![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to ip_input.c CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [Apple XNU] / XNU / bsd / netinet|
Return to ip_input.c CVS log | Up to [Apple XNU] / XNU / bsd / netinet |
1.1 root 1: /*
2: * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
3: *
4: * @APPLE_LICENSE_HEADER_START@
5: *
6: * The contents of this file constitute Original Code as defined in and
7: * are subject to the Apple Public Source License Version 1.1 (the
8: * "License"). You may not use this file except in compliance with the
9: * License. Please obtain a copy of the License at
10: * http://www.apple.com/publicsource and read it before using this file.
11: *
12: * This Original Code and all software distributed under the License are
13: * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
14: * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
15: * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
16: * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT. Please see the
17: * License for the specific language governing rights and limitations
18: * under the License.
19: *
20: * @APPLE_LICENSE_HEADER_END@
21: */
22: /*
23: * Copyright (c) 1982, 1986, 1988, 1993
24: * The Regents of the University of California. All rights reserved.
25: *
26: * Redistribution and use in source and binary forms, with or without
27: * modification, are permitted provided that the following conditions
28: * are met:
29: * 1. Redistributions of source code must retain the above copyright
30: * notice, this list of conditions and the following disclaimer.
31: * 2. Redistributions in binary form must reproduce the above copyright
32: * notice, this list of conditions and the following disclaimer in the
33: * documentation and/or other materials provided with the distribution.
34: * 3. All advertising materials mentioning features or use of this software
35: * must display the following acknowledgement:
36: * This product includes software developed by the University of
37: * California, Berkeley and its contributors.
38: * 4. Neither the name of the University nor the names of its contributors
39: * may be used to endorse or promote products derived from this software
40: * without specific prior written permission.
41: *
42: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
43: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
44: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
45: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
46: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
47: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
48: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
49: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
50: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
51: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
52: * SUCH DAMAGE.
53: *
54: * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
55: * $ANA: ip_input.c,v 1.5 1996/09/18 14:34:59 wollman Exp $
56: */
57:
58: #define _IP_VHL
59:
60: #include <stddef.h>
61:
62: #include <sys/param.h>
63: #include <sys/systm.h>
64: #include <sys/mbuf.h>
65: #include <sys/malloc.h>
66: #include <sys/domain.h>
67: #include <sys/protosw.h>
68: #include <sys/socket.h>
69: #include <sys/time.h>
70: #include <sys/kernel.h>
71: #include <sys/syslog.h>
72: #include <sys/sysctl.h>
73:
74: #include <kern/queue.h>
75:
76: #include <net/if.h>
77: #include <net/if_var.h>
78: #include <net/if_dl.h>
79: #include <net/route.h>
80: #include <net/netisr.h>
81:
82: #include <netinet/in.h>
83: #include <netinet/in_systm.h>
84: #include <netinet/in_var.h>
85: #include <netinet/ip.h>
86: #include <netinet/in_pcb.h>
87: #include <netinet/ip_var.h>
88: #include <netinet/ip_icmp.h>
89: #include <sys/socketvar.h>
90:
91: #if IPFIREWALL
92: #include <netinet/ip_fw.h>
93: #endif
94:
95: #if DUMMYNET
96: #include <netinet/ip_dummynet.h>
97: #endif
98:
99: int rsvp_on = 0;
100: static int ip_rsvp_on;
101: struct socket *ip_rsvpd;
102:
103: int ipforwarding = 0;
104: SYSCTL_INT(_net_inet_ip, IPCTL_FORWARDING, forwarding, CTLFLAG_RW,
105: &ipforwarding, 0, "");
106:
107: static int ipsendredirects = 1; /* XXX */
108: SYSCTL_INT(_net_inet_ip, IPCTL_SENDREDIRECTS, redirect, CTLFLAG_RW,
109: &ipsendredirects, 0, "");
110:
111: int ip_defttl = IPDEFTTL;
112: SYSCTL_INT(_net_inet_ip, IPCTL_DEFTTL, ttl, CTLFLAG_RW,
113: &ip_defttl, 0, "");
114:
115: static int ip_dosourceroute = 0;
116: SYSCTL_INT(_net_inet_ip, IPCTL_SOURCEROUTE, sourceroute, CTLFLAG_RW,
117: &ip_dosourceroute, 0, "");
118:
119: static int ip_acceptsourceroute = 0;
120: SYSCTL_INT(_net_inet_ip, IPCTL_ACCEPTSOURCEROUTE, accept_sourceroute,
121: CTLFLAG_RW, &ip_acceptsourceroute, 0, "");
122: #if DIAGNOSTIC
123: static int ipprintfs = 0;
124: #endif
125:
126: extern struct domain inetdomain;
127: extern struct protosw inetsw[];
128: struct protosw *ip_protox[IPPROTO_MAX];
129: static int ipqmaxlen = IFQ_MAXLEN;
130: struct in_ifaddrhead in_ifaddrhead; /* first inet address */
131: struct ifqueue ipintrq;
132: SYSCTL_INT(_net_inet_ip, IPCTL_INTRQMAXLEN, intr_queue_maxlen, CTLFLAG_RD,
133: &ipintrq.ifq_maxlen, 0, "");
134: SYSCTL_INT(_net_inet_ip, IPCTL_INTRQDROPS, intr_queue_drops, CTLFLAG_RD,
135: &ipintrq.ifq_drops, 0, "");
136:
137: struct ipstat ipstat;
138: SYSCTL_STRUCT(_net_inet_ip, IPCTL_STATS, stats, CTLFLAG_RD,
139: &ipstat, ipstat, "");
140:
141: /* Packet reassembly stuff */
142: #define IPREASS_NHASH_LOG2 6
143: #define IPREASS_NHASH (1 << IPREASS_NHASH_LOG2)
144: #define IPREASS_HMASK (IPREASS_NHASH - 1)
145: #define IPREASS_HASH(x,y) \
146: ((((x) & 0xF | ((((x) >> 8) & 0xF) << 4)) ^ (y)) & IPREASS_HMASK)
147:
148: static struct ipq ipq[IPREASS_NHASH];
149: static int nipq = 0; /* total # of reass queues */
150: static int maxnipq;
151:
152: #if IPCTL_DEFMTU
153: SYSCTL_INT(_net_inet_ip, IPCTL_DEFMTU, mtu, CTLFLAG_RW,
154: &ip_mtu, 0, "");
155: #endif
156:
157: #if !defined(COMPAT_IPFW) || COMPAT_IPFW == 1
158: #undef COMPAT_IPFW
159: #define COMPAT_IPFW 1
160: #else
161: #undef COMPAT_IPFW
162: #endif
163:
164: #if COMPAT_IPFW
165:
166: #include <netinet/ip_fw.h>
167:
168: /* Firewall hooks */
169: ip_fw_chk_t *ip_fw_chk_ptr;
170: ip_fw_ctl_t *ip_fw_ctl_ptr;
171:
172: #if DUMMYNET
173: ip_dn_ctl_t *ip_dn_ctl_ptr;
174: #endif
175:
176: /* IP Network Address Translation (NAT) hooks */
177: ip_nat_t *ip_nat_ptr;
178: ip_nat_ctl_t *ip_nat_ctl_ptr;
179: #endif
180:
181: #if defined(IPFILTER_LKM) || defined(IPFILTER)
182: int iplattach __P((void));
183: int (*fr_checkp) __P((struct ip *, int, struct ifnet *, int, struct mbuf **)) = NULL;
184: #endif
185:
186:
187: /*
188: * We need to save the IP options in case a protocol wants to respond
189: * to an incoming packet over the same route if the packet got here
190: * using IP source routing. This allows connection establishment and
191: * maintenance when the remote end is on a network that is not known
192: * to us.
