Context Navigation

source: rtems-libbsd/freebsd/sys/netinet6/in6_src.c @ 2017a6d

55-freebsd-126-freebsd-12freebsd-9.3

Last change on this file since 2017a6d was 2017a6d, checked in by Sebastian Huber <sebastian.huber@…>, on 04/07/16 at 07:48:12
Directly use <sys/time.h> provided by Newlib
Property mode set to `100644`
File size: 33.1 KB

Line
1	#include <machine/rtems-bsd-kernel-space.h>
2
3	/*-
4	* Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5	* All rights reserved.
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. Neither the name of the project 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.
18	*
19	* THIS SOFTWARE IS PROVIDED BY THE PROJECT 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 PROJECT 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
29	* SUCH DAMAGE.
30	*
31	* $KAME: in6_src.c,v 1.132 2003/08/26 04:42:27 keiichi Exp $
32	*/
33
34	/*-
35	* Copyright (c) 1982, 1986, 1991, 1993
36	* The Regents of the University of California. All rights reserved.
37	*
38	* Redistribution and use in source and binary forms, with or without
39	* modification, are permitted provided that the following conditions
40	* are met:
41	* 1. Redistributions of source code must retain the above copyright
42	* notice, this list of conditions and the following disclaimer.
43	* 2. Redistributions in binary form must reproduce the above copyright
44	* notice, this list of conditions and the following disclaimer in the
45	* documentation and/or other materials provided with the distribution.
46	* 4. Neither the name of the University nor the names of its contributors
47	* may be used to endorse or promote products derived from this software
48	* without specific prior written permission.
49	*
50	* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
51	* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
52	* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
53	* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
54	* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
55	* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
56	* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
57	* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
58	* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
59	* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
60	* SUCH DAMAGE.
61	*
62	* @(#)in_pcb.c 8.2 (Berkeley) 1/4/94
63	*/
64
65	#include <sys/cdefs.h>
66	__FBSDID("$FreeBSD$");
67
68	#include <rtems/bsd/local/opt_inet.h>
69	#include <rtems/bsd/local/opt_inet6.h>
70	#include <rtems/bsd/local/opt_mpath.h>
71
72	#include <rtems/bsd/sys/param.h>
73	#include <sys/systm.h>
74	#include <rtems/bsd/sys/lock.h>
75	#include <sys/malloc.h>
76	#include <sys/mbuf.h>
77	#include <sys/priv.h>
78	#include <sys/protosw.h>
79	#include <sys/socket.h>
80	#include <sys/socketvar.h>
81	#include <sys/sockio.h>
82	#include <sys/sysctl.h>
83	#include <rtems/bsd/sys/errno.h>
84	#include <sys/time.h>
85	#include <sys/jail.h>
86	#include <sys/kernel.h>
87	#include <sys/sx.h>
88
89	#include <net/if.h>
90	#include <net/if_dl.h>
91	#include <net/route.h>
92	#include <net/if_llatbl.h>
93	#ifdef RADIX_MPATH
94	#include <net/radix_mpath.h>
95	#endif
96
97	#include <netinet/in.h>
98	#include <netinet/in_var.h>
99	#include <netinet/in_systm.h>
100	#include <netinet/ip.h>
101	#include <netinet/in_pcb.h>
102	#include <netinet/ip_var.h>
103	#include <netinet/udp.h>
104	#include <netinet/udp_var.h>
105
106	#include <netinet6/in6_var.h>
107	#include <netinet/ip6.h>
108	#include <netinet6/in6_pcb.h>
109	#include <netinet6/ip6_var.h>
110	#include <netinet6/scope6_var.h>
111	#include <netinet6/nd6.h>
112
113	static struct mtx addrsel_lock;
114	#define ADDRSEL_LOCK_INIT() mtx_init(&addrsel_lock, "addrsel_lock", NULL, MTX_DEF)
115	#define ADDRSEL_LOCK() mtx_lock(&addrsel_lock)
116	#define ADDRSEL_UNLOCK() mtx_unlock(&addrsel_lock)
117	#define ADDRSEL_LOCK_ASSERT() mtx_assert(&addrsel_lock, MA_OWNED)
118
119	static struct sx addrsel_sxlock;
120	#define ADDRSEL_SXLOCK_INIT() sx_init(&addrsel_sxlock, "addrsel_sxlock")
121	#define ADDRSEL_SLOCK() sx_slock(&addrsel_sxlock)
122	#define ADDRSEL_SUNLOCK() sx_sunlock(&addrsel_sxlock)
123	#define ADDRSEL_XLOCK() sx_xlock(&addrsel_sxlock)
124	#define ADDRSEL_XUNLOCK() sx_xunlock(&addrsel_sxlock)
125
126	#define ADDR_LABEL_NOTAPP (-1)
127	static VNET_DEFINE(struct in6_addrpolicy, defaultaddrpolicy);
128	#define V_defaultaddrpolicy VNET(defaultaddrpolicy)
129
130	VNET_DEFINE(int, ip6_prefer_tempaddr) = 0;
131
132	static int selectroute(struct sockaddr_in6 , struct ip6_pktopts ,
133	struct ip6_moptions , struct route_in6 , struct ifnet **,
134	struct rtentry **, int, u_int);
135	static int in6_selectif(struct sockaddr_in6 , struct ip6_pktopts ,
136	struct ip6_moptions , struct route_in6 ro, struct ifnet **,
137	struct ifnet *, u_int);
138
139	static struct in6_addrpolicy lookup_addrsel_policy(struct sockaddr_in6 );
140
141	static void init_policy_queue(void);
142	static int add_addrsel_policyent(struct in6_addrpolicy *);
143	static int delete_addrsel_policyent(struct in6_addrpolicy *);
144	static int walk_addrsel_policy(int ()(struct in6_addrpolicy , void *),
145	void *);
146	static int dump_addrsel_policyent(struct in6_addrpolicy , void );
147	static struct in6_addrpolicy match_addrsel_policy(struct sockaddr_in6 );
148
149	/*
150	* Return an IPv6 address, which is the most appropriate for a given
151	* destination and user specified options.
