]> bbs.cooldavid.org Git - net-next-2.6.git/blame - net/ipv6/addrconf.c
fib: fib_rules_cleanup can be static
[net-next-2.6.git] / net / ipv6 / addrconf.c
CommitLineData
1da177e4
LT
1/*
2 * IPv6 Address [auto]configuration
3 * Linux INET6 implementation
4 *
5 * Authors:
1ab1457c 6 * Pedro Roque <roque@di.fc.ul.pt>
1da177e4
LT
7 * Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
8 *
1da177e4
LT
9 * This program is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU General Public License
11 * as published by the Free Software Foundation; either version
12 * 2 of the License, or (at your option) any later version.
13 */
14
15/*
16 * Changes:
17 *
18 * Janos Farkas : delete timer on ifdown
19 * <chexum@bankinf.banki.hu>
20 * Andi Kleen : kill double kfree on module
21 * unload.
22 * Maciej W. Rozycki : FDDI support
23 * sekiya@USAGI : Don't send too many RS
24 * packets.
25 * yoshfuji@USAGI : Fixed interval between DAD
26 * packets.
27 * YOSHIFUJI Hideaki @USAGI : improved accuracy of
28 * address validation timer.
29 * YOSHIFUJI Hideaki @USAGI : Privacy Extensions (RFC3041)
30 * support.
31 * Yuji SEKIYA @USAGI : Don't assign a same IPv6
32 * address on a same interface.
33 * YOSHIFUJI Hideaki @USAGI : ARCnet support
34 * YOSHIFUJI Hideaki @USAGI : convert /proc/net/if_inet6 to
35 * seq_file.
b1cacb68
YH
36 * YOSHIFUJI Hideaki @USAGI : improved source address
37 * selection; consider scope,
38 * status etc.
1da177e4
LT
39 */
40
1da177e4
LT
41#include <linux/errno.h>
42#include <linux/types.h>
a0bffffc 43#include <linux/kernel.h>
1da177e4
LT
44#include <linux/socket.h>
45#include <linux/sockios.h>
1da177e4
LT
46#include <linux/net.h>
47#include <linux/in6.h>
48#include <linux/netdevice.h>
1823730f 49#include <linux/if_addr.h>
1da177e4
LT
50#include <linux/if_arp.h>
51#include <linux/if_arcnet.h>
52#include <linux/if_infiniband.h>
53#include <linux/route.h>
54#include <linux/inetdevice.h>
55#include <linux/init.h>
5a0e3ad6 56#include <linux/slab.h>
1da177e4
LT
57#ifdef CONFIG_SYSCTL
58#include <linux/sysctl.h>
59#endif
4fc268d2 60#include <linux/capability.h>
1da177e4
LT
61#include <linux/delay.h>
62#include <linux/notifier.h>
543537bd 63#include <linux/string.h>
1da177e4 64
457c4cbc 65#include <net/net_namespace.h>
1da177e4
LT
66#include <net/sock.h>
67#include <net/snmp.h>
68
69#include <net/ipv6.h>
70#include <net/protocol.h>
71#include <net/ndisc.h>
72#include <net/ip6_route.h>
73#include <net/addrconf.h>
74#include <net/tcp.h>
75#include <net/ip.h>
5d620266 76#include <net/netlink.h>
f24e3d65 77#include <net/pkt_sched.h>
1da177e4
LT
78#include <linux/if_tunnel.h>
79#include <linux/rtnetlink.h>
80
81#ifdef CONFIG_IPV6_PRIVACY
82#include <linux/random.h>
1da177e4
LT
83#endif
84
e21e8467 85#include <linux/uaccess.h>
7f7d9a6b 86#include <asm/unaligned.h>
1da177e4
LT
87
88#include <linux/proc_fs.h>
89#include <linux/seq_file.h>
90
91/* Set to 3 to get tracing... */
92#define ACONF_DEBUG 2
93
94#if ACONF_DEBUG >= 3
95#define ADBG(x) printk x
96#else
97#define ADBG(x)
98#endif
99
100#define INFINITY_LIFE_TIME 0xFFFFFFFF
e21e8467 101#define TIME_DELTA(a, b) ((unsigned long)((long)(a) - (long)(b)))
1da177e4 102
b2db7564
YH
103#define ADDRCONF_TIMER_FUZZ_MINUS (HZ > 50 ? HZ/50 : 1)
104#define ADDRCONF_TIMER_FUZZ (HZ / 4)
105#define ADDRCONF_TIMER_FUZZ_MAX (HZ)
106
1da177e4 107#ifdef CONFIG_SYSCTL
f52295a9 108static void addrconf_sysctl_register(struct inet6_dev *idev);
408c4768
PE
109static void addrconf_sysctl_unregister(struct inet6_dev *idev);
110#else
111static inline void addrconf_sysctl_register(struct inet6_dev *idev)
112{
113}
114
115static inline void addrconf_sysctl_unregister(struct inet6_dev *idev)
116{
117}
1da177e4
LT
118#endif
119
120#ifdef CONFIG_IPV6_PRIVACY
121static int __ipv6_regen_rndid(struct inet6_dev *idev);
1ab1457c 122static int __ipv6_try_regen_rndid(struct inet6_dev *idev, struct in6_addr *tmpaddr);
1da177e4 123static void ipv6_regen_rndid(unsigned long data);
1da177e4
LT
124#endif
125
1b34be74 126static int ipv6_generate_eui64(u8 *eui, struct net_device *dev);
1da177e4
LT
127static int ipv6_count_addresses(struct inet6_dev *idev);
128
129/*
130 * Configured unicast address hash table
131 */
c2e21293 132static struct hlist_head inet6_addr_lst[IN6_ADDR_HSIZE];
5c578aed 133static DEFINE_SPINLOCK(addrconf_hash_lock);
1da177e4 134
1da177e4
LT
135static void addrconf_verify(unsigned long);
136
8d06afab 137static DEFINE_TIMER(addr_chk_timer, addrconf_verify, 0, 0);
1da177e4
LT
138static DEFINE_SPINLOCK(addrconf_verify_lock);
139
140static void addrconf_join_anycast(struct inet6_ifaddr *ifp);
141static void addrconf_leave_anycast(struct inet6_ifaddr *ifp);
142
93d9b7d7
JP
143static void addrconf_type_change(struct net_device *dev,
144 unsigned long event);
1da177e4
LT
145static int addrconf_ifdown(struct net_device *dev, int how);
146
e431b8c0 147static void addrconf_dad_start(struct inet6_ifaddr *ifp, u32 flags);
1da177e4
LT
148static void addrconf_dad_timer(unsigned long data);
149static void addrconf_dad_completed(struct inet6_ifaddr *ifp);
c5e33bdd 150static void addrconf_dad_run(struct inet6_dev *idev);
1da177e4
LT
151static void addrconf_rs_timer(unsigned long data);
152static void __ipv6_ifa_notify(int event, struct inet6_ifaddr *ifa);
153static void ipv6_ifa_notify(int event, struct inet6_ifaddr *ifa);
154
1ab1457c 155static void inet6_prefix_notify(int event, struct inet6_dev *idev,
1da177e4 156 struct prefix_info *pinfo);
3e81c6da
DM
157static bool ipv6_chk_same_addr(struct net *net, const struct in6_addr *addr,
158 struct net_device *dev);
1da177e4 159
e041c683 160static ATOMIC_NOTIFIER_HEAD(inet6addr_chain);
1da177e4 161
888c848e 162static struct ipv6_devconf ipv6_devconf __read_mostly = {
1da177e4
LT
163 .forwarding = 0,
164 .hop_limit = IPV6_DEFAULT_HOPLIMIT,
165 .mtu6 = IPV6_MIN_MTU,
166 .accept_ra = 1,
167 .accept_redirects = 1,
168 .autoconf = 1,
169 .force_mld_version = 0,
170 .dad_transmits = 1,
171 .rtr_solicits = MAX_RTR_SOLICITATIONS,
172 .rtr_solicit_interval = RTR_SOLICITATION_INTERVAL,
173 .rtr_solicit_delay = MAX_RTR_SOLICITATION_DELAY,
174#ifdef CONFIG_IPV6_PRIVACY
175 .use_tempaddr = 0,
176 .temp_valid_lft = TEMP_VALID_LIFETIME,
177 .temp_prefered_lft = TEMP_PREFERRED_LIFETIME,
178 .regen_max_retry = REGEN_MAX_RETRY,
179 .max_desync_factor = MAX_DESYNC_FACTOR,
180#endif
181 .max_addresses = IPV6_MAX_ADDRESSES,
65f5c7c1 182 .accept_ra_defrtr = 1,
c4fd30eb 183 .accept_ra_pinfo = 1,
930d6ff2
YH
184#ifdef CONFIG_IPV6_ROUTER_PREF
185 .accept_ra_rtr_pref = 1,
52e16356 186 .rtr_probe_interval = 60 * HZ,
09c884d4
YH
187#ifdef CONFIG_IPV6_ROUTE_INFO
188 .accept_ra_rt_info_max_plen = 0,
189#endif
930d6ff2 190#endif
fbea49e1 191 .proxy_ndp = 0,
0bcbc926 192 .accept_source_route = 0, /* we do not accept RH0 by default. */
778d80be 193 .disable_ipv6 = 0,
1b34be74 194 .accept_dad = 1,
1da177e4
LT
195};
196
ab32ea5d 197static struct ipv6_devconf ipv6_devconf_dflt __read_mostly = {
1da177e4
LT
198 .forwarding = 0,
199 .hop_limit = IPV6_DEFAULT_HOPLIMIT,
200 .mtu6 = IPV6_MIN_MTU,
201 .accept_ra = 1,
202 .accept_redirects = 1,
203 .autoconf = 1,
204 .dad_transmits = 1,
205 .rtr_solicits = MAX_RTR_SOLICITATIONS,
206 .rtr_solicit_interval = RTR_SOLICITATION_INTERVAL,
207 .rtr_solicit_delay = MAX_RTR_SOLICITATION_DELAY,
208#ifdef CONFIG_IPV6_PRIVACY
209 .use_tempaddr = 0,
210 .temp_valid_lft = TEMP_VALID_LIFETIME,
211 .temp_prefered_lft = TEMP_PREFERRED_LIFETIME,
212 .regen_max_retry = REGEN_MAX_RETRY,
213 .max_desync_factor = MAX_DESYNC_FACTOR,
214#endif
215 .max_addresses = IPV6_MAX_ADDRESSES,
65f5c7c1 216 .accept_ra_defrtr = 1,
c4fd30eb 217 .accept_ra_pinfo = 1,
930d6ff2
YH
218#ifdef CONFIG_IPV6_ROUTER_PREF
219 .accept_ra_rtr_pref = 1,
52e16356 220 .rtr_probe_interval = 60 * HZ,
09c884d4
YH
221#ifdef CONFIG_IPV6_ROUTE_INFO
222 .accept_ra_rt_info_max_plen = 0,
223#endif
930d6ff2 224#endif
fbea49e1 225 .proxy_ndp = 0,
0bcbc926 226 .accept_source_route = 0, /* we do not accept RH0 by default. */
778d80be 227 .disable_ipv6 = 0,
1b34be74 228 .accept_dad = 1,
1da177e4
LT
229};
230
231/* IPv6 Wildcard Address and Loopback Address defined by RFC2553 */
1da177e4 232const struct in6_addr in6addr_any = IN6ADDR_ANY_INIT;
1da177e4 233const struct in6_addr in6addr_loopback = IN6ADDR_LOOPBACK_INIT;
f3ee4010
YH
234const struct in6_addr in6addr_linklocal_allnodes = IN6ADDR_LINKLOCAL_ALLNODES_INIT;
235const struct in6_addr in6addr_linklocal_allrouters = IN6ADDR_LINKLOCAL_ALLROUTERS_INIT;
1da177e4 236
f24e3d65 237/* Check if a valid qdisc is available */
05297949 238static inline bool addrconf_qdisc_ok(const struct net_device *dev)
f24e3d65 239{
05297949 240 return !qdisc_tx_is_noop(dev);
f24e3d65
MC
241}
242
2b5ead46
YH
243/* Check if a route is valid prefix route */
244static inline int addrconf_is_prefix_route(const struct rt6_info *rt)
245{
a02cec21 246 return (rt->rt6i_flags & (RTF_GATEWAY | RTF_DEFAULT)) == 0;
2b5ead46
YH
247}
248
1da177e4
LT
249static void addrconf_del_timer(struct inet6_ifaddr *ifp)
250{
251 if (del_timer(&ifp->timer))
252 __in6_ifa_put(ifp);
253}
254
e21e8467 255enum addrconf_timer_t {
1da177e4
LT
256 AC_NONE,
257 AC_DAD,
258 AC_RS,
259};
260
261static void addrconf_mod_timer(struct inet6_ifaddr *ifp,
262 enum addrconf_timer_t what,
263 unsigned long when)
264{
265 if (!del_timer(&ifp->timer))
266 in6_ifa_hold(ifp);
267
268 switch (what) {
269 case AC_DAD:
270 ifp->timer.function = addrconf_dad_timer;
271 break;
272 case AC_RS:
273 ifp->timer.function = addrconf_rs_timer;
274 break;
e21e8467
SH
275 default:
276 break;
1da177e4
LT
277 }
278 ifp->timer.expires = jiffies + when;
279 add_timer(&ifp->timer);
280}
281
7f7d9a6b
HX
282static int snmp6_alloc_dev(struct inet6_dev *idev)
283{
7d720c3e 284 if (snmp_mib_init((void __percpu **)idev->stats.ipv6,
1823e4c8
ED
285 sizeof(struct ipstats_mib),
286 __alignof__(struct ipstats_mib)) < 0)
7f7d9a6b 287 goto err_ip;
7d720c3e 288 if (snmp_mib_init((void __percpu **)idev->stats.icmpv6,
1823e4c8
ED
289 sizeof(struct icmpv6_mib),
290 __alignof__(struct icmpv6_mib)) < 0)
7f7d9a6b 291 goto err_icmp;
7d720c3e 292 if (snmp_mib_init((void __percpu **)idev->stats.icmpv6msg,
1823e4c8
ED
293 sizeof(struct icmpv6msg_mib),
294 __alignof__(struct icmpv6msg_mib)) < 0)
14878f75 295 goto err_icmpmsg;
7f7d9a6b
HX
296
297 return 0;
298
14878f75 299err_icmpmsg:
7d720c3e 300 snmp_mib_free((void __percpu **)idev->stats.icmpv6);
7f7d9a6b 301err_icmp:
7d720c3e 302 snmp_mib_free((void __percpu **)idev->stats.ipv6);
7f7d9a6b 303err_ip:
aaf70ec7 304 return -ENOMEM;
7f7d9a6b
HX
305}
306
16910b98 307static void snmp6_free_dev(struct inet6_dev *idev)
7f7d9a6b 308{
7d720c3e
TH
309 snmp_mib_free((void __percpu **)idev->stats.icmpv6msg);
310 snmp_mib_free((void __percpu **)idev->stats.icmpv6);
311 snmp_mib_free((void __percpu **)idev->stats.ipv6);
7f7d9a6b
HX
312}
313
1da177e4
LT
314/* Nobody refers to this device, we may destroy it. */
315
8814c4b5
YH
316static void in6_dev_finish_destroy_rcu(struct rcu_head *head)
317{
318 struct inet6_dev *idev = container_of(head, struct inet6_dev, rcu);
319 kfree(idev);
320}
321
1da177e4
LT
322void in6_dev_finish_destroy(struct inet6_dev *idev)
323{
324 struct net_device *dev = idev->dev;
547b792c 325
502a2ffd 326 WARN_ON(!list_empty(&idev->addr_list));
547b792c
IJ
327 WARN_ON(idev->mc_list != NULL);
328
1da177e4
LT
329#ifdef NET_REFCNT_DEBUG
330 printk(KERN_DEBUG "in6_dev_finish_destroy: %s\n", dev ? dev->name : "NIL");
331#endif
332 dev_put(dev);
333 if (!idev->dead) {
e21e8467 334 pr_warning("Freeing alive inet6 device %p\n", idev);
1da177e4
LT
335 return;
336 }
337 snmp6_free_dev(idev);
8814c4b5 338 call_rcu(&idev->rcu, in6_dev_finish_destroy_rcu);
1da177e4
LT
339}
340
7159039a
YH
341EXPORT_SYMBOL(in6_dev_finish_destroy);
342
1da177e4
LT
343static struct inet6_dev * ipv6_add_dev(struct net_device *dev)
344{
345 struct inet6_dev *ndev;
346
347 ASSERT_RTNL();
348
349 if (dev->mtu < IPV6_MIN_MTU)
350 return NULL;
351
1ab1457c 352 ndev = kzalloc(sizeof(struct inet6_dev), GFP_KERNEL);
322f74a4 353
1ab1457c
YH
354 if (ndev == NULL)
355 return NULL;
322f74a4
IO
356
357 rwlock_init(&ndev->lock);
358 ndev->dev = dev;
502a2ffd 359 INIT_LIST_HEAD(&ndev->addr_list);
360
c346dca1 361 memcpy(&ndev->cnf, dev_net(dev)->ipv6.devconf_dflt, sizeof(ndev->cnf));
322f74a4
IO
362 ndev->cnf.mtu6 = dev->mtu;
363 ndev->cnf.sysctl = NULL;
364 ndev->nd_parms = neigh_parms_alloc(dev, &nd_tbl);
365 if (ndev->nd_parms == NULL) {
366 kfree(ndev);
367 return NULL;
368 }
0187bdfb
BH
369 if (ndev->cnf.forwarding)
370 dev_disable_lro(dev);
322f74a4
IO
371 /* We refer to the device */
372 dev_hold(dev);
1da177e4 373
322f74a4
IO
374 if (snmp6_alloc_dev(ndev) < 0) {
375 ADBG((KERN_WARNING
376 "%s(): cannot allocate memory for statistics; dev=%s.\n",
0dc47877 377 __func__, dev->name));
322f74a4
IO
378 neigh_parms_release(&nd_tbl, ndev->nd_parms);
379 ndev->dead = 1;
380 in6_dev_finish_destroy(ndev);
381 return NULL;
382 }
1da177e4 383
322f74a4
IO
384 if (snmp6_register_dev(ndev) < 0) {
385 ADBG((KERN_WARNING
386 "%s(): cannot create /proc/net/dev_snmp6/%s\n",
0dc47877 387 __func__, dev->name));
322f74a4
IO
388 neigh_parms_release(&nd_tbl, ndev->nd_parms);
389 ndev->dead = 1;
390 in6_dev_finish_destroy(ndev);
391 return NULL;
392 }
393
394 /* One reference from device. We must do this before
395 * we invoke __ipv6_regen_rndid().
396 */
397 in6_dev_hold(ndev);
1da177e4 398
1b34be74
YH
399 if (dev->flags & (IFF_NOARP | IFF_LOOPBACK))
400 ndev->cnf.accept_dad = -1;
401
b077d7ab
YH
402#if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
403 if (dev->type == ARPHRD_SIT && (dev->priv_flags & IFF_ISATAP)) {
404 printk(KERN_INFO
405 "%s: Disabled Multicast RS\n",
406 dev->name);
407 ndev->cnf.rtr_solicits = 0;
408 }
409#endif
410
1da177e4 411#ifdef CONFIG_IPV6_PRIVACY
372e6c8f 412 INIT_LIST_HEAD(&ndev->tempaddr_list);
b24b8a24 413 setup_timer(&ndev->regen_timer, ipv6_regen_rndid, (unsigned long)ndev);
322f74a4
IO
414 if ((dev->flags&IFF_LOOPBACK) ||
415 dev->type == ARPHRD_TUNNEL ||
9625ed72 416 dev->type == ARPHRD_TUNNEL6 ||
0be669bb 417 dev->type == ARPHRD_SIT ||
0be669bb 418 dev->type == ARPHRD_NONE) {
322f74a4
IO
419 printk(KERN_INFO
420 "%s: Disabled Privacy Extensions\n",
421 dev->name);
422 ndev->cnf.use_tempaddr = -1;
423 } else {
424 in6_dev_hold(ndev);
425 ipv6_regen_rndid((unsigned long) ndev);
426 }
1da177e4
LT
427#endif
428
f24e3d65 429 if (netif_running(dev) && addrconf_qdisc_ok(dev))
53aadcc9
HX
430 ndev->if_flags |= IF_READY;
431
322f74a4
IO
432 ipv6_mc_init_dev(ndev);
433 ndev->tstamp = jiffies;
f52295a9 434 addrconf_sysctl_register(ndev);
30c4cf57
DS
435 /* protected by rtnl_lock */
436 rcu_assign_pointer(dev->ip6_ptr, ndev);
d88ae4cc
YH
437
438 /* Join all-node multicast group */
f3ee4010 439 ipv6_dev_mc_inc(dev, &in6addr_linklocal_allnodes);
d88ae4cc 440
1da177e4
LT
441 return ndev;
442}
443
cee89473 444static struct inet6_dev * ipv6_find_idev(struct net_device *dev)
1da177e4
LT
445{
446 struct inet6_dev *idev;
447
448 ASSERT_RTNL();
449
e21e8467
SH
450 idev = __in6_dev_get(dev);
451 if (!idev) {
452 idev = ipv6_add_dev(dev);
453 if (!idev)
1da177e4
LT
454 return NULL;
455 }
c5e33bdd 456
1da177e4
LT
457 if (dev->flags&IFF_UP)
458 ipv6_mc_up(idev);
459 return idev;
460}
461
462#ifdef CONFIG_SYSCTL
463static void dev_forward_change(struct inet6_dev *idev)
464{
465 struct net_device *dev;
466 struct inet6_ifaddr *ifa;
1da177e4
LT
467
468 if (!idev)
469 return;
470 dev = idev->dev;
0187bdfb
BH
471 if (idev->cnf.forwarding)
472 dev_disable_lro(dev);
1da177e4 473 if (dev && (dev->flags & IFF_MULTICAST)) {
1da177e4 474 if (idev->cnf.forwarding)
f3ee4010 475 ipv6_dev_mc_inc(dev, &in6addr_linklocal_allrouters);
1da177e4 476 else
f3ee4010 477 ipv6_dev_mc_dec(dev, &in6addr_linklocal_allrouters);
1da177e4 478 }
502a2ffd 479
480 list_for_each_entry(ifa, &idev->addr_list, if_list) {
2c12a74c
MW
481 if (ifa->flags&IFA_F_TENTATIVE)
482 continue;
1da177e4
LT
483 if (idev->cnf.forwarding)
484 addrconf_join_anycast(ifa);
485 else
486 addrconf_leave_anycast(ifa);
487 }
488}
489
490
e186932b 491static void addrconf_forward_change(struct net *net, __s32 newf)
1da177e4
LT
492{
493 struct net_device *dev;
494 struct inet6_dev *idev;
495
c6d14c84
ED
496 rcu_read_lock();
497 for_each_netdev_rcu(net, dev) {
1da177e4
LT
498 idev = __in6_dev_get(dev);
499 if (idev) {
e186932b
PE
500 int changed = (!idev->cnf.forwarding) ^ (!newf);
501 idev->cnf.forwarding = newf;
1da177e4
LT
502 if (changed)
503 dev_forward_change(idev);
504 }
1da177e4 505 }
c6d14c84 506 rcu_read_unlock();
1da177e4 507}
c8fecf22 508
b325fddb 509static int addrconf_fixup_forwarding(struct ctl_table *table, int *p, int old)
c8fecf22 510{
bff16c2f
PE
511 struct net *net;
512
513 net = (struct net *)table->extra2;
441fc2a2 514 if (p == &net->ipv6.devconf_dflt->forwarding)
b325fddb
SH
515 return 0;
516
88af182e
EB
517 if (!rtnl_trylock()) {
518 /* Restore the original values before restarting */
519 *p = old;
5007392d 520 return restart_syscall();
88af182e 521 }
c8fecf22 522
e186932b
PE
523 if (p == &net->ipv6.devconf_all->forwarding) {
524 __s32 newf = net->ipv6.devconf_all->forwarding;
525 net->ipv6.devconf_dflt->forwarding = newf;
526 addrconf_forward_change(net, newf);
c8fecf22
PE
527 } else if ((!*p) ^ (!old))
528 dev_forward_change((struct inet6_dev *)table->extra1);
0187bdfb 529 rtnl_unlock();
c8fecf22
PE
530
531 if (*p)
7b4da532 532 rt6_purge_dflt_routers(net);
b325fddb 533 return 1;
c8fecf22 534}
1da177e4
LT
535#endif
536
5c578aed 537static void inet6_ifa_finish_destroy_rcu(struct rcu_head *head)
538{
539 struct inet6_ifaddr *ifp = container_of(head, struct inet6_ifaddr, rcu);
540 kfree(ifp);
541}
1da177e4 542
5c578aed 543/* Nobody refers to this ifaddr, destroy it */
1da177e4
LT
544void inet6_ifa_finish_destroy(struct inet6_ifaddr *ifp)
545{
c2e21293 546 WARN_ON(!hlist_unhashed(&ifp->addr_lst));
547b792c 547
1da177e4
LT
548#ifdef NET_REFCNT_DEBUG
549 printk(KERN_DEBUG "inet6_ifa_finish_destroy\n");
550#endif
551
552 in6_dev_put(ifp->idev);
553
554 if (del_timer(&ifp->timer))
e21e8467 555 pr_notice("Timer is still running, when freeing ifa=%p\n", ifp);
1da177e4 556
e9d3e084 557 if (ifp->state != INET6_IFADDR_STATE_DEAD) {
e21e8467 558 pr_warning("Freeing alive inet6 address %p\n", ifp);
1da177e4
LT
559 return;
560 }
d8d1f30b 561 dst_release(&ifp->rt->dst);
1da177e4 562
5c578aed 563 call_rcu(&ifp->rcu, inet6_ifa_finish_destroy_rcu);
1da177e4
LT
564}
565
e55ffac6
BH
566static void
567ipv6_link_dev_addr(struct inet6_dev *idev, struct inet6_ifaddr *ifp)
568{
502a2ffd 569 struct list_head *p;
8a6ce0c0 570 int ifp_scope = ipv6_addr_src_scope(&ifp->addr);
e55ffac6
BH
571
572 /*
573 * Each device address list is sorted in order of scope -
574 * global before linklocal.
