2 * Copyright (c) 1982, 1986, 1988, 1993
3 * The Regents of the University of California.
4 * Copyright (c) 2005 Andre Oppermann, Internet Business Solutions AG.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
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 * 4. Neither the name of the University nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 #include "opt_ipstealth.h"
37 #include <sys/param.h>
38 #include <sys/systm.h>
40 #include <sys/domain.h>
41 #include <sys/protosw.h>
42 #include <sys/socket.h>
44 #include <sys/kernel.h>
45 #include <sys/syslog.h>
46 #include <sys/sysctl.h>
49 #include <net/if_types.h>
50 #include <net/if_var.h>
51 #include <net/if_dl.h>
52 #include <net/route.h>
53 #include <net/netisr.h>
55 #include <netinet/in.h>
56 #include <netinet/in_systm.h>
57 #include <netinet/in_var.h>
58 #include <netinet/ip.h>
59 #include <netinet/in_pcb.h>
60 #include <netinet/ip_var.h>
61 #include <netinet/ip_options.h>
62 #include <netinet/ip_icmp.h>
63 #include <machine/in_cksum.h>
65 #include <sys/socketvar.h>
67 #include <security/mac/mac_framework.h>
69 static int ip_dosourceroute = 0;
70 SYSCTL_INT(_net_inet_ip, IPCTL_SOURCEROUTE, sourceroute, CTLFLAG_RW,
71 &ip_dosourceroute, 0, "Enable forwarding source routed IP packets");
73 static int ip_acceptsourceroute = 0;
74 SYSCTL_INT(_net_inet_ip, IPCTL_ACCEPTSOURCEROUTE, accept_sourceroute,
75 CTLFLAG_RW, &ip_acceptsourceroute, 0,
76 "Enable accepting source routed IP packets");
78 int ip_doopts = 1; /* 0 = ignore, 1 = process, 2 = reject */
79 SYSCTL_INT(_net_inet_ip, OID_AUTO, process_options, CTLFLAG_RW,
80 &ip_doopts, 0, "Enable IP options processing ([LS]SRR, RR, TS)");
82 static void save_rte(struct mbuf *m, u_char *, struct in_addr);
85 * Do option processing on a datagram, possibly discarding it if bad options
86 * are encountered, or forwarding it if source-routed.
88 * The pass argument is used when operating in the IPSTEALTH mode to tell
89 * what options to process: [LS]SRR (pass 0) or the others (pass 1). The
90 * reason for as many as two passes is that when doing IPSTEALTH, non-routing
91 * options should be processed only if the packet is for us.
93 * Returns 1 if packet has been forwarded/freed, 0 if the packet should be
97 ip_dooptions(struct mbuf *m, int pass)
99 struct ip *ip = mtod(m, struct ip *);
101 struct in_ifaddr *ia;
102 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
103 struct in_addr *sin, dst;
105 struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
107 /* Ignore or reject packets with IP options. */
110 else if (ip_doopts == 2) {
112 code = ICMP_UNREACH_FILTER_PROHIB;
117 cp = (u_char *)(ip + 1);
118 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
119 for (; cnt > 0; cnt -= optlen, cp += optlen) {
120 opt = cp[IPOPT_OPTVAL];
121 if (opt == IPOPT_EOL)
123 if (opt == IPOPT_NOP)
126 if (cnt < IPOPT_OLEN + sizeof(*cp)) {
127 code = &cp[IPOPT_OLEN] - (u_char *)ip;
130 optlen = cp[IPOPT_OLEN];
131 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
132 code = &cp[IPOPT_OLEN] - (u_char *)ip;
142 * Source routing with record. Find interface with current
143 * destination address. If none on this machine then drop if
144 * strictly routed, or do nothing if loosely routed. Record
145 * interface address and bring up next address component. If
146 * strictly routed make sure next address is on directly
152 if (ipstealth && pass > 0)
155 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
156 code = &cp[IPOPT_OLEN] - (u_char *)ip;
159 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
160 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
163 ipaddr.sin_addr = ip->ip_dst;
164 ia = (struct in_ifaddr *)
165 ifa_ifwithaddr((struct sockaddr *)&ipaddr);
167 if (opt == IPOPT_SSRR) {
169 code = ICMP_UNREACH_SRCFAIL;
172 if (!ip_dosourceroute)
173 goto nosourcerouting;
175 * Loose routing, and not at next destination
176 * yet; nothing to do except forward.