152	* If necessary, this function lookups the routing table and returns
153	* an entry to the caller for later use.
154	*/
155	#define REPLACE(r) do {\
156	IP6STAT_INC(ip6s_sources_rule[(r)]); \
157	rule = (r); \
158	/* { \
159	char ip6buf[INET6_ADDRSTRLEN], ip6b[INET6_ADDRSTRLEN]; \
160	printf("in6_selectsrc: replace %s with %s by %d\n", ia_best ? ip6_sprintf(ip6buf, &ia_best->ia_addr.sin6_addr) : "none", ip6_sprintf(ip6b, &ia->ia_addr.sin6_addr), (r)); \
161	} */ \
162	goto replace; \
163	} while(0)
164	#define NEXT(r) do {\
165	/* { \
166	char ip6buf[INET6_ADDRSTRLEN], ip6b[INET6_ADDRSTRLEN]; \
167	printf("in6_selectsrc: keep %s against %s by %d\n", ia_best ? ip6_sprintf(ip6buf, &ia_best->ia_addr.sin6_addr) : "none", ip6_sprintf(ip6b, &ia->ia_addr.sin6_addr), (r)); \
168	} */ \
169	goto next; /* XXX: we can't use 'continue' here */ \
170	} while(0)
171	#define BREAK(r) do { \
172	IP6STAT_INC(ip6s_sources_rule[(r)]); \
173	rule = (r); \
174	goto out; /* XXX: we can't use 'break' here */ \
175	} while(0)
176
177	int
178	in6_selectsrc(struct sockaddr_in6 dstsock, struct ip6_pktopts opts,
179	struct inpcb inp, struct route_in6 ro, struct ucred *cred,
180	struct ifnet *ifpp, struct in6_addr srcp)
181	{
182	struct in6_addr dst, tmp;
183	struct ifnet ifp = NULL, oifp = NULL;
184	struct in6_ifaddr ia = NULL, ia_best = NULL;
185	struct in6_pktinfo *pi = NULL;
186	int dst_scope = -1, best_scope = -1, best_matchlen = -1;
187	struct in6_addrpolicy dst_policy = NULL, best_policy = NULL;
188	u_int32_t odstzone;
189	int prefer_tempaddr;
190	int error, rule;
191	struct ip6_moptions *mopts;
192
193	KASSERT(srcp != NULL, ("%s: srcp is NULL", __func__));
194
195	dst = dstsock->sin6_addr; /* make a copy for local operation */
196	if (ifpp) {
197	/*
198	* Save a possibly passed in ifp for in6_selectsrc. Only
199	* neighbor discovery code should use this feature, where
200	* we may know the interface but not the FIB number holding
201	* the connected subnet in case someone deleted it from the
202	* default FIB and we need to check the interface.
203	*/
204	if (*ifpp != NULL)
205	oifp = *ifpp;
206	*ifpp = NULL;
207	}
208
209	if (inp != NULL) {
210	INP_LOCK_ASSERT(inp);
211	mopts = inp->in6p_moptions;
212	} else {
213	mopts = NULL;
214	}
215
216	/*
217	* If the source address is explicitly specified by the caller,
218	* check if the requested source address is indeed a unicast address
219	* assigned to the node, and can be used as the packet's source
220	* address. If everything is okay, use the address as source.
221	*/
222	if (opts && (pi = opts->ip6po_pktinfo) &&
223	!IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr)) {
224	struct sockaddr_in6 srcsock;
225	struct in6_ifaddr *ia6;
226
227	/* get the outgoing interface */
228	if ((error = in6_selectif(dstsock, opts, mopts, ro, &ifp, oifp,
229	(inp != NULL) ? inp->inp_inc.inc_fibnum : RT_DEFAULT_FIB))
230	!= 0)
231	return (error);
232
233	/*
234	* determine the appropriate zone id of the source based on
235	* the zone of the destination and the outgoing interface.
236	* If the specified address is ambiguous wrt the scope zone,
237	* the interface must be specified; otherwise, ifa_ifwithaddr()
238	* will fail matching the address.
239	*/
240	bzero(&srcsock, sizeof(srcsock));
241	srcsock.sin6_family = AF_INET6;
242	srcsock.sin6_len = sizeof(srcsock);
243	srcsock.sin6_addr = pi->ipi6_addr;
244	if (ifp) {
245	error = in6_setscope(&srcsock.sin6_addr, ifp, NULL);
246	if (error)
247	return (error);
248	}
249	if (cred != NULL && (error = prison_local_ip6(cred,
250	&srcsock.sin6_addr, (inp != NULL &&
251	(inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
252	return (error);
253
254	ia6 = (struct in6_ifaddr *)ifa_ifwithaddr(
255	(struct sockaddr *)&srcsock);
256	if (ia6 == NULL \|\|
257	(ia6->ia6_flags & (IN6_IFF_ANYCAST \| IN6_IFF_NOTREADY))) {
258	if (ia6 != NULL)
259	ifa_free(&ia6->ia_ifa);
260	return (EADDRNOTAVAIL);
261	}
262	pi->ipi6_addr = srcsock.sin6_addr; /* XXX: this overrides pi */
263	if (ifpp)
264	*ifpp = ifp;
265	bcopy(&ia6->ia_addr.sin6_addr, srcp, sizeof(*srcp));
266	ifa_free(&ia6->ia_ifa);
267	return (0);
268	}
269
270	/*
271	* Otherwise, if the socket has already bound the source, just use it.