575 */
502a2ffd 576 list_for_each(p, &idev->addr_list) {
577 struct inet6_ifaddr *ifa
578 = list_entry(p, struct inet6_ifaddr, if_list);
8a6ce0c0 579 if (ifp_scope >= ipv6_addr_src_scope(&ifa->addr))
e55ffac6
BH
580 break;
581 }
582
b54c9b98 583 list_add_tail(&ifp->if_list, p);
e55ffac6
BH
584}
585
3a88a81d 586static u32 ipv6_addr_hash(const struct in6_addr *addr)
3eb84f49 587{
3eb84f49
YH
588 /*
589 * We perform the hash function over the last 64 bits of the address
590 * This will include the IEEE address token on links that support it.
591 */
0eae88f3
ED
592 return jhash_2words((__force u32)addr->s6_addr32[2],
593 (__force u32)addr->s6_addr32[3], 0)
3a88a81d 594 & (IN6_ADDR_HSIZE - 1);
3eb84f49
YH
595}
596
1da177e4
LT
597/* On success it returns ifp with increased reference count */
598
599static struct inet6_ifaddr *
600ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr, int pfxlen,
e431b8c0 601 int scope, u32 flags)
1da177e4
LT
602{
603 struct inet6_ifaddr *ifa = NULL;
604 struct rt6_info *rt;
3a88a81d 605 unsigned int hash;
1da177e4 606 int err = 0;
d68b8270
YH
607 int addr_type = ipv6_addr_type(addr);
608
609 if (addr_type == IPV6_ADDR_ANY ||
610 addr_type & IPV6_ADDR_MULTICAST ||
611 (!(idev->dev->flags & IFF_LOOPBACK) &&
612 addr_type & IPV6_ADDR_LOOPBACK))
613 return ERR_PTR(-EADDRNOTAVAIL);
1da177e4 614
8814c4b5 615 rcu_read_lock_bh();
1da177e4
LT
616 if (idev->dead) {
617 err = -ENODEV; /*XXX*/
618 goto out2;
619 }
620
56d417b1 621 if (idev->cnf.disable_ipv6) {
9bdd8d40
BH
622 err = -EACCES;
623 goto out2;
624 }
625
5c578aed 626 spin_lock(&addrconf_hash_lock);
1da177e4
LT
627
628 /* Ignore adding duplicate addresses on an interface */
c346dca1 629 if (ipv6_chk_same_addr(dev_net(idev->dev), addr, idev->dev)) {
1da177e4
LT
630 ADBG(("ipv6_add_addr: already assigned\n"));
631 err = -EEXIST;
632 goto out;
633 }
634
322f74a4 635 ifa = kzalloc(sizeof(struct inet6_ifaddr), GFP_ATOMIC);
1da177e4
LT
636
637 if (ifa == NULL) {
638 ADBG(("ipv6_add_addr: malloc failed\n"));
639 err = -ENOBUFS;
640 goto out;
641 }
642
643 rt = addrconf_dst_alloc(idev, addr, 0);
644 if (IS_ERR(rt)) {
645 err = PTR_ERR(rt);
646 goto out;
647 }
648
1da177e4
LT
649 ipv6_addr_copy(&ifa->addr, addr);
650
651 spin_lock_init(&ifa->lock);
e9d3e084 652 spin_lock_init(&ifa->state_lock);
1da177e4 653 init_timer(&ifa->timer);
c2e21293 654 INIT_HLIST_NODE(&ifa->addr_lst);
1da177e4
LT
655 ifa->timer.data = (unsigned long) ifa;
656 ifa->scope = scope;
657 ifa->prefix_len = pfxlen;
658 ifa->flags = flags | IFA_F_TENTATIVE;
659 ifa->cstamp = ifa->tstamp = jiffies;
660
57f5f544
KF
661 ifa->rt = rt;
662
95c385b4
NH
663 /*
664 * part one of RFC 4429, section 3.3
665 * We should not configure an address as
666 * optimistic if we do not yet know the link
667 * layer address of our nexhop router
668 */
669
670 if (rt->rt6i_nexthop == NULL)
671 ifa->flags &= ~IFA_F_OPTIMISTIC;
672
1da177e4
LT
673 ifa->idev = idev;
674 in6_dev_hold(idev);
675 /* For caller */
676 in6_ifa_hold(ifa);
677
678 /* Add to big hash table */
679 hash = ipv6_addr_hash(addr);
680
5c578aed 681 hlist_add_head_rcu(&ifa->addr_lst, &inet6_addr_lst[hash]);
5c578aed 682 spin_unlock(&addrconf_hash_lock);
1da177e4
LT
683
684 write_lock(&idev->lock);
685 /* Add to inet6_dev unicast addr list. */
e55ffac6 686 ipv6_link_dev_addr(idev, ifa);
1da177e4
LT
687
688#ifdef CONFIG_IPV6_PRIVACY
689 if (ifa->flags&IFA_F_TEMPORARY) {
372e6c8f 690 list_add(&ifa->tmp_list, &idev->tempaddr_list);
1da177e4
LT
691 in6_ifa_hold(ifa);
692 }
693#endif
694
1da177e4
LT
695 in6_ifa_hold(ifa);
696 write_unlock(&idev->lock);
697out2:
8814c4b5 698 rcu_read_unlock_bh();
1da177e4 699
fd92833a 700 if (likely(err == 0))
e041c683 701 atomic_notifier_call_chain(&inet6addr_chain, NETDEV_UP, ifa);
1da177e4
LT
702 else {
703 kfree(ifa);
704 ifa = ERR_PTR(err);
705 }
706
707 return ifa;
708out:
5c578aed 709 spin_unlock(&addrconf_hash_lock);
1da177e4
LT
710 goto out2;
711}
712
713/* This function wants to get referenced ifp and releases it before return */
714
715static void ipv6_del_addr(struct inet6_ifaddr *ifp)
716{
502a2ffd 717 struct inet6_ifaddr *ifa, *ifn;
1da177e4 718 struct inet6_dev *idev = ifp->idev;
4c5ff6a6 719 int state;
1da177e4
LT
720 int hash;
721 int deleted = 0, onlink = 0;
722 unsigned long expires = jiffies;
723
724 hash = ipv6_addr_hash(&ifp->addr);
725
4c5ff6a6
HX
726 spin_lock_bh(&ifp->state_lock);
727 state = ifp->state;
e9d3e084 728 ifp->state = INET6_IFADDR_STATE_DEAD;
4c5ff6a6
HX
729 spin_unlock_bh(&ifp->state_lock);
730
731 if (state == INET6_IFADDR_STATE_DEAD)
732 goto out;
1da177e4 733
5c578aed 734 spin_lock_bh(&addrconf_hash_lock);
735 hlist_del_init_rcu(&ifp->addr_lst);
5c578aed 736 spin_unlock_bh(&addrconf_hash_lock);
1da177e4
LT
737
738 write_lock_bh(&idev->lock);
739#ifdef CONFIG_IPV6_PRIVACY
740 if (ifp->flags&IFA_F_TEMPORARY) {
372e6c8f 741 list_del(&ifp->tmp_list);
742 if (ifp->ifpub) {
743 in6_ifa_put(ifp->ifpub);
744 ifp->ifpub = NULL;
1da177e4 745 }
372e6c8f 746 __in6_ifa_put(ifp);
1da177e4
LT
747 }
748#endif
749
502a2ffd 750 list_for_each_entry_safe(ifa, ifn, &idev->addr_list, if_list) {
1da177e4 751 if (ifa == ifp) {
502a2ffd 752 list_del_init(&ifp->if_list);
1da177e4 753 __in6_ifa_put(ifp);
502a2ffd 754
1da177e4
LT
755 if (!(ifp->flags & IFA_F_PERMANENT) || onlink > 0)
756 break;
757 deleted = 1;
1d142804 758 continue;
1da177e4
LT
759 } else if (ifp->flags & IFA_F_PERMANENT) {
760 if (ipv6_prefix_equal(&ifa->addr, &ifp->addr,
761 ifp->prefix_len)) {
762 if (ifa->flags & IFA_F_PERMANENT) {
763 onlink = 1;
764 if (deleted)
765 break;
766 } else {
767 unsigned long lifetime;
768
769 if (!onlink)
770 onlink = -1;
771
772 spin_lock(&ifa->lock);
4bed72e4
YH
773
774 lifetime = addrconf_timeout_fixup(ifa->valid_lft, HZ);
775 /*
776 * Note: Because this address is
777 * not permanent, lifetime <
778 * LONG_MAX / HZ here.
779 */
1da177e4
LT
780 if (time_before(expires,
781 ifa->tstamp + lifetime * HZ))
782 expires = ifa->tstamp + lifetime * HZ;
783 spin_unlock(&ifa->lock);
784 }
785 }
786 }
787 }
788 write_unlock_bh(&idev->lock);
789
b2238566
AV
790 addrconf_del_timer(ifp);
791
1da177e4
LT
792 ipv6_ifa_notify(RTM_DELADDR, ifp);
793
e041c683 794 atomic_notifier_call_chain(&inet6addr_chain, NETDEV_DOWN, ifp);
1da177e4 795
1da177e4
LT
796 /*
797 * Purge or update corresponding prefix
798 *
799 * 1) we don't purge prefix here if address was not permanent.
800 * prefix is managed by its own lifetime.
801 * 2) if there're no addresses, delete prefix.
802 * 3) if there're still other permanent address(es),
803 * corresponding prefix is still permanent.
804 * 4) otherwise, update prefix lifetime to the
805 * longest valid lifetime among the corresponding
806 * addresses on the device.
807 * Note: subsequent RA will update lifetime.
808 *
809 * --yoshfuji
810 */
811 if ((ifp->flags & IFA_F_PERMANENT) && onlink < 1) {
812 struct in6_addr prefix;
813 struct rt6_info *rt;
c346dca1 814 struct net *net = dev_net(ifp->idev->dev);
1da177e4 815 ipv6_addr_prefix(&prefix, &ifp->addr, ifp->prefix_len);
6fda7350 816 rt = rt6_lookup(net, &prefix, NULL, ifp->idev->dev->ifindex, 1);
1da177e4 817
2b5ead46 818 if (rt && addrconf_is_prefix_route(rt)) {
1da177e4 819 if (onlink == 0) {
e0a1ad73 820 ip6_del_rt(rt);
1da177e4
LT
821 rt = NULL;
822 } else if (!(rt->rt6i_flags & RTF_EXPIRES)) {
823 rt->rt6i_expires = expires;
824 rt->rt6i_flags |= RTF_EXPIRES;
825 }
826 }
d8d1f30b 827 dst_release(&rt->dst);
1da177e4
LT
828 }
829
4c5ff6a6 830out:
1da177e4
LT
831 in6_ifa_put(ifp);
832}
833
834#ifdef CONFIG_IPV6_PRIVACY
835static int ipv6_create_tempaddr(struct inet6_ifaddr *ifp, struct inet6_ifaddr *ift)
836{
837 struct inet6_dev *idev = ifp->idev;
838 struct in6_addr addr, *tmpaddr;
839 unsigned long tmp_prefered_lft, tmp_valid_lft, tmp_cstamp, tmp_tstamp;
eac55bf9 840 unsigned long regen_advance;
1da177e4
LT
841 int tmp_plen;
842 int ret = 0;
843 int max_addresses;
95c385b4 844 u32 addr_flags;
1da177e4
LT
845
846 write_lock(&idev->lock);
847 if (ift) {
848 spin_lock_bh(&ift->lock);
849 memcpy(&addr.s6_addr[8], &ift->addr.s6_addr[8], 8);
850 spin_unlock_bh(&ift->lock);
851 tmpaddr = &addr;
852 } else {
853 tmpaddr = NULL;
854 }
855retry:
856 in6_dev_hold(idev);
857 if (idev->cnf.use_tempaddr <= 0) {
858 write_unlock(&idev->lock);
859 printk(KERN_INFO
860 "ipv6_create_tempaddr(): use_tempaddr is disabled.\n");
861 in6_dev_put(idev);
862 ret = -1;
863 goto out;
864 }
865 spin_lock_bh(&ifp->lock);
866 if (ifp->regen_count++ >= idev->cnf.regen_max_retry) {
867 idev->cnf.use_tempaddr = -1; /*XXX*/
868 spin_unlock_bh(&ifp->lock);
869 write_unlock(&idev->lock);
870 printk(KERN_WARNING
871 "ipv6_create_tempaddr(): regeneration time exceeded. disabled temporary address support.\n");
872 in6_dev_put(idev);
873 ret = -1;
874 goto out;
875 }
876 in6_ifa_hold(ifp);
877 memcpy(addr.s6_addr, ifp->addr.s6_addr, 8);
878 if (__ipv6_try_regen_rndid(idev, tmpaddr) < 0) {
879 spin_unlock_bh(&ifp->lock);
880 write_unlock(&idev->lock);
881 printk(KERN_WARNING
882 "ipv6_create_tempaddr(): regeneration of randomized interface id failed.\n");
883 in6_ifa_put(ifp);
884 in6_dev_put(idev);
885 ret = -1;
886 goto out;
887 }
888 memcpy(&addr.s6_addr[8], idev->rndid, 8);
889 tmp_valid_lft = min_t(__u32,
890 ifp->valid_lft,
891 idev->cnf.temp_valid_lft);
1ab1457c
YH
892 tmp_prefered_lft = min_t(__u32,
893 ifp->prefered_lft,
784e2710
BH
894 idev->cnf.temp_prefered_lft -
895 idev->cnf.max_desync_factor);
1da177e4
LT
896 tmp_plen = ifp->prefix_len;
897 max_addresses = idev->cnf.max_addresses;
898 tmp_cstamp = ifp->cstamp;
899 tmp_tstamp = ifp->tstamp;
900 spin_unlock_bh(&ifp->lock);
901
eac55bf9
BB
902 regen_advance = idev->cnf.regen_max_retry *
903 idev->cnf.dad_transmits *
904 idev->nd_parms->retrans_time / HZ;
1da177e4 905 write_unlock(&idev->lock);
95c385b4 906
eac55bf9
BB
907 /* A temporary address is created only if this calculated Preferred
908 * Lifetime is greater than REGEN_ADVANCE time units. In particular,
909 * an implementation must not create a temporary address with a zero
910 * Preferred Lifetime.
911 */
912 if (tmp_prefered_lft <= regen_advance) {
913 in6_ifa_put(ifp);
914 in6_dev_put(idev);
915 ret = -1;
916 goto out;
917 }
918
95c385b4
NH
919 addr_flags = IFA_F_TEMPORARY;
920 /* set in addrconf_prefix_rcv() */
921 if (ifp->flags & IFA_F_OPTIMISTIC)
922 addr_flags |= IFA_F_OPTIMISTIC;
923
1da177e4 924 ift = !max_addresses ||
1ab1457c 925 ipv6_count_addresses(idev) < max_addresses ?
1da177e4 926 ipv6_add_addr(idev, &addr, tmp_plen,
95c385b4
NH
927 ipv6_addr_type(&addr)&IPV6_ADDR_SCOPE_MASK,
928 addr_flags) : NULL;
1da177e4
LT
929 if (!ift || IS_ERR(ift)) {
930 in6_ifa_put(ifp);
931 in6_dev_put(idev);
932 printk(KERN_INFO
933 "ipv6_create_tempaddr(): retry temporary address regeneration.\n");
934 tmpaddr = &addr;
935 write_lock(&idev->lock);
936 goto retry;
937 }
938
939 spin_lock_bh(&ift->lock);
940 ift->ifpub = ifp;
941 ift->valid_lft = tmp_valid_lft;
942 ift->prefered_lft = tmp_prefered_lft;
943 ift->cstamp = tmp_cstamp;
944 ift->tstamp = tmp_tstamp;
945 spin_unlock_bh(&ift->lock);
946
947 addrconf_dad_start(ift, 0);
948 in6_ifa_put(ift);
949 in6_dev_put(idev);
950out:
951 return ret;
952}
953#endif
954
955/*
072047e4 956 * Choose an appropriate source address (RFC3484)
1da177e4 957 */
a9b05723
YH
958enum {
959 IPV6_SADDR_RULE_INIT = 0,
960 IPV6_SADDR_RULE_LOCAL,
961 IPV6_SADDR_RULE_SCOPE,
962 IPV6_SADDR_RULE_PREFERRED,
963#ifdef CONFIG_IPV6_MIP6
964 IPV6_SADDR_RULE_HOA,
965#endif
966 IPV6_SADDR_RULE_OIF,
967 IPV6_SADDR_RULE_LABEL,
968#ifdef CONFIG_IPV6_PRIVACY
969 IPV6_SADDR_RULE_PRIVACY,
970#endif
971 IPV6_SADDR_RULE_ORCHID,
972 IPV6_SADDR_RULE_PREFIX,
973 IPV6_SADDR_RULE_MAX
974};
975
072047e4 976struct ipv6_saddr_score {
a9b05723
YH
977 int rule;
978 int addr_type;
979 struct inet6_ifaddr *ifa;
980 DECLARE_BITMAP(scorebits, IPV6_SADDR_RULE_MAX);
981 int scopedist;
982 int matchlen;
072047e4
YH
983};
984
a9b05723 985struct ipv6_saddr_dst {
9acd9f3a 986 const struct in6_addr *addr;
a9b05723
YH
987 int ifindex;
988 int scope;
989 int label;
7cbca67c 990 unsigned int prefs;
a9b05723 991};
072047e4 992
b6f99a21 993static inline int ipv6_saddr_preferred(int type)
1da177e4 994{
45bb0060 995 if (type & (IPV6_ADDR_MAPPED|IPV6_ADDR_COMPATv4|IPV6_ADDR_LOOPBACK))
072047e4
YH
996 return 1;
997 return 0;
1da177e4
LT
998}
999
3de23255
BT
1000static int ipv6_get_saddr_eval(struct net *net,
1001 struct ipv6_saddr_score *score,
a9b05723
YH
1002 struct ipv6_saddr_dst *dst,
1003 int i)
1004{
1005 int ret;
1006
1007 if (i <= score->rule) {
1008 switch (i) {
1009 case IPV6_SADDR_RULE_SCOPE:
1010 ret = score->scopedist;
1011 break;
1012 case IPV6_SADDR_RULE_PREFIX:
1013 ret = score->matchlen;
1014 break;
1015 default:
1016 ret = !!test_bit(i, score->scorebits);
1017 }
1018 goto out;
1019 }
1020
1021 switch (i) {
1022 case IPV6_SADDR_RULE_INIT:
1023 /* Rule 0: remember if hiscore is not ready yet */
1024 ret = !!score->ifa;
1025 break;
1026 case IPV6_SADDR_RULE_LOCAL:
1027 /* Rule 1: Prefer same address */
1028 ret = ipv6_addr_equal(&score->ifa->addr, dst->addr);
1029 break;
1030 case IPV6_SADDR_RULE_SCOPE:
1031 /* Rule 2: Prefer appropriate scope
1032 *
1033 * ret
1034 * ^
1035 * -1 | d 15
1036 * ---+--+-+---> scope
1037 * |
1038 * | d is scope of the destination.
1039 * B-d | \
1040 * | \ <- smaller scope is better if
1041 * B-15 | \ if scope is enough for destinaion.
1042 * | ret = B - scope (-1 <= scope >= d <= 15).
1043 * d-C-1 | /
1044 * |/ <- greater is better
1045 * -C / if scope is not enough for destination.
1046 * /| ret = scope - C (-1 <= d < scope <= 15).
1047 *
1048 * d - C - 1 < B -15 (for all -1 <= d <= 15).
1049 * C > d + 14 - B >= 15 + 14 - B = 29 - B.
1050 * Assume B = 0 and we get C > 29.
1051 */
1052 ret = __ipv6_addr_src_scope(score->addr_type);
1053 if (ret >= dst->scope)
1054 ret = -ret;
1055 else
1056 ret -= 128; /* 30 is enough */
1057 score->scopedist = ret;
1058 break;
1059 case IPV6_SADDR_RULE_PREFERRED:
1060 /* Rule 3: Avoid deprecated and optimistic addresses */
1061 ret = ipv6_saddr_preferred(score->addr_type) ||
1062 !(score->ifa->flags & (IFA_F_DEPRECATED|IFA_F_OPTIMISTIC));
1063 break;
1064#ifdef CONFIG_IPV6_MIP6
1065 case IPV6_SADDR_RULE_HOA:
7cbca67c 1066 {
a9b05723 1067 /* Rule 4: Prefer home address */
7cbca67c
YH
1068 int prefhome = !(dst->prefs & IPV6_PREFER_SRC_COA);
1069 ret = !(score->ifa->flags & IFA_F_HOMEADDRESS) ^ prefhome;
a9b05723 1070 break;
7cbca67c 1071 }
a9b05723
YH
1072#endif
1073 case IPV6_SADDR_RULE_OIF:
1074 /* Rule 5: Prefer outgoing interface */
1075 ret = (!dst->ifindex ||
1076 dst->ifindex == score->ifa->idev->dev->ifindex);
1077 break;
1078 case IPV6_SADDR_RULE_LABEL:
1079 /* Rule 6: Prefer matching label */
3de23255
BT
1080 ret = ipv6_addr_label(net,
1081 &score->ifa->addr, score->addr_type,
a9b05723
YH
1082 score->ifa->idev->dev->ifindex) == dst->label;
1083 break;
1084#ifdef CONFIG_IPV6_PRIVACY
1085 case IPV6_SADDR_RULE_PRIVACY:
7cbca67c 1086 {
a9b05723
YH
1087 /* Rule 7: Prefer public address
1088 * Note: prefer temprary address if use_tempaddr >= 2
1089 */
7cbca67c
YH
1090 int preftmp = dst->prefs & (IPV6_PREFER_SRC_PUBLIC|IPV6_PREFER_SRC_TMP) ?
1091 !!(dst->prefs & IPV6_PREFER_SRC_TMP) :
1092 score->ifa->idev->cnf.use_tempaddr >= 2;
1093 ret = (!(score->ifa->flags & IFA_F_TEMPORARY)) ^ preftmp;
a9b05723 1094 break;
7cbca67c 1095 }
a9b05723
YH
1096#endif
1097 case IPV6_SADDR_RULE_ORCHID:
1098 /* Rule 8-: Prefer ORCHID vs ORCHID or
1099 * non-ORCHID vs non-ORCHID
1100 */
1101 ret = !(ipv6_addr_orchid(&score->ifa->addr) ^
1102 ipv6_addr_orchid(dst->addr));
1103 break;
1104 case IPV6_SADDR_RULE_PREFIX:
1105 /* Rule 8: Use longest matching prefix */
1106 score->matchlen = ret = ipv6_addr_diff(&score->ifa->addr,
1107 dst->addr);
1108 break;
1109 default:
1110 ret = 0;
1111 }
1112
1113 if (ret)
1114 __set_bit(i, score->scorebits);
1115 score->rule = i;
1116out:
1117 return ret;
1118}
1119
191cd582 1120int ipv6_dev_get_saddr(struct net *net, struct net_device *dst_dev,
9acd9f3a 1121 const struct in6_addr *daddr, unsigned int prefs,
7cbca67c 1122 struct in6_addr *saddr)
1da177e4 1123{
a9b05723
YH
1124 struct ipv6_saddr_score scores[2],
1125 *score = &scores[0], *hiscore = &scores[1];
a9b05723 1126 struct ipv6_saddr_dst dst;
072047e4 1127 struct net_device *dev;
a9b05723 1128 int dst_type;
1da177e4 1129
a9b05723
YH
1130 dst_type = __ipv6_addr_type(daddr);
1131 dst.addr = daddr;
1132 dst.ifindex = dst_dev ? dst_dev->ifindex : 0;
1133 dst.scope = __ipv6_addr_src_scope(dst_type);
3de23255 1134 dst.label = ipv6_addr_label(net, daddr, dst_type, dst.ifindex);
7cbca67c 1135 dst.prefs = prefs;
a9b05723
YH
1136
1137 hiscore->rule = -1;
1138 hiscore->ifa = NULL;
1da177e4 1139
8814c4b5 1140 rcu_read_lock();
1da177e4 1141
c6d14c84 1142 for_each_netdev_rcu(net, dev) {
072047e4 1143 struct inet6_dev *idev;
072047e4 1144
a9b05723 1145 /* Candidate Source Address (section 4)
072047e4
YH
1146 * - multicast and link-local destination address,
1147 * the set of candidate source address MUST only
1148 * include addresses assigned to interfaces
1149 * belonging to the same link as the outgoing
1150 * interface.
1151 * (- For site-local destination addresses, the
1152 * set of candidate source addresses MUST only
1153 * include addresses assigned to interfaces
1154 * belonging to the same site as the outgoing
1155 * interface.)
1156 */
a9b05723
YH
1157 if (((dst_type & IPV6_ADDR_MULTICAST) ||
1158 dst.scope <= IPV6_ADDR_SCOPE_LINKLOCAL) &&
1159 dst.ifindex && dev->ifindex != dst.ifindex)
072047e4 1160 continue;
1da177e4 1161
1da177e4 1162 idev = __in6_dev_get(dev);
072047e4
YH
1163 if (!idev)
1164 continue;
1165
1166 read_lock_bh(&idev->lock);
502a2ffd 1167 list_for_each_entry(score->ifa, &idev->addr_list, if_list) {
a9b05723 1168 int i;
072047e4 1169
a9b05723 1170 /*
6b3ae80a
YH
1171 * - Tentative Address (RFC2462 section 5.4)
1172 * - A tentative address is not considered
1173 * "assigned to an interface" in the traditional
95c385b4 1174 * sense, unless it is also flagged as optimistic.
6b3ae80a 1175 * - Candidate Source Address (section 4)
072047e4
YH
1176 * - In any case, anycast addresses, multicast
1177 * addresses, and the unspecified address MUST
1178 * NOT be included in a candidate set.
