Merge git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6
[wrt350n-kernel.git] / net / ipv6 / addrconf.c
blob81a1e44dde88e66961111e0d19c1f8ffaddb6fad
1 /*
2 * IPv6 Address [auto]configuration
3 * Linux INET6 implementation
5 * Authors:
6 * Pedro Roque <roque@di.fc.ul.pt>
7 * Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
9 * $Id: addrconf.c,v 1.69 2001/10/31 21:55:54 davem Exp $
11 * This program is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU General Public License
13 * as published by the Free Software Foundation; either version
14 * 2 of the License, or (at your option) any later version.
18 * Changes:
20 * Janos Farkas : delete timer on ifdown
21 * <chexum@bankinf.banki.hu>
22 * Andi Kleen : kill double kfree on module
23 * unload.
24 * Maciej W. Rozycki : FDDI support
25 * sekiya@USAGI : Don't send too many RS
26 * packets.
27 * yoshfuji@USAGI : Fixed interval between DAD
28 * packets.
29 * YOSHIFUJI Hideaki @USAGI : improved accuracy of
30 * address validation timer.
31 * YOSHIFUJI Hideaki @USAGI : Privacy Extensions (RFC3041)
32 * support.
33 * Yuji SEKIYA @USAGI : Don't assign a same IPv6
34 * address on a same interface.
35 * YOSHIFUJI Hideaki @USAGI : ARCnet support
36 * YOSHIFUJI Hideaki @USAGI : convert /proc/net/if_inet6 to
37 * seq_file.
38 * YOSHIFUJI Hideaki @USAGI : improved source address
39 * selection; consider scope,
40 * status etc.
43 #include <linux/errno.h>
44 #include <linux/types.h>
45 #include <linux/socket.h>
46 #include <linux/sockios.h>
47 #include <linux/net.h>
48 #include <linux/in6.h>
49 #include <linux/netdevice.h>
50 #include <linux/if_addr.h>
51 #include <linux/if_arp.h>
52 #include <linux/if_arcnet.h>
53 #include <linux/if_infiniband.h>
54 #include <linux/route.h>
55 #include <linux/inetdevice.h>
56 #include <linux/init.h>
57 #ifdef CONFIG_SYSCTL
58 #include <linux/sysctl.h>
59 #endif
60 #include <linux/capability.h>
61 #include <linux/delay.h>
62 #include <linux/notifier.h>
63 #include <linux/string.h>
65 #include <net/net_namespace.h>
66 #include <net/sock.h>
67 #include <net/snmp.h>
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>
76 #include <net/netlink.h>
77 #include <net/pkt_sched.h>
78 #include <linux/if_tunnel.h>
79 #include <linux/rtnetlink.h>
81 #ifdef CONFIG_IPV6_PRIVACY
82 #include <linux/random.h>
83 #endif
85 #include <asm/uaccess.h>
86 #include <asm/unaligned.h>
88 #include <linux/proc_fs.h>
89 #include <linux/seq_file.h>
91 /* Set to 3 to get tracing... */
92 #define ACONF_DEBUG 2
94 #if ACONF_DEBUG >= 3
95 #define ADBG(x) printk x
96 #else
97 #define ADBG(x)
98 #endif
100 #define INFINITY_LIFE_TIME 0xFFFFFFFF
101 #define TIME_DELTA(a,b) ((unsigned long)((long)(a) - (long)(b)))
103 #ifdef CONFIG_SYSCTL
104 static void addrconf_sysctl_register(struct inet6_dev *idev);
105 static void addrconf_sysctl_unregister(struct inet6_dev *idev);
106 #else
107 static inline void addrconf_sysctl_register(struct inet6_dev *idev)
111 static inline void addrconf_sysctl_unregister(struct inet6_dev *idev)
114 #endif
116 #ifdef CONFIG_IPV6_PRIVACY
117 static int __ipv6_regen_rndid(struct inet6_dev *idev);
118 static int __ipv6_try_regen_rndid(struct inet6_dev *idev, struct in6_addr *tmpaddr);
119 static void ipv6_regen_rndid(unsigned long data);
121 static int desync_factor = MAX_DESYNC_FACTOR * HZ;
122 #endif
124 static int ipv6_count_addresses(struct inet6_dev *idev);
127 * Configured unicast address hash table
129 static struct inet6_ifaddr *inet6_addr_lst[IN6_ADDR_HSIZE];
130 static DEFINE_RWLOCK(addrconf_hash_lock);
132 static void addrconf_verify(unsigned long);
134 static DEFINE_TIMER(addr_chk_timer, addrconf_verify, 0, 0);
135 static DEFINE_SPINLOCK(addrconf_verify_lock);
137 static void addrconf_join_anycast(struct inet6_ifaddr *ifp);
138 static void addrconf_leave_anycast(struct inet6_ifaddr *ifp);
140 static int addrconf_ifdown(struct net_device *dev, int how);
142 static void addrconf_dad_start(struct inet6_ifaddr *ifp, u32 flags);
143 static void addrconf_dad_timer(unsigned long data);
144 static void addrconf_dad_completed(struct inet6_ifaddr *ifp);
145 static void addrconf_dad_run(struct inet6_dev *idev);
146 static void addrconf_rs_timer(unsigned long data);
147 static void __ipv6_ifa_notify(int event, struct inet6_ifaddr *ifa);
148 static void ipv6_ifa_notify(int event, struct inet6_ifaddr *ifa);
150 static void inet6_prefix_notify(int event, struct inet6_dev *idev,
151 struct prefix_info *pinfo);
152 static int ipv6_chk_same_addr(struct net *net, const struct in6_addr *addr,
153 struct net_device *dev);
155 static ATOMIC_NOTIFIER_HEAD(inet6addr_chain);
157 struct ipv6_devconf ipv6_devconf __read_mostly = {
158 .forwarding = 0,
159 .hop_limit = IPV6_DEFAULT_HOPLIMIT,
160 .mtu6 = IPV6_MIN_MTU,
161 .accept_ra = 1,
162 .accept_redirects = 1,
163 .autoconf = 1,
164 .force_mld_version = 0,
165 .dad_transmits = 1,
166 .rtr_solicits = MAX_RTR_SOLICITATIONS,
167 .rtr_solicit_interval = RTR_SOLICITATION_INTERVAL,
168 .rtr_solicit_delay = MAX_RTR_SOLICITATION_DELAY,
169 #ifdef CONFIG_IPV6_PRIVACY
170 .use_tempaddr = 0,
171 .temp_valid_lft = TEMP_VALID_LIFETIME,
172 .temp_prefered_lft = TEMP_PREFERRED_LIFETIME,
173 .regen_max_retry = REGEN_MAX_RETRY,
174 .max_desync_factor = MAX_DESYNC_FACTOR,
175 #endif
176 .max_addresses = IPV6_MAX_ADDRESSES,
177 .accept_ra_defrtr = 1,
178 .accept_ra_pinfo = 1,
179 #ifdef CONFIG_IPV6_ROUTER_PREF
180 .accept_ra_rtr_pref = 1,
181 .rtr_probe_interval = 60 * HZ,
182 #ifdef CONFIG_IPV6_ROUTE_INFO
183 .accept_ra_rt_info_max_plen = 0,
184 #endif
185 #endif
186 .proxy_ndp = 0,
187 .accept_source_route = 0, /* we do not accept RH0 by default. */
190 static struct ipv6_devconf ipv6_devconf_dflt __read_mostly = {
191 .forwarding = 0,
192 .hop_limit = IPV6_DEFAULT_HOPLIMIT,
193 .mtu6 = IPV6_MIN_MTU,
194 .accept_ra = 1,
195 .accept_redirects = 1,
196 .autoconf = 1,
197 .dad_transmits = 1,
198 .rtr_solicits = MAX_RTR_SOLICITATIONS,
199 .rtr_solicit_interval = RTR_SOLICITATION_INTERVAL,
200 .rtr_solicit_delay = MAX_RTR_SOLICITATION_DELAY,
201 #ifdef CONFIG_IPV6_PRIVACY
202 .use_tempaddr = 0,
203 .temp_valid_lft = TEMP_VALID_LIFETIME,
204 .temp_prefered_lft = TEMP_PREFERRED_LIFETIME,
205 .regen_max_retry = REGEN_MAX_RETRY,
206 .max_desync_factor = MAX_DESYNC_FACTOR,
207 #endif
208 .max_addresses = IPV6_MAX_ADDRESSES,
209 .accept_ra_defrtr = 1,
210 .accept_ra_pinfo = 1,
211 #ifdef CONFIG_IPV6_ROUTER_PREF
212 .accept_ra_rtr_pref = 1,
213 .rtr_probe_interval = 60 * HZ,
214 #ifdef CONFIG_IPV6_ROUTE_INFO
215 .accept_ra_rt_info_max_plen = 0,
216 #endif
217 #endif
218 .proxy_ndp = 0,
219 .accept_source_route = 0, /* we do not accept RH0 by default. */
222 /* IPv6 Wildcard Address and Loopback Address defined by RFC2553 */
223 const struct in6_addr in6addr_any = IN6ADDR_ANY_INIT;
224 const struct in6_addr in6addr_loopback = IN6ADDR_LOOPBACK_INIT;
226 /* Check if a valid qdisc is available */
227 static inline int addrconf_qdisc_ok(struct net_device *dev)
229 return (dev->qdisc != &noop_qdisc);
232 static void addrconf_del_timer(struct inet6_ifaddr *ifp)
234 if (del_timer(&ifp->timer))
235 __in6_ifa_put(ifp);
238 enum addrconf_timer_t
240 AC_NONE,
241 AC_DAD,
242 AC_RS,
245 static void addrconf_mod_timer(struct inet6_ifaddr *ifp,
246 enum addrconf_timer_t what,
247 unsigned long when)
249 if (!del_timer(&ifp->timer))
250 in6_ifa_hold(ifp);
252 switch (what) {
253 case AC_DAD:
254 ifp->timer.function = addrconf_dad_timer;
255 break;
256 case AC_RS:
257 ifp->timer.function = addrconf_rs_timer;
258 break;
259 default:;
261 ifp->timer.expires = jiffies + when;
262 add_timer(&ifp->timer);
265 static int snmp6_alloc_dev(struct inet6_dev *idev)
267 if (snmp_mib_init((void **)idev->stats.ipv6,
268 sizeof(struct ipstats_mib)) < 0)
269 goto err_ip;
270 if (snmp_mib_init((void **)idev->stats.icmpv6,
271 sizeof(struct icmpv6_mib)) < 0)
272 goto err_icmp;
273 if (snmp_mib_init((void **)idev->stats.icmpv6msg,
274 sizeof(struct icmpv6msg_mib)) < 0)
275 goto err_icmpmsg;
277 return 0;
279 err_icmpmsg:
280 snmp_mib_free((void **)idev->stats.icmpv6);
281 err_icmp:
282 snmp_mib_free((void **)idev->stats.ipv6);
283 err_ip:
284 return -ENOMEM;
287 static void snmp6_free_dev(struct inet6_dev *idev)
289 snmp_mib_free((void **)idev->stats.icmpv6msg);
290 snmp_mib_free((void **)idev->stats.icmpv6);
291 snmp_mib_free((void **)idev->stats.ipv6);
294 /* Nobody refers to this device, we may destroy it. */
296 static void in6_dev_finish_destroy_rcu(struct rcu_head *head)
298 struct inet6_dev *idev = container_of(head, struct inet6_dev, rcu);
299 kfree(idev);
302 void in6_dev_finish_destroy(struct inet6_dev *idev)
304 struct net_device *dev = idev->dev;
305 BUG_TRAP(idev->addr_list==NULL);
306 BUG_TRAP(idev->mc_list==NULL);
307 #ifdef NET_REFCNT_DEBUG
308 printk(KERN_DEBUG "in6_dev_finish_destroy: %s\n", dev ? dev->name : "NIL");
309 #endif
310 dev_put(dev);
311 if (!idev->dead) {
312 printk("Freeing alive inet6 device %p\n", idev);
313 return;
315 snmp6_free_dev(idev);
316 call_rcu(&idev->rcu, in6_dev_finish_destroy_rcu);
319 EXPORT_SYMBOL(in6_dev_finish_destroy);
321 static struct inet6_dev * ipv6_add_dev(struct net_device *dev)
323 struct inet6_dev *ndev;
324 struct in6_addr maddr;
326 ASSERT_RTNL();
328 if (dev->mtu < IPV6_MIN_MTU)
329 return NULL;
331 ndev = kzalloc(sizeof(struct inet6_dev), GFP_KERNEL);
333 if (ndev == NULL)
334 return NULL;
336 rwlock_init(&ndev->lock);
337 ndev->dev = dev;
338 memcpy(&ndev->cnf, dev->nd_net->ipv6.devconf_dflt, sizeof(ndev->cnf));
339 ndev->cnf.mtu6 = dev->mtu;
340 ndev->cnf.sysctl = NULL;
341 ndev->nd_parms = neigh_parms_alloc(dev, &nd_tbl);
342 if (ndev->nd_parms == NULL) {
343 kfree(ndev);
344 return NULL;
346 /* We refer to the device */
347 dev_hold(dev);
349 if (snmp6_alloc_dev(ndev) < 0) {
350 ADBG((KERN_WARNING
351 "%s(): cannot allocate memory for statistics; dev=%s.\n",
352 __FUNCTION__, dev->name));
353 neigh_parms_release(&nd_tbl, ndev->nd_parms);
354 ndev->dead = 1;
355 in6_dev_finish_destroy(ndev);
356 return NULL;
359 if (snmp6_register_dev(ndev) < 0) {
360 ADBG((KERN_WARNING
361 "%s(): cannot create /proc/net/dev_snmp6/%s\n",
362 __FUNCTION__, dev->name));
363 neigh_parms_release(&nd_tbl, ndev->nd_parms);
364 ndev->dead = 1;
365 in6_dev_finish_destroy(ndev);
366 return NULL;
369 /* One reference from device. We must do this before
370 * we invoke __ipv6_regen_rndid().
372 in6_dev_hold(ndev);
374 #ifdef CONFIG_IPV6_PRIVACY
375 setup_timer(&ndev->regen_timer, ipv6_regen_rndid, (unsigned long)ndev);
376 if ((dev->flags&IFF_LOOPBACK) ||
377 dev->type == ARPHRD_TUNNEL ||
378 #if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
379 dev->type == ARPHRD_SIT ||
380 #endif
381 dev->type == ARPHRD_NONE) {
382 printk(KERN_INFO
383 "%s: Disabled Privacy Extensions\n",
384 dev->name);
385 ndev->cnf.use_tempaddr = -1;
387 if (dev->type == ARPHRD_SIT && (dev->priv_flags & IFF_ISATAP)) {
388 printk(KERN_INFO
389 "%s: Disabled Multicast RS\n",
390 dev->name);
391 ndev->cnf.rtr_solicits = 0;
393 } else {
394 in6_dev_hold(ndev);
395 ipv6_regen_rndid((unsigned long) ndev);
397 #endif
399 if (netif_running(dev) && addrconf_qdisc_ok(dev))
400 ndev->if_flags |= IF_READY;
402 ipv6_mc_init_dev(ndev);
403 ndev->tstamp = jiffies;
404 addrconf_sysctl_register(ndev);
405 /* protected by rtnl_lock */
406 rcu_assign_pointer(dev->ip6_ptr, ndev);
408 /* Join all-node multicast group */
409 ipv6_addr_all_nodes(&maddr);
410 ipv6_dev_mc_inc(dev, &maddr);
412 return ndev;
415 static struct inet6_dev * ipv6_find_idev(struct net_device *dev)
417 struct inet6_dev *idev;
419 ASSERT_RTNL();
421 if ((idev = __in6_dev_get(dev)) == NULL) {
422 if ((idev = ipv6_add_dev(dev)) == NULL)
423 return NULL;
426 if (dev->flags&IFF_UP)
427 ipv6_mc_up(idev);
428 return idev;
431 #ifdef CONFIG_SYSCTL
432 static void dev_forward_change(struct inet6_dev *idev)
434 struct net_device *dev;
435 struct inet6_ifaddr *ifa;
436 struct in6_addr addr;
438 if (!idev)
439 return;
440 dev = idev->dev;
441 if (dev && (dev->flags & IFF_MULTICAST)) {
442 ipv6_addr_all_routers(&addr);
444 if (idev->cnf.forwarding)
445 ipv6_dev_mc_inc(dev, &addr);
446 else
447 ipv6_dev_mc_dec(dev, &addr);
449 for (ifa=idev->addr_list; ifa; ifa=ifa->if_next) {
450 if (ifa->flags&IFA_F_TENTATIVE)
451 continue;
452 if (idev->cnf.forwarding)
453 addrconf_join_anycast(ifa);
454 else
455 addrconf_leave_anycast(ifa);
460 static void addrconf_forward_change(struct net *net, __s32 newf)
462 struct net_device *dev;
463 struct inet6_dev *idev;
465 read_lock(&dev_base_lock);
466 for_each_netdev(net, dev) {
467 rcu_read_lock();
468 idev = __in6_dev_get(dev);
469 if (idev) {
470 int changed = (!idev->cnf.forwarding) ^ (!newf);
471 idev->cnf.forwarding = newf;
472 if (changed)
473 dev_forward_change(idev);
475 rcu_read_unlock();
477 read_unlock(&dev_base_lock);
480 static void addrconf_fixup_forwarding(struct ctl_table *table, int *p, int old)
482 struct net *net;
484 net = (struct net *)table->extra2;
485 if (p == &net->ipv6.devconf_dflt->forwarding)
486 return;
488 if (p == &net->ipv6.devconf_all->forwarding) {
489 __s32 newf = net->ipv6.devconf_all->forwarding;
490 net->ipv6.devconf_dflt->forwarding = newf;
491 addrconf_forward_change(net, newf);
492 } else if ((!*p) ^ (!old))
493 dev_forward_change((struct inet6_dev *)table->extra1);
495 if (*p)
496 rt6_purge_dflt_routers();
498 #endif
500 /* Nobody refers to this ifaddr, destroy it */
502 void inet6_ifa_finish_destroy(struct inet6_ifaddr *ifp)
504 BUG_TRAP(ifp->if_next==NULL);
505 BUG_TRAP(ifp->lst_next==NULL);
506 #ifdef NET_REFCNT_DEBUG
507 printk(KERN_DEBUG "inet6_ifa_finish_destroy\n");
508 #endif
510 in6_dev_put(ifp->idev);
512 if (del_timer(&ifp->timer))
513 printk("Timer is still running, when freeing ifa=%p\n", ifp);
515 if (!ifp->dead) {
516 printk("Freeing alive inet6 address %p\n", ifp);
517 return;
519 dst_release(&ifp->rt->u.dst);
521 kfree(ifp);
524 static void
525 ipv6_link_dev_addr(struct inet6_dev *idev, struct inet6_ifaddr *ifp)
527 struct inet6_ifaddr *ifa, **ifap;
528 int ifp_scope = ipv6_addr_src_scope(&ifp->addr);
531 * Each device address list is sorted in order of scope -
532 * global before linklocal.