193: */
194: static int ip_nhops = 0;
195: static struct ip_srcrt {
196: struct in_addr dst; /* final destination */
197: char nop; /* one NOP to align */
198: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
199: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
200: } ip_srcrt;
201:
202: #if IPDIVERT
203: /*
204: * Shared variable between ip_input() and ip_reass() to communicate
205: * about which packets, once assembled from fragments, get diverted,
206: * and to which port.
207: */
208: static u_short frag_divert_port;
209: #endif
210:
211: struct sockaddr_in *ip_fw_fwd_addr;
212:
213: static void save_rte __P((u_char *, struct in_addr));
214: static int ip_dooptions __P((struct mbuf *));
215: static void ip_forward __P((struct mbuf *, int));
216: static void ip_freef __P((struct ipq *));
217: static struct ip *
218: ip_reass __P((struct mbuf *, struct ipq *, struct ipq *));
219: static struct in_ifaddr *
220: ip_rtaddr __P((struct in_addr));
221: void ipintr __P((void));
222: /*
223: * IP initialization: fill in IP protocol switch table.
224: * All protocols not implemented in kernel go to raw IP protocol handler.
225: */
226: void
227: ip_init()
228: {
229: register struct protosw *pr;
230: register int i;
231: static ip_initialized = 0;
232:
233: if (!ip_initialized)
234: {
235: TAILQ_INIT(&in_ifaddrhead);
236: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
237: if (pr == 0)
238: panic("ip_init");
239: for (i = 0; i < IPPROTO_MAX; i++)
240: ip_protox[i] = pr;
241: for (pr = inetdomain.dom_protosw; pr; pr = pr->pr_next)
242: { if(!((unsigned int)pr->pr_domain)) continue; /* If uninitialized, skip */
243: if (pr->pr_domain->dom_family == PF_INET &&
244: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
245: ip_protox[pr->pr_protocol] = pr;
246: }
247: for (i = 0; i < IPREASS_NHASH; i++)
248: ipq[i].next = ipq[i].prev = &ipq[i];
249:
250: maxnipq = nmbclusters/4;
251:
252: ip_id = time_second & 0xffff;
253: ipintrq.ifq_maxlen = ipqmaxlen;
254: #if DUMMYNET
255: ip_dn_init();
256: #endif
257: #if IPNAT
258: ip_nat_init();
259: #endif
260: #if IPFILTER
261: iplattach();
262: #endif
263: ip_initialized = 1;
264: }
265: }
266:
267: /* Initialize the PF_INET domain, and add in the pre-defined protos */
268: void
269: in_dinit()
270: { register int i;
271: register struct protosw *pr;
272: register struct domain *dp;
273: static inetdomain_initted = 0;
274: extern int in_proto_count;
275:
276: if (!inetdomain_initted)
277: { kprintf("Initing %d protosw entries\n", in_proto_count);
278: dp = &inetdomain;
279:
280: for (i=0, pr = &inetsw[0]; i<in_proto_count; i++, pr++)
281: net_add_proto(pr, dp);
282: inetdomain_initted = 1;
283: }
284: }
285:
286: static struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
287: static struct route ipforward_rt;
288:
289: /*
290: * Ip input routine. Checksum and byte swap header. If fragmented
291: * try to reassemble. Process options. Pass to next level.
292: */
293: void
294: ip_input(struct mbuf *m)
295: {
296: struct ip *ip;
297: struct ipq *fp;
298: struct in_ifaddr *ia;
299: int i, hlen, mff;
300: u_short sum;
301: #if !IPDIVERT /* dummy variable for the firewall code to play with */
302: u_short ip_divert_cookie = 0 ;
303: #endif
304: #if COMPAT_IPFW
305: struct ip_fw_chain *rule = NULL ;
306: #endif
307:
308: #if IPFIREWALL && DUMMYNET
309: /*
310: * dummynet packet are prepended a vestigial mbuf with
311: * m_type = MT_DUMMYNET and m_data pointing to the matching
312: * rule.
313: */
314: if (m->m_type == MT_DUMMYNET) {
315: struct mbuf *m0 = m ;
316: rule = (struct ip_fw_chain *)(m->m_data) ;
317: m = m->m_next ;
318: FREE(m0, M_IPFW);
319: ip = mtod(m, struct ip *);
320: hlen = IP_VHL_HL(ip->ip_vhl) << 2;
321: goto iphack ;
322: } else
323: rule = NULL ;
324: #endif
325:
326: #if DIAGNOSTIC
327: if (m == NULL || (m->m_flags & M_PKTHDR) == 0)
328: panic("ip_input no HDR");
329: #endif
330: /*
331: * If no IP addresses have been set yet but the interfaces
332: * are receiving, can't do anything with incoming packets yet.
333: * XXX This is broken! We should be able to receive broadcasts
334: * and multicasts even without any local addresses configured.
335: */
336: if (TAILQ_EMPTY(&in_ifaddrhead))
337: goto bad;
338: ipstat.ips_total++;
339:
340: if (m->m_pkthdr.len < sizeof(struct ip))
341: goto tooshort;
342:
343: if (m->m_len < sizeof (struct ip) &&
344: (m = m_pullup(m, sizeof (struct ip))) == 0) {
345: ipstat.ips_toosmall++;
346: return;
347: }
348: ip = mtod(m, struct ip *);
349:
350: if (IP_VHL_V(ip->ip_vhl) != IPVERSION) {
351: ipstat.ips_badvers++;
352: goto bad;
353: }
354:
355: hlen = IP_VHL_HL(ip->ip_vhl) << 2;
356: if (hlen < sizeof(struct ip)) { /* minimum header length */
357: ipstat.ips_badhlen++;
358: goto bad;
359: }
360: if (hlen > m->m_len) {
361: if ((m = m_pullup(m, hlen)) == 0) {
362: ipstat.ips_badhlen++;
363: return;
364: }
365: ip = mtod(m, struct ip *);
366: }
367:
368: sum = in_cksum(m, hlen);
369:
370: if (sum) {
371: ipstat.ips_badsum++;
372: goto bad;
373: }
374:
375: /*
376: * Convert fields to host representation.
377: */
378: NTOHS(ip->ip_len);
379: if (ip->ip_len < hlen) {
380: ipstat.ips_badlen++;
381: goto bad;
382: }
383: NTOHS(ip->ip_id);
384: NTOHS(ip->ip_off);
385:
386: /*
387: * Check that the amount of data in the buffers
388: * is as at least much as the IP header would have us expect.
389: * Trim mbufs if longer than we expect.
390: * Drop packet if shorter than we expect.