272	*/
273	if (inp != NULL && !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
274	if (cred != NULL &&
275	(error = prison_local_ip6(cred, &inp->in6p_laddr,
276	((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
277	return (error);
278	bcopy(&inp->in6p_laddr, srcp, sizeof(*srcp));
279	return (0);
280	}
281
282	/*
283	* Bypass source address selection and use the primary jail IP
284	* if requested.
285	*/
286	if (cred != NULL && !prison_saddrsel_ip6(cred, srcp))
287	return (0);
288
289	/*
290	* If the address is not specified, choose the best one based on
291	* the outgoing interface and the destination address.
292	*/
293	/* get the outgoing interface */
294	if ((error = in6_selectif(dstsock, opts, mopts, ro, &ifp, oifp,
295	(inp != NULL) ? inp->inp_inc.inc_fibnum : RT_DEFAULT_FIB)) != 0)
296	return (error);
297
298	#ifdef DIAGNOSTIC
299	if (ifp == NULL) /* this should not happen */
300	panic("in6_selectsrc: NULL ifp");
301	#endif
302	error = in6_setscope(&dst, ifp, &odstzone);
303	if (error)
304	return (error);
305
306	rule = 0;
307	IN6_IFADDR_RLOCK();
308	TAILQ_FOREACH(ia, &V_in6_ifaddrhead, ia_link) {
309	int new_scope = -1, new_matchlen = -1;
310	struct in6_addrpolicy *new_policy = NULL;
311	u_int32_t srczone, osrczone, dstzone;
312	struct in6_addr src;
313	struct ifnet *ifp1 = ia->ia_ifp;
314
315	/*
316	* We'll never take an address that breaks the scope zone
317	* of the destination. We also skip an address if its zone
318	* does not contain the outgoing interface.
319	* XXX: we should probably use sin6_scope_id here.
320	*/
321	if (in6_setscope(&dst, ifp1, &dstzone) \|\|
322	odstzone != dstzone) {
323	continue;
324	}
325	src = ia->ia_addr.sin6_addr;
326	if (in6_setscope(&src, ifp, &osrczone) \|\|
327	in6_setscope(&src, ifp1, &srczone) \|\|
328	osrczone != srczone) {
329	continue;
330	}
331
332	/* avoid unusable addresses */
333	if ((ia->ia6_flags &
334	(IN6_IFF_NOTREADY \| IN6_IFF_ANYCAST \| IN6_IFF_DETACHED))) {
335	continue;
336	}
337	if (!V_ip6_use_deprecated && IFA6_IS_DEPRECATED(ia))
338	continue;
339
340	/* If jailed only take addresses of the jail into account. */
341	if (cred != NULL &&
342	prison_check_ip6(cred, &ia->ia_addr.sin6_addr) != 0)
343	continue;
344
345	/* Rule 1: Prefer same address */
346	if (IN6_ARE_ADDR_EQUAL(&dst, &ia->ia_addr.sin6_addr)) {
347	ia_best = ia;
348	BREAK(1); /* there should be no better candidate */
349	}
350
351	if (ia_best == NULL)
352	REPLACE(0);
353
354	/* Rule 2: Prefer appropriate scope */
355	if (dst_scope < 0)
356	dst_scope = in6_addrscope(&dst);
357	new_scope = in6_addrscope(&ia->ia_addr.sin6_addr);
358	if (IN6_ARE_SCOPE_CMP(best_scope, new_scope) < 0) {
359	if (IN6_ARE_SCOPE_CMP(best_scope, dst_scope) < 0)
360	REPLACE(2);
361	NEXT(2);
362	} else if (IN6_ARE_SCOPE_CMP(new_scope, best_scope) < 0) {
363	if (IN6_ARE_SCOPE_CMP(new_scope, dst_scope) < 0)
364	NEXT(2);
365	REPLACE(2);
366	}
367
368	/*
369	* Rule 3: Avoid deprecated addresses. Note that the case of
370	* !ip6_use_deprecated is already rejected above.
371	*/
372	if (!IFA6_IS_DEPRECATED(ia_best) && IFA6_IS_DEPRECATED(ia))
373	NEXT(3);
374	if (IFA6_IS_DEPRECATED(ia_best) && !IFA6_IS_DEPRECATED(ia))
375	REPLACE(3);
376
377	/* Rule 4: Prefer home addresses */
378	/*
379	* XXX: This is a TODO. We should probably merge the MIP6
380	* case above.
381	*/
382
383	/* Rule 5: Prefer outgoing interface */
384	if (!(ND_IFINFO(ifp)->flags & ND6_IFF_NO_PREFER_IFACE)) {
385	if (ia_best->ia_ifp == ifp && ia->ia_ifp != ifp)
386	NEXT(5);
387	if (ia_best->ia_ifp != ifp && ia->ia_ifp == ifp)
388	REPLACE(5);
389	}
390
391	/*
392	* Rule 6: Prefer matching label
393	* Note that best_policy should be non-NULL here.
394	*/
395	if (dst_policy == NULL)
396	dst_policy = lookup_addrsel_policy(dstsock);
397	if (dst_policy->label != ADDR_LABEL_NOTAPP) {
398	new_policy = lookup_addrsel_policy(&ia->ia_addr);
399	if (dst_policy->label == best_policy->label &&
400	dst_policy->label != new_policy->label)
401	NEXT(6);
402	if (dst_policy->label != best_policy->label &&
403	dst_policy->label == new_policy->label)
404	REPLACE(6);
405	}
406
407	/*
408	* Rule 7: Prefer public addresses.
409	* We allow users to reverse the logic by configuring
410	* a sysctl variable, so that privacy conscious users can
411	* always prefer temporary addresses.