534 for (ifap = &idev->addr_list; (ifa = *ifap) != NULL;
535 ifap = &ifa->if_next) {
536 if (ifp_scope >= ipv6_addr_src_scope(&ifa->addr))
537 break;
540 ifp->if_next = *ifap;
541 *ifap = ifp;
544 /* On success it returns ifp with increased reference count */
546 static struct inet6_ifaddr *
547 ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr, int pfxlen,
548 int scope, u32 flags)
550 struct inet6_ifaddr *ifa = NULL;
551 struct rt6_info *rt;
552 int hash;
553 int err = 0;
555 rcu_read_lock_bh();
556 if (idev->dead) {
557 err = -ENODEV; /*XXX*/
558 goto out2;
561 write_lock(&addrconf_hash_lock);
563 /* Ignore adding duplicate addresses on an interface */
564 if (ipv6_chk_same_addr(&init_net, addr, idev->dev)) {
565 ADBG(("ipv6_add_addr: already assigned\n"));
566 err = -EEXIST;
567 goto out;
570 ifa = kzalloc(sizeof(struct inet6_ifaddr), GFP_ATOMIC);
572 if (ifa == NULL) {
573 ADBG(("ipv6_add_addr: malloc failed\n"));
574 err = -ENOBUFS;
575 goto out;
578 rt = addrconf_dst_alloc(idev, addr, 0);
579 if (IS_ERR(rt)) {
580 err = PTR_ERR(rt);
581 goto out;
584 ipv6_addr_copy(&ifa->addr, addr);
586 spin_lock_init(&ifa->lock);
587 init_timer(&ifa->timer);
588 ifa->timer.data = (unsigned long) ifa;
589 ifa->scope = scope;
590 ifa->prefix_len = pfxlen;
591 ifa->flags = flags | IFA_F_TENTATIVE;
592 ifa->cstamp = ifa->tstamp = jiffies;
594 ifa->rt = rt;
597 * part one of RFC 4429, section 3.3
598 * We should not configure an address as
599 * optimistic if we do not yet know the link
600 * layer address of our nexhop router
603 if (rt->rt6i_nexthop == NULL)
604 ifa->flags &= ~IFA_F_OPTIMISTIC;
606 ifa->idev = idev;
607 in6_dev_hold(idev);
608 /* For caller */
609 in6_ifa_hold(ifa);
611 /* Add to big hash table */
612 hash = ipv6_addr_hash(addr);
614 ifa->lst_next = inet6_addr_lst[hash];
615 inet6_addr_lst[hash] = ifa;
616 in6_ifa_hold(ifa);
617 write_unlock(&addrconf_hash_lock);
619 write_lock(&idev->lock);
620 /* Add to inet6_dev unicast addr list. */
621 ipv6_link_dev_addr(idev, ifa);
623 #ifdef CONFIG_IPV6_PRIVACY
624 if (ifa->flags&IFA_F_TEMPORARY) {
625 ifa->tmp_next = idev->tempaddr_list;
626 idev->tempaddr_list = ifa;
627 in6_ifa_hold(ifa);
629 #endif
631 in6_ifa_hold(ifa);
632 write_unlock(&idev->lock);
633 out2:
634 rcu_read_unlock_bh();
636 if (likely(err == 0))
637 atomic_notifier_call_chain(&inet6addr_chain, NETDEV_UP, ifa);
638 else {
639 kfree(ifa);
640 ifa = ERR_PTR(err);
643 return ifa;
644 out:
645 write_unlock(&addrconf_hash_lock);
646 goto out2;
649 /* This function wants to get referenced ifp and releases it before return */
651 static void ipv6_del_addr(struct inet6_ifaddr *ifp)
653 struct inet6_ifaddr *ifa, **ifap;
654 struct inet6_dev *idev = ifp->idev;
655 int hash;
656 int deleted = 0, onlink = 0;
657 unsigned long expires = jiffies;
659 hash = ipv6_addr_hash(&ifp->addr);
661 ifp->dead = 1;
663 write_lock_bh(&addrconf_hash_lock);
664 for (ifap = &inet6_addr_lst[hash]; (ifa=*ifap) != NULL;
665 ifap = &ifa->lst_next) {
666 if (ifa == ifp) {
667 *ifap = ifa->lst_next;
668 __in6_ifa_put(ifp);
669 ifa->lst_next = NULL;
670 break;
673 write_unlock_bh(&addrconf_hash_lock);
675 write_lock_bh(&idev->lock);
676 #ifdef CONFIG_IPV6_PRIVACY
677 if (ifp->flags&IFA_F_TEMPORARY) {
678 for (ifap = &idev->tempaddr_list; (ifa=*ifap) != NULL;
679 ifap = &ifa->tmp_next) {
680 if (ifa == ifp) {
681 *ifap = ifa->tmp_next;
682 if (ifp->ifpub) {
683 in6_ifa_put(ifp->ifpub);
684 ifp->ifpub = NULL;
686 __in6_ifa_put(ifp);
687 ifa->tmp_next = NULL;
688 break;
692 #endif
694 for (ifap = &idev->addr_list; (ifa=*ifap) != NULL;) {
695 if (ifa == ifp) {
696 *ifap = ifa->if_next;
697 __in6_ifa_put(ifp);
698 ifa->if_next = NULL;
699 if (!(ifp->flags & IFA_F_PERMANENT) || onlink > 0)
700 break;
701 deleted = 1;
702 continue;
703 } else if (ifp->flags & IFA_F_PERMANENT) {
704 if (ipv6_prefix_equal(&ifa->addr, &ifp->addr,
705 ifp->prefix_len)) {
706 if (ifa->flags & IFA_F_PERMANENT) {
707 onlink = 1;
708 if (deleted)
709 break;
710 } else {
711 unsigned long lifetime;
713 if (!onlink)
714 onlink = -1;
716 spin_lock(&ifa->lock);
717 lifetime = min_t(unsigned long,
718 ifa->valid_lft, 0x7fffffffUL/HZ);
719 if (time_before(expires,
720 ifa->tstamp + lifetime * HZ))
721 expires = ifa->tstamp + lifetime * HZ;
722 spin_unlock(&ifa->lock);
726 ifap = &ifa->if_next;
728 write_unlock_bh(&idev->lock);
730 ipv6_ifa_notify(RTM_DELADDR, ifp);
732 atomic_notifier_call_chain(&inet6addr_chain, NETDEV_DOWN, ifp);
734 addrconf_del_timer(ifp);
737 * Purge or update corresponding prefix
739 * 1) we don't purge prefix here if address was not permanent.
740 * prefix is managed by its own lifetime.
741 * 2) if there're no addresses, delete prefix.
742 * 3) if there're still other permanent address(es),
743 * corresponding prefix is still permanent.
744 * 4) otherwise, update prefix lifetime to the
745 * longest valid lifetime among the corresponding
746 * addresses on the device.
747 * Note: subsequent RA will update lifetime.
749 * --yoshfuji
751 if ((ifp->flags & IFA_F_PERMANENT) && onlink < 1) {
752 struct in6_addr prefix;
753 struct rt6_info *rt;
755 ipv6_addr_prefix(&prefix, &ifp->addr, ifp->prefix_len);
756 rt = rt6_lookup(&prefix, NULL, ifp->idev->dev->ifindex, 1);
758 if (rt && ((rt->rt6i_flags & (RTF_GATEWAY | RTF_DEFAULT)) == 0)) {
759 if (onlink == 0) {
760 ip6_del_rt(rt);
761 rt = NULL;
762 } else if (!(rt->rt6i_flags & RTF_EXPIRES)) {
763 rt->rt6i_expires = expires;
764 rt->rt6i_flags |= RTF_EXPIRES;
767 dst_release(&rt->u.dst);
770 in6_ifa_put(ifp);
773 #ifdef CONFIG_IPV6_PRIVACY
774 static int ipv6_create_tempaddr(struct inet6_ifaddr *ifp, struct inet6_ifaddr *ift)
776 struct inet6_dev *idev = ifp->idev;
777 struct in6_addr addr, *tmpaddr;
778 unsigned long tmp_prefered_lft, tmp_valid_lft, tmp_cstamp, tmp_tstamp;
779 int tmp_plen;
780 int ret = 0;
781 int max_addresses;
782 u32 addr_flags;
784 write_lock(&idev->lock);
785 if (ift) {
786 spin_lock_bh(&ift->lock);
787 memcpy(&addr.s6_addr[8], &ift->addr.s6_addr[8], 8);
788 spin_unlock_bh(&ift->lock);
789 tmpaddr = &addr;
790 } else {
791 tmpaddr = NULL;
793 retry:
794 in6_dev_hold(idev);
795 if (idev->cnf.use_tempaddr <= 0) {
796 write_unlock(&idev->lock);
797 printk(KERN_INFO
798 "ipv6_create_tempaddr(): use_tempaddr is disabled.\n");
799 in6_dev_put(idev);
800 ret = -1;
801 goto out;
803 spin_lock_bh(&ifp->lock);
804 if (ifp->regen_count++ >= idev->cnf.regen_max_retry) {
805 idev->cnf.use_tempaddr = -1; /*XXX*/
806 spin_unlock_bh(&ifp->lock);
807 write_unlock(&idev->lock);
808 printk(KERN_WARNING
809 "ipv6_create_tempaddr(): regeneration time exceeded. disabled temporary address support.\n");
810 in6_dev_put(idev);
811 ret = -1;
812 goto out;
814 in6_ifa_hold(ifp);
815 memcpy(addr.s6_addr, ifp->addr.s6_addr, 8);
816 if (__ipv6_try_regen_rndid(idev, tmpaddr) < 0) {
817 spin_unlock_bh(&ifp->lock);
818 write_unlock(&idev->lock);
819 printk(KERN_WARNING
820 "ipv6_create_tempaddr(): regeneration of randomized interface id failed.\n");
821 in6_ifa_put(ifp);
822 in6_dev_put(idev);
823 ret = -1;
824 goto out;
826 memcpy(&addr.s6_addr[8], idev->rndid, 8);
827 tmp_valid_lft = min_t(__u32,
828 ifp->valid_lft,
829 idev->cnf.temp_valid_lft);
830 tmp_prefered_lft = min_t(__u32,
831 ifp->prefered_lft,
832 idev->cnf.temp_prefered_lft - desync_factor / HZ);
833 tmp_plen = ifp->prefix_len;
834 max_addresses = idev->cnf.max_addresses;
835 tmp_cstamp = ifp->cstamp;
836 tmp_tstamp = ifp->tstamp;
837 spin_unlock_bh(&ifp->lock);
839 write_unlock(&idev->lock);
841 addr_flags = IFA_F_TEMPORARY;
842 /* set in addrconf_prefix_rcv() */
843 if (ifp->flags & IFA_F_OPTIMISTIC)
844 addr_flags |= IFA_F_OPTIMISTIC;
846 ift = !max_addresses ||
847 ipv6_count_addresses(idev) < max_addresses ?
848 ipv6_add_addr(idev, &addr, tmp_plen,
849 ipv6_addr_type(&addr)&IPV6_ADDR_SCOPE_MASK,
850 addr_flags) : NULL;
851 if (!ift || IS_ERR(ift)) {
852 in6_ifa_put(ifp);
853 in6_dev_put(idev);
854 printk(KERN_INFO
855 "ipv6_create_tempaddr(): retry temporary address regeneration.\n");
856 tmpaddr = &addr;
857 write_lock(&idev->lock);
858 goto retry;
861 spin_lock_bh(&ift->lock);
862 ift->ifpub = ifp;
863 ift->valid_lft = tmp_valid_lft;
864 ift->prefered_lft = tmp_prefered_lft;
865 ift->cstamp = tmp_cstamp;
866 ift->tstamp = tmp_tstamp;
867 spin_unlock_bh(&ift->lock);
869 addrconf_dad_start(ift, 0);
870 in6_ifa_put(ift);
871 in6_dev_put(idev);
872 out:
873 return ret;
875 #endif
878 * Choose an appropriate source address (RFC3484)
880 struct ipv6_saddr_score {
881 int addr_type;
882 unsigned int attrs;
883 int matchlen;
884 int scope;
885 unsigned int rule;
888 #define IPV6_SADDR_SCORE_LOCAL 0x0001
889 #define IPV6_SADDR_SCORE_PREFERRED 0x0004
890 #define IPV6_SADDR_SCORE_HOA 0x0008
891 #define IPV6_SADDR_SCORE_OIF 0x0010
892 #define IPV6_SADDR_SCORE_LABEL 0x0020
893 #define IPV6_SADDR_SCORE_PRIVACY 0x0040
895 static inline int ipv6_saddr_preferred(int type)
897 if (type & (IPV6_ADDR_MAPPED|IPV6_ADDR_COMPATv4|
898 IPV6_ADDR_LOOPBACK|IPV6_ADDR_RESERVED))
899 return 1;
900 return 0;
903 int ipv6_dev_get_saddr(struct net_device *daddr_dev,
904 struct in6_addr *daddr, struct in6_addr *saddr)
906 struct ipv6_saddr_score hiscore;
907 struct inet6_ifaddr *ifa_result = NULL;
908 int daddr_type = __ipv6_addr_type(daddr);
909 int daddr_scope = __ipv6_addr_src_scope(daddr_type);
910 int daddr_ifindex = daddr_dev ? daddr_dev->ifindex : 0;
911 u32 daddr_label = ipv6_addr_label(daddr, daddr_type, daddr_ifindex);
912 struct net_device *dev;
914 memset(&hiscore, 0, sizeof(hiscore));
916 read_lock(&dev_base_lock);
917 rcu_read_lock();
919 for_each_netdev(&init_net, dev) {
920 struct inet6_dev *idev;
921 struct inet6_ifaddr *ifa;
923 /* Rule 0: Candidate Source Address (section 4)
924 * - multicast and link-local destination address,
925 * the set of candidate source address MUST only
926 * include addresses assigned to interfaces
927 * belonging to the same link as the outgoing
928 * interface.
929 * (- For site-local destination addresses, the
930 * set of candidate source addresses MUST only
931 * include addresses assigned to interfaces
932 * belonging to the same site as the outgoing
933 * interface.)
935 if ((daddr_type & IPV6_ADDR_MULTICAST ||
936 daddr_scope <= IPV6_ADDR_SCOPE_LINKLOCAL) &&
937 daddr_dev && dev != daddr_dev)
938 continue;
940 idev = __in6_dev_get(dev);
941 if (!idev)
942 continue;
944 read_lock_bh(&idev->lock);
945 for (ifa = idev->addr_list; ifa; ifa = ifa->if_next) {
946 struct ipv6_saddr_score score;
948 score.addr_type = __ipv6_addr_type(&ifa->addr);
950 /* Rule 0:
951 * - Tentative Address (RFC2462 section 5.4)
952 * - A tentative address is not considered
953 * "assigned to an interface" in the traditional
954 * sense, unless it is also flagged as optimistic.
955 * - Candidate Source Address (section 4)
956 * - In any case, anycast addresses, multicast
957 * addresses, and the unspecified address MUST
958 * NOT be included in a candidate set.