391: */
392: if (m->m_pkthdr.len < ip->ip_len) {
393: tooshort:
394: ipstat.ips_tooshort++;
395: goto bad;
396: }
397: if (m->m_pkthdr.len > ip->ip_len) {
398: if (m->m_len == m->m_pkthdr.len) {
399: m->m_len = ip->ip_len;
400: m->m_pkthdr.len = ip->ip_len;
401: } else
402: m_adj(m, ip->ip_len - m->m_pkthdr.len);
403: }
404: /*
405: * IpHack's section.
406: * Right now when no processing on packet has done
407: * and it is still fresh out of network we do our black
408: * deals with it.
409: * - Firewall: deny/allow/divert
410: * - Xlate: translate packet's addr/port (NAT).
411: * - Pipe: pass pkt through dummynet.
412: * - Wrap: fake packet's addr/port <unimpl.>
413: * - Encapsulate: put it in another IP and send out. <unimp.>
414: */
415:
416: #if defined(IPFIREWALL) && defined(DUMMYNET)
417: iphack:
418: #endif
419: #if defined(IPFILTER) || defined(IPFILTER_LKM)
420: /*
421: * Check if we want to allow this packet to be processed.
422: * Consider it to be bad if not.
423: */
424: if (fr_checkp) {
425: struct mbuf *m1 = m;
426:
427: if ((*fr_checkp)(ip, hlen, m->m_pkthdr.rcvif, 0, &m1) || !m1)
428: return;
429: ip = mtod(m = m1, struct ip *);
430: }
431: #endif
432: #if COMPAT_IPFW
433: if (ip_fw_chk_ptr) {
434: #if IPFIREWALL_FORWARD
435: /*
436: * If we've been forwarded from the output side, then
437: * skip the firewall a second time
438: */
439: if (ip_fw_fwd_addr)
440: goto ours;
441: #endif /* IPFIREWALL_FORWARD */
442: i = (*ip_fw_chk_ptr)(&ip, hlen, NULL, &ip_divert_cookie,
443: &m, &rule, &ip_fw_fwd_addr);
444: /*
445: * see the comment in ip_output for the return values
446: * produced by the firewall.
447: */
448: if (!m) /* packet discarded by firewall */
449: return ;
450: if (i == 0 && ip_fw_fwd_addr == NULL) /* common case */
451: goto pass ;
452: #if DUMMYNET
453: if (i & 0x10000) {
454: /* send packet to the appropriate pipe */
455: dummynet_io(i&0xffff,DN_TO_IP_IN,m,NULL,NULL,0, rule);
456: return ;
457: }
458: #endif
459: #if IPDIVERT
460: if (i > 0 && i < 0x10000) {
461: /* Divert packet */
462: frag_divert_port = i & 0xffff ;
463: goto ours;
464: }
465: #endif
466: #if IPFIREWALL_FORWARD
467: if (i == 0 && ip_fw_fwd_addr != NULL)
468: goto pass ;
469: #endif
470: /*
471: * if we get here, the packet must be dropped
472: */
473: m_freem(m);
474: return;
475: }
476: pass:
477:
478: if (ip_nat_ptr && !(*ip_nat_ptr)(&ip, &m, m->m_pkthdr.rcvif, IP_NAT_IN))
479: return;
480: #endif /* !COMPAT_IPFW */
481:
482: /*
483: * Process options and, if not destined for us,
484: * ship it on. ip_dooptions returns 1 when an
485: * error was detected (causing an icmp message
486: * to be sent and the original packet to be freed).
487: */
488: ip_nhops = 0; /* for source routed packets */
489: if (hlen > sizeof (struct ip) && ip_dooptions(m))
490: return;
491:
492: /* greedy RSVP, snatches any PATH packet of the RSVP protocol and no
493: * matter if it is destined to another node, or whether it is
494: * a multicast one, RSVP wants it! and prevents it from being forwarded
495: * anywhere else. Also checks if the rsvp daemon is running before
496: * grabbing the packet.
497: */
498: if (rsvp_on && ip->ip_p==IPPROTO_RSVP)
499: goto ours;
500:
501: /*
502: * Check our list of addresses, to see if the packet is for us.
503: */
504: for (ia = TAILQ_FIRST(&in_ifaddrhead); ia;
505: ia = TAILQ_NEXT(ia, ia_link)) {
506: #define satosin(sa) ((struct sockaddr_in *)(sa))
507:
508: if (IA_SIN(ia)->sin_addr.s_addr == ip->ip_dst.s_addr)
509: goto ours;
510:
511: if (IA_SIN(ia)->sin_addr.s_addr == INADDR_ANY)
512: goto ours;
513:
514: #if IPFIREWALL_FORWARD
515: /*
516: * If the addr to forward to is one of ours, we pretend to
517: * be the destination for this packet.
518: */
519: if (ip_fw_fwd_addr != NULL &&
520: IA_SIN(ia)->sin_addr.s_addr ==
521: ip_fw_fwd_addr->sin_addr.s_addr)
522: goto ours;
523: #endif
524: if (ia->ia_ifp && ia->ia_ifp->if_flags & IFF_BROADCAST) {
525: if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr ==
526: ip->ip_dst.s_addr)
527: goto ours;
528: if (ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr)
529: goto ours;
530: }
531: }
532: if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) {
533: struct in_multi *inm;
534: if (ip_mrouter) {
535: /*
536: * If we are acting as a multicast router, all
537: * incoming multicast packets are passed to the
538: * kernel-level multicast forwarding function.
539: * The packet is returned (relatively) intact; if
540: * ip_mforward() returns a non-zero value, the packet
541: * must be discarded, else it may be accepted below.
542: *
543: * (The IP ident field is put in the same byte order
544: * as expected when ip_mforward() is called from
545: * ip_output().)
546: */
547: ip->ip_id = htons(ip->ip_id);
548: if (ip_mforward(ip, m->m_pkthdr.rcvif, m, 0) != 0) {
549: ipstat.ips_cantforward++;
550: m_freem(m);
551: return;
552: }
553: ip->ip_id = ntohs(ip->ip_id);
554:
555: /*
556: * The process-level routing demon needs to receive
557: * all multicast IGMP packets, whether or not this
558: * host belongs to their destination groups.
559: */
560: if (ip->ip_p == IPPROTO_IGMP)
561: goto ours;
562: ipstat.ips_forward++;
563: }
564: /*
565: * See if we belong to the destination multicast group on the
566: * arrival interface.
567: */
568: IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
569: if (inm == NULL) {
570: ipstat.ips_notmember++;
571: m_freem(m);
572: return;
573: }
574: goto ours;
575: }
576: if (ip->ip_dst.s_addr == (u_long)INADDR_BROADCAST)
577: goto ours;
578: if (ip->ip_dst.s_addr == INADDR_ANY)
579: goto ours;
580:
581: /*
582: * Not for us; forward if possible and desirable.
583: */
584: if (ipforwarding == 0) {
585: ipstat.ips_cantforward++;
586: m_freem(m);
587: } else
588: ip_forward(m, 0);
589: return;
590:
591: ours:
592:
593: /*
594: * If offset or IP_MF are set, must reassemble.
595: * Otherwise, nothing need be done.
596: * (We could look in the reassembly queue to see
597: * if the packet was previously fragmented,
598: * but it's not worth the time; just let them time out.)