1179 */
a9b05723
YH
1180 if ((score->ifa->flags & IFA_F_TENTATIVE) &&
1181 (!(score->ifa->flags & IFA_F_OPTIMISTIC)))
6b3ae80a 1182 continue;
a9b05723
YH
1183
1184 score->addr_type = __ipv6_addr_type(&score->ifa->addr);
1185
1186 if (unlikely(score->addr_type == IPV6_ADDR_ANY ||
1187 score->addr_type & IPV6_ADDR_MULTICAST)) {
072047e4 1188 LIMIT_NETDEBUG(KERN_DEBUG
3b6d821c 1189 "ADDRCONF: unspecified / multicast address "
072047e4
YH
1190 "assigned as unicast address on %s",
1191 dev->name);
1192 continue;
1193 }
1194
a9b05723
YH
1195 score->rule = -1;
1196 bitmap_zero(score->scorebits, IPV6_SADDR_RULE_MAX);
1197
1198 for (i = 0; i < IPV6_SADDR_RULE_MAX; i++) {
1199 int minihiscore, miniscore;
1200
3de23255
BT
1201 minihiscore = ipv6_get_saddr_eval(net, hiscore, &dst, i);
1202 miniscore = ipv6_get_saddr_eval(net, score, &dst, i);
a9b05723
YH
1203
1204 if (minihiscore > miniscore) {
1205 if (i == IPV6_SADDR_RULE_SCOPE &&
1206 score->scopedist > 0) {
1207 /*
1208 * special case:
1209 * each remaining entry
1210 * has too small (not enough)
1211 * scope, because ifa entries
1212 * are sorted by their scope
1213 * values.
1214 */
1215 goto try_nextdev;
1216 }
1217 break;
1218 } else if (minihiscore < miniscore) {
a9b05723
YH
1219 if (hiscore->ifa)
1220 in6_ifa_put(hiscore->ifa);
1221
1222 in6_ifa_hold(score->ifa);
1223
a0bffffc 1224 swap(hiscore, score);
a9b05723
YH
1225
1226 /* restore our iterator */
1227 score->ifa = hiscore->ifa;
1228
1229 break;
1da177e4 1230 }
072047e4 1231 }
1da177e4 1232 }
a9b05723 1233try_nextdev:
072047e4 1234 read_unlock_bh(&idev->lock);
1da177e4 1235 }
8814c4b5 1236 rcu_read_unlock();
1da177e4 1237
a9b05723 1238 if (!hiscore->ifa)
072047e4 1239 return -EADDRNOTAVAIL;
1ab1457c 1240
a9b05723
YH
1241 ipv6_addr_copy(saddr, &hiscore->ifa->addr);
1242 in6_ifa_put(hiscore->ifa);
072047e4 1243 return 0;
1da177e4 1244}
5e5f3f0f 1245EXPORT_SYMBOL(ipv6_dev_get_saddr);
1da177e4 1246
95c385b4
NH
1247int ipv6_get_lladdr(struct net_device *dev, struct in6_addr *addr,
1248 unsigned char banned_flags)
1da177e4
LT
1249{
1250 struct inet6_dev *idev;
1251 int err = -EADDRNOTAVAIL;
1252
8814c4b5 1253 rcu_read_lock();
e21e8467
SH
1254 idev = __in6_dev_get(dev);
1255 if (idev) {
1da177e4
LT
1256 struct inet6_ifaddr *ifp;
1257
1258 read_lock_bh(&idev->lock);
502a2ffd 1259 list_for_each_entry(ifp, &idev->addr_list, if_list) {
1260 if (ifp->scope == IFA_LINK &&
1261 !(ifp->flags & banned_flags)) {
1da177e4
LT
1262 ipv6_addr_copy(addr, &ifp->addr);
1263 err = 0;
1264 break;
1265 }
1266 }
1267 read_unlock_bh(&idev->lock);
1268 }
8814c4b5 1269 rcu_read_unlock();
1da177e4
LT
1270 return err;
1271}
1272
1273static int ipv6_count_addresses(struct inet6_dev *idev)
1274{
1275 int cnt = 0;
1276 struct inet6_ifaddr *ifp;
1277
1278 read_lock_bh(&idev->lock);
502a2ffd 1279 list_for_each_entry(ifp, &idev->addr_list, if_list)
1da177e4
LT
1280 cnt++;
1281 read_unlock_bh(&idev->lock);
1282 return cnt;
1283}
1284
bfeade08
DL
1285int ipv6_chk_addr(struct net *net, struct in6_addr *addr,
1286 struct net_device *dev, int strict)
1da177e4 1287{
eedf042a 1288 struct inet6_ifaddr *ifp;
c2e21293 1289 struct hlist_node *node;
3a88a81d 1290 unsigned int hash = ipv6_addr_hash(addr);
1da177e4 1291
5c578aed 1292 rcu_read_lock_bh();
1293 hlist_for_each_entry_rcu(ifp, node, &inet6_addr_lst[hash], addr_lst) {
878628fb 1294 if (!net_eq(dev_net(ifp->idev->dev), net))
bfeade08 1295 continue;
1da177e4 1296 if (ipv6_addr_equal(&ifp->addr, addr) &&
eedf042a
SH
1297 !(ifp->flags&IFA_F_TENTATIVE) &&
1298 (dev == NULL || ifp->idev->dev == dev ||
1299 !(ifp->scope&(IFA_LINK|IFA_HOST) || strict))) {
1300 rcu_read_unlock_bh();
1301 return 1;
1da177e4
LT
1302 }
1303 }
5c578aed 1304
eedf042a
SH
1305 rcu_read_unlock_bh();
1306 return 0;
1da177e4 1307}
7159039a
YH
1308EXPORT_SYMBOL(ipv6_chk_addr);
1309
3e81c6da
DM
1310static bool ipv6_chk_same_addr(struct net *net, const struct in6_addr *addr,
1311 struct net_device *dev)
1da177e4 1312{
3e81c6da 1313 unsigned int hash = ipv6_addr_hash(addr);
c2e21293 1314 struct inet6_ifaddr *ifp;
1315 struct hlist_node *node;
1da177e4 1316
c2e21293 1317 hlist_for_each_entry(ifp, node, &inet6_addr_lst[hash], addr_lst) {
878628fb 1318 if (!net_eq(dev_net(ifp->idev->dev), net))
06bfe655 1319 continue;
1da177e4
LT
1320 if (ipv6_addr_equal(&ifp->addr, addr)) {
1321 if (dev == NULL || ifp->idev->dev == dev)
3e81c6da 1322 return true;
1da177e4
LT
1323 }
1324 }
3e81c6da 1325 return false;
1da177e4
LT
1326}
1327
52eeeb84
YH
1328int ipv6_chk_prefix(struct in6_addr *addr, struct net_device *dev)
1329{
1330 struct inet6_dev *idev;
1331 struct inet6_ifaddr *ifa;
1332 int onlink;
1333
1334 onlink = 0;
1335 rcu_read_lock();
1336 idev = __in6_dev_get(dev);
1337 if (idev) {
1338 read_lock_bh(&idev->lock);
502a2ffd 1339 list_for_each_entry(ifa, &idev->addr_list, if_list) {
52eeeb84
YH
1340 onlink = ipv6_prefix_equal(addr, &ifa->addr,
1341 ifa->prefix_len);
1342 if (onlink)
1343 break;
1344 }
1345 read_unlock_bh(&idev->lock);
1346 }
1347 rcu_read_unlock();
1348 return onlink;
1349}
1350
1351EXPORT_SYMBOL(ipv6_chk_prefix);
1352
9acd9f3a 1353struct inet6_ifaddr *ipv6_get_ifaddr(struct net *net, const struct in6_addr *addr,
1cab3da6 1354 struct net_device *dev, int strict)
1da177e4 1355{
b79d1d54 1356 struct inet6_ifaddr *ifp, *result = NULL;
3a88a81d 1357 unsigned int hash = ipv6_addr_hash(addr);
b79d1d54 1358 struct hlist_node *node;
1da177e4 1359
5c578aed 1360 rcu_read_lock_bh();
4f70ecca 1361 hlist_for_each_entry_rcu_bh(ifp, node, &inet6_addr_lst[hash], addr_lst) {
878628fb 1362 if (!net_eq(dev_net(ifp->idev->dev), net))
1cab3da6 1363 continue;
1da177e4
LT
1364 if (ipv6_addr_equal(&ifp->addr, addr)) {
1365 if (dev == NULL || ifp->idev->dev == dev ||
1366 !(ifp->scope&(IFA_LINK|IFA_HOST) || strict)) {
b79d1d54 1367 result = ifp;
1da177e4
LT
1368 in6_ifa_hold(ifp);
1369 break;
1370 }
1371 }
1372 }
5c578aed 1373 rcu_read_unlock_bh();
1da177e4 1374
b79d1d54 1375 return result;
1da177e4
LT
1376}
1377
1da177e4
LT
1378/* Gets referenced address, destroys ifaddr */
1379
cc411d0b 1380static void addrconf_dad_stop(struct inet6_ifaddr *ifp, int dad_failed)
1da177e4 1381{
1da177e4
LT
1382 if (ifp->flags&IFA_F_PERMANENT) {
1383 spin_lock_bh(&ifp->lock);
1384 addrconf_del_timer(ifp);
1385 ifp->flags |= IFA_F_TENTATIVE;
cc411d0b
BH
1386 if (dad_failed)
1387 ifp->flags |= IFA_F_DADFAILED;
1da177e4 1388 spin_unlock_bh(&ifp->lock);
e2577a06
HX
1389 if (dad_failed)
1390 ipv6_ifa_notify(0, ifp);
1da177e4
LT
1391 in6_ifa_put(ifp);
1392#ifdef CONFIG_IPV6_PRIVACY
1393 } else if (ifp->flags&IFA_F_TEMPORARY) {
1394 struct inet6_ifaddr *ifpub;
1395 spin_lock_bh(&ifp->lock);
1396 ifpub = ifp->ifpub;
1397 if (ifpub) {
1398 in6_ifa_hold(ifpub);
1399 spin_unlock_bh(&ifp->lock);
1400 ipv6_create_tempaddr(ifpub, ifp);
1401 in6_ifa_put(ifpub);
1402 } else {
1403 spin_unlock_bh(&ifp->lock);
1404 }
1405 ipv6_del_addr(ifp);
1406#endif
1407 } else
1408 ipv6_del_addr(ifp);
1409}
1410
f2344a13
HX
1411static int addrconf_dad_end(struct inet6_ifaddr *ifp)
1412{
1413 int err = -ENOENT;
1414
1415 spin_lock(&ifp->state_lock);
1416 if (ifp->state == INET6_IFADDR_STATE_DAD) {
1417 ifp->state = INET6_IFADDR_STATE_POSTDAD;
1418 err = 0;
1419 }
1420 spin_unlock(&ifp->state_lock);
1421
1422 return err;
1423}
1424
3c21edbd
YH
1425void addrconf_dad_failure(struct inet6_ifaddr *ifp)
1426{
1b34be74 1427 struct inet6_dev *idev = ifp->idev;
9bdd8d40 1428
f2344a13
HX
1429 if (addrconf_dad_end(ifp))
1430 return;
1431
9bdd8d40 1432 if (net_ratelimit())
0522fea6
JR
1433 printk(KERN_INFO "%s: IPv6 duplicate address %pI6c detected!\n",
1434 ifp->idev->dev->name, &ifp->addr);
9bdd8d40 1435
1b34be74
YH
1436 if (idev->cnf.accept_dad > 1 && !idev->cnf.disable_ipv6) {
1437 struct in6_addr addr;
1438
1439 addr.s6_addr32[0] = htonl(0xfe800000);
1440 addr.s6_addr32[1] = 0;
1441
1442 if (!ipv6_generate_eui64(addr.s6_addr + 8, idev->dev) &&
1443 ipv6_addr_equal(&ifp->addr, &addr)) {
1444 /* DAD failed for link-local based on MAC address */
1445 idev->cnf.disable_ipv6 = 1;
9bdd8d40
BH
1446
1447 printk(KERN_INFO "%s: IPv6 being disabled!\n",
1448 ifp->idev->dev->name);
1b34be74
YH
1449 }
1450 }
1451
cc411d0b 1452 addrconf_dad_stop(ifp, 1);
3c21edbd 1453}
1da177e4
LT
1454
1455/* Join to solicited addr multicast group. */
1456
1457void addrconf_join_solict(struct net_device *dev, struct in6_addr *addr)
1458{
1459 struct in6_addr maddr;
1460
1461 if (dev->flags&(IFF_LOOPBACK|IFF_NOARP))
1462 return;
1463
1464 addrconf_addr_solict_mult(addr, &maddr);
1465 ipv6_dev_mc_inc(dev, &maddr);
1466}
1467
1468void addrconf_leave_solict(struct inet6_dev *idev, struct in6_addr *addr)
1469{
1470 struct in6_addr maddr;
1471
1472 if (idev->dev->flags&(IFF_LOOPBACK|IFF_NOARP))
1473 return;
1474
1475 addrconf_addr_solict_mult(addr, &maddr);
1476 __ipv6_dev_mc_dec(idev, &maddr);
1477}
1478
20380731 1479static void addrconf_join_anycast(struct inet6_ifaddr *ifp)
1da177e4
LT
1480{
1481 struct in6_addr addr;
1482 ipv6_addr_prefix(&addr, &ifp->addr, ifp->prefix_len);
1483 if (ipv6_addr_any(&addr))
1484 return;
1485 ipv6_dev_ac_inc(ifp->idev->dev, &addr);
1486}
1487
20380731 1488static void addrconf_leave_anycast(struct inet6_ifaddr *ifp)
1da177e4
LT
1489{
1490 struct in6_addr addr;
1491 ipv6_addr_prefix(&addr, &ifp->addr, ifp->prefix_len);
1492 if (ipv6_addr_any(&addr))
1493 return;
1494 __ipv6_dev_ac_dec(ifp->idev, &addr);
1495}
1496
073a8e0e
YH
1497static int addrconf_ifid_eui48(u8 *eui, struct net_device *dev)
1498{
1499 if (dev->addr_len != ETH_ALEN)
1500 return -1;
1501 memcpy(eui, dev->dev_addr, 3);
1502 memcpy(eui + 5, dev->dev_addr + 3, 3);
1503
1504 /*
1505 * The zSeries OSA network cards can be shared among various
1506 * OS instances, but the OSA cards have only one MAC address.
1507 * This leads to duplicate address conflicts in conjunction
1508 * with IPv6 if more than one instance uses the same card.
1509 *
1510 * The driver for these cards can deliver a unique 16-bit
1511 * identifier for each instance sharing the same card. It is
1512 * placed instead of 0xFFFE in the interface identifier. The
1513 * "u" bit of the interface identifier is not inverted in this
1514 * case. Hence the resulting interface identifier has local
1515 * scope according to RFC2373.
1516 */
1517 if (dev->dev_id) {
1518 eui[3] = (dev->dev_id >> 8) & 0xFF;
1519 eui[4] = dev->dev_id & 0xFF;
1520 } else {
1521 eui[3] = 0xFF;
1522 eui[4] = 0xFE;
1523 eui[0] ^= 2;
1524 }
1525 return 0;
1526}
1527
1528static int addrconf_ifid_arcnet(u8 *eui, struct net_device *dev)
1529{
1530 /* XXX: inherit EUI-64 from other interface -- yoshfuji */
1531 if (dev->addr_len != ARCNET_ALEN)
1532 return -1;
1533 memset(eui, 0, 7);
1534 eui[7] = *(u8*)dev->dev_addr;
1535 return 0;
1536}
1537
1538static int addrconf_ifid_infiniband(u8 *eui, struct net_device *dev)
1539{
1540 if (dev->addr_len != INFINIBAND_ALEN)
1541 return -1;
1542 memcpy(eui, dev->dev_addr + 12, 8);
1543 eui[0] |= 2;
1544 return 0;
1545}
1546
dfd982ba
YH
1547int __ipv6_isatap_ifid(u8 *eui, __be32 addr)
1548{
9af28511
SH
1549 if (addr == 0)
1550 return -1;
dfd982ba
YH
1551 eui[0] = (ipv4_is_zeronet(addr) || ipv4_is_private_10(addr) ||
1552 ipv4_is_loopback(addr) || ipv4_is_linklocal_169(addr) ||
1553 ipv4_is_private_172(addr) || ipv4_is_test_192(addr) ||
1554 ipv4_is_anycast_6to4(addr) || ipv4_is_private_192(addr) ||
1555 ipv4_is_test_198(addr) || ipv4_is_multicast(addr) ||
1556 ipv4_is_lbcast(addr)) ? 0x00 : 0x02;
1557 eui[1] = 0;
1558 eui[2] = 0x5E;
1559 eui[3] = 0xFE;
1560 memcpy(eui + 4, &addr, 4);
1561 return 0;
1562}
1563EXPORT_SYMBOL(__ipv6_isatap_ifid);
1564
1565static int addrconf_ifid_sit(u8 *eui, struct net_device *dev)
1566{
1567 if (dev->priv_flags & IFF_ISATAP)
1568 return __ipv6_isatap_ifid(eui, *(__be32 *)dev->dev_addr);
1569 return -1;
1570}
1571
1da177e4
LT
1572static int ipv6_generate_eui64(u8 *eui, struct net_device *dev)
1573{
1574 switch (dev->type) {
1575 case ARPHRD_ETHER:
1576 case ARPHRD_FDDI:
1577 case ARPHRD_IEEE802_TR:
073a8e0e 1578 return addrconf_ifid_eui48(eui, dev);
1da177e4 1579 case ARPHRD_ARCNET:
073a8e0e 1580 return addrconf_ifid_arcnet(eui, dev);
1da177e4 1581 case ARPHRD_INFINIBAND:
073a8e0e 1582 return addrconf_ifid_infiniband(eui, dev);
c7dc89c0 1583 case ARPHRD_SIT:
dfd982ba 1584 return addrconf_ifid_sit(eui, dev);
1da177e4
LT
1585 }
1586 return -1;
1587}
1588
1589static int ipv6_inherit_eui64(u8 *eui, struct inet6_dev *idev)
1590{
1591 int err = -1;
1592 struct inet6_ifaddr *ifp;
1593
1594 read_lock_bh(&idev->lock);
502a2ffd 1595 list_for_each_entry(ifp, &idev->addr_list, if_list) {
1da177e4
LT
1596 if (ifp->scope == IFA_LINK && !(ifp->flags&IFA_F_TENTATIVE)) {
1597 memcpy(eui, ifp->addr.s6_addr+8, 8);
1598 err = 0;
1599 break;
1600 }
1601 }
1602 read_unlock_bh(&idev->lock);
1603 return err;
1604}
1605
1606#ifdef CONFIG_IPV6_PRIVACY
1607/* (re)generation of randomized interface identifier (RFC 3041 3.2, 3.5) */
1608static int __ipv6_regen_rndid(struct inet6_dev *idev)
1609{
1da177e4 1610regen:
955189ef 1611 get_random_bytes(idev->rndid, sizeof(idev->rndid));
1da177e4 1612 idev->rndid[0] &= ~0x02;
1da177e4
LT
1613
1614 /*
1615 * <draft-ietf-ipngwg-temp-addresses-v2-00.txt>:
1616 * check if generated address is not inappropriate
1617 *
1618 * - Reserved subnet anycast (RFC 2526)
1619 * 11111101 11....11 1xxxxxxx
c7dc89c0 1620 * - ISATAP (RFC4214) 6.1
1da177e4
LT
1621 * 00-00-5E-FE-xx-xx-xx-xx
1622 * - value 0
1623 * - XXX: already assigned to an address on the device
1624 */
1ab1457c 1625 if (idev->rndid[0] == 0xfd &&
1da177e4
LT
1626 (idev->rndid[1]&idev->rndid[2]&idev->rndid[3]&idev->rndid[4]&idev->rndid[5]&idev->rndid[6]) == 0xff &&
1627 (idev->rndid[7]&0x80))
1628 goto regen;
1629 if ((idev->rndid[0]|idev->rndid[1]) == 0) {
1630 if (idev->rndid[2] == 0x5e && idev->rndid[3] == 0xfe)
1631 goto regen;
1632 if ((idev->rndid[2]|idev->rndid[3]|idev->rndid[4]|idev->rndid[5]|idev->rndid[6]|idev->rndid[7]) == 0x00)
1633 goto regen;
1634 }
1635
1636 return 0;
1637}
1638
1639static void ipv6_regen_rndid(unsigned long data)
1640{
1641 struct inet6_dev *idev = (struct inet6_dev *) data;
1642 unsigned long expires;
1643
8814c4b5 1644 rcu_read_lock_bh();
1da177e4
LT
1645 write_lock_bh(&idev->lock);
1646
1647 if (idev->dead)
1648 goto out;
1649
1650 if (__ipv6_regen_rndid(idev) < 0)
1651 goto out;
1ab1457c 1652
1da177e4 1653 expires = jiffies +
1ab1457c 1654 idev->cnf.temp_prefered_lft * HZ -
784e2710
BH
1655 idev->cnf.regen_max_retry * idev->cnf.dad_transmits * idev->nd_parms->retrans_time -
1656 idev->cnf.max_desync_factor * HZ;
1da177e4
LT
1657 if (time_before(expires, jiffies)) {
1658 printk(KERN_WARNING
1659 "ipv6_regen_rndid(): too short regeneration interval; timer disabled for %s.\n",
1660 idev->dev->name);
1661 goto out;
1662 }
1663
1664 if (!mod_timer(&idev->regen_timer, expires))
1665 in6_dev_hold(idev);
1666
1667out:
1668 write_unlock_bh(&idev->lock);
8814c4b5 1669 rcu_read_unlock_bh();
1da177e4
LT
1670 in6_dev_put(idev);
1671}
1672
1673static int __ipv6_try_regen_rndid(struct inet6_dev *idev, struct in6_addr *tmpaddr) {
1674 int ret = 0;
1675
1676 if (tmpaddr && memcmp(idev->rndid, &tmpaddr->s6_addr[8], 8) == 0)
1677 ret = __ipv6_regen_rndid(idev);
1678 return ret;
1679}
1680#endif
1681
1682/*
1683 * Add prefix route.
1684 */
1685
1686static void
1687addrconf_prefix_route(struct in6_addr *pfx, int plen, struct net_device *dev,
e431b8c0 1688 unsigned long expires, u32 flags)
1da177e4 1689{
86872cb5
TG
1690 struct fib6_config cfg = {
1691 .fc_table = RT6_TABLE_PREFIX,
1692 .fc_metric = IP6_RT_PRIO_ADDRCONF,
1693 .fc_ifindex = dev->ifindex,
1694 .fc_expires = expires,
1695 .fc_dst_len = plen,
1696 .fc_flags = RTF_UP | flags,
c346dca1 1697 .fc_nlinfo.nl_net = dev_net(dev),
f410a1fb 1698 .fc_protocol = RTPROT_KERNEL,
86872cb5 1699 };
1da177e4 1700
86872cb5 1701 ipv6_addr_copy(&cfg.fc_dst, pfx);
1da177e4
LT
1702
1703 /* Prevent useless cloning on PtP SIT.
1704 This thing is done here expecting that the whole
1705 class of non-broadcast devices need not cloning.
1706 */
0be669bb 1707#if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
86872cb5
TG
1708 if (dev->type == ARPHRD_SIT && (dev->flags & IFF_POINTOPOINT))
1709 cfg.fc_flags |= RTF_NONEXTHOP;
0be669bb 1710#endif
1da177e4 1711
86872cb5 1712 ip6_route_add(&cfg);
1da177e4
LT
1713}
1714
1715/* Create "default" multicast route to the interface */
1716
1717static void addrconf_add_mroute(struct net_device *dev)
1718{
86872cb5
TG
1719 struct fib6_config cfg = {
1720 .fc_table = RT6_TABLE_LOCAL,
1721 .fc_metric = IP6_RT_PRIO_ADDRCONF,
1722 .fc_ifindex = dev->ifindex,
1723 .fc_dst_len = 8,
1724 .fc_flags = RTF_UP,
c346dca1 1725 .fc_nlinfo.nl_net = dev_net(dev),
86872cb5
TG
1726 };
1727
1728 ipv6_addr_set(&cfg.fc_dst, htonl(0xFF000000), 0, 0, 0);
1729
1730 ip6_route_add(&cfg);
1da177e4
LT
1731}
1732
0be669bb 1733#if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
1da177e4
LT
1734static void sit_route_add(struct net_device *dev)
1735{
86872cb5
TG
1736 struct fib6_config cfg = {
1737 .fc_table = RT6_TABLE_MAIN,
1738 .fc_metric = IP6_RT_PRIO_ADDRCONF,
1739 .fc_ifindex = dev->ifindex,
1740 .fc_dst_len = 96,
1741 .fc_flags = RTF_UP | RTF_NONEXTHOP,
c346dca1 1742 .fc_nlinfo.nl_net = dev_net(dev),
86872cb5 1743 };
1da177e4
LT
1744
1745 /* prefix length - 96 bits "::d.d.d.d" */
86872cb5 1746 ip6_route_add(&cfg);
1da177e4 1747}
0be669bb 1748#endif
1da177e4
LT
1749
1750static void addrconf_add_lroute(struct net_device *dev)
1751{
1752 struct in6_addr addr;
1753
1754 ipv6_addr_set(&addr, htonl(0xFE800000), 0, 0, 0);
1755 addrconf_prefix_route(&addr, 64, dev, 0, 0);
1756}
1757
1758static struct inet6_dev *addrconf_add_dev(struct net_device *dev)
1759{
1760 struct inet6_dev *idev;
1761
1762 ASSERT_RTNL();
1763
e21e8467
SH
1764 idev = ipv6_find_idev(dev);
1765 if (!idev)
64e724f6
BH
1766 return ERR_PTR(-ENOBUFS);
1767
1768 if (idev->cnf.disable_ipv6)
1769 return ERR_PTR(-EACCES);
1da177e4
LT
1770
1771 /* Add default multicast route */
1772 addrconf_add_mroute(dev);
1773
1774 /* Add link local route */
1775 addrconf_add_lroute(dev);
1776 return idev;
1777}
1778
1779void addrconf_prefix_rcv(struct net_device *dev, u8 *opt, int len)
1780{
1781 struct prefix_info *pinfo;
1782 __u32 valid_lft;
1783 __u32 prefered_lft;
1784 int addr_type;
1da177e4 1785 struct inet6_dev *in6_dev;
56d417b1 1786 struct net *net = dev_net(dev);
1da177e4
LT
1787
1788 pinfo = (struct prefix_info *) opt;
1ab1457c 1789
1da177e4
LT
1790 if (len < sizeof(struct prefix_info)) {
1791 ADBG(("addrconf: prefix option too short\n"));
1792 return;
1793 }
1ab1457c 1794
1da177e4
LT
1795 /*
1796 * Validation checks ([ADDRCONF], page 19)
1797 */
1798
1799 addr_type = ipv6_addr_type(&pinfo->prefix);
1800
1801 if (addr_type & (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL))
1802 return;
1803
1804 valid_lft = ntohl(pinfo->valid);
1805 prefered_lft = ntohl(pinfo->prefered);
1806
1807 if (prefered_lft > valid_lft) {
1808 if (net_ratelimit())
1809 printk(KERN_WARNING "addrconf: prefix option has invalid lifetime\n");
1810 return;
1811 }
1812
1813 in6_dev = in6_dev_get(dev);
1814
1815 if (in6_dev == NULL) {
1816 if (net_ratelimit())
1817 printk(KERN_DEBUG "addrconf: device %s not configured\n", dev->name);
1818 return;
1819 }
1820
1821 /*
1822 * Two things going on here:
1823 * 1) Add routes for on-link prefixes
1824 * 2) Configure prefixes with the auto flag set
1825 */
1826
4bed72e4
YH
1827 if (pinfo->onlink) {
1828 struct rt6_info *rt;
1829 unsigned long rt_expires;
1830
6f704992
YH
1831 /* Avoid arithmetic overflow. Really, we could
1832 * save rt_expires in seconds, likely valid_lft,
1833 * but it would require division in fib gc, that it
1834 * not good.