412	*/
413	if (opts == NULL \|\|
414	opts->ip6po_prefer_tempaddr == IP6PO_TEMPADDR_SYSTEM) {
415	prefer_tempaddr = V_ip6_prefer_tempaddr;
416	} else if (opts->ip6po_prefer_tempaddr ==
417	IP6PO_TEMPADDR_NOTPREFER) {
418	prefer_tempaddr = 0;
419	} else
420	prefer_tempaddr = 1;
421	if (!(ia_best->ia6_flags & IN6_IFF_TEMPORARY) &&
422	(ia->ia6_flags & IN6_IFF_TEMPORARY)) {
423	if (prefer_tempaddr)
424	REPLACE(7);
425	else
426	NEXT(7);
427	}
428	if ((ia_best->ia6_flags & IN6_IFF_TEMPORARY) &&
429	!(ia->ia6_flags & IN6_IFF_TEMPORARY)) {
430	if (prefer_tempaddr)
431	NEXT(7);
432	else
433	REPLACE(7);
434	}
435
436	/*
437	* Rule 8: prefer addresses on alive interfaces.
438	* This is a KAME specific rule.
439	*/
440	if ((ia_best->ia_ifp->if_flags & IFF_UP) &&
441	!(ia->ia_ifp->if_flags & IFF_UP))
442	NEXT(8);
443	if (!(ia_best->ia_ifp->if_flags & IFF_UP) &&
444	(ia->ia_ifp->if_flags & IFF_UP))
445	REPLACE(8);
446
447	/*
448	* Rule 10: prefer address with `prefer_source' flag.
449	*/
450	if ((ia_best->ia6_flags & IN6_IFF_PREFER_SOURCE) == 0 &&
451	(ia->ia6_flags & IN6_IFF_PREFER_SOURCE) != 0)
452	REPLACE(10);
453	if ((ia_best->ia6_flags & IN6_IFF_PREFER_SOURCE) != 0 &&
454	(ia->ia6_flags & IN6_IFF_PREFER_SOURCE) == 0)
455	NEXT(10);
456
457	/*
458	* Rule 14: Use longest matching prefix.
459	* Note: in the address selection draft, this rule is
460	* documented as "Rule 8". However, since it is also
461	* documented that this rule can be overridden, we assign
462	* a large number so that it is easy to assign smaller numbers
463	* to more preferred rules.
464	*/
465	new_matchlen = in6_matchlen(&ia->ia_addr.sin6_addr, &dst);
466	if (best_matchlen < new_matchlen)
467	REPLACE(14);
468	if (new_matchlen < best_matchlen)
469	NEXT(14);
470
471	/* Rule 15 is reserved. */
472
473	/*
474	* Last resort: just keep the current candidate.
475	* Or, do we need more rules?
476	*/
477	continue;
478
479	replace:
480	ia_best = ia;
481	best_scope = (new_scope >= 0 ? new_scope :
482	in6_addrscope(&ia_best->ia_addr.sin6_addr));
483	best_policy = (new_policy ? new_policy :
484	lookup_addrsel_policy(&ia_best->ia_addr));
485	best_matchlen = (new_matchlen >= 0 ? new_matchlen :
486	in6_matchlen(&ia_best->ia_addr.sin6_addr,
487	&dst));
488
489	next:
490	continue;
491
492	out:
493	break;
494	}
495
496	if ((ia = ia_best) == NULL) {
497	IN6_IFADDR_RUNLOCK();
498	IP6STAT_INC(ip6s_sources_none);
499	return (EADDRNOTAVAIL);
500	}
501
502	/*
503	* At this point at least one of the addresses belonged to the jail
504	* but it could still be, that we want to further restrict it, e.g.
505	* theoratically IN6_IS_ADDR_LOOPBACK.
506	* It must not be IN6_IS_ADDR_UNSPECIFIED anymore.
507	* prison_local_ip6() will fix an IN6_IS_ADDR_LOOPBACK but should
508	* let all others previously selected pass.
509	* Use tmp to not change ::1 on lo0 to the primary jail address.
510	*/
511	tmp = ia->ia_addr.sin6_addr;
512	if (cred != NULL && prison_local_ip6(cred, &tmp, (inp != NULL &&
513	(inp->inp_flags & IN6P_IPV6_V6ONLY) != 0)) != 0) {
514	IN6_IFADDR_RUNLOCK();
515	IP6STAT_INC(ip6s_sources_none);
516	return (EADDRNOTAVAIL);
517	}
518
519	if (ifpp)
520	*ifpp = ifp;
521
522	bcopy(&tmp, srcp, sizeof(*srcp));
523	if (ia->ia_ifp == ifp)
524	IP6STAT_INC(ip6s_sources_sameif[best_scope]);
525	else
526	IP6STAT_INC(ip6s_sources_otherif[best_scope]);
527	if (dst_scope == best_scope)
528	IP6STAT_INC(ip6s_sources_samescope[best_scope]);
529	else
530	IP6STAT_INC(ip6s_sources_otherscope[best_scope]);
531	if (IFA6_IS_DEPRECATED(ia))
532	IP6STAT_INC(ip6s_sources_deprecated[best_scope]);
533	IN6_IFADDR_RUNLOCK();
534	return (0);
535	}
536
537	/*
538	* clone - meaningful only for bsdi and freebsd
539	*/
540	static int
541	selectroute(struct sockaddr_in6 dstsock, struct ip6_pktopts opts,
542	struct ip6_moptions mopts, struct route_in6 ro,
543	struct ifnet retifp, struct rtentry retrt, int norouteok, u_int fibnum)
544	{
545	int error = 0;
546	struct ifnet *ifp = NULL;
547	struct rtentry *rt = NULL;
548	struct sockaddr_in6 *sin6_next;
549	struct in6_pktinfo *pi = NULL;
550	struct in6_addr *dst = &dstsock->sin6_addr;
551	#if 0
552	char ip6buf[INET6_ADDRSTRLEN];
553
554	if (dstsock->sin6_addr.s6_addr32[0] == 0 &&
555	dstsock->sin6_addr.s6_addr32[1] == 0 &&
556	!IN6_IS_ADDR_LOOPBACK(&dstsock->sin6_addr)) {
557	printf("in6_selectroute: strange destination %s\n",
558	ip6_sprintf(ip6buf, &dstsock->sin6_addr));
559	} else {
560	printf("in6_selectroute: destination = %s%%%d\n",
561	ip6_sprintf(ip6buf, &dstsock->sin6_addr),
562	dstsock->sin6_scope_id); /* for debug */
563	}
564	#endif
565
566	/* If the caller specify the outgoing interface explicitly, use it. */
567	if (opts && (pi = opts->ip6po_pktinfo) != NULL && pi->ipi6_ifindex) {
568	/* XXX boundary check is assumed to be already done. */
569	ifp = ifnet_byindex(pi->ipi6_ifindex);
570	if (ifp != NULL &&
571	(norouteok \|\| retrt == NULL \|\|
572	IN6_IS_ADDR_MULTICAST(dst))) {
573	/*
574	* we do not have to check or get the route for
575	* multicast.
