2 * IPv6 Address [auto]configuration
3 * Linux INET6 implementation
6 * Pedro Roque <roque@di.fc.ul.pt>
7 * Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
9 * This program is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU General Public License
11 * as published by the Free Software Foundation; either version
12 * 2 of the License, or (at your option) any later version.
18 * Janos Farkas : delete timer on ifdown
19 * <chexum@bankinf.banki.hu>
20 * Andi Kleen : kill double kfree on module
22 * Maciej W. Rozycki : FDDI support
23 * sekiya@USAGI : Don't send too many RS
25 * yoshfuji@USAGI : Fixed interval between DAD
27 * YOSHIFUJI Hideaki @USAGI : improved accuracy of
28 * address validation timer.
29 * YOSHIFUJI Hideaki @USAGI : Privacy Extensions (RFC3041)
31 * Yuji SEKIYA @USAGI : Don't assign a same IPv6
32 * address on a same interface.
33 * YOSHIFUJI Hideaki @USAGI : ARCnet support
34 * YOSHIFUJI Hideaki @USAGI : convert /proc/net/if_inet6 to
36 * YOSHIFUJI Hideaki @USAGI : improved source address
37 * selection; consider scope,
41 #define pr_fmt(fmt) "IPv6: " fmt
43 #include <linux/errno.h>
44 #include <linux/types.h>
45 #include <linux/kernel.h>
46 #include <linux/socket.h>
47 #include <linux/sockios.h>
48 #include <linux/net.h>
49 #include <linux/inet.h>
50 #include <linux/in6.h>
51 #include <linux/netdevice.h>
52 #include <linux/if_addr.h>
53 #include <linux/if_arp.h>
54 #include <linux/if_arcnet.h>
55 #include <linux/if_infiniband.h>
56 #include <linux/route.h>
57 #include <linux/inetdevice.h>
58 #include <linux/init.h>
59 #include <linux/slab.h>
61 #include <linux/sysctl.h>
63 #include <linux/capability.h>
64 #include <linux/delay.h>
65 #include <linux/notifier.h>
66 #include <linux/string.h>
67 #include <linux/hash.h>
69 #include <net/net_namespace.h>
73 #include <net/af_ieee802154.h>
74 #include <net/firewire.h>
76 #include <net/protocol.h>
77 #include <net/ndisc.h>
78 #include <net/ip6_route.h>
79 #include <net/addrconf.h>
82 #include <net/netlink.h>
83 #include <net/pkt_sched.h>
84 #include <linux/if_tunnel.h>
85 #include <linux/rtnetlink.h>
86 #include <linux/netconf.h>
87 #include <linux/random.h>
88 #include <linux/uaccess.h>
89 #include <asm/unaligned.h>
91 #include <linux/proc_fs.h>
92 #include <linux/seq_file.h>
93 #include <linux/export.h>
95 /* Set to 3 to get tracing... */
99 #define ADBG(fmt, ...) printk(fmt, ##__VA_ARGS__)
101 #define ADBG(fmt, ...) do { if (0) printk(fmt, ##__VA_ARGS__); } while (0)
104 #define INFINITY_LIFE_TIME 0xFFFFFFFF
106 #define IPV6_MAX_STRLEN \
107 sizeof("ffff:ffff:ffff:ffff:ffff:ffff:255.255.255.255")
109 static inline u32
cstamp_delta(unsigned long cstamp
)
111 return (cstamp
- INITIAL_JIFFIES
) * 100UL / HZ
;
115 static int addrconf_sysctl_register(struct inet6_dev
*idev
);
116 static void addrconf_sysctl_unregister(struct inet6_dev
*idev
);
118 static inline int addrconf_sysctl_register(struct inet6_dev
*idev
)
123 static inline void addrconf_sysctl_unregister(struct inet6_dev
*idev
)
128 static void __ipv6_regen_rndid(struct inet6_dev
*idev
);
129 static void __ipv6_try_regen_rndid(struct inet6_dev
*idev
, struct in6_addr
*tmpaddr
);
130 static void ipv6_regen_rndid(unsigned long data
);
132 static int ipv6_generate_eui64(u8
*eui
, struct net_device
*dev
);
133 static int ipv6_count_addresses(struct inet6_dev
*idev
);
134 static int ipv6_generate_stable_address(struct in6_addr
*addr
,
136 const struct inet6_dev
*idev
);
139 * Configured unicast address hash table
141 static struct hlist_head inet6_addr_lst
[IN6_ADDR_HSIZE
];
142 static DEFINE_SPINLOCK(addrconf_hash_lock
);
144 static void addrconf_verify(void);
145 static void addrconf_verify_rtnl(void);
146 static void addrconf_verify_work(struct work_struct
*);
148 static struct workqueue_struct
*addrconf_wq
;
149 static DECLARE_DELAYED_WORK(addr_chk_work
, addrconf_verify_work
);
151 static void addrconf_join_anycast(struct inet6_ifaddr
*ifp
);
152 static void addrconf_leave_anycast(struct inet6_ifaddr
*ifp
);
154 static void addrconf_type_change(struct net_device
*dev
,
155 unsigned long event
);
156 static int addrconf_ifdown(struct net_device
*dev
, int how
);
158 static struct rt6_info
*addrconf_get_prefix_route(const struct in6_addr
*pfx
,
160 const struct net_device
*dev
,
161 u32 flags
, u32 noflags
);
163 static void addrconf_dad_start(struct inet6_ifaddr
*ifp
);
164 static void addrconf_dad_work(struct work_struct
*w
);
165 static void addrconf_dad_completed(struct inet6_ifaddr
*ifp
);
166 static void addrconf_dad_run(struct inet6_dev
*idev
);
167 static void addrconf_rs_timer(unsigned long data
);
168 static void __ipv6_ifa_notify(int event
, struct inet6_ifaddr
*ifa
);
169 static void ipv6_ifa_notify(int event
, struct inet6_ifaddr
*ifa
);
171 static void inet6_prefix_notify(int event
, struct inet6_dev
*idev
,
172 struct prefix_info
*pinfo
);
173 static bool ipv6_chk_same_addr(struct net
*net
, const struct in6_addr
*addr
,
174 struct net_device
*dev
);
176 static struct ipv6_devconf ipv6_devconf __read_mostly
= {
178 .hop_limit
= IPV6_DEFAULT_HOPLIMIT
,
179 .mtu6
= IPV6_MIN_MTU
,
181 .accept_redirects
= 1,
183 .force_mld_version
= 0,
184 .mldv1_unsolicited_report_interval
= 10 * HZ
,
185 .mldv2_unsolicited_report_interval
= HZ
,
187 .rtr_solicits
= MAX_RTR_SOLICITATIONS
,
188 .rtr_solicit_interval
= RTR_SOLICITATION_INTERVAL
,
189 .rtr_solicit_delay
= MAX_RTR_SOLICITATION_DELAY
,
191 .temp_valid_lft
= TEMP_VALID_LIFETIME
,
192 .temp_prefered_lft
= TEMP_PREFERRED_LIFETIME
,
193 .regen_max_retry
= REGEN_MAX_RETRY
,
194 .max_desync_factor
= MAX_DESYNC_FACTOR
,
195 .max_addresses
= IPV6_MAX_ADDRESSES
,
196 .accept_ra_defrtr
= 1,
197 .accept_ra_from_local
= 0,
198 .accept_ra_min_hop_limit
= 1,
199 .accept_ra_pinfo
= 1,
200 #ifdef CONFIG_IPV6_ROUTER_PREF
201 .accept_ra_rtr_pref
= 1,
202 .rtr_probe_interval
= 60 * HZ
,
203 #ifdef CONFIG_IPV6_ROUTE_INFO
204 .accept_ra_rt_info_max_plen
= 0,
208 .accept_source_route
= 0, /* we do not accept RH0 by default. */
211 .suppress_frag_ndisc
= 1,
214 .initialized
= false,
216 .use_oif_addrs_only
= 0,
217 .ignore_routes_with_linkdown
= 0,
220 static struct ipv6_devconf ipv6_devconf_dflt __read_mostly
= {
222 .hop_limit
= IPV6_DEFAULT_HOPLIMIT
,
223 .mtu6
= IPV6_MIN_MTU
,
225 .accept_redirects
= 1,
227 .force_mld_version
= 0,
228 .mldv1_unsolicited_report_interval
= 10 * HZ
,
229 .mldv2_unsolicited_report_interval
= HZ
,
231 .rtr_solicits
= MAX_RTR_SOLICITATIONS
,
232 .rtr_solicit_interval
= RTR_SOLICITATION_INTERVAL
,
233 .rtr_solicit_delay
= MAX_RTR_SOLICITATION_DELAY
,
235 .temp_valid_lft
= TEMP_VALID_LIFETIME
,
236 .temp_prefered_lft
= TEMP_PREFERRED_LIFETIME
,
237 .regen_max_retry
= REGEN_MAX_RETRY
,
238 .max_desync_factor
= MAX_DESYNC_FACTOR
,
239 .max_addresses
= IPV6_MAX_ADDRESSES
,
240 .accept_ra_defrtr
= 1,
241 .accept_ra_from_local
= 0,
242 .accept_ra_min_hop_limit
= 1,
243 .accept_ra_pinfo
= 1,
244 #ifdef CONFIG_IPV6_ROUTER_PREF
245 .accept_ra_rtr_pref
= 1,
246 .rtr_probe_interval
= 60 * HZ
,
247 #ifdef CONFIG_IPV6_ROUTE_INFO
248 .accept_ra_rt_info_max_plen
= 0,
252 .accept_source_route
= 0, /* we do not accept RH0 by default. */
255 .suppress_frag_ndisc
= 1,
258 .initialized
= false,
260 .use_oif_addrs_only
= 0,
261 .ignore_routes_with_linkdown
= 0,
264 /* Check if a valid qdisc is available */
265 static inline bool addrconf_qdisc_ok(const struct net_device
*dev
)
267 return !qdisc_tx_is_noop(dev
);
270 static void addrconf_del_rs_timer(struct inet6_dev
*idev
)
272 if (del_timer(&idev
->rs_timer
))
276 static void addrconf_del_dad_work(struct inet6_ifaddr
*ifp
)
278 if (cancel_delayed_work(&ifp
->dad_work
))
282 static void addrconf_mod_rs_timer(struct inet6_dev
*idev
,
285 if (!timer_pending(&idev
->rs_timer
))
287 mod_timer(&idev
->rs_timer
, jiffies
+ when
);
290 static void addrconf_mod_dad_work(struct inet6_ifaddr
*ifp
,
293 if (!delayed_work_pending(&ifp
->dad_work
))
295 mod_delayed_work(addrconf_wq
, &ifp
->dad_work
, delay
);
298 static int snmp6_alloc_dev(struct inet6_dev
*idev
)
302 idev
->stats
.ipv6
= alloc_percpu(struct ipstats_mib
);
303 if (!idev
->stats
.ipv6
)
306 for_each_possible_cpu(i
) {
307 struct ipstats_mib
*addrconf_stats
;
308 addrconf_stats
= per_cpu_ptr(idev
->stats
.ipv6
, i
);
309 u64_stats_init(&addrconf_stats
->syncp
);
313 idev
->stats
.icmpv6dev
= kzalloc(sizeof(struct icmpv6_mib_device
),
315 if (!idev
->stats
.icmpv6dev
)
317 idev
->stats
.icmpv6msgdev
= kzalloc(sizeof(struct icmpv6msg_mib_device
),
319 if (!idev
->stats
.icmpv6msgdev
)
325 kfree(idev
->stats
.icmpv6dev
);
327 free_percpu(idev
->stats
.ipv6
);
332 static struct inet6_dev
*ipv6_add_dev(struct net_device
*dev
)
334 struct inet6_dev
*ndev
;
339 if (dev
->mtu
< IPV6_MIN_MTU
)
340 return ERR_PTR(-EINVAL
);
342 ndev
= kzalloc(sizeof(struct inet6_dev
), GFP_KERNEL
);
346 rwlock_init(&ndev
->lock
);
348 INIT_LIST_HEAD(&ndev
->addr_list
);
349 setup_timer(&ndev
->rs_timer
, addrconf_rs_timer
,
350 (unsigned long)ndev
);
351 memcpy(&ndev
->cnf
, dev_net(dev
)->ipv6
.devconf_dflt
, sizeof(ndev
->cnf
));
352 ndev
->cnf
.mtu6
= dev
->mtu
;
353 ndev
->cnf
.sysctl
= NULL
;
354 ndev
->nd_parms
= neigh_parms_alloc(dev
, &nd_tbl
);
355 if (!ndev
->nd_parms
) {
359 if (ndev
->cnf
.forwarding
)
360 dev_disable_lro(dev
);
361 /* We refer to the device */
364 if (snmp6_alloc_dev(ndev
) < 0) {
366 "%s: cannot allocate memory for statistics; dev=%s.\n",
367 __func__
, dev
->name
);
368 neigh_parms_release(&nd_tbl
, ndev
->nd_parms
);
374 if (snmp6_register_dev(ndev
) < 0) {
376 "%s: cannot create /proc/net/dev_snmp6/%s\n",
377 __func__
, dev
->name
);
381 /* One reference from device. We must do this before
382 * we invoke __ipv6_regen_rndid().
386 if (dev
->flags
& (IFF_NOARP
| IFF_LOOPBACK
))
387 ndev
->cnf
.accept_dad
= -1;
389 #if IS_ENABLED(CONFIG_IPV6_SIT)
390 if (dev
->type
== ARPHRD_SIT
&& (dev
->priv_flags
& IFF_ISATAP
)) {
391 pr_info("%s: Disabled Multicast RS\n", dev
->name
);
392 ndev
->cnf
.rtr_solicits
= 0;
396 INIT_LIST_HEAD(&ndev
->tempaddr_list
);
397 setup_timer(&ndev
->regen_timer
, ipv6_regen_rndid
, (unsigned long)ndev
);
398 if ((dev
->flags
&IFF_LOOPBACK
) ||
399 dev
->type
== ARPHRD_TUNNEL
||
400 dev
->type
== ARPHRD_TUNNEL6
||
401 dev
->type
== ARPHRD_SIT
||
402 dev
->type
== ARPHRD_NONE
) {
403 ndev
->cnf
.use_tempaddr
= -1;
406 ipv6_regen_rndid((unsigned long) ndev
);
409 ndev
->token
= in6addr_any
;
411 if (netif_running(dev
) && addrconf_qdisc_ok(dev
))
412 ndev
->if_flags
|= IF_READY
;
414 ipv6_mc_init_dev(ndev
);
415 ndev
->tstamp
= jiffies
;
416 err
= addrconf_sysctl_register(ndev
);
418 ipv6_mc_destroy_dev(ndev
);
419 del_timer(&ndev
->regen_timer
);
422 /* protected by rtnl_lock */
423 rcu_assign_pointer(dev
->ip6_ptr
, ndev
);
425 /* Join interface-local all-node multicast group */
426 ipv6_dev_mc_inc(dev
, &in6addr_interfacelocal_allnodes
);
428 /* Join all-node multicast group */
429 ipv6_dev_mc_inc(dev
, &in6addr_linklocal_allnodes
);
431 /* Join all-router multicast group if forwarding is set */
432 if (ndev
->cnf
.forwarding
&& (dev
->flags
& IFF_MULTICAST
))
433 ipv6_dev_mc_inc(dev
, &in6addr_linklocal_allrouters
);
438 neigh_parms_release(&nd_tbl
, ndev
->nd_parms
);
440 in6_dev_finish_destroy(ndev
);
444 static struct inet6_dev
*ipv6_find_idev(struct net_device
*dev
)
446 struct inet6_dev
*idev
;
450 idev
= __in6_dev_get(dev
);
452 idev
= ipv6_add_dev(dev
);
457 if (dev
->flags
&IFF_UP
)
462 static int inet6_netconf_msgsize_devconf(int type
)
464 int size
= NLMSG_ALIGN(sizeof(struct netconfmsg
))
465 + nla_total_size(4); /* NETCONFA_IFINDEX */
467 /* type -1 is used for ALL */
468 if (type
== -1 || type
== NETCONFA_FORWARDING
)
469 size
+= nla_total_size(4);
470 #ifdef CONFIG_IPV6_MROUTE
471 if (type
== -1 || type
== NETCONFA_MC_FORWARDING
)
472 size
+= nla_total_size(4);
474 if (type
== -1 || type
== NETCONFA_PROXY_NEIGH
)
475 size
+= nla_total_size(4);
477 if (type
== -1 || type
== NETCONFA_IGNORE_ROUTES_WITH_LINKDOWN
)
478 size
+= nla_total_size(4);
483 static int inet6_netconf_fill_devconf(struct sk_buff
*skb
, int ifindex
,
484 struct ipv6_devconf
*devconf
, u32 portid
,
485 u32 seq
, int event
, unsigned int flags
,
488 struct nlmsghdr
*nlh
;
489 struct netconfmsg
*ncm
;
491 nlh
= nlmsg_put(skb
, portid
, seq
, event
, sizeof(struct netconfmsg
),
496 ncm
= nlmsg_data(nlh
);
497 ncm
->ncm_family
= AF_INET6
;
499 if (nla_put_s32(skb
, NETCONFA_IFINDEX
, ifindex
) < 0)
500 goto nla_put_failure
;
502 /* type -1 is used for ALL */
503 if ((type
== -1 || type
== NETCONFA_FORWARDING
) &&
504 nla_put_s32(skb
, NETCONFA_FORWARDING
, devconf
->forwarding
) < 0)
505 goto nla_put_failure
;
506 #ifdef CONFIG_IPV6_MROUTE
507 if ((type
== -1 || type
== NETCONFA_MC_FORWARDING
) &&
508 nla_put_s32(skb
, NETCONFA_MC_FORWARDING
,
509 devconf
->mc_forwarding
) < 0)
510 goto nla_put_failure
;
512 if ((type
== -1 || type
== NETCONFA_PROXY_NEIGH
) &&
513 nla_put_s32(skb
, NETCONFA_PROXY_NEIGH
, devconf
->proxy_ndp
) < 0)
514 goto nla_put_failure
;
516 if ((type
== -1 || type
== NETCONFA_IGNORE_ROUTES_WITH_LINKDOWN
) &&
517 nla_put_s32(skb
, NETCONFA_IGNORE_ROUTES_WITH_LINKDOWN
,
518 devconf
->ignore_routes_with_linkdown
) < 0)
519 goto nla_put_failure
;
525 nlmsg_cancel(skb
, nlh
);
529 void inet6_netconf_notify_devconf(struct net
*net
, int type
, int ifindex
,
530 struct ipv6_devconf
*devconf
)
535 skb
= nlmsg_new(inet6_netconf_msgsize_devconf(type
), GFP_ATOMIC
);
539 err
= inet6_netconf_fill_devconf(skb
, ifindex
, devconf
, 0, 0,
540 RTM_NEWNETCONF
, 0, type
);
542 /* -EMSGSIZE implies BUG in inet6_netconf_msgsize_devconf() */
543 WARN_ON(err
== -EMSGSIZE
);
547 rtnl_notify(skb
, net
, 0, RTNLGRP_IPV6_NETCONF
, NULL
, GFP_ATOMIC
);
550 rtnl_set_sk_err(net
, RTNLGRP_IPV6_NETCONF
, err
);
553 static const struct nla_policy devconf_ipv6_policy
[NETCONFA_MAX
+1] = {
554 [NETCONFA_IFINDEX
] = { .len
= sizeof(int) },
555 [NETCONFA_FORWARDING
] = { .len
= sizeof(int) },
556 [NETCONFA_PROXY_NEIGH
] = { .len
= sizeof(int) },
557 [NETCONFA_IGNORE_ROUTES_WITH_LINKDOWN
] = { .len
= sizeof(int) },
560 static int inet6_netconf_get_devconf(struct sk_buff
*in_skb
,
561 struct nlmsghdr
*nlh
)
563 struct net
*net
= sock_net(in_skb
->sk
);
564 struct nlattr
*tb
[NETCONFA_MAX
+1];
565 struct netconfmsg
*ncm
;
567 struct ipv6_devconf
*devconf
;
568 struct inet6_dev
*in6_dev
;
569 struct net_device
*dev
;
573 err
= nlmsg_parse(nlh
, sizeof(*ncm
), tb
, NETCONFA_MAX
,
574 devconf_ipv6_policy
);
579 if (!tb
[NETCONFA_IFINDEX
])
582 ifindex
= nla_get_s32(tb
[NETCONFA_IFINDEX
]);
584 case NETCONFA_IFINDEX_ALL
:
585 devconf
= net
->ipv6
.devconf_all
;
587 case NETCONFA_IFINDEX_DEFAULT
:
588 devconf
= net
->ipv6
.devconf_dflt
;
591 dev
= __dev_get_by_index(net
, ifindex
);
594 in6_dev
= __in6_dev_get(dev
);
597 devconf
= &in6_dev
->cnf
;
602 skb
= nlmsg_new(inet6_netconf_msgsize_devconf(-1), GFP_ATOMIC
);
606 err
= inet6_netconf_fill_devconf(skb
, ifindex
, devconf
,
607 NETLINK_CB(in_skb
).portid
,
608 nlh
->nlmsg_seq
, RTM_NEWNETCONF
, 0,
611 /* -EMSGSIZE implies BUG in inet6_netconf_msgsize_devconf() */
612 WARN_ON(err
== -EMSGSIZE
);
616 err
= rtnl_unicast(skb
, net
, NETLINK_CB(in_skb
).portid
);
621 static int inet6_netconf_dump_devconf(struct sk_buff
*skb
,
622 struct netlink_callback
*cb
)
624 struct net
*net
= sock_net(skb
->sk
);
627 struct net_device
*dev
;
628 struct inet6_dev
*idev
;
629 struct hlist_head
*head
;
632 s_idx
= idx
= cb
->args
[1];
634 for (h
= s_h
; h
< NETDEV_HASHENTRIES
; h
++, s_idx
= 0) {
636 head
= &net
->dev_index_head
[h
];
638 cb
->seq
= atomic_read(&net
->ipv6
.dev_addr_genid
) ^
640 hlist_for_each_entry_rcu(dev
, head
, index_hlist
) {
643 idev
= __in6_dev_get(dev
);
647 if (inet6_netconf_fill_devconf(skb
, dev
->ifindex
,
649 NETLINK_CB(cb
->skb
).portid
,
657 nl_dump_check_consistent(cb
, nlmsg_hdr(skb
));
663 if (h
== NETDEV_HASHENTRIES
) {
664 if (inet6_netconf_fill_devconf(skb
, NETCONFA_IFINDEX_ALL
,
665 net
->ipv6
.devconf_all
,
666 NETLINK_CB(cb
->skb
).portid
,
668 RTM_NEWNETCONF
, NLM_F_MULTI
,
674 if (h
== NETDEV_HASHENTRIES
+ 1) {
675 if (inet6_netconf_fill_devconf(skb
, NETCONFA_IFINDEX_DEFAULT
,
676 net
->ipv6
.devconf_dflt
,
677 NETLINK_CB(cb
->skb
).portid
,
679 RTM_NEWNETCONF
, NLM_F_MULTI
,
693 static void dev_forward_change(struct inet6_dev
*idev
)
695 struct net_device
*dev
;
696 struct inet6_ifaddr
*ifa
;
701 if (idev
->cnf
.forwarding
)
702 dev_disable_lro(dev
);
703 if (dev
->flags
& IFF_MULTICAST
) {
704 if (idev
->cnf
.forwarding
) {
705 ipv6_dev_mc_inc(dev
, &in6addr_linklocal_allrouters
);
706 ipv6_dev_mc_inc(dev
, &in6addr_interfacelocal_allrouters
);
707 ipv6_dev_mc_inc(dev
, &in6addr_sitelocal_allrouters
);
709 ipv6_dev_mc_dec(dev
, &in6addr_linklocal_allrouters
);
710 ipv6_dev_mc_dec(dev
, &in6addr_interfacelocal_allrouters
);
711 ipv6_dev_mc_dec(dev
, &in6addr_sitelocal_allrouters
);
715 list_for_each_entry(ifa
, &idev
->addr_list
, if_list
) {
716 if (ifa
->flags
&IFA_F_TENTATIVE
)
718 if (idev
->cnf
.forwarding
)
719 addrconf_join_anycast(ifa
);
721 addrconf_leave_anycast(ifa
);
723 inet6_netconf_notify_devconf(dev_net(dev
), NETCONFA_FORWARDING
,
724 dev
->ifindex
, &idev
->cnf
);
728 static void addrconf_forward_change(struct net
*net
, __s32 newf
)
730 struct net_device
*dev
;
731 struct inet6_dev
*idev
;
733 for_each_netdev(net
, dev
) {
734 idev
= __in6_dev_get(dev
);
736 int changed
= (!idev
->cnf
.forwarding
) ^ (!newf
);
737 idev
->cnf
.forwarding
= newf
;
739 dev_forward_change(idev
);
744 static int addrconf_fixup_forwarding(struct ctl_table
*table
, int *p
, int newf
)
750 return restart_syscall();
752 net
= (struct net
*)table
->extra2
;
756 if (p
== &net
->ipv6
.devconf_dflt
->forwarding
) {
757 if ((!newf
) ^ (!old
))
758 inet6_netconf_notify_devconf(net
, NETCONFA_FORWARDING
,
759 NETCONFA_IFINDEX_DEFAULT
,
760 net
->ipv6
.devconf_dflt
);
765 if (p
== &net
->ipv6
.devconf_all
->forwarding
) {
766 net
->ipv6
.devconf_dflt
->forwarding
= newf
;
767 addrconf_forward_change(net
, newf
);
768 if ((!newf
) ^ (!old
))
769 inet6_netconf_notify_devconf(net
, NETCONFA_FORWARDING
,
770 NETCONFA_IFINDEX_ALL
,
771 net
->ipv6
.devconf_all
);
772 } else if ((!newf
) ^ (!old
))
773 dev_forward_change((struct inet6_dev
*)table
->extra1
);
777 rt6_purge_dflt_routers(net
);
781 static void addrconf_linkdown_change(struct net
*net
, __s32 newf
)
783 struct net_device
*dev
;
784 struct inet6_dev
*idev
;
786 for_each_netdev(net
, dev
) {
787 idev
= __in6_dev_get(dev
);
789 int changed
= (!idev
->cnf
.ignore_routes_with_linkdown
) ^ (!newf
);
791 idev
->cnf
.ignore_routes_with_linkdown
= newf
;
793 inet6_netconf_notify_devconf(dev_net(dev
),
794 NETCONFA_IGNORE_ROUTES_WITH_LINKDOWN
,
801 static int addrconf_fixup_linkdown(struct ctl_table
*table
, int *p
, int newf
)
807 return restart_syscall();
809 net
= (struct net
*)table
->extra2
;
813 if (p
== &net
->ipv6
.devconf_dflt
->ignore_routes_with_linkdown
) {
814 if ((!newf
) ^ (!old
))
815 inet6_netconf_notify_devconf(net
,
816 NETCONFA_IGNORE_ROUTES_WITH_LINKDOWN
,
817 NETCONFA_IFINDEX_DEFAULT
,
818 net
->ipv6
.devconf_dflt
);
823 if (p
== &net
->ipv6
.devconf_all
->ignore_routes_with_linkdown
) {
824 net
->ipv6
.devconf_dflt
->ignore_routes_with_linkdown
= newf
;
825 addrconf_linkdown_change(net
, newf
);
826 if ((!newf
) ^ (!old
))
827 inet6_netconf_notify_devconf(net
,
828 NETCONFA_IGNORE_ROUTES_WITH_LINKDOWN
,
829 NETCONFA_IFINDEX_ALL
,
830 net
->ipv6
.devconf_all
);
839 /* Nobody refers to this ifaddr, destroy it */
840 void inet6_ifa_finish_destroy(struct inet6_ifaddr
*ifp
)
842 WARN_ON(!hlist_unhashed(&ifp
->addr_lst
));
844 #ifdef NET_REFCNT_DEBUG
845 pr_debug("%s\n", __func__
);
848 in6_dev_put(ifp
->idev
);
850 if (cancel_delayed_work(&ifp
->dad_work
))
851 pr_notice("delayed DAD work was pending while freeing ifa=%p\n",
854 if (ifp
->state
!= INET6_IFADDR_STATE_DEAD
) {
855 pr_warn("Freeing alive inet6 address %p\n", ifp
);
864 ipv6_link_dev_addr(struct inet6_dev
*idev
, struct inet6_ifaddr
*ifp
)
867 int ifp_scope
= ipv6_addr_src_scope(&ifp
->addr
);
870 * Each device address list is sorted in order of scope -
871 * global before linklocal.
873 list_for_each(p
, &idev
->addr_list
) {
874 struct inet6_ifaddr
*ifa
875 = list_entry(p
, struct inet6_ifaddr
, if_list
);
876 if (ifp_scope
>= ipv6_addr_src_scope(&ifa
->addr
))
880 list_add_tail(&ifp
->if_list
, p
);
883 static u32
inet6_addr_hash(const struct in6_addr
*addr
)
885 return hash_32(ipv6_addr_hash(addr
), IN6_ADDR_HSIZE_SHIFT
);
888 /* On success it returns ifp with increased reference count */
890 static struct inet6_ifaddr
*
891 ipv6_add_addr(struct inet6_dev
*idev
, const struct in6_addr
*addr
,
892 const struct in6_addr
*peer_addr
, int pfxlen
,
893 int scope
, u32 flags
, u32 valid_lft
, u32 prefered_lft
)
895 struct inet6_ifaddr
*ifa
= NULL
;
899 int addr_type
= ipv6_addr_type(addr
);
901 if (addr_type
== IPV6_ADDR_ANY
||
902 addr_type
& IPV6_ADDR_MULTICAST
||
903 (!(idev
->dev
->flags
& IFF_LOOPBACK
) &&
904 addr_type
& IPV6_ADDR_LOOPBACK
))
905 return ERR_PTR(-EADDRNOTAVAIL
);
909 err
= -ENODEV
; /*XXX*/
913 if (idev
->cnf
.disable_ipv6
) {
918 spin_lock(&addrconf_hash_lock
);
920 /* Ignore adding duplicate addresses on an interface */
921 if (ipv6_chk_same_addr(dev_net(idev
->dev
), addr
, idev
->dev
)) {
922 ADBG("ipv6_add_addr: already assigned\n");
927 ifa
= kzalloc(sizeof(struct inet6_ifaddr
), GFP_ATOMIC
);
930 ADBG("ipv6_add_addr: malloc failed\n");
935 rt
= addrconf_dst_alloc(idev
, addr
, false);
941 neigh_parms_data_state_setall(idev
->nd_parms
);
945 ifa
->peer_addr
= *peer_addr
;
947 spin_lock_init(&ifa
->lock
);
948 INIT_DELAYED_WORK(&ifa
->dad_work
, addrconf_dad_work
);
949 INIT_HLIST_NODE(&ifa
->addr_lst
);
951 ifa
->prefix_len
= pfxlen
;
952 ifa
->flags
= flags
| IFA_F_TENTATIVE
;
953 ifa
->valid_lft
= valid_lft
;
954 ifa
->prefered_lft
= prefered_lft
;
955 ifa
->cstamp
= ifa
->tstamp
= jiffies
;
956 ifa
->tokenized
= false;
965 /* Add to big hash table */
966 hash
= inet6_addr_hash(addr
);
968 hlist_add_head_rcu(&ifa
->addr_lst
, &inet6_addr_lst
[hash
]);
969 spin_unlock(&addrconf_hash_lock
);
971 write_lock(&idev
->lock
);
972 /* Add to inet6_dev unicast addr list. */
973 ipv6_link_dev_addr(idev
, ifa
);
975 if (ifa
->flags
&IFA_F_TEMPORARY
) {
976 list_add(&ifa
->tmp_list
, &idev
->tempaddr_list
);
981 write_unlock(&idev
->lock
);
983 rcu_read_unlock_bh();
985 if (likely(err
== 0))
986 inet6addr_notifier_call_chain(NETDEV_UP
, ifa
);
994 spin_unlock(&addrconf_hash_lock
);
998 enum cleanup_prefix_rt_t
{
999 CLEANUP_PREFIX_RT_NOP
, /* no cleanup action for prefix route */
1000 CLEANUP_PREFIX_RT_DEL
, /* delete the prefix route */
1001 CLEANUP_PREFIX_RT_EXPIRE
, /* update the lifetime of the prefix route */
1005 * Check, whether the prefix for ifp would still need a prefix route
1006 * after deleting ifp. The function returns one of the CLEANUP_PREFIX_RT_*
1009 * 1) we don't purge prefix if address was not permanent.
1010 * prefix is managed by its own lifetime.
1011 * 2) we also don't purge, if the address was IFA_F_NOPREFIXROUTE.
1012 * 3) if there are no addresses, delete prefix.
1013 * 4) if there are still other permanent address(es),
1014 * corresponding prefix is still permanent.
1015 * 5) if there are still other addresses with IFA_F_NOPREFIXROUTE,
1016 * don't purge the prefix, assume user space is managing it.
1017 * 6) otherwise, update prefix lifetime to the
1018 * longest valid lifetime among the corresponding
1019 * addresses on the device.
1020 * Note: subsequent RA will update lifetime.
1022 static enum cleanup_prefix_rt_t
1023 check_cleanup_prefix_route(struct inet6_ifaddr
*ifp
, unsigned long *expires
)
1025 struct inet6_ifaddr
*ifa
;
1026 struct inet6_dev
*idev
= ifp
->idev
;
1027 unsigned long lifetime
;
1028 enum cleanup_prefix_rt_t action
= CLEANUP_PREFIX_RT_DEL
;
1032 list_for_each_entry(ifa
, &idev
->addr_list
, if_list
) {
1035 if (!ipv6_prefix_equal(&ifa
->addr
, &ifp
->addr
,
1038 if (ifa
->flags
& (IFA_F_PERMANENT
| IFA_F_NOPREFIXROUTE
))
1039 return CLEANUP_PREFIX_RT_NOP
;
1041 action
= CLEANUP_PREFIX_RT_EXPIRE
;
1043 spin_lock(&ifa
->lock
);
1045 lifetime
= addrconf_timeout_fixup(ifa
->valid_lft
, HZ
);
1047 * Note: Because this address is
1048 * not permanent, lifetime <
1049 * LONG_MAX / HZ here.
1051 if (time_before(*expires
, ifa
->tstamp
+ lifetime
* HZ
))
1052 *expires
= ifa
->tstamp
+ lifetime
* HZ
;
1053 spin_unlock(&ifa
->lock
);
1060 cleanup_prefix_route(struct inet6_ifaddr
*ifp
, unsigned long expires
, bool del_rt
)
1062 struct rt6_info
*rt
;
1064 rt
= addrconf_get_prefix_route(&ifp
->addr
,
1067 0, RTF_GATEWAY
| RTF_DEFAULT
);
1072 if (!(rt
->rt6i_flags
& RTF_EXPIRES
))
1073 rt6_set_expires(rt
, expires
);
1080 /* This function wants to get referenced ifp and releases it before return */
1082 static void ipv6_del_addr(struct inet6_ifaddr
*ifp
)
1085 enum cleanup_prefix_rt_t action
= CLEANUP_PREFIX_RT_NOP
;
1086 unsigned long expires
;
1090 spin_lock_bh(&ifp
->lock
);
1092 ifp
->state
= INET6_IFADDR_STATE_DEAD
;
1093 spin_unlock_bh(&ifp
->lock
);
1095 if (state
== INET6_IFADDR_STATE_DEAD
)
1098 spin_lock_bh(&addrconf_hash_lock
);
1099 hlist_del_init_rcu(&ifp
->addr_lst
);
1100 spin_unlock_bh(&addrconf_hash_lock
);
1102 write_lock_bh(&ifp
->idev
->lock
);
1104 if (ifp
->flags
&IFA_F_TEMPORARY
) {
1105 list_del(&ifp
->tmp_list
);
1107 in6_ifa_put(ifp
->ifpub
);
1113 if (ifp
->flags
& IFA_F_PERMANENT
&& !(ifp
->flags
& IFA_F_NOPREFIXROUTE
))
1114 action
= check_cleanup_prefix_route(ifp
, &expires
);
1116 list_del_init(&ifp
->if_list
);
1119 write_unlock_bh(&ifp
->idev
->lock
);
1121 addrconf_del_dad_work(ifp
);
1123 ipv6_ifa_notify(RTM_DELADDR
, ifp
);
1125 inet6addr_notifier_call_chain(NETDEV_DOWN
, ifp
);
1127 if (action
!= CLEANUP_PREFIX_RT_NOP
) {
1128 cleanup_prefix_route(ifp
, expires
,
1129 action
== CLEANUP_PREFIX_RT_DEL
);
1132 /* clean up prefsrc entries */
1133 rt6_remove_prefsrc(ifp
);
1138 static int ipv6_create_tempaddr(struct inet6_ifaddr
*ifp
, struct inet6_ifaddr
*ift
)
1140 struct inet6_dev
*idev
= ifp
->idev
;
1141 struct in6_addr addr
, *tmpaddr
;
1142 unsigned long tmp_prefered_lft
, tmp_valid_lft
, tmp_tstamp
, age
;
1143 unsigned long regen_advance
;
1147 unsigned long now
= jiffies
;
1149 write_lock_bh(&idev
->lock
);
1151 spin_lock_bh(&ift
->lock
);
1152 memcpy(&addr
.s6_addr
[8], &ift
->addr
.s6_addr
[8], 8);
1153 spin_unlock_bh(&ift
->lock
);
1160 if (idev
->cnf
.use_tempaddr
<= 0) {
1161 write_unlock_bh(&idev
->lock
);
1162 pr_info("%s: use_tempaddr is disabled\n", __func__
);
1167 spin_lock_bh(&ifp
->lock
);
1168 if (ifp
->regen_count
++ >= idev
->cnf
.regen_max_retry
) {
1169 idev
->cnf
.use_tempaddr
= -1; /*XXX*/
1170 spin_unlock_bh(&ifp
->lock
);
1171 write_unlock_bh(&idev
->lock
);
1172 pr_warn("%s: regeneration time exceeded - disabled temporary address support\n",
1179 memcpy(addr
.s6_addr
, ifp
->addr
.s6_addr
, 8);
1180 __ipv6_try_regen_rndid(idev
, tmpaddr
);
1181 memcpy(&addr
.s6_addr
[8], idev
->rndid
, 8);
1182 age
= (now
- ifp
->tstamp
) / HZ
;
1183 tmp_valid_lft
= min_t(__u32
,
1185 idev
->cnf
.temp_valid_lft
+ age
);
1186 tmp_prefered_lft
= min_t(__u32
,
1188 idev
->cnf
.temp_prefered_lft
+ age
-
1189 idev
->cnf
.max_desync_factor
);
1190 tmp_plen
= ifp
->prefix_len
;
1191 tmp_tstamp
= ifp
->tstamp
;
1192 spin_unlock_bh(&ifp
->lock
);
1194 regen_advance
= idev
->cnf
.regen_max_retry
*
1195 idev
->cnf
.dad_transmits
*
1196 NEIGH_VAR(idev
->nd_parms
, RETRANS_TIME
) / HZ
;
1197 write_unlock_bh(&idev
->lock
);
1199 /* A temporary address is created only if this calculated Preferred
1200 * Lifetime is greater than REGEN_ADVANCE time units. In particular,
1201 * an implementation must not create a temporary address with a zero
1202 * Preferred Lifetime.
1203 * Use age calculation as in addrconf_verify to avoid unnecessary
1204 * temporary addresses being generated.
1206 age
= (now
- tmp_tstamp
+ ADDRCONF_TIMER_FUZZ_MINUS
) / HZ
;
1207 if (tmp_prefered_lft
<= regen_advance
+ age
) {
1214 addr_flags
= IFA_F_TEMPORARY
;
1215 /* set in addrconf_prefix_rcv() */
1216 if (ifp
->flags
& IFA_F_OPTIMISTIC
)
1217 addr_flags
|= IFA_F_OPTIMISTIC
;
1219 ift
= ipv6_add_addr(idev
, &addr
, NULL
, tmp_plen
,
1220 ipv6_addr_scope(&addr
), addr_flags
,
1221 tmp_valid_lft
, tmp_prefered_lft
);
1225 pr_info("%s: retry temporary address regeneration\n", __func__
);
1227 write_lock_bh(&idev
->lock
);
1231 spin_lock_bh(&ift
->lock
);
1234 ift
->tstamp
= tmp_tstamp
;
1235 spin_unlock_bh(&ift
->lock
);
1237 addrconf_dad_start(ift
);
1245 * Choose an appropriate source address (RFC3484)
1248 IPV6_SADDR_RULE_INIT
= 0,
1249 IPV6_SADDR_RULE_LOCAL
,
1250 IPV6_SADDR_RULE_SCOPE
,
1251 IPV6_SADDR_RULE_PREFERRED
,
1252 #ifdef CONFIG_IPV6_MIP6
1253 IPV6_SADDR_RULE_HOA
,
1255 IPV6_SADDR_RULE_OIF
,
1256 IPV6_SADDR_RULE_LABEL
,
1257 IPV6_SADDR_RULE_PRIVACY
,
1258 IPV6_SADDR_RULE_ORCHID
,
1259 IPV6_SADDR_RULE_PREFIX
,
1260 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
1261 IPV6_SADDR_RULE_NOT_OPTIMISTIC
,
1266 struct ipv6_saddr_score
{
1269 struct inet6_ifaddr
*ifa
;
1270 DECLARE_BITMAP(scorebits
, IPV6_SADDR_RULE_MAX
);
1275 struct ipv6_saddr_dst
{
1276 const struct in6_addr
*addr
;
1283 static inline int ipv6_saddr_preferred(int type
)
1285 if (type
& (IPV6_ADDR_MAPPED
|IPV6_ADDR_COMPATv4
|IPV6_ADDR_LOOPBACK
))
1290 static inline bool ipv6_use_optimistic_addr(struct inet6_dev
*idev
)
1292 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
1293 return idev
&& idev
->cnf
.optimistic_dad
&& idev
->cnf
.use_optimistic
;
1299 static int ipv6_get_saddr_eval(struct net
*net
,
1300 struct ipv6_saddr_score
*score
,
1301 struct ipv6_saddr_dst
*dst
,
1306 if (i
<= score
->rule
) {
1308 case IPV6_SADDR_RULE_SCOPE
:
1309 ret
= score
->scopedist
;
1311 case IPV6_SADDR_RULE_PREFIX
:
1312 ret
= score
->matchlen
;
1315 ret
= !!test_bit(i
, score
->scorebits
);
1321 case IPV6_SADDR_RULE_INIT
:
1322 /* Rule 0: remember if hiscore is not ready yet */
1325 case IPV6_SADDR_RULE_LOCAL
:
1326 /* Rule 1: Prefer same address */
1327 ret
= ipv6_addr_equal(&score
->ifa
->addr
, dst
->addr
);
1329 case IPV6_SADDR_RULE_SCOPE
:
1330 /* Rule 2: Prefer appropriate scope
1335 * ---+--+-+---> scope
1337 * | d is scope of the destination.
1339 * | \ <- smaller scope is better if
1340 * B-15 | \ if scope is enough for destination.
1341 * | ret = B - scope (-1 <= scope >= d <= 15).
1343 * |/ <- greater is better
1344 * -C / if scope is not enough for destination.
1345 * /| ret = scope - C (-1 <= d < scope <= 15).
1347 * d - C - 1 < B -15 (for all -1 <= d <= 15).
1348 * C > d + 14 - B >= 15 + 14 - B = 29 - B.
1349 * Assume B = 0 and we get C > 29.
1351 ret
= __ipv6_addr_src_scope(score
->addr_type
);
1352 if (ret
>= dst
->scope
)
1355 ret
-= 128; /* 30 is enough */
1356 score
->scopedist
= ret
;
1358 case IPV6_SADDR_RULE_PREFERRED
:
1360 /* Rule 3: Avoid deprecated and optimistic addresses */
1361 u8 avoid
= IFA_F_DEPRECATED
;
1363 if (!ipv6_use_optimistic_addr(score
->ifa
->idev
))
1364 avoid
|= IFA_F_OPTIMISTIC
;
1365 ret
= ipv6_saddr_preferred(score
->addr_type
) ||
1366 !(score
->ifa
->flags
& avoid
);
1369 #ifdef CONFIG_IPV6_MIP6
1370 case IPV6_SADDR_RULE_HOA
:
1372 /* Rule 4: Prefer home address */
1373 int prefhome
= !(dst
->prefs
& IPV6_PREFER_SRC_COA
);
1374 ret
= !(score
->ifa
->flags
& IFA_F_HOMEADDRESS
) ^ prefhome
;
1378 case IPV6_SADDR_RULE_OIF
:
1379 /* Rule 5: Prefer outgoing interface */
1380 ret
= (!dst
->ifindex
||
1381 dst
->ifindex
== score
->ifa
->idev
->dev
->ifindex
);
1383 case IPV6_SADDR_RULE_LABEL
:
1384 /* Rule 6: Prefer matching label */
1385 ret
= ipv6_addr_label(net
,
1386 &score
->ifa
->addr
, score
->addr_type
,
1387 score
->ifa
->idev
->dev
->ifindex
) == dst
->label
;
1389 case IPV6_SADDR_RULE_PRIVACY
:
1391 /* Rule 7: Prefer public address
1392 * Note: prefer temporary address if use_tempaddr >= 2
1394 int preftmp
= dst
->prefs
& (IPV6_PREFER_SRC_PUBLIC
|IPV6_PREFER_SRC_TMP
) ?
1395 !!(dst
->prefs
& IPV6_PREFER_SRC_TMP
) :
1396 score
->ifa
->idev
->cnf
.use_tempaddr
>= 2;
1397 ret
= (!(score
->ifa
->flags
& IFA_F_TEMPORARY
)) ^ preftmp
;
1400 case IPV6_SADDR_RULE_ORCHID
:
1401 /* Rule 8-: Prefer ORCHID vs ORCHID or
1402 * non-ORCHID vs non-ORCHID
1404 ret
= !(ipv6_addr_orchid(&score
->ifa
->addr
) ^
1405 ipv6_addr_orchid(dst
->addr
));
1407 case IPV6_SADDR_RULE_PREFIX
:
1408 /* Rule 8: Use longest matching prefix */
1409 ret
= ipv6_addr_diff(&score
->ifa
->addr
, dst
->addr
);
1410 if (ret
> score
->ifa
->prefix_len
)
1411 ret
= score
->ifa
->prefix_len
;
1412 score
->matchlen
= ret
;
1414 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
1415 case IPV6_SADDR_RULE_NOT_OPTIMISTIC
:
1416 /* Optimistic addresses still have lower precedence than other
1417 * preferred addresses.
1419 ret
= !(score
->ifa
->flags
& IFA_F_OPTIMISTIC
);
1427 __set_bit(i
, score
->scorebits
);
1433 static int __ipv6_dev_get_saddr(struct net
*net
,
1434 struct ipv6_saddr_dst
*dst
,
1435 struct inet6_dev
*idev
,
1436 struct ipv6_saddr_score
*scores
,
1439 struct ipv6_saddr_score
*score
= &scores
[1 - hiscore_idx
], *hiscore
= &scores
[hiscore_idx
];
1441 read_lock_bh(&idev
->lock
);
1442 list_for_each_entry(score
->ifa
, &idev
->addr_list
, if_list
) {
1446 * - Tentative Address (RFC2462 section 5.4)
1447 * - A tentative address is not considered
1448 * "assigned to an interface" in the traditional
1449 * sense, unless it is also flagged as optimistic.
1450 * - Candidate Source Address (section 4)
1451 * - In any case, anycast addresses, multicast
1452 * addresses, and the unspecified address MUST
1453 * NOT be included in a candidate set.
1455 if ((score
->ifa
->flags
& IFA_F_TENTATIVE
) &&
1456 (!(score
->ifa
->flags
& IFA_F_OPTIMISTIC
)))
1459 score
->addr_type
= __ipv6_addr_type(&score
->ifa
->addr
);
1461 if (unlikely(score
->addr_type
== IPV6_ADDR_ANY
||
1462 score
->addr_type
& IPV6_ADDR_MULTICAST
)) {
1463 net_dbg_ratelimited("ADDRCONF: unspecified / multicast address assigned as unicast address on %s",
1469 bitmap_zero(score
->scorebits
, IPV6_SADDR_RULE_MAX
);
1471 for (i
= 0; i
< IPV6_SADDR_RULE_MAX
; i
++) {
1472 int minihiscore
, miniscore
;
1474 minihiscore
= ipv6_get_saddr_eval(net
, hiscore
, dst
, i
);
1475 miniscore
= ipv6_get_saddr_eval(net
, score
, dst
, i
);
1477 if (minihiscore
> miniscore
) {
1478 if (i
== IPV6_SADDR_RULE_SCOPE
&&
1479 score
->scopedist
> 0) {
1482 * each remaining entry
1483 * has too small (not enough)
1484 * scope, because ifa entries
1485 * are sorted by their scope
1491 } else if (minihiscore
< miniscore
) {
1493 in6_ifa_put(hiscore
->ifa
);
1495 in6_ifa_hold(score
->ifa
);
1497 swap(hiscore
, score
);
1498 hiscore_idx
= 1 - hiscore_idx
;
1500 /* restore our iterator */
1501 score
->ifa
= hiscore
->ifa
;
1508 read_unlock_bh(&idev
->lock
);
1512 int ipv6_dev_get_saddr(struct net
*net
, const struct net_device
*dst_dev
,
1513 const struct in6_addr
*daddr
, unsigned int prefs
,
1514 struct in6_addr
*saddr
)
1516 struct ipv6_saddr_score scores
[2], *hiscore
;
1517 struct ipv6_saddr_dst dst
;
1518 struct inet6_dev
*idev
;
1519 struct net_device
*dev
;
1521 bool use_oif_addr
= false;
1522 int hiscore_idx
= 0;
1524 dst_type
= __ipv6_addr_type(daddr
);
1526 dst
.ifindex
= dst_dev
? dst_dev
->ifindex
: 0;
1527 dst
.scope
= __ipv6_addr_src_scope(dst_type
);
1528 dst
.label
= ipv6_addr_label(net
, daddr
, dst_type
, dst
.ifindex
);
1531 scores
[hiscore_idx
].rule
= -1;
1532 scores
[hiscore_idx
].ifa
= NULL
;
1536 /* Candidate Source Address (section 4)
1537 * - multicast and link-local destination address,
1538 * the set of candidate source address MUST only
1539 * include addresses assigned to interfaces
1540 * belonging to the same link as the outgoing
1542 * (- For site-local destination addresses, the
1543 * set of candidate source addresses MUST only
1544 * include addresses assigned to interfaces
1545 * belonging to the same site as the outgoing
1547 * - "It is RECOMMENDED that the candidate source addresses
1548 * be the set of unicast addresses assigned to the
1549 * interface that will be used to send to the destination
1550 * (the 'outgoing' interface)." (RFC 6724)
1553 idev
= __in6_dev_get(dst_dev
);
1554 if ((dst_type
& IPV6_ADDR_MULTICAST
) ||
1555 dst
.scope
<= IPV6_ADDR_SCOPE_LINKLOCAL
||
1556 (idev
&& idev
->cnf
.use_oif_addrs_only
)) {
1557 use_oif_addr
= true;
1563 hiscore_idx
= __ipv6_dev_get_saddr(net
, &dst
, idev
, scores
, hiscore_idx
);
1565 for_each_netdev_rcu(net
, dev
) {
1566 idev
= __in6_dev_get(dev
);
1569 hiscore_idx
= __ipv6_dev_get_saddr(net
, &dst
, idev
, scores
, hiscore_idx
);
1574 hiscore
= &scores
[hiscore_idx
];
1576 return -EADDRNOTAVAIL
;
1578 *saddr
= hiscore
->ifa
->addr
;
1579 in6_ifa_put(hiscore
->ifa
);
1582 EXPORT_SYMBOL(ipv6_dev_get_saddr
);
1584 int __ipv6_get_lladdr(struct inet6_dev
*idev
, struct in6_addr
*addr
,
1587 struct inet6_ifaddr
*ifp
;
1588 int err
= -EADDRNOTAVAIL
;
1590 list_for_each_entry_reverse(ifp
, &idev
->addr_list
, if_list
) {
1591 if (ifp
->scope
> IFA_LINK
)
1593 if (ifp
->scope
== IFA_LINK
&&
1594 !(ifp
->flags
& banned_flags
)) {
1603 int ipv6_get_lladdr(struct net_device
*dev
, struct in6_addr
*addr
,
1606 struct inet6_dev
*idev
;
1607 int err
= -EADDRNOTAVAIL
;
1610 idev
= __in6_dev_get(dev
);
1612 read_lock_bh(&idev
->lock
);
1613 err
= __ipv6_get_lladdr(idev
, addr
, banned_flags
);
1614 read_unlock_bh(&idev
->lock
);
1620 static int ipv6_count_addresses(struct inet6_dev
*idev
)
1623 struct inet6_ifaddr
*ifp
;
1625 read_lock_bh(&idev
->lock
);
1626 list_for_each_entry(ifp
, &idev
->addr_list
, if_list
)
1628 read_unlock_bh(&idev
->lock
);
1632 int ipv6_chk_addr(struct net
*net
, const struct in6_addr
*addr
,
1633 const struct net_device
*dev
, int strict
)
1635 return ipv6_chk_addr_and_flags(net
, addr
, dev
, strict
, IFA_F_TENTATIVE
);
1637 EXPORT_SYMBOL(ipv6_chk_addr
);
1639 int ipv6_chk_addr_and_flags(struct net
*net
, const struct in6_addr
*addr
,
1640 const struct net_device
*dev
, int strict
,
1643 struct inet6_ifaddr
*ifp
;
1644 unsigned int hash
= inet6_addr_hash(addr
);
1648 hlist_for_each_entry_rcu(ifp
, &inet6_addr_lst
[hash
], addr_lst
) {
1649 if (!net_eq(dev_net(ifp
->idev
->dev
), net
))
1651 /* Decouple optimistic from tentative for evaluation here.
1652 * Ban optimistic addresses explicitly, when required.
1654 ifp_flags
= (ifp
->flags
&IFA_F_OPTIMISTIC
)
1655 ? (ifp
->flags
&~IFA_F_TENTATIVE
)
1657 if (ipv6_addr_equal(&ifp
->addr
, addr
) &&
1658 !(ifp_flags
&banned_flags
) &&
1659 (!dev
|| ifp
->idev
->dev
== dev
||
1660 !(ifp
->scope
&(IFA_LINK
|IFA_HOST
) || strict
))) {
1661 rcu_read_unlock_bh();
1666 rcu_read_unlock_bh();
1669 EXPORT_SYMBOL(ipv6_chk_addr_and_flags
);
1671 static bool ipv6_chk_same_addr(struct net
*net
, const struct in6_addr
*addr
,
1672 struct net_device
*dev
)
1674 unsigned int hash
= inet6_addr_hash(addr
);
1675 struct inet6_ifaddr
*ifp
;
1677 hlist_for_each_entry(ifp
, &inet6_addr_lst
[hash
], addr_lst
) {
1678 if (!net_eq(dev_net(ifp
->idev
->dev
), net
))
1680 if (ipv6_addr_equal(&ifp
->addr
, addr
)) {
1681 if (!dev
|| ifp
->idev
->dev
== dev
)
1688 /* Compares an address/prefix_len with addresses on device @dev.
1689 * If one is found it returns true.
1691 bool ipv6_chk_custom_prefix(const struct in6_addr
*addr
,
1692 const unsigned int prefix_len
, struct net_device
*dev
)
1694 struct inet6_dev
*idev
;
1695 struct inet6_ifaddr
*ifa
;
1699 idev
= __in6_dev_get(dev
);
1701 read_lock_bh(&idev
->lock
);
1702 list_for_each_entry(ifa
, &idev
->addr_list
, if_list
) {
1703 ret
= ipv6_prefix_equal(addr
, &ifa
->addr
, prefix_len
);
1707 read_unlock_bh(&idev
->lock
);
1713 EXPORT_SYMBOL(ipv6_chk_custom_prefix
);
1715 int ipv6_chk_prefix(const struct in6_addr
*addr
, struct net_device
*dev
)
1717 struct inet6_dev
*idev
;
1718 struct inet6_ifaddr
*ifa
;
1723 idev
= __in6_dev_get(dev
);
1725 read_lock_bh(&idev
->lock
);
1726 list_for_each_entry(ifa
, &idev
->addr_list
, if_list
) {
1727 onlink
= ipv6_prefix_equal(addr
, &ifa
->addr
,
1732 read_unlock_bh(&idev
->lock
);
1737 EXPORT_SYMBOL(ipv6_chk_prefix
);
1739 struct inet6_ifaddr
*ipv6_get_ifaddr(struct net
*net
, const struct in6_addr
*addr
,
1740 struct net_device
*dev
, int strict
)
1742 struct inet6_ifaddr
*ifp
, *result
= NULL
;
1743 unsigned int hash
= inet6_addr_hash(addr
);
1746 hlist_for_each_entry_rcu_bh(ifp
, &inet6_addr_lst
[hash
], addr_lst
) {
1747 if (!net_eq(dev_net(ifp
->idev
->dev
), net
))
1749 if (ipv6_addr_equal(&ifp
->addr
, addr
)) {
1750 if (!dev
|| ifp
->idev
->dev
== dev
||
1751 !(ifp
->scope
&(IFA_LINK
|IFA_HOST
) || strict
)) {
1758 rcu_read_unlock_bh();
1763 /* Gets referenced address, destroys ifaddr */
1765 static void addrconf_dad_stop(struct inet6_ifaddr
*ifp
, int dad_failed
)
1767 if (ifp
->flags
&IFA_F_PERMANENT
) {
1768 spin_lock_bh(&ifp
->lock
);
1769 addrconf_del_dad_work(ifp
);
1770 ifp
->flags
|= IFA_F_TENTATIVE
;
1772 ifp
->flags
|= IFA_F_DADFAILED
;
1773 spin_unlock_bh(&ifp
->lock
);
1775 ipv6_ifa_notify(0, ifp
);
1777 } else if (ifp
->flags
&IFA_F_TEMPORARY
) {
1778 struct inet6_ifaddr
*ifpub
;
1779 spin_lock_bh(&ifp
->lock
);
1782 in6_ifa_hold(ifpub
);
1783 spin_unlock_bh(&ifp
->lock
);
1784 ipv6_create_tempaddr(ifpub
, ifp
);
1787 spin_unlock_bh(&ifp
->lock
);
1795 static int addrconf_dad_end(struct inet6_ifaddr
*ifp
)
1799 spin_lock_bh(&ifp
->lock
);
1800 if (ifp
->state
== INET6_IFADDR_STATE_DAD
) {
1801 ifp
->state
= INET6_IFADDR_STATE_POSTDAD
;
1804 spin_unlock_bh(&ifp
->lock
);
1809 void addrconf_dad_failure(struct inet6_ifaddr
*ifp
)
1811 struct in6_addr addr
;
1812 struct inet6_dev
*idev
= ifp
->idev
;
1813 struct net
*net
= dev_net(ifp
->idev
->dev
);
1815 if (addrconf_dad_end(ifp
)) {
1820 net_info_ratelimited("%s: IPv6 duplicate address %pI6c detected!\n",
1821 ifp
->idev
->dev
->name
, &ifp
->addr
);
1823 spin_lock_bh(&ifp
->lock
);
1825 if (ifp
->flags
& IFA_F_STABLE_PRIVACY
) {
1826 int scope
= ifp
->scope
;
1827 u32 flags
= ifp
->flags
;
1828 struct in6_addr new_addr
;
1829 struct inet6_ifaddr
*ifp2
;
1830 u32 valid_lft
, preferred_lft
;
1831 int pfxlen
= ifp
->prefix_len
;
1832 int retries
= ifp
->stable_privacy_retry
+ 1;
1834 if (retries
> net
->ipv6
.sysctl
.idgen_retries
) {
1835 net_info_ratelimited("%s: privacy stable address generation failed because of DAD conflicts!\n",
1836 ifp
->idev
->dev
->name
);
1840 new_addr
= ifp
->addr
;
1841 if (ipv6_generate_stable_address(&new_addr
, retries
,
1845 valid_lft
= ifp
->valid_lft
;
1846 preferred_lft
= ifp
->prefered_lft
;
1848 spin_unlock_bh(&ifp
->lock
);
1850 if (idev
->cnf
.max_addresses
&&
1851 ipv6_count_addresses(idev
) >=
1852 idev
->cnf
.max_addresses
)
1855 net_info_ratelimited("%s: generating new stable privacy address because of DAD conflict\n",
1856 ifp
->idev
->dev
->name
);
1858 ifp2
= ipv6_add_addr(idev
, &new_addr
, NULL
, pfxlen
,
1859 scope
, flags
, valid_lft
,
1864 spin_lock_bh(&ifp2
->lock
);
1865 ifp2
->stable_privacy_retry
= retries
;
1866 ifp2
->state
= INET6_IFADDR_STATE_PREDAD
;
1867 spin_unlock_bh(&ifp2
->lock
);
1869 addrconf_mod_dad_work(ifp2
, net
->ipv6
.sysctl
.idgen_delay
);
1872 spin_lock_bh(&ifp
->lock
);
1873 } else if (idev
->cnf
.accept_dad
> 1 && !idev
->cnf
.disable_ipv6
) {
1874 addr
.s6_addr32
[0] = htonl(0xfe800000);
1875 addr
.s6_addr32
[1] = 0;
1877 if (!ipv6_generate_eui64(addr
.s6_addr
+ 8, idev
->dev
) &&
1878 ipv6_addr_equal(&ifp
->addr
, &addr
)) {
1879 /* DAD failed for link-local based on MAC address */
1880 idev
->cnf
.disable_ipv6
= 1;
1882 pr_info("%s: IPv6 being disabled!\n",
1883 ifp
->idev
->dev
->name
);
1888 /* transition from _POSTDAD to _ERRDAD */
1889 ifp
->state
= INET6_IFADDR_STATE_ERRDAD
;
1890 spin_unlock_bh(&ifp
->lock
);
1892 addrconf_mod_dad_work(ifp
, 0);
1895 /* Join to solicited addr multicast group.
1896 * caller must hold RTNL */
1897 void addrconf_join_solict(struct net_device
*dev
, const struct in6_addr
*addr
)
1899 struct in6_addr maddr
;
1901 if (dev
->flags
&(IFF_LOOPBACK
|IFF_NOARP
))
1904 addrconf_addr_solict_mult(addr
, &maddr
);
1905 ipv6_dev_mc_inc(dev
, &maddr
);
1908 /* caller must hold RTNL */
1909 void addrconf_leave_solict(struct inet6_dev
*idev
, const struct in6_addr
*addr
)
1911 struct in6_addr maddr
;
1913 if (idev
->dev
->flags
&(IFF_LOOPBACK
|IFF_NOARP
))
1916 addrconf_addr_solict_mult(addr
, &maddr
);
1917 __ipv6_dev_mc_dec(idev
, &maddr
);
1920 /* caller must hold RTNL */
1921 static void addrconf_join_anycast(struct inet6_ifaddr
*ifp
)
1923 struct in6_addr addr
;
1925 if (ifp
->prefix_len
>= 127) /* RFC 6164 */
1927 ipv6_addr_prefix(&addr
, &ifp
->addr
, ifp
->prefix_len
);
1928 if (ipv6_addr_any(&addr
))
1930 __ipv6_dev_ac_inc(ifp
->idev
, &addr
);
1933 /* caller must hold RTNL */
1934 static void addrconf_leave_anycast(struct inet6_ifaddr
*ifp
)
1936 struct in6_addr addr
;
1938 if (ifp
->prefix_len
>= 127) /* RFC 6164 */
1940 ipv6_addr_prefix(&addr
, &ifp
->addr
, ifp
->prefix_len
);
1941 if (ipv6_addr_any(&addr
))
1943 __ipv6_dev_ac_dec(ifp
->idev
, &addr
);
1946 static int addrconf_ifid_eui48(u8
*eui
, struct net_device
*dev
)
1948 if (dev
->addr_len
!= ETH_ALEN
)
1950 memcpy(eui
, dev
->dev_addr
, 3);
1951 memcpy(eui
+ 5, dev
->dev_addr
+ 3, 3);
1954 * The zSeries OSA network cards can be shared among various
1955 * OS instances, but the OSA cards have only one MAC address.
1956 * This leads to duplicate address conflicts in conjunction
1957 * with IPv6 if more than one instance uses the same card.
1959 * The driver for these cards can deliver a unique 16-bit
1960 * identifier for each instance sharing the same card. It is
1961 * placed instead of 0xFFFE in the interface identifier. The
1962 * "u" bit of the interface identifier is not inverted in this
1963 * case. Hence the resulting interface identifier has local
1964 * scope according to RFC2373.
1967 eui
[3] = (dev
->dev_id
>> 8) & 0xFF;
1968 eui
[4] = dev
->dev_id
& 0xFF;
1977 static int addrconf_ifid_eui64(u8
*eui
, struct net_device
*dev
)
1979 if (dev
->addr_len
!= IEEE802154_ADDR_LEN
)
1981 memcpy(eui
, dev
->dev_addr
, 8);
1986 static int addrconf_ifid_ieee1394(u8
*eui
, struct net_device
*dev
)
1988 union fwnet_hwaddr
*ha
;
1990 if (dev
->addr_len
!= FWNET_ALEN
)
1993 ha
= (union fwnet_hwaddr
*)dev
->dev_addr
;
1995 memcpy(eui
, &ha
->uc
.uniq_id
, sizeof(ha
->uc
.uniq_id
));
2000 static int addrconf_ifid_arcnet(u8
*eui
, struct net_device
*dev
)
2002 /* XXX: inherit EUI-64 from other interface -- yoshfuji */
2003 if (dev
->addr_len
!= ARCNET_ALEN
)
2006 eui
[7] = *(u8
*)dev
->dev_addr
;
2010 static int addrconf_ifid_infiniband(u8
*eui
, struct net_device
*dev
)
2012 if (dev
->addr_len
!= INFINIBAND_ALEN
)
2014 memcpy(eui
, dev
->dev_addr
+ 12, 8);
2019 static int __ipv6_isatap_ifid(u8
*eui
, __be32 addr
)
2023 eui
[0] = (ipv4_is_zeronet(addr
) || ipv4_is_private_10(addr
) ||
2024 ipv4_is_loopback(addr
) || ipv4_is_linklocal_169(addr
) ||
2025 ipv4_is_private_172(addr
) || ipv4_is_test_192(addr
) ||
2026 ipv4_is_anycast_6to4(addr
) || ipv4_is_private_192(addr
) ||
2027 ipv4_is_test_198(addr
) || ipv4_is_multicast(addr
) ||
2028 ipv4_is_lbcast(addr
)) ? 0x00 : 0x02;
2032 memcpy(eui
+ 4, &addr
, 4);
2036 static int addrconf_ifid_sit(u8
*eui
, struct net_device
*dev
)
2038 if (dev
->priv_flags
& IFF_ISATAP
)
2039 return __ipv6_isatap_ifid(eui
, *(__be32
*)dev
->dev_addr
);
2043 static int addrconf_ifid_gre(u8
*eui
, struct net_device
*dev
)
2045 return __ipv6_isatap_ifid(eui
, *(__be32
*)dev
->dev_addr
);
2048 static int addrconf_ifid_ip6tnl(u8
*eui
, struct net_device
*dev
)
2050 memcpy(eui
, dev
->perm_addr
, 3);
2051 memcpy(eui
+ 5, dev
->perm_addr
+ 3, 3);
2058 static int ipv6_generate_eui64(u8
*eui
, struct net_device
*dev
)
2060 switch (dev
->type
) {
2063 return addrconf_ifid_eui48(eui
, dev
);
2065 return addrconf_ifid_arcnet(eui
, dev
);
2066 case ARPHRD_INFINIBAND
:
2067 return addrconf_ifid_infiniband(eui
, dev
);
2069 return addrconf_ifid_sit(eui
, dev
);
2071 return addrconf_ifid_gre(eui
, dev
);
2072 case ARPHRD_6LOWPAN
:
2073 case ARPHRD_IEEE802154
:
2074 return addrconf_ifid_eui64(eui
, dev
);
2075 case ARPHRD_IEEE1394
:
2076 return addrconf_ifid_ieee1394(eui
, dev
);
2077 case ARPHRD_TUNNEL6
:
2078 return addrconf_ifid_ip6tnl(eui
, dev
);
2083 static int ipv6_inherit_eui64(u8
*eui
, struct inet6_dev
*idev
)
2086 struct inet6_ifaddr
*ifp
;
2088 read_lock_bh(&idev
->lock
);
2089 list_for_each_entry_reverse(ifp
, &idev
->addr_list
, if_list
) {
2090 if (ifp
->scope
> IFA_LINK
)
2092 if (ifp
->scope
== IFA_LINK
&& !(ifp
->flags
&IFA_F_TENTATIVE
)) {
2093 memcpy(eui
, ifp
->addr
.s6_addr
+8, 8);
2098 read_unlock_bh(&idev
->lock
);
2102 /* (re)generation of randomized interface identifier (RFC 3041 3.2, 3.5) */
2103 static void __ipv6_regen_rndid(struct inet6_dev
*idev
)
2106 get_random_bytes(idev
->rndid
, sizeof(idev
->rndid
));
2107 idev
->rndid
[0] &= ~0x02;
2110 * <draft-ietf-ipngwg-temp-addresses-v2-00.txt>:
2111 * check if generated address is not inappropriate
2113 * - Reserved subnet anycast (RFC 2526)
2114 * 11111101 11....11 1xxxxxxx
2115 * - ISATAP (RFC4214) 6.1
2116 * 00-00-5E-FE-xx-xx-xx-xx
2118 * - XXX: already assigned to an address on the device
2120 if (idev
->rndid
[0] == 0xfd &&
2121 (idev
->rndid
[1]&idev
->rndid
[2]&idev
->rndid
[3]&idev
->rndid
[4]&idev
->rndid
[5]&idev
->rndid
[6]) == 0xff &&
2122 (idev
->rndid
[7]&0x80))
2124 if ((idev
->rndid
[0]|idev
->rndid
[1]) == 0) {
2125 if (idev
->rndid
[2] == 0x5e && idev
->rndid
[3] == 0xfe)
2127 if ((idev
->rndid
[2]|idev
->rndid
[3]|idev
->rndid
[4]|idev
->rndid
[5]|idev
->rndid
[6]|idev
->rndid
[7]) == 0x00)
2132 static void ipv6_regen_rndid(unsigned long data
)
2134 struct inet6_dev
*idev
= (struct inet6_dev
*) data
;
2135 unsigned long expires
;
2138 write_lock_bh(&idev
->lock
);
2143 __ipv6_regen_rndid(idev
);
2146 idev
->cnf
.temp_prefered_lft
* HZ
-
2147 idev
->cnf
.regen_max_retry
* idev
->cnf
.dad_transmits
*
2148 NEIGH_VAR(idev
->nd_parms
, RETRANS_TIME
) -
2149 idev
->cnf
.max_desync_factor
* HZ
;
2150 if (time_before(expires
, jiffies
)) {
2151 pr_warn("%s: too short regeneration interval; timer disabled for %s\n",
2152 __func__
, idev
->dev
->name
);
2156 if (!mod_timer(&idev
->regen_timer
, expires
))
2160 write_unlock_bh(&idev
->lock
);
2161 rcu_read_unlock_bh();
2165 static void __ipv6_try_regen_rndid(struct inet6_dev
*idev
, struct in6_addr
*tmpaddr
)
2167 if (tmpaddr
&& memcmp(idev
->rndid
, &tmpaddr
->s6_addr
[8], 8) == 0)
2168 __ipv6_regen_rndid(idev
);
2176 addrconf_prefix_route(struct in6_addr
*pfx
, int plen
, struct net_device
*dev
,
2177 unsigned long expires
, u32 flags
)
2179 struct fib6_config cfg
= {
2180 .fc_table
= RT6_TABLE_PREFIX
,
2181 .fc_metric
= IP6_RT_PRIO_ADDRCONF
,
2182 .fc_ifindex
= dev
->ifindex
,
2183 .fc_expires
= expires
,
2185 .fc_flags
= RTF_UP
| flags
,
2186 .fc_nlinfo
.nl_net
= dev_net(dev
),
2187 .fc_protocol
= RTPROT_KERNEL
,
2192 /* Prevent useless cloning on PtP SIT.
2193 This thing is done here expecting that the whole
2194 class of non-broadcast devices need not cloning.
2196 #if IS_ENABLED(CONFIG_IPV6_SIT)
2197 if (dev
->type
== ARPHRD_SIT
&& (dev
->flags
& IFF_POINTOPOINT
))
2198 cfg
.fc_flags
|= RTF_NONEXTHOP
;
2201 ip6_route_add(&cfg
);
2205 static struct rt6_info
*addrconf_get_prefix_route(const struct in6_addr
*pfx
,
2207 const struct net_device
*dev
,
2208 u32 flags
, u32 noflags
)
2210 struct fib6_node
*fn
;
2211 struct rt6_info
*rt
= NULL
;
2212 struct fib6_table
*table
;
2214 table
= fib6_get_table(dev_net(dev
), RT6_TABLE_PREFIX
);
2218 read_lock_bh(&table
->tb6_lock
);
2219 fn
= fib6_locate(&table
->tb6_root
, pfx
, plen
, NULL
, 0);
2223 noflags
|= RTF_CACHE
;
2224 for (rt
= fn
->leaf
; rt
; rt
= rt
->dst
.rt6_next
) {
2225 if (rt
->dst
.dev
->ifindex
!= dev
->ifindex
)
2227 if ((rt
->rt6i_flags
& flags
) != flags
)
2229 if ((rt
->rt6i_flags
& noflags
) != 0)
2235 read_unlock_bh(&table
->tb6_lock
);
2240 /* Create "default" multicast route to the interface */
2242 static void addrconf_add_mroute(struct net_device
*dev
)
2244 struct fib6_config cfg
= {
2245 .fc_table
= RT6_TABLE_LOCAL
,
2246 .fc_metric
= IP6_RT_PRIO_ADDRCONF
,
2247 .fc_ifindex
= dev
->ifindex
,
2250 .fc_nlinfo
.nl_net
= dev_net(dev
),
2253 ipv6_addr_set(&cfg
.fc_dst
, htonl(0xFF000000), 0, 0, 0);
2255 ip6_route_add(&cfg
);
2258 static struct inet6_dev
*addrconf_add_dev(struct net_device
*dev
)
2260 struct inet6_dev
*idev
;
2264 idev
= ipv6_find_idev(dev
);
2266 return ERR_PTR(-ENOBUFS
);
2268 if (idev
->cnf
.disable_ipv6
)
2269 return ERR_PTR(-EACCES
);
2271 /* Add default multicast route */
2272 if (!(dev
->flags
& IFF_LOOPBACK
))
2273 addrconf_add_mroute(dev
);
2278 static void manage_tempaddrs(struct inet6_dev
*idev
,
2279 struct inet6_ifaddr
*ifp
,
2280 __u32 valid_lft
, __u32 prefered_lft
,
2281 bool create
, unsigned long now
)
2284 struct inet6_ifaddr
*ift
;
2286 read_lock_bh(&idev
->lock
);
2287 /* update all temporary addresses in the list */
2288 list_for_each_entry(ift
, &idev
->tempaddr_list
, tmp_list
) {
2289 int age
, max_valid
, max_prefered
;
2291 if (ifp
!= ift
->ifpub
)
2294 /* RFC 4941 section 3.3:
2295 * If a received option will extend the lifetime of a public
2296 * address, the lifetimes of temporary addresses should
2297 * be extended, subject to the overall constraint that no
2298 * temporary addresses should ever remain "valid" or "preferred"
2299 * for a time longer than (TEMP_VALID_LIFETIME) or
2300 * (TEMP_PREFERRED_LIFETIME - DESYNC_FACTOR), respectively.
2302 age
= (now
- ift
->cstamp
) / HZ
;
2303 max_valid
= idev
->cnf
.temp_valid_lft
- age
;
2307 max_prefered
= idev
->cnf
.temp_prefered_lft
-
2308 idev
->cnf
.max_desync_factor
- age
;
2309 if (max_prefered
< 0)
2312 if (valid_lft
> max_valid
)
2313 valid_lft
= max_valid
;
2315 if (prefered_lft
> max_prefered
)
2316 prefered_lft
= max_prefered
;
2318 spin_lock(&ift
->lock
);
2320 ift
->valid_lft
= valid_lft
;
2321 ift
->prefered_lft
= prefered_lft
;
2323 if (prefered_lft
> 0)
2324 ift
->flags
&= ~IFA_F_DEPRECATED
;
2326 spin_unlock(&ift
->lock
);
2327 if (!(flags
&IFA_F_TENTATIVE
))
2328 ipv6_ifa_notify(0, ift
);
2331 if ((create
|| list_empty(&idev
->tempaddr_list
)) &&
2332 idev
->cnf
.use_tempaddr
> 0) {
2333 /* When a new public address is created as described
2334 * in [ADDRCONF], also create a new temporary address.
2335 * Also create a temporary address if it's enabled but
2336 * no temporary address currently exists.
2338 read_unlock_bh(&idev
->lock
);
2339 ipv6_create_tempaddr(ifp
, NULL
);
2341 read_unlock_bh(&idev
->lock
);
2345 void addrconf_prefix_rcv(struct net_device
*dev
, u8
*opt
, int len
, bool sllao
)
2347 struct prefix_info
*pinfo
;
2352 struct inet6_dev
*in6_dev
;
2353 struct net
*net
= dev_net(dev
);
2355 pinfo
= (struct prefix_info
*) opt
;
2357 if (len
< sizeof(struct prefix_info
)) {
2358 ADBG("addrconf: prefix option too short\n");
2363 * Validation checks ([ADDRCONF], page 19)
2366 addr_type
= ipv6_addr_type(&pinfo
->prefix
);
2368 if (addr_type
& (IPV6_ADDR_MULTICAST
|IPV6_ADDR_LINKLOCAL
))
2371 valid_lft
= ntohl(pinfo
->valid
);
2372 prefered_lft
= ntohl(pinfo
->prefered
);
2374 if (prefered_lft
> valid_lft
) {
2375 net_warn_ratelimited("addrconf: prefix option has invalid lifetime\n");
2379 in6_dev
= in6_dev_get(dev
);
2382 net_dbg_ratelimited("addrconf: device %s not configured\n",
2388 * Two things going on here:
2389 * 1) Add routes for on-link prefixes
2390 * 2) Configure prefixes with the auto flag set
2393 if (pinfo
->onlink
) {
2394 struct rt6_info
*rt
;
2395 unsigned long rt_expires
;
2397 /* Avoid arithmetic overflow. Really, we could
2398 * save rt_expires in seconds, likely valid_lft,
2399 * but it would require division in fib gc, that it
2403 rt_expires
= addrconf_timeout_fixup(valid_lft
, HZ
);
2405 rt_expires
= addrconf_timeout_fixup(valid_lft
, USER_HZ
);
2407 if (addrconf_finite_timeout(rt_expires
))
2410 rt
= addrconf_get_prefix_route(&pinfo
->prefix
,
2413 RTF_ADDRCONF
| RTF_PREFIX_RT
,
2414 RTF_GATEWAY
| RTF_DEFAULT
);
2417 /* Autoconf prefix route */
2418 if (valid_lft
== 0) {
2421 } else if (addrconf_finite_timeout(rt_expires
)) {
2423 rt6_set_expires(rt
, jiffies
+ rt_expires
);
2425 rt6_clean_expires(rt
);
2427 } else if (valid_lft
) {
2428 clock_t expires
= 0;
2429 int flags
= RTF_ADDRCONF
| RTF_PREFIX_RT
;
2430 if (addrconf_finite_timeout(rt_expires
)) {
2432 flags
|= RTF_EXPIRES
;
2433 expires
= jiffies_to_clock_t(rt_expires
);
2435 addrconf_prefix_route(&pinfo
->prefix
, pinfo
->prefix_len
,
2436 dev
, expires
, flags
);
2441 /* Try to figure out our local address for this prefix */
2443 if (pinfo
->autoconf
&& in6_dev
->cnf
.autoconf
) {
2444 struct inet6_ifaddr
*ifp
;
2445 struct in6_addr addr
;
2446 int create
= 0, update_lft
= 0;
2447 bool tokenized
= false;
2449 if (pinfo
->prefix_len
== 64) {
2450 memcpy(&addr
, &pinfo
->prefix
, 8);
2452 if (!ipv6_addr_any(&in6_dev
->token
)) {
2453 read_lock_bh(&in6_dev
->lock
);
2454 memcpy(addr
.s6_addr
+ 8,
2455 in6_dev
->token
.s6_addr
+ 8, 8);
2456 read_unlock_bh(&in6_dev
->lock
);
2458 } else if (in6_dev
->addr_gen_mode
==
2459 IN6_ADDR_GEN_MODE_STABLE_PRIVACY
&&
2460 !ipv6_generate_stable_address(&addr
, 0,
2462 addr_flags
|= IFA_F_STABLE_PRIVACY
;
2464 } else if (ipv6_generate_eui64(addr
.s6_addr
+ 8, dev
) &&
2465 ipv6_inherit_eui64(addr
.s6_addr
+ 8, in6_dev
)) {
2466 in6_dev_put(in6_dev
);
2471 net_dbg_ratelimited("IPv6 addrconf: prefix with wrong length %d\n",
2473 in6_dev_put(in6_dev
);
2478 ifp
= ipv6_get_ifaddr(net
, &addr
, dev
, 1);
2480 if (!ifp
&& valid_lft
) {
2481 int max_addresses
= in6_dev
->cnf
.max_addresses
;
2483 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
2484 if (in6_dev
->cnf
.optimistic_dad
&&
2485 !net
->ipv6
.devconf_all
->forwarding
&& sllao
)
2486 addr_flags
= IFA_F_OPTIMISTIC
;
2489 /* Do not allow to create too much of autoconfigured
2490 * addresses; this would be too easy way to crash kernel.
2492 if (!max_addresses
||
2493 ipv6_count_addresses(in6_dev
) < max_addresses
)
2494 ifp
= ipv6_add_addr(in6_dev
, &addr
, NULL
,
2496 addr_type
&IPV6_ADDR_SCOPE_MASK
,
2497 addr_flags
, valid_lft
,
2500 if (IS_ERR_OR_NULL(ifp
)) {
2501 in6_dev_put(in6_dev
);
2507 spin_lock_bh(&ifp
->lock
);
2508 ifp
->flags
|= IFA_F_MANAGETEMPADDR
;
2509 ifp
->cstamp
= jiffies
;
2510 ifp
->tokenized
= tokenized
;
2511 spin_unlock_bh(&ifp
->lock
);
2512 addrconf_dad_start(ifp
);
2520 /* update lifetime (RFC2462 5.5.3 e) */
2521 spin_lock_bh(&ifp
->lock
);
2523 if (ifp
->valid_lft
> (now
- ifp
->tstamp
) / HZ
)
2524 stored_lft
= ifp
->valid_lft
- (now
- ifp
->tstamp
) / HZ
;
2527 if (!update_lft
&& !create
&& stored_lft
) {
2528 const u32 minimum_lft
= min_t(u32
,
2529 stored_lft
, MIN_VALID_LIFETIME
);
2530 valid_lft
= max(valid_lft
, minimum_lft
);
2532 /* RFC4862 Section 5.5.3e:
2533 * "Note that the preferred lifetime of the
2534 * corresponding address is always reset to
2535 * the Preferred Lifetime in the received
2536 * Prefix Information option, regardless of
2537 * whether the valid lifetime is also reset or
2540 * So we should always update prefered_lft here.
2546 ifp
->valid_lft
= valid_lft
;
2547 ifp
->prefered_lft
= prefered_lft
;
2550 ifp
->flags
&= ~IFA_F_DEPRECATED
;
2551 spin_unlock_bh(&ifp
->lock
);
2553 if (!(flags
&IFA_F_TENTATIVE
))
2554 ipv6_ifa_notify(0, ifp
);
2556 spin_unlock_bh(&ifp
->lock
);
2558 manage_tempaddrs(in6_dev
, ifp
, valid_lft
, prefered_lft
,
2565 inet6_prefix_notify(RTM_NEWPREFIX
, in6_dev
, pinfo
);
2566 in6_dev_put(in6_dev
);
2570 * Set destination address.
2571 * Special case for SIT interfaces where we create a new "virtual"
2574 int addrconf_set_dstaddr(struct net
*net
, void __user
*arg
)
2576 struct in6_ifreq ireq
;
2577 struct net_device
*dev
;
2583 if (copy_from_user(&ireq
, arg
, sizeof(struct in6_ifreq
)))
2586 dev
= __dev_get_by_index(net
, ireq
.ifr6_ifindex
);
2592 #if IS_ENABLED(CONFIG_IPV6_SIT)
2593 if (dev
->type
== ARPHRD_SIT
) {
2594 const struct net_device_ops
*ops
= dev
->netdev_ops
;
2596 struct ip_tunnel_parm p
;
2598 err
= -EADDRNOTAVAIL
;
2599 if (!(ipv6_addr_type(&ireq
.ifr6_addr
) & IPV6_ADDR_COMPATv4
))
2602 memset(&p
, 0, sizeof(p
));
2603 p
.iph
.daddr
= ireq
.ifr6_addr
.s6_addr32
[3];
2607 p
.iph
.protocol
= IPPROTO_IPV6
;
2609 ifr
.ifr_ifru
.ifru_data
= (__force
void __user
*)&p
;
2611 if (ops
->ndo_do_ioctl
) {
2612 mm_segment_t oldfs
= get_fs();
2615 err
= ops
->ndo_do_ioctl(dev
, &ifr
, SIOCADDTUNNEL
);
2622 dev
= __dev_get_by_name(net
, p
.name
);
2625 err
= dev_open(dev
);
2635 static int ipv6_mc_config(struct sock
*sk
, bool join
,
2636 const struct in6_addr
*addr
, int ifindex
)
2644 ret
= ipv6_sock_mc_join(sk
, ifindex
, addr
);
2646 ret
= ipv6_sock_mc_drop(sk
, ifindex
, addr
);
2653 * Manual configuration of address on an interface
2655 static int inet6_addr_add(struct net
*net
, int ifindex
,
2656 const struct in6_addr
*pfx
,
2657 const struct in6_addr
*peer_pfx
,
2658 unsigned int plen
, __u32 ifa_flags
,
2659 __u32 prefered_lft
, __u32 valid_lft
)
2661 struct inet6_ifaddr
*ifp
;
2662 struct inet6_dev
*idev
;
2663 struct net_device
*dev
;
2664 unsigned long timeout
;
2674 /* check the lifetime */
2675 if (!valid_lft
|| prefered_lft
> valid_lft
)
2678 if (ifa_flags
& IFA_F_MANAGETEMPADDR
&& plen
!= 64)
2681 dev
= __dev_get_by_index(net
, ifindex
);
2685 idev
= addrconf_add_dev(dev
);
2687 return PTR_ERR(idev
);
2689 if (ifa_flags
& IFA_F_MCAUTOJOIN
) {
2690 int ret
= ipv6_mc_config(net
->ipv6
.mc_autojoin_sk
,
2691 true, pfx
, ifindex
);
2697 scope
= ipv6_addr_scope(pfx
);
2699 timeout
= addrconf_timeout_fixup(valid_lft
, HZ
);
2700 if (addrconf_finite_timeout(timeout
)) {
2701 expires
= jiffies_to_clock_t(timeout
* HZ
);
2702 valid_lft
= timeout
;
2703 flags
= RTF_EXPIRES
;
2707 ifa_flags
|= IFA_F_PERMANENT
;
2710 timeout
= addrconf_timeout_fixup(prefered_lft
, HZ
);
2711 if (addrconf_finite_timeout(timeout
)) {
2713 ifa_flags
|= IFA_F_DEPRECATED
;
2714 prefered_lft
= timeout
;
2717 ifp
= ipv6_add_addr(idev
, pfx
, peer_pfx
, plen
, scope
, ifa_flags
,
2718 valid_lft
, prefered_lft
);
2721 if (!(ifa_flags
& IFA_F_NOPREFIXROUTE
)) {
2722 addrconf_prefix_route(&ifp
->addr
, ifp
->prefix_len
, dev
,
2727 * Note that section 3.1 of RFC 4429 indicates
2728 * that the Optimistic flag should not be set for
2729 * manually configured addresses
2731 addrconf_dad_start(ifp
);
2732 if (ifa_flags
& IFA_F_MANAGETEMPADDR
)
2733 manage_tempaddrs(idev
, ifp
, valid_lft
, prefered_lft
,
2736 addrconf_verify_rtnl();
2738 } else if (ifa_flags
& IFA_F_MCAUTOJOIN
) {
2739 ipv6_mc_config(net
->ipv6
.mc_autojoin_sk
,
2740 false, pfx
, ifindex
);
2743 return PTR_ERR(ifp
);
2746 static int inet6_addr_del(struct net
*net
, int ifindex
, u32 ifa_flags
,
2747 const struct in6_addr
*pfx
, unsigned int plen
)
2749 struct inet6_ifaddr
*ifp
;
2750 struct inet6_dev
*idev
;
2751 struct net_device
*dev
;
2756 dev
= __dev_get_by_index(net
, ifindex
);
2760 idev
= __in6_dev_get(dev
);
2764 read_lock_bh(&idev
->lock
);
2765 list_for_each_entry(ifp
, &idev
->addr_list
, if_list
) {
2766 if (ifp
->prefix_len
== plen
&&
2767 ipv6_addr_equal(pfx
, &ifp
->addr
)) {
2769 read_unlock_bh(&idev
->lock
);
2771 if (!(ifp
->flags
& IFA_F_TEMPORARY
) &&
2772 (ifa_flags
& IFA_F_MANAGETEMPADDR
))
2773 manage_tempaddrs(idev
, ifp
, 0, 0, false,
2776 addrconf_verify_rtnl();
2777 if (ipv6_addr_is_multicast(pfx
)) {
2778 ipv6_mc_config(net
->ipv6
.mc_autojoin_sk
,
2779 false, pfx
, dev
->ifindex
);
2784 read_unlock_bh(&idev
->lock
);
2785 return -EADDRNOTAVAIL
;
2789 int addrconf_add_ifaddr(struct net
*net
, void __user
*arg
)
2791 struct in6_ifreq ireq
;
2794 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
2797 if (copy_from_user(&ireq
, arg
, sizeof(struct in6_ifreq
)))
2801 err
= inet6_addr_add(net
, ireq
.ifr6_ifindex
, &ireq
.ifr6_addr
, NULL
,
2802 ireq
.ifr6_prefixlen
, IFA_F_PERMANENT
,
2803 INFINITY_LIFE_TIME
, INFINITY_LIFE_TIME
);
2808 int addrconf_del_ifaddr(struct net
*net
, void __user
*arg
)
2810 struct in6_ifreq ireq
;
2813 if (!ns_capable(net
->user_ns
, CAP_NET_ADMIN
))
2816 if (copy_from_user(&ireq
, arg
, sizeof(struct in6_ifreq
)))
2820 err
= inet6_addr_del(net
, ireq
.ifr6_ifindex
, 0, &ireq
.ifr6_addr
,
2821 ireq
.ifr6_prefixlen
);
2826 static void add_addr(struct inet6_dev
*idev
, const struct in6_addr
*addr
,
2827 int plen
, int scope
)
2829 struct inet6_ifaddr
*ifp
;
2831 ifp
= ipv6_add_addr(idev
, addr
, NULL
, plen
,
2832 scope
, IFA_F_PERMANENT
,
2833 INFINITY_LIFE_TIME
, INFINITY_LIFE_TIME
);
2835 spin_lock_bh(&ifp
->lock
);
2836 ifp
->flags
&= ~IFA_F_TENTATIVE
;
2837 spin_unlock_bh(&ifp
->lock
);
2838 ipv6_ifa_notify(RTM_NEWADDR
, ifp
);
2843 #if IS_ENABLED(CONFIG_IPV6_SIT)
2844 static void sit_add_v4_addrs(struct inet6_dev
*idev
)
2846 struct in6_addr addr
;
2847 struct net_device
*dev
;
2848 struct net
*net
= dev_net(idev
->dev
);
2854 memset(&addr
, 0, sizeof(struct in6_addr
));
2855 memcpy(&addr
.s6_addr32
[3], idev
->dev
->dev_addr
, 4);
2857 if (idev
->dev
->flags
&IFF_POINTOPOINT
) {
2858 addr
.s6_addr32
[0] = htonl(0xfe800000);
2862 scope
= IPV6_ADDR_COMPATv4
;
2864 pflags
|= RTF_NONEXTHOP
;
2867 if (addr
.s6_addr32
[3]) {
2868 add_addr(idev
, &addr
, plen
, scope
);
2869 addrconf_prefix_route(&addr
, plen
, idev
->dev
, 0, pflags
);
2873 for_each_netdev(net
, dev
) {
2874 struct in_device
*in_dev
= __in_dev_get_rtnl(dev
);
2875 if (in_dev
&& (dev
->flags
& IFF_UP
)) {
2876 struct in_ifaddr
*ifa
;
2880 for (ifa
= in_dev
->ifa_list
; ifa
; ifa
= ifa
->ifa_next
) {
2882 addr
.s6_addr32
[3] = ifa
->ifa_local
;
2884 if (ifa
->ifa_scope
== RT_SCOPE_LINK
)
2886 if (ifa
->ifa_scope
>= RT_SCOPE_HOST
) {
2887 if (idev
->dev
->flags
&IFF_POINTOPOINT
)
2892 add_addr(idev
, &addr
, plen
, flag
);
2893 addrconf_prefix_route(&addr
, plen
, idev
->dev
, 0,
2901 static void init_loopback(struct net_device
*dev
)
2903 struct inet6_dev
*idev
;
2904 struct net_device
*sp_dev
;
2905 struct inet6_ifaddr
*sp_ifa
;
2906 struct rt6_info
*sp_rt
;
2912 idev
= ipv6_find_idev(dev
);
2914 pr_debug("%s: add_dev failed\n", __func__
);
2918 add_addr(idev
, &in6addr_loopback
, 128, IFA_HOST
);
2920 /* Add routes to other interface's IPv6 addresses */
2921 for_each_netdev(dev_net(dev
), sp_dev
) {
2922 if (!strcmp(sp_dev
->name
, dev
->name
))
2925 idev
= __in6_dev_get(sp_dev
);
2929 read_lock_bh(&idev
->lock
);
2930 list_for_each_entry(sp_ifa
, &idev
->addr_list
, if_list
) {
2932 if (sp_ifa
->flags
& (IFA_F_DADFAILED
| IFA_F_TENTATIVE
))
2936 /* This dst has been added to garbage list when
2937 * lo device down, release this obsolete dst and
2938 * reallocate a new router for ifa.
2940 if (sp_ifa
->rt
->dst
.obsolete
> 0) {
2941 ip6_rt_put(sp_ifa
->rt
);
2948 sp_rt
= addrconf_dst_alloc(idev
, &sp_ifa
->addr
, false);
2950 /* Failure cases are ignored */
2951 if (!IS_ERR(sp_rt
)) {
2956 read_unlock_bh(&idev
->lock
);
2960 static void addrconf_add_linklocal(struct inet6_dev
*idev
,
2961 const struct in6_addr
*addr
, u32 flags
)
2963 struct inet6_ifaddr
*ifp
;
2964 u32 addr_flags
= flags
| IFA_F_PERMANENT
;
2966 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
2967 if (idev
->cnf
.optimistic_dad
&&
2968 !dev_net(idev
->dev
)->ipv6
.devconf_all
->forwarding
)
2969 addr_flags
|= IFA_F_OPTIMISTIC
;
2972 ifp
= ipv6_add_addr(idev
, addr
, NULL
, 64, IFA_LINK
, addr_flags
,
2973 INFINITY_LIFE_TIME
, INFINITY_LIFE_TIME
);
2975 addrconf_prefix_route(&ifp
->addr
, ifp
->prefix_len
, idev
->dev
, 0, 0);
2976 addrconf_dad_start(ifp
);
2981 static bool ipv6_reserved_interfaceid(struct in6_addr address
)
2983 if ((address
.s6_addr32
[2] | address
.s6_addr32
[3]) == 0)
2986 if (address
.s6_addr32
[2] == htonl(0x02005eff) &&
2987 ((address
.s6_addr32
[3] & htonl(0xfe000000)) == htonl(0xfe000000)))
2990 if (address
.s6_addr32
[2] == htonl(0xfdffffff) &&
2991 ((address
.s6_addr32
[3] & htonl(0xffffff80)) == htonl(0xffffff80)))
2997 static int ipv6_generate_stable_address(struct in6_addr
*address
,
2999 const struct inet6_dev
*idev
)
3001 static DEFINE_SPINLOCK(lock
);
3002 static __u32 digest
[SHA_DIGEST_WORDS
];
3003 static __u32 workspace
[SHA_WORKSPACE_WORDS
];
3006 char __data
[SHA_MESSAGE_BYTES
];
3008 struct in6_addr secret
;
3010 unsigned char hwaddr
[MAX_ADDR_LEN
];
3015 struct in6_addr secret
;
3016 struct in6_addr temp
;
3017 struct net
*net
= dev_net(idev
->dev
);
3019 BUILD_BUG_ON(sizeof(data
.__data
) != sizeof(data
));
3021 if (idev
->cnf
.stable_secret
.initialized
)
3022 secret
= idev
->cnf
.stable_secret
.secret
;
3023 else if (net
->ipv6
.devconf_dflt
->stable_secret
.initialized
)
3024 secret
= net
->ipv6
.devconf_dflt
->stable_secret
.secret
;
3029 spin_lock_bh(&lock
);
3032 memset(&data
, 0, sizeof(data
));
3033 memset(workspace
, 0, sizeof(workspace
));
3034 memcpy(data
.hwaddr
, idev
->dev
->perm_addr
, idev
->dev
->addr_len
);
3035 data
.prefix
[0] = address
->s6_addr32
[0];
3036 data
.prefix
[1] = address
->s6_addr32
[1];
3037 data
.secret
= secret
;
3038 data
.dad_count
= dad_count
;
3040 sha_transform(digest
, data
.__data
, workspace
);
3043 temp
.s6_addr32
[2] = (__force __be32
)digest
[0];
3044 temp
.s6_addr32
[3] = (__force __be32
)digest
[1];
3046 spin_unlock_bh(&lock
);
3048 if (ipv6_reserved_interfaceid(temp
)) {
3050 if (dad_count
> dev_net(idev
->dev
)->ipv6
.sysctl
.idgen_retries
)
3059 static void addrconf_addr_gen(struct inet6_dev
*idev
, bool prefix_route
)
3061 struct in6_addr addr
;
3063 ipv6_addr_set(&addr
, htonl(0xFE800000), 0, 0, 0);
3065 if (idev
->addr_gen_mode
== IN6_ADDR_GEN_MODE_STABLE_PRIVACY
) {
3066 if (!ipv6_generate_stable_address(&addr
, 0, idev
))
3067 addrconf_add_linklocal(idev
, &addr
,
3068 IFA_F_STABLE_PRIVACY
);
3069 else if (prefix_route
)
3070 addrconf_prefix_route(&addr
, 64, idev
->dev
, 0, 0);
3071 } else if (idev
->addr_gen_mode
== IN6_ADDR_GEN_MODE_EUI64
) {
3072 /* addrconf_add_linklocal also adds a prefix_route and we
3073 * only need to care about prefix routes if ipv6_generate_eui64
3074 * couldn't generate one.
3076 if (ipv6_generate_eui64(addr
.s6_addr
+ 8, idev
->dev
) == 0)
3077 addrconf_add_linklocal(idev
, &addr
, 0);
3078 else if (prefix_route
)
3079 addrconf_prefix_route(&addr
, 64, idev
->dev
, 0, 0);
3083 static void addrconf_dev_config(struct net_device
*dev
)
3085 struct inet6_dev
*idev
;
3089 if ((dev
->type
!= ARPHRD_ETHER
) &&
3090 (dev
->type
!= ARPHRD_FDDI
) &&
3091 (dev
->type
!= ARPHRD_ARCNET
) &&
3092 (dev
->type
!= ARPHRD_INFINIBAND
) &&
3093 (dev
->type
!= ARPHRD_IEEE802154
) &&
3094 (dev
->type
!= ARPHRD_IEEE1394
) &&
3095 (dev
->type
!= ARPHRD_TUNNEL6
) &&
3096 (dev
->type
!= ARPHRD_6LOWPAN
)) {
3097 /* Alas, we support only Ethernet autoconfiguration. */
3101 idev
= addrconf_add_dev(dev
);
3105 addrconf_addr_gen(idev
, false);
3108 #if IS_ENABLED(CONFIG_IPV6_SIT)
3109 static void addrconf_sit_config(struct net_device
*dev
)
3111 struct inet6_dev
*idev
;
3116 * Configure the tunnel with one of our IPv4
3117 * addresses... we should configure all of
3118 * our v4 addrs in the tunnel
3121 idev
= ipv6_find_idev(dev
);
3123 pr_debug("%s: add_dev failed\n", __func__
);
3127 if (dev
->priv_flags
& IFF_ISATAP
) {
3128 addrconf_addr_gen(idev
, false);
3132 sit_add_v4_addrs(idev
);
3134 if (dev
->flags
&IFF_POINTOPOINT
)
3135 addrconf_add_mroute(dev
);
3139 #if IS_ENABLED(CONFIG_NET_IPGRE)
3140 static void addrconf_gre_config(struct net_device
*dev
)
3142 struct inet6_dev
*idev
;
3146 idev
= ipv6_find_idev(dev
);
3148 pr_debug("%s: add_dev failed\n", __func__
);
3152 addrconf_addr_gen(idev
, true);
3156 static int addrconf_notify(struct notifier_block
*this, unsigned long event
,
3159 struct net_device
*dev
= netdev_notifier_info_to_dev(ptr
);
3160 struct inet6_dev
*idev
= __in6_dev_get(dev
);
3161 int run_pending
= 0;
3165 case NETDEV_REGISTER
:
3166 if (!idev
&& dev
->mtu
>= IPV6_MIN_MTU
) {
3167 idev
= ipv6_add_dev(dev
);
3169 return notifier_from_errno(PTR_ERR(idev
));
3175 if (dev
->flags
& IFF_SLAVE
)
3178 if (idev
&& idev
->cnf
.disable_ipv6
)
3181 if (event
== NETDEV_UP
) {
3182 if (!addrconf_qdisc_ok(dev
)) {
3183 /* device is not ready yet. */
3184 pr_info("ADDRCONF(NETDEV_UP): %s: link is not ready\n",
3189 if (!idev
&& dev
->mtu
>= IPV6_MIN_MTU
)
3190 idev
= ipv6_add_dev(dev
);
3192 if (!IS_ERR_OR_NULL(idev
)) {
3193 idev
->if_flags
|= IF_READY
;
3197 if (!addrconf_qdisc_ok(dev
)) {
3198 /* device is still not ready. */
3203 if (idev
->if_flags
& IF_READY
)
3204 /* device is already configured. */
3206 idev
->if_flags
|= IF_READY
;
3209 pr_info("ADDRCONF(NETDEV_CHANGE): %s: link becomes ready\n",
3215 switch (dev
->type
) {
3216 #if IS_ENABLED(CONFIG_IPV6_SIT)
3218 addrconf_sit_config(dev
);
3221 #if IS_ENABLED(CONFIG_NET_IPGRE)
3223 addrconf_gre_config(dev
);
3226 case ARPHRD_LOOPBACK
:
3231 addrconf_dev_config(dev
);
3235 if (!IS_ERR_OR_NULL(idev
)) {
3237 addrconf_dad_run(idev
);
3240 * If the MTU changed during the interface down,
3241 * when the interface up, the changed MTU must be
3242 * reflected in the idev as well as routers.
3244 if (idev
->cnf
.mtu6
!= dev
->mtu
&&
3245 dev
->mtu
>= IPV6_MIN_MTU
) {
3246 rt6_mtu_change(dev
, dev
->mtu
);
3247 idev
->cnf
.mtu6
= dev
->mtu
;
3249 idev
->tstamp
= jiffies
;
3250 inet6_ifinfo_notify(RTM_NEWLINK
, idev
);
3253 * If the changed mtu during down is lower than
3254 * IPV6_MIN_MTU stop IPv6 on this interface.
3256 if (dev
->mtu
< IPV6_MIN_MTU
)
3257 addrconf_ifdown(dev
, 1);
3261 case NETDEV_CHANGEMTU
:
3262 if (idev
&& dev
->mtu
>= IPV6_MIN_MTU
) {
3263 rt6_mtu_change(dev
, dev
->mtu
);
3264 idev
->cnf
.mtu6
= dev
->mtu
;
3268 if (!idev
&& dev
->mtu
>= IPV6_MIN_MTU
) {
3269 idev
= ipv6_add_dev(dev
);
3275 * if MTU under IPV6_MIN_MTU.
3276 * Stop IPv6 on this interface.
3280 case NETDEV_UNREGISTER
:
3282 * Remove all addresses from this interface.
3284 addrconf_ifdown(dev
, event
!= NETDEV_DOWN
);
3287 case NETDEV_CHANGENAME
:
3289 snmp6_unregister_dev(idev
);
3290 addrconf_sysctl_unregister(idev
);
3291 err
= addrconf_sysctl_register(idev
);
3293 return notifier_from_errno(err
);
3294 err
= snmp6_register_dev(idev
);
3296 addrconf_sysctl_unregister(idev
);
3297 return notifier_from_errno(err
);
3302 case NETDEV_PRE_TYPE_CHANGE
:
3303 case NETDEV_POST_TYPE_CHANGE
:
3304 addrconf_type_change(dev
, event
);
3312 * addrconf module should be notified of a device going up
3314 static struct notifier_block ipv6_dev_notf
= {
3315 .notifier_call
= addrconf_notify
,
3318 static void addrconf_type_change(struct net_device
*dev
, unsigned long event
)
3320 struct inet6_dev
*idev
;
3323 idev
= __in6_dev_get(dev
);
3325 if (event
== NETDEV_POST_TYPE_CHANGE
)
3326 ipv6_mc_remap(idev
);
3327 else if (event
== NETDEV_PRE_TYPE_CHANGE
)
3328 ipv6_mc_unmap(idev
);
3331 static int addrconf_ifdown(struct net_device
*dev
, int how
)
3333 struct net
*net
= dev_net(dev
);
3334 struct inet6_dev
*idev
;
3335 struct inet6_ifaddr
*ifa
;
3340 rt6_ifdown(net
, dev
);
3341 neigh_ifdown(&nd_tbl
, dev
);
3343 idev
= __in6_dev_get(dev
);
3348 * Step 1: remove reference to ipv6 device from parent device.
3354 /* protected by rtnl_lock */
3355 RCU_INIT_POINTER(dev
->ip6_ptr
, NULL
);
3357 /* Step 1.5: remove snmp6 entry */
3358 snmp6_unregister_dev(idev
);
3362 /* Step 2: clear hash table */
3363 for (i
= 0; i
< IN6_ADDR_HSIZE
; i
++) {
3364 struct hlist_head
*h
= &inet6_addr_lst
[i
];
3366 spin_lock_bh(&addrconf_hash_lock
);
3368 hlist_for_each_entry_rcu(ifa
, h
, addr_lst
) {
3369 if (ifa
->idev
== idev
) {
3370 hlist_del_init_rcu(&ifa
->addr_lst
);
3371 addrconf_del_dad_work(ifa
);
3375 spin_unlock_bh(&addrconf_hash_lock
);
3378 write_lock_bh(&idev
->lock
);
3380 addrconf_del_rs_timer(idev
);
3382 /* Step 2: clear flags for stateless addrconf */
3384 idev
->if_flags
&= ~(IF_RS_SENT
|IF_RA_RCVD
|IF_READY
);
3386 if (how
&& del_timer(&idev
->regen_timer
))
3389 /* Step 3: clear tempaddr list */
3390 while (!list_empty(&idev
->tempaddr_list
)) {
3391 ifa
= list_first_entry(&idev
->tempaddr_list
,
3392 struct inet6_ifaddr
, tmp_list
);
3393 list_del(&ifa
->tmp_list
);
3394 write_unlock_bh(&idev
->lock
);
3395 spin_lock_bh(&ifa
->lock
);
3398 in6_ifa_put(ifa
->ifpub
);
3401 spin_unlock_bh(&ifa
->lock
);
3403 write_lock_bh(&idev
->lock
);
3406 while (!list_empty(&idev
->addr_list
)) {
3407 ifa
= list_first_entry(&idev
->addr_list
,
3408 struct inet6_ifaddr
, if_list
);
3409 addrconf_del_dad_work(ifa
);
3411 list_del(&ifa
->if_list
);
3413 write_unlock_bh(&idev
->lock
);
3415 spin_lock_bh(&ifa
->lock
);
3417 ifa
->state
= INET6_IFADDR_STATE_DEAD
;
3418 spin_unlock_bh(&ifa
->lock
);
3420 if (state
!= INET6_IFADDR_STATE_DEAD
) {
3421 __ipv6_ifa_notify(RTM_DELADDR
, ifa
);
3422 inet6addr_notifier_call_chain(NETDEV_DOWN
, ifa
);
3426 write_lock_bh(&idev
->lock
);
3429 write_unlock_bh(&idev
->lock
);
3431 /* Step 5: Discard anycast and multicast list */
3433 ipv6_ac_destroy_dev(idev
);
3434 ipv6_mc_destroy_dev(idev
);
3439 idev
->tstamp
= jiffies
;
3441 /* Last: Shot the device (if unregistered) */
3443 addrconf_sysctl_unregister(idev
);
3444 neigh_parms_release(&nd_tbl
, idev
->nd_parms
);
3445 neigh_ifdown(&nd_tbl
, dev
);
3451 static void addrconf_rs_timer(unsigned long data
)
3453 struct inet6_dev
*idev
= (struct inet6_dev
*)data
;
3454 struct net_device
*dev
= idev
->dev
;
3455 struct in6_addr lladdr
;
3457 write_lock(&idev
->lock
);
3458 if (idev
->dead
|| !(idev
->if_flags
& IF_READY
))
3461 if (!ipv6_accept_ra(idev
))
3464 /* Announcement received after solicitation was sent */
3465 if (idev
->if_flags
& IF_RA_RCVD
)
3468 if (idev
->rs_probes
++ < idev
->cnf
.rtr_solicits
) {
3469 write_unlock(&idev
->lock
);
3470 if (!ipv6_get_lladdr(dev
, &lladdr
, IFA_F_TENTATIVE
))
3471 ndisc_send_rs(dev
, &lladdr
,
3472 &in6addr_linklocal_allrouters
);
3476 write_lock(&idev
->lock
);
3477 /* The wait after the last probe can be shorter */
3478 addrconf_mod_rs_timer(idev
, (idev
->rs_probes
==
3479 idev
->cnf
.rtr_solicits
) ?
3480 idev
->cnf
.rtr_solicit_delay
:
3481 idev
->cnf
.rtr_solicit_interval
);
3484 * Note: we do not support deprecated "all on-link"
3485 * assumption any longer.
3487 pr_debug("%s: no IPv6 routers present\n", idev
->dev
->name
);
3491 write_unlock(&idev
->lock
);
3497 * Duplicate Address Detection
3499 static void addrconf_dad_kick(struct inet6_ifaddr
*ifp
)
3501 unsigned long rand_num
;
3502 struct inet6_dev
*idev
= ifp
->idev
;
3504 if (ifp
->flags
& IFA_F_OPTIMISTIC
)
3507 rand_num
= prandom_u32() % (idev
->cnf
.rtr_solicit_delay
? : 1);
3509 ifp
->dad_probes
= idev
->cnf
.dad_transmits
;
3510 addrconf_mod_dad_work(ifp
, rand_num
);
3513 static void addrconf_dad_begin(struct inet6_ifaddr
*ifp
)
3515 struct inet6_dev
*idev
= ifp
->idev
;
3516 struct net_device
*dev
= idev
->dev
;
3518 addrconf_join_solict(dev
, &ifp
->addr
);
3520 prandom_seed((__force u32
) ifp
->addr
.s6_addr32
[3]);
3522 read_lock_bh(&idev
->lock
);
3523 spin_lock(&ifp
->lock
);
3524 if (ifp
->state
== INET6_IFADDR_STATE_DEAD
)
3527 if (dev
->flags
&(IFF_NOARP
|IFF_LOOPBACK
) ||
3528 idev
->cnf
.accept_dad
< 1 ||
3529 !(ifp
->flags
&IFA_F_TENTATIVE
) ||
3530 ifp
->flags
& IFA_F_NODAD
) {
3531 ifp
->flags
&= ~(IFA_F_TENTATIVE
|IFA_F_OPTIMISTIC
|IFA_F_DADFAILED
);
3532 spin_unlock(&ifp
->lock
);
3533 read_unlock_bh(&idev
->lock
);
3535 addrconf_dad_completed(ifp
);
3539 if (!(idev
->if_flags
& IF_READY
)) {
3540 spin_unlock(&ifp
->lock
);
3541 read_unlock_bh(&idev
->lock
);
3543 * If the device is not ready:
3544 * - keep it tentative if it is a permanent address.
3545 * - otherwise, kill it.
3548 addrconf_dad_stop(ifp
, 0);
3553 * Optimistic nodes can start receiving
3556 if (ifp
->flags
& IFA_F_OPTIMISTIC
) {
3557 ip6_ins_rt(ifp
->rt
);
3558 if (ipv6_use_optimistic_addr(idev
)) {
3559 /* Because optimistic nodes can use this address,
3560 * notify listeners. If DAD fails, RTM_DELADDR is sent.
3562 ipv6_ifa_notify(RTM_NEWADDR
, ifp
);
3566 addrconf_dad_kick(ifp
);
3568 spin_unlock(&ifp
->lock
);
3569 read_unlock_bh(&idev
->lock
);
3572 static void addrconf_dad_start(struct inet6_ifaddr
*ifp
)
3574 bool begin_dad
= false;
3576 spin_lock_bh(&ifp
->lock
);
3577 if (ifp
->state
!= INET6_IFADDR_STATE_DEAD
) {
3578 ifp
->state
= INET6_IFADDR_STATE_PREDAD
;
3581 spin_unlock_bh(&ifp
->lock
);
3584 addrconf_mod_dad_work(ifp
, 0);
3587 static void addrconf_dad_work(struct work_struct
*w
)
3589 struct inet6_ifaddr
*ifp
= container_of(to_delayed_work(w
),
3590 struct inet6_ifaddr
,
3592 struct inet6_dev
*idev
= ifp
->idev
;
3593 struct in6_addr mcaddr
;
3599 } action
= DAD_PROCESS
;
3603 spin_lock_bh(&ifp
->lock
);
3604 if (ifp
->state
== INET6_IFADDR_STATE_PREDAD
) {
3606 ifp
->state
= INET6_IFADDR_STATE_DAD
;
3607 } else if (ifp
->state
== INET6_IFADDR_STATE_ERRDAD
) {
3609 ifp
->state
= INET6_IFADDR_STATE_POSTDAD
;
3611 spin_unlock_bh(&ifp
->lock
);
3613 if (action
== DAD_BEGIN
) {
3614 addrconf_dad_begin(ifp
);
3616 } else if (action
== DAD_ABORT
) {
3617 addrconf_dad_stop(ifp
, 1);
3621 if (!ifp
->dad_probes
&& addrconf_dad_end(ifp
))
3624 write_lock_bh(&idev
->lock
);
3625 if (idev
->dead
|| !(idev
->if_flags
& IF_READY
)) {
3626 write_unlock_bh(&idev
->lock
);
3630 spin_lock(&ifp
->lock
);
3631 if (ifp
->state
== INET6_IFADDR_STATE_DEAD
) {
3632 spin_unlock(&ifp
->lock
);
3633 write_unlock_bh(&idev
->lock
);
3637 if (ifp
->dad_probes
== 0) {
3639 * DAD was successful
3642 ifp
->flags
&= ~(IFA_F_TENTATIVE
|IFA_F_OPTIMISTIC
|IFA_F_DADFAILED
);
3643 spin_unlock(&ifp
->lock
);
3644 write_unlock_bh(&idev
->lock
);
3646 addrconf_dad_completed(ifp
);
3652 addrconf_mod_dad_work(ifp
,
3653 NEIGH_VAR(ifp
->idev
->nd_parms
, RETRANS_TIME
));
3654 spin_unlock(&ifp
->lock
);
3655 write_unlock_bh(&idev
->lock
);
3657 /* send a neighbour solicitation for our addr */
3658 addrconf_addr_solict_mult(&ifp
->addr
, &mcaddr
);
3659 ndisc_send_ns(ifp
->idev
->dev
, NULL
, &ifp
->addr
, &mcaddr
, &in6addr_any
, NULL
);
3665 /* ifp->idev must be at least read locked */
3666 static bool ipv6_lonely_lladdr(struct inet6_ifaddr
*ifp
)
3668 struct inet6_ifaddr
*ifpiter
;
3669 struct inet6_dev
*idev
= ifp
->idev
;
3671 list_for_each_entry_reverse(ifpiter
, &idev
->addr_list
, if_list
) {
3672 if (ifpiter
->scope
> IFA_LINK
)
3674 if (ifp
!= ifpiter
&& ifpiter
->scope
== IFA_LINK
&&
3675 (ifpiter
->flags
& (IFA_F_PERMANENT
|IFA_F_TENTATIVE
|
3676 IFA_F_OPTIMISTIC
|IFA_F_DADFAILED
)) ==
3683 static void addrconf_dad_completed(struct inet6_ifaddr
*ifp
)
3685 struct net_device
*dev
= ifp
->idev
->dev
;
3686 struct in6_addr lladdr
;
3687 bool send_rs
, send_mld
;
3689 addrconf_del_dad_work(ifp
);
3692 * Configure the address for reception. Now it is valid.
3695 ipv6_ifa_notify(RTM_NEWADDR
, ifp
);
3697 /* If added prefix is link local and we are prepared to process
3698 router advertisements, start sending router solicitations.
3701 read_lock_bh(&ifp
->idev
->lock
);
3702 send_mld
= ifp
->scope
== IFA_LINK
&& ipv6_lonely_lladdr(ifp
);
3703 send_rs
= send_mld
&&
3704 ipv6_accept_ra(ifp
->idev
) &&
3705 ifp
->idev
->cnf
.rtr_solicits
> 0 &&
3706 (dev
->flags
&IFF_LOOPBACK
) == 0;
3707 read_unlock_bh(&ifp
->idev
->lock
);
3709 /* While dad is in progress mld report's source address is in6_addrany.
3710 * Resend with proper ll now.
3713 ipv6_mc_dad_complete(ifp
->idev
);
3717 * If a host as already performed a random delay
3718 * [...] as part of DAD [...] there is no need
3719 * to delay again before sending the first RS
3721 if (ipv6_get_lladdr(dev
, &lladdr
, IFA_F_TENTATIVE
))
3723 ndisc_send_rs(dev
, &lladdr
, &in6addr_linklocal_allrouters
);
3725 write_lock_bh(&ifp
->idev
->lock
);
3726 spin_lock(&ifp
->lock
);
3727 ifp
->idev
->rs_probes
= 1;
3728 ifp
->idev
->if_flags
|= IF_RS_SENT
;
3729 addrconf_mod_rs_timer(ifp
->idev
,
3730 ifp
->idev
->cnf
.rtr_solicit_interval
);
3731 spin_unlock(&ifp
->lock
);
3732 write_unlock_bh(&ifp
->idev
->lock
);
3736 static void addrconf_dad_run(struct inet6_dev
*idev
)
3738 struct inet6_ifaddr
*ifp
;
3740 read_lock_bh(&idev
->lock
);
3741 list_for_each_entry(ifp
, &idev
->addr_list
, if_list
) {
3742 spin_lock(&ifp
->lock
);
3743 if (ifp
->flags
& IFA_F_TENTATIVE
&&
3744 ifp
->state
== INET6_IFADDR_STATE_DAD
)
3745 addrconf_dad_kick(ifp
);
3746 spin_unlock(&ifp
->lock
);
3748 read_unlock_bh(&idev
->lock
);
3751 #ifdef CONFIG_PROC_FS
3752 struct if6_iter_state
{
3753 struct seq_net_private p
;
3758 static struct inet6_ifaddr
*if6_get_first(struct seq_file
*seq
, loff_t pos
)
3760 struct inet6_ifaddr
*ifa
= NULL
;
3761 struct if6_iter_state
*state
= seq
->private;
3762 struct net
*net
= seq_file_net(seq
);
3765 /* initial bucket if pos is 0 */
3771 for (; state
->bucket
< IN6_ADDR_HSIZE
; ++state
->bucket
) {
3772 hlist_for_each_entry_rcu_bh(ifa
, &inet6_addr_lst
[state
->bucket
],
3774 if (!net_eq(dev_net(ifa
->idev
->dev
), net
))
3776 /* sync with offset */
3777 if (p
< state
->offset
) {
3785 /* prepare for next bucket */
3792 static struct inet6_ifaddr
*if6_get_next(struct seq_file
*seq
,
3793 struct inet6_ifaddr
*ifa
)
3795 struct if6_iter_state
*state
= seq
->private;
3796 struct net
*net
= seq_file_net(seq
);
3798 hlist_for_each_entry_continue_rcu_bh(ifa
, addr_lst
) {
3799 if (!net_eq(dev_net(ifa
->idev
->dev
), net
))
3805 while (++state
->bucket
< IN6_ADDR_HSIZE
) {
3807 hlist_for_each_entry_rcu_bh(ifa
,
3808 &inet6_addr_lst
[state
->bucket
], addr_lst
) {
3809 if (!net_eq(dev_net(ifa
->idev
->dev
), net
))
3819 static void *if6_seq_start(struct seq_file
*seq
, loff_t
*pos
)
3823 return if6_get_first(seq
, *pos
);
3826 static void *if6_seq_next(struct seq_file
*seq
, void *v
, loff_t
*pos
)
3828 struct inet6_ifaddr
*ifa
;
3830 ifa
= if6_get_next(seq
, v
);
3835 static void if6_seq_stop(struct seq_file
*seq
, void *v
)
3838 rcu_read_unlock_bh();
3841 static int if6_seq_show(struct seq_file
*seq
, void *v
)
3843 struct inet6_ifaddr
*ifp
= (struct inet6_ifaddr
*)v
;
3844 seq_printf(seq
, "%pi6 %02x %02x %02x %02x %8s\n",
3846 ifp
->idev
->dev
->ifindex
,
3850 ifp
->idev
->dev
->name
);
3854 static const struct seq_operations if6_seq_ops
= {
3855 .start
= if6_seq_start
,
3856 .next
= if6_seq_next
,
3857 .show
= if6_seq_show
,
3858 .stop
= if6_seq_stop
,
3861 static int if6_seq_open(struct inode
*inode
, struct file
*file
)
3863 return seq_open_net(inode
, file
, &if6_seq_ops
,
3864 sizeof(struct if6_iter_state
));
3867 static const struct file_operations if6_fops
= {
3868 .owner
= THIS_MODULE
,
3869 .open
= if6_seq_open
,
3871 .llseek
= seq_lseek
,
3872 .release
= seq_release_net
,
3875 static int __net_init
if6_proc_net_init(struct net
*net
)
3877 if (!proc_create("if_inet6", S_IRUGO
, net
->proc_net
, &if6_fops
))
3882 static void __net_exit
if6_proc_net_exit(struct net
*net
)
3884 remove_proc_entry("if_inet6", net
->proc_net
);
3887 static struct pernet_operations if6_proc_net_ops
= {
3888 .init
= if6_proc_net_init
,
3889 .exit
= if6_proc_net_exit
,
3892 int __init
if6_proc_init(void)
3894 return register_pernet_subsys(&if6_proc_net_ops
);
3897 void if6_proc_exit(void)
3899 unregister_pernet_subsys(&if6_proc_net_ops
);
3901 #endif /* CONFIG_PROC_FS */
3903 #if IS_ENABLED(CONFIG_IPV6_MIP6)
3904 /* Check if address is a home address configured on any interface. */
3905 int ipv6_chk_home_addr(struct net
*net
, const struct in6_addr
*addr
)
3908 struct inet6_ifaddr
*ifp
= NULL
;
3909 unsigned int hash
= inet6_addr_hash(addr
);
3912 hlist_for_each_entry_rcu_bh(ifp
, &inet6_addr_lst
[hash
], addr_lst
) {
3913 if (!net_eq(dev_net(ifp
->idev
->dev
), net
))
3915 if (ipv6_addr_equal(&ifp
->addr
, addr
) &&
3916 (ifp
->flags
& IFA_F_HOMEADDRESS
)) {
3921 rcu_read_unlock_bh();
3927 * Periodic address status verification
3930 static void addrconf_verify_rtnl(void)
3932 unsigned long now
, next
, next_sec
, next_sched
;
3933 struct inet6_ifaddr
*ifp
;
3940 next
= round_jiffies_up(now
+ ADDR_CHECK_FREQUENCY
);
3942 cancel_delayed_work(&addr_chk_work
);
3944 for (i
= 0; i
< IN6_ADDR_HSIZE
; i
++) {
3946 hlist_for_each_entry_rcu_bh(ifp
, &inet6_addr_lst
[i
], addr_lst
) {
3949 /* When setting preferred_lft to a value not zero or
3950 * infinity, while valid_lft is infinity
3951 * IFA_F_PERMANENT has a non-infinity life time.
3953 if ((ifp
->flags
& IFA_F_PERMANENT
) &&
3954 (ifp
->prefered_lft
== INFINITY_LIFE_TIME
))
3957 spin_lock(&ifp
->lock
);
3958 /* We try to batch several events at once. */
3959 age
= (now
- ifp
->tstamp
+ ADDRCONF_TIMER_FUZZ_MINUS
) / HZ
;
3961 if (ifp
->valid_lft
!= INFINITY_LIFE_TIME
&&
3962 age
>= ifp
->valid_lft
) {
3963 spin_unlock(&ifp
->lock
);
3967 } else if (ifp
->prefered_lft
== INFINITY_LIFE_TIME
) {
3968 spin_unlock(&ifp
->lock
);
3970 } else if (age
>= ifp
->prefered_lft
) {
3971 /* jiffies - ifp->tstamp > age >= ifp->prefered_lft */
3974 if (!(ifp
->flags
&IFA_F_DEPRECATED
)) {
3976 ifp
->flags
|= IFA_F_DEPRECATED
;
3979 if ((ifp
->valid_lft
!= INFINITY_LIFE_TIME
) &&
3980 (time_before(ifp
->tstamp
+ ifp
->valid_lft
* HZ
, next
)))
3981 next
= ifp
->tstamp
+ ifp
->valid_lft
* HZ
;
3983 spin_unlock(&ifp
->lock
);
3988 ipv6_ifa_notify(0, ifp
);
3992 } else if ((ifp
->flags
&IFA_F_TEMPORARY
) &&
3993 !(ifp
->flags
&IFA_F_TENTATIVE
)) {
3994 unsigned long regen_advance
= ifp
->idev
->cnf
.regen_max_retry
*
3995 ifp
->idev
->cnf
.dad_transmits
*
3996 NEIGH_VAR(ifp
->idev
->nd_parms
, RETRANS_TIME
) / HZ
;
3998 if (age
>= ifp
->prefered_lft
- regen_advance
) {
3999 struct inet6_ifaddr
*ifpub
= ifp
->ifpub
;
4000 if (time_before(ifp
->tstamp
+ ifp
->prefered_lft
* HZ
, next
))
4001 next
= ifp
->tstamp
+ ifp
->prefered_lft
* HZ
;
4002 if (!ifp
->regen_count
&& ifpub
) {
4005 in6_ifa_hold(ifpub
);
4006 spin_unlock(&ifp
->lock
);
4008 spin_lock(&ifpub
->lock
);
4009 ifpub
->regen_count
= 0;
4010 spin_unlock(&ifpub
->lock
);
4011 ipv6_create_tempaddr(ifpub
, ifp
);
4016 } else if (time_before(ifp
->tstamp
+ ifp
->prefered_lft
* HZ
- regen_advance
* HZ
, next
))
4017 next
= ifp
->tstamp
+ ifp
->prefered_lft
* HZ
- regen_advance
* HZ
;
4018 spin_unlock(&ifp
->lock
);
4020 /* ifp->prefered_lft <= ifp->valid_lft */
4021 if (time_before(ifp
->tstamp
+ ifp
->prefered_lft
* HZ
, next
))
4022 next
= ifp
->tstamp
+ ifp
->prefered_lft
* HZ
;
4023 spin_unlock(&ifp
->lock
);
4028 next_sec
= round_jiffies_up(next
);
4031 /* If rounded timeout is accurate enough, accept it. */
4032 if (time_before(next_sec
, next
+ ADDRCONF_TIMER_FUZZ
))
4033 next_sched
= next_sec
;
4035 /* And minimum interval is ADDRCONF_TIMER_FUZZ_MAX. */
4036 if (time_before(next_sched
, jiffies
+ ADDRCONF_TIMER_FUZZ_MAX
))
4037 next_sched
= jiffies
+ ADDRCONF_TIMER_FUZZ_MAX
;
4039 ADBG(KERN_DEBUG
"now = %lu, schedule = %lu, rounded schedule = %lu => %lu\n",
4040 now
, next
, next_sec
, next_sched
);
4041 mod_delayed_work(addrconf_wq
, &addr_chk_work
, next_sched
- now
);
4042 rcu_read_unlock_bh();
4045 static void addrconf_verify_work(struct work_struct
*w
)
4048 addrconf_verify_rtnl();
4052 static void addrconf_verify(void)
4054 mod_delayed_work(addrconf_wq
, &addr_chk_work
, 0);
4057 static struct in6_addr
*extract_addr(struct nlattr
*addr
, struct nlattr
*local
,
4058 struct in6_addr
**peer_pfx
)
4060 struct in6_addr
*pfx
= NULL
;
4065 pfx
= nla_data(addr
);
4068 if (pfx
&& nla_memcmp(local
, pfx
, sizeof(*pfx
)))
4070 pfx
= nla_data(local
);
4076 static const struct nla_policy ifa_ipv6_policy
[IFA_MAX
+1] = {
4077 [IFA_ADDRESS
] = { .len
= sizeof(struct in6_addr
) },
4078 [IFA_LOCAL
] = { .len
= sizeof(struct in6_addr
) },
4079 [IFA_CACHEINFO
] = { .len
= sizeof(struct ifa_cacheinfo
) },
4080 [IFA_FLAGS
] = { .len
= sizeof(u32
) },
4084 inet6_rtm_deladdr(struct sk_buff
*skb
, struct nlmsghdr
*nlh
)
4086 struct net
*net
= sock_net(skb
->sk
);
4087 struct ifaddrmsg
*ifm
;
4088 struct nlattr
*tb
[IFA_MAX
+1];
4089 struct in6_addr
*pfx
, *peer_pfx
;
4093 err
= nlmsg_parse(nlh
, sizeof(*ifm
), tb
, IFA_MAX
, ifa_ipv6_policy
);
4097 ifm
= nlmsg_data(nlh
);
4098 pfx
= extract_addr(tb
[IFA_ADDRESS
], tb
[IFA_LOCAL
], &peer_pfx
);
4102 ifa_flags
= tb
[IFA_FLAGS
] ? nla_get_u32(tb
[IFA_FLAGS
]) : ifm
->ifa_flags
;
4104 /* We ignore other flags so far. */
4105 ifa_flags
&= IFA_F_MANAGETEMPADDR
;
4107 return inet6_addr_del(net
, ifm
->ifa_index
, ifa_flags
, pfx
,
4108 ifm
->ifa_prefixlen
);
4111 static int inet6_addr_modify(struct inet6_ifaddr
*ifp
, u32 ifa_flags
,
4112 u32 prefered_lft
, u32 valid_lft
)
4116 unsigned long timeout
;
4117 bool was_managetempaddr
;
4118 bool had_prefixroute
;
4122 if (!valid_lft
|| (prefered_lft
> valid_lft
))
4125 if (ifa_flags
& IFA_F_MANAGETEMPADDR
&&
4126 (ifp
->flags
& IFA_F_TEMPORARY
|| ifp
->prefix_len
!= 64))
4129 timeout
= addrconf_timeout_fixup(valid_lft
, HZ
);
4130 if (addrconf_finite_timeout(timeout
)) {
4131 expires
= jiffies_to_clock_t(timeout
* HZ
);
4132 valid_lft
= timeout
;
4133 flags
= RTF_EXPIRES
;
4137 ifa_flags
|= IFA_F_PERMANENT
;
4140 timeout
= addrconf_timeout_fixup(prefered_lft
, HZ
);
4141 if (addrconf_finite_timeout(timeout
)) {
4143 ifa_flags
|= IFA_F_DEPRECATED
;
4144 prefered_lft
= timeout
;
4147 spin_lock_bh(&ifp
->lock
);
4148 was_managetempaddr
= ifp
->flags
& IFA_F_MANAGETEMPADDR
;
4149 had_prefixroute
= ifp
->flags
& IFA_F_PERMANENT
&&
4150 !(ifp
->flags
& IFA_F_NOPREFIXROUTE
);
4151 ifp
->flags
&= ~(IFA_F_DEPRECATED
| IFA_F_PERMANENT
| IFA_F_NODAD
|
4152 IFA_F_HOMEADDRESS
| IFA_F_MANAGETEMPADDR
|
4153 IFA_F_NOPREFIXROUTE
);
4154 ifp
->flags
|= ifa_flags
;
4155 ifp
->tstamp
= jiffies
;
4156 ifp
->valid_lft
= valid_lft
;
4157 ifp
->prefered_lft
= prefered_lft
;
4159 spin_unlock_bh(&ifp
->lock
);
4160 if (!(ifp
->flags
&IFA_F_TENTATIVE
))
4161 ipv6_ifa_notify(0, ifp
);
4163 if (!(ifa_flags
& IFA_F_NOPREFIXROUTE
)) {
4164 addrconf_prefix_route(&ifp
->addr
, ifp
->prefix_len
, ifp
->idev
->dev
,
4166 } else if (had_prefixroute
) {
4167 enum cleanup_prefix_rt_t action
;
4168 unsigned long rt_expires
;
4170 write_lock_bh(&ifp
->idev
->lock
);
4171 action
= check_cleanup_prefix_route(ifp
, &rt_expires
);
4172 write_unlock_bh(&ifp
->idev
->lock
);
4174 if (action
!= CLEANUP_PREFIX_RT_NOP
) {
4175 cleanup_prefix_route(ifp
, rt_expires
,
4176 action
== CLEANUP_PREFIX_RT_DEL
);
4180 if (was_managetempaddr
|| ifp
->flags
& IFA_F_MANAGETEMPADDR
) {
4181 if (was_managetempaddr
&& !(ifp
->flags
& IFA_F_MANAGETEMPADDR
))
4182 valid_lft
= prefered_lft
= 0;
4183 manage_tempaddrs(ifp
->idev
, ifp
, valid_lft
, prefered_lft
,
4184 !was_managetempaddr
, jiffies
);
4187 addrconf_verify_rtnl();
4193 inet6_rtm_newaddr(struct sk_buff
*skb
, struct nlmsghdr
*nlh
)
4195 struct net
*net
= sock_net(skb
->sk
);
4196 struct ifaddrmsg
*ifm
;
4197 struct nlattr
*tb
[IFA_MAX
+1];
4198 struct in6_addr
*pfx
, *peer_pfx
;
4199 struct inet6_ifaddr
*ifa
;
4200 struct net_device
*dev
;
4201 u32 valid_lft
= INFINITY_LIFE_TIME
, preferred_lft
= INFINITY_LIFE_TIME
;
4205 err
= nlmsg_parse(nlh
, sizeof(*ifm
), tb
, IFA_MAX
, ifa_ipv6_policy
);
4209 ifm
= nlmsg_data(nlh
);
4210 pfx
= extract_addr(tb
[IFA_ADDRESS
], tb
[IFA_LOCAL
], &peer_pfx
);
4214 if (tb
[IFA_CACHEINFO
]) {
4215 struct ifa_cacheinfo
*ci
;
4217 ci
= nla_data(tb
[IFA_CACHEINFO
]);
4218 valid_lft
= ci
->ifa_valid
;
4219 preferred_lft
= ci
->ifa_prefered
;
4221 preferred_lft
= INFINITY_LIFE_TIME
;
4222 valid_lft
= INFINITY_LIFE_TIME
;
4225 dev
= __dev_get_by_index(net
, ifm
->ifa_index
);
4229 ifa_flags
= tb
[IFA_FLAGS
] ? nla_get_u32(tb
[IFA_FLAGS
]) : ifm
->ifa_flags
;
4231 /* We ignore other flags so far. */
4232 ifa_flags
&= IFA_F_NODAD
| IFA_F_HOMEADDRESS
| IFA_F_MANAGETEMPADDR
|
4233 IFA_F_NOPREFIXROUTE
| IFA_F_MCAUTOJOIN
;
4235 ifa
= ipv6_get_ifaddr(net
, pfx
, dev
, 1);
4238 * It would be best to check for !NLM_F_CREATE here but
4239 * userspace already relies on not having to provide this.
4241 return inet6_addr_add(net
, ifm
->ifa_index
, pfx
, peer_pfx
,
4242 ifm
->ifa_prefixlen
, ifa_flags
,
4243 preferred_lft
, valid_lft
);
4246 if (nlh
->nlmsg_flags
& NLM_F_EXCL
||
4247 !(nlh
->nlmsg_flags
& NLM_F_REPLACE
))
4250 err
= inet6_addr_modify(ifa
, ifa_flags
, preferred_lft
, valid_lft
);
4257 static void put_ifaddrmsg(struct nlmsghdr
*nlh
, u8 prefixlen
, u32 flags
,
4258 u8 scope
, int ifindex
)
4260 struct ifaddrmsg
*ifm
;
4262 ifm
= nlmsg_data(nlh
);
4263 ifm
->ifa_family
= AF_INET6
;
4264 ifm
->ifa_prefixlen
= prefixlen
;
4265 ifm
->ifa_flags
= flags
;
4266 ifm
->ifa_scope
= scope
;
4267 ifm
->ifa_index
= ifindex
;
4270 static int put_cacheinfo(struct sk_buff
*skb
, unsigned long cstamp
,
4271 unsigned long tstamp
, u32 preferred
, u32 valid
)
4273 struct ifa_cacheinfo ci
;
4275 ci
.cstamp
= cstamp_delta(cstamp
);
4276 ci
.tstamp
= cstamp_delta(tstamp
);
4277 ci
.ifa_prefered
= preferred
;
4278 ci
.ifa_valid
= valid
;
4280 return nla_put(skb
, IFA_CACHEINFO
, sizeof(ci
), &ci
);
4283 static inline int rt_scope(int ifa_scope
)
4285 if (ifa_scope
& IFA_HOST
)
4286 return RT_SCOPE_HOST
;
4287 else if (ifa_scope
& IFA_LINK
)
4288 return RT_SCOPE_LINK
;
4289 else if (ifa_scope
& IFA_SITE
)
4290 return RT_SCOPE_SITE
;
4292 return RT_SCOPE_UNIVERSE
;
4295 static inline int inet6_ifaddr_msgsize(void)
4297 return NLMSG_ALIGN(sizeof(struct ifaddrmsg
))
4298 + nla_total_size(16) /* IFA_LOCAL */
4299 + nla_total_size(16) /* IFA_ADDRESS */
4300 + nla_total_size(sizeof(struct ifa_cacheinfo
))
4301 + nla_total_size(4) /* IFA_FLAGS */;
4304 static int inet6_fill_ifaddr(struct sk_buff
*skb
, struct inet6_ifaddr
*ifa
,
4305 u32 portid
, u32 seq
, int event
, unsigned int flags
)
4307 struct nlmsghdr
*nlh
;
4308 u32 preferred
, valid
;
4310 nlh
= nlmsg_put(skb
, portid
, seq
, event
, sizeof(struct ifaddrmsg
), flags
);
4314 put_ifaddrmsg(nlh
, ifa
->prefix_len
, ifa
->flags
, rt_scope(ifa
->scope
),
4315 ifa
->idev
->dev
->ifindex
);
4317 if (!((ifa
->flags
&IFA_F_PERMANENT
) &&
4318 (ifa
->prefered_lft
== INFINITY_LIFE_TIME
))) {
4319 preferred
= ifa
->prefered_lft
;
4320 valid
= ifa
->valid_lft
;
4321 if (preferred
!= INFINITY_LIFE_TIME
) {
4322 long tval
= (jiffies
- ifa
->tstamp
)/HZ
;
4323 if (preferred
> tval
)
4327 if (valid
!= INFINITY_LIFE_TIME
) {
4335 preferred
= INFINITY_LIFE_TIME
;
4336 valid
= INFINITY_LIFE_TIME
;
4339 if (!ipv6_addr_any(&ifa
->peer_addr
)) {
4340 if (nla_put_in6_addr(skb
, IFA_LOCAL
, &ifa
->addr
) < 0 ||
4341 nla_put_in6_addr(skb
, IFA_ADDRESS
, &ifa
->peer_addr
) < 0)
4344 if (nla_put_in6_addr(skb
, IFA_ADDRESS
, &ifa
->addr
) < 0)
4347 if (put_cacheinfo(skb
, ifa
->cstamp
, ifa
->tstamp
, preferred
, valid
) < 0)
4350 if (nla_put_u32(skb
, IFA_FLAGS
, ifa
->flags
) < 0)
4353 nlmsg_end(skb
, nlh
);
4357 nlmsg_cancel(skb
, nlh
);
4361 static int inet6_fill_ifmcaddr(struct sk_buff
*skb
, struct ifmcaddr6
*ifmca
,
4362 u32 portid
, u32 seq
, int event
, u16 flags
)
4364 struct nlmsghdr
*nlh
;
4365 u8 scope
= RT_SCOPE_UNIVERSE
;
4366 int ifindex
= ifmca
->idev
->dev
->ifindex
;
4368 if (ipv6_addr_scope(&ifmca
->mca_addr
) & IFA_SITE
)
4369 scope
= RT_SCOPE_SITE
;
4371 nlh
= nlmsg_put(skb
, portid
, seq
, event
, sizeof(struct ifaddrmsg
), flags
);
4375 put_ifaddrmsg(nlh
, 128, IFA_F_PERMANENT
, scope
, ifindex
);
4376 if (nla_put_in6_addr(skb
, IFA_MULTICAST
, &ifmca
->mca_addr
) < 0 ||
4377 put_cacheinfo(skb
, ifmca
->mca_cstamp
, ifmca
->mca_tstamp
,
4378 INFINITY_LIFE_TIME
, INFINITY_LIFE_TIME
) < 0) {
4379 nlmsg_cancel(skb
, nlh
);
4383 nlmsg_end(skb
, nlh
);
4387 static int inet6_fill_ifacaddr(struct sk_buff
*skb
, struct ifacaddr6
*ifaca
,
4388 u32 portid
, u32 seq
, int event
, unsigned int flags
)
4390 struct nlmsghdr
*nlh
;
4391 u8 scope
= RT_SCOPE_UNIVERSE
;
4392 int ifindex
= ifaca
->aca_idev
->dev
->ifindex
;
4394 if (ipv6_addr_scope(&ifaca
->aca_addr
) & IFA_SITE
)
4395 scope
= RT_SCOPE_SITE
;
4397 nlh
= nlmsg_put(skb
, portid
, seq
, event
, sizeof(struct ifaddrmsg
), flags
);
4401 put_ifaddrmsg(nlh
, 128, IFA_F_PERMANENT
, scope
, ifindex
);
4402 if (nla_put_in6_addr(skb
, IFA_ANYCAST
, &ifaca
->aca_addr
) < 0 ||
4403 put_cacheinfo(skb
, ifaca
->aca_cstamp
, ifaca
->aca_tstamp
,
4404 INFINITY_LIFE_TIME
, INFINITY_LIFE_TIME
) < 0) {
4405 nlmsg_cancel(skb
, nlh
);
4409 nlmsg_end(skb
, nlh
);
4419 /* called with rcu_read_lock() */
4420 static int in6_dump_addrs(struct inet6_dev
*idev
, struct sk_buff
*skb
,
4421 struct netlink_callback
*cb
, enum addr_type_t type
,
4422 int s_ip_idx
, int *p_ip_idx
)
4424 struct ifmcaddr6
*ifmca
;
4425 struct ifacaddr6
*ifaca
;
4427 int ip_idx
= *p_ip_idx
;
4429 read_lock_bh(&idev
->lock
);
4431 case UNICAST_ADDR
: {
4432 struct inet6_ifaddr
*ifa
;
4434 /* unicast address incl. temp addr */
4435 list_for_each_entry(ifa
, &idev
->addr_list
, if_list
) {
4436 if (++ip_idx
< s_ip_idx
)
4438 err
= inet6_fill_ifaddr(skb
, ifa
,
4439 NETLINK_CB(cb
->skb
).portid
,
4445 nl_dump_check_consistent(cb
, nlmsg_hdr(skb
));
4449 case MULTICAST_ADDR
:
4450 /* multicast address */
4451 for (ifmca
= idev
->mc_list
; ifmca
;
4452 ifmca
= ifmca
->next
, ip_idx
++) {
4453 if (ip_idx
< s_ip_idx
)
4455 err
= inet6_fill_ifmcaddr(skb
, ifmca
,
4456 NETLINK_CB(cb
->skb
).portid
,
4465 /* anycast address */
4466 for (ifaca
= idev
->ac_list
; ifaca
;
4467 ifaca
= ifaca
->aca_next
, ip_idx
++) {
4468 if (ip_idx
< s_ip_idx
)
4470 err
= inet6_fill_ifacaddr(skb
, ifaca
,
4471 NETLINK_CB(cb
->skb
).portid
,
4482 read_unlock_bh(&idev
->lock
);
4487 static int inet6_dump_addr(struct sk_buff
*skb
, struct netlink_callback
*cb
,
4488 enum addr_type_t type
)
4490 struct net
*net
= sock_net(skb
->sk
);
4493 int s_idx
, s_ip_idx
;
4494 struct net_device
*dev
;
4495 struct inet6_dev
*idev
;
4496 struct hlist_head
*head
;
4499 s_idx
= idx
= cb
->args
[1];
4500 s_ip_idx
= ip_idx
= cb
->args
[2];
4503 cb
->seq
= atomic_read(&net
->ipv6
.dev_addr_genid
) ^ net
->dev_base_seq
;
4504 for (h
= s_h
; h
< NETDEV_HASHENTRIES
; h
++, s_idx
= 0) {
4506 head
= &net
->dev_index_head
[h
];
4507 hlist_for_each_entry_rcu(dev
, head
, index_hlist
) {
4510 if (h
> s_h
|| idx
> s_idx
)
4513 idev
= __in6_dev_get(dev
);
4517 if (in6_dump_addrs(idev
, skb
, cb
, type
,
4518 s_ip_idx
, &ip_idx
) < 0)
4528 cb
->args
[2] = ip_idx
;
4533 static int inet6_dump_ifaddr(struct sk_buff
*skb
, struct netlink_callback
*cb
)
4535 enum addr_type_t type
= UNICAST_ADDR
;
4537 return inet6_dump_addr(skb
, cb
, type
);
4540 static int inet6_dump_ifmcaddr(struct sk_buff
*skb
, struct netlink_callback
*cb
)
4542 enum addr_type_t type
= MULTICAST_ADDR
;
4544 return inet6_dump_addr(skb
, cb
, type
);
4548 static int inet6_dump_ifacaddr(struct sk_buff
*skb
, struct netlink_callback
*cb
)
4550 enum addr_type_t type
= ANYCAST_ADDR
;
4552 return inet6_dump_addr(skb
, cb
, type
);
4555 static int inet6_rtm_getaddr(struct sk_buff
*in_skb
, struct nlmsghdr
*nlh
)
4557 struct net
*net
= sock_net(in_skb
->sk
);
4558 struct ifaddrmsg
*ifm
;
4559 struct nlattr
*tb
[IFA_MAX
+1];
4560 struct in6_addr
*addr
= NULL
, *peer
;
4561 struct net_device
*dev
= NULL
;
4562 struct inet6_ifaddr
*ifa
;
4563 struct sk_buff
*skb
;
4566 err
= nlmsg_parse(nlh
, sizeof(*ifm
), tb
, IFA_MAX
, ifa_ipv6_policy
);
4570 addr
= extract_addr(tb
[IFA_ADDRESS
], tb
[IFA_LOCAL
], &peer
);
4576 ifm
= nlmsg_data(nlh
);
4578 dev
= __dev_get_by_index(net
, ifm
->ifa_index
);
4580 ifa
= ipv6_get_ifaddr(net
, addr
, dev
, 1);
4582 err
= -EADDRNOTAVAIL
;
4586 skb
= nlmsg_new(inet6_ifaddr_msgsize(), GFP_KERNEL
);
4592 err
= inet6_fill_ifaddr(skb
, ifa
, NETLINK_CB(in_skb
).portid
,
4593 nlh
->nlmsg_seq
, RTM_NEWADDR
, 0);
4595 /* -EMSGSIZE implies BUG in inet6_ifaddr_msgsize() */
4596 WARN_ON(err
== -EMSGSIZE
);
4600 err
= rtnl_unicast(skb
, net
, NETLINK_CB(in_skb
).portid
);
4607 static void inet6_ifa_notify(int event
, struct inet6_ifaddr
*ifa
)
4609 struct sk_buff
*skb
;
4610 struct net
*net
= dev_net(ifa
->idev
->dev
);
4613 skb
= nlmsg_new(inet6_ifaddr_msgsize(), GFP_ATOMIC
);
4617 err
= inet6_fill_ifaddr(skb
, ifa
, 0, 0, event
, 0);
