",
1140: ntohl(ip->ip_src.s_addr),
1141: ntohl(ip->ip_dst.s_addr), ip->ip_ttl);
1.1 cgd 1142: #endif
1143: if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1144: ipstat.ips_cantforward++;
1145: m_freem(m);
1146: return;
1147: }
1148: HTONS(ip->ip_id);
1149: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1150: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1151: return;
1152: }
1153: ip->ip_ttl -= IPTTLDEC;
1154:
1.19 mycroft 1155: sin = satosin(&ipforward_rt.ro_dst);
1.1 cgd 1156: if ((rt = ipforward_rt.ro_rt) == 0 ||
1.35 mycroft 1157: !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1.1 cgd 1158: if (ipforward_rt.ro_rt) {
1159: RTFREE(ipforward_rt.ro_rt);
1160: ipforward_rt.ro_rt = 0;
1161: }
1162: sin->sin_family = AF_INET;
1.35 mycroft 1163: sin->sin_len = sizeof(struct sockaddr_in);
1.1 cgd 1164: sin->sin_addr = ip->ip_dst;
1165:
1166: rtalloc(&ipforward_rt);
1167: if (ipforward_rt.ro_rt == 0) {
1.13 mycroft 1168: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1.1 cgd 1169: return;
1170: }
1171: rt = ipforward_rt.ro_rt;
1172: }
1173:
1174: /*
1.34 mycroft 1175: * Save at most 68 bytes of the packet in case
1.1 cgd 1176: * we need to generate an ICMP message to the src.
1177: */
1.34 mycroft 1178: mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68));
1.1 cgd 1179:
1180: /*
1181: * If forwarding packet using same interface that it came in on,
1182: * perhaps should send a redirect to sender to shortcut a hop.
1183: * Only send redirect if source is sending directly to us,
1184: * and if packet was not source routed (or has any options).
1185: * Also, don't send redirect if forwarding using a default route
1186: * or a route modified by a redirect.
1187: */
1188: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1189: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.35 mycroft 1190: !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1.1 cgd 1191: ipsendredirects && !srcrt) {
1.19 mycroft 1192: if (rt->rt_ifa &&
1193: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1194: ifatoia(rt->rt_ifa)->ia_subnet) {
1.1 cgd 1195: if (rt->rt_flags & RTF_GATEWAY)
1.13 mycroft 1196: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1.1 cgd 1197: else
1.13 mycroft 1198: dest = ip->ip_dst.s_addr;
1199: /* Router requirements says to only send host redirects */
1.1 cgd 1200: type = ICMP_REDIRECT;
1.13 mycroft 1201: code = ICMP_REDIRECT_HOST;
1.1 cgd 1202: #ifdef DIAGNOSTIC
1203: if (ipprintfs)
1.39 christos 1204: printf("redirect (%d) to %x\n", code, (u_int32_t)dest);
1.1 cgd 1205: #endif
1206: }
1207: }
1208:
1.27 thorpej 1209: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1210: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1.1 cgd 1211: if (error)
1212: ipstat.ips_cantforward++;
1213: else {
1214: ipstat.ips_forward++;
1215: if (type)
1216: ipstat.ips_redirectsent++;
1217: else {
1.63 matt 1218: if (mcopy) {
1219: #ifdef GATEWAY
1.64 thorpej 1220: if (mcopy->m_flags & M_CANFASTFWD)
1221: ipflow_create(&ipforward_rt, mcopy);
1.63 matt 1222: #endif
1.1 cgd 1223: m_freem(mcopy);
1.63 matt 1224: }
1.1 cgd 1225: return;
1226: }
1227: }
1228: if (mcopy == NULL)
1229: return;
1.13 mycroft 1230: destifp = NULL;
1231:
1.1 cgd 1232: switch (error) {
1233:
1234: case 0: /* forwarded, but need redirect */
1235: /* type, code set above */
1236: break;
1237:
1238: case ENETUNREACH: /* shouldn't happen, checked above */
1239: case EHOSTUNREACH:
1240: case ENETDOWN:
1241: case EHOSTDOWN:
1242: default:
1243: type = ICMP_UNREACH;
1244: code = ICMP_UNREACH_HOST;
1245: break;
1246:
1247: case EMSGSIZE:
1248: type = ICMP_UNREACH;
1249: code = ICMP_UNREACH_NEEDFRAG;
1.13 mycroft 1250: if (ipforward_rt.ro_rt)
1251: destifp = ipforward_rt.ro_rt->rt_ifp;
1.1 cgd 1252: ipstat.ips_cantfrag++;
1253: break;
1254:
1255: case ENOBUFS:
1256: type = ICMP_SOURCEQUENCH;
1257: code = 0;
1258: break;
1259: }
1.13 mycroft 1260: icmp_error(mcopy, type, code, dest, destifp);
1.44 thorpej 1261: }
1262:
1263: void
1264: ip_savecontrol(inp, mp, ip, m)
1265: register struct inpcb *inp;
1266: register struct mbuf **mp;
1267: register struct ip *ip;
1268: register struct mbuf *m;
1269: {
1270:
1271: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1272: struct timeval tv;
1273:
1274: microtime(&tv);
1275: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1276: SCM_TIMESTAMP, SOL_SOCKET);
1277: if (*mp)
1278: mp = &(*mp)->m_next;
1279: }
1280: if (inp->inp_flags & INP_RECVDSTADDR) {
1281: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1282: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1283: if (*mp)
1284: mp = &(*mp)->m_next;
1285: }
1286: #ifdef notyet
1287: /*
1288: * XXX
1289: * Moving these out of udp_input() made them even more broken
1290: * than they already were.