1835 */
4bed72e4
YH
1836 if (HZ > USER_HZ)
1837 rt_expires = addrconf_timeout_fixup(valid_lft, HZ);
1838 else
1839 rt_expires = addrconf_timeout_fixup(valid_lft, USER_HZ);
3dd4bc68 1840
4bed72e4
YH
1841 if (addrconf_finite_timeout(rt_expires))
1842 rt_expires *= HZ;
1da177e4 1843
56d417b1 1844 rt = rt6_lookup(net, &pinfo->prefix, NULL,
6fda7350 1845 dev->ifindex, 1);
1da177e4 1846
2b5ead46 1847 if (rt && addrconf_is_prefix_route(rt)) {
6f704992
YH
1848 /* Autoconf prefix route */
1849 if (valid_lft == 0) {
1850 ip6_del_rt(rt);
1851 rt = NULL;
4bed72e4 1852 } else if (addrconf_finite_timeout(rt_expires)) {
6f704992
YH
1853 /* not infinity */
1854 rt->rt6i_expires = jiffies + rt_expires;
1855 rt->rt6i_flags |= RTF_EXPIRES;
1856 } else {
1857 rt->rt6i_flags &= ~RTF_EXPIRES;
1858 rt->rt6i_expires = 0;
1da177e4
LT
1859 }
1860 } else if (valid_lft) {
6f704992 1861 clock_t expires = 0;
4bed72e4
YH
1862 int flags = RTF_ADDRCONF | RTF_PREFIX_RT;
1863 if (addrconf_finite_timeout(rt_expires)) {
6f704992
YH
1864 /* not infinity */
1865 flags |= RTF_EXPIRES;
1866 expires = jiffies_to_clock_t(rt_expires);
1867 }
1da177e4 1868 addrconf_prefix_route(&pinfo->prefix, pinfo->prefix_len,
6f704992 1869 dev, expires, flags);
1da177e4
LT
1870 }
1871 if (rt)
d8d1f30b 1872 dst_release(&rt->dst);
1da177e4
LT
1873 }
1874
1875 /* Try to figure out our local address for this prefix */
1876
1877 if (pinfo->autoconf && in6_dev->cnf.autoconf) {
1878 struct inet6_ifaddr * ifp;
1879 struct in6_addr addr;
1880 int create = 0, update_lft = 0;
1881
1882 if (pinfo->prefix_len == 64) {
1883 memcpy(&addr, &pinfo->prefix, 8);
1884 if (ipv6_generate_eui64(addr.s6_addr + 8, dev) &&
1885 ipv6_inherit_eui64(addr.s6_addr + 8, in6_dev)) {
1886 in6_dev_put(in6_dev);
1887 return;
1888 }
1889 goto ok;
1890 }
1891 if (net_ratelimit())
1892 printk(KERN_DEBUG "IPv6 addrconf: prefix with wrong length %d\n",
1893 pinfo->prefix_len);
1894 in6_dev_put(in6_dev);
1895 return;
1896
1897ok:
1898
53b7997f 1899 ifp = ipv6_get_ifaddr(net, &addr, dev, 1);
1da177e4
LT
1900
1901 if (ifp == NULL && valid_lft) {
1902 int max_addresses = in6_dev->cnf.max_addresses;
95c385b4
NH
1903 u32 addr_flags = 0;
1904
1905#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
1906 if (in6_dev->cnf.optimistic_dad &&
53b7997f 1907 !net->ipv6.devconf_all->forwarding)
95c385b4
NH
1908 addr_flags = IFA_F_OPTIMISTIC;
1909#endif
1da177e4
LT
1910
1911 /* Do not allow to create too much of autoconfigured
1912 * addresses; this would be too easy way to crash kernel.
1913 */
1914 if (!max_addresses ||
1915 ipv6_count_addresses(in6_dev) < max_addresses)
1916 ifp = ipv6_add_addr(in6_dev, &addr, pinfo->prefix_len,
95c385b4
NH
1917 addr_type&IPV6_ADDR_SCOPE_MASK,
1918 addr_flags);
1da177e4
LT
1919
1920 if (!ifp || IS_ERR(ifp)) {
1921 in6_dev_put(in6_dev);
1922 return;
1923 }
1924
1925 update_lft = create = 1;
1926 ifp->cstamp = jiffies;
1927 addrconf_dad_start(ifp, RTF_ADDRCONF|RTF_PREFIX_RT);
1928 }
1929
1930 if (ifp) {
1931 int flags;
1932 unsigned long now;
1933#ifdef CONFIG_IPV6_PRIVACY
1934 struct inet6_ifaddr *ift;
1935#endif
1936 u32 stored_lft;
1937
1938 /* update lifetime (RFC2462 5.5.3 e) */
1939 spin_lock(&ifp->lock);
1940 now = jiffies;
1941 if (ifp->valid_lft > (now - ifp->tstamp) / HZ)
1942 stored_lft = ifp->valid_lft - (now - ifp->tstamp) / HZ;
1943 else
1944 stored_lft = 0;
1945 if (!update_lft && stored_lft) {
1946 if (valid_lft > MIN_VALID_LIFETIME ||
1947 valid_lft > stored_lft)
1948 update_lft = 1;
1949 else if (stored_lft <= MIN_VALID_LIFETIME) {
1950 /* valid_lft <= stored_lft is always true */
a1ed0526
BH
1951 /*
1952 * RFC 4862 Section 5.5.3e:
1953 * "Note that the preferred lifetime of
1954 * the corresponding address is always
1955 * reset to the Preferred Lifetime in
1956 * the received Prefix Information
1957 * option, regardless of whether the
1958 * valid lifetime is also reset or
1959 * ignored."
1960 *
1961 * So if the preferred lifetime in
1962 * this advertisement is different
1963 * than what we have stored, but the
1964 * valid lifetime is invalid, just
1965 * reset prefered_lft.
1966 *
1967 * We must set the valid lifetime
1968 * to the stored lifetime since we'll
1969 * be updating the timestamp below,
1970 * else we'll set it back to the
1971 * minumum.
1972 */
1973 if (prefered_lft != ifp->prefered_lft) {
1974 valid_lft = stored_lft;
1975 update_lft = 1;
1976 }
1da177e4
LT
1977 } else {
1978 valid_lft = MIN_VALID_LIFETIME;
1979 if (valid_lft < prefered_lft)
1980 prefered_lft = valid_lft;
1981 update_lft = 1;
1982 }
1983 }
1984
1985 if (update_lft) {
1986 ifp->valid_lft = valid_lft;
1987 ifp->prefered_lft = prefered_lft;
1988 ifp->tstamp = now;
1989 flags = ifp->flags;
1990 ifp->flags &= ~IFA_F_DEPRECATED;
1991 spin_unlock(&ifp->lock);
1992
1993 if (!(flags&IFA_F_TENTATIVE))
1994 ipv6_ifa_notify(0, ifp);
1995 } else
1996 spin_unlock(&ifp->lock);
1997
1998#ifdef CONFIG_IPV6_PRIVACY
1999 read_lock_bh(&in6_dev->lock);
2000 /* update all temporary addresses in the list */
372e6c8f 2001 list_for_each_entry(ift, &in6_dev->tempaddr_list, tmp_list) {
1da177e4
LT
2002 /*
2003 * When adjusting the lifetimes of an existing
2004 * temporary address, only lower the lifetimes.
2005 * Implementations must not increase the
2006 * lifetimes of an existing temporary address
2007 * when processing a Prefix Information Option.
2008 */
c6fbfac2
BB
2009 if (ifp != ift->ifpub)
2010 continue;
2011
1da177e4
LT
2012 spin_lock(&ift->lock);
2013 flags = ift->flags;
2014 if (ift->valid_lft > valid_lft &&
2015 ift->valid_lft - valid_lft > (jiffies - ift->tstamp) / HZ)
2016 ift->valid_lft = valid_lft + (jiffies - ift->tstamp) / HZ;
2017 if (ift->prefered_lft > prefered_lft &&
2018 ift->prefered_lft - prefered_lft > (jiffies - ift->tstamp) / HZ)
2019 ift->prefered_lft = prefered_lft + (jiffies - ift->tstamp) / HZ;
2020 spin_unlock(&ift->lock);
2021 if (!(flags&IFA_F_TENTATIVE))
2022 ipv6_ifa_notify(0, ift);
2023 }
2024
2025 if (create && in6_dev->cnf.use_tempaddr > 0) {
2026 /*
2027 * When a new public address is created as described in [ADDRCONF],
2028 * also create a new temporary address.
2029 */
1ab1457c 2030 read_unlock_bh(&in6_dev->lock);
1da177e4
LT
2031 ipv6_create_tempaddr(ifp, NULL);
2032 } else {
2033 read_unlock_bh(&in6_dev->lock);
2034 }
2035#endif
2036 in6_ifa_put(ifp);
2037 addrconf_verify(0);
2038 }
2039 }
2040 inet6_prefix_notify(RTM_NEWPREFIX, in6_dev, pinfo);
2041 in6_dev_put(in6_dev);
2042}
2043
2044/*
2045 * Set destination address.
2046 * Special case for SIT interfaces where we create a new "virtual"
2047 * device.
2048 */
af284937 2049int addrconf_set_dstaddr(struct net *net, void __user *arg)
1da177e4
LT
2050{
2051 struct in6_ifreq ireq;
2052 struct net_device *dev;
2053 int err = -EINVAL;
2054
2055 rtnl_lock();
2056
2057 err = -EFAULT;
2058 if (copy_from_user(&ireq, arg, sizeof(struct in6_ifreq)))
2059 goto err_exit;
2060
af284937 2061 dev = __dev_get_by_index(net, ireq.ifr6_ifindex);
1da177e4
LT
2062
2063 err = -ENODEV;
2064 if (dev == NULL)
2065 goto err_exit;
2066
0be669bb 2067#if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
1da177e4 2068 if (dev->type == ARPHRD_SIT) {
5bc3eb7e 2069 const struct net_device_ops *ops = dev->netdev_ops;
1da177e4 2070 struct ifreq ifr;
1da177e4
LT
2071 struct ip_tunnel_parm p;
2072
2073 err = -EADDRNOTAVAIL;
2074 if (!(ipv6_addr_type(&ireq.ifr6_addr) & IPV6_ADDR_COMPATv4))
2075 goto err_exit;
2076
2077 memset(&p, 0, sizeof(p));
2078 p.iph.daddr = ireq.ifr6_addr.s6_addr32[3];
2079 p.iph.saddr = 0;
2080 p.iph.version = 4;
2081 p.iph.ihl = 5;
2082 p.iph.protocol = IPPROTO_IPV6;
2083 p.iph.ttl = 64;
d20b3109 2084 ifr.ifr_ifru.ifru_data = (__force void __user *)&p;
1da177e4 2085
5bc3eb7e
SH
2086 if (ops->ndo_do_ioctl) {
2087 mm_segment_t oldfs = get_fs();
2088
2089 set_fs(KERNEL_DS);
2090 err = ops->ndo_do_ioctl(dev, &ifr, SIOCADDTUNNEL);
2091 set_fs(oldfs);
2092 } else
2093 err = -EOPNOTSUPP;
1da177e4
LT
2094
2095 if (err == 0) {
2096 err = -ENOBUFS;
af284937
DL
2097 dev = __dev_get_by_name(net, p.name);
2098 if (!dev)
1da177e4
LT
2099 goto err_exit;
2100 err = dev_open(dev);
2101 }
2102 }
0be669bb 2103#endif
1da177e4
LT
2104
2105err_exit:
2106 rtnl_unlock();
2107 return err;
2108}
2109
2110/*
2111 * Manual configuration of address on an interface
2112 */
af284937 2113static int inet6_addr_add(struct net *net, int ifindex, struct in6_addr *pfx,
24ef0da7 2114 unsigned int plen, __u8 ifa_flags, __u32 prefered_lft,
af284937 2115 __u32 valid_lft)
1da177e4
LT
2116{
2117 struct inet6_ifaddr *ifp;
2118 struct inet6_dev *idev;
2119 struct net_device *dev;
2120 int scope;
6f704992
YH
2121 u32 flags;
2122 clock_t expires;
4bed72e4 2123 unsigned long timeout;
1da177e4
LT
2124
2125 ASSERT_RTNL();
1ab1457c 2126
24ef0da7
TG
2127 if (plen > 128)
2128 return -EINVAL;
2129
0778769d
NT
2130 /* check the lifetime */
2131 if (!valid_lft || prefered_lft > valid_lft)
2132 return -EINVAL;
2133
af284937
DL
2134 dev = __dev_get_by_index(net, ifindex);
2135 if (!dev)
1da177e4 2136 return -ENODEV;
1ab1457c 2137
64e724f6
BH
2138 idev = addrconf_add_dev(dev);
2139 if (IS_ERR(idev))
2140 return PTR_ERR(idev);
1da177e4
LT
2141
2142 scope = ipv6_addr_scope(pfx);
2143
4bed72e4
YH
2144 timeout = addrconf_timeout_fixup(valid_lft, HZ);
2145 if (addrconf_finite_timeout(timeout)) {
2146 expires = jiffies_to_clock_t(timeout * HZ);
2147 valid_lft = timeout;
6f704992 2148 flags = RTF_EXPIRES;
4bed72e4
YH
2149 } else {
2150 expires = 0;
2151 flags = 0;
2152 ifa_flags |= IFA_F_PERMANENT;
6f704992 2153 }
0778769d 2154
4bed72e4
YH
2155 timeout = addrconf_timeout_fixup(prefered_lft, HZ);
2156 if (addrconf_finite_timeout(timeout)) {
2157 if (timeout == 0)
2158 ifa_flags |= IFA_F_DEPRECATED;
2159 prefered_lft = timeout;
2160 }
0778769d
NT
2161
2162 ifp = ipv6_add_addr(idev, pfx, plen, scope, ifa_flags);
2163
1da177e4 2164 if (!IS_ERR(ifp)) {
06aebfb7 2165 spin_lock_bh(&ifp->lock);
0778769d
NT
2166 ifp->valid_lft = valid_lft;
2167 ifp->prefered_lft = prefered_lft;
2168 ifp->tstamp = jiffies;
06aebfb7 2169 spin_unlock_bh(&ifp->lock);
0778769d 2170
46d48046 2171 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, dev,
6f704992 2172 expires, flags);
95c385b4
NH
2173 /*
2174 * Note that section 3.1 of RFC 4429 indicates
2175 * that the Optimistic flag should not be set for
2176 * manually configured addresses
2177 */
1da177e4
LT
2178 addrconf_dad_start(ifp, 0);
2179 in6_ifa_put(ifp);
0778769d 2180 addrconf_verify(0);
1da177e4
LT
2181 return 0;
2182 }
2183
2184 return PTR_ERR(ifp);
2185}
2186
af284937 2187static int inet6_addr_del(struct net *net, int ifindex, struct in6_addr *pfx,
24ef0da7 2188 unsigned int plen)
1da177e4
LT
2189{
2190 struct inet6_ifaddr *ifp;
2191 struct inet6_dev *idev;
2192 struct net_device *dev;
1ab1457c 2193
24ef0da7
TG
2194 if (plen > 128)
2195 return -EINVAL;
2196
af284937
DL
2197 dev = __dev_get_by_index(net, ifindex);
2198 if (!dev)
1da177e4
LT
2199 return -ENODEV;
2200
2201 if ((idev = __in6_dev_get(dev)) == NULL)
2202 return -ENXIO;
2203
2204 read_lock_bh(&idev->lock);
502a2ffd 2205 list_for_each_entry(ifp, &idev->addr_list, if_list) {
1da177e4
LT
2206 if (ifp->prefix_len == plen &&
2207 ipv6_addr_equal(pfx, &ifp->addr)) {
2208 in6_ifa_hold(ifp);
2209 read_unlock_bh(&idev->lock);
1ab1457c 2210
1da177e4
LT
2211 ipv6_del_addr(ifp);
2212
2213 /* If the last address is deleted administratively,
2214 disable IPv6 on this interface.
2215 */
502a2ffd 2216 if (list_empty(&idev->addr_list))
1da177e4
LT
2217 addrconf_ifdown(idev->dev, 1);
2218 return 0;
2219 }
2220 }
2221 read_unlock_bh(&idev->lock);
2222 return -EADDRNOTAVAIL;
2223}
2224
2225
af284937 2226int addrconf_add_ifaddr(struct net *net, void __user *arg)
1da177e4
LT
2227{
2228 struct in6_ifreq ireq;
2229 int err;
1ab1457c 2230
1da177e4
LT
2231 if (!capable(CAP_NET_ADMIN))
2232 return -EPERM;
1ab1457c 2233
1da177e4
LT
2234 if (copy_from_user(&ireq, arg, sizeof(struct in6_ifreq)))
2235 return -EFAULT;
2236
2237 rtnl_lock();
af284937
DL
2238 err = inet6_addr_add(net, ireq.ifr6_ifindex, &ireq.ifr6_addr,
2239 ireq.ifr6_prefixlen, IFA_F_PERMANENT,
2240 INFINITY_LIFE_TIME, INFINITY_LIFE_TIME);
1da177e4
LT
2241 rtnl_unlock();
2242 return err;
2243}
2244
af284937 2245int addrconf_del_ifaddr(struct net *net, void __user *arg)
1da177e4
LT
2246{
2247 struct in6_ifreq ireq;
2248 int err;
1ab1457c 2249
1da177e4
LT
2250 if (!capable(CAP_NET_ADMIN))
2251 return -EPERM;
2252
2253 if (copy_from_user(&ireq, arg, sizeof(struct in6_ifreq)))
2254 return -EFAULT;
2255
2256 rtnl_lock();
af284937
DL
2257 err = inet6_addr_del(net, ireq.ifr6_ifindex, &ireq.ifr6_addr,
2258 ireq.ifr6_prefixlen);
1da177e4
LT
2259 rtnl_unlock();
2260 return err;
2261}
2262
b5f348e5
IJ
2263static void add_addr(struct inet6_dev *idev, const struct in6_addr *addr,
2264 int plen, int scope)
2265{
2266 struct inet6_ifaddr *ifp;
2267
2268 ifp = ipv6_add_addr(idev, addr, plen, scope, IFA_F_PERMANENT);
2269 if (!IS_ERR(ifp)) {
2270 spin_lock_bh(&ifp->lock);
2271 ifp->flags &= ~IFA_F_TENTATIVE;
2272 spin_unlock_bh(&ifp->lock);
2273 ipv6_ifa_notify(RTM_NEWADDR, ifp);
2274 in6_ifa_put(ifp);
2275 }
2276}
2277
0be669bb 2278#if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
1da177e4
LT
2279static void sit_add_v4_addrs(struct inet6_dev *idev)
2280{
1da177e4
LT
2281 struct in6_addr addr;
2282 struct net_device *dev;
c346dca1 2283 struct net *net = dev_net(idev->dev);
1da177e4
LT
2284 int scope;
2285
2286 ASSERT_RTNL();
2287
2288 memset(&addr, 0, sizeof(struct in6_addr));
2289 memcpy(&addr.s6_addr32[3], idev->dev->dev_addr, 4);
2290
2291 if (idev->dev->flags&IFF_POINTOPOINT) {
2292 addr.s6_addr32[0] = htonl(0xfe800000);
2293 scope = IFA_LINK;
2294 } else {
2295 scope = IPV6_ADDR_COMPATv4;
2296 }
2297
2298 if (addr.s6_addr32[3]) {
b5f348e5 2299 add_addr(idev, &addr, 128, scope);
1da177e4
LT
2300 return;
2301 }
2302
6fda7350 2303 for_each_netdev(net, dev) {
e5ed6399 2304 struct in_device * in_dev = __in_dev_get_rtnl(dev);
1da177e4
LT
2305 if (in_dev && (dev->flags & IFF_UP)) {
2306 struct in_ifaddr * ifa;
2307
2308 int flag = scope;
2309
2310 for (ifa = in_dev->ifa_list; ifa; ifa = ifa->ifa_next) {
2311 int plen;
2312
2313 addr.s6_addr32[3] = ifa->ifa_local;
2314
2315 if (ifa->ifa_scope == RT_SCOPE_LINK)
2316 continue;
2317 if (ifa->ifa_scope >= RT_SCOPE_HOST) {
2318 if (idev->dev->flags&IFF_POINTOPOINT)
2319 continue;
2320 flag |= IFA_HOST;
2321 }
2322 if (idev->dev->flags&IFF_POINTOPOINT)
2323 plen = 64;
2324 else
2325 plen = 96;
2326
b5f348e5 2327 add_addr(idev, &addr, plen, flag);
1da177e4
LT
2328 }
2329 }
1ab1457c 2330 }
1da177e4 2331}
0be669bb 2332#endif
1da177e4
LT
2333
2334static void init_loopback(struct net_device *dev)
2335{
2336 struct inet6_dev *idev;
1da177e4
LT
2337
2338 /* ::1 */
2339
2340 ASSERT_RTNL();
2341
2342 if ((idev = ipv6_find_idev(dev)) == NULL) {
2343 printk(KERN_DEBUG "init loopback: add_dev failed\n");
2344 return;
2345 }
2346
b5f348e5 2347 add_addr(idev, &in6addr_loopback, 128, IFA_HOST);
1da177e4
LT
2348}
2349
2350static void addrconf_add_linklocal(struct inet6_dev *idev, struct in6_addr *addr)
2351{
2352 struct inet6_ifaddr * ifp;
95c385b4
NH
2353 u32 addr_flags = IFA_F_PERMANENT;
2354
2355#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
2356 if (idev->cnf.optimistic_dad &&
702beb87 2357 !dev_net(idev->dev)->ipv6.devconf_all->forwarding)
95c385b4
NH
2358 addr_flags |= IFA_F_OPTIMISTIC;
2359#endif
1da177e4 2360
95c385b4
NH
2361
2362 ifp = ipv6_add_addr(idev, addr, 64, IFA_LINK, addr_flags);
1da177e4 2363 if (!IS_ERR(ifp)) {
46d48046 2364 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, idev->dev, 0, 0);
1da177e4
LT
2365 addrconf_dad_start(ifp, 0);
2366 in6_ifa_put(ifp);
2367 }
2368}
2369
2370static void addrconf_dev_config(struct net_device *dev)
2371{
2372 struct in6_addr addr;
2373 struct inet6_dev * idev;
2374
2375 ASSERT_RTNL();
2376
74235a25
HX
2377 if ((dev->type != ARPHRD_ETHER) &&
2378 (dev->type != ARPHRD_FDDI) &&
2379 (dev->type != ARPHRD_IEEE802_TR) &&
2380 (dev->type != ARPHRD_ARCNET) &&
2381 (dev->type != ARPHRD_INFINIBAND)) {
2382 /* Alas, we support only Ethernet autoconfiguration. */
2383 return;
2384 }
2385
1da177e4 2386 idev = addrconf_add_dev(dev);
64e724f6 2387 if (IS_ERR(idev))
1da177e4
LT
2388 return;
2389
2390 memset(&addr, 0, sizeof(struct in6_addr));
2391 addr.s6_addr32[0] = htonl(0xFE800000);
2392
2393 if (ipv6_generate_eui64(addr.s6_addr + 8, dev) == 0)
2394 addrconf_add_linklocal(idev, &addr);
2395}
2396
0be669bb 2397#if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
1da177e4
LT
2398static void addrconf_sit_config(struct net_device *dev)
2399{
2400 struct inet6_dev *idev;
2401
2402 ASSERT_RTNL();
2403
1ab1457c
YH
2404 /*
2405 * Configure the tunnel with one of our IPv4
2406 * addresses... we should configure all of
1da177e4
LT
2407 * our v4 addrs in the tunnel
2408 */
2409
2410 if ((idev = ipv6_find_idev(dev)) == NULL) {
2411 printk(KERN_DEBUG "init sit: add_dev failed\n");
2412 return;
2413 }
2414
c7dc89c0
FT
2415 if (dev->priv_flags & IFF_ISATAP) {
2416 struct in6_addr addr;
2417
2418 ipv6_addr_set(&addr, htonl(0xFE800000), 0, 0, 0);
2419 addrconf_prefix_route(&addr, 64, dev, 0, 0);
2420 if (!ipv6_generate_eui64(addr.s6_addr + 8, dev))
2421 addrconf_add_linklocal(idev, &addr);
2422 return;
2423 }
2424
1da177e4
LT
2425 sit_add_v4_addrs(idev);
2426
2427 if (dev->flags&IFF_POINTOPOINT) {
2428 addrconf_add_mroute(dev);
2429 addrconf_add_lroute(dev);
2430 } else
2431 sit_route_add(dev);
2432}
0be669bb 2433#endif
1da177e4
LT
2434
2435static inline int
2436ipv6_inherit_linklocal(struct inet6_dev *idev, struct net_device *link_dev)
2437{
2438 struct in6_addr lladdr;
2439
95c385b4 2440 if (!ipv6_get_lladdr(link_dev, &lladdr, IFA_F_TENTATIVE)) {
1da177e4
LT
2441 addrconf_add_linklocal(idev, &lladdr);
2442 return 0;
2443 }
2444 return -1;
2445}
2446
2447static void ip6_tnl_add_linklocal(struct inet6_dev *idev)
2448{
2449 struct net_device *link_dev;
c346dca1 2450 struct net *net = dev_net(idev->dev);
1da177e4
LT
2451
2452 /* first try to inherit the link-local address from the link device */
2453 if (idev->dev->iflink &&
6fda7350 2454 (link_dev = __dev_get_by_index(net, idev->dev->iflink))) {
1da177e4
LT
2455 if (!ipv6_inherit_linklocal(idev, link_dev))
2456 return;
2457 }
2458 /* then try to inherit it from any device */
6fda7350 2459 for_each_netdev(net, link_dev) {
1da177e4
LT
2460 if (!ipv6_inherit_linklocal(idev, link_dev))
2461 return;
2462 }
2463 printk(KERN_DEBUG "init ip6-ip6: add_linklocal failed\n");
2464}
2465
2466/*
2467 * Autoconfigure tunnel with a link-local address so routing protocols,
2468 * DHCPv6, MLD etc. can be run over the virtual link
2469 */
2470
2471static void addrconf_ip6_tnl_config(struct net_device *dev)
2472{
2473 struct inet6_dev *idev;
2474
2475 ASSERT_RTNL();
2476
e21e8467 2477 idev = addrconf_add_dev(dev);
64e724f6 2478 if (IS_ERR(idev)) {
1da177e4
LT
2479 printk(KERN_DEBUG "init ip6-ip6: add_dev failed\n");
2480 return;
2481 }
2482 ip6_tnl_add_linklocal(idev);
1da177e4
LT
2483}
2484
1ab1457c 2485static int addrconf_notify(struct notifier_block *this, unsigned long event,
1da177e4
LT
2486 void * data)
2487{
2488 struct net_device *dev = (struct net_device *) data;
74235a25 2489 struct inet6_dev *idev = __in6_dev_get(dev);
c5e33bdd 2490 int run_pending = 0;
b217d616 2491 int err;
1da177e4 2492
e21e8467 2493 switch (event) {
45ba9dd2 2494 case NETDEV_REGISTER:
74235a25 2495 if (!idev && dev->mtu >= IPV6_MIN_MTU) {
45ba9dd2
YH
2496 idev = ipv6_add_dev(dev);
2497 if (!idev)
b217d616 2498 return notifier_from_errno(-ENOMEM);
45ba9dd2
YH
2499 }
2500 break;
bcdd553f 2501
1da177e4 2502 case NETDEV_UP:
3c21edbd 2503 case NETDEV_CHANGE:
c2edacf8
JV
2504 if (dev->flags & IFF_SLAVE)
2505 break;
2506
3c21edbd 2507 if (event == NETDEV_UP) {
f24e3d65 2508 if (!addrconf_qdisc_ok(dev)) {
3c21edbd
YH
2509 /* device is not ready yet. */
2510 printk(KERN_INFO
2511 "ADDRCONF(NETDEV_UP): %s: "
2512 "link is not ready\n",
2513 dev->name);
2514 break;
2515 }
99081049 2516
d31c7b8f
EP
2517 if (!idev && dev->mtu >= IPV6_MIN_MTU)
2518 idev = ipv6_add_dev(dev);
2519
e3ec6cfc 2520 if (idev) {
99081049 2521 idev->if_flags |= IF_READY;
e3ec6cfc
BT
2522 run_pending = 1;
2523 }
3c21edbd 2524 } else {
f24e3d65 2525 if (!addrconf_qdisc_ok(dev)) {
3c21edbd
YH
2526 /* device is still not ready. */
2527 break;
2528 }
2529
2530 if (idev) {
bcdd553f 2531 if (idev->if_flags & IF_READY)
3c21edbd
YH
2532 /* device is already configured. */
2533 break;
3c21edbd
YH
2534 idev->if_flags |= IF_READY;
2535 }
2536
2537 printk(KERN_INFO
2538 "ADDRCONF(NETDEV_CHANGE): %s: "
2539 "link becomes ready\n",
2540 dev->name);
2541
c5e33bdd 2542 run_pending = 1;
3c21edbd
YH
2543 }
2544
e21e8467 2545 switch (dev->type) {
0be669bb 2546#if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
1da177e4
LT
2547 case ARPHRD_SIT:
2548 addrconf_sit_config(dev);
2549 break;
0be669bb 2550#endif
1da177e4
LT
2551 case ARPHRD_TUNNEL6:
2552 addrconf_ip6_tnl_config(dev);
2553 break;
2554 case ARPHRD_LOOPBACK:
2555 init_loopback(dev);
2556 break;
2557
2558 default:
2559 addrconf_dev_config(dev);
2560 break;
3ff50b79 2561 }
bcdd553f 2562
1da177e4 2563 if (idev) {
c5e33bdd
YH
2564 if (run_pending)
2565 addrconf_dad_run(idev);
2566
bcdd553f
SH
2567 /*
2568 * If the MTU changed during the interface down,
2569 * when the interface up, the changed MTU must be
2570 * reflected in the idev as well as routers.