960 if ((ifa->flags & IFA_F_TENTATIVE) &&
961 (!(ifa->flags & IFA_F_OPTIMISTIC)))
962 continue;
963 if (unlikely(score.addr_type == IPV6_ADDR_ANY ||
964 score.addr_type & IPV6_ADDR_MULTICAST)) {
965 LIMIT_NETDEBUG(KERN_DEBUG
966 "ADDRCONF: unspecified / multicast address "
967 "assigned as unicast address on %s",
968 dev->name);
969 continue;
972 score.attrs = 0;
973 score.matchlen = 0;
974 score.scope = 0;
975 score.rule = 0;
977 if (ifa_result == NULL) {
978 /* record it if the first available entry */
979 goto record_it;
982 /* Rule 1: Prefer same address */
983 if (hiscore.rule < 1) {
984 if (ipv6_addr_equal(&ifa_result->addr, daddr))
985 hiscore.attrs |= IPV6_SADDR_SCORE_LOCAL;
986 hiscore.rule++;
988 if (ipv6_addr_equal(&ifa->addr, daddr)) {
989 score.attrs |= IPV6_SADDR_SCORE_LOCAL;
990 if (!(hiscore.attrs & IPV6_SADDR_SCORE_LOCAL)) {
991 score.rule = 1;
992 goto record_it;
994 } else {
995 if (hiscore.attrs & IPV6_SADDR_SCORE_LOCAL)
996 continue;
999 /* Rule 2: Prefer appropriate scope */
1000 if (hiscore.rule < 2) {
1001 hiscore.scope = __ipv6_addr_src_scope(hiscore.addr_type);
1002 hiscore.rule++;
1004 score.scope = __ipv6_addr_src_scope(score.addr_type);
1005 if (hiscore.scope < score.scope) {
1006 if (hiscore.scope < daddr_scope) {
1007 score.rule = 2;
1008 goto record_it;
1009 } else
1010 continue;
1011 } else if (score.scope < hiscore.scope) {
1012 if (score.scope < daddr_scope)
1013 break; /* addresses sorted by scope */
1014 else {
1015 score.rule = 2;
1016 goto record_it;
1020 /* Rule 3: Avoid deprecated and optimistic addresses */
1021 if (hiscore.rule < 3) {
1022 if (ipv6_saddr_preferred(hiscore.addr_type) ||
1023 (((ifa_result->flags &
1024 (IFA_F_DEPRECATED|IFA_F_OPTIMISTIC)) == 0)))
1025 hiscore.attrs |= IPV6_SADDR_SCORE_PREFERRED;
1026 hiscore.rule++;
1028 if (ipv6_saddr_preferred(score.addr_type) ||
1029 (((ifa->flags &
1030 (IFA_F_DEPRECATED|IFA_F_OPTIMISTIC)) == 0))) {
1031 score.attrs |= IPV6_SADDR_SCORE_PREFERRED;
1032 if (!(hiscore.attrs & IPV6_SADDR_SCORE_PREFERRED)) {
1033 score.rule = 3;
1034 goto record_it;
1036 } else {
1037 if (hiscore.attrs & IPV6_SADDR_SCORE_PREFERRED)
1038 continue;
1041 /* Rule 4: Prefer home address */
1042 #if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
1043 if (hiscore.rule < 4) {
1044 if (ifa_result->flags & IFA_F_HOMEADDRESS)
1045 hiscore.attrs |= IPV6_SADDR_SCORE_HOA;
1046 hiscore.rule++;
1048 if (ifa->flags & IFA_F_HOMEADDRESS) {
1049 score.attrs |= IPV6_SADDR_SCORE_HOA;
1050 if (!(ifa_result->flags & IFA_F_HOMEADDRESS)) {
1051 score.rule = 4;
1052 goto record_it;
1054 } else {
1055 if (hiscore.attrs & IPV6_SADDR_SCORE_HOA)
1056 continue;
1058 #else
1059 if (hiscore.rule < 4)
1060 hiscore.rule++;
1061 #endif
1063 /* Rule 5: Prefer outgoing interface */
1064 if (hiscore.rule < 5) {
1065 if (daddr_dev == NULL ||
1066 daddr_dev == ifa_result->idev->dev)
1067 hiscore.attrs |= IPV6_SADDR_SCORE_OIF;
1068 hiscore.rule++;
1070 if (daddr_dev == NULL ||
1071 daddr_dev == ifa->idev->dev) {
1072 score.attrs |= IPV6_SADDR_SCORE_OIF;
1073 if (!(hiscore.attrs & IPV6_SADDR_SCORE_OIF)) {
1074 score.rule = 5;
1075 goto record_it;
1077 } else {
1078 if (hiscore.attrs & IPV6_SADDR_SCORE_OIF)
1079 continue;
1082 /* Rule 6: Prefer matching label */
1083 if (hiscore.rule < 6) {
1084 if (ipv6_addr_label(&ifa_result->addr,
1085 hiscore.addr_type,
1086 ifa_result->idev->dev->ifindex) == daddr_label)
1087 hiscore.attrs |= IPV6_SADDR_SCORE_LABEL;
1088 hiscore.rule++;
1090 if (ipv6_addr_label(&ifa->addr,
1091 score.addr_type,
1092 ifa->idev->dev->ifindex) == daddr_label) {
1093 score.attrs |= IPV6_SADDR_SCORE_LABEL;
1094 if (!(hiscore.attrs & IPV6_SADDR_SCORE_LABEL)) {
1095 score.rule = 6;
1096 goto record_it;
1098 } else {
1099 if (hiscore.attrs & IPV6_SADDR_SCORE_LABEL)
1100 continue;
1103 #ifdef CONFIG_IPV6_PRIVACY
1104 /* Rule 7: Prefer public address
1105 * Note: prefer temprary address if use_tempaddr >= 2
1107 if (hiscore.rule < 7) {
1108 if ((!(ifa_result->flags & IFA_F_TEMPORARY)) ^
1109 (ifa_result->idev->cnf.use_tempaddr >= 2))
1110 hiscore.attrs |= IPV6_SADDR_SCORE_PRIVACY;
1111 hiscore.rule++;
1113 if ((!(ifa->flags & IFA_F_TEMPORARY)) ^
1114 (ifa->idev->cnf.use_tempaddr >= 2)) {
1115 score.attrs |= IPV6_SADDR_SCORE_PRIVACY;
1116 if (!(hiscore.attrs & IPV6_SADDR_SCORE_PRIVACY)) {
1117 score.rule = 7;
1118 goto record_it;
1120 } else {
1121 if (hiscore.attrs & IPV6_SADDR_SCORE_PRIVACY)
1122 continue;
1124 #else
1125 if (hiscore.rule < 7)
1126 hiscore.rule++;
1127 #endif
1128 /* Rule 8: Use longest matching prefix */
1129 if (hiscore.rule < 8) {
1130 hiscore.matchlen = ipv6_addr_diff(&ifa_result->addr, daddr);
1131 hiscore.rule++;
1133 score.matchlen = ipv6_addr_diff(&ifa->addr, daddr);
1134 if (score.matchlen > hiscore.matchlen) {
1135 score.rule = 8;
1136 goto record_it;
1138 #if 0
1139 else if (score.matchlen < hiscore.matchlen)
1140 continue;
1141 #endif
1143 /* Final Rule: choose first available one */
1144 continue;
1145 record_it:
1146 if (ifa_result)
1147 in6_ifa_put(ifa_result);
1148 in6_ifa_hold(ifa);
1149 ifa_result = ifa;
1150 hiscore = score;
1152 read_unlock_bh(&idev->lock);
1154 rcu_read_unlock();
1155 read_unlock(&dev_base_lock);
1157 if (!ifa_result)
1158 return -EADDRNOTAVAIL;
1160 ipv6_addr_copy(saddr, &ifa_result->addr);
1161 in6_ifa_put(ifa_result);
1162 return 0;
1166 int ipv6_get_saddr(struct dst_entry *dst,
1167 struct in6_addr *daddr, struct in6_addr *saddr)
1169 return ipv6_dev_get_saddr(dst ? ip6_dst_idev(dst)->dev : NULL, daddr, saddr);
1172 EXPORT_SYMBOL(ipv6_get_saddr);
1174 int ipv6_get_lladdr(struct net_device *dev, struct in6_addr *addr,
1175 unsigned char banned_flags)
1177 struct inet6_dev *idev;
1178 int err = -EADDRNOTAVAIL;
1180 rcu_read_lock();
1181 if ((idev = __in6_dev_get(dev)) != NULL) {
1182 struct inet6_ifaddr *ifp;
1184 read_lock_bh(&idev->lock);
1185 for (ifp=idev->addr_list; ifp; ifp=ifp->if_next) {
1186 if (ifp->scope == IFA_LINK && !(ifp->flags & banned_flags)) {
1187 ipv6_addr_copy(addr, &ifp->addr);
1188 err = 0;
1189 break;
1192 read_unlock_bh(&idev->lock);
1194 rcu_read_unlock();
1195 return err;
1198 static int ipv6_count_addresses(struct inet6_dev *idev)
1200 int cnt = 0;
1201 struct inet6_ifaddr *ifp;
1203 read_lock_bh(&idev->lock);
1204 for (ifp=idev->addr_list; ifp; ifp=ifp->if_next)
1205 cnt++;
1206 read_unlock_bh(&idev->lock);
1207 return cnt;
1210 int ipv6_chk_addr(struct net *net, struct in6_addr *addr,
1211 struct net_device *dev, int strict)
1213 struct inet6_ifaddr * ifp;
1214 u8 hash = ipv6_addr_hash(addr);
1216 read_lock_bh(&addrconf_hash_lock);
1217 for(ifp = inet6_addr_lst[hash]; ifp; ifp=ifp->lst_next) {
1218 if (ifp->idev->dev->nd_net != net)
1219 continue;
1220 if (ipv6_addr_equal(&ifp->addr, addr) &&
1221 !(ifp->flags&IFA_F_TENTATIVE)) {
1222 if (dev == NULL || ifp->idev->dev == dev ||
1223 !(ifp->scope&(IFA_LINK|IFA_HOST) || strict))
1224 break;
1227 read_unlock_bh(&addrconf_hash_lock);
1228 return ifp != NULL;
1230 EXPORT_SYMBOL(ipv6_chk_addr);
1232 static
1233 int ipv6_chk_same_addr(struct net *net, const struct in6_addr *addr,
1234 struct net_device *dev)
1236 struct inet6_ifaddr * ifp;
1237 u8 hash = ipv6_addr_hash(addr);
1239 for(ifp = inet6_addr_lst[hash]; ifp; ifp=ifp->lst_next) {
1240 if (ifp->idev->dev->nd_net != net)
1241 continue;
1242 if (ipv6_addr_equal(&ifp->addr, addr)) {
1243 if (dev == NULL || ifp->idev->dev == dev)
1244 break;
1247 return ifp != NULL;
1250 struct inet6_ifaddr *ipv6_get_ifaddr(struct net *net, struct in6_addr *addr,
1251 struct net_device *dev, int strict)
1253 struct inet6_ifaddr * ifp;
1254 u8 hash = ipv6_addr_hash(addr);
1256 read_lock_bh(&addrconf_hash_lock);
1257 for(ifp = inet6_addr_lst[hash]; ifp; ifp=ifp->lst_next) {
1258 if (ifp->idev->dev->nd_net != net)
1259 continue;
1260 if (ipv6_addr_equal(&ifp->addr, addr)) {
1261 if (dev == NULL || ifp->idev->dev == dev ||
1262 !(ifp->scope&(IFA_LINK|IFA_HOST) || strict)) {
1263 in6_ifa_hold(ifp);
1264 break;
1268 read_unlock_bh(&addrconf_hash_lock);
1270 return ifp;
1273 int ipv6_rcv_saddr_equal(const struct sock *sk, const struct sock *sk2)
1275 const struct in6_addr *sk_rcv_saddr6 = &inet6_sk(sk)->rcv_saddr;
1276 const struct in6_addr *sk2_rcv_saddr6 = inet6_rcv_saddr(sk2);
1277 __be32 sk_rcv_saddr = inet_sk(sk)->rcv_saddr;
1278 __be32 sk2_rcv_saddr = inet_rcv_saddr(sk2);
1279 int sk_ipv6only = ipv6_only_sock(sk);
1280 int sk2_ipv6only = inet_v6_ipv6only(sk2);
1281 int addr_type = ipv6_addr_type(sk_rcv_saddr6);
1282 int addr_type2 = sk2_rcv_saddr6 ? ipv6_addr_type(sk2_rcv_saddr6) : IPV6_ADDR_MAPPED;
1284 if (!sk2_rcv_saddr && !sk_ipv6only)
1285 return 1;
1287 if (addr_type2 == IPV6_ADDR_ANY &&
1288 !(sk2_ipv6only && addr_type == IPV6_ADDR_MAPPED))
1289 return 1;
1291 if (addr_type == IPV6_ADDR_ANY &&
1292 !(sk_ipv6only && addr_type2 == IPV6_ADDR_MAPPED))
1293 return 1;
1295 if (sk2_rcv_saddr6 &&
1296 ipv6_addr_equal(sk_rcv_saddr6, sk2_rcv_saddr6))
1297 return 1;
1299 if (addr_type == IPV6_ADDR_MAPPED &&
1300 !sk2_ipv6only &&
1301 (!sk2_rcv_saddr || !sk_rcv_saddr || sk_rcv_saddr == sk2_rcv_saddr))
1302 return 1;
1304 return 0;
1307 /* Gets referenced address, destroys ifaddr */
1309 static void addrconf_dad_stop(struct inet6_ifaddr *ifp)
1311 if (ifp->flags&IFA_F_PERMANENT) {
1312 spin_lock_bh(&ifp->lock);
1313 addrconf_del_timer(ifp);
1314 ifp->flags |= IFA_F_TENTATIVE;
1315 spin_unlock_bh(&ifp->lock);
1316 in6_ifa_put(ifp);
1317 #ifdef CONFIG_IPV6_PRIVACY
1318 } else if (ifp->flags&IFA_F_TEMPORARY) {
1319 struct inet6_ifaddr *ifpub;
1320 spin_lock_bh(&ifp->lock);
1321 ifpub = ifp->ifpub;
1322 if (ifpub) {
1323 in6_ifa_hold(ifpub);
1324 spin_unlock_bh(&ifp->lock);
1325 ipv6_create_tempaddr(ifpub, ifp);
1326 in6_ifa_put(ifpub);
1327 } else {
1328 spin_unlock_bh(&ifp->lock);
1330 ipv6_del_addr(ifp);
1331 #endif
1332 } else
1333 ipv6_del_addr(ifp);
1336 void addrconf_dad_failure(struct inet6_ifaddr *ifp)
1338 if (net_ratelimit())
1339 printk(KERN_INFO "%s: duplicate address detected!\n", ifp->idev->dev->name);
1340 addrconf_dad_stop(ifp);
1343 /* Join to solicited addr multicast group. */
1345 void addrconf_join_solict(struct net_device *dev, struct in6_addr *addr)
1347 struct in6_addr maddr;
1349 if (dev->flags&(IFF_LOOPBACK|IFF_NOARP))
1350 return;
1352 addrconf_addr_solict_mult(addr, &maddr);
1353 ipv6_dev_mc_inc(dev, &maddr);
1356 void addrconf_leave_solict(struct inet6_dev *idev, struct in6_addr *addr)
1358 struct in6_addr maddr;
1360 if (idev->dev->flags&(IFF_LOOPBACK|IFF_NOARP))
1361 return;
1363 addrconf_addr_solict_mult(addr, &maddr);
1364 __ipv6_dev_mc_dec(idev, &maddr);
1367 static void addrconf_join_anycast(struct inet6_ifaddr *ifp)
1369 struct in6_addr addr;
1370 ipv6_addr_prefix(&addr, &ifp->addr, ifp->prefix_len);
1371 if (ipv6_addr_any(&addr))
1372 return;
1373 ipv6_dev_ac_inc(ifp->idev->dev, &addr);
1376 static void addrconf_leave_anycast(struct inet6_ifaddr *ifp)
1378 struct in6_addr addr;
1379 ipv6_addr_prefix(&addr, &ifp->addr, ifp->prefix_len);
1380 if (ipv6_addr_any(&addr))
1381 return;
1382 __ipv6_dev_ac_dec(ifp->idev, &addr);
1385 static int addrconf_ifid_eui48(u8 *eui, struct net_device *dev)
1387 if (dev->addr_len != ETH_ALEN)
1388 return -1;
1389 memcpy(eui, dev->dev_addr, 3);
1390 memcpy(eui + 5, dev->dev_addr + 3, 3);
1393 * The zSeries OSA network cards can be shared among various
1394 * OS instances, but the OSA cards have only one MAC address.
1395 * This leads to duplicate address conflicts in conjunction
1396 * with IPv6 if more than one instance uses the same card.
1398 * The driver for these cards can deliver a unique 16-bit
1399 * identifier for each instance sharing the same card. It is
1400 * placed instead of 0xFFFE in the interface identifier. The
1401 * "u" bit of the interface identifier is not inverted in this
1402 * case. Hence the resulting interface identifier has local
1403 * scope according to RFC2373.
1405 if (dev->dev_id) {
1406 eui[3] = (dev->dev_id >> 8) & 0xFF;
1407 eui[4] = dev->dev_id & 0xFF;
1408 } else {
1409 eui[3] = 0xFF;
1410 eui[4] = 0xFE;
1411 eui[0] ^= 2;
1413 return 0;
1416 static int addrconf_ifid_arcnet(u8 *eui, struct net_device *dev)
1418 /* XXX: inherit EUI-64 from other interface -- yoshfuji */
1419 if (dev->addr_len != ARCNET_ALEN)
1420 return -1;
1421 memset(eui, 0, 7);
1422 eui[7] = *(u8*)dev->dev_addr;
1423 return 0;
1426 static int addrconf_ifid_infiniband(u8 *eui, struct net_device *dev)
1428 if (dev->addr_len != INFINIBAND_ALEN)
1429 return -1;
1430 memcpy(eui, dev->dev_addr + 12, 8);
1431 eui[0] |= 2;
1432 return 0;
1435 static int ipv6_generate_eui64(u8 *eui, struct net_device *dev)
1437 switch (dev->type) {
1438 case ARPHRD_ETHER:
1439 case ARPHRD_FDDI:
1440 case ARPHRD_IEEE802_TR:
1441 return addrconf_ifid_eui48(eui, dev);
1442 case ARPHRD_ARCNET:
1443 return addrconf_ifid_arcnet(eui, dev);
1444 case ARPHRD_INFINIBAND:
1445 return addrconf_ifid_infiniband(eui, dev);
1446 case ARPHRD_SIT:
1447 if (dev->priv_flags & IFF_ISATAP)
1448 return ipv6_isatap_eui64(eui, *(__be32 *)dev->dev_addr);
1450 return -1;
1453 static int ipv6_inherit_eui64(u8 *eui, struct inet6_dev *idev)
1455 int err = -1;
1456 struct inet6_ifaddr *ifp;
1458 read_lock_bh(&idev->lock);
1459 for (ifp=idev->addr_list; ifp; ifp=ifp->if_next) {
1460 if (ifp->scope == IFA_LINK && !(ifp->flags&IFA_F_TENTATIVE)) {
1461 memcpy(eui, ifp->addr.s6_addr+8, 8);
1462 err = 0;
1463 break;
1466 read_unlock_bh(&idev->lock);
1467 return err;
1470 #ifdef CONFIG_IPV6_PRIVACY
1471 /* (re)generation of randomized interface identifier (RFC 3041 3.2, 3.5) */
1472 static int __ipv6_regen_rndid(struct inet6_dev *idev)
1474 regen:
1475 get_random_bytes(idev->rndid, sizeof(idev->rndid));
1476 idev->rndid[0] &= ~0x02;
1479 * <draft-ietf-ipngwg-temp-addresses-v2-00.txt>:
1480 * check if generated address is not inappropriate
1482 * - Reserved subnet anycast (RFC 2526)
1483 * 11111101 11....11 1xxxxxxx
1484 * - ISATAP (RFC4214) 6.1
1485 * 00-00-5E-FE-xx-xx-xx-xx
1486 * - value 0
1487 * - XXX: already assigned to an address on the device
1489 if (idev->rndid[0] == 0xfd &&
1490 (idev->rndid[1]&idev->rndid[2]&idev->rndid[3]&idev->rndid[4]&idev->rndid[5]&idev->rndid[6]) == 0xff &&
1491 (idev->rndid[7]&0x80))
1492 goto regen;
1493 if ((idev->rndid[0]|idev->rndid[1]) == 0) {
1494 if (idev->rndid[2] == 0x5e && idev->rndid[3] == 0xfe)
1495 goto regen;
1496 if ((idev->rndid[2]|idev->rndid[3]|idev->rndid[4]|idev->rndid[5]|idev->rndid[6]|idev->rndid[7]) == 0x00)
1497 goto regen;
1500 return 0;
1503 static void ipv6_regen_rndid(unsigned long data)
1505 struct inet6_dev *idev = (struct inet6_dev *) data;
1506 unsigned long expires;
1508 rcu_read_lock_bh();
1509 write_lock_bh(&idev->lock);
1511 if (idev->dead)
1512 goto out;
1514 if (__ipv6_regen_rndid(idev) < 0)
1515 goto out;
1517 expires = jiffies +
1518 idev->cnf.temp_prefered_lft * HZ -
1519 idev->cnf.regen_max_retry * idev->cnf.dad_transmits * idev->nd_parms->retrans_time - desync_factor;
1520 if (time_before(expires, jiffies)) {
1521 printk(KERN_WARNING
1522 "ipv6_regen_rndid(): too short regeneration interval; timer disabled for %s.\n",
1523 idev->dev->name);
1524 goto out;
1527 if (!mod_timer(&idev->regen_timer, expires))
1528 in6_dev_hold(idev);
1530 out:
1531 write_unlock_bh(&idev->lock);
1532 rcu_read_unlock_bh();
1533 in6_dev_put(idev);
1536 static int __ipv6_try_regen_rndid(struct inet6_dev *idev, struct in6_addr *tmpaddr) {
1537 int ret = 0;
1539 if (tmpaddr && memcmp(idev->rndid, &tmpaddr->s6_addr[8], 8) == 0)
1540 ret = __ipv6_regen_rndid(idev);
1541 return ret;
1543 #endif
1546 * Add prefix route.
1549 static void
1550 addrconf_prefix_route(struct in6_addr *pfx, int plen, struct net_device *dev,
1551 unsigned long expires, u32 flags)
1553 struct fib6_config cfg = {
1554 .fc_table = RT6_TABLE_PREFIX,
1555 .fc_metric = IP6_RT_PRIO_ADDRCONF,
1556 .fc_ifindex = dev->ifindex,
1557 .fc_expires = expires,
1558 .fc_dst_len = plen,
1559 .fc_flags = RTF_UP | flags,
1560 <<<<<<< HEAD:net/ipv6/addrconf.c
1561 =======
1562 .fc_nlinfo.nl_net = &init_net,
1563 >>>>>>> 264e3e889d86e552b4191d69bb60f4f3b383135a:net/ipv6/addrconf.c
1566 ipv6_addr_copy(&cfg.fc_dst, pfx);
1568 /* Prevent useless cloning on PtP SIT.
1569 This thing is done here expecting that the whole
1570 class of non-broadcast devices need not cloning.
1572 #if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
1573 if (dev->type == ARPHRD_SIT && (dev->flags & IFF_POINTOPOINT))
1574 cfg.fc_flags |= RTF_NONEXTHOP;
1575 #endif
1577 ip6_route_add(&cfg);
1580 /* Create "default" multicast route to the interface */
1582 static void addrconf_add_mroute(struct net_device *dev)
1584 struct fib6_config cfg = {
1585 .fc_table = RT6_TABLE_LOCAL,
1586 .fc_metric = IP6_RT_PRIO_ADDRCONF,
1587 .fc_ifindex = dev->ifindex,
1588 .fc_dst_len = 8,
1589 .fc_flags = RTF_UP,
1590 <<<<<<< HEAD:net/ipv6/addrconf.c
1591 =======
1592 .fc_nlinfo.nl_net = &init_net,
1593 >>>>>>> 264e3e889d86e552b4191d69bb60f4f3b383135a:net/ipv6/addrconf.c
1596 ipv6_addr_set(&cfg.fc_dst, htonl(0xFF000000), 0, 0, 0);
1598 ip6_route_add(&cfg);
1601 #if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
1602 static void sit_route_add(struct net_device *dev)
1604 struct fib6_config cfg = {
1605 .fc_table = RT6_TABLE_MAIN,
1606 .fc_metric = IP6_RT_PRIO_ADDRCONF,
1607 .fc_ifindex = dev->ifindex,
1608 .fc_dst_len = 96,
1609 .fc_flags = RTF_UP | RTF_NONEXTHOP,
1610 <<<<<<< HEAD:net/ipv6/addrconf.c
1611 =======
1612 .fc_nlinfo.nl_net = &init_net,
1613 >>>>>>> 264e3e889d86e552b4191d69bb60f4f3b383135a:net/ipv6/addrconf.c
1616 /* prefix length - 96 bits "::d.d.d.d" */
1617 ip6_route_add(&cfg);
1619 #endif
1621 static void addrconf_add_lroute(struct net_device *dev)
1623 struct in6_addr addr;
1625 ipv6_addr_set(&addr, htonl(0xFE800000), 0, 0, 0);
1626 addrconf_prefix_route(&addr, 64, dev, 0, 0);
1629 static struct inet6_dev *addrconf_add_dev(struct net_device *dev)
1631 struct inet6_dev *idev;
1633 ASSERT_RTNL();
1635 if ((idev = ipv6_find_idev(dev)) == NULL)
1636 return NULL;
1638 /* Add default multicast route */
1639 addrconf_add_mroute(dev);
1641 /* Add link local route */
1642 addrconf_add_lroute(dev);
1643 return idev;
1646 void addrconf_prefix_rcv(struct net_device *dev, u8 *opt, int len)
1648 struct prefix_info *pinfo;
1649 __u32 valid_lft;
1650 __u32 prefered_lft;
1651 int addr_type;
1652 unsigned long rt_expires;
1653 struct inet6_dev *in6_dev;
1655 pinfo = (struct prefix_info *) opt;
1657 if (len < sizeof(struct prefix_info)) {
1658 ADBG(("addrconf: prefix option too short\n"));
1659 return;
1663 * Validation checks ([ADDRCONF], page 19)
1666 addr_type = ipv6_addr_type(&pinfo->prefix);
1668 if (addr_type & (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL))
1669 return;
1671 valid_lft = ntohl(pinfo->valid);
1672 prefered_lft = ntohl(pinfo->prefered);
1674 if (prefered_lft > valid_lft) {
1675 if (net_ratelimit())
1676 printk(KERN_WARNING "addrconf: prefix option has invalid lifetime\n");
1677 return;
1680 in6_dev = in6_dev_get(dev);
1682 if (in6_dev == NULL) {
1683 if (net_ratelimit())
1684 printk(KERN_DEBUG "addrconf: device %s not configured\n", dev->name);
1685 return;
1689 * Two things going on here:
1690 * 1) Add routes for on-link prefixes
1691 * 2) Configure prefixes with the auto flag set
1694 /* Avoid arithmetic overflow. Really, we could
1695 save rt_expires in seconds, likely valid_lft,
1696 but it would require division in fib gc, that it
1697 not good.