576	*/
577	goto done;
578	} else
579	goto getroute;
580	}
581
582	/*
583	* If the destination address is a multicast address and the outgoing
584	* interface for the address is specified by the caller, use it.
585	*/
586	if (IN6_IS_ADDR_MULTICAST(dst) &&
587	mopts != NULL && (ifp = mopts->im6o_multicast_ifp) != NULL) {
588	goto done; /* we do not need a route for multicast. */
589	}
590
591	getroute:
592	/*
593	* If the next hop address for the packet is specified by the caller,
594	* use it as the gateway.
595	*/
596	if (opts && opts->ip6po_nexthop) {
597	struct route_in6 *ron;
598	struct llentry *la;
599
600	sin6_next = satosin6(opts->ip6po_nexthop);
601
602	/* at this moment, we only support AF_INET6 next hops */
603	if (sin6_next->sin6_family != AF_INET6) {
604	error = EAFNOSUPPORT; /* or should we proceed? */
605	goto done;
606	}
607
608	/*
609	* If the next hop is an IPv6 address, then the node identified
610	* by that address must be a neighbor of the sending host.
611	*/
612	ron = &opts->ip6po_nextroute;
613	/*
614	* XXX what do we do here?
615	* PLZ to be fixing
616	*/
617
618
619	if (ron->ro_rt == NULL) {
620	in6_rtalloc(ron, fibnum); /* multi path case? */
621	if (ron->ro_rt == NULL) {
622	if (ron->ro_rt) {
623	RTFREE(ron->ro_rt);
624	ron->ro_rt = NULL;
625	}
626	error = EHOSTUNREACH;
627	goto done;
628	}
629	}
630
631	rt = ron->ro_rt;
632	ifp = rt->rt_ifp;
633	IF_AFDATA_RLOCK(ifp);
634	la = lla_lookup(LLTABLE6(ifp), 0, (struct sockaddr *)&sin6_next->sin6_addr);
635	IF_AFDATA_RUNLOCK(ifp);
636	if (la != NULL)
637	LLE_RUNLOCK(la);
638	else {
639	error = EHOSTUNREACH;
640	goto done;
641	}
642	#if 0
643	if ((ron->ro_rt &&
644	(ron->ro_rt->rt_flags & (RTF_UP \| RTF_LLINFO)) !=
645	(RTF_UP \| RTF_LLINFO)) \|\|
646	!IN6_ARE_ADDR_EQUAL(&satosin6(&ron->ro_dst)->sin6_addr,
647	&sin6_next->sin6_addr)) {
648	if (ron->ro_rt) {
649	RTFREE(ron->ro_rt);
650	ron->ro_rt = NULL;
651	}
652	satosin6(&ron->ro_dst) = sin6_next;
653	}
654	if (ron->ro_rt == NULL) {
655	in6_rtalloc(ron, fibnum); /* multi path case? */
656	if (ron->ro_rt == NULL \|\|
657	!(ron->ro_rt->rt_flags & RTF_LLINFO)) {
658	if (ron->ro_rt) {
659	RTFREE(ron->ro_rt);
660	ron->ro_rt = NULL;
661	}
662	error = EHOSTUNREACH;
663	goto done;
664	}
665	}
666	#endif
667
668	/*
669	* When cloning is required, try to allocate a route to the
670	* destination so that the caller can store path MTU
671	* information.
672	*/
673	goto done;
674	}
675
676	/*
677	* Use a cached route if it exists and is valid, else try to allocate
678	* a new one. Note that we should check the address family of the
679	* cached destination, in case of sharing the cache with IPv4.
680	*/
681	if (ro) {
682	if (ro->ro_rt &&
683	(!(ro->ro_rt->rt_flags & RTF_UP) \|\|
684	((struct sockaddr *)(&ro->ro_dst))->sa_family != AF_INET6 \|\|
685	!IN6_ARE_ADDR_EQUAL(&satosin6(&ro->ro_dst)->sin6_addr,
686	dst))) {
687	RTFREE(ro->ro_rt);
688	ro->ro_rt = (struct rtentry *)NULL;
689	}
690	if (ro->ro_rt == (struct rtentry *)NULL) {
691	struct sockaddr_in6 *sa6;
692
693	/* No route yet, so try to acquire one */
694	bzero(&ro->ro_dst, sizeof(struct sockaddr_in6));
695	sa6 = (struct sockaddr_in6 *)&ro->ro_dst;
696	sa6 = dstsock;
697	sa6->sin6_scope_id = 0;
698
699	#ifdef RADIX_MPATH
700	rtalloc_mpath_fib((struct route *)ro,
701	ntohl(sa6->sin6_addr.s6_addr32[3]), fibnum);
702	#else
703	ro->ro_rt = in6_rtalloc1((struct sockaddr *)
704	&ro->ro_dst, 0, 0UL, fibnum);
705	if (ro->ro_rt)
706	RT_UNLOCK(ro->ro_rt);
707	#endif
708	}
709
710	/*
711	* do not care about the result if we have the nexthop
712	* explicitly specified.