1da177e4 2571 */
bcdd553f
SH
2572 if (idev->cnf.mtu6 != dev->mtu &&
2573 dev->mtu >= IPV6_MIN_MTU) {
1da177e4
LT
2574 rt6_mtu_change(dev, dev->mtu);
2575 idev->cnf.mtu6 = dev->mtu;
2576 }
2577 idev->tstamp = jiffies;
2578 inet6_ifinfo_notify(RTM_NEWLINK, idev);
bcdd553f
SH
2579
2580 /*
2581 * If the changed mtu during down is lower than
2582 * IPV6_MIN_MTU stop IPv6 on this interface.
1da177e4
LT
2583 */
2584 if (dev->mtu < IPV6_MIN_MTU)
bcdd553f 2585 addrconf_ifdown(dev, 1);
1da177e4
LT
2586 }
2587 break;
2588
2589 case NETDEV_CHANGEMTU:
d31c7b8f 2590 if (idev && dev->mtu >= IPV6_MIN_MTU) {
1da177e4
LT
2591 rt6_mtu_change(dev, dev->mtu);
2592 idev->cnf.mtu6 = dev->mtu;
2593 break;
2594 }
2595
d31c7b8f
EP
2596 if (!idev && dev->mtu >= IPV6_MIN_MTU) {
2597 idev = ipv6_add_dev(dev);
2598 if (idev)
2599 break;
2600 }
2601
bcdd553f
SH
2602 /*
2603 * MTU falled under IPV6_MIN_MTU.
2604 * Stop IPv6 on this interface.
2605 */
1da177e4
LT
2606
2607 case NETDEV_DOWN:
2608 case NETDEV_UNREGISTER:
2609 /*
2610 * Remove all addresses from this interface.
2611 */
2612 addrconf_ifdown(dev, event != NETDEV_DOWN);
2613 break;
3c21edbd 2614
1da177e4 2615 case NETDEV_CHANGENAME:
1da177e4 2616 if (idev) {
5632c515 2617 snmp6_unregister_dev(idev);
408c4768 2618 addrconf_sysctl_unregister(idev);
f52295a9 2619 addrconf_sysctl_register(idev);
b217d616
HX
2620 err = snmp6_register_dev(idev);
2621 if (err)
2622 return notifier_from_errno(err);
5632c515 2623 }
1da177e4 2624 break;
bcdd553f 2625
93d9b7d7
JP
2626 case NETDEV_PRE_TYPE_CHANGE:
2627 case NETDEV_POST_TYPE_CHANGE:
2628 addrconf_type_change(dev, event);
75c78500 2629 break;
3ff50b79 2630 }
1da177e4
LT
2631
2632 return NOTIFY_OK;
2633}
2634
2635/*
2636 * addrconf module should be notified of a device going up
2637 */
2638static struct notifier_block ipv6_dev_notf = {
2639 .notifier_call = addrconf_notify,
1da177e4
LT
2640};
2641
93d9b7d7 2642static void addrconf_type_change(struct net_device *dev, unsigned long event)
75c78500
MS
2643{
2644 struct inet6_dev *idev;
2645 ASSERT_RTNL();
2646
2647 idev = __in6_dev_get(dev);
2648
93d9b7d7 2649 if (event == NETDEV_POST_TYPE_CHANGE)
75c78500 2650 ipv6_mc_remap(idev);
93d9b7d7 2651 else if (event == NETDEV_PRE_TYPE_CHANGE)
75c78500
MS
2652 ipv6_mc_unmap(idev);
2653}
2654
1da177e4
LT
2655static int addrconf_ifdown(struct net_device *dev, int how)
2656{
c346dca1 2657 struct net *net = dev_net(dev);
502a2ffd 2658 struct inet6_dev *idev;
2659 struct inet6_ifaddr *ifa;
2660 LIST_HEAD(keep_list);
4c5ff6a6 2661 int state;
1da177e4
LT
2662
2663 ASSERT_RTNL();
2664
f3db4851 2665 rt6_ifdown(net, dev);
1da177e4
LT
2666 neigh_ifdown(&nd_tbl, dev);
2667
2668 idev = __in6_dev_get(dev);
2669 if (idev == NULL)
2670 return -ENODEV;
2671
bcdd553f
SH
2672 /*
2673 * Step 1: remove reference to ipv6 device from parent device.
2674 * Do not dev_put!
1da177e4 2675 */
439e2385 2676 if (how) {
1da177e4 2677 idev->dead = 1;
8814c4b5
YH
2678
2679 /* protected by rtnl_lock */
2680 rcu_assign_pointer(dev->ip6_ptr, NULL);
1da177e4
LT
2681
2682 /* Step 1.5: remove snmp6 entry */
2683 snmp6_unregister_dev(idev);
2684
2685 }
2686
1da177e4
LT
2687 write_lock_bh(&idev->lock);
2688
bcdd553f 2689 /* Step 2: clear flags for stateless addrconf */
439e2385 2690 if (!how)
3c21edbd 2691 idev->if_flags &= ~(IF_RS_SENT|IF_RA_RCVD|IF_READY);
1da177e4 2692
1da177e4 2693#ifdef CONFIG_IPV6_PRIVACY
439e2385 2694 if (how && del_timer(&idev->regen_timer))
1da177e4
LT
2695 in6_dev_put(idev);
2696
bcdd553f 2697 /* Step 3: clear tempaddr list */
372e6c8f 2698 while (!list_empty(&idev->tempaddr_list)) {
2699 ifa = list_first_entry(&idev->tempaddr_list,
2700 struct inet6_ifaddr, tmp_list);
2701 list_del(&ifa->tmp_list);
1da177e4
LT
2702 write_unlock_bh(&idev->lock);
2703 spin_lock_bh(&ifa->lock);
2704
2705 if (ifa->ifpub) {
2706 in6_ifa_put(ifa->ifpub);
2707 ifa->ifpub = NULL;
2708 }
2709 spin_unlock_bh(&ifa->lock);
2710 in6_ifa_put(ifa);
2711 write_lock_bh(&idev->lock);
2712 }
2713#endif
8f37ada5 2714
502a2ffd 2715 while (!list_empty(&idev->addr_list)) {
2716 ifa = list_first_entry(&idev->addr_list,
2717 struct inet6_ifaddr, if_list);
84e8b803 2718 addrconf_del_timer(ifa);
2719
2720 /* If just doing link down, and address is permanent
2721 and not link-local, then retain it. */
bcdd553f 2722 if (!how &&
84e8b803 2723 (ifa->flags&IFA_F_PERMANENT) &&
2724 !(ipv6_addr_type(&ifa->addr) & IPV6_ADDR_LINKLOCAL)) {
502a2ffd 2725 list_move_tail(&ifa->if_list, &keep_list);
dc2b99f7 2726
84e8b803 2727 /* If not doing DAD on this address, just keep it. */
2728 if ((dev->flags&(IFF_NOARP|IFF_LOOPBACK)) ||
2729 idev->cnf.accept_dad <= 0 ||
2730 (ifa->flags & IFA_F_NODAD))
2731 continue;
2732
2733 /* If it was tentative already, no need to notify */
2734 if (ifa->flags & IFA_F_TENTATIVE)
2735 continue;
2736
2737 /* Flag it for later restoration when link comes up */
2738 ifa->flags |= IFA_F_TENTATIVE;
f2344a13 2739 ifa->state = INET6_IFADDR_STATE_DAD;
4c5ff6a6 2740
27bdb2ab 2741 write_unlock_bh(&idev->lock);
4c5ff6a6
HX
2742
2743 in6_ifa_hold(ifa);
dc2b99f7 2744 } else {
502a2ffd 2745 list_del(&ifa->if_list);
1da177e4 2746
27bdb2ab 2747 /* clear hash table */
2748 spin_lock_bh(&addrconf_hash_lock);
2749 hlist_del_init_rcu(&ifa->addr_lst);
27bdb2ab 2750 spin_unlock_bh(&addrconf_hash_lock);
4c5ff6a6
HX
2751
2752 write_unlock_bh(&idev->lock);
2753 spin_lock_bh(&ifa->state_lock);
2754 state = ifa->state;
2755 ifa->state = INET6_IFADDR_STATE_DEAD;
2756 spin_unlock_bh(&ifa->state_lock);
2757
2758 if (state == INET6_IFADDR_STATE_DEAD)
2759 goto put_ifa;
27bdb2ab 2760 }
c2e21293 2761
84e8b803 2762 __ipv6_ifa_notify(RTM_DELADDR, ifa);
e9d3e084 2763 if (ifa->state == INET6_IFADDR_STATE_DEAD)
8595805a 2764 atomic_notifier_call_chain(&inet6addr_chain,
2765 NETDEV_DOWN, ifa);
4c5ff6a6
HX
2766
2767put_ifa:
84e8b803 2768 in6_ifa_put(ifa);
dc2b99f7 2769
84e8b803 2770 write_lock_bh(&idev->lock);
1da177e4 2771 }
8f37ada5 2772
502a2ffd 2773 list_splice(&keep_list, &idev->addr_list);
8f37ada5 2774
1da177e4
LT
2775 write_unlock_bh(&idev->lock);
2776
2777 /* Step 5: Discard multicast list */
439e2385 2778 if (how)
1da177e4
LT
2779 ipv6_mc_destroy_dev(idev);
2780 else
2781 ipv6_mc_down(idev);
2782
1da177e4 2783 idev->tstamp = jiffies;
1ab1457c 2784
bcdd553f 2785 /* Last: Shot the device (if unregistered) */
439e2385 2786 if (how) {
408c4768 2787 addrconf_sysctl_unregister(idev);
1da177e4
LT
2788 neigh_parms_release(&nd_tbl, idev->nd_parms);
2789 neigh_ifdown(&nd_tbl, dev);
2790 in6_dev_put(idev);
2791 }
2792 return 0;
2793}
2794
2795static void addrconf_rs_timer(unsigned long data)
2796{
2797 struct inet6_ifaddr *ifp = (struct inet6_ifaddr *) data;
5b2a1953 2798 struct inet6_dev *idev = ifp->idev;
1da177e4 2799
5b2a1953 2800 read_lock(&idev->lock);
2801 if (idev->dead || !(idev->if_flags & IF_READY))
1da177e4
LT
2802 goto out;
2803
5b2a1953 2804 if (idev->cnf.forwarding)
2805 goto out;
2806
2807 /* Announcement received after solicitation was sent */
2808 if (idev->if_flags & IF_RA_RCVD)
1da177e4 2809 goto out;
1da177e4
LT
2810
2811 spin_lock(&ifp->lock);
5b2a1953 2812 if (ifp->probes++ < idev->cnf.rtr_solicits) {
1da177e4
LT
2813 /* The wait after the last probe can be shorter */
2814 addrconf_mod_timer(ifp, AC_RS,
5b2a1953 2815 (ifp->probes == idev->cnf.rtr_solicits) ?
2816 idev->cnf.rtr_solicit_delay :
2817 idev->cnf.rtr_solicit_interval);
1da177e4
LT
2818 spin_unlock(&ifp->lock);
2819
5b2a1953 2820 ndisc_send_rs(idev->dev, &ifp->addr, &in6addr_linklocal_allrouters);
1da177e4
LT
2821 } else {
2822 spin_unlock(&ifp->lock);
2823 /*
2824 * Note: we do not support deprecated "all on-link"
2825 * assumption any longer.
2826 */
2827 printk(KERN_DEBUG "%s: no IPv6 routers present\n",
5b2a1953 2828 idev->dev->name);
1da177e4
LT
2829 }
2830
2831out:
5b2a1953 2832 read_unlock(&idev->lock);
1da177e4
LT
2833 in6_ifa_put(ifp);
2834}
2835
2836/*
2837 * Duplicate Address Detection
2838 */
3c21edbd
YH
2839static void addrconf_dad_kick(struct inet6_ifaddr *ifp)
2840{
2841 unsigned long rand_num;
2842 struct inet6_dev *idev = ifp->idev;
2843
95c385b4
NH
2844 if (ifp->flags & IFA_F_OPTIMISTIC)
2845 rand_num = 0;
2846 else
2847 rand_num = net_random() % (idev->cnf.rtr_solicit_delay ? : 1);
2848
3c21edbd
YH
2849 ifp->probes = idev->cnf.dad_transmits;
2850 addrconf_mod_timer(ifp, AC_DAD, rand_num);
2851}
2852
e431b8c0 2853static void addrconf_dad_start(struct inet6_ifaddr *ifp, u32 flags)
1da177e4
LT
2854{
2855 struct inet6_dev *idev = ifp->idev;
2856 struct net_device *dev = idev->dev;
1da177e4
LT
2857
2858 addrconf_join_solict(dev, &ifp->addr);
2859
1da177e4 2860 net_srandom(ifp->addr.s6_addr32[3]);
1da177e4
LT
2861
2862 read_lock_bh(&idev->lock);
622ccdf1 2863 spin_lock(&ifp->lock);
e9d3e084 2864 if (ifp->state == INET6_IFADDR_STATE_DEAD)
1da177e4 2865 goto out;
1da177e4
LT
2866
2867 if (dev->flags&(IFF_NOARP|IFF_LOOPBACK) ||
1b34be74 2868 idev->cnf.accept_dad < 1 ||
55ebaef1
NT
2869 !(ifp->flags&IFA_F_TENTATIVE) ||
2870 ifp->flags & IFA_F_NODAD) {
cc411d0b 2871 ifp->flags &= ~(IFA_F_TENTATIVE|IFA_F_OPTIMISTIC|IFA_F_DADFAILED);
21809faf 2872 spin_unlock(&ifp->lock);
1da177e4
LT
2873 read_unlock_bh(&idev->lock);
2874
2875 addrconf_dad_completed(ifp);
2876 return;
2877 }
2878
6732bade 2879 if (!(idev->if_flags & IF_READY)) {
21809faf 2880 spin_unlock(&ifp->lock);
6732bade 2881 read_unlock_bh(&idev->lock);
3c21edbd 2882 /*
590a9887 2883 * If the device is not ready:
3c21edbd
YH
2884 * - keep it tentative if it is a permanent address.
2885 * - otherwise, kill it.
2886 */
2887 in6_ifa_hold(ifp);
cc411d0b 2888 addrconf_dad_stop(ifp, 0);
6732bade 2889 return;
3c21edbd 2890 }
95c385b4
NH
2891
2892 /*
2893 * Optimistic nodes can start receiving
2894 * Frames right away
2895 */
e21e8467 2896 if (ifp->flags & IFA_F_OPTIMISTIC)
95c385b4
NH
2897 ip6_ins_rt(ifp->rt);
2898
6732bade 2899 addrconf_dad_kick(ifp);
1da177e4 2900out:
622ccdf1 2901 spin_unlock(&ifp->lock);
1da177e4
LT
2902 read_unlock_bh(&idev->lock);
2903}
2904
2905static void addrconf_dad_timer(unsigned long data)
2906{
2907 struct inet6_ifaddr *ifp = (struct inet6_ifaddr *) data;
2908 struct inet6_dev *idev = ifp->idev;
1da177e4
LT
2909 struct in6_addr mcaddr;
2910
f2344a13
HX
2911 if (!ifp->probes && addrconf_dad_end(ifp))
2912 goto out;
2913
122e4519 2914 read_lock(&idev->lock);
2915 if (idev->dead || !(idev->if_flags & IF_READY)) {
2916 read_unlock(&idev->lock);
1da177e4
LT
2917 goto out;
2918 }
21809faf 2919
2920 spin_lock(&ifp->lock);
622ccdf1
HX
2921 if (ifp->state == INET6_IFADDR_STATE_DEAD) {
2922 spin_unlock(&ifp->lock);
2923 read_unlock(&idev->lock);
2924 goto out;
2925 }
2926
1da177e4
LT
2927 if (ifp->probes == 0) {
2928 /*
2929 * DAD was successful
2930 */
2931
cc411d0b 2932 ifp->flags &= ~(IFA_F_TENTATIVE|IFA_F_OPTIMISTIC|IFA_F_DADFAILED);
21809faf 2933 spin_unlock(&ifp->lock);
122e4519 2934 read_unlock(&idev->lock);
1da177e4
LT
2935
2936 addrconf_dad_completed(ifp);
2937
2938 goto out;
2939 }
2940
2941 ifp->probes--;
2942 addrconf_mod_timer(ifp, AC_DAD, ifp->idev->nd_parms->retrans_time);
21809faf 2943 spin_unlock(&ifp->lock);
122e4519 2944 read_unlock(&idev->lock);
1da177e4
LT
2945
2946 /* send a neighbour solicitation for our addr */
1da177e4 2947 addrconf_addr_solict_mult(&ifp->addr, &mcaddr);
d7aabf22 2948 ndisc_send_ns(ifp->idev->dev, NULL, &ifp->addr, &mcaddr, &in6addr_any);
1da177e4
LT
2949out:
2950 in6_ifa_put(ifp);
2951}
2952
2953static void addrconf_dad_completed(struct inet6_ifaddr *ifp)
2954{
e21e8467 2955 struct net_device *dev = ifp->idev->dev;
1da177e4
LT
2956
2957 /*
2958 * Configure the address for reception. Now it is valid.
2959 */
2960
2961 ipv6_ifa_notify(RTM_NEWADDR, ifp);
2962
2963 /* If added prefix is link local and forwarding is off,
2964 start sending router solicitations.