1699 if (valid_lft >= 0x7FFFFFFF/HZ)
1700 rt_expires = 0x7FFFFFFF - (0x7FFFFFFF % HZ);
1701 else
1702 rt_expires = valid_lft * HZ;
1705 * We convert this (in jiffies) to clock_t later.
1706 * Avoid arithmetic overflow there as well.
1707 * Overflow can happen only if HZ < USER_HZ.
1709 if (HZ < USER_HZ && rt_expires > 0x7FFFFFFF / USER_HZ)
1710 rt_expires = 0x7FFFFFFF / USER_HZ;
1712 if (pinfo->onlink) {
1713 struct rt6_info *rt;
1714 rt = rt6_lookup(&pinfo->prefix, NULL, dev->ifindex, 1);
1716 if (rt && ((rt->rt6i_flags & (RTF_GATEWAY | RTF_DEFAULT)) == 0)) {
1717 if (rt->rt6i_flags&RTF_EXPIRES) {
1718 if (valid_lft == 0) {
1719 ip6_del_rt(rt);
1720 rt = NULL;
1721 } else {
1722 rt->rt6i_expires = jiffies + rt_expires;
1725 } else if (valid_lft) {
1726 addrconf_prefix_route(&pinfo->prefix, pinfo->prefix_len,
1727 dev, jiffies_to_clock_t(rt_expires), RTF_ADDRCONF|RTF_EXPIRES|RTF_PREFIX_RT);
1729 if (rt)
1730 dst_release(&rt->u.dst);
1733 /* Try to figure out our local address for this prefix */
1735 if (pinfo->autoconf && in6_dev->cnf.autoconf) {
1736 struct inet6_ifaddr * ifp;
1737 struct in6_addr addr;
1738 int create = 0, update_lft = 0;
1740 if (pinfo->prefix_len == 64) {
1741 memcpy(&addr, &pinfo->prefix, 8);
1742 if (ipv6_generate_eui64(addr.s6_addr + 8, dev) &&
1743 ipv6_inherit_eui64(addr.s6_addr + 8, in6_dev)) {
1744 in6_dev_put(in6_dev);
1745 return;
1747 goto ok;
1749 if (net_ratelimit())
1750 printk(KERN_DEBUG "IPv6 addrconf: prefix with wrong length %d\n",
1751 pinfo->prefix_len);
1752 in6_dev_put(in6_dev);
1753 return;
1757 ifp = ipv6_get_ifaddr(&init_net, &addr, dev, 1);
1759 if (ifp == NULL && valid_lft) {
1760 int max_addresses = in6_dev->cnf.max_addresses;
1761 u32 addr_flags = 0;
1763 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
1764 if (in6_dev->cnf.optimistic_dad &&
1765 !ipv6_devconf.forwarding)
1766 addr_flags = IFA_F_OPTIMISTIC;
1767 #endif
1769 /* Do not allow to create too much of autoconfigured
1770 * addresses; this would be too easy way to crash kernel.
1772 if (!max_addresses ||
1773 ipv6_count_addresses(in6_dev) < max_addresses)
1774 ifp = ipv6_add_addr(in6_dev, &addr, pinfo->prefix_len,
1775 addr_type&IPV6_ADDR_SCOPE_MASK,
1776 addr_flags);
1778 if (!ifp || IS_ERR(ifp)) {
1779 in6_dev_put(in6_dev);
1780 return;
1783 update_lft = create = 1;
1784 ifp->cstamp = jiffies;
1785 addrconf_dad_start(ifp, RTF_ADDRCONF|RTF_PREFIX_RT);
1788 if (ifp) {
1789 int flags;
1790 unsigned long now;
1791 #ifdef CONFIG_IPV6_PRIVACY
1792 struct inet6_ifaddr *ift;
1793 #endif
1794 u32 stored_lft;
1796 /* update lifetime (RFC2462 5.5.3 e) */
1797 spin_lock(&ifp->lock);
1798 now = jiffies;
1799 if (ifp->valid_lft > (now - ifp->tstamp) / HZ)
1800 stored_lft = ifp->valid_lft - (now - ifp->tstamp) / HZ;
1801 else
1802 stored_lft = 0;
1803 if (!update_lft && stored_lft) {
1804 if (valid_lft > MIN_VALID_LIFETIME ||
1805 valid_lft > stored_lft)
1806 update_lft = 1;
1807 else if (stored_lft <= MIN_VALID_LIFETIME) {
1808 /* valid_lft <= stored_lft is always true */
1809 /* XXX: IPsec */
1810 update_lft = 0;
1811 } else {
1812 valid_lft = MIN_VALID_LIFETIME;
1813 if (valid_lft < prefered_lft)
1814 prefered_lft = valid_lft;
1815 update_lft = 1;
1819 if (update_lft) {
1820 ifp->valid_lft = valid_lft;
1821 ifp->prefered_lft = prefered_lft;
1822 ifp->tstamp = now;
1823 flags = ifp->flags;
1824 ifp->flags &= ~IFA_F_DEPRECATED;
1825 spin_unlock(&ifp->lock);
1827 if (!(flags&IFA_F_TENTATIVE))
1828 ipv6_ifa_notify(0, ifp);
1829 } else
1830 spin_unlock(&ifp->lock);
1832 #ifdef CONFIG_IPV6_PRIVACY
1833 read_lock_bh(&in6_dev->lock);
1834 /* update all temporary addresses in the list */
1835 for (ift=in6_dev->tempaddr_list; ift; ift=ift->tmp_next) {
1837 * When adjusting the lifetimes of an existing
1838 * temporary address, only lower the lifetimes.
1839 * Implementations must not increase the
1840 * lifetimes of an existing temporary address
1841 * when processing a Prefix Information Option.
1843 spin_lock(&ift->lock);
1844 flags = ift->flags;
1845 if (ift->valid_lft > valid_lft &&
1846 ift->valid_lft - valid_lft > (jiffies - ift->tstamp) / HZ)
1847 ift->valid_lft = valid_lft + (jiffies - ift->tstamp) / HZ;
1848 if (ift->prefered_lft > prefered_lft &&
1849 ift->prefered_lft - prefered_lft > (jiffies - ift->tstamp) / HZ)
1850 ift->prefered_lft = prefered_lft + (jiffies - ift->tstamp) / HZ;
1851 spin_unlock(&ift->lock);
1852 if (!(flags&IFA_F_TENTATIVE))
1853 ipv6_ifa_notify(0, ift);
1856 if (create && in6_dev->cnf.use_tempaddr > 0) {
1858 * When a new public address is created as described in [ADDRCONF],
1859 * also create a new temporary address.
1861 read_unlock_bh(&in6_dev->lock);
1862 ipv6_create_tempaddr(ifp, NULL);
1863 } else {
1864 read_unlock_bh(&in6_dev->lock);
1866 #endif
1867 in6_ifa_put(ifp);
1868 addrconf_verify(0);
1871 inet6_prefix_notify(RTM_NEWPREFIX, in6_dev, pinfo);
1872 in6_dev_put(in6_dev);
1876 * Set destination address.
1877 * Special case for SIT interfaces where we create a new "virtual"
1878 * device.
1880 int addrconf_set_dstaddr(void __user *arg)
1882 struct in6_ifreq ireq;
1883 struct net_device *dev;
1884 int err = -EINVAL;
1886 rtnl_lock();
1888 err = -EFAULT;
1889 if (copy_from_user(&ireq, arg, sizeof(struct in6_ifreq)))
1890 goto err_exit;
1892 dev = __dev_get_by_index(&init_net, ireq.ifr6_ifindex);
1894 err = -ENODEV;
1895 if (dev == NULL)
1896 goto err_exit;
1898 #if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
1899 if (dev->type == ARPHRD_SIT) {
1900 struct ifreq ifr;
1901 mm_segment_t oldfs;
1902 struct ip_tunnel_parm p;
1904 err = -EADDRNOTAVAIL;
1905 if (!(ipv6_addr_type(&ireq.ifr6_addr) & IPV6_ADDR_COMPATv4))
1906 goto err_exit;
1908 memset(&p, 0, sizeof(p));
1909 p.iph.daddr = ireq.ifr6_addr.s6_addr32[3];
1910 p.iph.saddr = 0;
1911 p.iph.version = 4;
1912 p.iph.ihl = 5;
1913 p.iph.protocol = IPPROTO_IPV6;
1914 p.iph.ttl = 64;
1915 ifr.ifr_ifru.ifru_data = (__force void __user *)&p;
1917 oldfs = get_fs(); set_fs(KERNEL_DS);
1918 err = dev->do_ioctl(dev, &ifr, SIOCADDTUNNEL);
1919 set_fs(oldfs);
1921 if (err == 0) {
1922 err = -ENOBUFS;
1923 if ((dev = __dev_get_by_name(&init_net, p.name)) == NULL)
1924 goto err_exit;
1925 err = dev_open(dev);
1928 #endif
1930 err_exit:
1931 rtnl_unlock();
1932 return err;
1936 * Manual configuration of address on an interface
1938 static int inet6_addr_add(int ifindex, struct in6_addr *pfx, int plen,
1939 __u8 ifa_flags, __u32 prefered_lft, __u32 valid_lft)
1941 struct inet6_ifaddr *ifp;
1942 struct inet6_dev *idev;
1943 struct net_device *dev;
1944 int scope;
1945 u32 flags = RTF_EXPIRES;
1947 ASSERT_RTNL();
1949 /* check the lifetime */
1950 if (!valid_lft || prefered_lft > valid_lft)
1951 return -EINVAL;
1953 if ((dev = __dev_get_by_index(&init_net, ifindex)) == NULL)
1954 return -ENODEV;
1956 if ((idev = addrconf_add_dev(dev)) == NULL)
1957 return -ENOBUFS;
1959 scope = ipv6_addr_scope(pfx);
1961 if (valid_lft == INFINITY_LIFE_TIME) {
1962 ifa_flags |= IFA_F_PERMANENT;
1963 flags = 0;
1964 } else if (valid_lft >= 0x7FFFFFFF/HZ)
1965 valid_lft = 0x7FFFFFFF/HZ;
1967 if (prefered_lft == 0)
1968 ifa_flags |= IFA_F_DEPRECATED;
1969 else if ((prefered_lft >= 0x7FFFFFFF/HZ) &&
1970 (prefered_lft != INFINITY_LIFE_TIME))
1971 prefered_lft = 0x7FFFFFFF/HZ;
1973 ifp = ipv6_add_addr(idev, pfx, plen, scope, ifa_flags);
1975 if (!IS_ERR(ifp)) {
1976 spin_lock_bh(&ifp->lock);
1977 ifp->valid_lft = valid_lft;
1978 ifp->prefered_lft = prefered_lft;
1979 ifp->tstamp = jiffies;
1980 spin_unlock_bh(&ifp->lock);
1982 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, dev,
1983 jiffies_to_clock_t(valid_lft * HZ), flags);
1985 * Note that section 3.1 of RFC 4429 indicates
1986 * that the Optimistic flag should not be set for
1987 * manually configured addresses
1989 addrconf_dad_start(ifp, 0);
1990 in6_ifa_put(ifp);
1991 addrconf_verify(0);
1992 return 0;
1995 return PTR_ERR(ifp);
1998 static int inet6_addr_del(int ifindex, struct in6_addr *pfx, int plen)
2000 struct inet6_ifaddr *ifp;
2001 struct inet6_dev *idev;
2002 struct net_device *dev;
2004 if ((dev = __dev_get_by_index(&init_net, ifindex)) == NULL)
2005 return -ENODEV;
2007 if ((idev = __in6_dev_get(dev)) == NULL)
2008 return -ENXIO;
2010 read_lock_bh(&idev->lock);
2011 for (ifp = idev->addr_list; ifp; ifp=ifp->if_next) {
2012 if (ifp->prefix_len == plen &&
2013 ipv6_addr_equal(pfx, &ifp->addr)) {
2014 in6_ifa_hold(ifp);
2015 read_unlock_bh(&idev->lock);
2017 ipv6_del_addr(ifp);
2019 /* If the last address is deleted administratively,
2020 disable IPv6 on this interface.
2022 if (idev->addr_list == NULL)
2023 addrconf_ifdown(idev->dev, 1);
2024 return 0;
2027 read_unlock_bh(&idev->lock);
2028 return -EADDRNOTAVAIL;
2032 int addrconf_add_ifaddr(void __user *arg)
2034 struct in6_ifreq ireq;
2035 int err;
2037 if (!capable(CAP_NET_ADMIN))
2038 return -EPERM;
2040 if (copy_from_user(&ireq, arg, sizeof(struct in6_ifreq)))
2041 return -EFAULT;
2043 rtnl_lock();
2044 err = inet6_addr_add(ireq.ifr6_ifindex, &ireq.ifr6_addr, ireq.ifr6_prefixlen,
2045 IFA_F_PERMANENT, INFINITY_LIFE_TIME, INFINITY_LIFE_TIME);
2046 rtnl_unlock();
2047 return err;
2050 int addrconf_del_ifaddr(void __user *arg)
2052 struct in6_ifreq ireq;
2053 int err;
2055 if (!capable(CAP_NET_ADMIN))
2056 return -EPERM;
2058 if (copy_from_user(&ireq, arg, sizeof(struct in6_ifreq)))
2059 return -EFAULT;
2061 rtnl_lock();
2062 err = inet6_addr_del(ireq.ifr6_ifindex, &ireq.ifr6_addr, ireq.ifr6_prefixlen);
2063 rtnl_unlock();
2064 return err;
2067 #if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
2068 static void sit_add_v4_addrs(struct inet6_dev *idev)
2070 struct inet6_ifaddr * ifp;
2071 struct in6_addr addr;
2072 struct net_device *dev;
2073 int scope;
2075 ASSERT_RTNL();
2077 memset(&addr, 0, sizeof(struct in6_addr));
2078 memcpy(&addr.s6_addr32[3], idev->dev->dev_addr, 4);
2080 if (idev->dev->flags&IFF_POINTOPOINT) {
2081 addr.s6_addr32[0] = htonl(0xfe800000);
2082 scope = IFA_LINK;
2083 } else {
2084 scope = IPV6_ADDR_COMPATv4;
2087 if (addr.s6_addr32[3]) {
2088 ifp = ipv6_add_addr(idev, &addr, 128, scope, IFA_F_PERMANENT);
2089 if (!IS_ERR(ifp)) {
2090 spin_lock_bh(&ifp->lock);
2091 ifp->flags &= ~IFA_F_TENTATIVE;
2092 spin_unlock_bh(&ifp->lock);
2093 ipv6_ifa_notify(RTM_NEWADDR, ifp);
2094 in6_ifa_put(ifp);
2096 return;
2099 for_each_netdev(&init_net, dev) {
2100 struct in_device * in_dev = __in_dev_get_rtnl(dev);
2101 if (in_dev && (dev->flags & IFF_UP)) {
2102 struct in_ifaddr * ifa;
2104 int flag = scope;
2106 for (ifa = in_dev->ifa_list; ifa; ifa = ifa->ifa_next) {
2107 int plen;
2109 addr.s6_addr32[3] = ifa->ifa_local;
2111 if (ifa->ifa_scope == RT_SCOPE_LINK)
2112 continue;
2113 if (ifa->ifa_scope >= RT_SCOPE_HOST) {
2114 if (idev->dev->flags&IFF_POINTOPOINT)
2115 continue;
2116 flag |= IFA_HOST;
2118 if (idev->dev->flags&IFF_POINTOPOINT)
2119 plen = 64;
2120 else
2121 plen = 96;
2123 ifp = ipv6_add_addr(idev, &addr, plen, flag,
2124 IFA_F_PERMANENT);
2125 if (!IS_ERR(ifp)) {
2126 spin_lock_bh(&ifp->lock);
2127 ifp->flags &= ~IFA_F_TENTATIVE;
2128 spin_unlock_bh(&ifp->lock);
2129 ipv6_ifa_notify(RTM_NEWADDR, ifp);
2130 in6_ifa_put(ifp);
2136 #endif
2138 static void init_loopback(struct net_device *dev)
2140 struct inet6_dev *idev;
2141 struct inet6_ifaddr * ifp;
2143 /* ::1 */
2145 ASSERT_RTNL();
2147 if ((idev = ipv6_find_idev(dev)) == NULL) {
2148 printk(KERN_DEBUG "init loopback: add_dev failed\n");
2149 return;
2152 ifp = ipv6_add_addr(idev, &in6addr_loopback, 128, IFA_HOST, IFA_F_PERMANENT);
2153 if (!IS_ERR(ifp)) {
2154 spin_lock_bh(&ifp->lock);
2155 ifp->flags &= ~IFA_F_TENTATIVE;
2156 spin_unlock_bh(&ifp->lock);
2157 ipv6_ifa_notify(RTM_NEWADDR, ifp);
2158 in6_ifa_put(ifp);
2162 static void addrconf_add_linklocal(struct inet6_dev *idev, struct in6_addr *addr)
2164 struct inet6_ifaddr * ifp;
2165 u32 addr_flags = IFA_F_PERMANENT;
2167 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
2168 if (idev->cnf.optimistic_dad &&
2169 !ipv6_devconf.forwarding)
2170 addr_flags |= IFA_F_OPTIMISTIC;
2171 #endif
2174 ifp = ipv6_add_addr(idev, addr, 64, IFA_LINK, addr_flags);
2175 if (!IS_ERR(ifp)) {
2176 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, idev->dev, 0, 0);
2177 addrconf_dad_start(ifp, 0);
2178 in6_ifa_put(ifp);
2182 static void addrconf_dev_config(struct net_device *dev)
2184 struct in6_addr addr;
2185 struct inet6_dev * idev;
2187 ASSERT_RTNL();
2189 if ((dev->type != ARPHRD_ETHER) &&
2190 (dev->type != ARPHRD_FDDI) &&
2191 (dev->type != ARPHRD_IEEE802_TR) &&
2192 (dev->type != ARPHRD_ARCNET) &&
2193 (dev->type != ARPHRD_INFINIBAND)) {
2194 /* Alas, we support only Ethernet autoconfiguration. */
2195 return;
2198 idev = addrconf_add_dev(dev);
2199 if (idev == NULL)
2200 return;
2202 memset(&addr, 0, sizeof(struct in6_addr));
2203 addr.s6_addr32[0] = htonl(0xFE800000);
2205 if (ipv6_generate_eui64(addr.s6_addr + 8, dev) == 0)
2206 addrconf_add_linklocal(idev, &addr);
2209 #if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
2210 static void addrconf_sit_config(struct net_device *dev)
2212 struct inet6_dev *idev;
2214 ASSERT_RTNL();
2217 * Configure the tunnel with one of our IPv4
2218 * addresses... we should configure all of
2219 * our v4 addrs in the tunnel
2222 if ((idev = ipv6_find_idev(dev)) == NULL) {
2223 printk(KERN_DEBUG "init sit: add_dev failed\n");
2224 return;
2227 if (dev->priv_flags & IFF_ISATAP) {
2228 struct in6_addr addr;
2230 ipv6_addr_set(&addr, htonl(0xFE800000), 0, 0, 0);
2231 addrconf_prefix_route(&addr, 64, dev, 0, 0);
2232 if (!ipv6_generate_eui64(addr.s6_addr + 8, dev))
2233 addrconf_add_linklocal(idev, &addr);
2234 return;
2237 sit_add_v4_addrs(idev);
2239 if (dev->flags&IFF_POINTOPOINT) {
2240 addrconf_add_mroute(dev);
2241 addrconf_add_lroute(dev);
2242 } else
2243 sit_route_add(dev);
2245 #endif
2247 static inline int
2248 ipv6_inherit_linklocal(struct inet6_dev *idev, struct net_device *link_dev)
2250 struct in6_addr lladdr;
2252 if (!ipv6_get_lladdr(link_dev, &lladdr, IFA_F_TENTATIVE)) {
2253 addrconf_add_linklocal(idev, &lladdr);
2254 return 0;
2256 return -1;
2259 static void ip6_tnl_add_linklocal(struct inet6_dev *idev)
2261 struct net_device *link_dev;
2263 /* first try to inherit the link-local address from the link device */
2264 if (idev->dev->iflink &&
2265 (link_dev = __dev_get_by_index(&init_net, idev->dev->iflink))) {
2266 if (!ipv6_inherit_linklocal(idev, link_dev))
2267 return;
2269 /* then try to inherit it from any device */
2270 for_each_netdev(&init_net, link_dev) {
2271 if (!ipv6_inherit_linklocal(idev, link_dev))
2272 return;
2274 printk(KERN_DEBUG "init ip6-ip6: add_linklocal failed\n");
2278 * Autoconfigure tunnel with a link-local address so routing protocols,
2279 * DHCPv6, MLD etc. can be run over the virtual link
2282 static void addrconf_ip6_tnl_config(struct net_device *dev)
2284 struct inet6_dev *idev;
2286 ASSERT_RTNL();
2288 if ((idev = addrconf_add_dev(dev)) == NULL) {
2289 printk(KERN_DEBUG "init ip6-ip6: add_dev failed\n");
2290 return;
2292 ip6_tnl_add_linklocal(idev);
2295 static int addrconf_notify(struct notifier_block *this, unsigned long event,
2296 void * data)
2298 struct net_device *dev = (struct net_device *) data;
2299 struct inet6_dev *idev = __in6_dev_get(dev);
2300 int run_pending = 0;
2301 int err;
2303 if (dev->nd_net != &init_net)
2304 return NOTIFY_DONE;
2306 switch(event) {
2307 case NETDEV_REGISTER:
2308 if (!idev && dev->mtu >= IPV6_MIN_MTU) {
2309 idev = ipv6_add_dev(dev);
2310 if (!idev)
2311 return notifier_from_errno(-ENOMEM);
2313 break;
2314 case NETDEV_UP:
2315 case NETDEV_CHANGE:
2316 if (dev->flags & IFF_SLAVE)
2317 break;
2319 if (event == NETDEV_UP) {
2320 if (!addrconf_qdisc_ok(dev)) {
2321 /* device is not ready yet. */
2322 printk(KERN_INFO
2323 "ADDRCONF(NETDEV_UP): %s: "
2324 "link is not ready\n",
2325 dev->name);
2326 break;
2329 if (!idev && dev->mtu >= IPV6_MIN_MTU)
2330 idev = ipv6_add_dev(dev);
2332 if (idev)
2333 idev->if_flags |= IF_READY;
2334 } else {
2335 if (!addrconf_qdisc_ok(dev)) {
2336 /* device is still not ready. */
2337 break;
2340 if (idev) {
2341 if (idev->if_flags & IF_READY) {
2342 /* device is already configured. */
2343 break;
2345 idev->if_flags |= IF_READY;
2348 printk(KERN_INFO
2349 "ADDRCONF(NETDEV_CHANGE): %s: "
2350 "link becomes ready\n",
2351 dev->name);
2353 run_pending = 1;
2356 switch(dev->type) {
2357 #if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
2358 case ARPHRD_SIT:
2359 addrconf_sit_config(dev);
2360 break;
2361 #endif
2362 case ARPHRD_TUNNEL6:
2363 addrconf_ip6_tnl_config(dev);
2364 break;
2365 case ARPHRD_LOOPBACK:
2366 init_loopback(dev);
2367 break;
2369 default:
2370 addrconf_dev_config(dev);
2371 break;
2373 if (idev) {
2374 if (run_pending)
2375 addrconf_dad_run(idev);
2377 /* If the MTU changed during the interface down, when the
2378 interface up, the changed MTU must be reflected in the
2379 idev as well as routers.