713	*/
714	if (opts && opts->ip6po_nexthop)
715	goto done;
716
717	if (ro->ro_rt) {
718	ifp = ro->ro_rt->rt_ifp;
719
720	if (ifp == NULL) { /* can this really happen? */
721	RTFREE(ro->ro_rt);
722	ro->ro_rt = NULL;
723	}
724	}
725	if (ro->ro_rt == NULL)
726	error = EHOSTUNREACH;
727	rt = ro->ro_rt;
728
729	/*
730	* Check if the outgoing interface conflicts with
731	* the interface specified by ipi6_ifindex (if specified).
732	* Note that loopback interface is always okay.
733	* (this may happen when we are sending a packet to one of
734	* our own addresses.)
735	*/
736	if (ifp && opts && opts->ip6po_pktinfo &&
737	opts->ip6po_pktinfo->ipi6_ifindex) {
738	if (!(ifp->if_flags & IFF_LOOPBACK) &&
739	ifp->if_index !=
740	opts->ip6po_pktinfo->ipi6_ifindex) {
741	error = EHOSTUNREACH;
742	goto done;
743	}
744	}
745	}
746
747	done:
748	if (ifp == NULL && rt == NULL) {
749	/*
750	* This can happen if the caller did not pass a cached route
751	* nor any other hints. We treat this case an error.
752	*/
753	error = EHOSTUNREACH;
754	}
755	if (error == EHOSTUNREACH)
756	IP6STAT_INC(ip6s_noroute);
757
758	if (retifp != NULL) {
759	*retifp = ifp;
760
761	/*
762	* Adjust the "outgoing" interface. If we're going to loop
763	* the packet back to ourselves, the ifp would be the loopback
764	* interface. However, we'd rather know the interface associated
765	* to the destination address (which should probably be one of
766	* our own addresses.)
767	*/
768	if (rt) {
769	if ((rt->rt_ifp->if_flags & IFF_LOOPBACK) &&
770	(rt->rt_gateway->sa_family == AF_LINK))
771	*retifp =
772	ifnet_byindex(((struct sockaddr_dl *)
773	rt->rt_gateway)->sdl_index);
774	}
775	}
776
777	if (retrt != NULL)
778	retrt = rt; / rt may be NULL */
779
780	return (error);
781	}
782
783	static int
784	in6_selectif(struct sockaddr_in6 dstsock, struct ip6_pktopts opts,
785	struct ip6_moptions mopts, struct route_in6 ro, struct ifnet **retifp,
786	struct ifnet *oifp, u_int fibnum)
787	{
788	int error;
789	struct route_in6 sro;
790	struct rtentry *rt = NULL;
791
792	KASSERT(retifp != NULL, ("%s: retifp is NULL", __func__));
793
794	if (ro == NULL) {
795	bzero(&sro, sizeof(sro));
796	ro = &sro;
797	}
798
799	if ((error = selectroute(dstsock, opts, mopts, ro, retifp,
800	&rt, 1, fibnum)) != 0) {
801	if (ro == &sro && rt && rt == sro.ro_rt)
802	RTFREE(rt);
803	/* Help ND. See oifp comment in in6_selectsrc(). */
804	if (oifp != NULL && fibnum == RT_DEFAULT_FIB) {
805	*retifp = oifp;
806	error = 0;
807	}
808	return (error);
809	}
810
811	/*
812	* do not use a rejected or black hole route.
813	* XXX: this check should be done in the L2 output routine.
814	* However, if we skipped this check here, we'd see the following
815	* scenario:
816	* - install a rejected route for a scoped address prefix
817	* (like fe80::/10)
818	* - send a packet to a destination that matches the scoped prefix,
819	* with ambiguity about the scope zone.
820	* - pick the outgoing interface from the route, and disambiguate the
821	* scope zone with the interface.
822	* - ip6_output() would try to get another route with the "new"
823	* destination, which may be valid.
824	* - we'd see no error on output.
825	* Although this may not be very harmful, it should still be confusing.
826	* We thus reject the case here.
827	*/
828	if (rt && (rt->rt_flags & (RTF_REJECT \| RTF_BLACKHOLE))) {
829	int flags = (rt->rt_flags & RTF_HOST ? EHOSTUNREACH : ENETUNREACH);
830
831	if (ro == &sro && rt && rt == sro.ro_rt)
832	RTFREE(rt);
833	return (flags);
834	}
835
836	if (ro == &sro && rt && rt == sro.ro_rt)
837	RTFREE(rt);
838	return (0);
839	}
840
841	/*
842	* Public wrapper function to selectroute().
843	*
844	* XXX-BZ in6_selectroute() should and will grow the FIB argument. The
845	* in6_selectroute_fib() function is only there for backward compat on stable.
846	*/
847	int
848	in6_selectroute(struct sockaddr_in6 dstsock, struct ip6_pktopts opts,
849	struct ip6_moptions mopts, struct route_in6 ro,
850	struct ifnet retifp, struct rtentry retrt)
851	{
852
853	return (selectroute(dstsock, opts, mopts, ro, retifp,
854	retrt, 0, RT_DEFAULT_FIB));
855	}
856
857	#ifndef BURN_BRIDGES
858	int
859	in6_selectroute_fib(struct sockaddr_in6 dstsock, struct ip6_pktopts opts,
860	struct ip6_moptions mopts, struct route_in6 ro,
861	struct ifnet retifp, struct rtentry retrt, u_int fibnum)
862	{
863
864	return (selectroute(dstsock, opts, mopts, ro, retifp,
865	retrt, 0, fibnum));
866	}
867	#endif
868
869	/*
870	* Default hop limit selection. The precedence is as follows:
871	* 1. Hoplimit value specified via ioctl.
872	* 2. (If the outgoing interface is detected) the current
873	* hop limit of the interface specified by router advertisement.