2965 */
2966
c3bccac2
TG
2967 if ((ifp->idev->cnf.forwarding == 0 ||
2968 ifp->idev->cnf.forwarding == 2) &&
1da177e4
LT
2969 ifp->idev->cnf.rtr_solicits > 0 &&
2970 (dev->flags&IFF_LOOPBACK) == 0 &&
2971 (ipv6_addr_type(&ifp->addr) & IPV6_ADDR_LINKLOCAL)) {
1da177e4
LT
2972 /*
2973 * If a host as already performed a random delay
2974 * [...] as part of DAD [...] there is no need
2975 * to delay again before sending the first RS
2976 */
f3ee4010 2977 ndisc_send_rs(ifp->idev->dev, &ifp->addr, &in6addr_linklocal_allrouters);
1da177e4
LT
2978
2979 spin_lock_bh(&ifp->lock);
2980 ifp->probes = 1;
2981 ifp->idev->if_flags |= IF_RS_SENT;
2982 addrconf_mod_timer(ifp, AC_RS, ifp->idev->cnf.rtr_solicit_interval);
2983 spin_unlock_bh(&ifp->lock);
2984 }
2985}
2986
e21e8467
SH
2987static void addrconf_dad_run(struct inet6_dev *idev)
2988{
c5e33bdd
YH
2989 struct inet6_ifaddr *ifp;
2990
2991 read_lock_bh(&idev->lock);
502a2ffd 2992 list_for_each_entry(ifp, &idev->addr_list, if_list) {
21809faf 2993 spin_lock(&ifp->lock);
f2344a13
HX
2994 if (ifp->flags & IFA_F_TENTATIVE &&
2995 ifp->state == INET6_IFADDR_STATE_DAD)
2996 addrconf_dad_kick(ifp);
21809faf 2997 spin_unlock(&ifp->lock);
c5e33bdd
YH
2998 }
2999 read_unlock_bh(&idev->lock);
3000}
3001
1da177e4
LT
3002#ifdef CONFIG_PROC_FS
3003struct if6_iter_state {
3c40090a 3004 struct seq_net_private p;
1da177e4
LT
3005 int bucket;
3006};
3007
3008static struct inet6_ifaddr *if6_get_first(struct seq_file *seq)
3009{
3010 struct inet6_ifaddr *ifa = NULL;
3011 struct if6_iter_state *state = seq->private;
1218854a 3012 struct net *net = seq_file_net(seq);
1da177e4
LT
3013
3014 for (state->bucket = 0; state->bucket < IN6_ADDR_HSIZE; ++state->bucket) {
c2e21293 3015 struct hlist_node *n;
4f70ecca 3016 hlist_for_each_entry_rcu_bh(ifa, n, &inet6_addr_lst[state->bucket],
5c578aed 3017 addr_lst)
c2e21293 3018 if (net_eq(dev_net(ifa->idev->dev), net))
3019 return ifa;
1da177e4 3020 }
c2e21293 3021 return NULL;
1da177e4
LT
3022}
3023
c2e21293 3024static struct inet6_ifaddr *if6_get_next(struct seq_file *seq,
3025 struct inet6_ifaddr *ifa)
1da177e4
LT
3026{
3027 struct if6_iter_state *state = seq->private;
1218854a 3028 struct net *net = seq_file_net(seq);
c2e21293 3029 struct hlist_node *n = &ifa->addr_lst;
1da177e4 3030
4f70ecca 3031 hlist_for_each_entry_continue_rcu_bh(ifa, n, addr_lst)
c2e21293 3032 if (net_eq(dev_net(ifa->idev->dev), net))
3033 return ifa;
3c40090a 3034
c2e21293 3035 while (++state->bucket < IN6_ADDR_HSIZE) {
4f70ecca 3036 hlist_for_each_entry_rcu_bh(ifa, n,
c2e21293 3037 &inet6_addr_lst[state->bucket], addr_lst) {
3038 if (net_eq(dev_net(ifa->idev->dev), net))
3039 return ifa;
3040 }
1da177e4 3041 }
3c40090a 3042
c2e21293 3043 return NULL;
1da177e4
LT
3044}
3045
3046static struct inet6_ifaddr *if6_get_idx(struct seq_file *seq, loff_t pos)
3047{
3048 struct inet6_ifaddr *ifa = if6_get_first(seq);
3049
3050 if (ifa)
e21e8467 3051 while (pos && (ifa = if6_get_next(seq, ifa)) != NULL)
1da177e4
LT
3052 --pos;
3053 return pos ? NULL : ifa;
3054}
3055
3056static void *if6_seq_start(struct seq_file *seq, loff_t *pos)
4f70ecca 3057 __acquires(rcu_bh)
1da177e4 3058{
5c578aed 3059 rcu_read_lock_bh();
1da177e4
LT
3060 return if6_get_idx(seq, *pos);
3061}
3062
3063static void *if6_seq_next(struct seq_file *seq, void *v, loff_t *pos)
3064{
3065 struct inet6_ifaddr *ifa;
3066
3067 ifa = if6_get_next(seq, v);
3068 ++*pos;
3069 return ifa;
3070}
3071
3072static void if6_seq_stop(struct seq_file *seq, void *v)
4f70ecca 3073 __releases(rcu_bh)
1da177e4 3074{
5c578aed 3075 rcu_read_unlock_bh();
1da177e4
LT
3076}
3077
3078static int if6_seq_show(struct seq_file *seq, void *v)
3079{
3080 struct inet6_ifaddr *ifp = (struct inet6_ifaddr *)v;
4b7a4274 3081 seq_printf(seq, "%pi6 %02x %02x %02x %02x %8s\n",
b071195d 3082 &ifp->addr,
1da177e4
LT
3083 ifp->idev->dev->ifindex,
3084 ifp->prefix_len,
3085 ifp->scope,
3086 ifp->flags,
3087 ifp->idev->dev->name);
3088 return 0;
3089}
3090
56b3d975 3091static const struct seq_operations if6_seq_ops = {
1da177e4
LT
3092 .start = if6_seq_start,
3093 .next = if6_seq_next,
3094 .show = if6_seq_show,
3095 .stop = if6_seq_stop,
3096};
3097
3098static int if6_seq_open(struct inode *inode, struct file *file)
3099{
3c40090a
DL
3100 return seq_open_net(inode, file, &if6_seq_ops,
3101 sizeof(struct if6_iter_state));
1da177e4
LT
3102}
3103
9a32144e 3104static const struct file_operations if6_fops = {
1da177e4
LT
3105 .owner = THIS_MODULE,
3106 .open = if6_seq_open,
3107 .read = seq_read,
3108 .llseek = seq_lseek,
3c40090a 3109 .release = seq_release_net,
1da177e4
LT
3110};
3111
2c8c1e72 3112static int __net_init if6_proc_net_init(struct net *net)
1da177e4 3113{
3c40090a 3114 if (!proc_net_fops_create(net, "if_inet6", S_IRUGO, &if6_fops))
1da177e4
LT
3115 return -ENOMEM;
3116 return 0;
3117}
3118
2c8c1e72 3119static void __net_exit if6_proc_net_exit(struct net *net)
3c40090a
DL
3120{
3121 proc_net_remove(net, "if_inet6");
3122}
3123
3124static struct pernet_operations if6_proc_net_ops = {
3125 .init = if6_proc_net_init,
3126 .exit = if6_proc_net_exit,
3127};
3128
3129int __init if6_proc_init(void)
3130{
3131 return register_pernet_subsys(&if6_proc_net_ops);
3132}
3133
1da177e4
LT
3134void if6_proc_exit(void)
3135{
3c40090a 3136 unregister_pernet_subsys(&if6_proc_net_ops);
1da177e4
LT
3137}
3138#endif /* CONFIG_PROC_FS */
3139
59fbb3a6 3140#if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
3b9f9a1c 3141/* Check if address is a home address configured on any interface. */
389f6612 3142int ipv6_chk_home_addr(struct net *net, struct in6_addr *addr)
3b9f9a1c
NT
3143{
3144 int ret = 0;
c2e21293 3145 struct inet6_ifaddr *ifp = NULL;
3146 struct hlist_node *n;
3a88a81d 3147 unsigned int hash = ipv6_addr_hash(addr);
c2e21293 3148
5c578aed 3149 rcu_read_lock_bh();
4f70ecca 3150 hlist_for_each_entry_rcu_bh(ifp, n, &inet6_addr_lst[hash], addr_lst) {
878628fb 3151 if (!net_eq(dev_net(ifp->idev->dev), net))
389f6612 3152 continue;
caad295f 3153 if (ipv6_addr_equal(&ifp->addr, addr) &&
3b9f9a1c
NT
3154 (ifp->flags & IFA_F_HOMEADDRESS)) {
3155 ret = 1;
3156 break;
3157 }
3158 }
5c578aed 3159 rcu_read_unlock_bh();
3b9f9a1c
NT
3160 return ret;
3161}
3162#endif
3163
1da177e4
LT
3164/*
3165 * Periodic address status verification
3166 */
3167
3168static void addrconf_verify(unsigned long foo)
3169{
b2db7564 3170 unsigned long now, next, next_sec, next_sched;
1da177e4 3171 struct inet6_ifaddr *ifp;
c2e21293 3172 struct hlist_node *node;
1da177e4
LT
3173 int i;
3174
5c578aed 3175 rcu_read_lock_bh();
3176 spin_lock(&addrconf_verify_lock);
1da177e4 3177 now = jiffies;
b2db7564 3178 next = round_jiffies_up(now + ADDR_CHECK_FREQUENCY);
1da177e4
LT
3179
3180 del_timer(&addr_chk_timer);
3181
bcdd553f 3182 for (i = 0; i < IN6_ADDR_HSIZE; i++) {
1da177e4 3183restart:
4f70ecca 3184 hlist_for_each_entry_rcu_bh(ifp, node,
5c578aed 3185 &inet6_addr_lst[i], addr_lst) {
1da177e4 3186 unsigned long age;
1da177e4
LT
3187
3188 if (ifp->flags & IFA_F_PERMANENT)
3189 continue;
3190
3191 spin_lock(&ifp->lock);
b2db7564
YH
3192 /* We try to batch several events at once. */
3193 age = (now - ifp->tstamp + ADDRCONF_TIMER_FUZZ_MINUS) / HZ;
1da177e4 3194
8f27ebb9
YH
3195 if (ifp->valid_lft != INFINITY_LIFE_TIME &&
3196 age >= ifp->valid_lft) {
1da177e4
LT
3197 spin_unlock(&ifp->lock);
3198 in6_ifa_hold(ifp);
1da177e4
LT
3199 ipv6_del_addr(ifp);
3200 goto restart;
8f27ebb9
YH
3201 } else if (ifp->prefered_lft == INFINITY_LIFE_TIME) {
3202 spin_unlock(&ifp->lock);
3203 continue;
1da177e4 3204 } else if (age >= ifp->prefered_lft) {
a1ed0526 3205 /* jiffies - ifp->tstamp > age >= ifp->prefered_lft */
1da177e4
LT
3206 int deprecate = 0;
3207
3208 if (!(ifp->flags&IFA_F_DEPRECATED)) {
3209 deprecate = 1;
3210 ifp->flags |= IFA_F_DEPRECATED;
3211 }
3212
3213 if (time_before(ifp->tstamp + ifp->valid_lft * HZ, next))
3214 next = ifp->tstamp + ifp->valid_lft * HZ;
3215
3216 spin_unlock(&ifp->lock);
3217
3218 if (deprecate) {
3219 in6_ifa_hold(ifp);
1da177e4
LT
3220
3221 ipv6_ifa_notify(0, ifp);
3222 in6_ifa_put(ifp);
3223 goto restart;
3224 }
3225#ifdef CONFIG_IPV6_PRIVACY
3226 } else if ((ifp->flags&IFA_F_TEMPORARY) &&
3227 !(ifp->flags&IFA_F_TENTATIVE)) {
88949cf4 3228 unsigned long regen_advance = ifp->idev->cnf.regen_max_retry *
3229 ifp->idev->cnf.dad_transmits *
3230 ifp->idev->nd_parms->retrans_time / HZ;
3231
1da177e4
LT
3232 if (age >= ifp->prefered_lft - regen_advance) {
3233 struct inet6_ifaddr *ifpub = ifp->ifpub;
3234 if (time_before(ifp->tstamp + ifp->prefered_lft * HZ, next))
3235 next = ifp->tstamp + ifp->prefered_lft * HZ;
3236 if (!ifp->regen_count && ifpub) {
3237 ifp->regen_count++;
3238 in6_ifa_hold(ifp);
3239 in6_ifa_hold(ifpub);
3240 spin_unlock(&ifp->lock);
5c578aed 3241
291d809b
HY
3242 spin_lock(&ifpub->lock);
3243 ifpub->regen_count = 0;
3244 spin_unlock(&ifpub->lock);
1da177e4
LT
3245 ipv6_create_tempaddr(ifpub, ifp);
3246 in6_ifa_put(ifpub);
3247 in6_ifa_put(ifp);
3248 goto restart;
3249 }
3250 } else if (time_before(ifp->tstamp + ifp->prefered_lft * HZ - regen_advance * HZ, next))
3251 next = ifp->tstamp + ifp->prefered_lft * HZ - regen_advance * HZ;
3252 spin_unlock(&ifp->lock);
3253#endif
3254 } else {
3255 /* ifp->prefered_lft <= ifp->valid_lft */
3256 if (time_before(ifp->tstamp + ifp->prefered_lft * HZ, next))
3257 next = ifp->tstamp + ifp->prefered_lft * HZ;
3258 spin_unlock(&ifp->lock);
3259 }
3260 }
1da177e4
LT
3261 }
3262
b2db7564
YH
3263 next_sec = round_jiffies_up(next);
3264 next_sched = next;
3265
3266 /* If rounded timeout is accurate enough, accept it. */
3267 if (time_before(next_sec, next + ADDRCONF_TIMER_FUZZ))
3268 next_sched = next_sec;
3269
3270 /* And minimum interval is ADDRCONF_TIMER_FUZZ_MAX. */
3271 if (time_before(next_sched, jiffies + ADDRCONF_TIMER_FUZZ_MAX))
3272 next_sched = jiffies + ADDRCONF_TIMER_FUZZ_MAX;
3273
3274 ADBG((KERN_DEBUG "now = %lu, schedule = %lu, rounded schedule = %lu => %lu\n",
3275 now, next, next_sec, next_sched));
3276
3277 addr_chk_timer.expires = next_sched;
1da177e4 3278 add_timer(&addr_chk_timer);
5c578aed 3279 spin_unlock(&addrconf_verify_lock);
3280 rcu_read_unlock_bh();
1da177e4
LT
3281}
3282
461d8837
TG
3283static struct in6_addr *extract_addr(struct nlattr *addr, struct nlattr *local)
3284{
3285 struct in6_addr *pfx = NULL;
3286
3287 if (addr)
3288 pfx = nla_data(addr);
3289
3290 if (local) {
3291 if (pfx && nla_memcmp(local, pfx, sizeof(*pfx)))
3292 pfx = NULL;
3293 else
3294 pfx = nla_data(local);
3295 }
3296
3297 return pfx;
3298}
3299
ef7c79ed 3300static const struct nla_policy ifa_ipv6_policy[IFA_MAX+1] = {
461d8837
TG
3301 [IFA_ADDRESS] = { .len = sizeof(struct in6_addr) },
3302 [IFA_LOCAL] = { .len = sizeof(struct in6_addr) },
3303 [IFA_CACHEINFO] = { .len = sizeof(struct ifa_cacheinfo) },
3304};
3305
1da177e4
LT
3306static int
3307inet6_rtm_deladdr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
3308{
3b1e0a65 3309 struct net *net = sock_net(skb->sk);
b933f716
TG
3310 struct ifaddrmsg *ifm;
3311 struct nlattr *tb[IFA_MAX+1];
1da177e4 3312 struct in6_addr *pfx;
b933f716 3313 int err;
1da177e4 3314
b933f716
TG
3315 err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv6_policy);
3316 if (err < 0)
3317 return err;
3318
3319 ifm = nlmsg_data(nlh);
3320 pfx = extract_addr(tb[IFA_ADDRESS], tb[IFA_LOCAL]);
1da177e4
LT
3321 if (pfx == NULL)
3322 return -EINVAL;
3323
af284937 3324 return inet6_addr_del(net, ifm->ifa_index, pfx, ifm->ifa_prefixlen);
1da177e4
LT
3325}
3326
55ebaef1
NT
3327static int inet6_addr_modify(struct inet6_ifaddr *ifp, u8 ifa_flags,
3328 u32 prefered_lft, u32 valid_lft)
081bba5b 3329{
6f704992
YH
3330 u32 flags;
3331 clock_t expires;
4bed72e4 3332 unsigned long timeout;
46d48046 3333
081bba5b
NT
3334 if (!valid_lft || (prefered_lft > valid_lft))
3335 return -EINVAL;
3336
4bed72e4
YH
3337 timeout = addrconf_timeout_fixup(valid_lft, HZ);
3338 if (addrconf_finite_timeout(timeout)) {
3339 expires = jiffies_to_clock_t(timeout * HZ);
3340 valid_lft = timeout;
6f704992 3341 flags = RTF_EXPIRES;
4bed72e4
YH
3342 } else {
3343 expires = 0;
3344 flags = 0;
3345 ifa_flags |= IFA_F_PERMANENT;
6f704992 3346 }
081bba5b 3347
4bed72e4
YH
3348 timeout = addrconf_timeout_fixup(prefered_lft, HZ);
3349 if (addrconf_finite_timeout(timeout)) {
3350 if (timeout == 0)
3351 ifa_flags |= IFA_F_DEPRECATED;
3352 prefered_lft = timeout;
3353 }
081bba5b
NT
3354
3355 spin_lock_bh(&ifp->lock);
3b9f9a1c 3356 ifp->flags = (ifp->flags & ~(IFA_F_DEPRECATED | IFA_F_PERMANENT | IFA_F_NODAD | IFA_F_HOMEADDRESS)) | ifa_flags;
081bba5b
NT
3357 ifp->tstamp = jiffies;
3358 ifp->valid_lft = valid_lft;
3359 ifp->prefered_lft = prefered_lft;
3360
3361 spin_unlock_bh(&ifp->lock);
3362 if (!(ifp->flags&IFA_F_TENTATIVE))
3363 ipv6_ifa_notify(0, ifp);
081bba5b 3364
46d48046 3365 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, ifp->idev->dev,
6f704992 3366 expires, flags);
081bba5b
NT
3367 addrconf_verify(0);
3368
3369 return 0;
3370}
3371
1da177e4
LT
3372static int
3373inet6_rtm_newaddr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
3374{
3b1e0a65 3375 struct net *net = sock_net(skb->sk);
461d8837
TG
3376 struct ifaddrmsg *ifm;
3377 struct nlattr *tb[IFA_MAX+1];
1da177e4 3378 struct in6_addr *pfx;
7198f8ce
TG
3379 struct inet6_ifaddr *ifa;
3380 struct net_device *dev;
55ebaef1
NT
3381 u32 valid_lft = INFINITY_LIFE_TIME, preferred_lft = INFINITY_LIFE_TIME;
3382 u8 ifa_flags;
461d8837 3383 int err;
1da177e4 3384
461d8837
TG
3385 err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv6_policy);
3386 if (err < 0)
3387 return err;
3388
3389 ifm = nlmsg_data(nlh);
3390 pfx = extract_addr(tb[IFA_ADDRESS], tb[IFA_LOCAL]);
1da177e4
LT
3391 if (pfx == NULL)
3392 return -EINVAL;
3393
461d8837 3394 if (tb[IFA_CACHEINFO]) {
0778769d 3395 struct ifa_cacheinfo *ci;
461d8837
TG
3396
3397 ci = nla_data(tb[IFA_CACHEINFO]);
0778769d 3398 valid_lft = ci->ifa_valid;
461d8837
TG
3399 preferred_lft = ci->ifa_prefered;
3400 } else {
3401 preferred_lft = INFINITY_LIFE_TIME;
3402 valid_lft = INFINITY_LIFE_TIME;
0778769d
NT
3403 }
3404
af284937 3405 dev = __dev_get_by_index(net, ifm->ifa_index);
7198f8ce
TG
3406 if (dev == NULL)
3407 return -ENODEV;
3408
55ebaef1 3409 /* We ignore other flags so far. */
3b9f9a1c 3410 ifa_flags = ifm->ifa_flags & (IFA_F_NODAD | IFA_F_HOMEADDRESS);
55ebaef1 3411
1cab3da6 3412 ifa = ipv6_get_ifaddr(net, pfx, dev, 1);
7198f8ce
TG
3413 if (ifa == NULL) {
3414 /*
3415 * It would be best to check for !NLM_F_CREATE here but
3416 * userspace alreay relies on not having to provide this.
3417 */
af284937
DL
3418 return inet6_addr_add(net, ifm->ifa_index, pfx,
3419 ifm->ifa_prefixlen, ifa_flags,
3420 preferred_lft, valid_lft);
081bba5b
NT
3421 }
3422
7198f8ce
TG
3423 if (nlh->nlmsg_flags & NLM_F_EXCL ||
3424 !(nlh->nlmsg_flags & NLM_F_REPLACE))
3425 err = -EEXIST;
3426 else
55ebaef1 3427 err = inet6_addr_modify(ifa, ifa_flags, preferred_lft, valid_lft);
7198f8ce
TG
3428
3429 in6_ifa_put(ifa);
3430
3431 return err;
1da177e4
LT
3432}
3433
101bb229
TG
3434static void put_ifaddrmsg(struct nlmsghdr *nlh, u8 prefixlen, u8 flags,
3435 u8 scope, int ifindex)
3436{
3437 struct ifaddrmsg *ifm;
3438
3439 ifm = nlmsg_data(nlh);
3440 ifm->ifa_family = AF_INET6;
3441 ifm->ifa_prefixlen = prefixlen;
3442 ifm->ifa_flags = flags;
3443 ifm->ifa_scope = scope;
3444 ifm->ifa_index = ifindex;
3445}
3446
85486af0
TG
3447static int put_cacheinfo(struct sk_buff *skb, unsigned long cstamp,
3448 unsigned long tstamp, u32 preferred, u32 valid)
3449{
3450 struct ifa_cacheinfo ci;
3451
3452 ci.cstamp = (u32)(TIME_DELTA(cstamp, INITIAL_JIFFIES) / HZ * 100
3453 + TIME_DELTA(cstamp, INITIAL_JIFFIES) % HZ * 100 / HZ);
3454 ci.tstamp = (u32)(TIME_DELTA(tstamp, INITIAL_JIFFIES) / HZ * 100
3455 + TIME_DELTA(tstamp, INITIAL_JIFFIES) % HZ * 100 / HZ);
3456 ci.ifa_prefered = preferred;
3457 ci.ifa_valid = valid;
3458
3459 return nla_put(skb, IFA_CACHEINFO, sizeof(ci), &ci);
3460}
3461
101bb229
TG
3462static inline int rt_scope(int ifa_scope)
3463{
3464 if (ifa_scope & IFA_HOST)
3465 return RT_SCOPE_HOST;
3466 else if (ifa_scope & IFA_LINK)
3467 return RT_SCOPE_LINK;
3468 else if (ifa_scope & IFA_SITE)
3469 return RT_SCOPE_SITE;
3470 else
3471 return RT_SCOPE_UNIVERSE;
3472}
3473
0ab6803b
TG
3474static inline int inet6_ifaddr_msgsize(void)
3475{
339bf98f
TG
3476 return NLMSG_ALIGN(sizeof(struct ifaddrmsg))
3477 + nla_total_size(16) /* IFA_ADDRESS */
3478 + nla_total_size(sizeof(struct ifa_cacheinfo));
0ab6803b 3479}
c5396a31 3480
1da177e4 3481static int inet6_fill_ifaddr(struct sk_buff *skb, struct inet6_ifaddr *ifa,
b6544c0b 3482 u32 pid, u32 seq, int event, unsigned int flags)
1da177e4 3483{
1da177e4 3484 struct nlmsghdr *nlh;
85486af0 3485 u32 preferred, valid;
1da177e4 3486
0ab6803b
TG
3487 nlh = nlmsg_put(skb, pid, seq, event, sizeof(struct ifaddrmsg), flags);
3488 if (nlh == NULL)
26932566 3489 return -EMSGSIZE;
0ab6803b 3490
101bb229
TG
3491 put_ifaddrmsg(nlh, ifa->prefix_len, ifa->flags, rt_scope(ifa->scope),
3492 ifa->idev->dev->ifindex);
3493
1da177e4 3494 if (!(ifa->flags&IFA_F_PERMANENT)) {
85486af0
TG
3495 preferred = ifa->prefered_lft;
3496 valid = ifa->valid_lft;
3497 if (preferred != INFINITY_LIFE_TIME) {
1da177e4 3498 long tval = (jiffies - ifa->tstamp)/HZ;
a1faa698
JR
3499 if (preferred > tval)
3500 preferred -= tval;
3501 else
3502 preferred = 0;
f56619fc
BH
3503 if (valid != INFINITY_LIFE_TIME) {
3504 if (valid > tval)
3505 valid -= tval;
3506 else
3507 valid = 0;
3508 }
1da177e4
LT
3509 }
3510 } else {
85486af0
TG
3511 preferred = INFINITY_LIFE_TIME;
3512 valid = INFINITY_LIFE_TIME;
3513 }
3514
0ab6803b 3515 if (nla_put(skb, IFA_ADDRESS, 16, &ifa->addr) < 0 ||
26932566
PM
3516 put_cacheinfo(skb, ifa->cstamp, ifa->tstamp, preferred, valid) < 0) {
3517 nlmsg_cancel(skb, nlh);
3518 return -EMSGSIZE;
3519 }
1da177e4 3520
0ab6803b 3521 return nlmsg_end(skb, nlh);
1da177e4
LT
3522}
3523
3524static int inet6_fill_ifmcaddr(struct sk_buff *skb, struct ifmcaddr6 *ifmca,
e431b8c0 3525 u32 pid, u32 seq, int event, u16 flags)
1da177e4 3526{
1da177e4 3527 struct nlmsghdr *nlh;
101bb229
TG
3528 u8 scope = RT_SCOPE_UNIVERSE;
3529 int ifindex = ifmca->idev->dev->ifindex;
3530
3531 if (ipv6_addr_scope(&ifmca->mca_addr) & IFA_SITE)
3532 scope = RT_SCOPE_SITE;
1da177e4 3533
0ab6803b
TG
3534 nlh = nlmsg_put(skb, pid, seq, event, sizeof(struct ifaddrmsg), flags);
3535 if (nlh == NULL)
26932566 3536 return -EMSGSIZE;
85486af0 3537
0ab6803b
TG
3538 put_ifaddrmsg(nlh, 128, IFA_F_PERMANENT, scope, ifindex);
3539 if (nla_put(skb, IFA_MULTICAST, 16, &ifmca->mca_addr) < 0 ||
3540 put_cacheinfo(skb, ifmca->mca_cstamp, ifmca->mca_tstamp,
26932566
PM
3541 INFINITY_LIFE_TIME, INFINITY_LIFE_TIME) < 0) {
3542 nlmsg_cancel(skb, nlh);
3543 return -EMSGSIZE;
3544 }
85486af0 3545
0ab6803b 3546 return nlmsg_end(skb, nlh);