2381 if (idev->cnf.mtu6 != dev->mtu && dev->mtu >= IPV6_MIN_MTU) {
2382 rt6_mtu_change(dev, dev->mtu);
2383 idev->cnf.mtu6 = dev->mtu;
2385 idev->tstamp = jiffies;
2386 inet6_ifinfo_notify(RTM_NEWLINK, idev);
2387 /* If the changed mtu during down is lower than IPV6_MIN_MTU
2388 stop IPv6 on this interface.
2390 if (dev->mtu < IPV6_MIN_MTU)
2391 addrconf_ifdown(dev, event != NETDEV_DOWN);
2393 break;
2395 case NETDEV_CHANGEMTU:
2396 if (idev && dev->mtu >= IPV6_MIN_MTU) {
2397 rt6_mtu_change(dev, dev->mtu);
2398 idev->cnf.mtu6 = dev->mtu;
2399 break;
2402 if (!idev && dev->mtu >= IPV6_MIN_MTU) {
2403 idev = ipv6_add_dev(dev);
2404 if (idev)
2405 break;
2408 /* MTU falled under IPV6_MIN_MTU. Stop IPv6 on this interface. */
2410 case NETDEV_DOWN:
2411 case NETDEV_UNREGISTER:
2413 * Remove all addresses from this interface.
2415 addrconf_ifdown(dev, event != NETDEV_DOWN);
2416 break;
2418 case NETDEV_CHANGENAME:
2419 if (idev) {
2420 snmp6_unregister_dev(idev);
2421 addrconf_sysctl_unregister(idev);
2422 addrconf_sysctl_register(idev);
2423 err = snmp6_register_dev(idev);
2424 if (err)
2425 return notifier_from_errno(err);
2427 break;
2430 return NOTIFY_OK;
2434 * addrconf module should be notified of a device going up
2436 static struct notifier_block ipv6_dev_notf = {
2437 .notifier_call = addrconf_notify,
2438 .priority = 0
2441 static int addrconf_ifdown(struct net_device *dev, int how)
2443 struct inet6_dev *idev;
2444 struct inet6_ifaddr *ifa, **bifa;
2445 int i;
2447 ASSERT_RTNL();
2449 if (dev == init_net.loopback_dev && how == 1)
2450 how = 0;
2452 rt6_ifdown(dev);
2453 neigh_ifdown(&nd_tbl, dev);
2455 idev = __in6_dev_get(dev);
2456 if (idev == NULL)
2457 return -ENODEV;
2459 /* Step 1: remove reference to ipv6 device from parent device.
2460 Do not dev_put!
2462 if (how == 1) {
2463 idev->dead = 1;
2465 /* protected by rtnl_lock */
2466 rcu_assign_pointer(dev->ip6_ptr, NULL);
2468 /* Step 1.5: remove snmp6 entry */
2469 snmp6_unregister_dev(idev);
2473 /* Step 2: clear hash table */
2474 for (i=0; i<IN6_ADDR_HSIZE; i++) {
2475 bifa = &inet6_addr_lst[i];
2477 write_lock_bh(&addrconf_hash_lock);
2478 while ((ifa = *bifa) != NULL) {
2479 if (ifa->idev == idev) {
2480 *bifa = ifa->lst_next;
2481 ifa->lst_next = NULL;
2482 addrconf_del_timer(ifa);
2483 in6_ifa_put(ifa);
2484 continue;
2486 bifa = &ifa->lst_next;
2488 write_unlock_bh(&addrconf_hash_lock);
2491 write_lock_bh(&idev->lock);
2493 /* Step 3: clear flags for stateless addrconf */
2494 if (how != 1)
2495 idev->if_flags &= ~(IF_RS_SENT|IF_RA_RCVD|IF_READY);
2497 /* Step 4: clear address list */
2498 #ifdef CONFIG_IPV6_PRIVACY
2499 if (how == 1 && del_timer(&idev->regen_timer))
2500 in6_dev_put(idev);
2502 /* clear tempaddr list */
2503 while ((ifa = idev->tempaddr_list) != NULL) {
2504 idev->tempaddr_list = ifa->tmp_next;
2505 ifa->tmp_next = NULL;
2506 ifa->dead = 1;
2507 write_unlock_bh(&idev->lock);
2508 spin_lock_bh(&ifa->lock);
2510 if (ifa->ifpub) {
2511 in6_ifa_put(ifa->ifpub);
2512 ifa->ifpub = NULL;
2514 spin_unlock_bh(&ifa->lock);
2515 in6_ifa_put(ifa);
2516 write_lock_bh(&idev->lock);
2518 #endif
2519 while ((ifa = idev->addr_list) != NULL) {
2520 idev->addr_list = ifa->if_next;
2521 ifa->if_next = NULL;
2522 ifa->dead = 1;
2523 addrconf_del_timer(ifa);
2524 write_unlock_bh(&idev->lock);
2526 __ipv6_ifa_notify(RTM_DELADDR, ifa);
2527 atomic_notifier_call_chain(&inet6addr_chain, NETDEV_DOWN, ifa);
2528 in6_ifa_put(ifa);
2530 write_lock_bh(&idev->lock);
2532 write_unlock_bh(&idev->lock);
2534 /* Step 5: Discard multicast list */
2536 if (how == 1)
2537 ipv6_mc_destroy_dev(idev);
2538 else
2539 ipv6_mc_down(idev);
2541 idev->tstamp = jiffies;
2543 /* Shot the device (if unregistered) */
2545 if (how == 1) {
2546 addrconf_sysctl_unregister(idev);
2547 neigh_parms_release(&nd_tbl, idev->nd_parms);
2548 neigh_ifdown(&nd_tbl, dev);
2549 in6_dev_put(idev);
2551 return 0;
2554 static void addrconf_rs_timer(unsigned long data)
2556 struct inet6_ifaddr *ifp = (struct inet6_ifaddr *) data;
2558 if (ifp->idev->cnf.forwarding)
2559 goto out;
2561 if (ifp->idev->if_flags & IF_RA_RCVD) {
2563 * Announcement received after solicitation
2564 * was sent
2566 goto out;
2569 spin_lock(&ifp->lock);
2570 if (ifp->probes++ < ifp->idev->cnf.rtr_solicits) {
2571 struct in6_addr all_routers;
2573 /* The wait after the last probe can be shorter */
2574 addrconf_mod_timer(ifp, AC_RS,
2575 (ifp->probes == ifp->idev->cnf.rtr_solicits) ?
2576 ifp->idev->cnf.rtr_solicit_delay :
2577 ifp->idev->cnf.rtr_solicit_interval);
2578 spin_unlock(&ifp->lock);
2580 ipv6_addr_all_routers(&all_routers);
2582 ndisc_send_rs(ifp->idev->dev, &ifp->addr, &all_routers);
2583 } else {
2584 spin_unlock(&ifp->lock);
2586 * Note: we do not support deprecated "all on-link"
2587 * assumption any longer.
2589 printk(KERN_DEBUG "%s: no IPv6 routers present\n",
2590 ifp->idev->dev->name);
2593 out:
2594 in6_ifa_put(ifp);
2598 * Duplicate Address Detection
2600 static void addrconf_dad_kick(struct inet6_ifaddr *ifp)
2602 unsigned long rand_num;
2603 struct inet6_dev *idev = ifp->idev;
2605 if (ifp->flags & IFA_F_OPTIMISTIC)
2606 rand_num = 0;
2607 else
2608 rand_num = net_random() % (idev->cnf.rtr_solicit_delay ? : 1);
2610 ifp->probes = idev->cnf.dad_transmits;
2611 addrconf_mod_timer(ifp, AC_DAD, rand_num);
2614 static void addrconf_dad_start(struct inet6_ifaddr *ifp, u32 flags)
2616 struct inet6_dev *idev = ifp->idev;
2617 struct net_device *dev = idev->dev;
2619 addrconf_join_solict(dev, &ifp->addr);
2621 net_srandom(ifp->addr.s6_addr32[3]);
2623 read_lock_bh(&idev->lock);
2624 if (ifp->dead)
2625 goto out;
2626 spin_lock_bh(&ifp->lock);
2628 if (dev->flags&(IFF_NOARP|IFF_LOOPBACK) ||
2629 !(ifp->flags&IFA_F_TENTATIVE) ||
2630 ifp->flags & IFA_F_NODAD) {
2631 ifp->flags &= ~(IFA_F_TENTATIVE|IFA_F_OPTIMISTIC);
2632 spin_unlock_bh(&ifp->lock);
2633 read_unlock_bh(&idev->lock);
2635 addrconf_dad_completed(ifp);
2636 return;
2639 if (!(idev->if_flags & IF_READY)) {
2640 spin_unlock_bh(&ifp->lock);
2641 read_unlock_bh(&idev->lock);
2643 * If the defice is not ready:
2644 * - keep it tentative if it is a permanent address.
2645 * - otherwise, kill it.
2647 in6_ifa_hold(ifp);
2648 addrconf_dad_stop(ifp);
2649 return;
2653 * Optimistic nodes can start receiving
2654 * Frames right away
2656 if(ifp->flags & IFA_F_OPTIMISTIC)
2657 ip6_ins_rt(ifp->rt);
2659 addrconf_dad_kick(ifp);
2660 spin_unlock_bh(&ifp->lock);
2661 out:
2662 read_unlock_bh(&idev->lock);
2665 static void addrconf_dad_timer(unsigned long data)
2667 struct inet6_ifaddr *ifp = (struct inet6_ifaddr *) data;
2668 struct inet6_dev *idev = ifp->idev;
2669 struct in6_addr unspec;
2670 struct in6_addr mcaddr;
2672 read_lock_bh(&idev->lock);
2673 if (idev->dead) {
2674 read_unlock_bh(&idev->lock);
2675 goto out;
2677 spin_lock_bh(&ifp->lock);
2678 if (ifp->probes == 0) {
2680 * DAD was successful
2683 ifp->flags &= ~(IFA_F_TENTATIVE|IFA_F_OPTIMISTIC);
2684 spin_unlock_bh(&ifp->lock);
2685 read_unlock_bh(&idev->lock);
2687 addrconf_dad_completed(ifp);
2689 goto out;
2692 ifp->probes--;
2693 addrconf_mod_timer(ifp, AC_DAD, ifp->idev->nd_parms->retrans_time);
2694 spin_unlock_bh(&ifp->lock);
2695 read_unlock_bh(&idev->lock);
2697 /* send a neighbour solicitation for our addr */
2698 memset(&unspec, 0, sizeof(unspec));
2699 addrconf_addr_solict_mult(&ifp->addr, &mcaddr);
2700 ndisc_send_ns(ifp->idev->dev, NULL, &ifp->addr, &mcaddr, &unspec);
2701 out:
2702 in6_ifa_put(ifp);
2705 static void addrconf_dad_completed(struct inet6_ifaddr *ifp)
2707 struct net_device * dev = ifp->idev->dev;
2710 * Configure the address for reception. Now it is valid.
2713 ipv6_ifa_notify(RTM_NEWADDR, ifp);
2715 /* If added prefix is link local and forwarding is off,
2716 start sending router solicitations.