874	* 3. The system default hoplimit.
875	*/
876	int
877	in6_selecthlim(struct inpcb in6p, struct ifnet ifp)
878	{
879
880	if (in6p && in6p->in6p_hops >= 0)
881	return (in6p->in6p_hops);
882	else if (ifp)
883	return (ND_IFINFO(ifp)->chlim);
884	else if (in6p && !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) {
885	struct route_in6 ro6;
886	struct ifnet *lifp;
887
888	bzero(&ro6, sizeof(ro6));
889	ro6.ro_dst.sin6_family = AF_INET6;
890	ro6.ro_dst.sin6_len = sizeof(struct sockaddr_in6);
891	ro6.ro_dst.sin6_addr = in6p->in6p_faddr;
892	in6_rtalloc(&ro6, in6p->inp_inc.inc_fibnum);
893	if (ro6.ro_rt) {
894	lifp = ro6.ro_rt->rt_ifp;
895	RTFREE(ro6.ro_rt);
896	if (lifp)
897	return (ND_IFINFO(lifp)->chlim);
898	}
899	}
900	return (V_ip6_defhlim);
901	}
902
903	/*
904	* XXX: this is borrowed from in6_pcbbind(). If possible, we should
905	* share this function by all bsd...
906	*/
907	int
908	in6_pcbsetport(struct in6_addr laddr, struct inpcb inp, struct ucred *cred)
909	{
910	struct socket *so = inp->inp_socket;
911	u_int16_t lport = 0;
912	int error, lookupflags = 0;
913	#ifdef INVARIANTS
914	struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
915	#endif
916
917	INP_WLOCK_ASSERT(inp);
918	INP_HASH_WLOCK_ASSERT(pcbinfo);
919
920	error = prison_local_ip6(cred, laddr,
921	((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0));
922	if (error)
923	return(error);
924
925	/* XXX: this is redundant when called from in6_pcbbind */
926	if ((so->so_options & (SO_REUSEADDR\|SO_REUSEPORT)) == 0)
927	lookupflags = INPLOOKUP_WILDCARD;
928
929	inp->inp_flags \|= INP_ANONPORT;
930
931	error = in_pcb_lport(inp, NULL, &lport, cred, lookupflags);
932	if (error != 0)
933	return (error);
934
935	inp->inp_lport = lport;
936	if (in_pcbinshash(inp) != 0) {
937	inp->in6p_laddr = in6addr_any;
938	inp->inp_lport = 0;
939	return (EAGAIN);
940	}
941
942	return (0);
943	}
944
945	void
946	addrsel_policy_init(void)
947	{
948
949	init_policy_queue();
950
951	/* initialize the "last resort" policy */
952	bzero(&V_defaultaddrpolicy, sizeof(V_defaultaddrpolicy));
953	V_defaultaddrpolicy.label = ADDR_LABEL_NOTAPP;
954
955	if (!IS_DEFAULT_VNET(curvnet))
956	return;
957
958	ADDRSEL_LOCK_INIT();
959	ADDRSEL_SXLOCK_INIT();
960	}
961
962	static struct in6_addrpolicy *
963	lookup_addrsel_policy(struct sockaddr_in6 *key)
964	{
965	struct in6_addrpolicy *match = NULL;
966
967	ADDRSEL_LOCK();
968	match = match_addrsel_policy(key);
969
970	if (match == NULL)
971	match = &V_defaultaddrpolicy;
972	else
973	match->use++;
974	ADDRSEL_UNLOCK();
975
976	return (match);
977	}
978
979	/*
980	* Subroutines to manage the address selection policy table via sysctl.
981	*/
982	struct walkarg {
983	struct sysctl_req *w_req;
984	};
985
986	static int in6_src_sysctl(SYSCTL_HANDLER_ARGS);
987	SYSCTL_DECL(_net_inet6_ip6);
988	static SYSCTL_NODE(_net_inet6_ip6, IPV6CTL_ADDRCTLPOLICY, addrctlpolicy,
989	CTLFLAG_RD, in6_src_sysctl, "");
990
991	static int
992	in6_src_sysctl(SYSCTL_HANDLER_ARGS)
993	{
994	struct walkarg w;
995
996	if (req->newptr)
997	return EPERM;
998
999	bzero(&w, sizeof(w));
1000	w.w_req = req;
1001
1002	return (walk_addrsel_policy(dump_addrsel_policyent, &w));
1003	}
1004
1005	int
1006	in6_src_ioctl(u_long cmd, caddr_t data)
1007	{
1008	int i;
1009	struct in6_addrpolicy ent0;
1010
1011	if (cmd != SIOCAADDRCTL_POLICY && cmd != SIOCDADDRCTL_POLICY)
1012	return (EOPNOTSUPP); /* check for safety */
1013
1014	ent0 = (struct in6_addrpolicy )data;
1015
1016	if (ent0.label == ADDR_LABEL_NOTAPP)
1017	return (EINVAL);
1018	/* check if the prefix mask is consecutive. */
1019	if (in6_mask2len(&ent0.addrmask.sin6_addr, NULL) < 0)
1020	return (EINVAL);
1021	/* clear trailing garbages (if any) of the prefix address. */
1022	for (i = 0; i < 4; i++) {
1023	ent0.addr.sin6_addr.s6_addr32[i] &=
1024	ent0.addrmask.sin6_addr.s6_addr32[i];
1025	}
1026	ent0.use = 0;
1027
1028	switch (cmd) {
1029	case SIOCAADDRCTL_POLICY:
1030	return (add_addrsel_policyent(&ent0));
1031	case SIOCDADDRCTL_POLICY:
1032	return (delete_addrsel_policyent(&ent0));
1033	}
1034
1035	return (0); /* XXX: compromise compilers */
1036	}
1037
1038	/*
1039	* The followings are implementation of the policy table using a
1040	* simple tail queue.