599: */
600: if (ip->ip_off & (IP_MF | IP_OFFMASK | IP_RF)) {
601: if (m->m_flags & M_EXT) { /* XXX */
602: if ((m = m_pullup(m, hlen)) == 0) {
603: ipstat.ips_toosmall++;
604: #if IPDIVERT
605: frag_divert_port = 0;
606: ip_divert_cookie = 0;
607: #endif
608: return;
609: }
610: ip = mtod(m, struct ip *);
611: }
612: sum = IPREASS_HASH(ip->ip_src.s_addr, ip->ip_id);
613: /*
614: * Look for queue of fragments
615: * of this datagram.
616: */
617: for (fp = ipq[sum].next; fp != &ipq[sum]; fp = fp->next)
618: if (ip->ip_id == fp->ipq_id &&
619: ip->ip_src.s_addr == fp->ipq_src.s_addr &&
620: ip->ip_dst.s_addr == fp->ipq_dst.s_addr &&
621: ip->ip_p == fp->ipq_p)
622: goto found;
623:
624: fp = 0;
625:
626: /* check if there's a place for the new queue */
627: if (nipq > maxnipq) {
628: /*
629: * drop something from the tail of the current queue
630: * before proceeding further
631: */
632: if (ipq[sum].prev == &ipq[sum]) { /* gak */
633: for (i = 0; i < IPREASS_NHASH; i++) {
634: if (ipq[i].prev != &ipq[i]) {
635: ip_freef(ipq[i].prev);
636: break;
637: }
638: }
639: } else
640: ip_freef(ipq[sum].prev);
641: }
642: found:
643: /*
644: * Adjust ip_len to not reflect header,
645: * set ip_mff if more fragments are expected,
646: * convert offset of this to bytes.
647: */
648: ip->ip_len -= hlen;
649: mff = (ip->ip_off & IP_MF) != 0;
650: if (mff) {
651: /*
652: * Make sure that fragments have a data length
653: * that's a non-zero multiple of 8 bytes.
654: */
655: if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
656: ipstat.ips_toosmall++; /* XXX */
657: goto bad;
658: }
659: m->m_flags |= M_FRAG;
660: }
661: ip->ip_off <<= 3;
662:
663: /*
664: * If datagram marked as having more fragments
665: * or if this is not the first fragment,
666: * attempt reassembly; if it succeeds, proceed.
667: */
668: if (mff || ip->ip_off) {
669: ipstat.ips_fragments++;
670: m->m_pkthdr.header = ip;
671: ip = ip_reass(m, fp, &ipq[sum]);
672: if (ip == 0) {
673: #if IPFIREWALL_FORWARD
674: ip_fw_fwd_addr = NULL;
675: #endif
676: return;
677: }
678: /* Get the length of the reassembled packets header */
679: hlen = IP_VHL_HL(ip->ip_vhl) << 2;
680: ipstat.ips_reassembled++;
681: m = dtom(ip);
682: #if IPDIVERT
683: if (frag_divert_port) {
684: struct mbuf m;
685: m.m_next = 0;
686: m.m_len = hlen;
687: m.m_data = (char *) ip;
688: ip->ip_len += hlen;
689: HTONS(ip->ip_len);
690: HTONS(ip->ip_off);
691: HTONS(ip->ip_id);
692: ip->ip_sum = 0;
693: ip->ip_sum = in_cksum(&m, hlen);
694: NTOHS(ip->ip_id);
695: NTOHS(ip->ip_off);
696: NTOHS(ip->ip_len);
697: ip->ip_len -= hlen;
698: }
699: #endif
700: } else
701: if (fp)
702: ip_freef(fp);
703: } else
704: ip->ip_len -= hlen;
705:
706: #if IPDIVERT
707: /*
708: * Divert reassembled packets to the divert protocol if required
709: * If divert port is null then cookie should be too,
710: * so we shouldn't need to clear them here. Assume ip_divert does so.
711: */
712: if (frag_divert_port) {
713: ipstat.ips_delivered++;
714: ip_divert_port = frag_divert_port;
715: frag_divert_port = 0;
716: (*ip_protox[IPPROTO_DIVERT]->pr_input)(m, hlen);
717: return;
718: }
719:
720: /* Don't let packets divert themselves */
721: if (ip->ip_p == IPPROTO_DIVERT) {
722: ipstat.ips_noproto++;
723: goto bad;
724: }
725:
726: #endif
727:
728: /*
729: * Switch out to protocol's input routine.
730: */
731: ipstat.ips_delivered++;
732: (*ip_protox[ip->ip_p]->pr_input)(m, hlen);
733: #if IPFIREWALL_FORWARD
734: ip_fw_fwd_addr = NULL; /* tcp needed it */
735: #endif
736: return;
737: bad:
738: #if IPFIREWALL_FORWARD
739: ip_fw_fwd_addr = NULL;
740: #endif
741: m_freem(m);
742: }
743:
744: /*
745: * IP software interrupt routine - to go away sometime soon
746: */
747: void
748: ipintr(void)
749: {
750: int s;
751: struct mbuf *m;
752:
753: while(1) {
754: s = splimp();
755: IF_DEQUEUE(&ipintrq, m);
756: splx(s);
757: if (m == 0)
758: return;
759:
760: ip_input(m);
761: }
762: }
763:
764: NETISR_SET(NETISR_IP, ipintr);
765:
766: /*
767: * Take incoming datagram fragment and try to
768: * reassemble it into whole datagram. If a chain for
769: * reassembly of this datagram already exists, then it
770: * is given as fp; otherwise have to make a chain.
771: */
772: static struct ip *
773: ip_reass(m, fp, where)
774: register struct mbuf *m;
775: register struct ipq *fp;
776: struct ipq *where;
777: {
778: struct ip *ip = mtod(m, struct ip *);
779: register struct mbuf *p = 0, *q, *nq;
780: struct mbuf *t;
781: int hlen = IP_VHL_HL(ip->ip_vhl) << 2;
782: int i, next;
783:
784: /*
785: * Presence of header sizes in mbufs
786: * would confuse code below.
787: */
788: m->m_data += hlen;
789: m->m_len -= hlen;
790:
791: /*
792: * If first fragment to arrive, create a reassembly queue.
793: */
794: if (fp == 0) {
795: if ((t = m_get(M_DONTWAIT, MT_FTABLE)) == NULL)
796: goto dropfrag;
797: fp = mtod(t, struct ipq *);
798: insque((void *) fp, (void *) where);
799: nipq++;
800: fp->ipq_ttl = IPFRAGTTL;
801: fp->ipq_p = ip->ip_p;
802: fp->ipq_id = ip->ip_id;
803: fp->ipq_src = ip->ip_src;
804: fp->ipq_dst = ip->ip_dst;
805: fp->ipq_frags = m;
806: m->m_nextpkt = NULL;
807: #if IPDIVERT
808: fp->ipq_divert = 0;
809: fp->ipq_div_cookie = 0;
810: #endif
811: goto inserted;
812: }
813:
814: #define GETIP(m) ((struct ip*)((m)->m_pkthdr.header))
815:
816: /*
817: * Find a segment which begins after this one does.