2719 if (ifp->idev->cnf.forwarding == 0 &&
2720 ifp->idev->cnf.rtr_solicits > 0 &&
2721 (dev->flags&IFF_LOOPBACK) == 0 &&
2722 (ipv6_addr_type(&ifp->addr) & IPV6_ADDR_LINKLOCAL)) {
2723 struct in6_addr all_routers;
2725 ipv6_addr_all_routers(&all_routers);
2728 * If a host as already performed a random delay
2729 * [...] as part of DAD [...] there is no need
2730 * to delay again before sending the first RS
2732 ndisc_send_rs(ifp->idev->dev, &ifp->addr, &all_routers);
2734 spin_lock_bh(&ifp->lock);
2735 ifp->probes = 1;
2736 ifp->idev->if_flags |= IF_RS_SENT;
2737 addrconf_mod_timer(ifp, AC_RS, ifp->idev->cnf.rtr_solicit_interval);
2738 spin_unlock_bh(&ifp->lock);
2742 static void addrconf_dad_run(struct inet6_dev *idev) {
2743 struct inet6_ifaddr *ifp;
2745 read_lock_bh(&idev->lock);
2746 for (ifp = idev->addr_list; ifp; ifp = ifp->if_next) {
2747 spin_lock_bh(&ifp->lock);
2748 if (!(ifp->flags & IFA_F_TENTATIVE)) {
2749 spin_unlock_bh(&ifp->lock);
2750 continue;
2752 spin_unlock_bh(&ifp->lock);
2753 addrconf_dad_kick(ifp);
2755 read_unlock_bh(&idev->lock);
2758 #ifdef CONFIG_PROC_FS
2759 struct if6_iter_state {
2760 struct seq_net_private p;
2761 int bucket;
2764 static struct inet6_ifaddr *if6_get_first(struct seq_file *seq)
2766 struct inet6_ifaddr *ifa = NULL;
2767 struct if6_iter_state *state = seq->private;
2768 struct net *net = state->p.net;
2770 for (state->bucket = 0; state->bucket < IN6_ADDR_HSIZE; ++state->bucket) {
2771 ifa = inet6_addr_lst[state->bucket];
2773 while (ifa && ifa->idev->dev->nd_net != net)
2774 ifa = ifa->lst_next;
2775 if (ifa)
2776 break;
2778 return ifa;
2781 static struct inet6_ifaddr *if6_get_next(struct seq_file *seq, struct inet6_ifaddr *ifa)
2783 struct if6_iter_state *state = seq->private;
2784 struct net *net = state->p.net;
2786 ifa = ifa->lst_next;
2787 try_again:
2788 if (ifa) {
2789 if (ifa->idev->dev->nd_net != net) {
2790 ifa = ifa->lst_next;
2791 goto try_again;
2795 if (!ifa && ++state->bucket < IN6_ADDR_HSIZE) {
2796 ifa = inet6_addr_lst[state->bucket];
2797 goto try_again;
2800 return ifa;
2803 static struct inet6_ifaddr *if6_get_idx(struct seq_file *seq, loff_t pos)
2805 struct inet6_ifaddr *ifa = if6_get_first(seq);
2807 if (ifa)
2808 while(pos && (ifa = if6_get_next(seq, ifa)) != NULL)
2809 --pos;
2810 return pos ? NULL : ifa;
2813 static void *if6_seq_start(struct seq_file *seq, loff_t *pos)
2814 __acquires(addrconf_hash_lock)
2816 read_lock_bh(&addrconf_hash_lock);
2817 return if6_get_idx(seq, *pos);
2820 static void *if6_seq_next(struct seq_file *seq, void *v, loff_t *pos)
2822 struct inet6_ifaddr *ifa;
2824 ifa = if6_get_next(seq, v);
2825 ++*pos;
2826 return ifa;
2829 static void if6_seq_stop(struct seq_file *seq, void *v)
2830 __releases(addrconf_hash_lock)
2832 read_unlock_bh(&addrconf_hash_lock);
2835 static int if6_seq_show(struct seq_file *seq, void *v)
2837 struct inet6_ifaddr *ifp = (struct inet6_ifaddr *)v;
2838 seq_printf(seq,
2839 NIP6_SEQFMT " %02x %02x %02x %02x %8s\n",
2840 NIP6(ifp->addr),
2841 ifp->idev->dev->ifindex,
2842 ifp->prefix_len,
2843 ifp->scope,
2844 ifp->flags,
2845 ifp->idev->dev->name);
2846 return 0;
2849 static const struct seq_operations if6_seq_ops = {
2850 .start = if6_seq_start,
2851 .next = if6_seq_next,
2852 .show = if6_seq_show,
2853 .stop = if6_seq_stop,
2856 static int if6_seq_open(struct inode *inode, struct file *file)
2858 return seq_open_net(inode, file, &if6_seq_ops,
2859 sizeof(struct if6_iter_state));
2862 static const struct file_operations if6_fops = {
2863 .owner = THIS_MODULE,
2864 .open = if6_seq_open,
2865 .read = seq_read,
2866 .llseek = seq_lseek,
2867 .release = seq_release_net,
2870 static int if6_proc_net_init(struct net *net)
2872 if (!proc_net_fops_create(net, "if_inet6", S_IRUGO, &if6_fops))
2873 return -ENOMEM;
2874 return 0;
2877 static void if6_proc_net_exit(struct net *net)
2879 proc_net_remove(net, "if_inet6");
2882 static struct pernet_operations if6_proc_net_ops = {
2883 .init = if6_proc_net_init,
2884 .exit = if6_proc_net_exit,
2887 int __init if6_proc_init(void)
2889 return register_pernet_subsys(&if6_proc_net_ops);
2892 void if6_proc_exit(void)
2894 unregister_pernet_subsys(&if6_proc_net_ops);
2896 #endif /* CONFIG_PROC_FS */
2898 #if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
2899 /* Check if address is a home address configured on any interface. */
2900 int ipv6_chk_home_addr(struct net *net, struct in6_addr *addr)
2902 int ret = 0;
2903 struct inet6_ifaddr * ifp;
2904 u8 hash = ipv6_addr_hash(addr);
2905 read_lock_bh(&addrconf_hash_lock);
2906 for (ifp = inet6_addr_lst[hash]; ifp; ifp = ifp->lst_next) {
2907 if (ifp->idev->dev->nd_net != net)
2908 continue;
2909 if (ipv6_addr_cmp(&ifp->addr, addr) == 0 &&
2910 (ifp->flags & IFA_F_HOMEADDRESS)) {
2911 ret = 1;
2912 break;
2915 read_unlock_bh(&addrconf_hash_lock);
2916 return ret;
2918 #endif
2921 * Periodic address status verification
2924 static void addrconf_verify(unsigned long foo)
2926 struct inet6_ifaddr *ifp;
2927 unsigned long now, next;
2928 int i;
2930 spin_lock_bh(&addrconf_verify_lock);
2931 now = jiffies;
2932 next = now + ADDR_CHECK_FREQUENCY;
2934 del_timer(&addr_chk_timer);
2936 for (i=0; i < IN6_ADDR_HSIZE; i++) {
2938 restart:
2939 read_lock(&addrconf_hash_lock);
2940 for (ifp=inet6_addr_lst[i]; ifp; ifp=ifp->lst_next) {
2941 unsigned long age;
2942 #ifdef CONFIG_IPV6_PRIVACY
2943 unsigned long regen_advance;
2944 #endif
2946 if (ifp->flags & IFA_F_PERMANENT)
2947 continue;
2949 spin_lock(&ifp->lock);
2950 age = (now - ifp->tstamp) / HZ;
2952 #ifdef CONFIG_IPV6_PRIVACY
2953 regen_advance = ifp->idev->cnf.regen_max_retry *
2954 ifp->idev->cnf.dad_transmits *
2955 ifp->idev->nd_parms->retrans_time / HZ;
2956 #endif
2958 if (ifp->valid_lft != INFINITY_LIFE_TIME &&
2959 age >= ifp->valid_lft) {
2960 spin_unlock(&ifp->lock);
2961 in6_ifa_hold(ifp);
2962 read_unlock(&addrconf_hash_lock);
2963 ipv6_del_addr(ifp);
2964 goto restart;
2965 } else if (ifp->prefered_lft == INFINITY_LIFE_TIME) {
2966 spin_unlock(&ifp->lock);
2967 continue;
2968 } else if (age >= ifp->prefered_lft) {
2969 /* jiffies - ifp->tsamp > age >= ifp->prefered_lft */
2970 int deprecate = 0;
2972 if (!(ifp->flags&IFA_F_DEPRECATED)) {
2973 deprecate = 1;
2974 ifp->flags |= IFA_F_DEPRECATED;
2977 if (time_before(ifp->tstamp + ifp->valid_lft * HZ, next))
2978 next = ifp->tstamp + ifp->valid_lft * HZ;
2980 spin_unlock(&ifp->lock);
2982 if (deprecate) {
2983 in6_ifa_hold(ifp);
2984 read_unlock(&addrconf_hash_lock);
2986 ipv6_ifa_notify(0, ifp);
2987 in6_ifa_put(ifp);
2988 goto restart;
2990 #ifdef CONFIG_IPV6_PRIVACY
2991 } else if ((ifp->flags&IFA_F_TEMPORARY) &&
2992 !(ifp->flags&IFA_F_TENTATIVE)) {
2993 if (age >= ifp->prefered_lft - regen_advance) {
2994 struct inet6_ifaddr *ifpub = ifp->ifpub;
2995 if (time_before(ifp->tstamp + ifp->prefered_lft * HZ, next))
2996 next = ifp->tstamp + ifp->prefered_lft * HZ;
2997 if (!ifp->regen_count && ifpub) {
2998 ifp->regen_count++;
2999 in6_ifa_hold(ifp);
3000 in6_ifa_hold(ifpub);
3001 spin_unlock(&ifp->lock);
3002 read_unlock(&addrconf_hash_lock);
3003 spin_lock(&ifpub->lock);
3004 ifpub->regen_count = 0;
3005 spin_unlock(&ifpub->lock);
3006 ipv6_create_tempaddr(ifpub, ifp);
3007 in6_ifa_put(ifpub);
3008 in6_ifa_put(ifp);
3009 goto restart;
3011 } else if (time_before(ifp->tstamp + ifp->prefered_lft * HZ - regen_advance * HZ, next))
3012 next = ifp->tstamp + ifp->prefered_lft * HZ - regen_advance * HZ;
3013 spin_unlock(&ifp->lock);
3014 #endif
3015 } else {
3016 /* ifp->prefered_lft <= ifp->valid_lft */
3017 if (time_before(ifp->tstamp + ifp->prefered_lft * HZ, next))
3018 next = ifp->tstamp + ifp->prefered_lft * HZ;
3019 spin_unlock(&ifp->lock);
3022 read_unlock(&addrconf_hash_lock);
3025 addr_chk_timer.expires = time_before(next, jiffies + HZ) ? jiffies + HZ : next;
3026 add_timer(&addr_chk_timer);
3027 spin_unlock_bh(&addrconf_verify_lock);
3030 static struct in6_addr *extract_addr(struct nlattr *addr, struct nlattr *local)
3032 struct in6_addr *pfx = NULL;
3034 if (addr)
3035 pfx = nla_data(addr);
3037 if (local) {
3038 if (pfx && nla_memcmp(local, pfx, sizeof(*pfx)))
3039 pfx = NULL;
3040 else
3041 pfx = nla_data(local);
3044 return pfx;
3047 static const struct nla_policy ifa_ipv6_policy[IFA_MAX+1] = {
3048 [IFA_ADDRESS] = { .len = sizeof(struct in6_addr) },
3049 [IFA_LOCAL] = { .len = sizeof(struct in6_addr) },
3050 [IFA_CACHEINFO] = { .len = sizeof(struct ifa_cacheinfo) },
3053 static int
3054 inet6_rtm_deladdr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
3056 struct net *net = skb->sk->sk_net;
3057 struct ifaddrmsg *ifm;
3058 struct nlattr *tb[IFA_MAX+1];
3059 struct in6_addr *pfx;
3060 int err;
3062 if (net != &init_net)
3063 return -EINVAL;
3065 err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv6_policy);
3066 if (err < 0)
3067 return err;
3069 ifm = nlmsg_data(nlh);
3070 pfx = extract_addr(tb[IFA_ADDRESS], tb[IFA_LOCAL]);
3071 if (pfx == NULL)
3072 return -EINVAL;
3074 return inet6_addr_del(ifm->ifa_index, pfx, ifm->ifa_prefixlen);
3077 static int inet6_addr_modify(struct inet6_ifaddr *ifp, u8 ifa_flags,
3078 u32 prefered_lft, u32 valid_lft)
3080 u32 flags = RTF_EXPIRES;
3082 if (!valid_lft || (prefered_lft > valid_lft))
3083 return -EINVAL;
3085 if (valid_lft == INFINITY_LIFE_TIME) {
3086 ifa_flags |= IFA_F_PERMANENT;
3087 flags = 0;
3088 } else if (valid_lft >= 0x7FFFFFFF/HZ)
3089 valid_lft = 0x7FFFFFFF/HZ;
3091 if (prefered_lft == 0)
3092 ifa_flags |= IFA_F_DEPRECATED;
3093 else if ((prefered_lft >= 0x7FFFFFFF/HZ) &&
3094 (prefered_lft != INFINITY_LIFE_TIME))
3095 prefered_lft = 0x7FFFFFFF/HZ;
3097 spin_lock_bh(&ifp->lock);
3098 ifp->flags = (ifp->flags & ~(IFA_F_DEPRECATED | IFA_F_PERMANENT | IFA_F_NODAD | IFA_F_HOMEADDRESS)) | ifa_flags;
3099 ifp->tstamp = jiffies;
3100 ifp->valid_lft = valid_lft;
3101 ifp->prefered_lft = prefered_lft;
3103 spin_unlock_bh(&ifp->lock);
3104 if (!(ifp->flags&IFA_F_TENTATIVE))
3105 ipv6_ifa_notify(0, ifp);
3107 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, ifp->idev->dev,
3108 jiffies_to_clock_t(valid_lft * HZ), flags);
3109 addrconf_verify(0);
3111 return 0;
3114 static int
3115 inet6_rtm_newaddr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
3117 struct net *net = skb->sk->sk_net;
3118 struct ifaddrmsg *ifm;
3119 struct nlattr *tb[IFA_MAX+1];
3120 struct in6_addr *pfx;
3121 struct inet6_ifaddr *ifa;
3122 struct net_device *dev;
3123 u32 valid_lft = INFINITY_LIFE_TIME, preferred_lft = INFINITY_LIFE_TIME;
3124 u8 ifa_flags;
3125 int err;
3127 if (net != &init_net)
3128 return -EINVAL;
3130 err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv6_policy);
3131 if (err < 0)
3132 return err;
3134 ifm = nlmsg_data(nlh);
3135 pfx = extract_addr(tb[IFA_ADDRESS], tb[IFA_LOCAL]);
3136 if (pfx == NULL)
3137 return -EINVAL;
3139 if (tb[IFA_CACHEINFO]) {
3140 struct ifa_cacheinfo *ci;
3142 ci = nla_data(tb[IFA_CACHEINFO]);
3143 valid_lft = ci->ifa_valid;
3144 preferred_lft = ci->ifa_prefered;
3145 } else {
3146 preferred_lft = INFINITY_LIFE_TIME;
3147 valid_lft = INFINITY_LIFE_TIME;
3150 dev = __dev_get_by_index(&init_net, ifm->ifa_index);
3151 if (dev == NULL)
3152 return -ENODEV;
3154 /* We ignore other flags so far. */
3155 ifa_flags = ifm->ifa_flags & (IFA_F_NODAD | IFA_F_HOMEADDRESS);
3157 ifa = ipv6_get_ifaddr(net, pfx, dev, 1);
3158 if (ifa == NULL) {
3160 * It would be best to check for !NLM_F_CREATE here but
3161 * userspace alreay relies on not having to provide this.
3163 return inet6_addr_add(ifm->ifa_index, pfx, ifm->ifa_prefixlen,
3164 ifa_flags, preferred_lft, valid_lft);
3167 if (nlh->nlmsg_flags & NLM_F_EXCL ||
3168 !(nlh->nlmsg_flags & NLM_F_REPLACE))
3169 err = -EEXIST;
3170 else
3171 err = inet6_addr_modify(ifa, ifa_flags, preferred_lft, valid_lft);
3173 in6_ifa_put(ifa);
3175 return err;
3178 static void put_ifaddrmsg(struct nlmsghdr *nlh, u8 prefixlen, u8 flags,
3179 u8 scope, int ifindex)
3181 struct ifaddrmsg *ifm;
3183 ifm = nlmsg_data(nlh);
3184 ifm->ifa_family = AF_INET6;
3185 ifm->ifa_prefixlen = prefixlen;
3186 ifm->ifa_flags = flags;
3187 ifm->ifa_scope = scope;
3188 ifm->ifa_index = ifindex;
3191 static int put_cacheinfo(struct sk_buff *skb, unsigned long cstamp,
3192 unsigned long tstamp, u32 preferred, u32 valid)
3194 struct ifa_cacheinfo ci;
3196 ci.cstamp = (u32)(TIME_DELTA(cstamp, INITIAL_JIFFIES) / HZ * 100
3197 + TIME_DELTA(cstamp, INITIAL_JIFFIES) % HZ * 100 / HZ);
3198 ci.tstamp = (u32)(TIME_DELTA(tstamp, INITIAL_JIFFIES) / HZ * 100
3199 + TIME_DELTA(tstamp, INITIAL_JIFFIES) % HZ * 100 / HZ);
3200 ci.ifa_prefered = preferred;
3201 ci.ifa_valid = valid;
3203 return nla_put(skb, IFA_CACHEINFO, sizeof(ci), &ci);
3206 static inline int rt_scope(int ifa_scope)
3208 if (ifa_scope & IFA_HOST)
3209 return RT_SCOPE_HOST;
3210 else if (ifa_scope & IFA_LINK)
3211 return RT_SCOPE_LINK;
3212 else if (ifa_scope & IFA_SITE)
3213 return RT_SCOPE_SITE;
3214 else
3215 return RT_SCOPE_UNIVERSE;
3218 static inline int inet6_ifaddr_msgsize(void)
3220 return NLMSG_ALIGN(sizeof(struct ifaddrmsg))
3221 + nla_total_size(16) /* IFA_ADDRESS */
3222 + nla_total_size(sizeof(struct ifa_cacheinfo));
3225 static int inet6_fill_ifaddr(struct sk_buff *skb, struct inet6_ifaddr *ifa,
3226 u32 pid, u32 seq, int event, unsigned int flags)
3228 struct nlmsghdr *nlh;
3229 u32 preferred, valid;
3231 nlh = nlmsg_put(skb, pid, seq, event, sizeof(struct ifaddrmsg), flags);
3232 if (nlh == NULL)
3233 return -EMSGSIZE;
3235 put_ifaddrmsg(nlh, ifa->prefix_len, ifa->flags, rt_scope(ifa->scope),
3236 ifa->idev->dev->ifindex);
3238 if (!(ifa->flags&IFA_F_PERMANENT)) {
3239 preferred = ifa->prefered_lft;
3240 valid = ifa->valid_lft;
3241 if (preferred != INFINITY_LIFE_TIME) {
3242 long tval = (jiffies - ifa->tstamp)/HZ;
3243 preferred -= tval;
3244 if (valid != INFINITY_LIFE_TIME)
3245 valid -= tval;
3247 } else {
3248 preferred = INFINITY_LIFE_TIME;
3249 valid = INFINITY_LIFE_TIME;
3252 if (nla_put(skb, IFA_ADDRESS, 16, &ifa->addr) < 0 ||
3253 put_cacheinfo(skb, ifa->cstamp, ifa->tstamp, preferred, valid) < 0) {
3254 nlmsg_cancel(skb, nlh);
3255 return -EMSGSIZE;
3258 return nlmsg_end(skb, nlh);
3261 static int inet6_fill_ifmcaddr(struct sk_buff *skb, struct ifmcaddr6 *ifmca,
3262 u32 pid, u32 seq, int event, u16 flags)
3264 struct nlmsghdr *nlh;
3265 u8 scope = RT_SCOPE_UNIVERSE;
3266 int ifindex = ifmca->idev->dev->ifindex;
3268 if (ipv6_addr_scope(&ifmca->mca_addr) & IFA_SITE)
3269 scope = RT_SCOPE_SITE;
3271 nlh = nlmsg_put(skb, pid, seq, event, sizeof(struct ifaddrmsg), flags);
3272 if (nlh == NULL)
3273 return -EMSGSIZE;
3275 put_ifaddrmsg(nlh, 128, IFA_F_PERMANENT, scope, ifindex);
3276 if (nla_put(skb, IFA_MULTICAST, 16, &ifmca->mca_addr) < 0 ||
3277 put_cacheinfo(skb, ifmca->mca_cstamp, ifmca->mca_tstamp,
3278 INFINITY_LIFE_TIME, INFINITY_LIFE_TIME) < 0) {
3279 nlmsg_cancel(skb, nlh);
3280 return -EMSGSIZE;