180 off--; /* 0 origin */
181 if (off > optlen - (int)sizeof(struct in_addr)) {
183 * End of source route. Should be for us.
185 if (!ip_acceptsourceroute)
186 goto nosourcerouting;
187 save_rte(m, cp, ip->ip_src);
194 if (!ip_dosourceroute) {
196 char buf[16]; /* aaa.bbb.ccc.ddd\0 */
198 * Acting as a router, so generate
202 strcpy(buf, inet_ntoa(ip->ip_dst));
204 "attempted source route from %s to %s\n",
205 inet_ntoa(ip->ip_src), buf);
207 code = ICMP_UNREACH_SRCFAIL;
211 * Not acting as a router, so
217 ipstat.ips_cantforward++;
224 * locate outgoing interface
226 (void)memcpy(&ipaddr.sin_addr, cp + off,
227 sizeof(ipaddr.sin_addr));
229 if (opt == IPOPT_SSRR) {
230 #define INA struct in_ifaddr *
231 #define SA struct sockaddr *
232 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == NULL)
233 ia = (INA)ifa_ifwithnet((SA)&ipaddr);
235 ia = ip_rtaddr(ipaddr.sin_addr);
238 code = ICMP_UNREACH_SRCFAIL;
241 ip->ip_dst = ipaddr.sin_addr;
242 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
243 sizeof(struct in_addr));
244 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
246 * Let ip_intr's mcast routing check handle mcast pkts
248 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr));
253 if (ipstealth && pass == 0)
256 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
257 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
260 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
261 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
265 * If no space remains, ignore.
267 off--; /* 0 origin */
268 if (off > optlen - (int)sizeof(struct in_addr))
270 (void)memcpy(&ipaddr.sin_addr, &ip->ip_dst,
271 sizeof(ipaddr.sin_addr));
273 * Locate outgoing interface; if we're the
274 * destination, use the incoming interface (should be
277 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == NULL &&
278 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
280 code = ICMP_UNREACH_HOST;
283 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
284 sizeof(struct in_addr));
285 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
290 if (ipstealth && pass == 0)
293 code = cp - (u_char *)ip;
294 if (optlen < 4 || optlen > 40) {
295 code = &cp[IPOPT_OLEN] - (u_char *)ip;
298 if ((off = cp[IPOPT_OFFSET]) < 5) {
299 code = &cp[IPOPT_OLEN] - (u_char *)ip;
302 if (off > optlen - (int)sizeof(int32_t)) {
303 cp[IPOPT_OFFSET + 1] += (1 << 4);
304 if ((cp[IPOPT_OFFSET + 1] & 0xf0) == 0) {
305 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
310 off--; /* 0 origin */
311 sin = (struct in_addr *)(cp + off);
312 switch (cp[IPOPT_OFFSET + 1] & 0x0f) {
314 case IPOPT_TS_TSONLY:
317 case IPOPT_TS_TSANDADDR:
318 if (off + sizeof(n_time) +
319 sizeof(struct in_addr) > optlen) {
320 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
323 ipaddr.sin_addr = dst;
324 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
328 (void)memcpy(sin, &IA_SIN(ia)->sin_addr,
329 sizeof(struct in_addr));
330 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
331 off += sizeof(struct in_addr);
334 case IPOPT_TS_PRESPEC:
335 if (off + sizeof(n_time) +
336 sizeof(struct in_addr) > optlen) {
337 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
340 (void)memcpy(&ipaddr.sin_addr, sin,
341 sizeof(struct in_addr));
342 if (ifa_ifwithaddr((SA)&ipaddr) == NULL)
344 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
345 off += sizeof(struct in_addr);
349 code = &cp[IPOPT_OFFSET + 1] - (u_char *)ip;
353 (void)memcpy(cp + off, &ntime, sizeof(n_time));
354 cp[IPOPT_OFFSET] += sizeof(n_time);
357 if (forward && ipforwarding) {
363 icmp_error(m, type, code, 0, 0);
364 ipstat.ips_badoptions++;
369 * Save incoming source route for use in replies, to be picked up later by
370 * ip_srcroute if the receiver is interested.