1da177e4
LT
3547}
3548
3549static int inet6_fill_ifacaddr(struct sk_buff *skb, struct ifacaddr6 *ifaca,
b6544c0b 3550 u32 pid, u32 seq, int event, unsigned int flags)
1da177e4 3551{
1da177e4 3552 struct nlmsghdr *nlh;
101bb229
TG
3553 u8 scope = RT_SCOPE_UNIVERSE;
3554 int ifindex = ifaca->aca_idev->dev->ifindex;
3555
3556 if (ipv6_addr_scope(&ifaca->aca_addr) & IFA_SITE)
3557 scope = RT_SCOPE_SITE;
1da177e4 3558
0ab6803b
TG
3559 nlh = nlmsg_put(skb, pid, seq, event, sizeof(struct ifaddrmsg), flags);
3560 if (nlh == NULL)
26932566 3561 return -EMSGSIZE;
85486af0 3562
0ab6803b
TG
3563 put_ifaddrmsg(nlh, 128, IFA_F_PERMANENT, scope, ifindex);
3564 if (nla_put(skb, IFA_ANYCAST, 16, &ifaca->aca_addr) < 0 ||
3565 put_cacheinfo(skb, ifaca->aca_cstamp, ifaca->aca_tstamp,
26932566
PM
3566 INFINITY_LIFE_TIME, INFINITY_LIFE_TIME) < 0) {
3567 nlmsg_cancel(skb, nlh);
3568 return -EMSGSIZE;
3569 }
1da177e4 3570
0ab6803b 3571 return nlmsg_end(skb, nlh);
1da177e4
LT
3572}
3573
e21e8467 3574enum addr_type_t {
1da177e4
LT
3575 UNICAST_ADDR,
3576 MULTICAST_ADDR,
3577 ANYCAST_ADDR,
3578};
3579
234b27c3
ED
3580/* called with rcu_read_lock() */
3581static int in6_dump_addrs(struct inet6_dev *idev, struct sk_buff *skb,
3582 struct netlink_callback *cb, enum addr_type_t type,
3583 int s_ip_idx, int *p_ip_idx)
3584{
234b27c3
ED
3585 struct ifmcaddr6 *ifmca;
3586 struct ifacaddr6 *ifaca;
3587 int err = 1;
3588 int ip_idx = *p_ip_idx;
3589
3590 read_lock_bh(&idev->lock);
3591 switch (type) {
502a2ffd 3592 case UNICAST_ADDR: {
3593 struct inet6_ifaddr *ifa;
3594
234b27c3 3595 /* unicast address incl. temp addr */
502a2ffd 3596 list_for_each_entry(ifa, &idev->addr_list, if_list) {
3597 if (++ip_idx < s_ip_idx)
234b27c3
ED
3598 continue;
3599 err = inet6_fill_ifaddr(skb, ifa,
3600 NETLINK_CB(cb->skb).pid,
3601 cb->nlh->nlmsg_seq,
3602 RTM_NEWADDR,
3603 NLM_F_MULTI);
3604 if (err <= 0)
3605 break;
3606 }
3607 break;
502a2ffd 3608 }
234b27c3
ED
3609 case MULTICAST_ADDR:
3610 /* multicast address */
3611 for (ifmca = idev->mc_list; ifmca;
3612 ifmca = ifmca->next, ip_idx++) {
3613 if (ip_idx < s_ip_idx)
3614 continue;
3615 err = inet6_fill_ifmcaddr(skb, ifmca,
3616 NETLINK_CB(cb->skb).pid,
3617 cb->nlh->nlmsg_seq,
3618 RTM_GETMULTICAST,
3619 NLM_F_MULTI);
3620 if (err <= 0)
3621 break;
3622 }
3623 break;
3624 case ANYCAST_ADDR:
3625 /* anycast address */
3626 for (ifaca = idev->ac_list; ifaca;
3627 ifaca = ifaca->aca_next, ip_idx++) {
3628 if (ip_idx < s_ip_idx)
3629 continue;
3630 err = inet6_fill_ifacaddr(skb, ifaca,
3631 NETLINK_CB(cb->skb).pid,
3632 cb->nlh->nlmsg_seq,
3633 RTM_GETANYCAST,
3634 NLM_F_MULTI);
3635 if (err <= 0)
3636 break;
3637 }
3638 break;
3639 default:
3640 break;
3641 }
3642 read_unlock_bh(&idev->lock);
3643 *p_ip_idx = ip_idx;
3644 return err;
3645}
3646
1da177e4
LT
3647static int inet6_dump_addr(struct sk_buff *skb, struct netlink_callback *cb,
3648 enum addr_type_t type)
3649{
234b27c3
ED
3650 struct net *net = sock_net(skb->sk);
3651 int h, s_h;
1da177e4
LT
3652 int idx, ip_idx;
3653 int s_idx, s_ip_idx;
1da177e4 3654 struct net_device *dev;
234b27c3
ED
3655 struct inet6_dev *idev;
3656 struct hlist_head *head;
3657 struct hlist_node *node;
1ab1457c 3658
234b27c3
ED
3659 s_h = cb->args[0];
3660 s_idx = idx = cb->args[1];
3661 s_ip_idx = ip_idx = cb->args[2];
5d5619b4 3662
234b27c3
ED
3663 rcu_read_lock();
3664 for (h = s_h; h < NETDEV_HASHENTRIES; h++, s_idx = 0) {
3665 idx = 0;
3666 head = &net->dev_index_head[h];
3667 hlist_for_each_entry_rcu(dev, node, head, index_hlist) {
3668 if (idx < s_idx)
3669 goto cont;
4b97efdf 3670 if (h > s_h || idx > s_idx)
234b27c3
ED
3671 s_ip_idx = 0;
3672 ip_idx = 0;
e21e8467
SH
3673 idev = __in6_dev_get(dev);
3674 if (!idev)
234b27c3
ED
3675 goto cont;
3676
3677 if (in6_dump_addrs(idev, skb, cb, type,
3678 s_ip_idx, &ip_idx) <= 0)
3679 goto done;
7562f876 3680cont:
234b27c3
ED
3681 idx++;
3682 }
1da177e4 3683 }
234b27c3
ED
3684done:
3685 rcu_read_unlock();
3686 cb->args[0] = h;
3687 cb->args[1] = idx;
3688 cb->args[2] = ip_idx;
3689
1da177e4
LT
3690 return skb->len;
3691}
3692
3693static int inet6_dump_ifaddr(struct sk_buff *skb, struct netlink_callback *cb)
3694{
3695 enum addr_type_t type = UNICAST_ADDR;
b854272b 3696
1da177e4
LT
3697 return inet6_dump_addr(skb, cb, type);
3698}
3699
3700static int inet6_dump_ifmcaddr(struct sk_buff *skb, struct netlink_callback *cb)
3701{
3702 enum addr_type_t type = MULTICAST_ADDR;
b854272b 3703
1da177e4
LT
3704 return inet6_dump_addr(skb, cb, type);
3705}
3706
3707
3708static int inet6_dump_ifacaddr(struct sk_buff *skb, struct netlink_callback *cb)
3709{
3710 enum addr_type_t type = ANYCAST_ADDR;
b854272b 3711
1da177e4
LT
3712 return inet6_dump_addr(skb, cb, type);
3713}
3714
1b29fc2c
TG
3715static int inet6_rtm_getaddr(struct sk_buff *in_skb, struct nlmsghdr* nlh,
3716 void *arg)
6c223828 3717{
3b1e0a65 3718 struct net *net = sock_net(in_skb->sk);
1b29fc2c
TG
3719 struct ifaddrmsg *ifm;
3720 struct nlattr *tb[IFA_MAX+1];
6c223828
NT
3721 struct in6_addr *addr = NULL;
3722 struct net_device *dev = NULL;
3723 struct inet6_ifaddr *ifa;
3724 struct sk_buff *skb;
6c223828
NT
3725 int err;
3726
1b29fc2c
TG
3727 err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv6_policy);
3728 if (err < 0)
3729 goto errout;
3730
3731 addr = extract_addr(tb[IFA_ADDRESS], tb[IFA_LOCAL]);
3732 if (addr == NULL) {
3733 err = -EINVAL;
3734 goto errout;
6c223828 3735 }
6c223828 3736
1b29fc2c 3737 ifm = nlmsg_data(nlh);
6c223828 3738 if (ifm->ifa_index)
6fda7350 3739 dev = __dev_get_by_index(net, ifm->ifa_index);
6c223828 3740
e21e8467
SH
3741 ifa = ipv6_get_ifaddr(net, addr, dev, 1);
3742 if (!ifa) {
1b29fc2c
TG
3743 err = -EADDRNOTAVAIL;
3744 goto errout;
3745 }
6c223828 3746
e21e8467
SH
3747 skb = nlmsg_new(inet6_ifaddr_msgsize(), GFP_KERNEL);
3748 if (!skb) {
6c223828 3749 err = -ENOBUFS;
1b29fc2c 3750 goto errout_ifa;
6c223828
NT
3751 }
3752
6c223828
NT
3753 err = inet6_fill_ifaddr(skb, ifa, NETLINK_CB(in_skb).pid,
3754 nlh->nlmsg_seq, RTM_NEWADDR, 0);
26932566
PM
3755 if (err < 0) {
3756 /* -EMSGSIZE implies BUG in inet6_ifaddr_msgsize() */
3757 WARN_ON(err == -EMSGSIZE);
3758 kfree_skb(skb);
3759 goto errout_ifa;
3760 }
6fda7350 3761 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).pid);
1b29fc2c 3762errout_ifa:
6c223828 3763 in6_ifa_put(ifa);
1b29fc2c 3764errout:
6c223828 3765 return err;
6c223828
NT
3766}
3767
1da177e4
LT
3768static void inet6_ifa_notify(int event, struct inet6_ifaddr *ifa)
3769{
3770 struct sk_buff *skb;
c346dca1 3771 struct net *net = dev_net(ifa->idev->dev);
5d620266 3772 int err = -ENOBUFS;
1da177e4 3773
0ab6803b 3774 skb = nlmsg_new(inet6_ifaddr_msgsize(), GFP_ATOMIC);
5d620266
TG
3775 if (skb == NULL)
3776 goto errout;
3777
3778 err = inet6_fill_ifaddr(skb, ifa, 0, 0, event, 0);
26932566
PM
3779 if (err < 0) {
3780 /* -EMSGSIZE implies BUG in inet6_ifaddr_msgsize() */
3781 WARN_ON(err == -EMSGSIZE);
3782 kfree_skb(skb);
3783 goto errout;
3784 }
1ce85fe4
PNA
3785 rtnl_notify(skb, net, 0, RTNLGRP_IPV6_IFADDR, NULL, GFP_ATOMIC);
3786 return;
5d620266
TG
3787errout:
3788 if (err < 0)
6fda7350 3789 rtnl_set_sk_err(net, RTNLGRP_IPV6_IFADDR, err);
1da177e4
LT
3790}
3791
b6f99a21 3792static inline void ipv6_store_devconf(struct ipv6_devconf *cnf,
1da177e4
LT
3793 __s32 *array, int bytes)
3794{
04561c1f
TG
3795 BUG_ON(bytes < (DEVCONF_MAX * 4));
3796
1da177e4
LT
3797 memset(array, 0, bytes);
3798 array[DEVCONF_FORWARDING] = cnf->forwarding;
3799 array[DEVCONF_HOPLIMIT] = cnf->hop_limit;
3800 array[DEVCONF_MTU6] = cnf->mtu6;
3801 array[DEVCONF_ACCEPT_RA] = cnf->accept_ra;
3802 array[DEVCONF_ACCEPT_REDIRECTS] = cnf->accept_redirects;
3803 array[DEVCONF_AUTOCONF] = cnf->autoconf;
3804 array[DEVCONF_DAD_TRANSMITS] = cnf->dad_transmits;
3805 array[DEVCONF_RTR_SOLICITS] = cnf->rtr_solicits;
3806 array[DEVCONF_RTR_SOLICIT_INTERVAL] = cnf->rtr_solicit_interval;
3807 array[DEVCONF_RTR_SOLICIT_DELAY] = cnf->rtr_solicit_delay;
3808 array[DEVCONF_FORCE_MLD_VERSION] = cnf->force_mld_version;
3809#ifdef CONFIG_IPV6_PRIVACY
3810 array[DEVCONF_USE_TEMPADDR] = cnf->use_tempaddr;
3811 array[DEVCONF_TEMP_VALID_LFT] = cnf->temp_valid_lft;
3812 array[DEVCONF_TEMP_PREFERED_LFT] = cnf->temp_prefered_lft;
3813 array[DEVCONF_REGEN_MAX_RETRY] = cnf->regen_max_retry;
3814 array[DEVCONF_MAX_DESYNC_FACTOR] = cnf->max_desync_factor;
3815#endif
3816 array[DEVCONF_MAX_ADDRESSES] = cnf->max_addresses;
65f5c7c1 3817 array[DEVCONF_ACCEPT_RA_DEFRTR] = cnf->accept_ra_defrtr;
c4fd30eb 3818 array[DEVCONF_ACCEPT_RA_PINFO] = cnf->accept_ra_pinfo;
930d6ff2
YH
3819#ifdef CONFIG_IPV6_ROUTER_PREF
3820 array[DEVCONF_ACCEPT_RA_RTR_PREF] = cnf->accept_ra_rtr_pref;
52e16356 3821 array[DEVCONF_RTR_PROBE_INTERVAL] = cnf->rtr_probe_interval;
fa03ef38 3822#ifdef CONFIG_IPV6_ROUTE_INFO
09c884d4
YH
3823 array[DEVCONF_ACCEPT_RA_RT_INFO_MAX_PLEN] = cnf->accept_ra_rt_info_max_plen;
3824#endif
930d6ff2 3825#endif
fbea49e1 3826 array[DEVCONF_PROXY_NDP] = cnf->proxy_ndp;
0bcbc926 3827 array[DEVCONF_ACCEPT_SOURCE_ROUTE] = cnf->accept_source_route;
95c385b4
NH
3828#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
3829 array[DEVCONF_OPTIMISTIC_DAD] = cnf->optimistic_dad;
3830#endif
7bc570c8
YH
3831#ifdef CONFIG_IPV6_MROUTE
3832 array[DEVCONF_MC_FORWARDING] = cnf->mc_forwarding;
3833#endif
778d80be 3834 array[DEVCONF_DISABLE_IPV6] = cnf->disable_ipv6;
1b34be74 3835 array[DEVCONF_ACCEPT_DAD] = cnf->accept_dad;
c3faca05 3836 array[DEVCONF_FORCE_TLLAO] = cnf->force_tllao;
1da177e4
LT
3837}
3838
339bf98f
TG
3839static inline size_t inet6_if_nlmsg_size(void)
3840{
3841 return NLMSG_ALIGN(sizeof(struct ifinfomsg))
3842 + nla_total_size(IFNAMSIZ) /* IFLA_IFNAME */
3843 + nla_total_size(MAX_ADDR_LEN) /* IFLA_ADDRESS */
3844 + nla_total_size(4) /* IFLA_MTU */
3845 + nla_total_size(4) /* IFLA_LINK */
3846 + nla_total_size( /* IFLA_PROTINFO */
3847 nla_total_size(4) /* IFLA_INET6_FLAGS */
3848 + nla_total_size(sizeof(struct ifla_cacheinfo))
3849 + nla_total_size(DEVCONF_MAX * 4) /* IFLA_INET6_CONF */
bf99f1bd
YH
3850 + nla_total_size(IPSTATS_MIB_MAX * 8) /* IFLA_INET6_STATS */
3851 + nla_total_size(ICMP6_MIB_MAX * 8) /* IFLA_INET6_ICMP6STATS */
339bf98f
TG
3852 );
3853}
c5396a31 3854
7d720c3e
TH
3855static inline void __snmp6_fill_stats(u64 *stats, void __percpu **mib,
3856 int items, int bytes)
7f7d9a6b
HX
3857{
3858 int i;
3859 int pad = bytes - sizeof(u64) * items;
3860 BUG_ON(pad < 0);
3861
3862 /* Use put_unaligned() because stats may not be aligned for u64. */
3863 put_unaligned(items, &stats[0]);
3864 for (i = 1; i < items; i++)
3865 put_unaligned(snmp_fold_field(mib, i), &stats[i]);
3866
3867 memset(&stats[items], 0, pad);
3868}
3869
4ce3c183
ED
3870static inline void __snmp6_fill_stats64(u64 *stats, void __percpu **mib,
3871 int items, int bytes, size_t syncpoff)
3872{
3873 int i;
3874 int pad = bytes - sizeof(u64) * items;
3875 BUG_ON(pad < 0);
3876
3877 /* Use put_unaligned() because stats may not be aligned for u64. */
3878 put_unaligned(items, &stats[0]);
3879 for (i = 1; i < items; i++)
3880 put_unaligned(snmp_fold_field64(mib, i, syncpoff), &stats[i]);
3881
3882 memset(&stats[items], 0, pad);
3883}
3884
7f7d9a6b
HX
3885static void snmp6_fill_stats(u64 *stats, struct inet6_dev *idev, int attrtype,
3886 int bytes)
3887{
e21e8467 3888 switch (attrtype) {
7f7d9a6b 3889 case IFLA_INET6_STATS:
4ce3c183
ED
3890 __snmp6_fill_stats64(stats, (void __percpu **)idev->stats.ipv6,
3891 IPSTATS_MIB_MAX, bytes, offsetof(struct ipstats_mib, syncp));
7f7d9a6b
HX
3892 break;
3893 case IFLA_INET6_ICMP6STATS:
7d720c3e 3894 __snmp6_fill_stats(stats, (void __percpu **)idev->stats.icmpv6, ICMP6_MIB_MAX, bytes);
7f7d9a6b
HX
3895 break;
3896 }
3897}
3898
1ab1457c 3899static int inet6_fill_ifinfo(struct sk_buff *skb, struct inet6_dev *idev,
b6544c0b 3900 u32 pid, u32 seq, int event, unsigned int flags)
1da177e4 3901{
04561c1f 3902 struct net_device *dev = idev->dev;
bf99f1bd 3903 struct nlattr *nla;
04561c1f
TG
3904 struct ifinfomsg *hdr;
3905 struct nlmsghdr *nlh;
3906 void *protoinfo;
3907 struct ifla_cacheinfo ci;
3908
3909 nlh = nlmsg_put(skb, pid, seq, event, sizeof(*hdr), flags);
3910 if (nlh == NULL)
26932566 3911 return -EMSGSIZE;
04561c1f
TG
3912
3913 hdr = nlmsg_data(nlh);
3914 hdr->ifi_family = AF_INET6;
3915 hdr->__ifi_pad = 0;
3916 hdr->ifi_type = dev->type;
3917 hdr->ifi_index = dev->ifindex;
3918 hdr->ifi_flags = dev_get_flags(dev);
3919 hdr->ifi_change = 0;
3920
3921 NLA_PUT_STRING(skb, IFLA_IFNAME, dev->name);
1da177e4
LT
3922
3923 if (dev->addr_len)
04561c1f 3924 NLA_PUT(skb, IFLA_ADDRESS, dev->addr_len, dev->dev_addr);
1da177e4 3925
04561c1f 3926 NLA_PUT_U32(skb, IFLA_MTU, dev->mtu);
1da177e4 3927 if (dev->ifindex != dev->iflink)
04561c1f 3928 NLA_PUT_U32(skb, IFLA_LINK, dev->iflink);
1da177e4 3929
04561c1f
TG
3930 protoinfo = nla_nest_start(skb, IFLA_PROTINFO);
3931 if (protoinfo == NULL)
3932 goto nla_put_failure;
1da177e4 3933
04561c1f 3934 NLA_PUT_U32(skb, IFLA_INET6_FLAGS, idev->if_flags);
1da177e4 3935
1da177e4
LT
3936 ci.max_reasm_len = IPV6_MAXPLEN;
3937 ci.tstamp = (__u32)(TIME_DELTA(idev->tstamp, INITIAL_JIFFIES) / HZ * 100
3938 + TIME_DELTA(idev->tstamp, INITIAL_JIFFIES) % HZ * 100 / HZ);
3939 ci.reachable_time = idev->nd_parms->reachable_time;
3940 ci.retrans_time = idev->nd_parms->retrans_time;
04561c1f
TG
3941 NLA_PUT(skb, IFLA_INET6_CACHEINFO, sizeof(ci), &ci);
3942
bf99f1bd
YH
3943 nla = nla_reserve(skb, IFLA_INET6_CONF, DEVCONF_MAX * sizeof(s32));
3944 if (nla == NULL)
04561c1f 3945 goto nla_put_failure;
bf99f1bd 3946 ipv6_store_devconf(&idev->cnf, nla_data(nla), nla_len(nla));
1da177e4 3947
bf99f1bd
YH
3948 /* XXX - MC not implemented */
3949
3950 nla = nla_reserve(skb, IFLA_INET6_STATS, IPSTATS_MIB_MAX * sizeof(u64));
3951 if (nla == NULL)
3952 goto nla_put_failure;
3953 snmp6_fill_stats(nla_data(nla), idev, IFLA_INET6_STATS, nla_len(nla));
3954
3955 nla = nla_reserve(skb, IFLA_INET6_ICMP6STATS, ICMP6_MIB_MAX * sizeof(u64));
3956 if (nla == NULL)
3957 goto nla_put_failure;
3958 snmp6_fill_stats(nla_data(nla), idev, IFLA_INET6_ICMP6STATS, nla_len(nla));
1da177e4 3959
04561c1f
TG
3960 nla_nest_end(skb, protoinfo);
3961 return nlmsg_end(skb, nlh);
1da177e4 3962
04561c1f 3963nla_put_failure:
26932566
PM
3964 nlmsg_cancel(skb, nlh);
3965 return -EMSGSIZE;
1da177e4
LT
3966}
3967
3968static int inet6_dump_ifinfo(struct sk_buff *skb, struct netlink_callback *cb)
3969{
3b1e0a65 3970 struct net *net = sock_net(skb->sk);
84d2697d 3971 int h, s_h;
434a8a58 3972 int idx = 0, s_idx;
1da177e4
LT
3973 struct net_device *dev;
3974 struct inet6_dev *idev;
84d2697d
ED
3975 struct hlist_head *head;
3976 struct hlist_node *node;
1da177e4 3977
84d2697d
ED
3978 s_h = cb->args[0];
3979 s_idx = cb->args[1];
3980
3981 rcu_read_lock();
3982 for (h = s_h; h < NETDEV_HASHENTRIES; h++, s_idx = 0) {
3983 idx = 0;
3984 head = &net->dev_index_head[h];
3985 hlist_for_each_entry_rcu(dev, node, head, index_hlist) {
3986 if (idx < s_idx)
3987 goto cont;
3988 idev = __in6_dev_get(dev);
3989 if (!idev)
3990 goto cont;
3991 if (inet6_fill_ifinfo(skb, idev,
3992 NETLINK_CB(cb->skb).pid,
3993 cb->nlh->nlmsg_seq,
3994 RTM_NEWLINK, NLM_F_MULTI) <= 0)
3995 goto out;
7562f876 3996cont:
84d2697d
ED
3997 idx++;
3998 }
1da177e4 3999 }
84d2697d
ED
4000out:
4001 rcu_read_unlock();
4002 cb->args[1] = idx;
4003 cb->args[0] = h;
1da177e4
LT
4004
4005 return skb->len;
4006}
4007
4008void inet6_ifinfo_notify(int event, struct inet6_dev *idev)
4009{
4010 struct sk_buff *skb;
c346dca1 4011 struct net *net = dev_net(idev->dev);
8d7a76c9 4012 int err = -ENOBUFS;
1ab1457c 4013
339bf98f 4014 skb = nlmsg_new(inet6_if_nlmsg_size(), GFP_ATOMIC);
8d7a76c9
TG
4015 if (skb == NULL)
4016 goto errout;
4017
4018 err = inet6_fill_ifinfo(skb, idev, 0, 0, event, 0);
26932566
PM
4019 if (err < 0) {
4020 /* -EMSGSIZE implies BUG in inet6_if_nlmsg_size() */
4021 WARN_ON(err == -EMSGSIZE);
4022 kfree_skb(skb);
4023 goto errout;
4024 }
1ce85fe4
PNA
4025 rtnl_notify(skb, net, 0, RTNLGRP_IPV6_IFADDR, NULL, GFP_ATOMIC);
4026 return;
8d7a76c9
TG
4027errout:
4028 if (err < 0)
6fda7350 4029 rtnl_set_sk_err(net, RTNLGRP_IPV6_IFADDR, err);
1da177e4
LT
4030}
4031
339bf98f
TG
4032static inline size_t inet6_prefix_nlmsg_size(void)
4033{
4034 return NLMSG_ALIGN(sizeof(struct prefixmsg))
4035 + nla_total_size(sizeof(struct in6_addr))
4036 + nla_total_size(sizeof(struct prefix_cacheinfo));
4037}
c5396a31 4038
1da177e4 4039static int inet6_fill_prefix(struct sk_buff *skb, struct inet6_dev *idev,
6051e2f4
TG
4040 struct prefix_info *pinfo, u32 pid, u32 seq,
4041 int event, unsigned int flags)
1da177e4 4042{
6051e2f4
TG
4043 struct prefixmsg *pmsg;
4044 struct nlmsghdr *nlh;
1da177e4
LT
4045 struct prefix_cacheinfo ci;
4046
6051e2f4
TG
4047 nlh = nlmsg_put(skb, pid, seq, event, sizeof(*pmsg), flags);
4048 if (nlh == NULL)
26932566 4049 return -EMSGSIZE;
6051e2f4
TG
4050
4051 pmsg = nlmsg_data(nlh);
1da177e4 4052 pmsg->prefix_family = AF_INET6;
8a47077a
PM
4053 pmsg->prefix_pad1 = 0;
4054 pmsg->prefix_pad2 = 0;
1da177e4
LT
4055 pmsg->prefix_ifindex = idev->dev->ifindex;
4056 pmsg->prefix_len = pinfo->prefix_len;
4057 pmsg->prefix_type = pinfo->type;
8a47077a 4058 pmsg->prefix_pad3 = 0;
1da177e4
LT
4059 pmsg->prefix_flags = 0;
4060 if (pinfo->onlink)
4061 pmsg->prefix_flags |= IF_PREFIX_ONLINK;
4062 if (pinfo->autoconf)
4063 pmsg->prefix_flags |= IF_PREFIX_AUTOCONF;
4064
6051e2f4 4065 NLA_PUT(skb, PREFIX_ADDRESS, sizeof(pinfo->prefix), &pinfo->prefix);
1da177e4
LT
4066
4067 ci.preferred_time = ntohl(pinfo->prefered);
4068 ci.valid_time = ntohl(pinfo->valid);
6051e2f4 4069 NLA_PUT(skb, PREFIX_CACHEINFO, sizeof(ci), &ci);
1da177e4 4070
6051e2f4 4071 return nlmsg_end(skb, nlh);
1da177e4 4072
6051e2f4 4073nla_put_failure:
26932566
PM
4074 nlmsg_cancel(skb, nlh);
4075 return -EMSGSIZE;
1da177e4
LT
4076}
4077
1ab1457c 4078static void inet6_prefix_notify(int event, struct inet6_dev *idev,
1da177e4
LT
4079 struct prefix_info *pinfo)
4080{
4081 struct sk_buff *skb;
c346dca1 4082 struct net *net = dev_net(idev->dev);
8c384bfa 4083 int err = -ENOBUFS;
1da177e4 4084
339bf98f 4085 skb = nlmsg_new(inet6_prefix_nlmsg_size(), GFP_ATOMIC);
8c384bfa
TG
4086 if (skb == NULL)
4087 goto errout;
4088
4089 err = inet6_fill_prefix(skb, idev, pinfo, 0, 0, event, 0);
26932566
PM
4090 if (err < 0) {
4091 /* -EMSGSIZE implies BUG in inet6_prefix_nlmsg_size() */
4092 WARN_ON(err == -EMSGSIZE);
4093 kfree_skb(skb);
4094 goto errout;
4095 }
1ce85fe4
PNA
4096 rtnl_notify(skb, net, 0, RTNLGRP_IPV6_PREFIX, NULL, GFP_ATOMIC);
4097 return;
8c384bfa
TG
4098errout:
4099 if (err < 0)
6fda7350 4100 rtnl_set_sk_err(net, RTNLGRP_IPV6_PREFIX, err);
1da177e4
LT
4101}
4102
1da177e4
LT
4103static void __ipv6_ifa_notify(int event, struct inet6_ifaddr *ifp)
4104{
4105 inet6_ifa_notify(event ? : RTM_NEWADDR, ifp);
4106
4107 switch (event) {