3283 return nlmsg_end(skb, nlh);
3286 static int inet6_fill_ifacaddr(struct sk_buff *skb, struct ifacaddr6 *ifaca,
3287 u32 pid, u32 seq, int event, unsigned int flags)
3289 struct nlmsghdr *nlh;
3290 u8 scope = RT_SCOPE_UNIVERSE;
3291 int ifindex = ifaca->aca_idev->dev->ifindex;
3293 if (ipv6_addr_scope(&ifaca->aca_addr) & IFA_SITE)
3294 scope = RT_SCOPE_SITE;
3296 nlh = nlmsg_put(skb, pid, seq, event, sizeof(struct ifaddrmsg), flags);
3297 if (nlh == NULL)
3298 return -EMSGSIZE;
3300 put_ifaddrmsg(nlh, 128, IFA_F_PERMANENT, scope, ifindex);
3301 if (nla_put(skb, IFA_ANYCAST, 16, &ifaca->aca_addr) < 0 ||
3302 put_cacheinfo(skb, ifaca->aca_cstamp, ifaca->aca_tstamp,
3303 INFINITY_LIFE_TIME, INFINITY_LIFE_TIME) < 0) {
3304 nlmsg_cancel(skb, nlh);
3305 return -EMSGSIZE;
3308 return nlmsg_end(skb, nlh);
3311 enum addr_type_t
3313 UNICAST_ADDR,
3314 MULTICAST_ADDR,
3315 ANYCAST_ADDR,
3318 static int inet6_dump_addr(struct sk_buff *skb, struct netlink_callback *cb,
3319 enum addr_type_t type)
3321 int idx, ip_idx;
3322 int s_idx, s_ip_idx;
3323 int err = 1;
3324 struct net_device *dev;
3325 struct inet6_dev *idev = NULL;
3326 struct inet6_ifaddr *ifa;
3327 struct ifmcaddr6 *ifmca;
3328 struct ifacaddr6 *ifaca;
3330 s_idx = cb->args[0];
3331 s_ip_idx = ip_idx = cb->args[1];
3333 idx = 0;
3334 for_each_netdev(&init_net, dev) {
3335 if (idx < s_idx)
3336 goto cont;
3337 if (idx > s_idx)
3338 s_ip_idx = 0;
3339 ip_idx = 0;
3340 if ((idev = in6_dev_get(dev)) == NULL)
3341 goto cont;
3342 read_lock_bh(&idev->lock);
3343 switch (type) {
3344 case UNICAST_ADDR:
3345 /* unicast address incl. temp addr */
3346 for (ifa = idev->addr_list; ifa;
3347 ifa = ifa->if_next, ip_idx++) {
3348 if (ip_idx < s_ip_idx)
3349 continue;
3350 err = inet6_fill_ifaddr(skb, ifa,
3351 NETLINK_CB(cb->skb).pid,
3352 cb->nlh->nlmsg_seq,
3353 RTM_NEWADDR,
3354 NLM_F_MULTI);
3356 break;
3357 case MULTICAST_ADDR:
3358 /* multicast address */
3359 for (ifmca = idev->mc_list; ifmca;
3360 ifmca = ifmca->next, ip_idx++) {
3361 if (ip_idx < s_ip_idx)
3362 continue;
3363 err = inet6_fill_ifmcaddr(skb, ifmca,
3364 NETLINK_CB(cb->skb).pid,
3365 cb->nlh->nlmsg_seq,
3366 RTM_GETMULTICAST,
3367 NLM_F_MULTI);
3369 break;
3370 case ANYCAST_ADDR:
3371 /* anycast address */
3372 for (ifaca = idev->ac_list; ifaca;
3373 ifaca = ifaca->aca_next, ip_idx++) {
3374 if (ip_idx < s_ip_idx)
3375 continue;
3376 err = inet6_fill_ifacaddr(skb, ifaca,
3377 NETLINK_CB(cb->skb).pid,
3378 cb->nlh->nlmsg_seq,
3379 RTM_GETANYCAST,
3380 NLM_F_MULTI);
3382 break;
3383 default:
3384 break;
3386 read_unlock_bh(&idev->lock);
3387 in6_dev_put(idev);
3389 if (err <= 0)
3390 break;
3391 cont:
3392 idx++;
3394 cb->args[0] = idx;
3395 cb->args[1] = ip_idx;
3396 return skb->len;
3399 static int inet6_dump_ifaddr(struct sk_buff *skb, struct netlink_callback *cb)
3401 struct net *net = skb->sk->sk_net;
3402 enum addr_type_t type = UNICAST_ADDR;
3404 if (net != &init_net)
3405 return 0;
3407 return inet6_dump_addr(skb, cb, type);
3410 static int inet6_dump_ifmcaddr(struct sk_buff *skb, struct netlink_callback *cb)
3412 struct net *net = skb->sk->sk_net;
3413 enum addr_type_t type = MULTICAST_ADDR;
3415 if (net != &init_net)
3416 return 0;
3418 return inet6_dump_addr(skb, cb, type);
3422 static int inet6_dump_ifacaddr(struct sk_buff *skb, struct netlink_callback *cb)
3424 struct net *net = skb->sk->sk_net;
3425 enum addr_type_t type = ANYCAST_ADDR;
3427 if (net != &init_net)
3428 return 0;
3430 return inet6_dump_addr(skb, cb, type);
3433 static int inet6_rtm_getaddr(struct sk_buff *in_skb, struct nlmsghdr* nlh,
3434 void *arg)
3436 struct net *net = in_skb->sk->sk_net;
3437 struct ifaddrmsg *ifm;
3438 struct nlattr *tb[IFA_MAX+1];
3439 struct in6_addr *addr = NULL;
3440 struct net_device *dev = NULL;
3441 struct inet6_ifaddr *ifa;
3442 struct sk_buff *skb;
3443 int err;
3445 if (net != &init_net)
3446 return -EINVAL;
3448 err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv6_policy);
3449 if (err < 0)
3450 goto errout;
3452 addr = extract_addr(tb[IFA_ADDRESS], tb[IFA_LOCAL]);
3453 if (addr == NULL) {
3454 err = -EINVAL;
3455 goto errout;
3458 ifm = nlmsg_data(nlh);
3459 if (ifm->ifa_index)
3460 dev = __dev_get_by_index(&init_net, ifm->ifa_index);
3462 if ((ifa = ipv6_get_ifaddr(net, addr, dev, 1)) == NULL) {
3463 err = -EADDRNOTAVAIL;
3464 goto errout;
3467 if ((skb = nlmsg_new(inet6_ifaddr_msgsize(), GFP_KERNEL)) == NULL) {
3468 err = -ENOBUFS;
3469 goto errout_ifa;
3472 err = inet6_fill_ifaddr(skb, ifa, NETLINK_CB(in_skb).pid,
3473 nlh->nlmsg_seq, RTM_NEWADDR, 0);
3474 if (err < 0) {
3475 /* -EMSGSIZE implies BUG in inet6_ifaddr_msgsize() */
3476 WARN_ON(err == -EMSGSIZE);
3477 kfree_skb(skb);
3478 goto errout_ifa;
3480 err = rtnl_unicast(skb, &init_net, NETLINK_CB(in_skb).pid);
3481 errout_ifa:
3482 in6_ifa_put(ifa);
3483 errout:
3484 return err;
3487 static void inet6_ifa_notify(int event, struct inet6_ifaddr *ifa)
3489 struct sk_buff *skb;
3490 int err = -ENOBUFS;
3492 skb = nlmsg_new(inet6_ifaddr_msgsize(), GFP_ATOMIC);
3493 if (skb == NULL)
3494 goto errout;
3496 err = inet6_fill_ifaddr(skb, ifa, 0, 0, event, 0);
3497 if (err < 0) {
3498 /* -EMSGSIZE implies BUG in inet6_ifaddr_msgsize() */
3499 WARN_ON(err == -EMSGSIZE);
3500 kfree_skb(skb);
3501 goto errout;
3503 err = rtnl_notify(skb, &init_net, 0, RTNLGRP_IPV6_IFADDR, NULL, GFP_ATOMIC);
3504 errout:
3505 if (err < 0)
3506 rtnl_set_sk_err(&init_net, RTNLGRP_IPV6_IFADDR, err);
3509 static inline void ipv6_store_devconf(struct ipv6_devconf *cnf,
3510 __s32 *array, int bytes)
3512 BUG_ON(bytes < (DEVCONF_MAX * 4));
3514 memset(array, 0, bytes);
3515 array[DEVCONF_FORWARDING] = cnf->forwarding;
3516 array[DEVCONF_HOPLIMIT] = cnf->hop_limit;
3517 array[DEVCONF_MTU6] = cnf->mtu6;
3518 array[DEVCONF_ACCEPT_RA] = cnf->accept_ra;
3519 array[DEVCONF_ACCEPT_REDIRECTS] = cnf->accept_redirects;
3520 array[DEVCONF_AUTOCONF] = cnf->autoconf;
3521 array[DEVCONF_DAD_TRANSMITS] = cnf->dad_transmits;
3522 array[DEVCONF_RTR_SOLICITS] = cnf->rtr_solicits;
3523 array[DEVCONF_RTR_SOLICIT_INTERVAL] = cnf->rtr_solicit_interval;
3524 array[DEVCONF_RTR_SOLICIT_DELAY] = cnf->rtr_solicit_delay;
3525 array[DEVCONF_FORCE_MLD_VERSION] = cnf->force_mld_version;
3526 #ifdef CONFIG_IPV6_PRIVACY
3527 array[DEVCONF_USE_TEMPADDR] = cnf->use_tempaddr;
3528 array[DEVCONF_TEMP_VALID_LFT] = cnf->temp_valid_lft;
3529 array[DEVCONF_TEMP_PREFERED_LFT] = cnf->temp_prefered_lft;
3530 array[DEVCONF_REGEN_MAX_RETRY] = cnf->regen_max_retry;
3531 array[DEVCONF_MAX_DESYNC_FACTOR] = cnf->max_desync_factor;
3532 #endif
3533 array[DEVCONF_MAX_ADDRESSES] = cnf->max_addresses;
3534 array[DEVCONF_ACCEPT_RA_DEFRTR] = cnf->accept_ra_defrtr;
3535 array[DEVCONF_ACCEPT_RA_PINFO] = cnf->accept_ra_pinfo;
3536 #ifdef CONFIG_IPV6_ROUTER_PREF
3537 array[DEVCONF_ACCEPT_RA_RTR_PREF] = cnf->accept_ra_rtr_pref;
3538 array[DEVCONF_RTR_PROBE_INTERVAL] = cnf->rtr_probe_interval;
3539 #ifdef CONFIG_IPV6_ROUTE_INFO
3540 array[DEVCONF_ACCEPT_RA_RT_INFO_MAX_PLEN] = cnf->accept_ra_rt_info_max_plen;
3541 #endif
3542 #endif
3543 array[DEVCONF_PROXY_NDP] = cnf->proxy_ndp;
3544 array[DEVCONF_ACCEPT_SOURCE_ROUTE] = cnf->accept_source_route;
3545 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
3546 array[DEVCONF_OPTIMISTIC_DAD] = cnf->optimistic_dad;
3547 #endif
3550 static inline size_t inet6_if_nlmsg_size(void)
3552 return NLMSG_ALIGN(sizeof(struct ifinfomsg))
3553 + nla_total_size(IFNAMSIZ) /* IFLA_IFNAME */
3554 + nla_total_size(MAX_ADDR_LEN) /* IFLA_ADDRESS */
3555 + nla_total_size(4) /* IFLA_MTU */
3556 + nla_total_size(4) /* IFLA_LINK */
3557 + nla_total_size( /* IFLA_PROTINFO */
3558 nla_total_size(4) /* IFLA_INET6_FLAGS */
3559 + nla_total_size(sizeof(struct ifla_cacheinfo))
3560 + nla_total_size(DEVCONF_MAX * 4) /* IFLA_INET6_CONF */
3561 + nla_total_size(IPSTATS_MIB_MAX * 8) /* IFLA_INET6_STATS */
3562 + nla_total_size(ICMP6_MIB_MAX * 8) /* IFLA_INET6_ICMP6STATS */
3566 static inline void __snmp6_fill_stats(u64 *stats, void **mib, int items,
3567 int bytes)
3569 int i;
3570 int pad = bytes - sizeof(u64) * items;
3571 BUG_ON(pad < 0);
3573 /* Use put_unaligned() because stats may not be aligned for u64. */
3574 put_unaligned(items, &stats[0]);
3575 for (i = 1; i < items; i++)
3576 put_unaligned(snmp_fold_field(mib, i), &stats[i]);
3578 memset(&stats[items], 0, pad);
3581 static void snmp6_fill_stats(u64 *stats, struct inet6_dev *idev, int attrtype,
3582 int bytes)
3584 switch(attrtype) {
3585 case IFLA_INET6_STATS:
3586 __snmp6_fill_stats(stats, (void **)idev->stats.ipv6, IPSTATS_MIB_MAX, bytes);
3587 break;
3588 case IFLA_INET6_ICMP6STATS:
3589 __snmp6_fill_stats(stats, (void **)idev->stats.icmpv6, ICMP6_MIB_MAX, bytes);
3590 break;
3594 static int inet6_fill_ifinfo(struct sk_buff *skb, struct inet6_dev *idev,
3595 u32 pid, u32 seq, int event, unsigned int flags)
3597 struct net_device *dev = idev->dev;
3598 struct nlattr *nla;
3599 struct ifinfomsg *hdr;
3600 struct nlmsghdr *nlh;
3601 void *protoinfo;
3602 struct ifla_cacheinfo ci;
3604 nlh = nlmsg_put(skb, pid, seq, event, sizeof(*hdr), flags);
3605 if (nlh == NULL)
3606 return -EMSGSIZE;
3608 hdr = nlmsg_data(nlh);
3609 hdr->ifi_family = AF_INET6;
3610 hdr->__ifi_pad = 0;
3611 hdr->ifi_type = dev->type;
3612 hdr->ifi_index = dev->ifindex;
3613 hdr->ifi_flags = dev_get_flags(dev);
3614 hdr->ifi_change = 0;
3616 NLA_PUT_STRING(skb, IFLA_IFNAME, dev->name);
3618 if (dev->addr_len)
3619 NLA_PUT(skb, IFLA_ADDRESS, dev->addr_len, dev->dev_addr);
3621 NLA_PUT_U32(skb, IFLA_MTU, dev->mtu);
3622 if (dev->ifindex != dev->iflink)
3623 NLA_PUT_U32(skb, IFLA_LINK, dev->iflink);
3625 protoinfo = nla_nest_start(skb, IFLA_PROTINFO);
3626 if (protoinfo == NULL)
3627 goto nla_put_failure;
3629 NLA_PUT_U32(skb, IFLA_INET6_FLAGS, idev->if_flags);
3631 ci.max_reasm_len = IPV6_MAXPLEN;
3632 ci.tstamp = (__u32)(TIME_DELTA(idev->tstamp, INITIAL_JIFFIES) / HZ * 100
3633 + TIME_DELTA(idev->tstamp, INITIAL_JIFFIES) % HZ * 100 / HZ);
3634 ci.reachable_time = idev->nd_parms->reachable_time;
3635 ci.retrans_time = idev->nd_parms->retrans_time;
3636 NLA_PUT(skb, IFLA_INET6_CACHEINFO, sizeof(ci), &ci);
3638 nla = nla_reserve(skb, IFLA_INET6_CONF, DEVCONF_MAX * sizeof(s32));
3639 if (nla == NULL)
3640 goto nla_put_failure;
3641 ipv6_store_devconf(&idev->cnf, nla_data(nla), nla_len(nla));
3643 /* XXX - MC not implemented */
3645 nla = nla_reserve(skb, IFLA_INET6_STATS, IPSTATS_MIB_MAX * sizeof(u64));
3646 if (nla == NULL)
3647 goto nla_put_failure;
3648 snmp6_fill_stats(nla_data(nla), idev, IFLA_INET6_STATS, nla_len(nla));
3650 nla = nla_reserve(skb, IFLA_INET6_ICMP6STATS, ICMP6_MIB_MAX * sizeof(u64));
3651 if (nla == NULL)
3652 goto nla_put_failure;
3653 snmp6_fill_stats(nla_data(nla), idev, IFLA_INET6_ICMP6STATS, nla_len(nla));
3655 nla_nest_end(skb, protoinfo);
3656 return nlmsg_end(skb, nlh);
3658 nla_put_failure:
3659 nlmsg_cancel(skb, nlh);
3660 return -EMSGSIZE;
3663 static int inet6_dump_ifinfo(struct sk_buff *skb, struct netlink_callback *cb)
3665 struct net *net = skb->sk->sk_net;
3666 int idx, err;
3667 int s_idx = cb->args[0];
3668 struct net_device *dev;
3669 struct inet6_dev *idev;
3671 if (net != &init_net)
3672 return 0;
3674 read_lock(&dev_base_lock);
3675 idx = 0;
3676 for_each_netdev(&init_net, dev) {
3677 if (idx < s_idx)
3678 goto cont;
3679 if ((idev = in6_dev_get(dev)) == NULL)
3680 goto cont;
3681 err = inet6_fill_ifinfo(skb, idev, NETLINK_CB(cb->skb).pid,
3682 cb->nlh->nlmsg_seq, RTM_NEWLINK, NLM_F_MULTI);
3683 in6_dev_put(idev);
3684 if (err <= 0)
3685 break;
3686 cont:
3687 idx++;
3689 read_unlock(&dev_base_lock);
3690 cb->args[0] = idx;
3692 return skb->len;
3695 void inet6_ifinfo_notify(int event, struct inet6_dev *idev)
3697 struct sk_buff *skb;
3698 int err = -ENOBUFS;
3700 skb = nlmsg_new(inet6_if_nlmsg_size(), GFP_ATOMIC);
3701 if (skb == NULL)
3702 goto errout;
3704 err = inet6_fill_ifinfo(skb, idev, 0, 0, event, 0);
3705 if (err < 0) {
3706 /* -EMSGSIZE implies BUG in inet6_if_nlmsg_size() */
3707 WARN_ON(err == -EMSGSIZE);
3708 kfree_skb(skb);
3709 goto errout;
3711 err = rtnl_notify(skb, &init_net, 0, RTNLGRP_IPV6_IFADDR, NULL, GFP_ATOMIC);
3712 errout:
3713 if (err < 0)
3714 rtnl_set_sk_err(&init_net, RTNLGRP_IPV6_IFADDR, err);
3717 static inline size_t inet6_prefix_nlmsg_size(void)
3719 return NLMSG_ALIGN(sizeof(struct prefixmsg))
3720 + nla_total_size(sizeof(struct in6_addr))
3721 + nla_total_size(sizeof(struct prefix_cacheinfo));
3724 static int inet6_fill_prefix(struct sk_buff *skb, struct inet6_dev *idev,
3725 struct prefix_info *pinfo, u32 pid, u32 seq,
3726 int event, unsigned int flags)
3728 struct prefixmsg *pmsg;
3729 struct nlmsghdr *nlh;
3730 struct prefix_cacheinfo ci;
3732 nlh = nlmsg_put(skb, pid, seq, event, sizeof(*pmsg), flags);
3733 if (nlh == NULL)
3734 return -EMSGSIZE;
3736 pmsg = nlmsg_data(nlh);
3737 pmsg->prefix_family = AF_INET6;
3738 pmsg->prefix_pad1 = 0;
3739 pmsg->prefix_pad2 = 0;
3740 pmsg->prefix_ifindex = idev->dev->ifindex;
3741 pmsg->prefix_len = pinfo->prefix_len;
3742 pmsg->prefix_type = pinfo->type;
3743 pmsg->prefix_pad3 = 0;
3744 pmsg->prefix_flags = 0;
3745 if (pinfo->onlink)
3746 pmsg->prefix_flags |= IF_PREFIX_ONLINK;
3747 if (pinfo->autoconf)
3748 pmsg->prefix_flags |= IF_PREFIX_AUTOCONF;
3750 NLA_PUT(skb, PREFIX_ADDRESS, sizeof(pinfo->prefix), &pinfo->prefix);
3752 ci.preferred_time = ntohl(pinfo->prefered);
3753 ci.valid_time = ntohl(pinfo->valid);
3754 NLA_PUT(skb, PREFIX_CACHEINFO, sizeof(ci), &ci);
3756 return nlmsg_end(skb, nlh);
3758 nla_put_failure:
3759 nlmsg_cancel(skb, nlh);
3760 return -EMSGSIZE;
3763 static void inet6_prefix_notify(int event, struct inet6_dev *idev,
3764 struct prefix_info *pinfo)
3766 struct sk_buff *skb;
3767 int err = -ENOBUFS;
3769 skb = nlmsg_new(inet6_prefix_nlmsg_size(), GFP_ATOMIC);
3770 if (skb == NULL)
3771 goto errout;
3773 err = inet6_fill_prefix(skb, idev, pinfo, 0, 0, event, 0);
3774 if (err < 0) {
3775 /* -EMSGSIZE implies BUG in inet6_prefix_nlmsg_size() */
3776 WARN_ON(err == -EMSGSIZE);
3777 kfree_skb(skb);
3778 goto errout;
3780 err = rtnl_notify(skb, &init_net, 0, RTNLGRP_IPV6_PREFIX, NULL, GFP_ATOMIC);
3781 errout:
3782 if (err < 0)
3783 rtnl_set_sk_err(&init_net, RTNLGRP_IPV6_PREFIX, err);
3786 static void __ipv6_ifa_notify(int event, struct inet6_ifaddr *ifp)
3788 inet6_ifa_notify(event ? : RTM_NEWADDR, ifp);
3790 switch (event) {
3791 case RTM_NEWADDR:
3793 * If the address was optimistic
3794 * we inserted the route at the start of
3795 * our DAD process, so we don't need
3796 * to do it again
3798 if (!(ifp->rt->rt6i_node))
3799 ip6_ins_rt(ifp->rt);
3800 if (ifp->idev->cnf.forwarding)
3801 addrconf_join_anycast(ifp);
3802 break;
3803 case RTM_DELADDR:
3804 if (ifp->idev->cnf.forwarding)
3805 addrconf_leave_anycast(ifp);
3806 addrconf_leave_solict(ifp->idev, &ifp->addr);
3807 dst_hold(&ifp->rt->u.dst);