818: */
819: for (p = NULL, q = fp->ipq_frags; q; p = q, q = q->m_nextpkt)
820: if (GETIP(q)->ip_off > ip->ip_off)
821: break;
822:
823: /*
824: * If there is a preceding segment, it may provide some of
825: * our data already. If so, drop the data from the incoming
826: * segment. If it provides all of our data, drop us, otherwise
827: * stick new segment in the proper place.
828: */
829: if (p) {
830: i = GETIP(p)->ip_off + GETIP(p)->ip_len - ip->ip_off;
831: if (i > 0) {
832: if (i >= ip->ip_len)
833: goto dropfrag;
834: m_adj(dtom(ip), i);
835: ip->ip_off += i;
836: ip->ip_len -= i;
837: }
838: m->m_nextpkt = p->m_nextpkt;
839: p->m_nextpkt = m;
840: } else {
841: m->m_nextpkt = fp->ipq_frags;
842: fp->ipq_frags = m;
843: }
844:
845: /*
846: * While we overlap succeeding segments trim them or,
847: * if they are completely covered, dequeue them.
848: */
849: for (; q != NULL && ip->ip_off + ip->ip_len > GETIP(q)->ip_off;
850: q = nq) {
851: i = (ip->ip_off + ip->ip_len) -
852: GETIP(q)->ip_off;
853: if (i < GETIP(q)->ip_len) {
854: GETIP(q)->ip_len -= i;
855: GETIP(q)->ip_off += i;
856: m_adj(q, i);
857: break;
858: }
859: nq = q->m_nextpkt;
860: m->m_nextpkt = nq;
861: m_freem(q);
862: }
863:
864: inserted:
865:
866: #if IPDIVERT
867: /*
868: * Any fragment diverting causes the whole packet to divert
869: */
870: if (frag_divert_port) {
871: fp->ipq_divert = frag_divert_port;
872: fp->ipq_div_cookie = ip_divert_cookie;
873: }
874: frag_divert_port = 0;
875: ip_divert_cookie = 0;
876: #endif
877:
878: /*
879: * Check for complete reassembly.
880: */
881: next = 0;
882: for (p = NULL, q = fp->ipq_frags; q; p = q, q = q->m_nextpkt) {
883: if (GETIP(q)->ip_off != next)
884: return (0);
885: next += GETIP(q)->ip_len;
886: }
887: /* Make sure the last packet didn't have the IP_MF flag */
888: if (p->m_flags & M_FRAG)
889: return (0);
890:
891: /*
892: * Reassembly is complete. Make sure the packet is a sane size.
893: */
894: q = fp->ipq_frags;
895: ip = GETIP(q);
896: if (next + (IP_VHL_HL(ip->ip_vhl) << 2) > IP_MAXPACKET) {
897: ipstat.ips_toolong++;
898: ip_freef(fp);
899: return (0);
900: }
901:
902: /*
903: * Concatenate fragments.
904: */
905: m = q;
906: t = m->m_next;
907: m->m_next = 0;
908: m_cat(m, t);
909: nq = q->m_nextpkt;
910: q->m_nextpkt = 0;
911: for (q = nq; q != NULL; q = nq) {
912: nq = q->m_nextpkt;
913: q->m_nextpkt = NULL;
914: m_cat(m, q);
915: }
916:
917: #if IPDIVERT
918: /*
919: * extract divert port for packet, if any
920: */
921: frag_divert_port = fp->ipq_divert;
922: ip_divert_cookie = fp->ipq_div_cookie;
923: #endif
924:
925: /*
926: * Create header for new ip packet by
927: * modifying header of first packet;
928: * dequeue and discard fragment reassembly header.
929: * Make header visible.
930: */
931: ip->ip_len = next;
932: ip->ip_src = fp->ipq_src;
933: ip->ip_dst = fp->ipq_dst;
934: remque((void *) fp);
935: nipq--;
936: (void) m_free(dtom(fp));
937: m->m_len += (IP_VHL_HL(ip->ip_vhl) << 2);
938: m->m_data -= (IP_VHL_HL(ip->ip_vhl) << 2);
939: /* some debugging cruft by sklower, below, will go away soon */
940: if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
941: register int plen = 0;
942: for (t = m; m; m = m->m_next)
943: plen += m->m_len;
944: t->m_pkthdr.len = plen;
945: }
946: return (ip);
947:
948: dropfrag:
949: #if IPDIVERT
950: frag_divert_port = 0;
951: ip_divert_cookie = 0;
952: #endif
953: ipstat.ips_fragdropped++;
954: m_freem(m);
955: return (0);
956:
957: #undef GETIP
958: }
959:
960: /*
961: * Free a fragment reassembly header and all
962: * associated datagrams.
963: */
964: static void
965: ip_freef(fp)
966: struct ipq *fp;
967: {
968: register struct mbuf *q;
969:
970: while (fp->ipq_frags) {
971: q = fp->ipq_frags;
972: fp->ipq_frags = q->m_nextpkt;
973: m_freem(q);
974: }
975: remque((void *) fp);
976: (void) m_free(dtom(fp));
977: nipq--;
978: }
979:
980: /*
981: * IP timer processing;
982: * if a timer expires on a reassembly
983: * queue, discard it.
984: */
985: void
986: ip_slowtimo()
987: {
988: register struct ipq *fp;
989: int s = splnet();
990: int i;
991:
992: for (i = 0; i < IPREASS_NHASH; i++) {
993: fp = ipq[i].next;
994: if (fp == 0)
995: continue;
996: while (fp != &ipq[i]) {
997: --fp->ipq_ttl;
998: fp = fp->next;
999: if (fp->prev->ipq_ttl == 0) {
1000: ipstat.ips_fragtimeout++;
1001: ip_freef(fp->prev);
1002: }
1003: }
1004: }
1005: ipflow_slowtimo();
1006: splx(s);
1007: }
1008:
1009: /*
1010: * Drain off all datagram fragments.
1011: */
1012: void
1013: ip_drain()
1014: {
1015: int i;
1016:
1017: for (i = 0; i < IPREASS_NHASH; i++) {
1018: while (ipq[i].next != &ipq[i]) {
1019: ipstat.ips_fragdropped++;
1020: ip_freef(ipq[i].next);
1021: }
1022: }
1023: in_rtqdrain();
1024: }
1025:
1026: /*
1027: * Do option processing on a datagram,
1028: * possibly discarding it if bad options are encountered,
1029: * or forwarding it if source-routed.
1030: * Returns 1 if packet has been forwarded/freed,
1031: * 0 if the packet should be processed further.
1032: */
1033: static int
1034: ip_dooptions(m)
1035: struct mbuf *m;
1036: {
1037: register struct ip *ip = mtod(m, struct ip *);
1038: register u_char *cp;
1039: register struct ip_timestamp *ipt;
1040: register struct in_ifaddr *ia;
1041: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1042: struct in_addr *sin, dst;
1043: n_time ntime;
1044:
1045: dst = ip->ip_dst;
1046: cp = (u_char *)(ip + 1);
1047: cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip);
1048: for (; cnt > 0; cnt -= optlen, cp += optlen) {
1049: opt = cp[IPOPT_OPTVAL];
1050: if (opt == IPOPT_EOL)
1051: break;
1052: if (opt == IPOPT_NOP)
1053: optlen = 1;
1054: else {
1055: optlen = cp[IPOPT_OLEN];
1056: if (optlen <= 0 || optlen > cnt) {
1057: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1058: goto bad;
1059: }
1060: }
1061: switch (opt) {
1062:
1063: default:
1064: break;
1065:
1066: /*
1067: * Source routing with record.
1068: * Find interface with current destination address.
1069: * If none on this machine then drop if strictly routed,
1070: * or do nothing if loosely routed.
1071: * Record interface address and bring up next address
1072: * component. If strictly routed make sure next
1073: * address is on directly accessible net.
1074: */
1075: case IPOPT_LSRR:
1076: case IPOPT_SSRR:
1077: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1078: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1079: goto bad;
1080: }
1081: ipaddr.sin_addr = ip->ip_dst;
1082: ia = (struct in_ifaddr *)
1083: ifa_ifwithaddr((struct sockaddr *)&ipaddr);
1084: if (ia == 0) {
1085: if (opt == IPOPT_SSRR) {
1086: type = ICMP_UNREACH;
1087: code = ICMP_UNREACH_SRCFAIL;
1088: goto bad;
1089: }
1090: if (!ip_dosourceroute)
1091: goto nosourcerouting;
1092: /*
1093: * Loose routing, and not at next destination
1094: * yet; nothing to do except forward.
1095: */
1096: break;
1097: }
1098: off--; /* 0 origin */
1099: if (off > optlen - sizeof(struct in_addr)) {
1100: /*
1101: * End of source route. Should be for us.
1102: */
1103: if (!ip_acceptsourceroute)
1104: goto nosourcerouting;
1105: save_rte(cp, ip->ip_src);
1106: break;
1107: }
1108:
1109: if (!ip_dosourceroute) {
1110: if (ipforwarding) {
1111: char buf[16]; /* aaa.bbb.ccc.ddd\0 */
1112: /*
1113: * Acting as a router, so generate ICMP
1114: */
1115: nosourcerouting:
1116: strcpy(buf, inet_ntoa(ip->ip_dst));
1117: log(LOG_WARNING,
1118: "attempted source route from %s to %s\n",
1119: inet_ntoa(ip->ip_src), buf);
1120: type = ICMP_UNREACH;
1121: code = ICMP_UNREACH_SRCFAIL;
1122: goto bad;
1123: } else {
1124: /*
1125: * Not acting as a router, so silently drop.
1126: */
1127: ipstat.ips_cantforward++;
1128: m_freem(m);
1129: return (1);
1130: }
1131: }
1132:
1133: /*
1134: * locate outgoing interface
1135: */
1136: (void)memcpy(&ipaddr.sin_addr, cp + off,
1137: sizeof(ipaddr.sin_addr));
1138:
1139: if (opt == IPOPT_SSRR) {
1140: #define INA struct in_ifaddr *
1141: #define SA struct sockaddr *
1142: if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == 0)
1143: ia = (INA)ifa_ifwithnet((SA)&ipaddr);
1144: } else
1145: ia = ip_rtaddr(ipaddr.sin_addr);
1146: if (ia == 0) {
1147: type = ICMP_UNREACH;
1148: code = ICMP_UNREACH_SRCFAIL;
1149: goto bad;
1150: }
1151: ip->ip_dst = ipaddr.sin_addr;
1152: (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
1153: sizeof(struct in_addr));
1154: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1155: /*
1156: * Let ip_intr's mcast routing check handle mcast pkts
1157: */
1158: forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr));
1159: break;
1160:
1161: case IPOPT_RR:
1162: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1163: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1164: goto bad;
1165: }
1166: /*
1167: * If no space remains, ignore.
1168: */
1169: off--; /* 0 origin */
1170: if (off > optlen - sizeof(struct in_addr))
1171: break;
1172: (void)memcpy(&ipaddr.sin_addr, &ip->ip_dst,
1173: sizeof(ipaddr.sin_addr));
1174: /*
1175: * locate outgoing interface; if we're the destination,
1176: * use the incoming interface (should be same).
1177: */
1178: if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
1179: (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
1180: type = ICMP_UNREACH;
1181: code = ICMP_UNREACH_HOST;
1182: goto bad;
1183: }
1184: (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
1185: sizeof(struct in_addr));
1186: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1187: break;
1188:
1189: case IPOPT_TS:
1190: code = cp - (u_char *)ip;
1191: ipt = (struct ip_timestamp *)cp;
1192: if (ipt->ipt_len < 5)
1193: goto bad;
1194: if (ipt->ipt_ptr > ipt->ipt_len - sizeof(int32_t)) {
1195: if (++ipt->ipt_oflw == 0)
1196: goto bad;
1197: break;
1198: }
1199: sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
1200: switch (ipt->ipt_flg) {
1201:
1202: case IPOPT_TS_TSONLY:
1203: break;
1204:
1205: case IPOPT_TS_TSANDADDR:
1206: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1207: sizeof(struct in_addr) > ipt->ipt_len)
1208: goto bad;
1209: ipaddr.sin_addr = dst;
1210: ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
1211: m->m_pkthdr.rcvif);
1212: if (ia == 0)
1213: continue;
1214: (void)memcpy(sin, &IA_SIN(ia)->sin_addr,
1215: sizeof(struct in_addr));
1216: ipt->ipt_ptr += sizeof(struct in_addr);
1217: break;
1218:
1219: case IPOPT_TS_PRESPEC:
1220: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1221: sizeof(struct in_addr) > ipt->ipt_len)
1222: goto bad;
1223: (void)memcpy(&ipaddr.sin_addr, sin,
1224: sizeof(struct in_addr));
1225: if (ifa_ifwithaddr((SA)&ipaddr) == 0)
1226: continue;
1227: ipt->ipt_ptr += sizeof(struct in_addr);
1228: break;
1229:
1230: default:
1231: goto bad;
1232: }
1233: ntime = iptime();
1234: (void)memcpy(cp + ipt->ipt_ptr - 1, &ntime,
1235: sizeof(n_time));
1236: ipt->ipt_ptr += sizeof(n_time);
1237: }
1238: }
1239: if (forward && ipforwarding) {
1240: ip_forward(m, 1);
1241: return (1);
1242: }
1243: return (0);
1244: bad:
1245: ip->ip_len -= IP_VHL_HL(ip->ip_vhl) << 2; /* XXX icmp_error adds in hdr length */
1246: icmp_error(m, type, code, 0, 0);
1247: ipstat.ips_badoptions++;
1248: return (1);
1249: }
1250:
1251: /*
1252: * Given address of next destination (final or next hop),
1253: * return internet address info of interface to be used to get there.
1254: */
1255: static struct in_ifaddr *
1256: ip_rtaddr(dst)
1257: struct in_addr dst;
1258: {
1259: register struct sockaddr_in *sin;
1260:
1261: sin = (struct sockaddr_in *) &ipforward_rt.ro_dst;
1262:
1263: if (ipforward_rt.ro_rt == 0 || dst.s_addr != sin->sin_addr.s_addr) {
1264: if (ipforward_rt.ro_rt) {
1265: RTFREE(ipforward_rt.ro_rt);
1266: ipforward_rt.ro_rt = 0;
1267: }
1268: sin->sin_family = AF_INET;
1269: sin->sin_len = sizeof(*sin);
1270: sin->sin_addr = dst;
1271:
1272: rtalloc_ign(&ipforward_rt, RTF_PRCLONING);
1273: }
1274: if (ipforward_rt.ro_rt == 0)
1275: return ((struct in_ifaddr *)0);
1276: return ((struct in_ifaddr *) ipforward_rt.ro_rt->rt_ifa);
1277: }
1278:
1279: /*
1280: * Save incoming source route for use in replies,
1281: * to be picked up later by ip_srcroute if the receiver is interested.
1282: */
1283: void
1284: save_rte(option, dst)
1285: u_char *option;
1286: struct in_addr dst;
1287: {
1288: unsigned olen;
1289:
1290: olen = option[IPOPT_OLEN];
1291: #if DIAGNOSTIC
1292: if (ipprintfs)
1293: printf("save_rte: olen %d\n", olen);
1294: #endif
1295: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1296: return;
1297: bcopy(option, ip_srcrt.srcopt, olen);
1298: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1299: ip_srcrt.dst = dst;
1300: }
1301:
1302: /*
1303: * Retrieve incoming source route for use in replies,
1304: * in the same form used by setsockopt.
1305: * The first hop is placed before the options, will be removed later.
1306: */
1307: struct mbuf *
1308: ip_srcroute()
1309: {
1310: register struct in_addr *p, *q;
1311: register struct mbuf *m;
1312:
1313: if (ip_nhops == 0)
1314: return ((struct mbuf *)0);
1315: m = m_get(M_DONTWAIT, MT_HEADER);
1316: if (m == 0)
1317: return ((struct mbuf *)0);
1318:
1319: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1320:
1321: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1322: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1323: OPTSIZ;
1324: #if DIAGNOSTIC
1325: if (ipprintfs)
1326: printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1327: #endif
1328:
1329: /*
1330: * First save first hop for return route
1331: */
1332: p = &ip_srcrt.route[ip_nhops - 1];
1333: *(mtod(m, struct in_addr *)) = *p--;
1334: #if DIAGNOSTIC
1335: if (ipprintfs)
1336: printf(" hops %lx", (u_long)ntohl(mtod(m, struct in_addr *)->s_addr));
1337: #endif
1338:
1339: /*
1340: * Copy option fields and padding (nop) to mbuf.
1341: */
1342: ip_srcrt.nop = IPOPT_NOP;
1343: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1344: (void)memcpy(mtod(m, caddr_t) + sizeof(struct in_addr),
1345: &ip_srcrt.nop, OPTSIZ);
1346: q = (struct in_addr *)(mtod(m, caddr_t) +
1347: sizeof(struct in_addr) + OPTSIZ);
1348: #undef OPTSIZ
1349: /*
1350: * Record return path as an IP source route,
1351: * reversing the path (pointers are now aligned).
1352: */
1353: while (p >= ip_srcrt.route) {
1354: #if DIAGNOSTIC
1355: if (ipprintfs)
1356: printf(" %lx", (u_long)ntohl(q->s_addr));
1357: #endif
1358: *q++ = *p--;
1359: }
1360: /*
1361: * Last hop goes to final destination.
1362: */
1363: *q = ip_srcrt.dst;
1364: #if DIAGNOSTIC
1365: if (ipprintfs)
1366: printf(" %lx\n", (u_long)ntohl(q->s_addr));
1367: #endif
1368: return (m);
1369: }
1370:
1371: /*
1372: * Strip out IP options, at higher
1373: * level protocol in the kernel.
1374: * Second argument is buffer to which options
1375: * will be moved, and return value is their length.
1376: * XXX should be deleted; last arg currently ignored.
1377: */
1378: void
1379: ip_stripoptions(m, mopt)
1380: register struct mbuf *m;
1381: struct mbuf *mopt;
1382: {
1383: register int i;
1384: struct ip *ip = mtod(m, struct ip *);
1385: register caddr_t opts;
1386: int olen;
1387:
1388: olen = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip);
1389: opts = (caddr_t)(ip + 1);
1390: i = m->m_len - (sizeof (struct ip) + olen);
1391: bcopy(opts + olen, opts, (unsigned)i);
1392: m->m_len -= olen;
1393: if (m->m_flags & M_PKTHDR)
1394: m->m_pkthdr.len -= olen;
1395: ip->ip_vhl = IP_MAKE_VHL(IPVERSION, sizeof(struct ip) >> 2);
1396: }
1397:
1398: u_char inetctlerrmap[PRC_NCMDS] = {
1399: 0, 0, 0, 0,
1400: 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1401: EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1402: EMSGSIZE, EHOSTUNREACH, 0, 0,
1403: 0, 0, 0, 0,
1404: ENOPROTOOPT
1405: };
1406:
1407: /*
1408: * Forward a packet. If some error occurs return the sender
1409: * an icmp packet. Note we can't always generate a meaningful
1410: * icmp message because icmp doesn't have a large enough repertoire
1411: * of codes and types.
1412: *
1413: * If not forwarding, just drop the packet. This could be confusing
1414: * if ipforwarding was zero but some routing protocol was advancing
1415: * us as a gateway to somewhere. However, we must let the routing
1416: * protocol deal with that.
1417: *
1418: * The srcrt parameter indicates whether the packet is being forwarded
1419: * via a source route.
1420: */
1421: static void
1422: ip_forward(m, srcrt)
1423: struct mbuf *m;
1424: int srcrt;
1425: {
1426: register struct ip *ip = mtod(m, struct ip *);
1427: register struct sockaddr_in *sin;
1428: register struct rtentry *rt;
1429: int error, type = 0, code = 0;
1430: struct mbuf *mcopy;
1431: n_long dest;
1432: struct ifnet *destifp;
1433:
1434: dest = 0;
1435: #if DIAGNOSTIC
1436: if (ipprintfs)
1437: printf("forward: src %lx dst %lx ttl %x\n",
1438: (u_long)ip->ip_src.s_addr, (u_long)ip->ip_dst.s_addr,
1439: ip->ip_ttl);
1440: #endif
1441:
1442:
1443: if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1444: ipstat.ips_cantforward++;
1445: m_freem(m);
1446: return;
1447: }
1448: HTONS(ip->ip_id);
1449: if (ip->ip_ttl <= IPTTLDEC) {
1450: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1451: return;
1452: }
1453: ip->ip_ttl -= IPTTLDEC;
1454:
1455: sin = (struct sockaddr_in *)&ipforward_rt.ro_dst;
1456: if ((rt = ipforward_rt.ro_rt) == 0 ||
1457: ip->ip_dst.s_addr != sin->sin_addr.s_addr) {
1458: if (ipforward_rt.ro_rt) {
1459: RTFREE(ipforward_rt.ro_rt);
1460: ipforward_rt.ro_rt = 0;
1461: }
1462: sin->sin_family = AF_INET;
1463: sin->sin_len = sizeof(*sin);
1464: sin->sin_addr = ip->ip_dst;
1465:
1466: rtalloc_ign(&ipforward_rt, RTF_PRCLONING);
1467: if (ipforward_rt.ro_rt == 0) {
1468: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1469: return;
1470: }
1471: rt = ipforward_rt.ro_rt;
1472: }
1473:
1474: /*
1475: * Save at most 64 bytes of the packet in case
1476: * we need to generate an ICMP message to the src.