4108 case RTM_NEWADDR:
95c385b4
NH
4109 /*
4110 * If the address was optimistic
4111 * we inserted the route at the start of
4112 * our DAD process, so we don't need
4113 * to do it again
4114 */
4115 if (!(ifp->rt->rt6i_node))
4116 ip6_ins_rt(ifp->rt);
1da177e4
LT
4117 if (ifp->idev->cnf.forwarding)
4118 addrconf_join_anycast(ifp);
4119 break;
4120 case RTM_DELADDR:
4121 if (ifp->idev->cnf.forwarding)
4122 addrconf_leave_anycast(ifp);
4123 addrconf_leave_solict(ifp->idev, &ifp->addr);
d8d1f30b 4124 dst_hold(&ifp->rt->dst);
93fa159a 4125
e9d3e084
HX
4126 if (ifp->state == INET6_IFADDR_STATE_DEAD &&
4127 ip6_del_rt(ifp->rt))
d8d1f30b 4128 dst_free(&ifp->rt->dst);
1da177e4
LT
4129 break;
4130 }
4131}
4132
4133static void ipv6_ifa_notify(int event, struct inet6_ifaddr *ifp)
4134{
8814c4b5 4135 rcu_read_lock_bh();
1da177e4
LT
4136 if (likely(ifp->idev->dead == 0))
4137 __ipv6_ifa_notify(event, ifp);
8814c4b5 4138 rcu_read_unlock_bh();
1da177e4
LT
4139}
4140
4141#ifdef CONFIG_SYSCTL
4142
4143static
8d65af78 4144int addrconf_sysctl_forward(ctl_table *ctl, int write,
1da177e4
LT
4145 void __user *buffer, size_t *lenp, loff_t *ppos)
4146{
4147 int *valp = ctl->data;
4148 int val = *valp;
88af182e 4149 loff_t pos = *ppos;
1da177e4
LT
4150 int ret;
4151
8d65af78 4152 ret = proc_dointvec(ctl, write, buffer, lenp, ppos);
1da177e4 4153
c8fecf22 4154 if (write)
b325fddb 4155 ret = addrconf_fixup_forwarding(ctl, valp, val);
88af182e
EB
4156 if (ret)
4157 *ppos = pos;
1ab1457c 4158 return ret;
1da177e4
LT
4159}
4160
56d417b1
BH
4161static void dev_disable_change(struct inet6_dev *idev)
4162{
4163 if (!idev || !idev->dev)
4164 return;
4165
4166 if (idev->cnf.disable_ipv6)
4167 addrconf_notify(NULL, NETDEV_DOWN, idev->dev);
4168 else
4169 addrconf_notify(NULL, NETDEV_UP, idev->dev);
4170}
4171
4172static void addrconf_disable_change(struct net *net, __s32 newf)
4173{
4174 struct net_device *dev;
4175 struct inet6_dev *idev;
4176
c6d14c84
ED
4177 rcu_read_lock();
4178 for_each_netdev_rcu(net, dev) {
56d417b1
BH
4179 idev = __in6_dev_get(dev);
4180 if (idev) {
4181 int changed = (!idev->cnf.disable_ipv6) ^ (!newf);
4182 idev->cnf.disable_ipv6 = newf;
4183 if (changed)
4184 dev_disable_change(idev);
4185 }
56d417b1 4186 }
c6d14c84 4187 rcu_read_unlock();
56d417b1
BH
4188}
4189
4190static int addrconf_disable_ipv6(struct ctl_table *table, int *p, int old)
4191{
4192 struct net *net;
4193
4194 net = (struct net *)table->extra2;
4195
4196 if (p == &net->ipv6.devconf_dflt->disable_ipv6)
4197 return 0;
4198
88af182e
EB
4199 if (!rtnl_trylock()) {
4200 /* Restore the original values before restarting */
4201 *p = old;
56d417b1 4202 return restart_syscall();
88af182e 4203 }
56d417b1
BH
4204
4205 if (p == &net->ipv6.devconf_all->disable_ipv6) {
4206 __s32 newf = net->ipv6.devconf_all->disable_ipv6;
4207 net->ipv6.devconf_dflt->disable_ipv6 = newf;
4208 addrconf_disable_change(net, newf);
4209 } else if ((!*p) ^ (!old))
4210 dev_disable_change((struct inet6_dev *)table->extra1);
4211
4212 rtnl_unlock();
4213 return 0;
4214}
4215
4216static
8d65af78 4217int addrconf_sysctl_disable(ctl_table *ctl, int write,
56d417b1
BH
4218 void __user *buffer, size_t *lenp, loff_t *ppos)
4219{
4220 int *valp = ctl->data;
4221 int val = *valp;
88af182e 4222 loff_t pos = *ppos;
56d417b1
BH
4223 int ret;
4224
8d65af78 4225 ret = proc_dointvec(ctl, write, buffer, lenp, ppos);
56d417b1
BH
4226
4227 if (write)
4228 ret = addrconf_disable_ipv6(ctl, valp, val);
88af182e
EB
4229 if (ret)
4230 *ppos = pos;
56d417b1
BH
4231 return ret;
4232}
4233
1da177e4
LT
4234static struct addrconf_sysctl_table
4235{
4236 struct ctl_table_header *sysctl_header;
f6a07b29 4237 ctl_table addrconf_vars[DEVCONF_MAX+1];
1dab6222 4238 char *dev_name;
ab32ea5d 4239} addrconf_sysctl __read_mostly = {
1da177e4
LT
4240 .sysctl_header = NULL,
4241 .addrconf_vars = {
1ab1457c 4242 {
e21e8467
SH
4243 .procname = "forwarding",
4244 .data = &ipv6_devconf.forwarding,
4245 .maxlen = sizeof(int),
4246 .mode = 0644,
4247 .proc_handler = addrconf_sysctl_forward,
1da177e4
LT
4248 },
4249 {
e21e8467
SH
4250 .procname = "hop_limit",
4251 .data = &ipv6_devconf.hop_limit,
4252 .maxlen = sizeof(int),
4253 .mode = 0644,
4254 .proc_handler = proc_dointvec,
1da177e4
LT
4255 },
4256 {
e21e8467
SH
4257 .procname = "mtu",
4258 .data = &ipv6_devconf.mtu6,
4259 .maxlen = sizeof(int),
4260 .mode = 0644,
4261 .proc_handler = proc_dointvec,
1da177e4
LT
4262 },
4263 {
e21e8467
SH
4264 .procname = "accept_ra",
4265 .data = &ipv6_devconf.accept_ra,
4266 .maxlen = sizeof(int),
4267 .mode = 0644,
4268 .proc_handler = proc_dointvec,
1da177e4
LT
4269 },
4270 {
e21e8467
SH
4271 .procname = "accept_redirects",
4272 .data = &ipv6_devconf.accept_redirects,
4273 .maxlen = sizeof(int),
4274 .mode = 0644,
4275 .proc_handler = proc_dointvec,
1da177e4
LT
4276 },
4277 {
e21e8467
SH
4278 .procname = "autoconf",
4279 .data = &ipv6_devconf.autoconf,
4280 .maxlen = sizeof(int),
4281 .mode = 0644,
4282 .proc_handler = proc_dointvec,
1da177e4
LT
4283 },
4284 {
e21e8467
SH
4285 .procname = "dad_transmits",
4286 .data = &ipv6_devconf.dad_transmits,
4287 .maxlen = sizeof(int),
4288 .mode = 0644,
4289 .proc_handler = proc_dointvec,
1da177e4
LT
4290 },
4291 {
e21e8467
SH
4292 .procname = "router_solicitations",
4293 .data = &ipv6_devconf.rtr_solicits,
4294 .maxlen = sizeof(int),
4295 .mode = 0644,
4296 .proc_handler = proc_dointvec,
1da177e4
LT
4297 },
4298 {
e21e8467
SH
4299 .procname = "router_solicitation_interval",
4300 .data = &ipv6_devconf.rtr_solicit_interval,
4301 .maxlen = sizeof(int),
4302 .mode = 0644,
4303 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT
4304 },
4305 {
e21e8467
SH
4306 .procname = "router_solicitation_delay",
4307 .data = &ipv6_devconf.rtr_solicit_delay,
4308 .maxlen = sizeof(int),
4309 .mode = 0644,
4310 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT
4311 },
4312 {
e21e8467
SH
4313 .procname = "force_mld_version",
4314 .data = &ipv6_devconf.force_mld_version,
4315 .maxlen = sizeof(int),
4316 .mode = 0644,
4317 .proc_handler = proc_dointvec,
1da177e4
LT
4318 },
4319#ifdef CONFIG_IPV6_PRIVACY
4320 {
e21e8467
SH
4321 .procname = "use_tempaddr",
4322 .data = &ipv6_devconf.use_tempaddr,
4323 .maxlen = sizeof(int),
4324 .mode = 0644,
4325 .proc_handler = proc_dointvec,
1da177e4
LT
4326 },
4327 {
e21e8467
SH
4328 .procname = "temp_valid_lft",
4329 .data = &ipv6_devconf.temp_valid_lft,
4330 .maxlen = sizeof(int),
4331 .mode = 0644,
4332 .proc_handler = proc_dointvec,
1da177e4
LT
4333 },
4334 {
e21e8467
SH
4335 .procname = "temp_prefered_lft",
4336 .data = &ipv6_devconf.temp_prefered_lft,
4337 .maxlen = sizeof(int),
4338 .mode = 0644,
4339 .proc_handler = proc_dointvec,
1da177e4
LT
4340 },
4341 {
e21e8467
SH
4342 .procname = "regen_max_retry",
4343 .data = &ipv6_devconf.regen_max_retry,
4344 .maxlen = sizeof(int),
4345 .mode = 0644,
4346 .proc_handler = proc_dointvec,
1da177e4
LT
4347 },
4348 {
e21e8467
SH
4349 .procname = "max_desync_factor",
4350 .data = &ipv6_devconf.max_desync_factor,
4351 .maxlen = sizeof(int),
4352 .mode = 0644,
4353 .proc_handler = proc_dointvec,
1da177e4
LT
4354 },
4355#endif
4356 {
e21e8467
SH
4357 .procname = "max_addresses",
4358 .data = &ipv6_devconf.max_addresses,
4359 .maxlen = sizeof(int),
4360 .mode = 0644,
4361 .proc_handler = proc_dointvec,
1da177e4 4362 },
65f5c7c1 4363 {
e21e8467
SH
4364 .procname = "accept_ra_defrtr",
4365 .data = &ipv6_devconf.accept_ra_defrtr,
4366 .maxlen = sizeof(int),
4367 .mode = 0644,
4368 .proc_handler = proc_dointvec,
65f5c7c1 4369 },
c4fd30eb 4370 {
e21e8467
SH
4371 .procname = "accept_ra_pinfo",
4372 .data = &ipv6_devconf.accept_ra_pinfo,
4373 .maxlen = sizeof(int),
4374 .mode = 0644,
4375 .proc_handler = proc_dointvec,
c4fd30eb 4376 },
930d6ff2
YH
4377#ifdef CONFIG_IPV6_ROUTER_PREF
4378 {
e21e8467
SH
4379 .procname = "accept_ra_rtr_pref",
4380 .data = &ipv6_devconf.accept_ra_rtr_pref,
4381 .maxlen = sizeof(int),
4382 .mode = 0644,
4383 .proc_handler = proc_dointvec,
930d6ff2 4384 },
52e16356 4385 {
e21e8467
SH
4386 .procname = "router_probe_interval",
4387 .data = &ipv6_devconf.rtr_probe_interval,
4388 .maxlen = sizeof(int),
4389 .mode = 0644,
4390 .proc_handler = proc_dointvec_jiffies,
52e16356 4391 },
fa03ef38 4392#ifdef CONFIG_IPV6_ROUTE_INFO
09c884d4 4393 {
e21e8467
SH
4394 .procname = "accept_ra_rt_info_max_plen",
4395 .data = &ipv6_devconf.accept_ra_rt_info_max_plen,
4396 .maxlen = sizeof(int),
4397 .mode = 0644,
4398 .proc_handler = proc_dointvec,
09c884d4
YH
4399 },
4400#endif
930d6ff2 4401#endif
fbea49e1 4402 {
e21e8467
SH
4403 .procname = "proxy_ndp",
4404 .data = &ipv6_devconf.proxy_ndp,
4405 .maxlen = sizeof(int),
4406 .mode = 0644,
4407 .proc_handler = proc_dointvec,
fbea49e1 4408 },
0bcbc926 4409 {
e21e8467
SH
4410 .procname = "accept_source_route",
4411 .data = &ipv6_devconf.accept_source_route,
4412 .maxlen = sizeof(int),
4413 .mode = 0644,
4414 .proc_handler = proc_dointvec,
0bcbc926 4415 },
95c385b4
NH
4416#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
4417 {
e21e8467
SH
4418 .procname = "optimistic_dad",
4419 .data = &ipv6_devconf.optimistic_dad,
4420 .maxlen = sizeof(int),
4421 .mode = 0644,
4422 .proc_handler = proc_dointvec,
95c385b4
NH
4423
4424 },
7bc570c8
YH
4425#endif
4426#ifdef CONFIG_IPV6_MROUTE
4427 {
e21e8467
SH
4428 .procname = "mc_forwarding",
4429 .data = &ipv6_devconf.mc_forwarding,
4430 .maxlen = sizeof(int),
4431 .mode = 0444,
4432 .proc_handler = proc_dointvec,
7bc570c8 4433 },
95c385b4 4434#endif
778d80be 4435 {
e21e8467
SH
4436 .procname = "disable_ipv6",
4437 .data = &ipv6_devconf.disable_ipv6,
4438 .maxlen = sizeof(int),
4439 .mode = 0644,
4440 .proc_handler = addrconf_sysctl_disable,
778d80be 4441 },
1b34be74 4442 {
e21e8467
SH
4443 .procname = "accept_dad",
4444 .data = &ipv6_devconf.accept_dad,
4445 .maxlen = sizeof(int),
4446 .mode = 0644,
4447 .proc_handler = proc_dointvec,
1b34be74 4448 },
f7734fdf 4449 {
f7734fdf
OP
4450 .procname = "force_tllao",
4451 .data = &ipv6_devconf.force_tllao,
4452 .maxlen = sizeof(int),
4453 .mode = 0644,
4454 .proc_handler = proc_dointvec
4455 },
1da177e4 4456 {
f8572d8f 4457 /* sentinel */
1da177e4
LT
4458 }
4459 },
1da177e4
LT
4460};
4461
bff16c2f 4462static int __addrconf_sysctl_register(struct net *net, char *dev_name,
f8572d8f 4463 struct inet6_dev *idev, struct ipv6_devconf *p)
1da177e4
LT
4464{
4465 int i;
1da177e4 4466 struct addrconf_sysctl_table *t;
1da177e4 4467
1dab6222
PE
4468#define ADDRCONF_CTL_PATH_DEV 3
4469
4470 struct ctl_path addrconf_ctl_path[] = {
f8572d8f
EB
4471 { .procname = "net", },
4472 { .procname = "ipv6", },
4473 { .procname = "conf", },
1dab6222
PE
4474 { /* to be set */ },
4475 { },
4476 };
4477
4478
af879cc7 4479 t = kmemdup(&addrconf_sysctl, sizeof(*t), GFP_KERNEL);
1da177e4 4480 if (t == NULL)
f68635e6
PE
4481 goto out;
4482
bcdd553f 4483 for (i = 0; t->addrconf_vars[i].data; i++) {
e21e8467 4484 t->addrconf_vars[i].data += (char *)p - (char *)&ipv6_devconf;
1da177e4 4485 t->addrconf_vars[i].extra1 = idev; /* embedded; no ref */
bff16c2f 4486 t->addrconf_vars[i].extra2 = net;
1da177e4 4487 }
1da177e4 4488
1ab1457c
YH
4489 /*
4490 * Make a copy of dev_name, because '.procname' is regarded as const
1da177e4
LT
4491 * by sysctl and we wouldn't want anyone to change it under our feet
4492 * (see SIOCSIFNAME).
1ab1457c 4493 */
1dab6222
PE
4494 t->dev_name = kstrdup(dev_name, GFP_KERNEL);
4495 if (!t->dev_name)
f68635e6 4496 goto free;
1da177e4 4497
1dab6222 4498 addrconf_ctl_path[ADDRCONF_CTL_PATH_DEV].procname = t->dev_name;
1da177e4 4499
bff16c2f 4500 t->sysctl_header = register_net_sysctl_table(net, addrconf_ctl_path,
1dab6222 4501 t->addrconf_vars);
1da177e4
LT
4502 if (t->sysctl_header == NULL)
4503 goto free_procname;
f68635e6
PE
4504
4505 p->sysctl = t;
95897312 4506 return 0;
1da177e4 4507
f68635e6 4508free_procname:
1dab6222 4509 kfree(t->dev_name);
f68635e6 4510free:
1da177e4 4511 kfree(t);
f68635e6 4512out:
95897312 4513 return -ENOBUFS;
1da177e4
LT
4514}
4515
408c4768
PE
4516static void __addrconf_sysctl_unregister(struct ipv6_devconf *p)
4517{
4518 struct addrconf_sysctl_table *t;
4519
4520 if (p->sysctl == NULL)
4521 return;
4522
4523 t = p->sysctl;
4524 p->sysctl = NULL;
4525 unregister_sysctl_table(t->sysctl_header);
4526 kfree(t->dev_name);
4527 kfree(t);
4528}
4529
f52295a9
PE
4530static void addrconf_sysctl_register(struct inet6_dev *idev)
4531{
54716e3b 4532 neigh_sysctl_register(idev->dev, idev->nd_parms, "ipv6",
f8572d8f 4533 &ndisc_ifinfo_sysctl_change);
c346dca1 4534 __addrconf_sysctl_register(dev_net(idev->dev), idev->dev->name,
f8572d8f 4535 idev, &idev->cnf);
f52295a9
PE
4536}
4537
408c4768 4538static void addrconf_sysctl_unregister(struct inet6_dev *idev)
1da177e4 4539{
408c4768
PE
4540 __addrconf_sysctl_unregister(&idev->cnf);
4541 neigh_sysctl_unregister(idev->nd_parms);
1da177e4
LT
4542}
4543
4544
4545#endif
4546
2c8c1e72 4547static int __net_init addrconf_init_net(struct net *net)
e0da5a48
PE
4548{
4549 int err;
4550 struct ipv6_devconf *all, *dflt;
4551
4552 err = -ENOMEM;
4553 all = &ipv6_devconf;
4554 dflt = &ipv6_devconf_dflt;
4555
09ad9bc7 4556 if (!net_eq(net, &init_net)) {
e0da5a48
PE
4557 all = kmemdup(all, sizeof(ipv6_devconf), GFP_KERNEL);
4558 if (all == NULL)
4559 goto err_alloc_all;
4560
4561 dflt = kmemdup(dflt, sizeof(ipv6_devconf_dflt), GFP_KERNEL);
4562 if (dflt == NULL)
4563 goto err_alloc_dflt;
56d417b1
BH
4564 } else {
4565 /* these will be inherited by all namespaces */
4566 dflt->autoconf = ipv6_defaults.autoconf;
4567 dflt->disable_ipv6 = ipv6_defaults.disable_ipv6;
e0da5a48
PE
4568 }
4569
4570 net->ipv6.devconf_all = all;
4571 net->ipv6.devconf_dflt = dflt;
4572
4573#ifdef CONFIG_SYSCTL
f8572d8f 4574 err = __addrconf_sysctl_register(net, "all", NULL, all);
e0da5a48
PE
4575 if (err < 0)
4576 goto err_reg_all;
4577
f8572d8f 4578 err = __addrconf_sysctl_register(net, "default", NULL, dflt);
e0da5a48
PE
4579 if (err < 0)
4580 goto err_reg_dflt;
4581#endif
4582 return 0;
4583
4584#ifdef CONFIG_SYSCTL
4585err_reg_dflt:
4586 __addrconf_sysctl_unregister(all);
4587err_reg_all:
4588 kfree(dflt);
4589#endif
4590err_alloc_dflt:
4591 kfree(all);
4592err_alloc_all:
4593 return err;
4594}
4595
2c8c1e72 4596static void __net_exit addrconf_exit_net(struct net *net)
e0da5a48
PE
4597{
4598#ifdef CONFIG_SYSCTL
4599 __addrconf_sysctl_unregister(net->ipv6.devconf_dflt);
4600 __addrconf_sysctl_unregister(net->ipv6.devconf_all);
4601#endif
09ad9bc7 4602 if (!net_eq(net, &init_net)) {
e0da5a48
PE
4603 kfree(net->ipv6.devconf_dflt);
4604 kfree(net->ipv6.devconf_all);
4605 }
4606}
4607
4608static struct pernet_operations addrconf_ops = {
4609 .init = addrconf_init_net,
4610 .exit = addrconf_exit_net,
4611};
4612
1da177e4
LT
4613/*
4614 * Device notifier
4615 */
4616
4617int register_inet6addr_notifier(struct notifier_block *nb)
4618{
1ab1457c 4619 return atomic_notifier_chain_register(&inet6addr_chain, nb);
1da177e4 4620}
7159039a
YH
4621EXPORT_SYMBOL(register_inet6addr_notifier);
4622
1da177e4
LT
4623int unregister_inet6addr_notifier(struct notifier_block *nb)
4624{
e21e8467 4625 return atomic_notifier_chain_unregister(&inet6addr_chain, nb);
1da177e4 4626}
7159039a
YH
4627EXPORT_SYMBOL(unregister_inet6addr_notifier);
4628
1da177e4
LT
4629/*
4630 * Init / cleanup code
4631 */
4632
4633int __init addrconf_init(void)
4634{
c2e21293 4635 int i, err;
2a8cc6c8 4636
e21e8467
SH
4637 err = ipv6_addr_label_init();
4638 if (err < 0) {
4639 printk(KERN_CRIT "IPv6 Addrconf:"
4640 " cannot initialize default policy table: %d.\n", err);
2cc6d2bf 4641 goto out;
2a8cc6c8 4642 }
1da177e4 4643
2cc6d2bf
NH
4644 err = register_pernet_subsys(&addrconf_ops);
4645 if (err < 0)
4646 goto out_addrlabel;
e0da5a48 4647
1da177e4
LT
4648 /* The addrconf netdev notifier requires that loopback_dev
4649 * has it's ipv6 private information allocated and setup
4650 * before it can bring up and give link-local addresses
4651 * to other devices which are up.
4652 *
4653 * Unfortunately, loopback_dev is not necessarily the first
4654 * entry in the global dev_base list of net devices. In fact,
4655 * it is likely to be the very last entry on that list.
4656 * So this causes the notifier registry below to try and
4657 * give link-local addresses to all devices besides loopback_dev
4658 * first, then loopback_dev, which cases all the non-loopback_dev
4659 * devices to fail to get a link-local address.
4660 *
4661 * So, as a temporary fix, allocate the ipv6 structure for
4662 * loopback_dev first by hand.
4663 * Longer term, all of the dependencies ipv6 has upon the loopback
4664 * device and it being up should be removed.
4665 */
4666 rtnl_lock();
2774c7ab 4667 if (!ipv6_add_dev(init_net.loopback_dev))
1da177e4
LT
4668 err = -ENOMEM;
4669 rtnl_unlock();
4670 if (err)
e0da5a48 4671 goto errlo;
1da177e4 4672
c2e21293 4673 for (i = 0; i < IN6_ADDR_HSIZE; i++)
4674 INIT_HLIST_HEAD(&inet6_addr_lst[i]);
4675
1da177e4
LT
4676 register_netdevice_notifier(&ipv6_dev_notf);
4677
1da177e4 4678 addrconf_verify(0);
c127ea2c
TG
4679
4680 err = __rtnl_register(PF_INET6, RTM_GETLINK, NULL, inet6_dump_ifinfo);
4681 if (err < 0)
4682 goto errout;
4683
4684 /* Only the first call to __rtnl_register can fail */
4685 __rtnl_register(PF_INET6, RTM_NEWADDR, inet6_rtm_newaddr, NULL);
4686 __rtnl_register(PF_INET6, RTM_DELADDR, inet6_rtm_deladdr, NULL);
4687 __rtnl_register(PF_INET6, RTM_GETADDR, inet6_rtm_getaddr, inet6_dump_ifaddr);
4688 __rtnl_register(PF_INET6, RTM_GETMULTICAST, NULL, inet6_dump_ifmcaddr);
4689 __rtnl_register(PF_INET6, RTM_GETANYCAST, NULL, inet6_dump_ifacaddr);
4690
2a8cc6c8
YH
4691 ipv6_addr_label_rtnl_register();
4692
1da177e4 4693 return 0;
c127ea2c
TG
4694errout:
4695 unregister_netdevice_notifier(&ipv6_dev_notf);
e0da5a48
PE
4696errlo:
4697 unregister_pernet_subsys(&addrconf_ops);
2cc6d2bf
NH
4698out_addrlabel:
4699 ipv6_addr_label_cleanup();
4700out:
c127ea2c 4701 return err;
1da177e4
LT
4702}
4703
09f7709f 4704void addrconf_cleanup(void)
1da177e4 4705{
176c39af 4706 struct net_device *dev;
1da177e4
LT
4707 int i;
4708
4709 unregister_netdevice_notifier(&ipv6_dev_notf);
e0da5a48 4710 unregister_pernet_subsys(&addrconf_ops);
2cc6d2bf 4711 ipv6_addr_label_cleanup();
1da177e4
LT
4712
4713 rtnl_lock();
4714
176c39af
DL
4715 /* clean dev list */
4716 for_each_netdev(&init_net, dev) {
4717 if (__in6_dev_get(dev) == NULL)
4718 continue;
4719 addrconf_ifdown(dev, 1);
4720 }
4721 addrconf_ifdown(init_net.loopback_dev, 2);
4722
1da177e4
LT
4723 /*
4724 * Check hash table.
4725 */
5c578aed 4726 spin_lock_bh(&addrconf_hash_lock);
c2e21293 4727 for (i = 0; i < IN6_ADDR_HSIZE; i++)
4728 WARN_ON(!hlist_empty(&inet6_addr_lst[i]));
5c578aed 4729 spin_unlock_bh(&addrconf_hash_lock);
1da177e4
LT
4730
4731 del_timer(&addr_chk_timer);
1da177e4 4732 rtnl_unlock();
1da177e4 4733}