4619 /* -EMSGSIZE implies BUG in inet6_ifaddr_msgsize() */
4620 WARN_ON(err
== -EMSGSIZE
);
4624 rtnl_notify(skb
, net
, 0, RTNLGRP_IPV6_IFADDR
, NULL
, GFP_ATOMIC
);
4628 rtnl_set_sk_err(net
, RTNLGRP_IPV6_IFADDR
, err
);
4631 static inline void ipv6_store_devconf(struct ipv6_devconf
*cnf
,
4632 __s32
*array
, int bytes
)
4634 BUG_ON(bytes
< (DEVCONF_MAX
* 4));
4636 memset(array
, 0, bytes
);
4637 array
[DEVCONF_FORWARDING
] = cnf
->forwarding
;
4638 array
[DEVCONF_HOPLIMIT
] = cnf
->hop_limit
;
4639 array
[DEVCONF_MTU6
] = cnf
->mtu6
;
4640 array
[DEVCONF_ACCEPT_RA
] = cnf
->accept_ra
;
4641 array
[DEVCONF_ACCEPT_REDIRECTS
] = cnf
->accept_redirects
;
4642 array
[DEVCONF_AUTOCONF
] = cnf
->autoconf
;
4643 array
[DEVCONF_DAD_TRANSMITS
] = cnf
->dad_transmits
;
4644 array
[DEVCONF_RTR_SOLICITS
] = cnf
->rtr_solicits
;
4645 array
[DEVCONF_RTR_SOLICIT_INTERVAL
] =
4646 jiffies_to_msecs(cnf
->rtr_solicit_interval
);
4647 array
[DEVCONF_RTR_SOLICIT_DELAY
] =
4648 jiffies_to_msecs(cnf
->rtr_solicit_delay
);
4649 array
[DEVCONF_FORCE_MLD_VERSION
] = cnf
->force_mld_version
;
4650 array
[DEVCONF_MLDV1_UNSOLICITED_REPORT_INTERVAL
] =
4651 jiffies_to_msecs(cnf
->mldv1_unsolicited_report_interval
);
4652 array
[DEVCONF_MLDV2_UNSOLICITED_REPORT_INTERVAL
] =
4653 jiffies_to_msecs(cnf
->mldv2_unsolicited_report_interval
);
4654 array
[DEVCONF_USE_TEMPADDR
] = cnf
->use_tempaddr
;
4655 array
[DEVCONF_TEMP_VALID_LFT
] = cnf
->temp_valid_lft
;
4656 array
[DEVCONF_TEMP_PREFERED_LFT
] = cnf
->temp_prefered_lft
;
4657 array
[DEVCONF_REGEN_MAX_RETRY
] = cnf
->regen_max_retry
;
4658 array
[DEVCONF_MAX_DESYNC_FACTOR
] = cnf
->max_desync_factor
;
4659 array
[DEVCONF_MAX_ADDRESSES
] = cnf
->max_addresses
;
4660 array
[DEVCONF_ACCEPT_RA_DEFRTR
] = cnf
->accept_ra_defrtr
;
4661 array
[DEVCONF_ACCEPT_RA_MIN_HOP_LIMIT
] = cnf
->accept_ra_min_hop_limit
;
4662 array
[DEVCONF_ACCEPT_RA_PINFO
] = cnf
->accept_ra_pinfo
;
4663 #ifdef CONFIG_IPV6_ROUTER_PREF
4664 array
[DEVCONF_ACCEPT_RA_RTR_PREF
] = cnf
->accept_ra_rtr_pref
;
4665 array
[DEVCONF_RTR_PROBE_INTERVAL
] =
4666 jiffies_to_msecs(cnf
->rtr_probe_interval
);
4667 #ifdef CONFIG_IPV6_ROUTE_INFO
4668 array
[DEVCONF_ACCEPT_RA_RT_INFO_MAX_PLEN
] = cnf
->accept_ra_rt_info_max_plen
;
4671 array
[DEVCONF_PROXY_NDP
] = cnf
->proxy_ndp
;
4672 array
[DEVCONF_ACCEPT_SOURCE_ROUTE
] = cnf
->accept_source_route
;
4673 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
4674 array
[DEVCONF_OPTIMISTIC_DAD
] = cnf
->optimistic_dad
;
4675 array
[DEVCONF_USE_OPTIMISTIC
] = cnf
->use_optimistic
;
4677 #ifdef CONFIG_IPV6_MROUTE
4678 array
[DEVCONF_MC_FORWARDING
] = cnf
->mc_forwarding
;
4680 array
[DEVCONF_DISABLE_IPV6
] = cnf
->disable_ipv6
;
4681 array
[DEVCONF_ACCEPT_DAD
] = cnf
->accept_dad
;
4682 array
[DEVCONF_FORCE_TLLAO
] = cnf
->force_tllao
;
4683 array
[DEVCONF_NDISC_NOTIFY
] = cnf
->ndisc_notify
;
4684 array
[DEVCONF_SUPPRESS_FRAG_NDISC
] = cnf
->suppress_frag_ndisc
;
4685 array
[DEVCONF_ACCEPT_RA_FROM_LOCAL
] = cnf
->accept_ra_from_local
;
4686 array
[DEVCONF_ACCEPT_RA_MTU
] = cnf
->accept_ra_mtu
;
4687 array
[DEVCONF_IGNORE_ROUTES_WITH_LINKDOWN
] = cnf
->ignore_routes_with_linkdown
;
4688 /* we omit DEVCONF_STABLE_SECRET for now */
4689 array
[DEVCONF_USE_OIF_ADDRS_ONLY
] = cnf
->use_oif_addrs_only
;
4692 static inline size_t inet6_ifla6_size(void)
4694 return nla_total_size(4) /* IFLA_INET6_FLAGS */
4695 + nla_total_size(sizeof(struct ifla_cacheinfo
))
4696 + nla_total_size(DEVCONF_MAX
* 4) /* IFLA_INET6_CONF */
4697 + nla_total_size(IPSTATS_MIB_MAX
* 8) /* IFLA_INET6_STATS */
4698 + nla_total_size(ICMP6_MIB_MAX
* 8) /* IFLA_INET6_ICMP6STATS */
4699 + nla_total_size(sizeof(struct in6_addr
)); /* IFLA_INET6_TOKEN */
4702 static inline size_t inet6_if_nlmsg_size(void)
4704 return NLMSG_ALIGN(sizeof(struct ifinfomsg
))
4705 + nla_total_size(IFNAMSIZ
) /* IFLA_IFNAME */
4706 + nla_total_size(MAX_ADDR_LEN
) /* IFLA_ADDRESS */
4707 + nla_total_size(4) /* IFLA_MTU */
4708 + nla_total_size(4) /* IFLA_LINK */
4709 + nla_total_size(1) /* IFLA_OPERSTATE */
4710 + nla_total_size(inet6_ifla6_size()); /* IFLA_PROTINFO */
4713 static inline void __snmp6_fill_statsdev(u64
*stats
, atomic_long_t
*mib
,
4714 int items
, int bytes
)
4717 int pad
= bytes
- sizeof(u64
) * items
;
4720 /* Use put_unaligned() because stats may not be aligned for u64. */
4721 put_unaligned(items
, &stats
[0]);
4722 for (i
= 1; i
< items
; i
++)
4723 put_unaligned(atomic_long_read(&mib
[i
]), &stats
[i
]);
4725 memset(&stats
[items
], 0, pad
);
4728 static inline void __snmp6_fill_stats64(u64
*stats
, void __percpu
*mib
,
4729 int items
, int bytes
, size_t syncpoff
)
4732 int pad
= bytes
- sizeof(u64
) * items
;
4735 /* Use put_unaligned() because stats may not be aligned for u64. */
4736 put_unaligned(items
, &stats
[0]);
4737 for (i
= 1; i
< items
; i
++)
4738 put_unaligned(snmp_fold_field64(mib
, i
, syncpoff
), &stats
[i
]);
4740 memset(&stats
[items
], 0, pad
);
4743 static void snmp6_fill_stats(u64
*stats
, struct inet6_dev
*idev
, int attrtype
,
4747 case IFLA_INET6_STATS
:
4748 __snmp6_fill_stats64(stats
, idev
->stats
.ipv6
,
4749 IPSTATS_MIB_MAX
, bytes
, offsetof(struct ipstats_mib
, syncp
));
4751 case IFLA_INET6_ICMP6STATS
:
4752 __snmp6_fill_statsdev(stats
, idev
->stats
.icmpv6dev
->mibs
, ICMP6_MIB_MAX
, bytes
);
4757 static int inet6_fill_ifla6_attrs(struct sk_buff
*skb
, struct inet6_dev
*idev
)
4760 struct ifla_cacheinfo ci
;
4762 if (nla_put_u32(skb
, IFLA_INET6_FLAGS
, idev
->if_flags
))
4763 goto nla_put_failure
;
4764 ci
.max_reasm_len
= IPV6_MAXPLEN
;
4765 ci
.tstamp
= cstamp_delta(idev
->tstamp
);
4766 ci
.reachable_time
= jiffies_to_msecs(idev
->nd_parms
->reachable_time
);
4767 ci
.retrans_time
= jiffies_to_msecs(NEIGH_VAR(idev
->nd_parms
, RETRANS_TIME
));
4768 if (nla_put(skb
, IFLA_INET6_CACHEINFO
, sizeof(ci
), &ci
))
4769 goto nla_put_failure
;
4770 nla
= nla_reserve(skb
, IFLA_INET6_CONF
, DEVCONF_MAX
* sizeof(s32
));
4772 goto nla_put_failure
;
4773 ipv6_store_devconf(&idev
->cnf
, nla_data(nla
), nla_len(nla
));
4775 /* XXX - MC not implemented */
4777 nla
= nla_reserve(skb
, IFLA_INET6_STATS
, IPSTATS_MIB_MAX
* sizeof(u64
));
4779 goto nla_put_failure
;
4780 snmp6_fill_stats(nla_data(nla
), idev
, IFLA_INET6_STATS
, nla_len(nla
));
4782 nla
= nla_reserve(skb
, IFLA_INET6_ICMP6STATS
, ICMP6_MIB_MAX
* sizeof(u64
));
4784 goto nla_put_failure
;
4785 snmp6_fill_stats(nla_data(nla
), idev
, IFLA_INET6_ICMP6STATS
, nla_len(nla
));
4787 nla
= nla_reserve(skb
, IFLA_INET6_TOKEN
, sizeof(struct in6_addr
));
4789 goto nla_put_failure
;
4791 if (nla_put_u8(skb
, IFLA_INET6_ADDR_GEN_MODE
, idev
->addr_gen_mode
))
4792 goto nla_put_failure
;
4794 read_lock_bh(&idev
->lock
);
4795 memcpy(nla_data(nla
), idev
->token
.s6_addr
, nla_len(nla
));
4796 read_unlock_bh(&idev
->lock
);
4804 static size_t inet6_get_link_af_size(const struct net_device
*dev
)
4806 if (!__in6_dev_get(dev
))
4809 return inet6_ifla6_size();
4812 static int inet6_fill_link_af(struct sk_buff
*skb
, const struct net_device
*dev
)
4814 struct inet6_dev
*idev
= __in6_dev_get(dev
);
4819 if (inet6_fill_ifla6_attrs(skb
, idev
) < 0)
4825 static int inet6_set_iftoken(struct inet6_dev
*idev
, struct in6_addr
*token
)
4827 struct inet6_ifaddr
*ifp
;
4828 struct net_device
*dev
= idev
->dev
;
4829 bool update_rs
= false;
4830 struct in6_addr ll_addr
;
4836 if (ipv6_addr_any(token
))
4838 if (dev
->flags
& (IFF_LOOPBACK
| IFF_NOARP
))
4840 if (!ipv6_accept_ra(idev
))
4842 if (idev
->cnf
.rtr_solicits
<= 0)
4845 write_lock_bh(&idev
->lock
);
4847 BUILD_BUG_ON(sizeof(token
->s6_addr
) != 16);
4848 memcpy(idev
->token
.s6_addr
+ 8, token
->s6_addr
+ 8, 8);
4850 write_unlock_bh(&idev
->lock
);
4852 if (!idev
->dead
&& (idev
->if_flags
& IF_READY
) &&
4853 !ipv6_get_lladdr(dev
, &ll_addr
, IFA_F_TENTATIVE
|
4854 IFA_F_OPTIMISTIC
)) {
4856 /* If we're not ready, then normal ifup will take care
4857 * of this. Otherwise, we need to request our rs here.
4859 ndisc_send_rs(dev
, &ll_addr
, &in6addr_linklocal_allrouters
);
4863 write_lock_bh(&idev
->lock
);
4866 idev
->if_flags
|= IF_RS_SENT
;
4867 idev
->rs_probes
= 1;
4868 addrconf_mod_rs_timer(idev
, idev
->cnf
.rtr_solicit_interval
);
4871 /* Well, that's kinda nasty ... */
4872 list_for_each_entry(ifp
, &idev
->addr_list
, if_list
) {
4873 spin_lock(&ifp
->lock
);
4874 if (ifp
->tokenized
) {
4876 ifp
->prefered_lft
= 0;
4878 spin_unlock(&ifp
->lock
);
4881 write_unlock_bh(&idev
->lock
);
4882 inet6_ifinfo_notify(RTM_NEWLINK
, idev
);
4883 addrconf_verify_rtnl();
4887 static const struct nla_policy inet6_af_policy
[IFLA_INET6_MAX
+ 1] = {
4888 [IFLA_INET6_ADDR_GEN_MODE
] = { .type
= NLA_U8
},
4889 [IFLA_INET6_TOKEN
] = { .len
= sizeof(struct in6_addr
) },
4892 static int inet6_validate_link_af(const struct net_device
*dev
,
4893 const struct nlattr
*nla
)
4895 struct nlattr
*tb
[IFLA_INET6_MAX
+ 1];
4897 if (dev
&& !__in6_dev_get(dev
))
4898 return -EAFNOSUPPORT
;
4900 return nla_parse_nested(tb
, IFLA_INET6_MAX
, nla
, inet6_af_policy
);
4903 static int inet6_set_link_af(struct net_device
*dev
, const struct nlattr
*nla
)
4906 struct inet6_dev
*idev
= __in6_dev_get(dev
);
4907 struct nlattr
*tb
[IFLA_INET6_MAX
+ 1];
4910 return -EAFNOSUPPORT
;
4912 if (nla_parse_nested(tb
, IFLA_INET6_MAX
, nla
, NULL
) < 0)
4915 if (tb
[IFLA_INET6_TOKEN
]) {
4916 err
= inet6_set_iftoken(idev
, nla_data(tb
[IFLA_INET6_TOKEN
]));
4921 if (tb
[IFLA_INET6_ADDR_GEN_MODE
]) {
4922 u8 mode
= nla_get_u8(tb
[IFLA_INET6_ADDR_GEN_MODE
]);
4924 if (mode
!= IN6_ADDR_GEN_MODE_EUI64
&&
4925 mode
!= IN6_ADDR_GEN_MODE_NONE
&&
4926 mode
!= IN6_ADDR_GEN_MODE_STABLE_PRIVACY
)
4929 if (mode
== IN6_ADDR_GEN_MODE_STABLE_PRIVACY
&&
4930 !idev
->cnf
.stable_secret
.initialized
&&
4931 !dev_net(dev
)->ipv6
.devconf_dflt
->stable_secret
.initialized
)
4934 idev
->addr_gen_mode
= mode
;
4941 static int inet6_fill_ifinfo(struct sk_buff
*skb
, struct inet6_dev
*idev
,
4942 u32 portid
, u32 seq
, int event
, unsigned int flags
)
4944 struct net_device
*dev
= idev
->dev
;
4945 struct ifinfomsg
*hdr
;
4946 struct nlmsghdr
*nlh
;
4949 nlh
= nlmsg_put(skb
, portid
, seq
, event
, sizeof(*hdr
), flags
);
4953 hdr
= nlmsg_data(nlh
);
4954 hdr
->ifi_family
= AF_INET6
;
4956 hdr
->ifi_type
= dev
->type
;
4957 hdr
->ifi_index
= dev
->ifindex
;
4958 hdr
->ifi_flags
= dev_get_flags(dev
);
4959 hdr
->ifi_change
= 0;
4961 if (nla_put_string(skb
, IFLA_IFNAME
, dev
->name
) ||
4963 nla_put(skb
, IFLA_ADDRESS
, dev
->addr_len
, dev
->dev_addr
)) ||
4964 nla_put_u32(skb
, IFLA_MTU
, dev
->mtu
) ||
4965 (dev
->ifindex
!= dev_get_iflink(dev
) &&
4966 nla_put_u32(skb
, IFLA_LINK
, dev_get_iflink(dev
))) ||
4967 nla_put_u8(skb
, IFLA_OPERSTATE
,
4968 netif_running(dev
) ? dev
->operstate
: IF_OPER_DOWN
))
4969 goto nla_put_failure
;
4970 protoinfo
= nla_nest_start(skb
, IFLA_PROTINFO
);
4972 goto nla_put_failure
;
4974 if (inet6_fill_ifla6_attrs(skb
, idev
) < 0)
4975 goto nla_put_failure
;
4977 nla_nest_end(skb
, protoinfo
);
4978 nlmsg_end(skb
, nlh
);
4982 nlmsg_cancel(skb
, nlh
);
4986 static int inet6_dump_ifinfo(struct sk_buff
*skb
, struct netlink_callback
*cb
)
4988 struct net
*net
= sock_net(skb
->sk
);
4991 struct net_device
*dev
;
4992 struct inet6_dev
*idev
;
4993 struct hlist_head
*head
;
4996 s_idx
= cb
->args
[1];
4999 for (h
= s_h
; h
< NETDEV_HASHENTRIES
; h
++, s_idx
= 0) {
5001 head
= &net
->dev_index_head
[h
];
5002 hlist_for_each_entry_rcu(dev
, head
, index_hlist
) {
5005 idev
= __in6_dev_get(dev
);
5008 if (inet6_fill_ifinfo(skb
, idev
,
5009 NETLINK_CB(cb
->skb
).portid
,
5011 RTM_NEWLINK
, NLM_F_MULTI
) < 0)
5025 void inet6_ifinfo_notify(int event
, struct inet6_dev
*idev
)
5027 struct sk_buff
*skb
;
5028 struct net
*net
= dev_net(idev
->dev
);
5031 skb
= nlmsg_new(inet6_if_nlmsg_size(), GFP_ATOMIC
);
5035 err
= inet6_fill_ifinfo(skb
, idev
, 0, 0, event
, 0);
5037 /* -EMSGSIZE implies BUG in inet6_if_nlmsg_size() */
5038 WARN_ON(err
== -EMSGSIZE
);
5042 rtnl_notify(skb
, net
, 0, RTNLGRP_IPV6_IFINFO
, NULL
, GFP_ATOMIC
);
5046 rtnl_set_sk_err(net
, RTNLGRP_IPV6_IFINFO
, err
);
5049 static inline size_t inet6_prefix_nlmsg_size(void)
5051 return NLMSG_ALIGN(sizeof(struct prefixmsg
))
5052 + nla_total_size(sizeof(struct in6_addr
))
5053 + nla_total_size(sizeof(struct prefix_cacheinfo
));
5056 static int inet6_fill_prefix(struct sk_buff
*skb
, struct inet6_dev
*idev
,
5057 struct prefix_info
*pinfo
, u32 portid
, u32 seq
,
5058 int event
, unsigned int flags
)
5060 struct prefixmsg
*pmsg
;
5061 struct nlmsghdr
*nlh
;
5062 struct prefix_cacheinfo ci
;
5064 nlh
= nlmsg_put(skb
, portid
, seq
, event
, sizeof(*pmsg
), flags
);
5068 pmsg
= nlmsg_data(nlh
);
5069 pmsg
->prefix_family
= AF_INET6
;
5070 pmsg
->prefix_pad1
= 0;
5071 pmsg
->prefix_pad2
= 0;
5072 pmsg
->prefix_ifindex
= idev
->dev
->ifindex
;
5073 pmsg
->prefix_len
= pinfo
->prefix_len
;
5074 pmsg
->prefix_type
= pinfo
->type
;
5075 pmsg
->prefix_pad3
= 0;
5076 pmsg
->prefix_flags
= 0;
5078 pmsg
->prefix_flags
|= IF_PREFIX_ONLINK
;
5079 if (pinfo
->autoconf
)
5080 pmsg
->prefix_flags
|= IF_PREFIX_AUTOCONF
;
5082 if (nla_put(skb
, PREFIX_ADDRESS
, sizeof(pinfo
->prefix
), &pinfo
->prefix
))
5083 goto nla_put_failure
;
5084 ci
.preferred_time
= ntohl(pinfo
->prefered
);
5085 ci
.valid_time
= ntohl(pinfo
->valid
);
5086 if (nla_put(skb
, PREFIX_CACHEINFO
, sizeof(ci
), &ci
))
5087 goto nla_put_failure
;
5088 nlmsg_end(skb
, nlh
);
5092 nlmsg_cancel(skb
, nlh
);
5096 static void inet6_prefix_notify(int event
, struct inet6_dev
*idev
,
5097 struct prefix_info
*pinfo
)
5099 struct sk_buff
*skb
;
5100 struct net
*net
= dev_net(idev
->dev
);
5103 skb
= nlmsg_new(inet6_prefix_nlmsg_size(), GFP_ATOMIC
);
5107 err
= inet6_fill_prefix(skb
, idev
, pinfo
, 0, 0, event
, 0);
5109 /* -EMSGSIZE implies BUG in inet6_prefix_nlmsg_size() */
5110 WARN_ON(err
== -EMSGSIZE
);
5114 rtnl_notify(skb
, net
, 0, RTNLGRP_IPV6_PREFIX
, NULL
, GFP_ATOMIC
);
5118 rtnl_set_sk_err(net
, RTNLGRP_IPV6_PREFIX
, err
);
5121 static void __ipv6_ifa_notify(int event
, struct inet6_ifaddr
*ifp
)
5123 struct net
*net
= dev_net(ifp
->idev
->dev
);
5128 inet6_ifa_notify(event
? : RTM_NEWADDR
, ifp
);
5133 * If the address was optimistic
5134 * we inserted the route at the start of
5135 * our DAD process, so we don't need
5138 if (!(ifp
->rt
->rt6i_node
))
5139 ip6_ins_rt(ifp
->rt
);
5140 if (ifp
->idev
->cnf
.forwarding
)
5141 addrconf_join_anycast(ifp
);
5142 if (!ipv6_addr_any(&ifp
->peer_addr
))
5143 addrconf_prefix_route(&ifp
->peer_addr
, 128,
5144 ifp
->idev
->dev
, 0, 0);
5147 if (ifp
->idev
->cnf
.forwarding
)
5148 addrconf_leave_anycast(ifp
);
5149 addrconf_leave_solict(ifp
->idev
, &ifp
->addr
);
5150 if (!ipv6_addr_any(&ifp
->peer_addr
)) {
5151 struct rt6_info
*rt
;
5153 rt
= addrconf_get_prefix_route(&ifp
->peer_addr
, 128,
5154 ifp
->idev
->dev
, 0, 0);
5155 if (rt
&& ip6_del_rt(rt
))
5158 dst_hold(&ifp
->rt
->dst
);
5160 if (ip6_del_rt(ifp
->rt
))
5161 dst_free(&ifp
->rt
->dst
);
5163 rt_genid_bump_ipv6(net
);
5166 atomic_inc(&net
->ipv6
.dev_addr_genid
);
5169 static void ipv6_ifa_notify(int event
, struct inet6_ifaddr
*ifp
)
5172 if (likely(ifp
->idev
->dead
== 0))
5173 __ipv6_ifa_notify(event
, ifp
);
5174 rcu_read_unlock_bh();
5177 #ifdef CONFIG_SYSCTL
5180 int addrconf_sysctl_forward(struct ctl_table
*ctl
, int write
,
5181 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
)
5183 int *valp
= ctl
->data
;
5186 struct ctl_table lctl
;
5190 * ctl->data points to idev->cnf.forwarding, we should
5191 * not modify it until we get the rtnl lock.
5196 ret
= proc_dointvec(&lctl
, write
, buffer
, lenp
, ppos
);
5199 ret
= addrconf_fixup_forwarding(ctl
, valp
, val
);
5206 int addrconf_sysctl_mtu(struct ctl_table
*ctl
, int write
,
5207 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
)
5209 struct inet6_dev
*idev
= ctl
->extra1
;
5210 int min_mtu
= IPV6_MIN_MTU
;
5211 struct ctl_table lctl
;
5214 lctl
.extra1
= &min_mtu
;
5215 lctl
.extra2
= idev
? &idev
->dev
->mtu
: NULL
;
5217 return proc_dointvec_minmax(&lctl
, write
, buffer
, lenp
, ppos
);
5220 static void dev_disable_change(struct inet6_dev
*idev
)
5222 struct netdev_notifier_info info
;
5224 if (!idev
|| !idev
->dev
)
5227 netdev_notifier_info_init(&info
, idev
->dev
);
5228 if (idev
->cnf
.disable_ipv6
)
5229 addrconf_notify(NULL
, NETDEV_DOWN
, &info
);
5231 addrconf_notify(NULL
, NETDEV_UP
, &info
);
5234 static void addrconf_disable_change(struct net
*net
, __s32 newf
)
5236 struct net_device
*dev
;
5237 struct inet6_dev
*idev
;
5240 for_each_netdev_rcu(net
, dev
) {
5241 idev
= __in6_dev_get(dev
);
5243 int changed
= (!idev
->cnf
.disable_ipv6
) ^ (!newf
);
5244 idev
->cnf
.disable_ipv6
= newf
;
5246 dev_disable_change(idev
);
5252 static int addrconf_disable_ipv6(struct ctl_table
*table
, int *p
, int newf
)
5257 if (!rtnl_trylock())
5258 return restart_syscall();
5260 net
= (struct net
*)table
->extra2
;
5264 if (p
== &net
->ipv6
.devconf_dflt
->disable_ipv6
) {
5269 if (p
== &net
->ipv6
.devconf_all
->disable_ipv6
) {
5270 net
->ipv6
.devconf_dflt
->disable_ipv6
= newf
;
5271 addrconf_disable_change(net
, newf
);
5272 } else if ((!newf
) ^ (!old
))
5273 dev_disable_change((struct inet6_dev
*)table
->extra1
);
5280 int addrconf_sysctl_disable(struct ctl_table
*ctl
, int write
,
5281 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
)
5283 int *valp
= ctl
->data
;
5286 struct ctl_table lctl
;
5290 * ctl->data points to idev->cnf.disable_ipv6, we should
5291 * not modify it until we get the rtnl lock.
5296 ret
= proc_dointvec(&lctl
, write
, buffer
, lenp
, ppos
);
5299 ret
= addrconf_disable_ipv6(ctl
, valp
, val
);
5306 int addrconf_sysctl_proxy_ndp(struct ctl_table
*ctl
, int write
,
5307 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
)
5309 int *valp
= ctl
->data
;
5314 ret
= proc_dointvec(ctl
, write
, buffer
, lenp
, ppos
);
5317 if (write
&& old
!= new) {
5318 struct net
*net
= ctl
->extra2
;
5320 if (!rtnl_trylock())
5321 return restart_syscall();
5323 if (valp
== &net
->ipv6
.devconf_dflt
->proxy_ndp
)
5324 inet6_netconf_notify_devconf(net
, NETCONFA_PROXY_NEIGH
,
5325 NETCONFA_IFINDEX_DEFAULT
,
5326 net
->ipv6
.devconf_dflt
);
5327 else if (valp
== &net
->ipv6
.devconf_all
->proxy_ndp
)
5328 inet6_netconf_notify_devconf(net
, NETCONFA_PROXY_NEIGH
,
5329 NETCONFA_IFINDEX_ALL
,
5330 net
->ipv6
.devconf_all
);
5332 struct inet6_dev
*idev
= ctl
->extra1
;
5334 inet6_netconf_notify_devconf(net
, NETCONFA_PROXY_NEIGH
,
5344 static int addrconf_sysctl_stable_secret(struct ctl_table
*ctl
, int write
,
5345 void __user
*buffer
, size_t *lenp
,
5349 struct in6_addr addr
;
5350 char str
[IPV6_MAX_STRLEN
];
5351 struct ctl_table lctl
= *ctl
;
5352 struct net
*net
= ctl
->extra2
;
5353 struct ipv6_stable_secret
*secret
= ctl
->data
;
5355 if (&net
->ipv6
.devconf_all
->stable_secret
== ctl
->data
)
5358 lctl
.maxlen
= IPV6_MAX_STRLEN
;
5361 if (!rtnl_trylock())
5362 return restart_syscall();
5364 if (!write
&& !secret
->initialized
) {
5370 err
= snprintf(str
, sizeof(str
), "%pI6",
5372 if (err
>= sizeof(str
)) {
5378 err
= proc_dostring(&lctl
, write
, buffer
, lenp
, ppos
);
5382 if (in6_pton(str
, -1, addr
.in6_u
.u6_addr8
, -1, NULL
) != 1) {
5387 secret
->initialized
= true;
5388 secret
->secret
= addr
;
5390 if (&net
->ipv6
.devconf_dflt
->stable_secret
== ctl
->data
) {
5391 struct net_device
*dev
;
5393 for_each_netdev(net
, dev
) {
5394 struct inet6_dev
*idev
= __in6_dev_get(dev
);
5397 idev
->addr_gen_mode
=
5398 IN6_ADDR_GEN_MODE_STABLE_PRIVACY
;
5402 struct inet6_dev
*idev
= ctl
->extra1
;
5404 idev
->addr_gen_mode
= IN6_ADDR_GEN_MODE_STABLE_PRIVACY
;
5414 int addrconf_sysctl_ignore_routes_with_linkdown(struct ctl_table
*ctl
,
5416 void __user
*buffer
,
5420 int *valp
= ctl
->data
;
5423 struct ctl_table lctl
;
5426 /* ctl->data points to idev->cnf.ignore_routes_when_linkdown
5427 * we should not modify it until we get the rtnl lock.
5432 ret
= proc_dointvec(&lctl
, write
, buffer
, lenp
, ppos
);
5435 ret
= addrconf_fixup_linkdown(ctl
, valp
, val
);
5441 static struct addrconf_sysctl_table
5443 struct ctl_table_header
*sysctl_header
;
5444 struct ctl_table addrconf_vars
[DEVCONF_MAX
+1];
5445 } addrconf_sysctl __read_mostly
= {
5446 .sysctl_header
= NULL
,
5449 .procname
= "forwarding",
5450 .data
= &ipv6_devconf
.forwarding
,
5451 .maxlen
= sizeof(int),
5453 .proc_handler
= addrconf_sysctl_forward
,
5456 .procname
= "hop_limit",
5457 .data
= &ipv6_devconf
.hop_limit
,
5458 .maxlen
= sizeof(int),
5460 .proc_handler
= proc_dointvec
,
5464 .data
= &ipv6_devconf
.mtu6
,
5465 .maxlen
= sizeof(int),
5467 .proc_handler
= addrconf_sysctl_mtu
,
5470 .procname
= "accept_ra",
5471 .data
= &ipv6_devconf
.accept_ra
,
5472 .maxlen
= sizeof(int),
5474 .proc_handler
= proc_dointvec
,
5477 .procname
= "accept_redirects",
5478 .data
= &ipv6_devconf
.accept_redirects
,
5479 .maxlen
= sizeof(int),
5481 .proc_handler
= proc_dointvec
,
5484 .procname
= "autoconf",
5485 .data
= &ipv6_devconf
.autoconf
,
5486 .maxlen
= sizeof(int),
5488 .proc_handler
= proc_dointvec
,
5491 .procname
= "dad_transmits",
5492 .data
= &ipv6_devconf
.dad_transmits
,
5493 .maxlen
= sizeof(int),
5495 .proc_handler
= proc_dointvec
,
5498 .procname
= "router_solicitations",
5499 .data
= &ipv6_devconf
.rtr_solicits
,
5500 .maxlen
= sizeof(int),
5502 .proc_handler
= proc_dointvec
,
5505 .procname
= "router_solicitation_interval",
5506 .data
= &ipv6_devconf
.rtr_solicit_interval
,
5507 .maxlen
= sizeof(int),
5509 .proc_handler
= proc_dointvec_jiffies
,
5512 .procname
= "router_solicitation_delay",
5513 .data
= &ipv6_devconf
.rtr_solicit_delay
,
5514 .maxlen
= sizeof(int),
5516 .proc_handler
= proc_dointvec_jiffies
,
5519 .procname
= "force_mld_version",
5520 .data
= &ipv6_devconf
.force_mld_version
,
5521 .maxlen
= sizeof(int),
5523 .proc_handler
= proc_dointvec
,
5526 .procname
= "mldv1_unsolicited_report_interval",
5528 &ipv6_devconf
.mldv1_unsolicited_report_interval
,
5529 .maxlen
= sizeof(int),
5531 .proc_handler
= proc_dointvec_ms_jiffies
,
5534 .procname
= "mldv2_unsolicited_report_interval",
5536 &ipv6_devconf
.mldv2_unsolicited_report_interval
,
5537 .maxlen
= sizeof(int),
5539 .proc_handler
= proc_dointvec_ms_jiffies
,
5542 .procname
= "use_tempaddr",
5543 .data
= &ipv6_devconf
.use_tempaddr
,
5544 .maxlen
= sizeof(int),
5546 .proc_handler
= proc_dointvec
,
5549 .procname
= "temp_valid_lft",
5550 .data
= &ipv6_devconf
.temp_valid_lft
,
5551 .maxlen
= sizeof(int),
5553 .proc_handler
= proc_dointvec
,
5556 .procname
= "temp_prefered_lft",
5557 .data
= &ipv6_devconf
.temp_prefered_lft
,
5558 .maxlen
= sizeof(int),
5560 .proc_handler
= proc_dointvec
,
5563 .procname
= "regen_max_retry",
5564 .data
= &ipv6_devconf
.regen_max_retry
,
5565 .maxlen
= sizeof(int),
5567 .proc_handler
= proc_dointvec
,
5570 .procname
= "max_desync_factor",
5571 .data
= &ipv6_devconf
.max_desync_factor
,
5572 .maxlen
= sizeof(int),
5574 .proc_handler
= proc_dointvec
,
5577 .procname
= "max_addresses",
5578 .data
= &ipv6_devconf
.max_addresses
,
5579 .maxlen
= sizeof(int),
5581 .proc_handler
= proc_dointvec
,
5584 .procname
= "accept_ra_defrtr",
5585 .data
= &ipv6_devconf
.accept_ra_defrtr
,
5586 .maxlen
= sizeof(int),
5588 .proc_handler
= proc_dointvec
,
5591 .procname
= "accept_ra_min_hop_limit",
5592 .data
= &ipv6_devconf
.accept_ra_min_hop_limit
,
5593 .maxlen
= sizeof(int),
5595 .proc_handler
= proc_dointvec
,
5598 .procname
= "accept_ra_pinfo",
5599 .data
= &ipv6_devconf
.accept_ra_pinfo
,
5600 .maxlen
= sizeof(int),
5602 .proc_handler
= proc_dointvec
,
5604 #ifdef CONFIG_IPV6_ROUTER_PREF
5606 .procname
= "accept_ra_rtr_pref",
5607 .data
= &ipv6_devconf
.accept_ra_rtr_pref
,
5608 .maxlen
= sizeof(int),
5610 .proc_handler
= proc_dointvec
,
5613 .procname
= "router_probe_interval",
5614 .data
= &ipv6_devconf
.rtr_probe_interval
,
5615 .maxlen
= sizeof(int),
5617 .proc_handler
= proc_dointvec_jiffies
,
5619 #ifdef CONFIG_IPV6_ROUTE_INFO
5621 .procname
= "accept_ra_rt_info_max_plen",
5622 .data
= &ipv6_devconf
.accept_ra_rt_info_max_plen
,
5623 .maxlen
= sizeof(int),
5625 .proc_handler
= proc_dointvec
,
5630 .procname
= "proxy_ndp",
5631 .data
= &ipv6_devconf
.proxy_ndp
,
5632 .maxlen
= sizeof(int),
5634 .proc_handler
= addrconf_sysctl_proxy_ndp
,
5637 .procname
= "accept_source_route",
5638 .data
= &ipv6_devconf
.accept_source_route
,
5639 .maxlen
= sizeof(int),
5641 .proc_handler
= proc_dointvec
,
5643 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
5645 .procname
= "optimistic_dad",
5646 .data
= &ipv6_devconf
.optimistic_dad
,
5647 .maxlen
= sizeof(int),
5649 .proc_handler
= proc_dointvec
,
5653 .procname
= "use_optimistic",
5654 .data
= &ipv6_devconf
.use_optimistic
,
5655 .maxlen
= sizeof(int),
5657 .proc_handler
= proc_dointvec
,
5661 #ifdef CONFIG_IPV6_MROUTE
5663 .procname
= "mc_forwarding",
5664 .data
= &ipv6_devconf
.mc_forwarding
,
5665 .maxlen
= sizeof(int),
5667 .proc_handler
= proc_dointvec
,
5671 .procname
= "disable_ipv6",
5672 .data
= &ipv6_devconf
.disable_ipv6
,
5673 .maxlen
= sizeof(int),
5675 .proc_handler
= addrconf_sysctl_disable
,
5678 .procname
= "accept_dad",
5679 .data
= &ipv6_devconf
.accept_dad
,
5680 .maxlen
= sizeof(int),
5682 .proc_handler
= proc_dointvec
,
5685 .procname
= "force_tllao",
5686 .data
= &ipv6_devconf
.force_tllao
,
5687 .maxlen
= sizeof(int),
5689 .proc_handler
= proc_dointvec
5692 .procname
= "ndisc_notify",
5693 .data
= &ipv6_devconf
.ndisc_notify
,
5694 .maxlen
= sizeof(int),
5696 .proc_handler
= proc_dointvec
5699 .procname
= "suppress_frag_ndisc",
5700 .data
= &ipv6_devconf
.suppress_frag_ndisc
,
5701 .maxlen
= sizeof(int),
5703 .proc_handler
= proc_dointvec
5706 .procname
= "accept_ra_from_local",
5707 .data
= &ipv6_devconf
.accept_ra_from_local
,
5708 .maxlen
= sizeof(int),
5710 .proc_handler
= proc_dointvec
,
5713 .procname
= "accept_ra_mtu",
5714 .data
= &ipv6_devconf
.accept_ra_mtu
,
5715 .maxlen
= sizeof(int),
5717 .proc_handler
= proc_dointvec
,
5720 .procname
= "stable_secret",
5721 .data
= &ipv6_devconf
.stable_secret
,
5722 .maxlen
= IPV6_MAX_STRLEN
,
5724 .proc_handler
= addrconf_sysctl_stable_secret
,
5727 .procname
= "use_oif_addrs_only",
5728 .data
= &ipv6_devconf
.use_oif_addrs_only
,
5729 .maxlen
= sizeof(int),
5731 .proc_handler
= proc_dointvec
,
5734 .procname
= "ignore_routes_with_linkdown",
5735 .data
= &ipv6_devconf
.ignore_routes_with_linkdown
,
5736 .maxlen
= sizeof(int),
5738 .proc_handler
= addrconf_sysctl_ignore_routes_with_linkdown
,
5746 static int __addrconf_sysctl_register(struct net
*net
, char *dev_name
,
5747 struct inet6_dev
*idev
, struct ipv6_devconf
*p
)
5750 struct addrconf_sysctl_table
*t
;
5751 char path
[sizeof("net/ipv6/conf/") + IFNAMSIZ
];
5753 t
= kmemdup(&addrconf_sysctl
, sizeof(*t
), GFP_KERNEL
);
5757 for (i
= 0; t
->addrconf_vars
[i
].data
; i
++) {
5758 t
->addrconf_vars
[i
].data
+= (char *)p
- (char *)&ipv6_devconf
;
5759 t
->addrconf_vars
[i
].extra1
= idev
; /* embedded; no ref */
5760 t
->addrconf_vars
[i
].extra2
= net
;
5763 snprintf(path
, sizeof(path
), "net/ipv6/conf/%s", dev_name
);
5765 t
->sysctl_header
= register_net_sysctl(net
, path
, t
->addrconf_vars
);
5766 if (!t
->sysctl_header
)
5778 static void __addrconf_sysctl_unregister(struct ipv6_devconf
*p
)
5780 struct addrconf_sysctl_table
*t
;
5787 unregister_net_sysctl_table(t
->sysctl_header
);
5791 static int addrconf_sysctl_register(struct inet6_dev
*idev
)
5795 if (!sysctl_dev_name_is_allowed(idev
->dev
->name
))
5798 err
= neigh_sysctl_register(idev
->dev
, idev
->nd_parms
,
5799 &ndisc_ifinfo_sysctl_change
);
5802 err
= __addrconf_sysctl_register(dev_net(idev
->dev
), idev
->dev
->name
,
5805 neigh_sysctl_unregister(idev
->nd_parms
);
5810 static void addrconf_sysctl_unregister(struct inet6_dev
*idev
)
5812 __addrconf_sysctl_unregister(&idev
->cnf
);
5813 neigh_sysctl_unregister(idev
->nd_parms
);
5819 static int __net_init
addrconf_init_net(struct net
*net
)
5822 struct ipv6_devconf
*all
, *dflt
;
5824 all
= kmemdup(&ipv6_devconf
, sizeof(ipv6_devconf
), GFP_KERNEL
);
5828 dflt
= kmemdup(&ipv6_devconf_dflt
, sizeof(ipv6_devconf_dflt
), GFP_KERNEL
);
5830 goto err_alloc_dflt
;
5832 /* these will be inherited by all namespaces */
5833 dflt
->autoconf
= ipv6_defaults
.autoconf
;
5834 dflt
->disable_ipv6
= ipv6_defaults
.disable_ipv6
;
5836 dflt
->stable_secret
.initialized
= false;
5837 all
->stable_secret
.initialized
= false;
5839 net
->ipv6
.devconf_all
= all
;
5840 net
->ipv6
.devconf_dflt
= dflt
;
5842 #ifdef CONFIG_SYSCTL
5843 err
= __addrconf_sysctl_register(net
, "all", NULL
, all
);
5847 err
= __addrconf_sysctl_register(net
, "default", NULL
, dflt
);
5853 #ifdef CONFIG_SYSCTL
5855 __addrconf_sysctl_unregister(all
);
5865 static void __net_exit
addrconf_exit_net(struct net
*net
)
5867 #ifdef CONFIG_SYSCTL
5868 __addrconf_sysctl_unregister(net
->ipv6
.devconf_dflt
);
5869 __addrconf_sysctl_unregister(net
->ipv6
.devconf_all
);
5871 kfree(net
->ipv6
.devconf_dflt
);
5872 kfree(net
->ipv6
.devconf_all
);
5875 static struct pernet_operations addrconf_ops
= {
5876 .init
= addrconf_init_net
,
5877 .exit
= addrconf_exit_net
,
5880 static struct rtnl_af_ops inet6_ops __read_mostly
= {
5882 .fill_link_af
= inet6_fill_link_af
,
5883 .get_link_af_size
= inet6_get_link_af_size
,
5884 .validate_link_af
= inet6_validate_link_af
,
5885 .set_link_af
= inet6_set_link_af
,
5889 * Init / cleanup code
5892 int __init
addrconf_init(void)
5894 struct inet6_dev
*idev
;
5897 err
= ipv6_addr_label_init();
5899 pr_crit("%s: cannot initialize default policy table: %d\n",
5904 err
= register_pernet_subsys(&addrconf_ops
);
5908 addrconf_wq
= create_workqueue("ipv6_addrconf");
5914 /* The addrconf netdev notifier requires that loopback_dev
5915 * has it's ipv6 private information allocated and setup
5916 * before it can bring up and give link-local addresses
5917 * to other devices which are up.
5919 * Unfortunately, loopback_dev is not necessarily the first
5920 * entry in the global dev_base list of net devices. In fact,
5921 * it is likely to be the very last entry on that list.
5922 * So this causes the notifier registry below to try and
5923 * give link-local addresses to all devices besides loopback_dev
5924 * first, then loopback_dev, which cases all the non-loopback_dev
5925 * devices to fail to get a link-local address.
5927 * So, as a temporary fix, allocate the ipv6 structure for
5928 * loopback_dev first by hand.
5929 * Longer term, all of the dependencies ipv6 has upon the loopback
5930 * device and it being up should be removed.
5933 idev
= ipv6_add_dev(init_net
.loopback_dev
);
5936 err
= PTR_ERR(idev
);
5940 for (i
= 0; i
< IN6_ADDR_HSIZE
; i
++)
5941 INIT_HLIST_HEAD(&inet6_addr_lst
[i
]);
5943 register_netdevice_notifier(&ipv6_dev_notf
);
5947 rtnl_af_register(&inet6_ops
);
5949 err
= __rtnl_register(PF_INET6
, RTM_GETLINK
, NULL
, inet6_dump_ifinfo
,
5954 /* Only the first call to __rtnl_register can fail */
5955 __rtnl_register(PF_INET6
, RTM_NEWADDR
, inet6_rtm_newaddr
, NULL
, NULL
);
5956 __rtnl_register(PF_INET6
, RTM_DELADDR
, inet6_rtm_deladdr
, NULL
, NULL
);
5957 __rtnl_register(PF_INET6
, RTM_GETADDR
, inet6_rtm_getaddr
,
5958 inet6_dump_ifaddr
, NULL
);
5959 __rtnl_register(PF_INET6
, RTM_GETMULTICAST
, NULL
,
5960 inet6_dump_ifmcaddr
, NULL
);
5961 __rtnl_register(PF_INET6
, RTM_GETANYCAST
, NULL
,
5962 inet6_dump_ifacaddr
, NULL
);
5963 __rtnl_register(PF_INET6
, RTM_GETNETCONF
, inet6_netconf_get_devconf
,
5964 inet6_netconf_dump_devconf
, NULL
);
5966 ipv6_addr_label_rtnl_register();
5970 rtnl_af_unregister(&inet6_ops
);
5971 unregister_netdevice_notifier(&ipv6_dev_notf
);
5973 destroy_workqueue(addrconf_wq
);
5975 unregister_pernet_subsys(&addrconf_ops
);
5977 ipv6_addr_label_cleanup();
5982 void addrconf_cleanup(void)
5984 struct net_device
*dev
;
5987 unregister_netdevice_notifier(&ipv6_dev_notf
);
5988 unregister_pernet_subsys(&addrconf_ops
);
5989 ipv6_addr_label_cleanup();
5993 __rtnl_af_unregister(&inet6_ops
);
5995 /* clean dev list */
5996 for_each_netdev(&init_net
, dev
) {
5997 if (__in6_dev_get(dev
) == NULL
)
5999 addrconf_ifdown(dev
, 1);
6001 addrconf_ifdown(init_net
.loopback_dev
, 2);
6006 spin_lock_bh(&addrconf_hash_lock
);
6007 for (i
= 0; i
< IN6_ADDR_HSIZE
; i
++)
6008 WARN_ON(!hlist_empty(&inet6_addr_lst
[i
]));
6009 spin_unlock_bh(&addrconf_hash_lock
);
6010 cancel_delayed_work(&addr_chk_work
);
6013 destroy_workqueue(addrconf_wq
);