1477: */
1478: mcopy = m_copy(m, 0, imin((int)ip->ip_len, 64));
1479:
1480: /*
1481: * If forwarding packet using same interface that it came in on,
1482: * perhaps should send a redirect to sender to shortcut a hop.
1483: * Only send redirect if source is sending directly to us,
1484: * and if packet was not source routed (or has any options).
1485: * Also, don't send redirect if forwarding using a default route
1486: * or a route modified by a redirect.
1487: */
1488: #define satosin(sa) ((struct sockaddr_in *)(sa))
1489: if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1490: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1491: satosin(rt_key(rt))->sin_addr.s_addr != 0 &&
1492: ipsendredirects && !srcrt) {
1493: #define RTA(rt) ((struct in_ifaddr *)(rt->rt_ifa))
1494: u_long src = ntohl(ip->ip_src.s_addr);
1495:
1496: if (RTA(rt) &&
1497: (src & RTA(rt)->ia_subnetmask) == RTA(rt)->ia_subnet) {
1498: if (rt->rt_flags & RTF_GATEWAY)
1499: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1500: else
1501: dest = ip->ip_dst.s_addr;
1502: /* Router requirements says to only send host redirects */
1503: type = ICMP_REDIRECT;
1504: code = ICMP_REDIRECT_HOST;
1505: #if DIAGNOSTIC
1506: if (ipprintfs)
1507: printf("redirect (%d) to %lx\n", code, (u_long)dest);
1508: #endif
1509: }
1510: }
1511:
1512: error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1513: IP_FORWARDING, 0);
1514: if (error)
1515: ipstat.ips_cantforward++;
1516: else {
1517: ipstat.ips_forward++;
1518: if (type)
1519: ipstat.ips_redirectsent++;
1520: else {
1521: if (mcopy) {
1522: ipflow_create(&ipforward_rt, mcopy);
1523: m_freem(mcopy);
1524: }
1525: return;
1526: }
1527: }
1528: if (mcopy == NULL)
1529: return;
1530: destifp = NULL;
1531:
1532: switch (error) {
1533:
1534: case 0: /* forwarded, but need redirect */
1535: /* type, code set above */
1536: break;
1537:
1538: case ENETUNREACH: /* shouldn't happen, checked above */
1539: case EHOSTUNREACH:
1540: case ENETDOWN:
1541: case EHOSTDOWN:
1542: default:
1543: type = ICMP_UNREACH;
1544: code = ICMP_UNREACH_HOST;
1545: break;
1546:
1547: case EMSGSIZE:
1548: type = ICMP_UNREACH;
1549: code = ICMP_UNREACH_NEEDFRAG;
1550: if (ipforward_rt.ro_rt)
1551: destifp = ipforward_rt.ro_rt->rt_ifp;
1552: ipstat.ips_cantfrag++;
1553: break;
1554:
1555: case ENOBUFS:
1556: type = ICMP_SOURCEQUENCH;
1557: code = 0;
1558: break;
1559: }
1560: icmp_error(mcopy, type, code, dest, destifp);
1561: }
1562:
1563: void
1564: ip_savecontrol(inp, mp, ip, m)
1565: register struct inpcb *inp;
1566: register struct mbuf **mp;
1567: register struct ip *ip;
1568: register struct mbuf *m;
1569: {
1570: if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1571: struct timeval tv;
1572:
1573: microtime(&tv);
1574: *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1575: SCM_TIMESTAMP, SOL_SOCKET);
1576: if (*mp)
1577: mp = &(*mp)->m_next;
1578: }
1579: if (inp->inp_flags & INP_RECVDSTADDR) {
1580: *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1581: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1582: if (*mp)
1583: mp = &(*mp)->m_next;
1584: }
1585: #ifdef notyet
1586: /* XXX
1587: * Moving these out of udp_input() made them even more broken
1588: * than they already were.
1589: */
1590: /* options were tossed already */
1591: if (inp->inp_flags & INP_RECVOPTS) {
1592: *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1593: sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1594: if (*mp)
1595: mp = &(*mp)->m_next;
1596: }
1597: /* ip_srcroute doesn't do what we want here, need to fix */
1598: if (inp->inp_flags & INP_RECVRETOPTS) {
1599: *mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1600: sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1601: if (*mp)
1602: mp = &(*mp)->m_next;
1603: }
1604: #endif
1605: if (inp->inp_flags & INP_RECVIF) {
1606: struct ifnet *ifp;
1607: struct sdlbuf {
1608: struct sockaddr_dl sdl;
1609: u_char pad[32];
1610: } sdlbuf;
1611: struct sockaddr_dl *sdp;
1612: struct sockaddr_dl *sdl2 = &sdlbuf.sdl;
1613:
1614: if (((ifp = m->m_pkthdr.rcvif))
1615: && ( ifp->if_index && (ifp->if_index <= if_index))) {
1616: sdp = (struct sockaddr_dl *)(ifnet_addrs
1617: [ifp->if_index - 1]->ifa_addr);
1618: /*
1619: * Change our mind and don't try copy.
1620: */
1621: if ((sdp->sdl_family != AF_LINK)
1622: || (sdp->sdl_len > sizeof(sdlbuf))) {
1623: goto makedummy;
1624: }
1625: bcopy(sdp, sdl2, sdp->sdl_len);
1626: } else {
1627: makedummy:
1628: sdl2->sdl_len
1629: = offsetof(struct sockaddr_dl, sdl_data[0]);
1630: sdl2->sdl_family = AF_LINK;
1631: sdl2->sdl_index = 0;
1632: sdl2->sdl_nlen = sdl2->sdl_alen = sdl2->sdl_slen = 0;
1633: }
1634: *mp = sbcreatecontrol((caddr_t) sdl2, sdl2->sdl_len,
1635: IP_RECVIF, IPPROTO_IP);
1636: if (*mp)
1637: mp = &(*mp)->m_next;
1638: }
1639: }
1640:
1641: int
1642: ip_rsvp_init(struct socket *so)
1643: {
1644: if (so->so_type != SOCK_RAW ||
1645: so->so_proto->pr_protocol != IPPROTO_RSVP)
1646: return EOPNOTSUPP;
1647:
1648: if (ip_rsvpd != NULL)
1649: return EADDRINUSE;
1650:
1651: ip_rsvpd = so;
1652: /*
1653: * This may seem silly, but we need to be sure we don't over-increment
1654: * the RSVP counter, in case something slips up.
1655: */
1656: if (!ip_rsvp_on) {
1657: ip_rsvp_on = 1;
1658: rsvp_on++;
1659: }
1660:
1661: return 0;
1662: }
1663:
1664: int
1665: ip_rsvp_done(void)
1666: {
1667: ip_rsvpd = NULL;
1668: /*
1669: * This may seem silly, but we need to be sure we don't over-decrement
1670: * the RSVP counter, in case something slips up.
1671: */
1672: if (ip_rsvp_on) {
1673: ip_rsvp_on = 0;
1674: rsvp_on--;
1675: }
1676: return 0;
1677: }
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.