1291: * - fenner@parc.xerox.com
1292: */
1293: /* options were tossed already */
1294: if (inp->inp_flags & INP_RECVOPTS) {
1295: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1296: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1297: if (*mp)
1298: mp = &(*mp)->m_next;
1299: }
1300: /* ip_srcroute doesn't do what we want here, need to fix */
1301: if (inp->inp_flags & INP_RECVRETOPTS) {
1302: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1303: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1304: if (*mp)
1305: mp = &(*mp)->m_next;
1306: }
1307: #endif
1308: if (inp->inp_flags & INP_RECVIF) {
1309: struct sockaddr_dl sdl;
1310:
1311: sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1312: sdl.sdl_family = AF_LINK;
1313: sdl.sdl_index = m->m_pkthdr.rcvif ?
1314: m->m_pkthdr.rcvif->if_index : 0;
1315: sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1316: *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1317: IP_RECVIF, IPPROTO_IP);
1318: if (*mp)
1319: mp = &(*mp)->m_next;
1320: }
1.13 mycroft 1321: }
1322:
1323: int
1324: ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1325: int *name;
1326: u_int namelen;
1327: void *oldp;
1328: size_t *oldlenp;
1329: void *newp;
1330: size_t newlen;
1331: {
1.52 thorpej 1332: extern int subnetsarelocal;
1333:
1.54 lukem 1334: int error, old;
1335:
1.13 mycroft 1336: /* All sysctl names at this level are terminal. */
1337: if (namelen != 1)
1338: return (ENOTDIR);
1339:
1340: switch (name[0]) {
1341: case IPCTL_FORWARDING:
1342: return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1343: case IPCTL_SENDREDIRECTS:
1344: return (sysctl_int(oldp, oldlenp, newp, newlen,
1345: &ipsendredirects));
1346: case IPCTL_DEFTTL:
1347: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1348: #ifdef notyet
1349: case IPCTL_DEFMTU:
1350: return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1351: #endif
1.26 thorpej 1352: case IPCTL_FORWSRCRT:
1.47 cjs 1353: /* Don't allow this to change in a secure environment. */
1.26 thorpej 1354: if (securelevel > 0)
1.46 cjs 1355: return (sysctl_rdint(oldp, oldlenp, newp,
1356: ip_forwsrcrt));
1357: else
1358: return (sysctl_int(oldp, oldlenp, newp, newlen,
1359: &ip_forwsrcrt));
1.27 thorpej 1360: case IPCTL_DIRECTEDBCAST:
1361: return (sysctl_int(oldp, oldlenp, newp, newlen,
1362: &ip_directedbcast));
1.47 cjs 1363: case IPCTL_ALLOWSRCRT:
1364: return (sysctl_int(oldp, oldlenp, newp, newlen,
1365: &ip_allowsrcrt));
1.52 thorpej 1366: case IPCTL_SUBNETSARELOCAL:
1367: return (sysctl_int(oldp, oldlenp, newp, newlen,
1368: &subnetsarelocal));
1.53 kml 1369: case IPCTL_MTUDISC:
1.60 kml 1370: error = sysctl_int(oldp, oldlenp, newp, newlen,
1371: &ip_mtudisc);
1372: if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) {
1373: ip_mtudisc_timeout_q =
1374: rt_timer_queue_create(ip_mtudisc_timeout);
1375: } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) {
1376: rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE);
1377: ip_mtudisc_timeout_q = NULL;
1378: }
1379: return error;
1.54 lukem 1380: case IPCTL_ANONPORTMIN:
1381: old = anonportmin;
1382: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1383: if (anonportmin >= anonportmax || anonportmin > 65535
1384: #ifndef IPNOPRIVPORTS
1385: || anonportmin < IPPORT_RESERVED
1386: #endif
1387: ) {
1388: anonportmin = old;
1389: return (EINVAL);
1390: }
1391: return (error);
1392: case IPCTL_ANONPORTMAX:
1393: old = anonportmax;
1394: error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1395: if (anonportmin >= anonportmax || anonportmax > 65535
1396: #ifndef IPNOPRIVPORTS
1397: || anonportmax < IPPORT_RESERVED
1398: #endif
1399: ) {
1400: anonportmax = old;
1401: return (EINVAL);
1402: }
1.60 kml 1403: return (error);
1404: case IPCTL_MTUDISCTIMEOUT:
1405: error = sysctl_int(oldp, oldlenp, newp, newlen,
1406: &ip_mtudisc_timeout);
1407: if (ip_mtudisc_timeout_q != NULL)
1408: rt_timer_queue_change(ip_mtudisc_timeout_q,
1409: ip_mtudisc_timeout);
1.54 lukem 1410: return (error);
1.65 matt 1411: #ifdef GATEWAY
1412: case IPCTL_MAXFLOWS:
1.67 thorpej 1413: {
1414: int s;
1415:
1.65 matt 1416: error = sysctl_int(oldp, oldlenp, newp, newlen,
1417: &ip_maxflows);
1.67 thorpej 1418: s = splsoftnet();
1.65 matt 1419: ipflow_reap(0);
1.67 thorpej 1420: splx(s);
1.65 matt 1421: return (error);
1.67 thorpej 1422: }
1.65 matt 1423: #endif
1.13 mycroft 1424: default:
1425: return (EOPNOTSUPP);
1426: }
1427: /* NOTREACHED */
1.1 cgd 1428: }
CVSweb <webmaster@jp.NetBSD.org>
![[BACK]](/icons/back.gif){kind=icon}
Return to ip_input.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [cvs.NetBSD.org] / src / sys / netinet