1041	* XXX such details should be hidden.
1042	* XXX implementation using binary tree should be more efficient.
1043	*/
1044	struct addrsel_policyent {
1045	TAILQ_ENTRY(addrsel_policyent) ape_entry;
1046	struct in6_addrpolicy ape_policy;
1047	};
1048
1049	TAILQ_HEAD(addrsel_policyhead, addrsel_policyent);
1050
1051	static VNET_DEFINE(struct addrsel_policyhead, addrsel_policytab);
1052	#define V_addrsel_policytab VNET(addrsel_policytab)
1053
1054	static void
1055	init_policy_queue(void)
1056	{
1057
1058	TAILQ_INIT(&V_addrsel_policytab);
1059	}
1060
1061	static int
1062	add_addrsel_policyent(struct in6_addrpolicy *newpolicy)
1063	{
1064	struct addrsel_policyent new, pol;
1065
1066	new = malloc(sizeof(*new), M_IFADDR,
1067	M_WAITOK);
1068	ADDRSEL_XLOCK();
1069	ADDRSEL_LOCK();
1070
1071	/* duplication check */
1072	TAILQ_FOREACH(pol, &V_addrsel_policytab, ape_entry) {
1073	if (IN6_ARE_ADDR_EQUAL(&newpolicy->addr.sin6_addr,
1074	&pol->ape_policy.addr.sin6_addr) &&
1075	IN6_ARE_ADDR_EQUAL(&newpolicy->addrmask.sin6_addr,
1076	&pol->ape_policy.addrmask.sin6_addr)) {
1077	ADDRSEL_UNLOCK();
1078	ADDRSEL_XUNLOCK();
1079	free(new, M_IFADDR);
1080	return (EEXIST); /* or override it? */
1081	}
1082	}
1083
1084	bzero(new, sizeof(*new));
1085
1086	/* XXX: should validate entry */
1087	new->ape_policy = *newpolicy;
1088
1089	TAILQ_INSERT_TAIL(&V_addrsel_policytab, new, ape_entry);
1090	ADDRSEL_UNLOCK();
1091	ADDRSEL_XUNLOCK();
1092
1093	return (0);
1094	}
1095
1096	static int
1097	delete_addrsel_policyent(struct in6_addrpolicy *key)
1098	{
1099	struct addrsel_policyent *pol;
1100
1101	ADDRSEL_XLOCK();
1102	ADDRSEL_LOCK();
1103
1104	/* search for the entry in the table */
1105	TAILQ_FOREACH(pol, &V_addrsel_policytab, ape_entry) {
1106	if (IN6_ARE_ADDR_EQUAL(&key->addr.sin6_addr,
1107	&pol->ape_policy.addr.sin6_addr) &&
1108	IN6_ARE_ADDR_EQUAL(&key->addrmask.sin6_addr,
1109	&pol->ape_policy.addrmask.sin6_addr)) {
1110	break;
1111	}
1112	}
1113	if (pol == NULL) {
1114	ADDRSEL_UNLOCK();
1115	ADDRSEL_XUNLOCK();
1116	return (ESRCH);
1117	}
1118
1119	TAILQ_REMOVE(&V_addrsel_policytab, pol, ape_entry);
1120	ADDRSEL_UNLOCK();
1121	ADDRSEL_XUNLOCK();
1122	free(pol, M_IFADDR);
1123
1124	return (0);
1125	}
1126
1127	static int
1128	walk_addrsel_policy(int (callback)(struct in6_addrpolicy , void *),
1129	void *w)
1130	{
1131	struct addrsel_policyent *pol;
1132	int error = 0;
1133
1134	ADDRSEL_SLOCK();
1135	TAILQ_FOREACH(pol, &V_addrsel_policytab, ape_entry) {
1136	if ((error = (*callback)(&pol->ape_policy, w)) != 0) {
1137	ADDRSEL_SUNLOCK();
1138	return (error);
1139	}
1140	}
1141	ADDRSEL_SUNLOCK();
1142	return (error);
1143	}
1144
1145	static int
1146	dump_addrsel_policyent(struct in6_addrpolicy pol, void arg)
1147	{
1148	int error = 0;
1149	struct walkarg *w = arg;
1150
1151	error = SYSCTL_OUT(w->w_req, pol, sizeof(*pol));
1152
1153	return (error);
1154	}
1155
1156	static struct in6_addrpolicy *
1157	match_addrsel_policy(struct sockaddr_in6 *key)
1158	{
1159	struct addrsel_policyent *pent;
1160	struct in6_addrpolicy bestpol = NULL, pol;
1161	int matchlen, bestmatchlen = -1;
1162	u_char mp, ep, k, p, m;
1163
1164	TAILQ_FOREACH(pent, &V_addrsel_policytab, ape_entry) {
1165	matchlen = 0;
1166
1167	pol = &pent->ape_policy;
1168	mp = (u_char *)&pol->addrmask.sin6_addr;
1169	ep = mp + 16; /* XXX: scope field? */
1170	k = (u_char *)&key->sin6_addr;
1171	p = (u_char *)&pol->addr.sin6_addr;
1172	for (; mp < ep && *mp; mp++, k++, p++) {
1173	m = *mp;
1174	if ((k & m) != p)
1175	goto next; /* not match */
1176	if (m == 0xff) /* short cut for a typical case */
1177	matchlen += 8;
1178	else {
1179	while (m >= 0x80) {
1180	matchlen++;
1181	m <<= 1;
1182	}
1183	}
1184	}
1185
1186	/* matched. check if this is better than the current best. */
1187	if (bestpol == NULL \|\|
1188	matchlen > bestmatchlen) {
1189	bestpol = pol;
1190	bestmatchlen = matchlen;
1191	}
1192
1193	next:
1194	continue;
1195	}
1196
1197	return (bestpol);
1198	}

Note: See TracBrowser for help on using the repository browser.

Download in other formats: