[IPV6]: inet6_dev on loopback should be kept until namespace stop.
[linux-2.6/openmoko-kernel/knife-kernel.git] / net / ipv6 / addrconf.c
bloba65935a9afd93e983200cd10a380ffa4eb7071ba
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 unsigned long regen_advance;
780 int tmp_plen;
781 int ret = 0;
782 int max_addresses;
783 u32 addr_flags;
785 write_lock(&idev->lock);
786 if (ift) {
787 spin_lock_bh(&ift->lock);
788 memcpy(&addr.s6_addr[8], &ift->addr.s6_addr[8], 8);
789 spin_unlock_bh(&ift->lock);
790 tmpaddr = &addr;
791 } else {
792 tmpaddr = NULL;
794 retry:
795 in6_dev_hold(idev);
796 if (idev->cnf.use_tempaddr <= 0) {
797 write_unlock(&idev->lock);
798 printk(KERN_INFO
799 "ipv6_create_tempaddr(): use_tempaddr is disabled.\n");
800 in6_dev_put(idev);
801 ret = -1;
802 goto out;
804 spin_lock_bh(&ifp->lock);
805 if (ifp->regen_count++ >= idev->cnf.regen_max_retry) {
806 idev->cnf.use_tempaddr = -1; /*XXX*/
807 spin_unlock_bh(&ifp->lock);
808 write_unlock(&idev->lock);
809 printk(KERN_WARNING
810 "ipv6_create_tempaddr(): regeneration time exceeded. disabled temporary address support.\n");
811 in6_dev_put(idev);
812 ret = -1;
813 goto out;
815 in6_ifa_hold(ifp);
816 memcpy(addr.s6_addr, ifp->addr.s6_addr, 8);
817 if (__ipv6_try_regen_rndid(idev, tmpaddr) < 0) {
818 spin_unlock_bh(&ifp->lock);
819 write_unlock(&idev->lock);
820 printk(KERN_WARNING
821 "ipv6_create_tempaddr(): regeneration of randomized interface id failed.\n");
822 in6_ifa_put(ifp);
823 in6_dev_put(idev);
824 ret = -1;
825 goto out;
827 memcpy(&addr.s6_addr[8], idev->rndid, 8);
828 tmp_valid_lft = min_t(__u32,
829 ifp->valid_lft,
830 idev->cnf.temp_valid_lft);
831 tmp_prefered_lft = min_t(__u32,
832 ifp->prefered_lft,
833 idev->cnf.temp_prefered_lft - desync_factor / HZ);
834 tmp_plen = ifp->prefix_len;
835 max_addresses = idev->cnf.max_addresses;
836 tmp_cstamp = ifp->cstamp;
837 tmp_tstamp = ifp->tstamp;
838 spin_unlock_bh(&ifp->lock);
840 regen_advance = idev->cnf.regen_max_retry *
841 idev->cnf.dad_transmits *
842 idev->nd_parms->retrans_time / HZ;
843 write_unlock(&idev->lock);
845 /* A temporary address is created only if this calculated Preferred
846 * Lifetime is greater than REGEN_ADVANCE time units. In particular,
847 * an implementation must not create a temporary address with a zero
848 * Preferred Lifetime.
850 if (tmp_prefered_lft <= regen_advance) {
851 in6_ifa_put(ifp);
852 in6_dev_put(idev);
853 ret = -1;
854 goto out;
857 addr_flags = IFA_F_TEMPORARY;
858 /* set in addrconf_prefix_rcv() */
859 if (ifp->flags & IFA_F_OPTIMISTIC)
860 addr_flags |= IFA_F_OPTIMISTIC;
862 ift = !max_addresses ||
863 ipv6_count_addresses(idev) < max_addresses ?
864 ipv6_add_addr(idev, &addr, tmp_plen,
865 ipv6_addr_type(&addr)&IPV6_ADDR_SCOPE_MASK,
866 addr_flags) : NULL;
867 if (!ift || IS_ERR(ift)) {
868 in6_ifa_put(ifp);
869 in6_dev_put(idev);
870 printk(KERN_INFO
871 "ipv6_create_tempaddr(): retry temporary address regeneration.\n");
872 tmpaddr = &addr;
873 write_lock(&idev->lock);
874 goto retry;
877 spin_lock_bh(&ift->lock);
878 ift->ifpub = ifp;
879 ift->valid_lft = tmp_valid_lft;
880 ift->prefered_lft = tmp_prefered_lft;
881 ift->cstamp = tmp_cstamp;
882 ift->tstamp = tmp_tstamp;
883 spin_unlock_bh(&ift->lock);
885 addrconf_dad_start(ift, 0);
886 in6_ifa_put(ift);
887 in6_dev_put(idev);
888 out:
889 return ret;
891 #endif
894 * Choose an appropriate source address (RFC3484)
896 struct ipv6_saddr_score {
897 int addr_type;
898 unsigned int attrs;
899 int matchlen;
900 int scope;
901 unsigned int rule;
904 #define IPV6_SADDR_SCORE_LOCAL 0x0001
905 #define IPV6_SADDR_SCORE_PREFERRED 0x0004
906 #define IPV6_SADDR_SCORE_HOA 0x0008
907 #define IPV6_SADDR_SCORE_OIF 0x0010
908 #define IPV6_SADDR_SCORE_LABEL 0x0020
909 #define IPV6_SADDR_SCORE_PRIVACY 0x0040
911 static inline int ipv6_saddr_preferred(int type)
913 if (type & (IPV6_ADDR_MAPPED|IPV6_ADDR_COMPATv4|
914 IPV6_ADDR_LOOPBACK|IPV6_ADDR_RESERVED))
915 return 1;
916 return 0;
919 int ipv6_dev_get_saddr(struct net_device *daddr_dev,
920 struct in6_addr *daddr, struct in6_addr *saddr)
922 struct ipv6_saddr_score hiscore;
923 struct inet6_ifaddr *ifa_result = NULL;
924 int daddr_type = __ipv6_addr_type(daddr);
925 int daddr_scope = __ipv6_addr_src_scope(daddr_type);
926 int daddr_ifindex = daddr_dev ? daddr_dev->ifindex : 0;
927 u32 daddr_label = ipv6_addr_label(daddr, daddr_type, daddr_ifindex);
928 struct net_device *dev;
930 memset(&hiscore, 0, sizeof(hiscore));
932 read_lock(&dev_base_lock);
933 rcu_read_lock();
935 for_each_netdev(&init_net, dev) {
936 struct inet6_dev *idev;
937 struct inet6_ifaddr *ifa;
939 /* Rule 0: Candidate Source Address (section 4)
940 * - multicast and link-local destination address,
941 * the set of candidate source address MUST only
942 * include addresses assigned to interfaces
943 * belonging to the same link as the outgoing
944 * interface.
945 * (- For site-local destination addresses, the
946 * set of candidate source addresses MUST only
947 * include addresses assigned to interfaces
948 * belonging to the same site as the outgoing
949 * interface.)
951 if ((daddr_type & IPV6_ADDR_MULTICAST ||
952 daddr_scope <= IPV6_ADDR_SCOPE_LINKLOCAL) &&
953 daddr_dev && dev != daddr_dev)
954 continue;
956 idev = __in6_dev_get(dev);
957 if (!idev)
958 continue;
960 read_lock_bh(&idev->lock);
961 for (ifa = idev->addr_list; ifa; ifa = ifa->if_next) {
962 struct ipv6_saddr_score score;
964 score.addr_type = __ipv6_addr_type(&ifa->addr);
966 /* Rule 0:
967 * - Tentative Address (RFC2462 section 5.4)
968 * - A tentative address is not considered
969 * "assigned to an interface" in the traditional
970 * sense, unless it is also flagged as optimistic.
971 * - Candidate Source Address (section 4)
972 * - In any case, anycast addresses, multicast
973 * addresses, and the unspecified address MUST
974 * NOT be included in a candidate set.
976 if ((ifa->flags & IFA_F_TENTATIVE) &&
977 (!(ifa->flags & IFA_F_OPTIMISTIC)))
978 continue;
979 if (unlikely(score.addr_type == IPV6_ADDR_ANY ||
980 score.addr_type & IPV6_ADDR_MULTICAST)) {
981 LIMIT_NETDEBUG(KERN_DEBUG
982 "ADDRCONF: unspecified / multicast address "
983 "assigned as unicast address on %s",
984 dev->name);
985 continue;
988 score.attrs = 0;
989 score.matchlen = 0;
990 score.scope = 0;
991 score.rule = 0;
993 if (ifa_result == NULL) {
994 /* record it if the first available entry */
995 goto record_it;
998 /* Rule 1: Prefer same address */
999 if (hiscore.rule < 1) {
1000 if (ipv6_addr_equal(&ifa_result->addr, daddr))
1001 hiscore.attrs |= IPV6_SADDR_SCORE_LOCAL;
1002 hiscore.rule++;
1004 if (ipv6_addr_equal(&ifa->addr, daddr)) {
1005 score.attrs |= IPV6_SADDR_SCORE_LOCAL;
1006 if (!(hiscore.attrs & IPV6_SADDR_SCORE_LOCAL)) {
1007 score.rule = 1;
1008 goto record_it;
1010 } else {
1011 if (hiscore.attrs & IPV6_SADDR_SCORE_LOCAL)
1012 continue;
1015 /* Rule 2: Prefer appropriate scope */
1016 if (hiscore.rule < 2) {
1017 hiscore.scope = __ipv6_addr_src_scope(hiscore.addr_type);
1018 hiscore.rule++;
1020 score.scope = __ipv6_addr_src_scope(score.addr_type);
1021 if (hiscore.scope < score.scope) {
1022 if (hiscore.scope < daddr_scope) {
1023 score.rule = 2;
1024 goto record_it;
1025 } else
1026 continue;
1027 } else if (score.scope < hiscore.scope) {
1028 if (score.scope < daddr_scope)
1029 break; /* addresses sorted by scope */
1030 else {
1031 score.rule = 2;
1032 goto record_it;
1036 /* Rule 3: Avoid deprecated and optimistic addresses */
1037 if (hiscore.rule < 3) {
1038 if (ipv6_saddr_preferred(hiscore.addr_type) ||
1039 (((ifa_result->flags &
1040 (IFA_F_DEPRECATED|IFA_F_OPTIMISTIC)) == 0)))
1041 hiscore.attrs |= IPV6_SADDR_SCORE_PREFERRED;
1042 hiscore.rule++;
1044 if (ipv6_saddr_preferred(score.addr_type) ||
1045 (((ifa->flags &
1046 (IFA_F_DEPRECATED|IFA_F_OPTIMISTIC)) == 0))) {
1047 score.attrs |= IPV6_SADDR_SCORE_PREFERRED;
1048 if (!(hiscore.attrs & IPV6_SADDR_SCORE_PREFERRED)) {
1049 score.rule = 3;
1050 goto record_it;
1052 } else {
1053 if (hiscore.attrs & IPV6_SADDR_SCORE_PREFERRED)
1054 continue;
1057 /* Rule 4: Prefer home address */
1058 #if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
1059 if (hiscore.rule < 4) {
1060 if (ifa_result->flags & IFA_F_HOMEADDRESS)
1061 hiscore.attrs |= IPV6_SADDR_SCORE_HOA;
1062 hiscore.rule++;
1064 if (ifa->flags & IFA_F_HOMEADDRESS) {
1065 score.attrs |= IPV6_SADDR_SCORE_HOA;
1066 if (!(ifa_result->flags & IFA_F_HOMEADDRESS)) {
1067 score.rule = 4;
1068 goto record_it;
1070 } else {
1071 if (hiscore.attrs & IPV6_SADDR_SCORE_HOA)
1072 continue;
1074 #else
1075 if (hiscore.rule < 4)
1076 hiscore.rule++;
1077 #endif
1079 /* Rule 5: Prefer outgoing interface */
1080 if (hiscore.rule < 5) {
1081 if (daddr_dev == NULL ||
1082 daddr_dev == ifa_result->idev->dev)
1083 hiscore.attrs |= IPV6_SADDR_SCORE_OIF;
1084 hiscore.rule++;
1086 if (daddr_dev == NULL ||
1087 daddr_dev == ifa->idev->dev) {
1088 score.attrs |= IPV6_SADDR_SCORE_OIF;
1089 if (!(hiscore.attrs & IPV6_SADDR_SCORE_OIF)) {
1090 score.rule = 5;
1091 goto record_it;
1093 } else {
1094 if (hiscore.attrs & IPV6_SADDR_SCORE_OIF)
1095 continue;
1098 /* Rule 6: Prefer matching label */
1099 if (hiscore.rule < 6) {
1100 if (ipv6_addr_label(&ifa_result->addr,
1101 hiscore.addr_type,
1102 ifa_result->idev->dev->ifindex) == daddr_label)
1103 hiscore.attrs |= IPV6_SADDR_SCORE_LABEL;
1104 hiscore.rule++;
1106 if (ipv6_addr_label(&ifa->addr,
1107 score.addr_type,
1108 ifa->idev->dev->ifindex) == daddr_label) {
1109 score.attrs |= IPV6_SADDR_SCORE_LABEL;
1110 if (!(hiscore.attrs & IPV6_SADDR_SCORE_LABEL)) {
1111 score.rule = 6;
1112 goto record_it;
1114 } else {
1115 if (hiscore.attrs & IPV6_SADDR_SCORE_LABEL)
1116 continue;
1119 #ifdef CONFIG_IPV6_PRIVACY
1120 /* Rule 7: Prefer public address
1121 * Note: prefer temprary address if use_tempaddr >= 2
1123 if (hiscore.rule < 7) {
1124 if ((!(ifa_result->flags & IFA_F_TEMPORARY)) ^
1125 (ifa_result->idev->cnf.use_tempaddr >= 2))
1126 hiscore.attrs |= IPV6_SADDR_SCORE_PRIVACY;
1127 hiscore.rule++;
1129 if ((!(ifa->flags & IFA_F_TEMPORARY)) ^
1130 (ifa->idev->cnf.use_tempaddr >= 2)) {
1131 score.attrs |= IPV6_SADDR_SCORE_PRIVACY;
1132 if (!(hiscore.attrs & IPV6_SADDR_SCORE_PRIVACY)) {
1133 score.rule = 7;
1134 goto record_it;
1136 } else {
1137 if (hiscore.attrs & IPV6_SADDR_SCORE_PRIVACY)
1138 continue;
1140 #else
1141 if (hiscore.rule < 7)
1142 hiscore.rule++;
1143 #endif
1144 /* Rule 8: Use longest matching prefix */
1145 if (hiscore.rule < 8) {
1146 hiscore.matchlen = ipv6_addr_diff(&ifa_result->addr, daddr);
1147 hiscore.rule++;
1149 score.matchlen = ipv6_addr_diff(&ifa->addr, daddr);
1150 if (score.matchlen > hiscore.matchlen) {
1151 score.rule = 8;
1152 goto record_it;
1154 #if 0
1155 else if (score.matchlen < hiscore.matchlen)
1156 continue;
1157 #endif
1159 /* Final Rule: choose first available one */
1160 continue;
1161 record_it:
1162 if (ifa_result)
1163 in6_ifa_put(ifa_result);
1164 in6_ifa_hold(ifa);
1165 ifa_result = ifa;
1166 hiscore = score;
1168 read_unlock_bh(&idev->lock);
1170 rcu_read_unlock();
1171 read_unlock(&dev_base_lock);
1173 if (!ifa_result)
1174 return -EADDRNOTAVAIL;
1176 ipv6_addr_copy(saddr, &ifa_result->addr);
1177 in6_ifa_put(ifa_result);
1178 return 0;
1182 int ipv6_get_saddr(struct dst_entry *dst,
1183 struct in6_addr *daddr, struct in6_addr *saddr)
1185 return ipv6_dev_get_saddr(dst ? ip6_dst_idev(dst)->dev : NULL, daddr, saddr);
1188 EXPORT_SYMBOL(ipv6_get_saddr);
1190 int ipv6_get_lladdr(struct net_device *dev, struct in6_addr *addr,
1191 unsigned char banned_flags)
1193 struct inet6_dev *idev;
1194 int err = -EADDRNOTAVAIL;
1196 rcu_read_lock();
1197 if ((idev = __in6_dev_get(dev)) != NULL) {
1198 struct inet6_ifaddr *ifp;
1200 read_lock_bh(&idev->lock);
1201 for (ifp=idev->addr_list; ifp; ifp=ifp->if_next) {
1202 if (ifp->scope == IFA_LINK && !(ifp->flags & banned_flags)) {
1203 ipv6_addr_copy(addr, &ifp->addr);
1204 err = 0;
1205 break;
1208 read_unlock_bh(&idev->lock);
1210 rcu_read_unlock();
1211 return err;
1214 static int ipv6_count_addresses(struct inet6_dev *idev)
1216 int cnt = 0;
1217 struct inet6_ifaddr *ifp;
1219 read_lock_bh(&idev->lock);
1220 for (ifp=idev->addr_list; ifp; ifp=ifp->if_next)
1221 cnt++;
1222 read_unlock_bh(&idev->lock);
1223 return cnt;
1226 int ipv6_chk_addr(struct net *net, struct in6_addr *addr,
1227 struct net_device *dev, int strict)
1229 struct inet6_ifaddr * ifp;
1230 u8 hash = ipv6_addr_hash(addr);
1232 read_lock_bh(&addrconf_hash_lock);
1233 for(ifp = inet6_addr_lst[hash]; ifp; ifp=ifp->lst_next) {
1234 if (ifp->idev->dev->nd_net != net)
1235 continue;
1236 if (ipv6_addr_equal(&ifp->addr, addr) &&
1237 !(ifp->flags&IFA_F_TENTATIVE)) {
1238 if (dev == NULL || ifp->idev->dev == dev ||
1239 !(ifp->scope&(IFA_LINK|IFA_HOST) || strict))
1240 break;
1243 read_unlock_bh(&addrconf_hash_lock);
1244 return ifp != NULL;
1246 EXPORT_SYMBOL(ipv6_chk_addr);
1248 static
1249 int ipv6_chk_same_addr(struct net *net, const struct in6_addr *addr,
1250 struct net_device *dev)
1252 struct inet6_ifaddr * ifp;
1253 u8 hash = ipv6_addr_hash(addr);
1255 for(ifp = inet6_addr_lst[hash]; ifp; ifp=ifp->lst_next) {
1256 if (ifp->idev->dev->nd_net != net)
1257 continue;
1258 if (ipv6_addr_equal(&ifp->addr, addr)) {
1259 if (dev == NULL || ifp->idev->dev == dev)
1260 break;
1263 return ifp != NULL;
1266 struct inet6_ifaddr *ipv6_get_ifaddr(struct net *net, struct in6_addr *addr,
1267 struct net_device *dev, int strict)
1269 struct inet6_ifaddr * ifp;
1270 u8 hash = ipv6_addr_hash(addr);
1272 read_lock_bh(&addrconf_hash_lock);
1273 for(ifp = inet6_addr_lst[hash]; ifp; ifp=ifp->lst_next) {
1274 if (ifp->idev->dev->nd_net != net)
1275 continue;
1276 if (ipv6_addr_equal(&ifp->addr, addr)) {
1277 if (dev == NULL || ifp->idev->dev == dev ||
1278 !(ifp->scope&(IFA_LINK|IFA_HOST) || strict)) {
1279 in6_ifa_hold(ifp);
1280 break;
1284 read_unlock_bh(&addrconf_hash_lock);
1286 return ifp;
1289 int ipv6_rcv_saddr_equal(const struct sock *sk, const struct sock *sk2)
1291 const struct in6_addr *sk_rcv_saddr6 = &inet6_sk(sk)->rcv_saddr;
1292 const struct in6_addr *sk2_rcv_saddr6 = inet6_rcv_saddr(sk2);
1293 __be32 sk_rcv_saddr = inet_sk(sk)->rcv_saddr;
1294 __be32 sk2_rcv_saddr = inet_rcv_saddr(sk2);
1295 int sk_ipv6only = ipv6_only_sock(sk);
1296 int sk2_ipv6only = inet_v6_ipv6only(sk2);
1297 int addr_type = ipv6_addr_type(sk_rcv_saddr6);
1298 int addr_type2 = sk2_rcv_saddr6 ? ipv6_addr_type(sk2_rcv_saddr6) : IPV6_ADDR_MAPPED;
1300 if (!sk2_rcv_saddr && !sk_ipv6only)
1301 return 1;
1303 if (addr_type2 == IPV6_ADDR_ANY &&
1304 !(sk2_ipv6only && addr_type == IPV6_ADDR_MAPPED))
1305 return 1;
1307 if (addr_type == IPV6_ADDR_ANY &&
1308 !(sk_ipv6only && addr_type2 == IPV6_ADDR_MAPPED))
1309 return 1;
1311 if (sk2_rcv_saddr6 &&
1312 ipv6_addr_equal(sk_rcv_saddr6, sk2_rcv_saddr6))
1313 return 1;
1315 if (addr_type == IPV6_ADDR_MAPPED &&
1316 !sk2_ipv6only &&
1317 (!sk2_rcv_saddr || !sk_rcv_saddr || sk_rcv_saddr == sk2_rcv_saddr))
1318 return 1;
1320 return 0;
1323 /* Gets referenced address, destroys ifaddr */
1325 static void addrconf_dad_stop(struct inet6_ifaddr *ifp)
1327 if (ifp->flags&IFA_F_PERMANENT) {
1328 spin_lock_bh(&ifp->lock);
1329 addrconf_del_timer(ifp);
1330 ifp->flags |= IFA_F_TENTATIVE;
1331 spin_unlock_bh(&ifp->lock);
1332 in6_ifa_put(ifp);
1333 #ifdef CONFIG_IPV6_PRIVACY
1334 } else if (ifp->flags&IFA_F_TEMPORARY) {
1335 struct inet6_ifaddr *ifpub;
1336 spin_lock_bh(&ifp->lock);
1337 ifpub = ifp->ifpub;
1338 if (ifpub) {
1339 in6_ifa_hold(ifpub);
1340 spin_unlock_bh(&ifp->lock);
1341 ipv6_create_tempaddr(ifpub, ifp);
1342 in6_ifa_put(ifpub);
1343 } else {
1344 spin_unlock_bh(&ifp->lock);
1346 ipv6_del_addr(ifp);
1347 #endif
1348 } else
1349 ipv6_del_addr(ifp);
1352 void addrconf_dad_failure(struct inet6_ifaddr *ifp)
1354 if (net_ratelimit())
1355 printk(KERN_INFO "%s: duplicate address detected!\n", ifp->idev->dev->name);
1356 addrconf_dad_stop(ifp);
1359 /* Join to solicited addr multicast group. */
1361 void addrconf_join_solict(struct net_device *dev, struct in6_addr *addr)
1363 struct in6_addr maddr;
1365 if (dev->flags&(IFF_LOOPBACK|IFF_NOARP))
1366 return;
1368 addrconf_addr_solict_mult(addr, &maddr);
1369 ipv6_dev_mc_inc(dev, &maddr);
1372 void addrconf_leave_solict(struct inet6_dev *idev, struct in6_addr *addr)
1374 struct in6_addr maddr;
1376 if (idev->dev->flags&(IFF_LOOPBACK|IFF_NOARP))
1377 return;
1379 addrconf_addr_solict_mult(addr, &maddr);
1380 __ipv6_dev_mc_dec(idev, &maddr);
1383 static void addrconf_join_anycast(struct inet6_ifaddr *ifp)
1385 struct in6_addr addr;
1386 ipv6_addr_prefix(&addr, &ifp->addr, ifp->prefix_len);
1387 if (ipv6_addr_any(&addr))
1388 return;
1389 ipv6_dev_ac_inc(ifp->idev->dev, &addr);
1392 static void addrconf_leave_anycast(struct inet6_ifaddr *ifp)
1394 struct in6_addr addr;
1395 ipv6_addr_prefix(&addr, &ifp->addr, ifp->prefix_len);
1396 if (ipv6_addr_any(&addr))
1397 return;
1398 __ipv6_dev_ac_dec(ifp->idev, &addr);
1401 static int addrconf_ifid_eui48(u8 *eui, struct net_device *dev)
1403 if (dev->addr_len != ETH_ALEN)
1404 return -1;
1405 memcpy(eui, dev->dev_addr, 3);
1406 memcpy(eui + 5, dev->dev_addr + 3, 3);
1409 * The zSeries OSA network cards can be shared among various
1410 * OS instances, but the OSA cards have only one MAC address.
1411 * This leads to duplicate address conflicts in conjunction
1412 * with IPv6 if more than one instance uses the same card.
1414 * The driver for these cards can deliver a unique 16-bit
1415 * identifier for each instance sharing the same card. It is
1416 * placed instead of 0xFFFE in the interface identifier. The
1417 * "u" bit of the interface identifier is not inverted in this
1418 * case. Hence the resulting interface identifier has local
1419 * scope according to RFC2373.
1421 if (dev->dev_id) {
1422 eui[3] = (dev->dev_id >> 8) & 0xFF;
1423 eui[4] = dev->dev_id & 0xFF;
1424 } else {
1425 eui[3] = 0xFF;
1426 eui[4] = 0xFE;
1427 eui[0] ^= 2;
1429 return 0;
1432 static int addrconf_ifid_arcnet(u8 *eui, struct net_device *dev)
1434 /* XXX: inherit EUI-64 from other interface -- yoshfuji */
1435 if (dev->addr_len != ARCNET_ALEN)
1436 return -1;
1437 memset(eui, 0, 7);
1438 eui[7] = *(u8*)dev->dev_addr;
1439 return 0;
1442 static int addrconf_ifid_infiniband(u8 *eui, struct net_device *dev)
1444 if (dev->addr_len != INFINIBAND_ALEN)
1445 return -1;
1446 memcpy(eui, dev->dev_addr + 12, 8);
1447 eui[0] |= 2;
1448 return 0;
1451 static int ipv6_generate_eui64(u8 *eui, struct net_device *dev)
1453 switch (dev->type) {
1454 case ARPHRD_ETHER:
1455 case ARPHRD_FDDI:
1456 case ARPHRD_IEEE802_TR:
1457 return addrconf_ifid_eui48(eui, dev);
1458 case ARPHRD_ARCNET:
1459 return addrconf_ifid_arcnet(eui, dev);
1460 case ARPHRD_INFINIBAND:
1461 return addrconf_ifid_infiniband(eui, dev);
1462 case ARPHRD_SIT:
1463 if (dev->priv_flags & IFF_ISATAP)
1464 return ipv6_isatap_eui64(eui, *(__be32 *)dev->dev_addr);
1466 return -1;
1469 static int ipv6_inherit_eui64(u8 *eui, struct inet6_dev *idev)
1471 int err = -1;
1472 struct inet6_ifaddr *ifp;
1474 read_lock_bh(&idev->lock);
1475 for (ifp=idev->addr_list; ifp; ifp=ifp->if_next) {
1476 if (ifp->scope == IFA_LINK && !(ifp->flags&IFA_F_TENTATIVE)) {
1477 memcpy(eui, ifp->addr.s6_addr+8, 8);
1478 err = 0;
1479 break;
1482 read_unlock_bh(&idev->lock);
1483 return err;
1486 #ifdef CONFIG_IPV6_PRIVACY
1487 /* (re)generation of randomized interface identifier (RFC 3041 3.2, 3.5) */
1488 static int __ipv6_regen_rndid(struct inet6_dev *idev)
1490 regen:
1491 get_random_bytes(idev->rndid, sizeof(idev->rndid));
1492 idev->rndid[0] &= ~0x02;
1495 * <draft-ietf-ipngwg-temp-addresses-v2-00.txt>:
1496 * check if generated address is not inappropriate
1498 * - Reserved subnet anycast (RFC 2526)
1499 * 11111101 11....11 1xxxxxxx
1500 * - ISATAP (RFC4214) 6.1
1501 * 00-00-5E-FE-xx-xx-xx-xx
1502 * - value 0
1503 * - XXX: already assigned to an address on the device
1505 if (idev->rndid[0] == 0xfd &&
1506 (idev->rndid[1]&idev->rndid[2]&idev->rndid[3]&idev->rndid[4]&idev->rndid[5]&idev->rndid[6]) == 0xff &&
1507 (idev->rndid[7]&0x80))
1508 goto regen;
1509 if ((idev->rndid[0]|idev->rndid[1]) == 0) {
1510 if (idev->rndid[2] == 0x5e && idev->rndid[3] == 0xfe)
1511 goto regen;
1512 if ((idev->rndid[2]|idev->rndid[3]|idev->rndid[4]|idev->rndid[5]|idev->rndid[6]|idev->rndid[7]) == 0x00)
1513 goto regen;
1516 return 0;
1519 static void ipv6_regen_rndid(unsigned long data)
1521 struct inet6_dev *idev = (struct inet6_dev *) data;
1522 unsigned long expires;
1524 rcu_read_lock_bh();
1525 write_lock_bh(&idev->lock);
1527 if (idev->dead)
1528 goto out;
1530 if (__ipv6_regen_rndid(idev) < 0)
1531 goto out;
1533 expires = jiffies +
1534 idev->cnf.temp_prefered_lft * HZ -
1535 idev->cnf.regen_max_retry * idev->cnf.dad_transmits * idev->nd_parms->retrans_time - desync_factor;
1536 if (time_before(expires, jiffies)) {
1537 printk(KERN_WARNING
1538 "ipv6_regen_rndid(): too short regeneration interval; timer disabled for %s.\n",
1539 idev->dev->name);
1540 goto out;
1543 if (!mod_timer(&idev->regen_timer, expires))
1544 in6_dev_hold(idev);
1546 out:
1547 write_unlock_bh(&idev->lock);
1548 rcu_read_unlock_bh();
1549 in6_dev_put(idev);
1552 static int __ipv6_try_regen_rndid(struct inet6_dev *idev, struct in6_addr *tmpaddr) {
1553 int ret = 0;
1555 if (tmpaddr && memcmp(idev->rndid, &tmpaddr->s6_addr[8], 8) == 0)
1556 ret = __ipv6_regen_rndid(idev);
1557 return ret;
1559 #endif
1562 * Add prefix route.
1565 static void
1566 addrconf_prefix_route(struct in6_addr *pfx, int plen, struct net_device *dev,
1567 unsigned long expires, u32 flags)
1569 struct fib6_config cfg = {
1570 .fc_table = RT6_TABLE_PREFIX,
1571 .fc_metric = IP6_RT_PRIO_ADDRCONF,
1572 .fc_ifindex = dev->ifindex,
1573 .fc_expires = expires,
1574 .fc_dst_len = plen,
1575 .fc_flags = RTF_UP | flags,
1576 .fc_nlinfo.nl_net = &init_net,
1579 ipv6_addr_copy(&cfg.fc_dst, pfx);
1581 /* Prevent useless cloning on PtP SIT.
1582 This thing is done here expecting that the whole
1583 class of non-broadcast devices need not cloning.
1585 #if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
1586 if (dev->type == ARPHRD_SIT && (dev->flags & IFF_POINTOPOINT))
1587 cfg.fc_flags |= RTF_NONEXTHOP;
1588 #endif
1590 ip6_route_add(&cfg);
1593 /* Create "default" multicast route to the interface */
1595 static void addrconf_add_mroute(struct net_device *dev)
1597 struct fib6_config cfg = {
1598 .fc_table = RT6_TABLE_LOCAL,
1599 .fc_metric = IP6_RT_PRIO_ADDRCONF,
1600 .fc_ifindex = dev->ifindex,
1601 .fc_dst_len = 8,
1602 .fc_flags = RTF_UP,
1603 .fc_nlinfo.nl_net = &init_net,
1606 ipv6_addr_set(&cfg.fc_dst, htonl(0xFF000000), 0, 0, 0);
1608 ip6_route_add(&cfg);
1611 #if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
1612 static void sit_route_add(struct net_device *dev)
1614 struct fib6_config cfg = {
1615 .fc_table = RT6_TABLE_MAIN,
1616 .fc_metric = IP6_RT_PRIO_ADDRCONF,
1617 .fc_ifindex = dev->ifindex,
1618 .fc_dst_len = 96,
1619 .fc_flags = RTF_UP | RTF_NONEXTHOP,
1620 .fc_nlinfo.nl_net = &init_net,
1623 /* prefix length - 96 bits "::d.d.d.d" */
1624 ip6_route_add(&cfg);
1626 #endif
1628 static void addrconf_add_lroute(struct net_device *dev)
1630 struct in6_addr addr;
1632 ipv6_addr_set(&addr, htonl(0xFE800000), 0, 0, 0);
1633 addrconf_prefix_route(&addr, 64, dev, 0, 0);
1636 static struct inet6_dev *addrconf_add_dev(struct net_device *dev)
1638 struct inet6_dev *idev;
1640 ASSERT_RTNL();
1642 if ((idev = ipv6_find_idev(dev)) == NULL)
1643 return NULL;
1645 /* Add default multicast route */
1646 addrconf_add_mroute(dev);
1648 /* Add link local route */
1649 addrconf_add_lroute(dev);
1650 return idev;
1653 void addrconf_prefix_rcv(struct net_device *dev, u8 *opt, int len)
1655 struct prefix_info *pinfo;
1656 __u32 valid_lft;
1657 __u32 prefered_lft;
1658 int addr_type;
1659 unsigned long rt_expires;
1660 struct inet6_dev *in6_dev;
1662 pinfo = (struct prefix_info *) opt;
1664 if (len < sizeof(struct prefix_info)) {
1665 ADBG(("addrconf: prefix option too short\n"));
1666 return;
1670 * Validation checks ([ADDRCONF], page 19)
1673 addr_type = ipv6_addr_type(&pinfo->prefix);
1675 if (addr_type & (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL))
1676 return;
1678 valid_lft = ntohl(pinfo->valid);
1679 prefered_lft = ntohl(pinfo->prefered);
1681 if (prefered_lft > valid_lft) {
1682 if (net_ratelimit())
1683 printk(KERN_WARNING "addrconf: prefix option has invalid lifetime\n");
1684 return;
1687 in6_dev = in6_dev_get(dev);
1689 if (in6_dev == NULL) {
1690 if (net_ratelimit())
1691 printk(KERN_DEBUG "addrconf: device %s not configured\n", dev->name);
1692 return;
1696 * Two things going on here:
1697 * 1) Add routes for on-link prefixes
1698 * 2) Configure prefixes with the auto flag set
1701 /* Avoid arithmetic overflow. Really, we could
1702 save rt_expires in seconds, likely valid_lft,
1703 but it would require division in fib gc, that it
1704 not good.
1706 if (valid_lft >= 0x7FFFFFFF/HZ)
1707 rt_expires = 0x7FFFFFFF - (0x7FFFFFFF % HZ);
1708 else
1709 rt_expires = valid_lft * HZ;
1712 * We convert this (in jiffies) to clock_t later.
1713 * Avoid arithmetic overflow there as well.
1714 * Overflow can happen only if HZ < USER_HZ.
1716 if (HZ < USER_HZ && rt_expires > 0x7FFFFFFF / USER_HZ)
1717 rt_expires = 0x7FFFFFFF / USER_HZ;
1719 if (pinfo->onlink) {
1720 struct rt6_info *rt;
1721 rt = rt6_lookup(&pinfo->prefix, NULL, dev->ifindex, 1);
1723 if (rt && ((rt->rt6i_flags & (RTF_GATEWAY | RTF_DEFAULT)) == 0)) {
1724 if (rt->rt6i_flags&RTF_EXPIRES) {
1725 if (valid_lft == 0) {
1726 ip6_del_rt(rt);
1727 rt = NULL;
1728 } else {
1729 rt->rt6i_expires = jiffies + rt_expires;
1732 } else if (valid_lft) {
1733 addrconf_prefix_route(&pinfo->prefix, pinfo->prefix_len,
1734 dev, jiffies_to_clock_t(rt_expires), RTF_ADDRCONF|RTF_EXPIRES|RTF_PREFIX_RT);
1736 if (rt)
1737 dst_release(&rt->u.dst);
1740 /* Try to figure out our local address for this prefix */
1742 if (pinfo->autoconf && in6_dev->cnf.autoconf) {
1743 struct inet6_ifaddr * ifp;
1744 struct in6_addr addr;
1745 int create = 0, update_lft = 0;
1747 if (pinfo->prefix_len == 64) {
1748 memcpy(&addr, &pinfo->prefix, 8);
1749 if (ipv6_generate_eui64(addr.s6_addr + 8, dev) &&
1750 ipv6_inherit_eui64(addr.s6_addr + 8, in6_dev)) {
1751 in6_dev_put(in6_dev);
1752 return;
1754 goto ok;
1756 if (net_ratelimit())
1757 printk(KERN_DEBUG "IPv6 addrconf: prefix with wrong length %d\n",
1758 pinfo->prefix_len);
1759 in6_dev_put(in6_dev);
1760 return;
1764 ifp = ipv6_get_ifaddr(&init_net, &addr, dev, 1);
1766 if (ifp == NULL && valid_lft) {
1767 int max_addresses = in6_dev->cnf.max_addresses;
1768 u32 addr_flags = 0;
1770 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
1771 if (in6_dev->cnf.optimistic_dad &&
1772 !ipv6_devconf.forwarding)
1773 addr_flags = IFA_F_OPTIMISTIC;
1774 #endif
1776 /* Do not allow to create too much of autoconfigured
1777 * addresses; this would be too easy way to crash kernel.
1779 if (!max_addresses ||
1780 ipv6_count_addresses(in6_dev) < max_addresses)
1781 ifp = ipv6_add_addr(in6_dev, &addr, pinfo->prefix_len,
1782 addr_type&IPV6_ADDR_SCOPE_MASK,
1783 addr_flags);
1785 if (!ifp || IS_ERR(ifp)) {
1786 in6_dev_put(in6_dev);
1787 return;
1790 update_lft = create = 1;
1791 ifp->cstamp = jiffies;
1792 addrconf_dad_start(ifp, RTF_ADDRCONF|RTF_PREFIX_RT);
1795 if (ifp) {
1796 int flags;
1797 unsigned long now;
1798 #ifdef CONFIG_IPV6_PRIVACY
1799 struct inet6_ifaddr *ift;
1800 #endif
1801 u32 stored_lft;
1803 /* update lifetime (RFC2462 5.5.3 e) */
1804 spin_lock(&ifp->lock);
1805 now = jiffies;
1806 if (ifp->valid_lft > (now - ifp->tstamp) / HZ)
1807 stored_lft = ifp->valid_lft - (now - ifp->tstamp) / HZ;
1808 else
1809 stored_lft = 0;
1810 if (!update_lft && stored_lft) {
1811 if (valid_lft > MIN_VALID_LIFETIME ||
1812 valid_lft > stored_lft)
1813 update_lft = 1;
1814 else if (stored_lft <= MIN_VALID_LIFETIME) {
1815 /* valid_lft <= stored_lft is always true */
1816 /* XXX: IPsec */
1817 update_lft = 0;
1818 } else {
1819 valid_lft = MIN_VALID_LIFETIME;
1820 if (valid_lft < prefered_lft)
1821 prefered_lft = valid_lft;
1822 update_lft = 1;
1826 if (update_lft) {
1827 ifp->valid_lft = valid_lft;
1828 ifp->prefered_lft = prefered_lft;
1829 ifp->tstamp = now;
1830 flags = ifp->flags;
1831 ifp->flags &= ~IFA_F_DEPRECATED;
1832 spin_unlock(&ifp->lock);
1834 if (!(flags&IFA_F_TENTATIVE))
1835 ipv6_ifa_notify(0, ifp);
1836 } else
1837 spin_unlock(&ifp->lock);
1839 #ifdef CONFIG_IPV6_PRIVACY
1840 read_lock_bh(&in6_dev->lock);
1841 /* update all temporary addresses in the list */
1842 for (ift=in6_dev->tempaddr_list; ift; ift=ift->tmp_next) {
1844 * When adjusting the lifetimes of an existing
1845 * temporary address, only lower the lifetimes.
1846 * Implementations must not increase the
1847 * lifetimes of an existing temporary address
1848 * when processing a Prefix Information Option.
1850 if (ifp != ift->ifpub)
1851 continue;
1853 spin_lock(&ift->lock);
1854 flags = ift->flags;
1855 if (ift->valid_lft > valid_lft &&
1856 ift->valid_lft - valid_lft > (jiffies - ift->tstamp) / HZ)
1857 ift->valid_lft = valid_lft + (jiffies - ift->tstamp) / HZ;
1858 if (ift->prefered_lft > prefered_lft &&
1859 ift->prefered_lft - prefered_lft > (jiffies - ift->tstamp) / HZ)
1860 ift->prefered_lft = prefered_lft + (jiffies - ift->tstamp) / HZ;
1861 spin_unlock(&ift->lock);
1862 if (!(flags&IFA_F_TENTATIVE))
1863 ipv6_ifa_notify(0, ift);
1866 if (create && in6_dev->cnf.use_tempaddr > 0) {
1868 * When a new public address is created as described in [ADDRCONF],
1869 * also create a new temporary address.
1871 read_unlock_bh(&in6_dev->lock);
1872 ipv6_create_tempaddr(ifp, NULL);
1873 } else {
1874 read_unlock_bh(&in6_dev->lock);
1876 #endif
1877 in6_ifa_put(ifp);
1878 addrconf_verify(0);
1881 inet6_prefix_notify(RTM_NEWPREFIX, in6_dev, pinfo);
1882 in6_dev_put(in6_dev);
1886 * Set destination address.
1887 * Special case for SIT interfaces where we create a new "virtual"
1888 * device.
1890 int addrconf_set_dstaddr(void __user *arg)
1892 struct in6_ifreq ireq;
1893 struct net_device *dev;
1894 int err = -EINVAL;
1896 rtnl_lock();
1898 err = -EFAULT;
1899 if (copy_from_user(&ireq, arg, sizeof(struct in6_ifreq)))
1900 goto err_exit;
1902 dev = __dev_get_by_index(&init_net, ireq.ifr6_ifindex);
1904 err = -ENODEV;
1905 if (dev == NULL)
1906 goto err_exit;
1908 #if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
1909 if (dev->type == ARPHRD_SIT) {
1910 struct ifreq ifr;
1911 mm_segment_t oldfs;
1912 struct ip_tunnel_parm p;
1914 err = -EADDRNOTAVAIL;
1915 if (!(ipv6_addr_type(&ireq.ifr6_addr) & IPV6_ADDR_COMPATv4))
1916 goto err_exit;
1918 memset(&p, 0, sizeof(p));
1919 p.iph.daddr = ireq.ifr6_addr.s6_addr32[3];
1920 p.iph.saddr = 0;
1921 p.iph.version = 4;
1922 p.iph.ihl = 5;
1923 p.iph.protocol = IPPROTO_IPV6;
1924 p.iph.ttl = 64;
1925 ifr.ifr_ifru.ifru_data = (__force void __user *)&p;
1927 oldfs = get_fs(); set_fs(KERNEL_DS);
1928 err = dev->do_ioctl(dev, &ifr, SIOCADDTUNNEL);
1929 set_fs(oldfs);
1931 if (err == 0) {
1932 err = -ENOBUFS;
1933 if ((dev = __dev_get_by_name(&init_net, p.name)) == NULL)
1934 goto err_exit;
1935 err = dev_open(dev);
1938 #endif
1940 err_exit:
1941 rtnl_unlock();
1942 return err;
1946 * Manual configuration of address on an interface
1948 static int inet6_addr_add(int ifindex, struct in6_addr *pfx, int plen,
1949 __u8 ifa_flags, __u32 prefered_lft, __u32 valid_lft)
1951 struct inet6_ifaddr *ifp;
1952 struct inet6_dev *idev;
1953 struct net_device *dev;
1954 int scope;
1955 u32 flags = RTF_EXPIRES;
1957 ASSERT_RTNL();
1959 /* check the lifetime */
1960 if (!valid_lft || prefered_lft > valid_lft)
1961 return -EINVAL;
1963 if ((dev = __dev_get_by_index(&init_net, ifindex)) == NULL)
1964 return -ENODEV;
1966 if ((idev = addrconf_add_dev(dev)) == NULL)
1967 return -ENOBUFS;
1969 scope = ipv6_addr_scope(pfx);
1971 if (valid_lft == INFINITY_LIFE_TIME) {
1972 ifa_flags |= IFA_F_PERMANENT;
1973 flags = 0;
1974 } else if (valid_lft >= 0x7FFFFFFF/HZ)
1975 valid_lft = 0x7FFFFFFF/HZ;
1977 if (prefered_lft == 0)
1978 ifa_flags |= IFA_F_DEPRECATED;
1979 else if ((prefered_lft >= 0x7FFFFFFF/HZ) &&
1980 (prefered_lft != INFINITY_LIFE_TIME))
1981 prefered_lft = 0x7FFFFFFF/HZ;
1983 ifp = ipv6_add_addr(idev, pfx, plen, scope, ifa_flags);
1985 if (!IS_ERR(ifp)) {
1986 spin_lock_bh(&ifp->lock);
1987 ifp->valid_lft = valid_lft;
1988 ifp->prefered_lft = prefered_lft;
1989 ifp->tstamp = jiffies;
1990 spin_unlock_bh(&ifp->lock);
1992 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, dev,
1993 jiffies_to_clock_t(valid_lft * HZ), flags);
1995 * Note that section 3.1 of RFC 4429 indicates
1996 * that the Optimistic flag should not be set for
1997 * manually configured addresses
1999 addrconf_dad_start(ifp, 0);
2000 in6_ifa_put(ifp);
2001 addrconf_verify(0);
2002 return 0;
2005 return PTR_ERR(ifp);
2008 static int inet6_addr_del(int ifindex, struct in6_addr *pfx, int plen)
2010 struct inet6_ifaddr *ifp;
2011 struct inet6_dev *idev;
2012 struct net_device *dev;
2014 if ((dev = __dev_get_by_index(&init_net, ifindex)) == NULL)
2015 return -ENODEV;
2017 if ((idev = __in6_dev_get(dev)) == NULL)
2018 return -ENXIO;
2020 read_lock_bh(&idev->lock);
2021 for (ifp = idev->addr_list; ifp; ifp=ifp->if_next) {
2022 if (ifp->prefix_len == plen &&
2023 ipv6_addr_equal(pfx, &ifp->addr)) {
2024 in6_ifa_hold(ifp);
2025 read_unlock_bh(&idev->lock);
2027 ipv6_del_addr(ifp);
2029 /* If the last address is deleted administratively,
2030 disable IPv6 on this interface.
2032 if (idev->addr_list == NULL)
2033 addrconf_ifdown(idev->dev, 1);
2034 return 0;
2037 read_unlock_bh(&idev->lock);
2038 return -EADDRNOTAVAIL;
2042 int addrconf_add_ifaddr(void __user *arg)
2044 struct in6_ifreq ireq;
2045 int err;
2047 if (!capable(CAP_NET_ADMIN))
2048 return -EPERM;
2050 if (copy_from_user(&ireq, arg, sizeof(struct in6_ifreq)))
2051 return -EFAULT;
2053 rtnl_lock();
2054 err = inet6_addr_add(ireq.ifr6_ifindex, &ireq.ifr6_addr, ireq.ifr6_prefixlen,
2055 IFA_F_PERMANENT, INFINITY_LIFE_TIME, INFINITY_LIFE_TIME);
2056 rtnl_unlock();
2057 return err;
2060 int addrconf_del_ifaddr(void __user *arg)
2062 struct in6_ifreq ireq;
2063 int err;
2065 if (!capable(CAP_NET_ADMIN))
2066 return -EPERM;
2068 if (copy_from_user(&ireq, arg, sizeof(struct in6_ifreq)))
2069 return -EFAULT;
2071 rtnl_lock();
2072 err = inet6_addr_del(ireq.ifr6_ifindex, &ireq.ifr6_addr, ireq.ifr6_prefixlen);
2073 rtnl_unlock();
2074 return err;
2077 #if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
2078 static void sit_add_v4_addrs(struct inet6_dev *idev)
2080 struct inet6_ifaddr * ifp;
2081 struct in6_addr addr;
2082 struct net_device *dev;
2083 int scope;
2085 ASSERT_RTNL();
2087 memset(&addr, 0, sizeof(struct in6_addr));
2088 memcpy(&addr.s6_addr32[3], idev->dev->dev_addr, 4);
2090 if (idev->dev->flags&IFF_POINTOPOINT) {
2091 addr.s6_addr32[0] = htonl(0xfe800000);
2092 scope = IFA_LINK;
2093 } else {
2094 scope = IPV6_ADDR_COMPATv4;
2097 if (addr.s6_addr32[3]) {
2098 ifp = ipv6_add_addr(idev, &addr, 128, scope, IFA_F_PERMANENT);
2099 if (!IS_ERR(ifp)) {
2100 spin_lock_bh(&ifp->lock);
2101 ifp->flags &= ~IFA_F_TENTATIVE;
2102 spin_unlock_bh(&ifp->lock);
2103 ipv6_ifa_notify(RTM_NEWADDR, ifp);
2104 in6_ifa_put(ifp);
2106 return;
2109 for_each_netdev(&init_net, dev) {
2110 struct in_device * in_dev = __in_dev_get_rtnl(dev);
2111 if (in_dev && (dev->flags & IFF_UP)) {
2112 struct in_ifaddr * ifa;
2114 int flag = scope;
2116 for (ifa = in_dev->ifa_list; ifa; ifa = ifa->ifa_next) {
2117 int plen;
2119 addr.s6_addr32[3] = ifa->ifa_local;
2121 if (ifa->ifa_scope == RT_SCOPE_LINK)
2122 continue;
2123 if (ifa->ifa_scope >= RT_SCOPE_HOST) {
2124 if (idev->dev->flags&IFF_POINTOPOINT)
2125 continue;
2126 flag |= IFA_HOST;
2128 if (idev->dev->flags&IFF_POINTOPOINT)
2129 plen = 64;
2130 else
2131 plen = 96;
2133 ifp = ipv6_add_addr(idev, &addr, plen, flag,
2134 IFA_F_PERMANENT);
2135 if (!IS_ERR(ifp)) {
2136 spin_lock_bh(&ifp->lock);
2137 ifp->flags &= ~IFA_F_TENTATIVE;
2138 spin_unlock_bh(&ifp->lock);
2139 ipv6_ifa_notify(RTM_NEWADDR, ifp);
2140 in6_ifa_put(ifp);
2146 #endif
2148 static void init_loopback(struct net_device *dev)
2150 struct inet6_dev *idev;
2151 struct inet6_ifaddr * ifp;
2153 /* ::1 */
2155 ASSERT_RTNL();
2157 if ((idev = ipv6_find_idev(dev)) == NULL) {
2158 printk(KERN_DEBUG "init loopback: add_dev failed\n");
2159 return;
2162 ifp = ipv6_add_addr(idev, &in6addr_loopback, 128, IFA_HOST, IFA_F_PERMANENT);
2163 if (!IS_ERR(ifp)) {
2164 spin_lock_bh(&ifp->lock);
2165 ifp->flags &= ~IFA_F_TENTATIVE;
2166 spin_unlock_bh(&ifp->lock);
2167 ipv6_ifa_notify(RTM_NEWADDR, ifp);
2168 in6_ifa_put(ifp);
2172 static void addrconf_add_linklocal(struct inet6_dev *idev, struct in6_addr *addr)
2174 struct inet6_ifaddr * ifp;
2175 u32 addr_flags = IFA_F_PERMANENT;
2177 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
2178 if (idev->cnf.optimistic_dad &&
2179 !ipv6_devconf.forwarding)
2180 addr_flags |= IFA_F_OPTIMISTIC;
2181 #endif
2184 ifp = ipv6_add_addr(idev, addr, 64, IFA_LINK, addr_flags);
2185 if (!IS_ERR(ifp)) {
2186 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, idev->dev, 0, 0);
2187 addrconf_dad_start(ifp, 0);
2188 in6_ifa_put(ifp);
2192 static void addrconf_dev_config(struct net_device *dev)
2194 struct in6_addr addr;
2195 struct inet6_dev * idev;
2197 ASSERT_RTNL();
2199 if ((dev->type != ARPHRD_ETHER) &&
2200 (dev->type != ARPHRD_FDDI) &&
2201 (dev->type != ARPHRD_IEEE802_TR) &&
2202 (dev->type != ARPHRD_ARCNET) &&
2203 (dev->type != ARPHRD_INFINIBAND)) {
2204 /* Alas, we support only Ethernet autoconfiguration. */
2205 return;
2208 idev = addrconf_add_dev(dev);
2209 if (idev == NULL)
2210 return;
2212 memset(&addr, 0, sizeof(struct in6_addr));
2213 addr.s6_addr32[0] = htonl(0xFE800000);
2215 if (ipv6_generate_eui64(addr.s6_addr + 8, dev) == 0)
2216 addrconf_add_linklocal(idev, &addr);
2219 #if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
2220 static void addrconf_sit_config(struct net_device *dev)
2222 struct inet6_dev *idev;
2224 ASSERT_RTNL();
2227 * Configure the tunnel with one of our IPv4
2228 * addresses... we should configure all of
2229 * our v4 addrs in the tunnel
2232 if ((idev = ipv6_find_idev(dev)) == NULL) {
2233 printk(KERN_DEBUG "init sit: add_dev failed\n");
2234 return;
2237 if (dev->priv_flags & IFF_ISATAP) {
2238 struct in6_addr addr;
2240 ipv6_addr_set(&addr, htonl(0xFE800000), 0, 0, 0);
2241 addrconf_prefix_route(&addr, 64, dev, 0, 0);
2242 if (!ipv6_generate_eui64(addr.s6_addr + 8, dev))
2243 addrconf_add_linklocal(idev, &addr);
2244 return;
2247 sit_add_v4_addrs(idev);
2249 if (dev->flags&IFF_POINTOPOINT) {
2250 addrconf_add_mroute(dev);
2251 addrconf_add_lroute(dev);
2252 } else
2253 sit_route_add(dev);
2255 #endif
2257 static inline int
2258 ipv6_inherit_linklocal(struct inet6_dev *idev, struct net_device *link_dev)
2260 struct in6_addr lladdr;
2262 if (!ipv6_get_lladdr(link_dev, &lladdr, IFA_F_TENTATIVE)) {
2263 addrconf_add_linklocal(idev, &lladdr);
2264 return 0;
2266 return -1;
2269 static void ip6_tnl_add_linklocal(struct inet6_dev *idev)
2271 struct net_device *link_dev;
2273 /* first try to inherit the link-local address from the link device */
2274 if (idev->dev->iflink &&
2275 (link_dev = __dev_get_by_index(&init_net, idev->dev->iflink))) {
2276 if (!ipv6_inherit_linklocal(idev, link_dev))
2277 return;
2279 /* then try to inherit it from any device */
2280 for_each_netdev(&init_net, link_dev) {
2281 if (!ipv6_inherit_linklocal(idev, link_dev))
2282 return;
2284 printk(KERN_DEBUG "init ip6-ip6: add_linklocal failed\n");
2288 * Autoconfigure tunnel with a link-local address so routing protocols,
2289 * DHCPv6, MLD etc. can be run over the virtual link
2292 static void addrconf_ip6_tnl_config(struct net_device *dev)
2294 struct inet6_dev *idev;
2296 ASSERT_RTNL();
2298 if ((idev = addrconf_add_dev(dev)) == NULL) {
2299 printk(KERN_DEBUG "init ip6-ip6: add_dev failed\n");
2300 return;
2302 ip6_tnl_add_linklocal(idev);
2305 static int addrconf_notify(struct notifier_block *this, unsigned long event,
2306 void * data)
2308 struct net_device *dev = (struct net_device *) data;
2309 struct inet6_dev *idev = __in6_dev_get(dev);
2310 int run_pending = 0;
2311 int err;
2313 if (dev->nd_net != &init_net)
2314 return NOTIFY_DONE;
2316 switch(event) {
2317 case NETDEV_REGISTER:
2318 if (!idev && dev->mtu >= IPV6_MIN_MTU) {
2319 idev = ipv6_add_dev(dev);
2320 if (!idev)
2321 return notifier_from_errno(-ENOMEM);
2323 break;
2324 case NETDEV_UP:
2325 case NETDEV_CHANGE:
2326 if (dev->flags & IFF_SLAVE)
2327 break;
2329 if (event == NETDEV_UP) {
2330 if (!addrconf_qdisc_ok(dev)) {
2331 /* device is not ready yet. */
2332 printk(KERN_INFO
2333 "ADDRCONF(NETDEV_UP): %s: "
2334 "link is not ready\n",
2335 dev->name);
2336 break;
2339 if (!idev && dev->mtu >= IPV6_MIN_MTU)
2340 idev = ipv6_add_dev(dev);
2342 if (idev)
2343 idev->if_flags |= IF_READY;
2344 } else {
2345 if (!addrconf_qdisc_ok(dev)) {
2346 /* device is still not ready. */
2347 break;
2350 if (idev) {
2351 if (idev->if_flags & IF_READY) {
2352 /* device is already configured. */
2353 break;
2355 idev->if_flags |= IF_READY;
2358 printk(KERN_INFO
2359 "ADDRCONF(NETDEV_CHANGE): %s: "
2360 "link becomes ready\n",
2361 dev->name);
2363 run_pending = 1;
2366 switch(dev->type) {
2367 #if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
2368 case ARPHRD_SIT:
2369 addrconf_sit_config(dev);
2370 break;
2371 #endif
2372 case ARPHRD_TUNNEL6:
2373 addrconf_ip6_tnl_config(dev);
2374 break;
2375 case ARPHRD_LOOPBACK:
2376 init_loopback(dev);
2377 break;
2379 default:
2380 addrconf_dev_config(dev);
2381 break;
2383 if (idev) {
2384 if (run_pending)
2385 addrconf_dad_run(idev);
2387 /* If the MTU changed during the interface down, when the
2388 interface up, the changed MTU must be reflected in the
2389 idev as well as routers.
2391 if (idev->cnf.mtu6 != dev->mtu && dev->mtu >= IPV6_MIN_MTU) {
2392 rt6_mtu_change(dev, dev->mtu);
2393 idev->cnf.mtu6 = dev->mtu;
2395 idev->tstamp = jiffies;
2396 inet6_ifinfo_notify(RTM_NEWLINK, idev);
2397 /* If the changed mtu during down is lower than IPV6_MIN_MTU
2398 stop IPv6 on this interface.
2400 if (dev->mtu < IPV6_MIN_MTU)
2401 addrconf_ifdown(dev, event != NETDEV_DOWN);
2403 break;
2405 case NETDEV_CHANGEMTU:
2406 if (idev && dev->mtu >= IPV6_MIN_MTU) {
2407 rt6_mtu_change(dev, dev->mtu);
2408 idev->cnf.mtu6 = dev->mtu;
2409 break;
2412 if (!idev && dev->mtu >= IPV6_MIN_MTU) {
2413 idev = ipv6_add_dev(dev);
2414 if (idev)
2415 break;
2418 /* MTU falled under IPV6_MIN_MTU. Stop IPv6 on this interface. */
2420 case NETDEV_DOWN:
2421 case NETDEV_UNREGISTER:
2423 * Remove all addresses from this interface.
2425 addrconf_ifdown(dev, event != NETDEV_DOWN);
2426 break;
2428 case NETDEV_CHANGENAME:
2429 if (idev) {
2430 snmp6_unregister_dev(idev);
2431 addrconf_sysctl_unregister(idev);
2432 addrconf_sysctl_register(idev);
2433 err = snmp6_register_dev(idev);
2434 if (err)
2435 return notifier_from_errno(err);
2437 break;
2440 return NOTIFY_OK;
2444 * addrconf module should be notified of a device going up
2446 static struct notifier_block ipv6_dev_notf = {
2447 .notifier_call = addrconf_notify,
2448 .priority = 0
2451 static int addrconf_ifdown(struct net_device *dev, int how)
2453 struct inet6_dev *idev;
2454 struct inet6_ifaddr *ifa, **bifa;
2455 int i;
2457 ASSERT_RTNL();
2459 if ((dev->flags & IFF_LOOPBACK) && how == 1)
2460 how = 0;
2462 rt6_ifdown(dev);
2463 neigh_ifdown(&nd_tbl, dev);
2465 idev = __in6_dev_get(dev);
2466 if (idev == NULL)
2467 return -ENODEV;
2469 /* Step 1: remove reference to ipv6 device from parent device.
2470 Do not dev_put!
2472 if (how) {
2473 idev->dead = 1;
2475 /* protected by rtnl_lock */
2476 rcu_assign_pointer(dev->ip6_ptr, NULL);
2478 /* Step 1.5: remove snmp6 entry */
2479 snmp6_unregister_dev(idev);
2483 /* Step 2: clear hash table */
2484 for (i=0; i<IN6_ADDR_HSIZE; i++) {
2485 bifa = &inet6_addr_lst[i];
2487 write_lock_bh(&addrconf_hash_lock);
2488 while ((ifa = *bifa) != NULL) {
2489 if (ifa->idev == idev) {
2490 *bifa = ifa->lst_next;
2491 ifa->lst_next = NULL;
2492 addrconf_del_timer(ifa);
2493 in6_ifa_put(ifa);
2494 continue;
2496 bifa = &ifa->lst_next;
2498 write_unlock_bh(&addrconf_hash_lock);
2501 write_lock_bh(&idev->lock);
2503 /* Step 3: clear flags for stateless addrconf */
2504 if (!how)
2505 idev->if_flags &= ~(IF_RS_SENT|IF_RA_RCVD|IF_READY);
2507 /* Step 4: clear address list */
2508 #ifdef CONFIG_IPV6_PRIVACY
2509 if (how && del_timer(&idev->regen_timer))
2510 in6_dev_put(idev);
2512 /* clear tempaddr list */
2513 while ((ifa = idev->tempaddr_list) != NULL) {
2514 idev->tempaddr_list = ifa->tmp_next;
2515 ifa->tmp_next = NULL;
2516 ifa->dead = 1;
2517 write_unlock_bh(&idev->lock);
2518 spin_lock_bh(&ifa->lock);
2520 if (ifa->ifpub) {
2521 in6_ifa_put(ifa->ifpub);
2522 ifa->ifpub = NULL;
2524 spin_unlock_bh(&ifa->lock);
2525 in6_ifa_put(ifa);
2526 write_lock_bh(&idev->lock);
2528 #endif
2529 while ((ifa = idev->addr_list) != NULL) {
2530 idev->addr_list = ifa->if_next;
2531 ifa->if_next = NULL;
2532 ifa->dead = 1;
2533 addrconf_del_timer(ifa);
2534 write_unlock_bh(&idev->lock);
2536 __ipv6_ifa_notify(RTM_DELADDR, ifa);
2537 atomic_notifier_call_chain(&inet6addr_chain, NETDEV_DOWN, ifa);
2538 in6_ifa_put(ifa);
2540 write_lock_bh(&idev->lock);
2542 write_unlock_bh(&idev->lock);
2544 /* Step 5: Discard multicast list */
2546 if (how)
2547 ipv6_mc_destroy_dev(idev);
2548 else
2549 ipv6_mc_down(idev);
2551 idev->tstamp = jiffies;
2553 /* Shot the device (if unregistered) */
2555 if (how) {
2556 addrconf_sysctl_unregister(idev);
2557 neigh_parms_release(&nd_tbl, idev->nd_parms);
2558 neigh_ifdown(&nd_tbl, dev);
2559 in6_dev_put(idev);
2561 return 0;
2564 static void addrconf_rs_timer(unsigned long data)
2566 struct inet6_ifaddr *ifp = (struct inet6_ifaddr *) data;
2568 if (ifp->idev->cnf.forwarding)
2569 goto out;
2571 if (ifp->idev->if_flags & IF_RA_RCVD) {
2573 * Announcement received after solicitation
2574 * was sent
2576 goto out;
2579 spin_lock(&ifp->lock);
2580 if (ifp->probes++ < ifp->idev->cnf.rtr_solicits) {
2581 struct in6_addr all_routers;
2583 /* The wait after the last probe can be shorter */
2584 addrconf_mod_timer(ifp, AC_RS,
2585 (ifp->probes == ifp->idev->cnf.rtr_solicits) ?
2586 ifp->idev->cnf.rtr_solicit_delay :
2587 ifp->idev->cnf.rtr_solicit_interval);
2588 spin_unlock(&ifp->lock);
2590 ipv6_addr_all_routers(&all_routers);
2592 ndisc_send_rs(ifp->idev->dev, &ifp->addr, &all_routers);
2593 } else {
2594 spin_unlock(&ifp->lock);
2596 * Note: we do not support deprecated "all on-link"
2597 * assumption any longer.
2599 printk(KERN_DEBUG "%s: no IPv6 routers present\n",
2600 ifp->idev->dev->name);
2603 out:
2604 in6_ifa_put(ifp);
2608 * Duplicate Address Detection
2610 static void addrconf_dad_kick(struct inet6_ifaddr *ifp)
2612 unsigned long rand_num;
2613 struct inet6_dev *idev = ifp->idev;
2615 if (ifp->flags & IFA_F_OPTIMISTIC)
2616 rand_num = 0;
2617 else
2618 rand_num = net_random() % (idev->cnf.rtr_solicit_delay ? : 1);
2620 ifp->probes = idev->cnf.dad_transmits;
2621 addrconf_mod_timer(ifp, AC_DAD, rand_num);
2624 static void addrconf_dad_start(struct inet6_ifaddr *ifp, u32 flags)
2626 struct inet6_dev *idev = ifp->idev;
2627 struct net_device *dev = idev->dev;
2629 addrconf_join_solict(dev, &ifp->addr);
2631 net_srandom(ifp->addr.s6_addr32[3]);
2633 read_lock_bh(&idev->lock);
2634 if (ifp->dead)
2635 goto out;
2636 spin_lock_bh(&ifp->lock);
2638 if (dev->flags&(IFF_NOARP|IFF_LOOPBACK) ||
2639 !(ifp->flags&IFA_F_TENTATIVE) ||
2640 ifp->flags & IFA_F_NODAD) {
2641 ifp->flags &= ~(IFA_F_TENTATIVE|IFA_F_OPTIMISTIC);
2642 spin_unlock_bh(&ifp->lock);
2643 read_unlock_bh(&idev->lock);
2645 addrconf_dad_completed(ifp);
2646 return;
2649 if (!(idev->if_flags & IF_READY)) {
2650 spin_unlock_bh(&ifp->lock);
2651 read_unlock_bh(&idev->lock);
2653 * If the defice is not ready:
2654 * - keep it tentative if it is a permanent address.
2655 * - otherwise, kill it.
2657 in6_ifa_hold(ifp);
2658 addrconf_dad_stop(ifp);
2659 return;
2663 * Optimistic nodes can start receiving
2664 * Frames right away
2666 if(ifp->flags & IFA_F_OPTIMISTIC)
2667 ip6_ins_rt(ifp->rt);
2669 addrconf_dad_kick(ifp);
2670 spin_unlock_bh(&ifp->lock);
2671 out:
2672 read_unlock_bh(&idev->lock);
2675 static void addrconf_dad_timer(unsigned long data)
2677 struct inet6_ifaddr *ifp = (struct inet6_ifaddr *) data;
2678 struct inet6_dev *idev = ifp->idev;
2679 struct in6_addr unspec;
2680 struct in6_addr mcaddr;
2682 read_lock_bh(&idev->lock);
2683 if (idev->dead) {
2684 read_unlock_bh(&idev->lock);
2685 goto out;
2687 spin_lock_bh(&ifp->lock);
2688 if (ifp->probes == 0) {
2690 * DAD was successful
2693 ifp->flags &= ~(IFA_F_TENTATIVE|IFA_F_OPTIMISTIC);
2694 spin_unlock_bh(&ifp->lock);
2695 read_unlock_bh(&idev->lock);
2697 addrconf_dad_completed(ifp);
2699 goto out;
2702 ifp->probes--;
2703 addrconf_mod_timer(ifp, AC_DAD, ifp->idev->nd_parms->retrans_time);
2704 spin_unlock_bh(&ifp->lock);
2705 read_unlock_bh(&idev->lock);
2707 /* send a neighbour solicitation for our addr */
2708 memset(&unspec, 0, sizeof(unspec));
2709 addrconf_addr_solict_mult(&ifp->addr, &mcaddr);
2710 ndisc_send_ns(ifp->idev->dev, NULL, &ifp->addr, &mcaddr, &unspec);
2711 out:
2712 in6_ifa_put(ifp);
2715 static void addrconf_dad_completed(struct inet6_ifaddr *ifp)
2717 struct net_device * dev = ifp->idev->dev;
2720 * Configure the address for reception. Now it is valid.
2723 ipv6_ifa_notify(RTM_NEWADDR, ifp);
2725 /* If added prefix is link local and forwarding is off,
2726 start sending router solicitations.
2729 if (ifp->idev->cnf.forwarding == 0 &&
2730 ifp->idev->cnf.rtr_solicits > 0 &&
2731 (dev->flags&IFF_LOOPBACK) == 0 &&
2732 (ipv6_addr_type(&ifp->addr) & IPV6_ADDR_LINKLOCAL)) {
2733 struct in6_addr all_routers;
2735 ipv6_addr_all_routers(&all_routers);
2738 * If a host as already performed a random delay
2739 * [...] as part of DAD [...] there is no need
2740 * to delay again before sending the first RS
2742 ndisc_send_rs(ifp->idev->dev, &ifp->addr, &all_routers);
2744 spin_lock_bh(&ifp->lock);
2745 ifp->probes = 1;
2746 ifp->idev->if_flags |= IF_RS_SENT;
2747 addrconf_mod_timer(ifp, AC_RS, ifp->idev->cnf.rtr_solicit_interval);
2748 spin_unlock_bh(&ifp->lock);
2752 static void addrconf_dad_run(struct inet6_dev *idev) {
2753 struct inet6_ifaddr *ifp;
2755 read_lock_bh(&idev->lock);
2756 for (ifp = idev->addr_list; ifp; ifp = ifp->if_next) {
2757 spin_lock_bh(&ifp->lock);
2758 if (!(ifp->flags & IFA_F_TENTATIVE)) {
2759 spin_unlock_bh(&ifp->lock);
2760 continue;
2762 spin_unlock_bh(&ifp->lock);
2763 addrconf_dad_kick(ifp);
2765 read_unlock_bh(&idev->lock);
2768 #ifdef CONFIG_PROC_FS
2769 struct if6_iter_state {
2770 struct seq_net_private p;
2771 int bucket;
2774 static struct inet6_ifaddr *if6_get_first(struct seq_file *seq)
2776 struct inet6_ifaddr *ifa = NULL;
2777 struct if6_iter_state *state = seq->private;
2778 struct net *net = state->p.net;
2780 for (state->bucket = 0; state->bucket < IN6_ADDR_HSIZE; ++state->bucket) {
2781 ifa = inet6_addr_lst[state->bucket];
2783 while (ifa && ifa->idev->dev->nd_net != net)
2784 ifa = ifa->lst_next;
2785 if (ifa)
2786 break;
2788 return ifa;
2791 static struct inet6_ifaddr *if6_get_next(struct seq_file *seq, struct inet6_ifaddr *ifa)
2793 struct if6_iter_state *state = seq->private;
2794 struct net *net = state->p.net;
2796 ifa = ifa->lst_next;
2797 try_again:
2798 if (ifa) {
2799 if (ifa->idev->dev->nd_net != net) {
2800 ifa = ifa->lst_next;
2801 goto try_again;
2805 if (!ifa && ++state->bucket < IN6_ADDR_HSIZE) {
2806 ifa = inet6_addr_lst[state->bucket];
2807 goto try_again;
2810 return ifa;
2813 static struct inet6_ifaddr *if6_get_idx(struct seq_file *seq, loff_t pos)
2815 struct inet6_ifaddr *ifa = if6_get_first(seq);
2817 if (ifa)
2818 while(pos && (ifa = if6_get_next(seq, ifa)) != NULL)
2819 --pos;
2820 return pos ? NULL : ifa;
2823 static void *if6_seq_start(struct seq_file *seq, loff_t *pos)
2824 __acquires(addrconf_hash_lock)
2826 read_lock_bh(&addrconf_hash_lock);
2827 return if6_get_idx(seq, *pos);
2830 static void *if6_seq_next(struct seq_file *seq, void *v, loff_t *pos)
2832 struct inet6_ifaddr *ifa;
2834 ifa = if6_get_next(seq, v);
2835 ++*pos;
2836 return ifa;
2839 static void if6_seq_stop(struct seq_file *seq, void *v)
2840 __releases(addrconf_hash_lock)
2842 read_unlock_bh(&addrconf_hash_lock);
2845 static int if6_seq_show(struct seq_file *seq, void *v)
2847 struct inet6_ifaddr *ifp = (struct inet6_ifaddr *)v;
2848 seq_printf(seq,
2849 NIP6_SEQFMT " %02x %02x %02x %02x %8s\n",
2850 NIP6(ifp->addr),
2851 ifp->idev->dev->ifindex,
2852 ifp->prefix_len,
2853 ifp->scope,
2854 ifp->flags,
2855 ifp->idev->dev->name);
2856 return 0;
2859 static const struct seq_operations if6_seq_ops = {
2860 .start = if6_seq_start,
2861 .next = if6_seq_next,
2862 .show = if6_seq_show,
2863 .stop = if6_seq_stop,
2866 static int if6_seq_open(struct inode *inode, struct file *file)
2868 return seq_open_net(inode, file, &if6_seq_ops,
2869 sizeof(struct if6_iter_state));
2872 static const struct file_operations if6_fops = {
2873 .owner = THIS_MODULE,
2874 .open = if6_seq_open,
2875 .read = seq_read,
2876 .llseek = seq_lseek,
2877 .release = seq_release_net,
2880 static int if6_proc_net_init(struct net *net)
2882 if (!proc_net_fops_create(net, "if_inet6", S_IRUGO, &if6_fops))
2883 return -ENOMEM;
2884 return 0;
2887 static void if6_proc_net_exit(struct net *net)
2889 proc_net_remove(net, "if_inet6");
2892 static struct pernet_operations if6_proc_net_ops = {
2893 .init = if6_proc_net_init,
2894 .exit = if6_proc_net_exit,
2897 int __init if6_proc_init(void)
2899 return register_pernet_subsys(&if6_proc_net_ops);
2902 void if6_proc_exit(void)
2904 unregister_pernet_subsys(&if6_proc_net_ops);
2906 #endif /* CONFIG_PROC_FS */
2908 #if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
2909 /* Check if address is a home address configured on any interface. */
2910 int ipv6_chk_home_addr(struct net *net, struct in6_addr *addr)
2912 int ret = 0;
2913 struct inet6_ifaddr * ifp;
2914 u8 hash = ipv6_addr_hash(addr);
2915 read_lock_bh(&addrconf_hash_lock);
2916 for (ifp = inet6_addr_lst[hash]; ifp; ifp = ifp->lst_next) {
2917 if (ifp->idev->dev->nd_net != net)
2918 continue;
2919 if (ipv6_addr_cmp(&ifp->addr, addr) == 0 &&
2920 (ifp->flags & IFA_F_HOMEADDRESS)) {
2921 ret = 1;
2922 break;
2925 read_unlock_bh(&addrconf_hash_lock);
2926 return ret;
2928 #endif
2931 * Periodic address status verification
2934 static void addrconf_verify(unsigned long foo)
2936 struct inet6_ifaddr *ifp;
2937 unsigned long now, next;
2938 int i;
2940 spin_lock_bh(&addrconf_verify_lock);
2941 now = jiffies;
2942 next = now + ADDR_CHECK_FREQUENCY;
2944 del_timer(&addr_chk_timer);
2946 for (i=0; i < IN6_ADDR_HSIZE; i++) {
2948 restart:
2949 read_lock(&addrconf_hash_lock);
2950 for (ifp=inet6_addr_lst[i]; ifp; ifp=ifp->lst_next) {
2951 unsigned long age;
2952 #ifdef CONFIG_IPV6_PRIVACY
2953 unsigned long regen_advance;
2954 #endif
2956 if (ifp->flags & IFA_F_PERMANENT)
2957 continue;
2959 spin_lock(&ifp->lock);
2960 age = (now - ifp->tstamp) / HZ;
2962 #ifdef CONFIG_IPV6_PRIVACY
2963 regen_advance = ifp->idev->cnf.regen_max_retry *
2964 ifp->idev->cnf.dad_transmits *
2965 ifp->idev->nd_parms->retrans_time / HZ;
2966 #endif
2968 if (ifp->valid_lft != INFINITY_LIFE_TIME &&
2969 age >= ifp->valid_lft) {
2970 spin_unlock(&ifp->lock);
2971 in6_ifa_hold(ifp);
2972 read_unlock(&addrconf_hash_lock);
2973 ipv6_del_addr(ifp);
2974 goto restart;
2975 } else if (ifp->prefered_lft == INFINITY_LIFE_TIME) {
2976 spin_unlock(&ifp->lock);
2977 continue;
2978 } else if (age >= ifp->prefered_lft) {
2979 /* jiffies - ifp->tsamp > age >= ifp->prefered_lft */
2980 int deprecate = 0;
2982 if (!(ifp->flags&IFA_F_DEPRECATED)) {
2983 deprecate = 1;
2984 ifp->flags |= IFA_F_DEPRECATED;
2987 if (time_before(ifp->tstamp + ifp->valid_lft * HZ, next))
2988 next = ifp->tstamp + ifp->valid_lft * HZ;
2990 spin_unlock(&ifp->lock);
2992 if (deprecate) {
2993 in6_ifa_hold(ifp);
2994 read_unlock(&addrconf_hash_lock);
2996 ipv6_ifa_notify(0, ifp);
2997 in6_ifa_put(ifp);
2998 goto restart;
3000 #ifdef CONFIG_IPV6_PRIVACY
3001 } else if ((ifp->flags&IFA_F_TEMPORARY) &&
3002 !(ifp->flags&IFA_F_TENTATIVE)) {
3003 if (age >= ifp->prefered_lft - regen_advance) {
3004 struct inet6_ifaddr *ifpub = ifp->ifpub;
3005 if (time_before(ifp->tstamp + ifp->prefered_lft * HZ, next))
3006 next = ifp->tstamp + ifp->prefered_lft * HZ;
3007 if (!ifp->regen_count && ifpub) {
3008 ifp->regen_count++;
3009 in6_ifa_hold(ifp);
3010 in6_ifa_hold(ifpub);
3011 spin_unlock(&ifp->lock);
3012 read_unlock(&addrconf_hash_lock);
3013 spin_lock(&ifpub->lock);
3014 ifpub->regen_count = 0;
3015 spin_unlock(&ifpub->lock);
3016 ipv6_create_tempaddr(ifpub, ifp);
3017 in6_ifa_put(ifpub);
3018 in6_ifa_put(ifp);
3019 goto restart;
3021 } else if (time_before(ifp->tstamp + ifp->prefered_lft * HZ - regen_advance * HZ, next))
3022 next = ifp->tstamp + ifp->prefered_lft * HZ - regen_advance * HZ;
3023 spin_unlock(&ifp->lock);
3024 #endif
3025 } else {
3026 /* ifp->prefered_lft <= ifp->valid_lft */
3027 if (time_before(ifp->tstamp + ifp->prefered_lft * HZ, next))
3028 next = ifp->tstamp + ifp->prefered_lft * HZ;
3029 spin_unlock(&ifp->lock);
3032 read_unlock(&addrconf_hash_lock);
3035 addr_chk_timer.expires = time_before(next, jiffies + HZ) ? jiffies + HZ : next;
3036 add_timer(&addr_chk_timer);
3037 spin_unlock_bh(&addrconf_verify_lock);
3040 static struct in6_addr *extract_addr(struct nlattr *addr, struct nlattr *local)
3042 struct in6_addr *pfx = NULL;
3044 if (addr)
3045 pfx = nla_data(addr);
3047 if (local) {
3048 if (pfx && nla_memcmp(local, pfx, sizeof(*pfx)))
3049 pfx = NULL;
3050 else
3051 pfx = nla_data(local);
3054 return pfx;
3057 static const struct nla_policy ifa_ipv6_policy[IFA_MAX+1] = {
3058 [IFA_ADDRESS] = { .len = sizeof(struct in6_addr) },
3059 [IFA_LOCAL] = { .len = sizeof(struct in6_addr) },
3060 [IFA_CACHEINFO] = { .len = sizeof(struct ifa_cacheinfo) },
3063 static int
3064 inet6_rtm_deladdr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
3066 struct net *net = skb->sk->sk_net;
3067 struct ifaddrmsg *ifm;
3068 struct nlattr *tb[IFA_MAX+1];
3069 struct in6_addr *pfx;
3070 int err;
3072 if (net != &init_net)
3073 return -EINVAL;
3075 err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv6_policy);
3076 if (err < 0)
3077 return err;
3079 ifm = nlmsg_data(nlh);
3080 pfx = extract_addr(tb[IFA_ADDRESS], tb[IFA_LOCAL]);
3081 if (pfx == NULL)
3082 return -EINVAL;
3084 return inet6_addr_del(ifm->ifa_index, pfx, ifm->ifa_prefixlen);
3087 static int inet6_addr_modify(struct inet6_ifaddr *ifp, u8 ifa_flags,
3088 u32 prefered_lft, u32 valid_lft)
3090 u32 flags = RTF_EXPIRES;
3092 if (!valid_lft || (prefered_lft > valid_lft))
3093 return -EINVAL;
3095 if (valid_lft == INFINITY_LIFE_TIME) {
3096 ifa_flags |= IFA_F_PERMANENT;
3097 flags = 0;
3098 } else if (valid_lft >= 0x7FFFFFFF/HZ)
3099 valid_lft = 0x7FFFFFFF/HZ;
3101 if (prefered_lft == 0)
3102 ifa_flags |= IFA_F_DEPRECATED;
3103 else if ((prefered_lft >= 0x7FFFFFFF/HZ) &&
3104 (prefered_lft != INFINITY_LIFE_TIME))
3105 prefered_lft = 0x7FFFFFFF/HZ;
3107 spin_lock_bh(&ifp->lock);
3108 ifp->flags = (ifp->flags & ~(IFA_F_DEPRECATED | IFA_F_PERMANENT | IFA_F_NODAD | IFA_F_HOMEADDRESS)) | ifa_flags;
3109 ifp->tstamp = jiffies;
3110 ifp->valid_lft = valid_lft;
3111 ifp->prefered_lft = prefered_lft;
3113 spin_unlock_bh(&ifp->lock);
3114 if (!(ifp->flags&IFA_F_TENTATIVE))
3115 ipv6_ifa_notify(0, ifp);
3117 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, ifp->idev->dev,
3118 jiffies_to_clock_t(valid_lft * HZ), flags);
3119 addrconf_verify(0);
3121 return 0;
3124 static int
3125 inet6_rtm_newaddr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
3127 struct net *net = skb->sk->sk_net;
3128 struct ifaddrmsg *ifm;
3129 struct nlattr *tb[IFA_MAX+1];
3130 struct in6_addr *pfx;
3131 struct inet6_ifaddr *ifa;
3132 struct net_device *dev;
3133 u32 valid_lft = INFINITY_LIFE_TIME, preferred_lft = INFINITY_LIFE_TIME;
3134 u8 ifa_flags;
3135 int err;
3137 if (net != &init_net)
3138 return -EINVAL;
3140 err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv6_policy);
3141 if (err < 0)
3142 return err;
3144 ifm = nlmsg_data(nlh);
3145 pfx = extract_addr(tb[IFA_ADDRESS], tb[IFA_LOCAL]);
3146 if (pfx == NULL)
3147 return -EINVAL;
3149 if (tb[IFA_CACHEINFO]) {
3150 struct ifa_cacheinfo *ci;
3152 ci = nla_data(tb[IFA_CACHEINFO]);
3153 valid_lft = ci->ifa_valid;
3154 preferred_lft = ci->ifa_prefered;
3155 } else {
3156 preferred_lft = INFINITY_LIFE_TIME;
3157 valid_lft = INFINITY_LIFE_TIME;
3160 dev = __dev_get_by_index(&init_net, ifm->ifa_index);
3161 if (dev == NULL)
3162 return -ENODEV;
3164 /* We ignore other flags so far. */
3165 ifa_flags = ifm->ifa_flags & (IFA_F_NODAD | IFA_F_HOMEADDRESS);
3167 ifa = ipv6_get_ifaddr(net, pfx, dev, 1);
3168 if (ifa == NULL) {
3170 * It would be best to check for !NLM_F_CREATE here but
3171 * userspace alreay relies on not having to provide this.
3173 return inet6_addr_add(ifm->ifa_index, pfx, ifm->ifa_prefixlen,
3174 ifa_flags, preferred_lft, valid_lft);
3177 if (nlh->nlmsg_flags & NLM_F_EXCL ||
3178 !(nlh->nlmsg_flags & NLM_F_REPLACE))
3179 err = -EEXIST;
3180 else
3181 err = inet6_addr_modify(ifa, ifa_flags, preferred_lft, valid_lft);
3183 in6_ifa_put(ifa);
3185 return err;
3188 static void put_ifaddrmsg(struct nlmsghdr *nlh, u8 prefixlen, u8 flags,
3189 u8 scope, int ifindex)
3191 struct ifaddrmsg *ifm;
3193 ifm = nlmsg_data(nlh);
3194 ifm->ifa_family = AF_INET6;
3195 ifm->ifa_prefixlen = prefixlen;
3196 ifm->ifa_flags = flags;
3197 ifm->ifa_scope = scope;
3198 ifm->ifa_index = ifindex;
3201 static int put_cacheinfo(struct sk_buff *skb, unsigned long cstamp,
3202 unsigned long tstamp, u32 preferred, u32 valid)
3204 struct ifa_cacheinfo ci;
3206 ci.cstamp = (u32)(TIME_DELTA(cstamp, INITIAL_JIFFIES) / HZ * 100
3207 + TIME_DELTA(cstamp, INITIAL_JIFFIES) % HZ * 100 / HZ);
3208 ci.tstamp = (u32)(TIME_DELTA(tstamp, INITIAL_JIFFIES) / HZ * 100
3209 + TIME_DELTA(tstamp, INITIAL_JIFFIES) % HZ * 100 / HZ);
3210 ci.ifa_prefered = preferred;
3211 ci.ifa_valid = valid;
3213 return nla_put(skb, IFA_CACHEINFO, sizeof(ci), &ci);
3216 static inline int rt_scope(int ifa_scope)
3218 if (ifa_scope & IFA_HOST)
3219 return RT_SCOPE_HOST;
3220 else if (ifa_scope & IFA_LINK)
3221 return RT_SCOPE_LINK;
3222 else if (ifa_scope & IFA_SITE)
3223 return RT_SCOPE_SITE;
3224 else
3225 return RT_SCOPE_UNIVERSE;
3228 static inline int inet6_ifaddr_msgsize(void)
3230 return NLMSG_ALIGN(sizeof(struct ifaddrmsg))
3231 + nla_total_size(16) /* IFA_ADDRESS */
3232 + nla_total_size(sizeof(struct ifa_cacheinfo));
3235 static int inet6_fill_ifaddr(struct sk_buff *skb, struct inet6_ifaddr *ifa,
3236 u32 pid, u32 seq, int event, unsigned int flags)
3238 struct nlmsghdr *nlh;
3239 u32 preferred, valid;
3241 nlh = nlmsg_put(skb, pid, seq, event, sizeof(struct ifaddrmsg), flags);
3242 if (nlh == NULL)
3243 return -EMSGSIZE;
3245 put_ifaddrmsg(nlh, ifa->prefix_len, ifa->flags, rt_scope(ifa->scope),
3246 ifa->idev->dev->ifindex);
3248 if (!(ifa->flags&IFA_F_PERMANENT)) {
3249 preferred = ifa->prefered_lft;
3250 valid = ifa->valid_lft;
3251 if (preferred != INFINITY_LIFE_TIME) {
3252 long tval = (jiffies - ifa->tstamp)/HZ;
3253 preferred -= tval;
3254 if (valid != INFINITY_LIFE_TIME)
3255 valid -= tval;
3257 } else {
3258 preferred = INFINITY_LIFE_TIME;
3259 valid = INFINITY_LIFE_TIME;
3262 if (nla_put(skb, IFA_ADDRESS, 16, &ifa->addr) < 0 ||
3263 put_cacheinfo(skb, ifa->cstamp, ifa->tstamp, preferred, valid) < 0) {
3264 nlmsg_cancel(skb, nlh);
3265 return -EMSGSIZE;
3268 return nlmsg_end(skb, nlh);
3271 static int inet6_fill_ifmcaddr(struct sk_buff *skb, struct ifmcaddr6 *ifmca,
3272 u32 pid, u32 seq, int event, u16 flags)
3274 struct nlmsghdr *nlh;
3275 u8 scope = RT_SCOPE_UNIVERSE;
3276 int ifindex = ifmca->idev->dev->ifindex;
3278 if (ipv6_addr_scope(&ifmca->mca_addr) & IFA_SITE)
3279 scope = RT_SCOPE_SITE;
3281 nlh = nlmsg_put(skb, pid, seq, event, sizeof(struct ifaddrmsg), flags);
3282 if (nlh == NULL)
3283 return -EMSGSIZE;
3285 put_ifaddrmsg(nlh, 128, IFA_F_PERMANENT, scope, ifindex);
3286 if (nla_put(skb, IFA_MULTICAST, 16, &ifmca->mca_addr) < 0 ||
3287 put_cacheinfo(skb, ifmca->mca_cstamp, ifmca->mca_tstamp,
3288 INFINITY_LIFE_TIME, INFINITY_LIFE_TIME) < 0) {
3289 nlmsg_cancel(skb, nlh);
3290 return -EMSGSIZE;
3293 return nlmsg_end(skb, nlh);
3296 static int inet6_fill_ifacaddr(struct sk_buff *skb, struct ifacaddr6 *ifaca,
3297 u32 pid, u32 seq, int event, unsigned int flags)
3299 struct nlmsghdr *nlh;
3300 u8 scope = RT_SCOPE_UNIVERSE;
3301 int ifindex = ifaca->aca_idev->dev->ifindex;
3303 if (ipv6_addr_scope(&ifaca->aca_addr) & IFA_SITE)
3304 scope = RT_SCOPE_SITE;
3306 nlh = nlmsg_put(skb, pid, seq, event, sizeof(struct ifaddrmsg), flags);
3307 if (nlh == NULL)
3308 return -EMSGSIZE;
3310 put_ifaddrmsg(nlh, 128, IFA_F_PERMANENT, scope, ifindex);
3311 if (nla_put(skb, IFA_ANYCAST, 16, &ifaca->aca_addr) < 0 ||
3312 put_cacheinfo(skb, ifaca->aca_cstamp, ifaca->aca_tstamp,
3313 INFINITY_LIFE_TIME, INFINITY_LIFE_TIME) < 0) {
3314 nlmsg_cancel(skb, nlh);
3315 return -EMSGSIZE;
3318 return nlmsg_end(skb, nlh);
3321 enum addr_type_t
3323 UNICAST_ADDR,
3324 MULTICAST_ADDR,
3325 ANYCAST_ADDR,
3328 static int inet6_dump_addr(struct sk_buff *skb, struct netlink_callback *cb,
3329 enum addr_type_t type)
3331 int idx, ip_idx;
3332 int s_idx, s_ip_idx;
3333 int err = 1;
3334 struct net_device *dev;
3335 struct inet6_dev *idev = NULL;
3336 struct inet6_ifaddr *ifa;
3337 struct ifmcaddr6 *ifmca;
3338 struct ifacaddr6 *ifaca;
3340 s_idx = cb->args[0];
3341 s_ip_idx = ip_idx = cb->args[1];
3343 idx = 0;
3344 for_each_netdev(&init_net, dev) {
3345 if (idx < s_idx)
3346 goto cont;
3347 if (idx > s_idx)
3348 s_ip_idx = 0;
3349 ip_idx = 0;
3350 if ((idev = in6_dev_get(dev)) == NULL)
3351 goto cont;
3352 read_lock_bh(&idev->lock);
3353 switch (type) {
3354 case UNICAST_ADDR:
3355 /* unicast address incl. temp addr */
3356 for (ifa = idev->addr_list; ifa;
3357 ifa = ifa->if_next, ip_idx++) {
3358 if (ip_idx < s_ip_idx)
3359 continue;
3360 err = inet6_fill_ifaddr(skb, ifa,
3361 NETLINK_CB(cb->skb).pid,
3362 cb->nlh->nlmsg_seq,
3363 RTM_NEWADDR,
3364 NLM_F_MULTI);
3366 break;
3367 case MULTICAST_ADDR:
3368 /* multicast address */
3369 for (ifmca = idev->mc_list; ifmca;
3370 ifmca = ifmca->next, ip_idx++) {
3371 if (ip_idx < s_ip_idx)
3372 continue;
3373 err = inet6_fill_ifmcaddr(skb, ifmca,
3374 NETLINK_CB(cb->skb).pid,
3375 cb->nlh->nlmsg_seq,
3376 RTM_GETMULTICAST,
3377 NLM_F_MULTI);
3379 break;
3380 case ANYCAST_ADDR:
3381 /* anycast address */
3382 for (ifaca = idev->ac_list; ifaca;
3383 ifaca = ifaca->aca_next, ip_idx++) {
3384 if (ip_idx < s_ip_idx)
3385 continue;
3386 err = inet6_fill_ifacaddr(skb, ifaca,
3387 NETLINK_CB(cb->skb).pid,
3388 cb->nlh->nlmsg_seq,
3389 RTM_GETANYCAST,
3390 NLM_F_MULTI);
3392 break;
3393 default:
3394 break;
3396 read_unlock_bh(&idev->lock);
3397 in6_dev_put(idev);
3399 if (err <= 0)
3400 break;
3401 cont:
3402 idx++;
3404 cb->args[0] = idx;
3405 cb->args[1] = ip_idx;
3406 return skb->len;
3409 static int inet6_dump_ifaddr(struct sk_buff *skb, struct netlink_callback *cb)
3411 struct net *net = skb->sk->sk_net;
3412 enum addr_type_t type = UNICAST_ADDR;
3414 if (net != &init_net)
3415 return 0;
3417 return inet6_dump_addr(skb, cb, type);
3420 static int inet6_dump_ifmcaddr(struct sk_buff *skb, struct netlink_callback *cb)
3422 struct net *net = skb->sk->sk_net;
3423 enum addr_type_t type = MULTICAST_ADDR;
3425 if (net != &init_net)
3426 return 0;
3428 return inet6_dump_addr(skb, cb, type);
3432 static int inet6_dump_ifacaddr(struct sk_buff *skb, struct netlink_callback *cb)
3434 struct net *net = skb->sk->sk_net;
3435 enum addr_type_t type = ANYCAST_ADDR;
3437 if (net != &init_net)
3438 return 0;
3440 return inet6_dump_addr(skb, cb, type);
3443 static int inet6_rtm_getaddr(struct sk_buff *in_skb, struct nlmsghdr* nlh,
3444 void *arg)
3446 struct net *net = in_skb->sk->sk_net;
3447 struct ifaddrmsg *ifm;
3448 struct nlattr *tb[IFA_MAX+1];
3449 struct in6_addr *addr = NULL;
3450 struct net_device *dev = NULL;
3451 struct inet6_ifaddr *ifa;
3452 struct sk_buff *skb;
3453 int err;
3455 if (net != &init_net)
3456 return -EINVAL;
3458 err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv6_policy);
3459 if (err < 0)
3460 goto errout;
3462 addr = extract_addr(tb[IFA_ADDRESS], tb[IFA_LOCAL]);
3463 if (addr == NULL) {
3464 err = -EINVAL;
3465 goto errout;
3468 ifm = nlmsg_data(nlh);
3469 if (ifm->ifa_index)
3470 dev = __dev_get_by_index(&init_net, ifm->ifa_index);
3472 if ((ifa = ipv6_get_ifaddr(net, addr, dev, 1)) == NULL) {
3473 err = -EADDRNOTAVAIL;
3474 goto errout;
3477 if ((skb = nlmsg_new(inet6_ifaddr_msgsize(), GFP_KERNEL)) == NULL) {
3478 err = -ENOBUFS;
3479 goto errout_ifa;
3482 err = inet6_fill_ifaddr(skb, ifa, NETLINK_CB(in_skb).pid,
3483 nlh->nlmsg_seq, RTM_NEWADDR, 0);
3484 if (err < 0) {
3485 /* -EMSGSIZE implies BUG in inet6_ifaddr_msgsize() */
3486 WARN_ON(err == -EMSGSIZE);
3487 kfree_skb(skb);
3488 goto errout_ifa;
3490 err = rtnl_unicast(skb, &init_net, NETLINK_CB(in_skb).pid);
3491 errout_ifa:
3492 in6_ifa_put(ifa);
3493 errout:
3494 return err;
3497 static void inet6_ifa_notify(int event, struct inet6_ifaddr *ifa)
3499 struct sk_buff *skb;
3500 int err = -ENOBUFS;
3502 skb = nlmsg_new(inet6_ifaddr_msgsize(), GFP_ATOMIC);
3503 if (skb == NULL)
3504 goto errout;
3506 err = inet6_fill_ifaddr(skb, ifa, 0, 0, event, 0);
3507 if (err < 0) {
3508 /* -EMSGSIZE implies BUG in inet6_ifaddr_msgsize() */
3509 WARN_ON(err == -EMSGSIZE);
3510 kfree_skb(skb);
3511 goto errout;
3513 err = rtnl_notify(skb, &init_net, 0, RTNLGRP_IPV6_IFADDR, NULL, GFP_ATOMIC);
3514 errout:
3515 if (err < 0)
3516 rtnl_set_sk_err(&init_net, RTNLGRP_IPV6_IFADDR, err);
3519 static inline void ipv6_store_devconf(struct ipv6_devconf *cnf,
3520 __s32 *array, int bytes)
3522 BUG_ON(bytes < (DEVCONF_MAX * 4));
3524 memset(array, 0, bytes);
3525 array[DEVCONF_FORWARDING] = cnf->forwarding;
3526 array[DEVCONF_HOPLIMIT] = cnf->hop_limit;
3527 array[DEVCONF_MTU6] = cnf->mtu6;
3528 array[DEVCONF_ACCEPT_RA] = cnf->accept_ra;
3529 array[DEVCONF_ACCEPT_REDIRECTS] = cnf->accept_redirects;
3530 array[DEVCONF_AUTOCONF] = cnf->autoconf;
3531 array[DEVCONF_DAD_TRANSMITS] = cnf->dad_transmits;
3532 array[DEVCONF_RTR_SOLICITS] = cnf->rtr_solicits;
3533 array[DEVCONF_RTR_SOLICIT_INTERVAL] = cnf->rtr_solicit_interval;
3534 array[DEVCONF_RTR_SOLICIT_DELAY] = cnf->rtr_solicit_delay;
3535 array[DEVCONF_FORCE_MLD_VERSION] = cnf->force_mld_version;
3536 #ifdef CONFIG_IPV6_PRIVACY
3537 array[DEVCONF_USE_TEMPADDR] = cnf->use_tempaddr;
3538 array[DEVCONF_TEMP_VALID_LFT] = cnf->temp_valid_lft;
3539 array[DEVCONF_TEMP_PREFERED_LFT] = cnf->temp_prefered_lft;
3540 array[DEVCONF_REGEN_MAX_RETRY] = cnf->regen_max_retry;
3541 array[DEVCONF_MAX_DESYNC_FACTOR] = cnf->max_desync_factor;
3542 #endif
3543 array[DEVCONF_MAX_ADDRESSES] = cnf->max_addresses;
3544 array[DEVCONF_ACCEPT_RA_DEFRTR] = cnf->accept_ra_defrtr;
3545 array[DEVCONF_ACCEPT_RA_PINFO] = cnf->accept_ra_pinfo;
3546 #ifdef CONFIG_IPV6_ROUTER_PREF
3547 array[DEVCONF_ACCEPT_RA_RTR_PREF] = cnf->accept_ra_rtr_pref;
3548 array[DEVCONF_RTR_PROBE_INTERVAL] = cnf->rtr_probe_interval;
3549 #ifdef CONFIG_IPV6_ROUTE_INFO
3550 array[DEVCONF_ACCEPT_RA_RT_INFO_MAX_PLEN] = cnf->accept_ra_rt_info_max_plen;
3551 #endif
3552 #endif
3553 array[DEVCONF_PROXY_NDP] = cnf->proxy_ndp;
3554 array[DEVCONF_ACCEPT_SOURCE_ROUTE] = cnf->accept_source_route;
3555 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
3556 array[DEVCONF_OPTIMISTIC_DAD] = cnf->optimistic_dad;
3557 #endif
3560 static inline size_t inet6_if_nlmsg_size(void)
3562 return NLMSG_ALIGN(sizeof(struct ifinfomsg))
3563 + nla_total_size(IFNAMSIZ) /* IFLA_IFNAME */
3564 + nla_total_size(MAX_ADDR_LEN) /* IFLA_ADDRESS */
3565 + nla_total_size(4) /* IFLA_MTU */
3566 + nla_total_size(4) /* IFLA_LINK */
3567 + nla_total_size( /* IFLA_PROTINFO */
3568 nla_total_size(4) /* IFLA_INET6_FLAGS */
3569 + nla_total_size(sizeof(struct ifla_cacheinfo))
3570 + nla_total_size(DEVCONF_MAX * 4) /* IFLA_INET6_CONF */
3571 + nla_total_size(IPSTATS_MIB_MAX * 8) /* IFLA_INET6_STATS */
3572 + nla_total_size(ICMP6_MIB_MAX * 8) /* IFLA_INET6_ICMP6STATS */
3576 static inline void __snmp6_fill_stats(u64 *stats, void **mib, int items,
3577 int bytes)
3579 int i;
3580 int pad = bytes - sizeof(u64) * items;
3581 BUG_ON(pad < 0);
3583 /* Use put_unaligned() because stats may not be aligned for u64. */
3584 put_unaligned(items, &stats[0]);
3585 for (i = 1; i < items; i++)
3586 put_unaligned(snmp_fold_field(mib, i), &stats[i]);
3588 memset(&stats[items], 0, pad);
3591 static void snmp6_fill_stats(u64 *stats, struct inet6_dev *idev, int attrtype,
3592 int bytes)
3594 switch(attrtype) {
3595 case IFLA_INET6_STATS:
3596 __snmp6_fill_stats(stats, (void **)idev->stats.ipv6, IPSTATS_MIB_MAX, bytes);
3597 break;
3598 case IFLA_INET6_ICMP6STATS:
3599 __snmp6_fill_stats(stats, (void **)idev->stats.icmpv6, ICMP6_MIB_MAX, bytes);
3600 break;
3604 static int inet6_fill_ifinfo(struct sk_buff *skb, struct inet6_dev *idev,
3605 u32 pid, u32 seq, int event, unsigned int flags)
3607 struct net_device *dev = idev->dev;
3608 struct nlattr *nla;
3609 struct ifinfomsg *hdr;
3610 struct nlmsghdr *nlh;
3611 void *protoinfo;
3612 struct ifla_cacheinfo ci;
3614 nlh = nlmsg_put(skb, pid, seq, event, sizeof(*hdr), flags);
3615 if (nlh == NULL)
3616 return -EMSGSIZE;
3618 hdr = nlmsg_data(nlh);
3619 hdr->ifi_family = AF_INET6;
3620 hdr->__ifi_pad = 0;
3621 hdr->ifi_type = dev->type;
3622 hdr->ifi_index = dev->ifindex;
3623 hdr->ifi_flags = dev_get_flags(dev);
3624 hdr->ifi_change = 0;
3626 NLA_PUT_STRING(skb, IFLA_IFNAME, dev->name);
3628 if (dev->addr_len)
3629 NLA_PUT(skb, IFLA_ADDRESS, dev->addr_len, dev->dev_addr);
3631 NLA_PUT_U32(skb, IFLA_MTU, dev->mtu);
3632 if (dev->ifindex != dev->iflink)
3633 NLA_PUT_U32(skb, IFLA_LINK, dev->iflink);
3635 protoinfo = nla_nest_start(skb, IFLA_PROTINFO);
3636 if (protoinfo == NULL)
3637 goto nla_put_failure;
3639 NLA_PUT_U32(skb, IFLA_INET6_FLAGS, idev->if_flags);
3641 ci.max_reasm_len = IPV6_MAXPLEN;
3642 ci.tstamp = (__u32)(TIME_DELTA(idev->tstamp, INITIAL_JIFFIES) / HZ * 100
3643 + TIME_DELTA(idev->tstamp, INITIAL_JIFFIES) % HZ * 100 / HZ);
3644 ci.reachable_time = idev->nd_parms->reachable_time;
3645 ci.retrans_time = idev->nd_parms->retrans_time;
3646 NLA_PUT(skb, IFLA_INET6_CACHEINFO, sizeof(ci), &ci);
3648 nla = nla_reserve(skb, IFLA_INET6_CONF, DEVCONF_MAX * sizeof(s32));
3649 if (nla == NULL)
3650 goto nla_put_failure;
3651 ipv6_store_devconf(&idev->cnf, nla_data(nla), nla_len(nla));
3653 /* XXX - MC not implemented */
3655 nla = nla_reserve(skb, IFLA_INET6_STATS, IPSTATS_MIB_MAX * sizeof(u64));
3656 if (nla == NULL)
3657 goto nla_put_failure;
3658 snmp6_fill_stats(nla_data(nla), idev, IFLA_INET6_STATS, nla_len(nla));
3660 nla = nla_reserve(skb, IFLA_INET6_ICMP6STATS, ICMP6_MIB_MAX * sizeof(u64));
3661 if (nla == NULL)
3662 goto nla_put_failure;
3663 snmp6_fill_stats(nla_data(nla), idev, IFLA_INET6_ICMP6STATS, nla_len(nla));
3665 nla_nest_end(skb, protoinfo);
3666 return nlmsg_end(skb, nlh);
3668 nla_put_failure:
3669 nlmsg_cancel(skb, nlh);
3670 return -EMSGSIZE;
3673 static int inet6_dump_ifinfo(struct sk_buff *skb, struct netlink_callback *cb)
3675 struct net *net = skb->sk->sk_net;
3676 int idx, err;
3677 int s_idx = cb->args[0];
3678 struct net_device *dev;
3679 struct inet6_dev *idev;
3681 if (net != &init_net)
3682 return 0;
3684 read_lock(&dev_base_lock);
3685 idx = 0;
3686 for_each_netdev(&init_net, dev) {
3687 if (idx < s_idx)
3688 goto cont;
3689 if ((idev = in6_dev_get(dev)) == NULL)
3690 goto cont;
3691 err = inet6_fill_ifinfo(skb, idev, NETLINK_CB(cb->skb).pid,
3692 cb->nlh->nlmsg_seq, RTM_NEWLINK, NLM_F_MULTI);
3693 in6_dev_put(idev);
3694 if (err <= 0)
3695 break;
3696 cont:
3697 idx++;
3699 read_unlock(&dev_base_lock);
3700 cb->args[0] = idx;
3702 return skb->len;
3705 void inet6_ifinfo_notify(int event, struct inet6_dev *idev)
3707 struct sk_buff *skb;
3708 int err = -ENOBUFS;
3710 skb = nlmsg_new(inet6_if_nlmsg_size(), GFP_ATOMIC);
3711 if (skb == NULL)
3712 goto errout;
3714 err = inet6_fill_ifinfo(skb, idev, 0, 0, event, 0);
3715 if (err < 0) {
3716 /* -EMSGSIZE implies BUG in inet6_if_nlmsg_size() */
3717 WARN_ON(err == -EMSGSIZE);
3718 kfree_skb(skb);
3719 goto errout;
3721 err = rtnl_notify(skb, &init_net, 0, RTNLGRP_IPV6_IFADDR, NULL, GFP_ATOMIC);
3722 errout:
3723 if (err < 0)
3724 rtnl_set_sk_err(&init_net, RTNLGRP_IPV6_IFADDR, err);
3727 static inline size_t inet6_prefix_nlmsg_size(void)
3729 return NLMSG_ALIGN(sizeof(struct prefixmsg))
3730 + nla_total_size(sizeof(struct in6_addr))
3731 + nla_total_size(sizeof(struct prefix_cacheinfo));
3734 static int inet6_fill_prefix(struct sk_buff *skb, struct inet6_dev *idev,
3735 struct prefix_info *pinfo, u32 pid, u32 seq,
3736 int event, unsigned int flags)
3738 struct prefixmsg *pmsg;
3739 struct nlmsghdr *nlh;
3740 struct prefix_cacheinfo ci;
3742 nlh = nlmsg_put(skb, pid, seq, event, sizeof(*pmsg), flags);
3743 if (nlh == NULL)
3744 return -EMSGSIZE;
3746 pmsg = nlmsg_data(nlh);
3747 pmsg->prefix_family = AF_INET6;
3748 pmsg->prefix_pad1 = 0;
3749 pmsg->prefix_pad2 = 0;
3750 pmsg->prefix_ifindex = idev->dev->ifindex;
3751 pmsg->prefix_len = pinfo->prefix_len;
3752 pmsg->prefix_type = pinfo->type;
3753 pmsg->prefix_pad3 = 0;
3754 pmsg->prefix_flags = 0;
3755 if (pinfo->onlink)
3756 pmsg->prefix_flags |= IF_PREFIX_ONLINK;
3757 if (pinfo->autoconf)
3758 pmsg->prefix_flags |= IF_PREFIX_AUTOCONF;
3760 NLA_PUT(skb, PREFIX_ADDRESS, sizeof(pinfo->prefix), &pinfo->prefix);
3762 ci.preferred_time = ntohl(pinfo->prefered);
3763 ci.valid_time = ntohl(pinfo->valid);
3764 NLA_PUT(skb, PREFIX_CACHEINFO, sizeof(ci), &ci);
3766 return nlmsg_end(skb, nlh);
3768 nla_put_failure:
3769 nlmsg_cancel(skb, nlh);
3770 return -EMSGSIZE;
3773 static void inet6_prefix_notify(int event, struct inet6_dev *idev,
3774 struct prefix_info *pinfo)
3776 struct sk_buff *skb;
3777 int err = -ENOBUFS;
3779 skb = nlmsg_new(inet6_prefix_nlmsg_size(), GFP_ATOMIC);
3780 if (skb == NULL)
3781 goto errout;
3783 err = inet6_fill_prefix(skb, idev, pinfo, 0, 0, event, 0);
3784 if (err < 0) {
3785 /* -EMSGSIZE implies BUG in inet6_prefix_nlmsg_size() */
3786 WARN_ON(err == -EMSGSIZE);
3787 kfree_skb(skb);
3788 goto errout;
3790 err = rtnl_notify(skb, &init_net, 0, RTNLGRP_IPV6_PREFIX, NULL, GFP_ATOMIC);
3791 errout:
3792 if (err < 0)
3793 rtnl_set_sk_err(&init_net, RTNLGRP_IPV6_PREFIX, err);
3796 static void __ipv6_ifa_notify(int event, struct inet6_ifaddr *ifp)
3798 inet6_ifa_notify(event ? : RTM_NEWADDR, ifp);
3800 switch (event) {
3801 case RTM_NEWADDR:
3803 * If the address was optimistic
3804 * we inserted the route at the start of
3805 * our DAD process, so we don't need
3806 * to do it again
3808 if (!(ifp->rt->rt6i_node))
3809 ip6_ins_rt(ifp->rt);
3810 if (ifp->idev->cnf.forwarding)
3811 addrconf_join_anycast(ifp);
3812 break;
3813 case RTM_DELADDR:
3814 if (ifp->idev->cnf.forwarding)
3815 addrconf_leave_anycast(ifp);
3816 addrconf_leave_solict(ifp->idev, &ifp->addr);
3817 dst_hold(&ifp->rt->u.dst);
3818 if (ip6_del_rt(ifp->rt))
3819 dst_free(&ifp->rt->u.dst);
3820 break;
3824 static void ipv6_ifa_notify(int event, struct inet6_ifaddr *ifp)
3826 rcu_read_lock_bh();
3827 if (likely(ifp->idev->dead == 0))
3828 __ipv6_ifa_notify(event, ifp);
3829 rcu_read_unlock_bh();
3832 #ifdef CONFIG_SYSCTL
3834 static
3835 int addrconf_sysctl_forward(ctl_table *ctl, int write, struct file * filp,
3836 void __user *buffer, size_t *lenp, loff_t *ppos)
3838 int *valp = ctl->data;
3839 int val = *valp;
3840 int ret;
3842 ret = proc_dointvec(ctl, write, filp, buffer, lenp, ppos);
3844 if (write)
3845 addrconf_fixup_forwarding(ctl, valp, val);
3846 return ret;
3849 static int addrconf_sysctl_forward_strategy(ctl_table *table,
3850 int __user *name, int nlen,
3851 void __user *oldval,
3852 size_t __user *oldlenp,
3853 void __user *newval, size_t newlen)
3855 int *valp = table->data;
3856 int val = *valp;
3857 int new;
3859 if (!newval || !newlen)
3860 return 0;
3861 if (newlen != sizeof(int))
3862 return -EINVAL;
3863 if (get_user(new, (int __user *)newval))
3864 return -EFAULT;
3865 if (new == *valp)
3866 return 0;
3867 if (oldval && oldlenp) {
3868 size_t len;
3869 if (get_user(len, oldlenp))
3870 return -EFAULT;
3871 if (len) {
3872 if (len > table->maxlen)
3873 len = table->maxlen;
3874 if (copy_to_user(oldval, valp, len))
3875 return -EFAULT;
3876 if (put_user(len, oldlenp))
3877 return -EFAULT;
3881 *valp = new;
3882 addrconf_fixup_forwarding(table, valp, val);
3883 return 1;
3886 static struct addrconf_sysctl_table
3888 struct ctl_table_header *sysctl_header;
3889 ctl_table addrconf_vars[__NET_IPV6_MAX];
3890 char *dev_name;
3891 } addrconf_sysctl __read_mostly = {
3892 .sysctl_header = NULL,
3893 .addrconf_vars = {
3895 .ctl_name = NET_IPV6_FORWARDING,
3896 .procname = "forwarding",
3897 .data = &ipv6_devconf.forwarding,
3898 .maxlen = sizeof(int),
3899 .mode = 0644,
3900 .proc_handler = &addrconf_sysctl_forward,
3901 .strategy = &addrconf_sysctl_forward_strategy,
3904 .ctl_name = NET_IPV6_HOP_LIMIT,
3905 .procname = "hop_limit",
3906 .data = &ipv6_devconf.hop_limit,
3907 .maxlen = sizeof(int),
3908 .mode = 0644,
3909 .proc_handler = proc_dointvec,
3912 .ctl_name = NET_IPV6_MTU,
3913 .procname = "mtu",
3914 .data = &ipv6_devconf.mtu6,
3915 .maxlen = sizeof(int),
3916 .mode = 0644,
3917 .proc_handler = &proc_dointvec,
3920 .ctl_name = NET_IPV6_ACCEPT_RA,
3921 .procname = "accept_ra",
3922 .data = &ipv6_devconf.accept_ra,
3923 .maxlen = sizeof(int),
3924 .mode = 0644,
3925 .proc_handler = &proc_dointvec,
3928 .ctl_name = NET_IPV6_ACCEPT_REDIRECTS,
3929 .procname = "accept_redirects",
3930 .data = &ipv6_devconf.accept_redirects,
3931 .maxlen = sizeof(int),
3932 .mode = 0644,
3933 .proc_handler = &proc_dointvec,
3936 .ctl_name = NET_IPV6_AUTOCONF,
3937 .procname = "autoconf",
3938 .data = &ipv6_devconf.autoconf,
3939 .maxlen = sizeof(int),
3940 .mode = 0644,
3941 .proc_handler = &proc_dointvec,
3944 .ctl_name = NET_IPV6_DAD_TRANSMITS,
3945 .procname = "dad_transmits",
3946 .data = &ipv6_devconf.dad_transmits,
3947 .maxlen = sizeof(int),
3948 .mode = 0644,
3949 .proc_handler = &proc_dointvec,
3952 .ctl_name = NET_IPV6_RTR_SOLICITS,
3953 .procname = "router_solicitations",
3954 .data = &ipv6_devconf.rtr_solicits,
3955 .maxlen = sizeof(int),
3956 .mode = 0644,
3957 .proc_handler = &proc_dointvec,
3960 .ctl_name = NET_IPV6_RTR_SOLICIT_INTERVAL,
3961 .procname = "router_solicitation_interval",
3962 .data = &ipv6_devconf.rtr_solicit_interval,
3963 .maxlen = sizeof(int),
3964 .mode = 0644,
3965 .proc_handler = &proc_dointvec_jiffies,
3966 .strategy = &sysctl_jiffies,
3969 .ctl_name = NET_IPV6_RTR_SOLICIT_DELAY,
3970 .procname = "router_solicitation_delay",
3971 .data = &ipv6_devconf.rtr_solicit_delay,
3972 .maxlen = sizeof(int),
3973 .mode = 0644,
3974 .proc_handler = &proc_dointvec_jiffies,