3808 if (ip6_del_rt(ifp->rt))
3809 dst_free(&ifp->rt->u.dst);
3810 break;
3814 static void ipv6_ifa_notify(int event, struct inet6_ifaddr *ifp)
3816 rcu_read_lock_bh();
3817 if (likely(ifp->idev->dead == 0))
3818 __ipv6_ifa_notify(event, ifp);
3819 rcu_read_unlock_bh();
3822 #ifdef CONFIG_SYSCTL
3824 static
3825 int addrconf_sysctl_forward(ctl_table *ctl, int write, struct file * filp,
3826 void __user *buffer, size_t *lenp, loff_t *ppos)
3828 int *valp = ctl->data;
3829 int val = *valp;
3830 int ret;
3832 ret = proc_dointvec(ctl, write, filp, buffer, lenp, ppos);
3834 if (write)
3835 addrconf_fixup_forwarding(ctl, valp, val);
3836 return ret;
3839 static int addrconf_sysctl_forward_strategy(ctl_table *table,
3840 int __user *name, int nlen,
3841 void __user *oldval,
3842 size_t __user *oldlenp,
3843 void __user *newval, size_t newlen)
3845 int *valp = table->data;
3846 int val = *valp;
3847 int new;
3849 if (!newval || !newlen)
3850 return 0;
3851 if (newlen != sizeof(int))
3852 return -EINVAL;
3853 if (get_user(new, (int __user *)newval))
3854 return -EFAULT;
3855 if (new == *valp)
3856 return 0;
3857 if (oldval && oldlenp) {
3858 size_t len;
3859 if (get_user(len, oldlenp))
3860 return -EFAULT;
3861 if (len) {
3862 if (len > table->maxlen)
3863 len = table->maxlen;
3864 if (copy_to_user(oldval, valp, len))
3865 return -EFAULT;
3866 if (put_user(len, oldlenp))
3867 return -EFAULT;
3871 *valp = new;
3872 addrconf_fixup_forwarding(table, valp, val);
3873 return 1;
3876 static struct addrconf_sysctl_table
3878 struct ctl_table_header *sysctl_header;
3879 ctl_table addrconf_vars[__NET_IPV6_MAX];
3880 char *dev_name;
3881 } addrconf_sysctl __read_mostly = {
3882 .sysctl_header = NULL,
3883 .addrconf_vars = {
3885 .ctl_name = NET_IPV6_FORWARDING,
3886 .procname = "forwarding",
3887 .data = &ipv6_devconf.forwarding,
3888 .maxlen = sizeof(int),
3889 .mode = 0644,
3890 .proc_handler = &addrconf_sysctl_forward,
3891 .strategy = &addrconf_sysctl_forward_strategy,
3894 .ctl_name = NET_IPV6_HOP_LIMIT,
3895 .procname = "hop_limit",
3896 .data = &ipv6_devconf.hop_limit,
3897 .maxlen = sizeof(int),
3898 .mode = 0644,
3899 .proc_handler = proc_dointvec,
3902 .ctl_name = NET_IPV6_MTU,
3903 .procname = "mtu",
3904 .data = &ipv6_devconf.mtu6,
3905 .maxlen = sizeof(int),
3906 .mode = 0644,
3907 .proc_handler = &proc_dointvec,
3910 .ctl_name = NET_IPV6_ACCEPT_RA,
3911 .procname = "accept_ra",
3912 .data = &ipv6_devconf.accept_ra,
3913 .maxlen = sizeof(int),
3914 .mode = 0644,
3915 .proc_handler = &proc_dointvec,
3918 .ctl_name = NET_IPV6_ACCEPT_REDIRECTS,
3919 .procname = "accept_redirects",
3920 .data = &ipv6_devconf.accept_redirects,
3921 .maxlen = sizeof(int),
3922 .mode = 0644,
3923 .proc_handler = &proc_dointvec,
3926 .ctl_name = NET_IPV6_AUTOCONF,
3927 .procname = "autoconf",
3928 .data = &ipv6_devconf.autoconf,
3929 .maxlen = sizeof(int),
3930 .mode = 0644,
3931 .proc_handler = &proc_dointvec,
3934 .ctl_name = NET_IPV6_DAD_TRANSMITS,
3935 .procname = "dad_transmits",
3936 .data = &ipv6_devconf.dad_transmits,
3937 .maxlen = sizeof(int),
3938 .mode = 0644,
3939 .proc_handler = &proc_dointvec,
3942 .ctl_name = NET_IPV6_RTR_SOLICITS,
3943 .procname = "router_solicitations",
3944 .data = &ipv6_devconf.rtr_solicits,
3945 .maxlen = sizeof(int),
3946 .mode = 0644,
3947 .proc_handler = &proc_dointvec,
3950 .ctl_name = NET_IPV6_RTR_SOLICIT_INTERVAL,
3951 .procname = "router_solicitation_interval",
3952 .data = &ipv6_devconf.rtr_solicit_interval,
3953 .maxlen = sizeof(int),
3954 .mode = 0644,
3955 .proc_handler = &proc_dointvec_jiffies,
3956 .strategy = &sysctl_jiffies,
3959 .ctl_name = NET_IPV6_RTR_SOLICIT_DELAY,
3960 .procname = "router_solicitation_delay",
3961 .data = &ipv6_devconf.rtr_solicit_delay,
3962 .maxlen = sizeof(int),
3963 .mode = 0644,
3964 .proc_handler = &proc_dointvec_jiffies,
3965 .strategy = &sysctl_jiffies,
3968 .ctl_name = NET_IPV6_FORCE_MLD_VERSION,
3969 .procname = "force_mld_version",
3970 .data = &ipv6_devconf.force_mld_version,
3971 .maxlen = sizeof(int),
3972 .mode = 0644,
3973 .proc_handler = &proc_dointvec,
3975 #ifdef CONFIG_IPV6_PRIVACY
3977 .ctl_name = NET_IPV6_USE_TEMPADDR,
3978 .procname = "use_tempaddr",
3979 .data = &ipv6_devconf.use_tempaddr,
3980 .maxlen = sizeof(int),
3981 .mode = 0644,
3982 .proc_handler = &proc_dointvec,
3985 .ctl_name = NET_IPV6_TEMP_VALID_LFT,
3986 .procname = "temp_valid_lft",
3987 .data = &ipv6_devconf.temp_valid_lft,
3988 .maxlen = sizeof(int),
3989 .mode = 0644,
3990 .proc_handler = &proc_dointvec,
3993 .ctl_name = NET_IPV6_TEMP_PREFERED_LFT,
3994 .procname = "temp_prefered_lft",
3995 .data = &ipv6_devconf.temp_prefered_lft,
3996 .maxlen = sizeof(int),
3997 .mode = 0644,
3998 .proc_handler = &proc_dointvec,
4001 .ctl_name = NET_IPV6_REGEN_MAX_RETRY,
4002 .procname = "regen_max_retry",
4003 .data = &ipv6_devconf.regen_max_retry,
4004 .maxlen = sizeof(int),
4005 .mode = 0644,
4006 .proc_handler = &proc_dointvec,
4009 .ctl_name = NET_IPV6_MAX_DESYNC_FACTOR,
4010 .procname = "max_desync_factor",
4011 .data = &ipv6_devconf.max_desync_factor,
4012 .maxlen = sizeof(int),
4013 .mode = 0644,
4014 .proc_handler = &proc_dointvec,
4016 #endif
4018 .ctl_name = NET_IPV6_MAX_ADDRESSES,
4019 .procname = "max_addresses",
4020 .data = &ipv6_devconf.max_addresses,
4021 .maxlen = sizeof(int),
4022 .mode = 0644,
4023 .proc_handler = &proc_dointvec,
4026 .ctl_name = NET_IPV6_ACCEPT_RA_DEFRTR,
4027 .procname = "accept_ra_defrtr",
4028 .data = &ipv6_devconf.accept_ra_defrtr,
4029 .maxlen = sizeof(int),
4030 .mode = 0644,
4031 .proc_handler = &proc_dointvec,
4034 .ctl_name = NET_IPV6_ACCEPT_RA_PINFO,
4035 .procname = "accept_ra_pinfo",
4036 .data = &ipv6_devconf.accept_ra_pinfo,
4037 .maxlen = sizeof(int),
4038 .mode = 0644,
4039 .proc_handler = &proc_dointvec,
4041 #ifdef CONFIG_IPV6_ROUTER_PREF
4043 .ctl_name = NET_IPV6_ACCEPT_RA_RTR_PREF,
4044 .procname = "accept_ra_rtr_pref",
4045 .data = &ipv6_devconf.accept_ra_rtr_pref,
4046 .maxlen = sizeof(int),
4047 .mode = 0644,
4048 .proc_handler = &proc_dointvec,
4051 .ctl_name = NET_IPV6_RTR_PROBE_INTERVAL,
4052 .procname = "router_probe_interval",
4053 .data = &ipv6_devconf.rtr_probe_interval,
4054 .maxlen = sizeof(int),
4055 .mode = 0644,
4056 .proc_handler = &proc_dointvec_jiffies,
4057 .strategy = &sysctl_jiffies,
4059 #ifdef CONFIG_IPV6_ROUTE_INFO
4061 .ctl_name = NET_IPV6_ACCEPT_RA_RT_INFO_MAX_PLEN,
4062 .procname = "accept_ra_rt_info_max_plen",
4063 .data = &ipv6_devconf.accept_ra_rt_info_max_plen,
4064 .maxlen = sizeof(int),
4065 .mode = 0644,
4066 .proc_handler = &proc_dointvec,
4068 #endif
4069 #endif
4071 .ctl_name = NET_IPV6_PROXY_NDP,
4072 .procname = "proxy_ndp",
4073 .data = &ipv6_devconf.proxy_ndp,
4074 .maxlen = sizeof(int),
4075 .mode = 0644,
4076 .proc_handler = &proc_dointvec,
4079 .ctl_name = NET_IPV6_ACCEPT_SOURCE_ROUTE,
4080 .procname = "accept_source_route",
4081 .data = &ipv6_devconf.accept_source_route,
4082 .maxlen = sizeof(int),
4083 .mode = 0644,
4084 .proc_handler = &proc_dointvec,
4086 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
4088 .ctl_name = CTL_UNNUMBERED,
4089 .procname = "optimistic_dad",
4090 .data = &ipv6_devconf.optimistic_dad,
4091 .maxlen = sizeof(int),
4092 .mode = 0644,
4093 .proc_handler = &proc_dointvec,
4096 #endif
4098 .ctl_name = 0, /* sentinel */
4103 static int __addrconf_sysctl_register(struct net *net, char *dev_name,
4104 int ctl_name, struct inet6_dev *idev, struct ipv6_devconf *p)
4106 int i;
4107 struct addrconf_sysctl_table *t;
4109 #define ADDRCONF_CTL_PATH_DEV 3
4111 struct ctl_path addrconf_ctl_path[] = {
4112 { .procname = "net", .ctl_name = CTL_NET, },
4113 { .procname = "ipv6", .ctl_name = NET_IPV6, },
4114 { .procname = "conf", .ctl_name = NET_IPV6_CONF, },
4115 { /* to be set */ },
4116 { },
4120 t = kmemdup(&addrconf_sysctl, sizeof(*t), GFP_KERNEL);
4121 if (t == NULL)
4122 goto out;
4124 for (i=0; t->addrconf_vars[i].data; i++) {
4125 t->addrconf_vars[i].data += (char*)p - (char*)&ipv6_devconf;
4126 t->addrconf_vars[i].extra1 = idev; /* embedded; no ref */
4127 t->addrconf_vars[i].extra2 = net;
4131 * Make a copy of dev_name, because '.procname' is regarded as const
4132 * by sysctl and we wouldn't want anyone to change it under our feet
4133 * (see SIOCSIFNAME).
4135 t->dev_name = kstrdup(dev_name, GFP_KERNEL);
4136 if (!t->dev_name)
4137 goto free;
4139 addrconf_ctl_path[ADDRCONF_CTL_PATH_DEV].procname = t->dev_name;
4140 addrconf_ctl_path[ADDRCONF_CTL_PATH_DEV].ctl_name = ctl_name;
4142 t->sysctl_header = register_net_sysctl_table(net, addrconf_ctl_path,
4143 t->addrconf_vars);
4144 if (t->sysctl_header == NULL)
4145 goto free_procname;
4147 p->sysctl = t;
4148 return 0;
4150 free_procname:
4151 kfree(t->dev_name);
4152 free:
4153 kfree(t);
4154 out:
4155 return -ENOBUFS;
4158 static void __addrconf_sysctl_unregister(struct ipv6_devconf *p)
4160 struct addrconf_sysctl_table *t;
4162 if (p->sysctl == NULL)
4163 return;
4165 t = p->sysctl;
4166 p->sysctl = NULL;
4167 unregister_sysctl_table(t->sysctl_header);
4168 kfree(t->dev_name);
4169 kfree(t);
4172 static void addrconf_sysctl_register(struct inet6_dev *idev)
4174 neigh_sysctl_register(idev->dev, idev->nd_parms, NET_IPV6,
4175 NET_IPV6_NEIGH, "ipv6",
4176 &ndisc_ifinfo_sysctl_change,
4177 NULL);
4178 __addrconf_sysctl_register(idev->dev->nd_net, idev->dev->name,
4179 idev->dev->ifindex, idev, &idev->cnf);
4182 static void addrconf_sysctl_unregister(struct inet6_dev *idev)
4184 __addrconf_sysctl_unregister(&idev->cnf);
4185 neigh_sysctl_unregister(idev->nd_parms);
4189 #endif
4191 static int addrconf_init_net(struct net *net)
4193 int err;
4194 struct ipv6_devconf *all, *dflt;
4196 err = -ENOMEM;
4197 all = &ipv6_devconf;
4198 dflt = &ipv6_devconf_dflt;
4200 if (net != &init_net) {
4201 all = kmemdup(all, sizeof(ipv6_devconf), GFP_KERNEL);
4202 if (all == NULL)
4203 goto err_alloc_all;
4205 dflt = kmemdup(dflt, sizeof(ipv6_devconf_dflt), GFP_KERNEL);
4206 if (dflt == NULL)
4207 goto err_alloc_dflt;
4210 net->ipv6.devconf_all = all;
4211 net->ipv6.devconf_dflt = dflt;
4213 #ifdef CONFIG_SYSCTL
4214 err = __addrconf_sysctl_register(net, "all", NET_PROTO_CONF_ALL,
4215 NULL, all);
4216 if (err < 0)
4217 goto err_reg_all;
4219 err = __addrconf_sysctl_register(net, "default", NET_PROTO_CONF_DEFAULT,
4220 NULL, dflt);
4221 if (err < 0)
4222 goto err_reg_dflt;
4223 #endif
4224 return 0;
4226 #ifdef CONFIG_SYSCTL
4227 err_reg_dflt:
4228 __addrconf_sysctl_unregister(all);
4229 err_reg_all:
4230 kfree(dflt);
4231 #endif
4232 err_alloc_dflt:
4233 kfree(all);
4234 err_alloc_all:
4235 return err;
4238 static void addrconf_exit_net(struct net *net)
4240 #ifdef CONFIG_SYSCTL
4241 __addrconf_sysctl_unregister(net->ipv6.devconf_dflt);
4242 __addrconf_sysctl_unregister(net->ipv6.devconf_all);
4243 #endif
4244 if (net != &init_net) {
4245 kfree(net->ipv6.devconf_dflt);
4246 kfree(net->ipv6.devconf_all);
4250 static struct pernet_operations addrconf_ops = {
4251 .init = addrconf_init_net,
4252 .exit = addrconf_exit_net,
4256 * Device notifier
4259 int register_inet6addr_notifier(struct notifier_block *nb)
4261 return atomic_notifier_chain_register(&inet6addr_chain, nb);
4264 EXPORT_SYMBOL(register_inet6addr_notifier);
4266 int unregister_inet6addr_notifier(struct notifier_block *nb)
4268 return atomic_notifier_chain_unregister(&inet6addr_chain,nb);
4271 EXPORT_SYMBOL(unregister_inet6addr_notifier);
4274 * Init / cleanup code
4277 int __init addrconf_init(void)
4279 int err;
4281 if ((err = ipv6_addr_label_init()) < 0) {
4282 printk(KERN_CRIT "IPv6 Addrconf: cannot initialize default policy table: %d.\n",
4283 err);
4284 return err;
4287 register_pernet_subsys(&addrconf_ops);
4289 /* The addrconf netdev notifier requires that loopback_dev
4290 * has it's ipv6 private information allocated and setup
4291 * before it can bring up and give link-local addresses
4292 * to other devices which are up.
4294 * Unfortunately, loopback_dev is not necessarily the first
4295 * entry in the global dev_base list of net devices. In fact,
4296 * it is likely to be the very last entry on that list.
4297 * So this causes the notifier registry below to try and
4298 * give link-local addresses to all devices besides loopback_dev
4299 * first, then loopback_dev, which cases all the non-loopback_dev
4300 * devices to fail to get a link-local address.
4302 * So, as a temporary fix, allocate the ipv6 structure for
4303 * loopback_dev first by hand.
4304 * Longer term, all of the dependencies ipv6 has upon the loopback
4305 * device and it being up should be removed.
4307 rtnl_lock();
4308 if (!ipv6_add_dev(init_net.loopback_dev))
4309 err = -ENOMEM;
4310 rtnl_unlock();
4311 if (err)
4312 goto errlo;
4314 ip6_null_entry.u.dst.dev = init_net.loopback_dev;
4315 ip6_null_entry.rt6i_idev = in6_dev_get(init_net.loopback_dev);
4316 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
4317 ip6_prohibit_entry.u.dst.dev = init_net.loopback_dev;
4318 ip6_prohibit_entry.rt6i_idev = in6_dev_get(init_net.loopback_dev);
4319 ip6_blk_hole_entry.u.dst.dev = init_net.loopback_dev;
4320 ip6_blk_hole_entry.rt6i_idev = in6_dev_get(init_net.loopback_dev);
4321 #endif
4323 register_netdevice_notifier(&ipv6_dev_notf);
4325 addrconf_verify(0);
4327 err = __rtnl_register(PF_INET6, RTM_GETLINK, NULL, inet6_dump_ifinfo);
4328 if (err < 0)
4329 goto errout;
4331 /* Only the first call to __rtnl_register can fail */
4332 __rtnl_register(PF_INET6, RTM_NEWADDR, inet6_rtm_newaddr, NULL);
4333 __rtnl_register(PF_INET6, RTM_DELADDR, inet6_rtm_deladdr, NULL);
4334 __rtnl_register(PF_INET6, RTM_GETADDR, inet6_rtm_getaddr, inet6_dump_ifaddr);
4335 __rtnl_register(PF_INET6, RTM_GETMULTICAST, NULL, inet6_dump_ifmcaddr);
4336 __rtnl_register(PF_INET6, RTM_GETANYCAST, NULL, inet6_dump_ifacaddr);
4338 ipv6_addr_label_rtnl_register();
4340 return 0;
4341 errout:
4342 unregister_netdevice_notifier(&ipv6_dev_notf);
4343 errlo:
4344 unregister_pernet_subsys(&addrconf_ops);
4346 return err;
4349 void addrconf_cleanup(void)
4351 struct net_device *dev;
4352 struct inet6_ifaddr *ifa;
4353 int i;
4355 unregister_netdevice_notifier(&ipv6_dev_notf);
4357 unregister_pernet_subsys(&addrconf_ops);
4359 rtnl_lock();
4362 * clean dev list.
4365 for_each_netdev(&init_net, dev) {
4366 if (__in6_dev_get(dev) == NULL)
4367 continue;
4368 addrconf_ifdown(dev, 1);
4370 addrconf_ifdown(init_net.loopback_dev, 2);
4373 * Check hash table.
4376 write_lock_bh(&addrconf_hash_lock);
4377 for (i=0; i < IN6_ADDR_HSIZE; i++) {
4378 for (ifa=inet6_addr_lst[i]; ifa; ) {
4379 struct inet6_ifaddr *bifa;
4381 bifa = ifa;
4382 ifa = ifa->lst_next;
4383 printk(KERN_DEBUG "bug: IPv6 address leakage detected: ifa=%p\n", bifa);
4384 /* Do not free it; something is wrong.
4385 Now we can investigate it with debugger.
4389 write_unlock_bh(&addrconf_hash_lock);
4391 del_timer(&addr_chk_timer);
4393 rtnl_unlock();