373 save_rte(struct mbuf *m, u_char *option, struct in_addr dst)
376 struct ipopt_tag *opts;
378 opts = (struct ipopt_tag *)m_tag_get(PACKET_TAG_IPOPTIONS,
379 sizeof(struct ipopt_tag), M_NOWAIT);
383 olen = option[IPOPT_OLEN];
384 if (olen > sizeof(opts->ip_srcrt) - (1 + sizeof(dst))) {
385 m_tag_free((struct m_tag *)opts);
388 bcopy(option, opts->ip_srcrt.srcopt, olen);
389 opts->ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
390 opts->ip_srcrt.dst = dst;
391 m_tag_prepend(m, (struct m_tag *)opts);
395 * Retrieve incoming source route for use in replies, in the same form used
396 * by setsockopt. The first hop is placed before the options, will be
400 ip_srcroute(struct mbuf *m0)
402 struct in_addr *p, *q;
404 struct ipopt_tag *opts;
406 opts = (struct ipopt_tag *)m_tag_find(m0, PACKET_TAG_IPOPTIONS, NULL);
410 if (opts->ip_nhops == 0)
412 m = m_get(M_DONTWAIT, MT_DATA);
416 #define OPTSIZ (sizeof(opts->ip_srcrt.nop) + sizeof(opts->ip_srcrt.srcopt))
418 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
419 m->m_len = opts->ip_nhops * sizeof(struct in_addr) +
420 sizeof(struct in_addr) + OPTSIZ;
423 * First, save first hop for return route.
425 p = &(opts->ip_srcrt.route[opts->ip_nhops - 1]);
426 *(mtod(m, struct in_addr *)) = *p--;
429 * Copy option fields and padding (nop) to mbuf.
431 opts->ip_srcrt.nop = IPOPT_NOP;
432 opts->ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
433 (void)memcpy(mtod(m, caddr_t) + sizeof(struct in_addr),
434 &(opts->ip_srcrt.nop), OPTSIZ);
435 q = (struct in_addr *)(mtod(m, caddr_t) +
436 sizeof(struct in_addr) + OPTSIZ);
439 * Record return path as an IP source route, reversing the path
440 * (pointers are now aligned).
442 while (p >= opts->ip_srcrt.route) {
446 * Last hop goes to final destination.
448 *q = opts->ip_srcrt.dst;
449 m_tag_delete(m0, (struct m_tag *)opts);
454 * Strip out IP options, at higher level protocol in the kernel. Second
455 * argument is buffer to which options will be moved, and return value is
458 * XXX should be deleted; last arg currently ignored.
461 ip_stripoptions(struct mbuf *m, struct mbuf *mopt)
464 struct ip *ip = mtod(m, struct ip *);
468 olen = (ip->ip_hl << 2) - sizeof (struct ip);
469 opts = (caddr_t)(ip + 1);
470 i = m->m_len - (sizeof (struct ip) + olen);
471 bcopy(opts + olen, opts, (unsigned)i);
473 if (m->m_flags & M_PKTHDR)
474 m->m_pkthdr.len -= olen;
475 ip->ip_v = IPVERSION;
476 ip->ip_hl = sizeof(struct ip) >> 2;
480 * Insert IP options into preformed packet. Adjust IP destination as
481 * required for IP source routing, as indicated by a non-zero in_addr at the
482 * start of the options.
484 * XXX This routine assumes that the packet has no options in place.