3975 .strategy = &sysctl_jiffies,
3978 .ctl_name = NET_IPV6_FORCE_MLD_VERSION,
3979 .procname = "force_mld_version",
3980 .data = &ipv6_devconf.force_mld_version,
3981 .maxlen = sizeof(int),
3982 .mode = 0644,
3983 .proc_handler = &proc_dointvec,
3985 #ifdef CONFIG_IPV6_PRIVACY
3987 .ctl_name = NET_IPV6_USE_TEMPADDR,
3988 .procname = "use_tempaddr",
3989 .data = &ipv6_devconf.use_tempaddr,
3990 .maxlen = sizeof(int),
3991 .mode = 0644,
3992 .proc_handler = &proc_dointvec,
3995 .ctl_name = NET_IPV6_TEMP_VALID_LFT,
3996 .procname = "temp_valid_lft",
3997 .data = &ipv6_devconf.temp_valid_lft,
3998 .maxlen = sizeof(int),
3999 .mode = 0644,
4000 .proc_handler = &proc_dointvec,
4003 .ctl_name = NET_IPV6_TEMP_PREFERED_LFT,
4004 .procname = "temp_prefered_lft",
4005 .data = &ipv6_devconf.temp_prefered_lft,
4006 .maxlen = sizeof(int),
4007 .mode = 0644,
4008 .proc_handler = &proc_dointvec,
4011 .ctl_name = NET_IPV6_REGEN_MAX_RETRY,
4012 .procname = "regen_max_retry",
4013 .data = &ipv6_devconf.regen_max_retry,
4014 .maxlen = sizeof(int),
4015 .mode = 0644,
4016 .proc_handler = &proc_dointvec,
4019 .ctl_name = NET_IPV6_MAX_DESYNC_FACTOR,
4020 .procname = "max_desync_factor",
4021 .data = &ipv6_devconf.max_desync_factor,
4022 .maxlen = sizeof(int),
4023 .mode = 0644,
4024 .proc_handler = &proc_dointvec,
4026 #endif
4028 .ctl_name = NET_IPV6_MAX_ADDRESSES,
4029 .procname = "max_addresses",
4030 .data = &ipv6_devconf.max_addresses,
4031 .maxlen = sizeof(int),
4032 .mode = 0644,
4033 .proc_handler = &proc_dointvec,
4036 .ctl_name = NET_IPV6_ACCEPT_RA_DEFRTR,
4037 .procname = "accept_ra_defrtr",
4038 .data = &ipv6_devconf.accept_ra_defrtr,
4039 .maxlen = sizeof(int),
4040 .mode = 0644,
4041 .proc_handler = &proc_dointvec,
4044 .ctl_name = NET_IPV6_ACCEPT_RA_PINFO,
4045 .procname = "accept_ra_pinfo",
4046 .data = &ipv6_devconf.accept_ra_pinfo,
4047 .maxlen = sizeof(int),
4048 .mode = 0644,
4049 .proc_handler = &proc_dointvec,
4051 #ifdef CONFIG_IPV6_ROUTER_PREF
4053 .ctl_name = NET_IPV6_ACCEPT_RA_RTR_PREF,
4054 .procname = "accept_ra_rtr_pref",
4055 .data = &ipv6_devconf.accept_ra_rtr_pref,
4056 .maxlen = sizeof(int),
4057 .mode = 0644,
4058 .proc_handler = &proc_dointvec,
4061 .ctl_name = NET_IPV6_RTR_PROBE_INTERVAL,
4062 .procname = "router_probe_interval",
4063 .data = &ipv6_devconf.rtr_probe_interval,
4064 .maxlen = sizeof(int),
4065 .mode = 0644,
4066 .proc_handler = &proc_dointvec_jiffies,
4067 .strategy = &sysctl_jiffies,
4069 #ifdef CONFIG_IPV6_ROUTE_INFO
4071 .ctl_name = NET_IPV6_ACCEPT_RA_RT_INFO_MAX_PLEN,
4072 .procname = "accept_ra_rt_info_max_plen",
4073 .data = &ipv6_devconf.accept_ra_rt_info_max_plen,
4074 .maxlen = sizeof(int),
4075 .mode = 0644,
4076 .proc_handler = &proc_dointvec,
4078 #endif
4079 #endif
4081 .ctl_name = NET_IPV6_PROXY_NDP,
4082 .procname = "proxy_ndp",
4083 .data = &ipv6_devconf.proxy_ndp,
4084 .maxlen = sizeof(int),
4085 .mode = 0644,
4086 .proc_handler = &proc_dointvec,
4089 .ctl_name = NET_IPV6_ACCEPT_SOURCE_ROUTE,
4090 .procname = "accept_source_route",
4091 .data = &ipv6_devconf.accept_source_route,
4092 .maxlen = sizeof(int),
4093 .mode = 0644,
4094 .proc_handler = &proc_dointvec,
4096 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
4098 .ctl_name = CTL_UNNUMBERED,
4099 .procname = "optimistic_dad",
4100 .data = &ipv6_devconf.optimistic_dad,
4101 .maxlen = sizeof(int),
4102 .mode = 0644,
4103 .proc_handler = &proc_dointvec,
4106 #endif
4108 .ctl_name = 0, /* sentinel */
4113 static int __addrconf_sysctl_register(struct net *net, char *dev_name,
4114 int ctl_name, struct inet6_dev *idev, struct ipv6_devconf *p)
4116 int i;
4117 struct addrconf_sysctl_table *t;
4119 #define ADDRCONF_CTL_PATH_DEV 3
4121 struct ctl_path addrconf_ctl_path[] = {
4122 { .procname = "net", .ctl_name = CTL_NET, },
4123 { .procname = "ipv6", .ctl_name = NET_IPV6, },
4124 { .procname = "conf", .ctl_name = NET_IPV6_CONF, },
4125 { /* to be set */ },
4126 { },
4130 t = kmemdup(&addrconf_sysctl, sizeof(*t), GFP_KERNEL);
4131 if (t == NULL)
4132 goto out;
4134 for (i=0; t->addrconf_vars[i].data; i++) {
4135 t->addrconf_vars[i].data += (char*)p - (char*)&ipv6_devconf;
4136 t->addrconf_vars[i].extra1 = idev; /* embedded; no ref */
4137 t->addrconf_vars[i].extra2 = net;
4141 * Make a copy of dev_name, because '.procname' is regarded as const
4142 * by sysctl and we wouldn't want anyone to change it under our feet
4143 * (see SIOCSIFNAME).
4145 t->dev_name = kstrdup(dev_name, GFP_KERNEL);
4146 if (!t->dev_name)
4147 goto free;
4149 addrconf_ctl_path[ADDRCONF_CTL_PATH_DEV].procname = t->dev_name;
4150 addrconf_ctl_path[ADDRCONF_CTL_PATH_DEV].ctl_name = ctl_name;
4152 t->sysctl_header = register_net_sysctl_table(net, addrconf_ctl_path,
4153 t->addrconf_vars);
4154 if (t->sysctl_header == NULL)
4155 goto free_procname;
4157 p->sysctl = t;
4158 return 0;
4160 free_procname:
4161 kfree(t->dev_name);
4162 free:
4163 kfree(t);
4164 out:
4165 return -ENOBUFS;
4168 static void __addrconf_sysctl_unregister(struct ipv6_devconf *p)
4170 struct addrconf_sysctl_table *t;
4172 if (p->sysctl == NULL)
4173 return;
4175 t = p->sysctl;
4176 p->sysctl = NULL;
4177 unregister_sysctl_table(t->sysctl_header);
4178 kfree(t->dev_name);
4179 kfree(t);
4182 static void addrconf_sysctl_register(struct inet6_dev *idev)
4184 neigh_sysctl_register(idev->dev, idev->nd_parms, NET_IPV6,
4185 NET_IPV6_NEIGH, "ipv6",
4186 &ndisc_ifinfo_sysctl_change,
4187 NULL);
4188 __addrconf_sysctl_register(idev->dev->nd_net, idev->dev->name,
4189 idev->dev->ifindex, idev, &idev->cnf);
4192 static void addrconf_sysctl_unregister(struct inet6_dev *idev)
4194 __addrconf_sysctl_unregister(&idev->cnf);
4195 neigh_sysctl_unregister(idev->nd_parms);
4199 #endif
4201 static int addrconf_init_net(struct net *net)
4203 int err;
4204 struct ipv6_devconf *all, *dflt;
4206 err = -ENOMEM;
4207 all = &ipv6_devconf;
4208 dflt = &ipv6_devconf_dflt;
4210 if (net != &init_net) {
4211 all = kmemdup(all, sizeof(ipv6_devconf), GFP_KERNEL);
4212 if (all == NULL)
4213 goto err_alloc_all;
4215 dflt = kmemdup(dflt, sizeof(ipv6_devconf_dflt), GFP_KERNEL);
4216 if (dflt == NULL)
4217 goto err_alloc_dflt;
4220 net->ipv6.devconf_all = all;
4221 net->ipv6.devconf_dflt = dflt;
4223 #ifdef CONFIG_SYSCTL
4224 err = __addrconf_sysctl_register(net, "all", NET_PROTO_CONF_ALL,
4225 NULL, all);
4226 if (err < 0)
4227 goto err_reg_all;
4229 err = __addrconf_sysctl_register(net, "default", NET_PROTO_CONF_DEFAULT,
4230 NULL, dflt);
4231 if (err < 0)
4232 goto err_reg_dflt;
4233 #endif
4234 return 0;
4236 #ifdef CONFIG_SYSCTL
4237 err_reg_dflt:
4238 __addrconf_sysctl_unregister(all);
4239 err_reg_all:
4240 kfree(dflt);
4241 #endif
4242 err_alloc_dflt:
4243 kfree(all);
4244 err_alloc_all:
4245 return err;
4248 static void addrconf_exit_net(struct net *net)
4250 #ifdef CONFIG_SYSCTL
4251 __addrconf_sysctl_unregister(net->ipv6.devconf_dflt);
4252 __addrconf_sysctl_unregister(net->ipv6.devconf_all);
4253 #endif
4254 if (net != &init_net) {
4255 kfree(net->ipv6.devconf_dflt);
4256 kfree(net->ipv6.devconf_all);
4260 static struct pernet_operations addrconf_ops = {
4261 .init = addrconf_init_net,
4262 .exit = addrconf_exit_net,
4266 * Device notifier
4269 int register_inet6addr_notifier(struct notifier_block *nb)
4271 return atomic_notifier_chain_register(&inet6addr_chain, nb);
4274 EXPORT_SYMBOL(register_inet6addr_notifier);
4276 int unregister_inet6addr_notifier(struct notifier_block *nb)
4278 return atomic_notifier_chain_unregister(&inet6addr_chain,nb);
4281 EXPORT_SYMBOL(unregister_inet6addr_notifier);
4284 * Init / cleanup code
4287 int __init addrconf_init(void)
4289 int err;
4291 if ((err = ipv6_addr_label_init()) < 0) {
4292 printk(KERN_CRIT "IPv6 Addrconf: cannot initialize default policy table: %d.\n",
4293 err);
4294 return err;
4297 register_pernet_subsys(&addrconf_ops);
4299 /* The addrconf netdev notifier requires that loopback_dev
4300 * has it's ipv6 private information allocated and setup
4301 * before it can bring up and give link-local addresses
4302 * to other devices which are up.
4304 * Unfortunately, loopback_dev is not necessarily the first
4305 * entry in the global dev_base list of net devices. In fact,
4306 * it is likely to be the very last entry on that list.
4307 * So this causes the notifier registry below to try and
4308 * give link-local addresses to all devices besides loopback_dev
4309 * first, then loopback_dev, which cases all the non-loopback_dev
4310 * devices to fail to get a link-local address.
4312 * So, as a temporary fix, allocate the ipv6 structure for
4313 * loopback_dev first by hand.
4314 * Longer term, all of the dependencies ipv6 has upon the loopback
4315 * device and it being up should be removed.
4317 rtnl_lock();
4318 if (!ipv6_add_dev(init_net.loopback_dev))
4319 err = -ENOMEM;
4320 rtnl_unlock();
4321 if (err)
4322 goto errlo;
4324 ip6_null_entry.u.dst.dev = init_net.loopback_dev;
4325 ip6_null_entry.rt6i_idev = in6_dev_get(init_net.loopback_dev);
4326 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
4327 ip6_prohibit_entry.u.dst.dev = init_net.loopback_dev;
4328 ip6_prohibit_entry.rt6i_idev = in6_dev_get(init_net.loopback_dev);
4329 ip6_blk_hole_entry.u.dst.dev = init_net.loopback_dev;
4330 ip6_blk_hole_entry.rt6i_idev = in6_dev_get(init_net.loopback_dev);
4331 #endif
4333 register_netdevice_notifier(&ipv6_dev_notf);
4335 addrconf_verify(0);
4337 err = __rtnl_register(PF_INET6, RTM_GETLINK, NULL, inet6_dump_ifinfo);
4338 if (err < 0)
4339 goto errout;
4341 /* Only the first call to __rtnl_register can fail */
4342 __rtnl_register(PF_INET6, RTM_NEWADDR, inet6_rtm_newaddr, NULL);
4343 __rtnl_register(PF_INET6, RTM_DELADDR, inet6_rtm_deladdr, NULL);
4344 __rtnl_register(PF_INET6, RTM_GETADDR, inet6_rtm_getaddr, inet6_dump_ifaddr);
4345 __rtnl_register(PF_INET6, RTM_GETMULTICAST, NULL, inet6_dump_ifmcaddr);
4346 __rtnl_register(PF_INET6, RTM_GETANYCAST, NULL, inet6_dump_ifacaddr);
4348 ipv6_addr_label_rtnl_register();
4350 return 0;
4351 errout:
4352 unregister_netdevice_notifier(&ipv6_dev_notf);
4353 errlo:
4354 unregister_pernet_subsys(&addrconf_ops);
4356 return err;
4359 void addrconf_cleanup(void)
4361 struct net_device *dev;
4362 struct inet6_ifaddr *ifa;
4363 int i;
4365 unregister_netdevice_notifier(&ipv6_dev_notf);
4367 unregister_pernet_subsys(&addrconf_ops);
4369 rtnl_lock();
4372 * clean dev list.
4375 for_each_netdev(&init_net, dev) {
4376 if (__in6_dev_get(dev) == NULL)
4377 continue;
4378 addrconf_ifdown(dev, 1);
4380 addrconf_ifdown(init_net.loopback_dev, 2);
4383 * Check hash table.
4386 write_lock_bh(&addrconf_hash_lock);
4387 for (i=0; i < IN6_ADDR_HSIZE; i++) {
4388 for (ifa=inet6_addr_lst[i]; ifa; ) {
4389 struct inet6_ifaddr *bifa;
4391 bifa = ifa;
4392 ifa = ifa->lst_next;
4393 printk(KERN_DEBUG "bug: IPv6 address leakage detected: ifa=%p\n", bifa);
4394 /* Do not free it; something is wrong.
4395 Now we can investigate it with debugger.
4399 write_unlock_bh(&addrconf_hash_lock);
4401 del_timer(&addr_chk_timer);
4403 rtnl_unlock();