487 ip_insertoptions(struct mbuf *m, struct mbuf *opt, int *phlen)
489 struct ipoption *p = mtod(opt, struct ipoption *);
491 struct ip *ip = mtod(m, struct ip *);
494 optlen = opt->m_len - sizeof(p->ipopt_dst);
495 if (optlen + ip->ip_len > IP_MAXPACKET) {
497 return (m); /* XXX should fail */
499 if (p->ipopt_dst.s_addr)
500 ip->ip_dst = p->ipopt_dst;
501 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) {
502 MGETHDR(n, M_DONTWAIT, MT_DATA);
508 n->m_pkthdr.rcvif = NULL;
512 n->m_pkthdr.len += optlen;
513 m->m_len -= sizeof(struct ip);
514 m->m_data += sizeof(struct ip);
517 m->m_len = optlen + sizeof(struct ip);
518 m->m_data += max_linkhdr;
519 bcopy(ip, mtod(m, void *), sizeof(struct ip));
523 m->m_pkthdr.len += optlen;
524 bcopy(ip, mtod(m, void *), sizeof(struct ip));
526 ip = mtod(m, struct ip *);
527 bcopy(p->ipopt_list, ip + 1, optlen);
528 *phlen = sizeof(struct ip) + optlen;
529 ip->ip_v = IPVERSION;
530 ip->ip_hl = *phlen >> 2;
531 ip->ip_len += optlen;
536 * Copy options from ip to jp, omitting those not copied during
540 ip_optcopy(struct ip *ip, struct ip *jp)
543 int opt, optlen, cnt;
545 cp = (u_char *)(ip + 1);
546 dp = (u_char *)(jp + 1);
547 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
548 for (; cnt > 0; cnt -= optlen, cp += optlen) {
550 if (opt == IPOPT_EOL)
552 if (opt == IPOPT_NOP) {
553 /* Preserve for IP mcast tunnel's LSRR alignment. */
559 KASSERT(cnt >= IPOPT_OLEN + sizeof(*cp),
560 ("ip_optcopy: malformed ipv4 option"));
561 optlen = cp[IPOPT_OLEN];
562 KASSERT(optlen >= IPOPT_OLEN + sizeof(*cp) && optlen <= cnt,
563 ("ip_optcopy: malformed ipv4 option"));
565 /* Bogus lengths should have been caught by ip_dooptions. */
568 if (IPOPT_COPIED(opt)) {
569 bcopy(cp, dp, optlen);
573 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++)
579 * Set up IP options in pcb for insertion in output packets. Store in mbuf
580 * with pointer in pcbopt, adding pseudo-option with destination address if
584 ip_pcbopts(struct inpcb *inp, int optname, struct mbuf *m)
588 struct mbuf **pcbopt;
591 INP_LOCK_ASSERT(inp);
593 pcbopt = &inp->inp_options;
595 /* turn off any old options */
597 (void)m_free(*pcbopt);
599 if (m == NULL || m->m_len == 0) {
601 * Only turning off any previous options.
608 if (m->m_len % sizeof(int32_t))
611 * IP first-hop destination address will be stored before actual
612 * options; move other options back and clear it when none present.
614 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN])
617 m->m_len += sizeof(struct in_addr);
618 cp = mtod(m, u_char *) + sizeof(struct in_addr);
619 bcopy(mtod(m, void *), cp, (unsigned)cnt);
620 bzero(mtod(m, void *), sizeof(struct in_addr));
622 for (; cnt > 0; cnt -= optlen, cp += optlen) {
623 opt = cp[IPOPT_OPTVAL];
624 if (opt == IPOPT_EOL)
626 if (opt == IPOPT_NOP)
629 if (cnt < IPOPT_OLEN + sizeof(*cp))
631 optlen = cp[IPOPT_OLEN];
632 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt)
643 * User process specifies route as:
647 * D must be our final destination (but we can't
648 * check that since we may not have connected yet).
649 * A is first hop destination, which doesn't appear
650 * in actual IP option, but is stored before the
653 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr))
655 m->m_len -= sizeof(struct in_addr);
656 cnt -= sizeof(struct in_addr);
657 optlen -= sizeof(struct in_addr);
658 cp[IPOPT_OLEN] = optlen;
660 * Move first hop before start of options.
662 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t),
663 sizeof(struct in_addr));
665 * Then copy rest of options back
666 * to close up the deleted entry.
668 bcopy((&cp[IPOPT_OFFSET+1] + sizeof(struct in_addr)),
670 (unsigned)cnt - (IPOPT_MINOFF - 1));
674 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr))