1 /*
2 * IPv6 Address [auto]configuration
3 * Linux INET6 implementation
4 *
5 * Authors:
6 * Pedro Roque <roque@di.fc.ul.pt>
7 * Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
8 *
9 * This program is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU General Public License
11 * as published by the Free Software Foundation; either version
12 * 2 of the License, or (at your option) any later version.
13 */
14
15 /*
16 * Changes:
17 *
18 * Janos Farkas : delete timer on ifdown
19 * <chexum@bankinf.banki.hu>
20 * Andi Kleen : kill double kfree on module
21 * unload.
22 * Maciej W. Rozycki : FDDI support
23 * sekiya@USAGI : Don't send too many RS
24 * packets.
25 * yoshfuji@USAGI : Fixed interval between DAD
26 * packets.
27 * YOSHIFUJI Hideaki @USAGI : improved accuracy of
28 * address validation timer.
29 * YOSHIFUJI Hideaki @USAGI : Privacy Extensions (RFC3041)
30 * support.
31 * Yuji SEKIYA @USAGI : Don't assign a same IPv6
32 * address on a same interface.
33 * YOSHIFUJI Hideaki @USAGI : ARCnet support
34 * YOSHIFUJI Hideaki @USAGI : convert /proc/net/if_inet6 to
35 * seq_file.
36 * YOSHIFUJI Hideaki @USAGI : improved source address
37 * selection; consider scope,
38 * status etc.
39 */
40
41 #define pr_fmt(fmt) "IPv6: " fmt
42
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/in6.h>
50 #include <linux/netdevice.h>
51 #include <linux/if_addr.h>
52 #include <linux/if_arp.h>
53 #include <linux/if_arcnet.h>
54 #include <linux/if_infiniband.h>
55 #include <linux/route.h>
56 #include <linux/inetdevice.h>
57 #include <linux/init.h>
58 #include <linux/slab.h>
59 #ifdef CONFIG_SYSCTL
60 #include <linux/sysctl.h>
61 #endif
62 #include <linux/capability.h>
63 #include <linux/delay.h>
64 #include <linux/notifier.h>
65 #include <linux/string.h>
66 #include <linux/hash.h>
67
68 #include <net/net_namespace.h>
69 #include <net/sock.h>
70 #include <net/snmp.h>
71
72 #include <net/af_ieee802154.h>
73 #include <net/firewire.h>
74 #include <net/ipv6.h>
75 #include <net/protocol.h>
76 #include <net/ndisc.h>
77 #include <net/ip6_route.h>
78 #include <net/addrconf.h>
79 #include <net/tcp.h>
80 #include <net/ip.h>
81 #include <net/netlink.h>
82 #include <net/pkt_sched.h>
83 #include <linux/if_tunnel.h>
84 #include <linux/rtnetlink.h>
85 #include <linux/netconf.h>
86 #include <linux/random.h>
87 #include <linux/uaccess.h>
88 #include <asm/unaligned.h>
89
90 #include <linux/proc_fs.h>
91 #include <linux/seq_file.h>
92 #include <linux/export.h>
93
94 /* Set to 3 to get tracing... */
95 #define ACONF_DEBUG 2
96
97 #if ACONF_DEBUG >= 3
98 #define ADBG(fmt, ...) printk(fmt, ##__VA_ARGS__)
99 #else
100 #define ADBG(fmt, ...) do { if (0) printk(fmt, ##__VA_ARGS__); } while (0)
101 #endif
102
103 #define INFINITY_LIFE_TIME 0xFFFFFFFF
104
cstamp_delta(unsigned long cstamp)105 static inline u32 cstamp_delta(unsigned long cstamp)
106 {
107 return (cstamp - INITIAL_JIFFIES) * 100UL / HZ;
108 }
109
rfc3315_s14_backoff_init(s32 irt)110 static inline s32 rfc3315_s14_backoff_init(s32 irt)
111 {
112 /* multiply 'initial retransmission time' by 0.9 .. 1.1 */
113 u64 tmp = (900000 + prandom_u32() % 200001) * (u64)irt;
114 do_div(tmp, 1000000);
115 return (s32)tmp;
116 }
117
rfc3315_s14_backoff_update(s32 rt,s32 mrt)118 static inline s32 rfc3315_s14_backoff_update(s32 rt, s32 mrt)
119 {
120 /* multiply 'retransmission timeout' by 1.9 .. 2.1 */
121 u64 tmp = (1900000 + prandom_u32() % 200001) * (u64)rt;
122 do_div(tmp, 1000000);
123 if ((s32)tmp > mrt) {
124 /* multiply 'maximum retransmission time' by 0.9 .. 1.1 */
125 tmp = (900000 + prandom_u32() % 200001) * (u64)mrt;
126 do_div(tmp, 1000000);
127 }
128 return (s32)tmp;
129 }
130
131 #ifdef CONFIG_SYSCTL
132 static int addrconf_sysctl_register(struct inet6_dev *idev);
133 static void addrconf_sysctl_unregister(struct inet6_dev *idev);
134 #else
addrconf_sysctl_register(struct inet6_dev * idev)135 static inline int addrconf_sysctl_register(struct inet6_dev *idev)
136 {
137 return 0;
138 }
139
addrconf_sysctl_unregister(struct inet6_dev * idev)140 static inline void addrconf_sysctl_unregister(struct inet6_dev *idev)
141 {
142 }
143 #endif
144
145 static void __ipv6_regen_rndid(struct inet6_dev *idev);
146 static void __ipv6_try_regen_rndid(struct inet6_dev *idev, struct in6_addr *tmpaddr);
147 static void ipv6_regen_rndid(unsigned long data);
148
149 static int ipv6_generate_eui64(u8 *eui, struct net_device *dev);
150 static int ipv6_count_addresses(struct inet6_dev *idev);
151
152 /*
153 * Configured unicast address hash table
154 */
155 static struct hlist_head inet6_addr_lst[IN6_ADDR_HSIZE];
156 static DEFINE_SPINLOCK(addrconf_hash_lock);
157
158 static void addrconf_verify(void);
159 static void addrconf_verify_rtnl(void);
160 static void addrconf_verify_work(struct work_struct *);
161
162 static struct workqueue_struct *addrconf_wq;
163 static DECLARE_DELAYED_WORK(addr_chk_work, addrconf_verify_work);
164
165 static void addrconf_join_anycast(struct inet6_ifaddr *ifp);
166 static void addrconf_leave_anycast(struct inet6_ifaddr *ifp);
167
168 static void addrconf_type_change(struct net_device *dev,
169 unsigned long event);
170 static int addrconf_ifdown(struct net_device *dev, int how);
171
172 static struct rt6_info *addrconf_get_prefix_route(const struct in6_addr *pfx,
173 int plen,
174 const struct net_device *dev,
175 u32 flags, u32 noflags);
176
177 static void addrconf_dad_start(struct inet6_ifaddr *ifp);
178 static void addrconf_dad_work(struct work_struct *w);
179 static void addrconf_dad_completed(struct inet6_ifaddr *ifp);
180 static void addrconf_dad_run(struct inet6_dev *idev);
181 static void addrconf_rs_timer(unsigned long data);
182 static void __ipv6_ifa_notify(int event, struct inet6_ifaddr *ifa);
183 static void ipv6_ifa_notify(int event, struct inet6_ifaddr *ifa);
184
185 static void inet6_prefix_notify(int event, struct inet6_dev *idev,
186 struct prefix_info *pinfo);
187 static bool ipv6_chk_same_addr(struct net *net, const struct in6_addr *addr,
188 struct net_device *dev);
189
190 static struct ipv6_devconf ipv6_devconf __read_mostly = {
191 .forwarding = 0,
192 .hop_limit = IPV6_DEFAULT_HOPLIMIT,
193 .mtu6 = IPV6_MIN_MTU,
194 .accept_ra = 1,
195 .accept_redirects = 1,
196 .autoconf = 1,
197 .force_mld_version = 0,
198 .mldv1_unsolicited_report_interval = 10 * HZ,
199 .mldv2_unsolicited_report_interval = HZ,
200 .dad_transmits = 1,
201 .rtr_solicits = MAX_RTR_SOLICITATIONS,
202 .rtr_solicit_interval = RTR_SOLICITATION_INTERVAL,
203 .rtr_solicit_max_interval = RTR_SOLICITATION_MAX_INTERVAL,
204 .rtr_solicit_delay = MAX_RTR_SOLICITATION_DELAY,
205 .use_tempaddr = 0,
206 .temp_valid_lft = TEMP_VALID_LIFETIME,
207 .temp_prefered_lft = TEMP_PREFERRED_LIFETIME,
208 .regen_max_retry = REGEN_MAX_RETRY,
209 .max_desync_factor = MAX_DESYNC_FACTOR,
210 .max_addresses = IPV6_MAX_ADDRESSES,
211 .accept_ra_defrtr = 1,
212 .accept_ra_from_local = 0,
213 .accept_ra_pinfo = 1,
214 #ifdef CONFIG_IPV6_ROUTER_PREF
215 .accept_ra_rtr_pref = 1,
216 .rtr_probe_interval = 60 * HZ,
217 #ifdef CONFIG_IPV6_ROUTE_INFO
218 .accept_ra_rt_info_min_plen = 0,
219 .accept_ra_rt_info_max_plen = 0,
220 #endif
221 #endif
222 .accept_ra_rt_table = 0,
223 .proxy_ndp = 0,
224 .accept_source_route = 0, /* we do not accept RH0 by default. */
225 .disable_ipv6 = 0,
226 .accept_dad = 1,
227 .suppress_frag_ndisc = 1,
228 .use_oif_addrs_only = 0,
229 };
230
231 static struct ipv6_devconf ipv6_devconf_dflt __read_mostly = {
232 .forwarding = 0,
233 .hop_limit = IPV6_DEFAULT_HOPLIMIT,
234 .mtu6 = IPV6_MIN_MTU,
235 .accept_ra = 1,
236 .accept_redirects = 1,
237 .autoconf = 1,
238 .force_mld_version = 0,
239 .mldv1_unsolicited_report_interval = 10 * HZ,
240 .mldv2_unsolicited_report_interval = HZ,
241 .dad_transmits = 1,
242 .rtr_solicits = MAX_RTR_SOLICITATIONS,
243 .rtr_solicit_interval = RTR_SOLICITATION_INTERVAL,
244 .rtr_solicit_max_interval = RTR_SOLICITATION_MAX_INTERVAL,
245 .rtr_solicit_delay = MAX_RTR_SOLICITATION_DELAY,
246 .use_tempaddr = 0,
247 .temp_valid_lft = TEMP_VALID_LIFETIME,
248 .temp_prefered_lft = TEMP_PREFERRED_LIFETIME,
249 .regen_max_retry = REGEN_MAX_RETRY,
250 .max_desync_factor = MAX_DESYNC_FACTOR,
251 .max_addresses = IPV6_MAX_ADDRESSES,
252 .accept_ra_defrtr = 1,
253 .accept_ra_from_local = 0,
254 .accept_ra_pinfo = 1,
255 #ifdef CONFIG_IPV6_ROUTER_PREF
256 .accept_ra_rtr_pref = 1,
257 .rtr_probe_interval = 60 * HZ,
258 #ifdef CONFIG_IPV6_ROUTE_INFO
259 .accept_ra_rt_info_min_plen = 0,
260 .accept_ra_rt_info_max_plen = 0,
261 #endif
262 #endif
263 .accept_ra_rt_table = 0,
264 .proxy_ndp = 0,
265 .accept_source_route = 0, /* we do not accept RH0 by default. */
266 .disable_ipv6 = 0,
267 .accept_dad = 1,
268 .suppress_frag_ndisc = 1,
269 .use_oif_addrs_only = 0,
270 };
271
272 /* Check if a valid qdisc is available */
addrconf_qdisc_ok(const struct net_device * dev)273 static inline bool addrconf_qdisc_ok(const struct net_device *dev)
274 {
275 return !qdisc_tx_is_noop(dev);
276 }
277
addrconf_del_rs_timer(struct inet6_dev * idev)278 static void addrconf_del_rs_timer(struct inet6_dev *idev)
279 {
280 if (del_timer(&idev->rs_timer))
281 __in6_dev_put(idev);
282 }
283
addrconf_del_dad_work(struct inet6_ifaddr * ifp)284 static void addrconf_del_dad_work(struct inet6_ifaddr *ifp)
285 {
286 if (cancel_delayed_work(&ifp->dad_work))
287 __in6_ifa_put(ifp);
288 }
289
addrconf_mod_rs_timer(struct inet6_dev * idev,unsigned long when)290 static void addrconf_mod_rs_timer(struct inet6_dev *idev,
291 unsigned long when)
292 {
293 if (!timer_pending(&idev->rs_timer))
294 in6_dev_hold(idev);
295 mod_timer(&idev->rs_timer, jiffies + when);
296 }
297
addrconf_mod_dad_work(struct inet6_ifaddr * ifp,unsigned long delay)298 static void addrconf_mod_dad_work(struct inet6_ifaddr *ifp,
299 unsigned long delay)
300 {
301 in6_ifa_hold(ifp);
302 if (mod_delayed_work(addrconf_wq, &ifp->dad_work, delay))
303 in6_ifa_put(ifp);
304 }
305
snmp6_alloc_dev(struct inet6_dev * idev)306 static int snmp6_alloc_dev(struct inet6_dev *idev)
307 {
308 int i;
309
310 idev->stats.ipv6 = alloc_percpu(struct ipstats_mib);
311 if (!idev->stats.ipv6)
312 goto err_ip;
313
314 for_each_possible_cpu(i) {
315 struct ipstats_mib *addrconf_stats;
316 addrconf_stats = per_cpu_ptr(idev->stats.ipv6, i);
317 u64_stats_init(&addrconf_stats->syncp);
318 }
319
320
321 idev->stats.icmpv6dev = kzalloc(sizeof(struct icmpv6_mib_device),
322 GFP_KERNEL);
323 if (!idev->stats.icmpv6dev)
324 goto err_icmp;
325 idev->stats.icmpv6msgdev = kzalloc(sizeof(struct icmpv6msg_mib_device),
326 GFP_KERNEL);
327 if (!idev->stats.icmpv6msgdev)
328 goto err_icmpmsg;
329
330 return 0;
331
332 err_icmpmsg:
333 kfree(idev->stats.icmpv6dev);
334 err_icmp:
335 free_percpu(idev->stats.ipv6);
336 err_ip:
337 return -ENOMEM;
338 }
339
ipv6_add_dev(struct net_device * dev)340 static struct inet6_dev *ipv6_add_dev(struct net_device *dev)
341 {
342 struct inet6_dev *ndev;
343 int err = -ENOMEM;
344
345 ASSERT_RTNL();
346
347 if (dev->mtu < IPV6_MIN_MTU)
348 return ERR_PTR(-EINVAL);
349
350 ndev = kzalloc(sizeof(struct inet6_dev), GFP_KERNEL);
351 if (ndev == NULL)
352 return ERR_PTR(err);
353
354 rwlock_init(&ndev->lock);
355 ndev->dev = dev;
356 INIT_LIST_HEAD(&ndev->addr_list);
357 setup_timer(&ndev->rs_timer, addrconf_rs_timer,
358 (unsigned long)ndev);
359 memcpy(&ndev->cnf, dev_net(dev)->ipv6.devconf_dflt, sizeof(ndev->cnf));
360 ndev->cnf.mtu6 = dev->mtu;
361 ndev->cnf.sysctl = NULL;
362 ndev->nd_parms = neigh_parms_alloc(dev, &nd_tbl);
363 if (ndev->nd_parms == NULL) {
364 kfree(ndev);
365 return ERR_PTR(err);
366 }
367 if (ndev->cnf.forwarding)
368 dev_disable_lro(dev);
369 /* We refer to the device */
370 dev_hold(dev);
371
372 if (snmp6_alloc_dev(ndev) < 0) {
373 ADBG(KERN_WARNING
374 "%s: cannot allocate memory for statistics; dev=%s.\n",
375 __func__, dev->name);
376 neigh_parms_release(&nd_tbl, ndev->nd_parms);
377 dev_put(dev);
378 kfree(ndev);
379 return ERR_PTR(err);
380 }
381
382 if (snmp6_register_dev(ndev) < 0) {
383 ADBG(KERN_WARNING
384 "%s: cannot create /proc/net/dev_snmp6/%s\n",
385 __func__, dev->name);
386 goto err_release;
387 }
388
389 /* One reference from device. We must do this before
390 * we invoke __ipv6_regen_rndid().
391 */
392 in6_dev_hold(ndev);
393
394 if (dev->flags & (IFF_NOARP | IFF_LOOPBACK))
395 ndev->cnf.accept_dad = -1;
396
397 #if IS_ENABLED(CONFIG_IPV6_SIT)
398 if (dev->type == ARPHRD_SIT && (dev->priv_flags & IFF_ISATAP)) {
399 pr_info("%s: Disabled Multicast RS\n", dev->name);
400 ndev->cnf.rtr_solicits = 0;
401 }
402 #endif
403
404 INIT_LIST_HEAD(&ndev->tempaddr_list);
405 setup_timer(&ndev->regen_timer, ipv6_regen_rndid, (unsigned long)ndev);
406 if ((dev->flags&IFF_LOOPBACK) ||
407 dev->type == ARPHRD_TUNNEL ||
408 dev->type == ARPHRD_TUNNEL6 ||
409 dev->type == ARPHRD_SIT ||
410 dev->type == ARPHRD_NONE) {
411 ndev->cnf.use_tempaddr = -1;
412 } else {
413 in6_dev_hold(ndev);
414 ipv6_regen_rndid((unsigned long) ndev);
415 }
416
417 ndev->token = in6addr_any;
418
419 if (netif_running(dev) && addrconf_qdisc_ok(dev))
420 ndev->if_flags |= IF_READY;
421
422 ipv6_mc_init_dev(ndev);
423 ndev->tstamp = jiffies;
424 err = addrconf_sysctl_register(ndev);
425 if (err) {
426 ipv6_mc_destroy_dev(ndev);
427 del_timer(&ndev->regen_timer);
428 snmp6_unregister_dev(ndev);
429 goto err_release;
430 }
431 /* protected by rtnl_lock */
432 rcu_assign_pointer(dev->ip6_ptr, ndev);
433
434 /* Join interface-local all-node multicast group */
435 ipv6_dev_mc_inc(dev, &in6addr_interfacelocal_allnodes);
436
437 /* Join all-node multicast group */
438 ipv6_dev_mc_inc(dev, &in6addr_linklocal_allnodes);
439
440 /* Join all-router multicast group if forwarding is set */
441 if (ndev->cnf.forwarding && (dev->flags & IFF_MULTICAST))
442 ipv6_dev_mc_inc(dev, &in6addr_linklocal_allrouters);
443
444 return ndev;
445
446 err_release:
447 neigh_parms_release(&nd_tbl, ndev->nd_parms);
448 ndev->dead = 1;
449 in6_dev_finish_destroy(ndev);
450 return ERR_PTR(err);
451 }
452
ipv6_find_idev(struct net_device * dev)453 static struct inet6_dev *ipv6_find_idev(struct net_device *dev)
454 {
455 struct inet6_dev *idev;
456
457 ASSERT_RTNL();
458
459 idev = __in6_dev_get(dev);
460 if (!idev) {
461 idev = ipv6_add_dev(dev);
462 if (IS_ERR(idev))
463 return NULL;
464 }
465
466 if (dev->flags&IFF_UP)
467 ipv6_mc_up(idev);
468 return idev;
469 }
470
inet6_netconf_msgsize_devconf(int type)471 static int inet6_netconf_msgsize_devconf(int type)
472 {
473 int size = NLMSG_ALIGN(sizeof(struct netconfmsg))
474 + nla_total_size(4); /* NETCONFA_IFINDEX */
475
476 /* type -1 is used for ALL */
477 if (type == -1 || type == NETCONFA_FORWARDING)
478 size += nla_total_size(4);
479 #ifdef CONFIG_IPV6_MROUTE
480 if (type == -1 || type == NETCONFA_MC_FORWARDING)
481 size += nla_total_size(4);
482 #endif
483 if (type == -1 || type == NETCONFA_PROXY_NEIGH)
484 size += nla_total_size(4);
485
486 return size;
487 }
488
inet6_netconf_fill_devconf(struct sk_buff * skb,int ifindex,struct ipv6_devconf * devconf,u32 portid,u32 seq,int event,unsigned int flags,int type)489 static int inet6_netconf_fill_devconf(struct sk_buff *skb, int ifindex,
490 struct ipv6_devconf *devconf, u32 portid,
491 u32 seq, int event, unsigned int flags,
492 int type)
493 {
494 struct nlmsghdr *nlh;
495 struct netconfmsg *ncm;
496
497 nlh = nlmsg_put(skb, portid, seq, event, sizeof(struct netconfmsg),
498 flags);
499 if (nlh == NULL)
500 return -EMSGSIZE;
501
502 ncm = nlmsg_data(nlh);
503 ncm->ncm_family = AF_INET6;
504
505 if (nla_put_s32(skb, NETCONFA_IFINDEX, ifindex) < 0)
506 goto nla_put_failure;
507
508 /* type -1 is used for ALL */
509 if ((type == -1 || type == NETCONFA_FORWARDING) &&
510 nla_put_s32(skb, NETCONFA_FORWARDING, devconf->forwarding) < 0)
511 goto nla_put_failure;
512 #ifdef CONFIG_IPV6_MROUTE
513 if ((type == -1 || type == NETCONFA_MC_FORWARDING) &&
514 nla_put_s32(skb, NETCONFA_MC_FORWARDING,
515 devconf->mc_forwarding) < 0)
516 goto nla_put_failure;
517 #endif
518 if ((type == -1 || type == NETCONFA_PROXY_NEIGH) &&
519 nla_put_s32(skb, NETCONFA_PROXY_NEIGH, devconf->proxy_ndp) < 0)
520 goto nla_put_failure;
521
522 return nlmsg_end(skb, nlh);
523
524 nla_put_failure:
525 nlmsg_cancel(skb, nlh);
526 return -EMSGSIZE;
527 }
528
inet6_netconf_notify_devconf(struct net * net,int type,int ifindex,struct ipv6_devconf * devconf)529 void inet6_netconf_notify_devconf(struct net *net, int type, int ifindex,
530 struct ipv6_devconf *devconf)
531 {
532 struct sk_buff *skb;
533 int err = -ENOBUFS;
534
535 skb = nlmsg_new(inet6_netconf_msgsize_devconf(type), GFP_ATOMIC);
536 if (skb == NULL)
537 goto errout;
538
539 err = inet6_netconf_fill_devconf(skb, ifindex, devconf, 0, 0,
540 RTM_NEWNETCONF, 0, type);
541 if (err < 0) {
542 /* -EMSGSIZE implies BUG in inet6_netconf_msgsize_devconf() */
543 WARN_ON(err == -EMSGSIZE);
544 kfree_skb(skb);
545 goto errout;
546 }
547 rtnl_notify(skb, net, 0, RTNLGRP_IPV6_NETCONF, NULL, GFP_ATOMIC);
548 return;
549 errout:
550 rtnl_set_sk_err(net, RTNLGRP_IPV6_NETCONF, err);
551 }
552
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 };
558
inet6_netconf_get_devconf(struct sk_buff * in_skb,struct nlmsghdr * nlh)559 static int inet6_netconf_get_devconf(struct sk_buff *in_skb,
560 struct nlmsghdr *nlh)
561 {
562 struct net *net = sock_net(in_skb->sk);
563 struct nlattr *tb[NETCONFA_MAX+1];
564 struct netconfmsg *ncm;
565 struct sk_buff *skb;
566 struct ipv6_devconf *devconf;
567 struct inet6_dev *in6_dev;
568 struct net_device *dev;
569 int ifindex;
570 int err;
571
572 err = nlmsg_parse(nlh, sizeof(*ncm), tb, NETCONFA_MAX,
573 devconf_ipv6_policy);
574 if (err < 0)
575 goto errout;
576
577 err = EINVAL;
578 if (!tb[NETCONFA_IFINDEX])
579 goto errout;
580
581 ifindex = nla_get_s32(tb[NETCONFA_IFINDEX]);
582 switch (ifindex) {
583 case NETCONFA_IFINDEX_ALL:
584 devconf = net->ipv6.devconf_all;
585 break;
586 case NETCONFA_IFINDEX_DEFAULT:
587 devconf = net->ipv6.devconf_dflt;
588 break;
589 default:
590 dev = __dev_get_by_index(net, ifindex);
591 if (dev == NULL)
592 goto errout;
593 in6_dev = __in6_dev_get(dev);
594 if (in6_dev == NULL)
595 goto errout;
596 devconf = &in6_dev->cnf;
597 break;
598 }
599
600 err = -ENOBUFS;
601 skb = nlmsg_new(inet6_netconf_msgsize_devconf(-1), GFP_ATOMIC);
602 if (skb == NULL)
603 goto errout;
604
605 err = inet6_netconf_fill_devconf(skb, ifindex, devconf,
606 NETLINK_CB(in_skb).portid,
607 nlh->nlmsg_seq, RTM_NEWNETCONF, 0,
608 -1);
609 if (err < 0) {
610 /* -EMSGSIZE implies BUG in inet6_netconf_msgsize_devconf() */
611 WARN_ON(err == -EMSGSIZE);
612 kfree_skb(skb);
613 goto errout;
614 }
615 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
616 errout:
617 return err;
618 }
619
inet6_netconf_dump_devconf(struct sk_buff * skb,struct netlink_callback * cb)620 static int inet6_netconf_dump_devconf(struct sk_buff *skb,
621 struct netlink_callback *cb)
622 {
623 struct net *net = sock_net(skb->sk);
624 int h, s_h;
625 int idx, s_idx;
626 struct net_device *dev;
627 struct inet6_dev *idev;
628 struct hlist_head *head;
629
630 s_h = cb->args[0];
631 s_idx = idx = cb->args[1];
632
633 for (h = s_h; h < NETDEV_HASHENTRIES; h++, s_idx = 0) {
634 idx = 0;
635 head = &net->dev_index_head[h];
636 rcu_read_lock();
637 cb->seq = atomic_read(&net->ipv6.dev_addr_genid) ^
638 net->dev_base_seq;
639 hlist_for_each_entry_rcu(dev, head, index_hlist) {
640 if (idx < s_idx)
641 goto cont;
642 idev = __in6_dev_get(dev);
643 if (!idev)
644 goto cont;
645
646 if (inet6_netconf_fill_devconf(skb, dev->ifindex,
647 &idev->cnf,
648 NETLINK_CB(cb->skb).portid,
649 cb->nlh->nlmsg_seq,
650 RTM_NEWNETCONF,
651 NLM_F_MULTI,
652 -1) <= 0) {
653 rcu_read_unlock();
654 goto done;
655 }
656 nl_dump_check_consistent(cb, nlmsg_hdr(skb));
657 cont:
658 idx++;
659 }
660 rcu_read_unlock();
661 }
662 if (h == NETDEV_HASHENTRIES) {
663 if (inet6_netconf_fill_devconf(skb, NETCONFA_IFINDEX_ALL,
664 net->ipv6.devconf_all,
665 NETLINK_CB(cb->skb).portid,
666 cb->nlh->nlmsg_seq,
667 RTM_NEWNETCONF, NLM_F_MULTI,
668 -1) <= 0)
669 goto done;
670 else
671 h++;
672 }
673 if (h == NETDEV_HASHENTRIES + 1) {
674 if (inet6_netconf_fill_devconf(skb, NETCONFA_IFINDEX_DEFAULT,
675 net->ipv6.devconf_dflt,
676 NETLINK_CB(cb->skb).portid,
677 cb->nlh->nlmsg_seq,
678 RTM_NEWNETCONF, NLM_F_MULTI,
679 -1) <= 0)
680 goto done;
681 else
682 h++;
683 }
684 done:
685 cb->args[0] = h;
686 cb->args[1] = idx;
687
688 return skb->len;
689 }
690
691 #ifdef CONFIG_SYSCTL
dev_forward_change(struct inet6_dev * idev)692 static void dev_forward_change(struct inet6_dev *idev)
693 {
694 struct net_device *dev;
695 struct inet6_ifaddr *ifa;
696
697 if (!idev)
698 return;
699 dev = idev->dev;
700 if (idev->cnf.forwarding)
701 dev_disable_lro(dev);
702 if (dev->flags & IFF_MULTICAST) {
703 if (idev->cnf.forwarding) {
704 ipv6_dev_mc_inc(dev, &in6addr_linklocal_allrouters);
705 ipv6_dev_mc_inc(dev, &in6addr_interfacelocal_allrouters);
706 ipv6_dev_mc_inc(dev, &in6addr_sitelocal_allrouters);
707 } else {
708 ipv6_dev_mc_dec(dev, &in6addr_linklocal_allrouters);
709 ipv6_dev_mc_dec(dev, &in6addr_interfacelocal_allrouters);
710 ipv6_dev_mc_dec(dev, &in6addr_sitelocal_allrouters);
711 }
712 }
713
714 list_for_each_entry(ifa, &idev->addr_list, if_list) {
715 if (ifa->flags&IFA_F_TENTATIVE)
716 continue;
717 if (idev->cnf.forwarding)
718 addrconf_join_anycast(ifa);
719 else
720 addrconf_leave_anycast(ifa);
721 }
722 inet6_netconf_notify_devconf(dev_net(dev), NETCONFA_FORWARDING,
723 dev->ifindex, &idev->cnf);
724 }
725
726
addrconf_forward_change(struct net * net,__s32 newf)727 static void addrconf_forward_change(struct net *net, __s32 newf)
728 {
729 struct net_device *dev;
730 struct inet6_dev *idev;
731
732 for_each_netdev(net, dev) {
733 idev = __in6_dev_get(dev);
734 if (idev) {
735 int changed = (!idev->cnf.forwarding) ^ (!newf);
736 idev->cnf.forwarding = newf;
737 if (changed)
738 dev_forward_change(idev);
739 }
740 }
741 }
742
addrconf_fixup_forwarding(struct ctl_table * table,int * p,int newf)743 static int addrconf_fixup_forwarding(struct ctl_table *table, int *p, int newf)
744 {
745 struct net *net;
746 int old;
747
748 if (!rtnl_trylock())
749 return restart_syscall();
750
751 net = (struct net *)table->extra2;
752 old = *p;
753 *p = newf;
754
755 if (p == &net->ipv6.devconf_dflt->forwarding) {
756 if ((!newf) ^ (!old))
757 inet6_netconf_notify_devconf(net, NETCONFA_FORWARDING,
758 NETCONFA_IFINDEX_DEFAULT,
759 net->ipv6.devconf_dflt);
760 rtnl_unlock();
761 return 0;
762 }
763
764 if (p == &net->ipv6.devconf_all->forwarding) {
765 net->ipv6.devconf_dflt->forwarding = newf;
766 addrconf_forward_change(net, newf);
767 if ((!newf) ^ (!old))
768 inet6_netconf_notify_devconf(net, NETCONFA_FORWARDING,
769 NETCONFA_IFINDEX_ALL,
770 net->ipv6.devconf_all);
771 } else if ((!newf) ^ (!old))
772 dev_forward_change((struct inet6_dev *)table->extra1);
773 rtnl_unlock();
774
775 if (newf)
776 rt6_purge_dflt_routers(net);
777 return 1;
778 }
779 #endif
780
781 /* Nobody refers to this ifaddr, destroy it */
inet6_ifa_finish_destroy(struct inet6_ifaddr * ifp)782 void inet6_ifa_finish_destroy(struct inet6_ifaddr *ifp)
783 {
784 WARN_ON(!hlist_unhashed(&ifp->addr_lst));
785
786 #ifdef NET_REFCNT_DEBUG
787 pr_debug("%s\n", __func__);
788 #endif
789
790 in6_dev_put(ifp->idev);
791
792 if (cancel_delayed_work(&ifp->dad_work))
793 pr_notice("delayed DAD work was pending while freeing ifa=%p\n",
794 ifp);
795
796 if (ifp->state != INET6_IFADDR_STATE_DEAD) {
797 pr_warn("Freeing alive inet6 address %p\n", ifp);
798 return;
799 }
800 ip6_rt_put(ifp->rt);
801
802 kfree_rcu(ifp, rcu);
803 }
804
805 static void
ipv6_link_dev_addr(struct inet6_dev * idev,struct inet6_ifaddr * ifp)806 ipv6_link_dev_addr(struct inet6_dev *idev, struct inet6_ifaddr *ifp)
807 {
808 struct list_head *p;
809 int ifp_scope = ipv6_addr_src_scope(&ifp->addr);
810
811 /*
812 * Each device address list is sorted in order of scope -
813 * global before linklocal.
814 */
815 list_for_each(p, &idev->addr_list) {
816 struct inet6_ifaddr *ifa
817 = list_entry(p, struct inet6_ifaddr, if_list);
818 if (ifp_scope >= ipv6_addr_src_scope(&ifa->addr))
819 break;
820 }
821
822 list_add_tail(&ifp->if_list, p);
823 }
824
inet6_addr_hash(const struct in6_addr * addr)825 static u32 inet6_addr_hash(const struct in6_addr *addr)
826 {
827 return hash_32(ipv6_addr_hash(addr), IN6_ADDR_HSIZE_SHIFT);
828 }
829
830 /* On success it returns ifp with increased reference count */
831
832 static struct inet6_ifaddr *
ipv6_add_addr(struct inet6_dev * idev,const struct in6_addr * addr,const struct in6_addr * peer_addr,int pfxlen,int scope,u32 flags,u32 valid_lft,u32 prefered_lft)833 ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr,
834 const struct in6_addr *peer_addr, int pfxlen,
835 int scope, u32 flags, u32 valid_lft, u32 prefered_lft)
836 {
837 struct inet6_ifaddr *ifa = NULL;
838 struct rt6_info *rt;
839 unsigned int hash;
840 int err = 0;
841 int addr_type = ipv6_addr_type(addr);
842
843 if (addr_type == IPV6_ADDR_ANY ||
844 addr_type & IPV6_ADDR_MULTICAST ||
845 (!(idev->dev->flags & IFF_LOOPBACK) &&
846 addr_type & IPV6_ADDR_LOOPBACK))
847 return ERR_PTR(-EADDRNOTAVAIL);
848
849 rcu_read_lock_bh();
850 if (idev->dead) {
851 err = -ENODEV; /*XXX*/
852 goto out2;
853 }
854
855 if (idev->cnf.disable_ipv6) {
856 err = -EACCES;
857 goto out2;
858 }
859
860 spin_lock(&addrconf_hash_lock);
861
862 /* Ignore adding duplicate addresses on an interface */
863 if (ipv6_chk_same_addr(dev_net(idev->dev), addr, idev->dev)) {
864 ADBG("ipv6_add_addr: already assigned\n");
865 err = -EEXIST;
866 goto out;
867 }
868
869 ifa = kzalloc(sizeof(struct inet6_ifaddr), GFP_ATOMIC);
870
871 if (ifa == NULL) {
872 ADBG("ipv6_add_addr: malloc failed\n");
873 err = -ENOBUFS;
874 goto out;
875 }
876
877 rt = addrconf_dst_alloc(idev, addr, false);
878 if (IS_ERR(rt)) {
879 err = PTR_ERR(rt);
880 goto out;
881 }
882
883 neigh_parms_data_state_setall(idev->nd_parms);
884
885 ifa->addr = *addr;
886 if (peer_addr)
887 ifa->peer_addr = *peer_addr;
888
889 spin_lock_init(&ifa->lock);
890 spin_lock_init(&ifa->state_lock);
891 INIT_DELAYED_WORK(&ifa->dad_work, addrconf_dad_work);
892 INIT_HLIST_NODE(&ifa->addr_lst);
893 ifa->scope = scope;
894 ifa->prefix_len = pfxlen;
895 ifa->flags = flags | IFA_F_TENTATIVE;
896 ifa->valid_lft = valid_lft;
897 ifa->prefered_lft = prefered_lft;
898 ifa->cstamp = ifa->tstamp = jiffies;
899 ifa->tokenized = false;
900
901 ifa->rt = rt;
902
903 ifa->idev = idev;
904 in6_dev_hold(idev);
905 /* For caller */
906 in6_ifa_hold(ifa);
907
908 /* Add to big hash table */
909 hash = inet6_addr_hash(addr);
910
911 hlist_add_head_rcu(&ifa->addr_lst, &inet6_addr_lst[hash]);
912 spin_unlock(&addrconf_hash_lock);
913
914 write_lock(&idev->lock);
915 /* Add to inet6_dev unicast addr list. */
916 ipv6_link_dev_addr(idev, ifa);
917
918 if (ifa->flags&IFA_F_TEMPORARY) {
919 list_add(&ifa->tmp_list, &idev->tempaddr_list);
920 in6_ifa_hold(ifa);
921 }
922
923 in6_ifa_hold(ifa);
924 write_unlock(&idev->lock);
925 out2:
926 rcu_read_unlock_bh();
927
928 if (likely(err == 0))
929 inet6addr_notifier_call_chain(NETDEV_UP, ifa);
930 else {
931 kfree(ifa);
932 ifa = ERR_PTR(err);
933 }
934
935 return ifa;
936 out:
937 spin_unlock(&addrconf_hash_lock);
938 goto out2;
939 }
940
941 enum cleanup_prefix_rt_t {
942 CLEANUP_PREFIX_RT_NOP, /* no cleanup action for prefix route */
943 CLEANUP_PREFIX_RT_DEL, /* delete the prefix route */
944 CLEANUP_PREFIX_RT_EXPIRE, /* update the lifetime of the prefix route */
945 };
946
947 /*
948 * Check, whether the prefix for ifp would still need a prefix route
949 * after deleting ifp. The function returns one of the CLEANUP_PREFIX_RT_*
950 * constants.
951 *
952 * 1) we don't purge prefix if address was not permanent.
953 * prefix is managed by its own lifetime.
954 * 2) we also don't purge, if the address was IFA_F_NOPREFIXROUTE.
955 * 3) if there are no addresses, delete prefix.
956 * 4) if there are still other permanent address(es),
957 * corresponding prefix is still permanent.
958 * 5) if there are still other addresses with IFA_F_NOPREFIXROUTE,
959 * don't purge the prefix, assume user space is managing it.
960 * 6) otherwise, update prefix lifetime to the
961 * longest valid lifetime among the corresponding
962 * addresses on the device.
963 * Note: subsequent RA will update lifetime.
964 **/
965 static enum cleanup_prefix_rt_t
check_cleanup_prefix_route(struct inet6_ifaddr * ifp,unsigned long * expires)966 check_cleanup_prefix_route(struct inet6_ifaddr *ifp, unsigned long *expires)
967 {
968 struct inet6_ifaddr *ifa;
969 struct inet6_dev *idev = ifp->idev;
970 unsigned long lifetime;
971 enum cleanup_prefix_rt_t action = CLEANUP_PREFIX_RT_DEL;
972
973 *expires = jiffies;
974
975 list_for_each_entry(ifa, &idev->addr_list, if_list) {
976 if (ifa == ifp)
977 continue;
978 if (!ipv6_prefix_equal(&ifa->addr, &ifp->addr,
979 ifp->prefix_len))
980 continue;
981 if (ifa->flags & (IFA_F_PERMANENT | IFA_F_NOPREFIXROUTE))
982 return CLEANUP_PREFIX_RT_NOP;
983
984 action = CLEANUP_PREFIX_RT_EXPIRE;
985
986 spin_lock(&ifa->lock);
987
988 lifetime = addrconf_timeout_fixup(ifa->valid_lft, HZ);
989 /*
990 * Note: Because this address is
991 * not permanent, lifetime <
992 * LONG_MAX / HZ here.
993 */
994 if (time_before(*expires, ifa->tstamp + lifetime * HZ))
995 *expires = ifa->tstamp + lifetime * HZ;
996 spin_unlock(&ifa->lock);
997 }
998
999 return action;
1000 }
1001
1002 static void
cleanup_prefix_route(struct inet6_ifaddr * ifp,unsigned long expires,bool del_rt)1003 cleanup_prefix_route(struct inet6_ifaddr *ifp, unsigned long expires, bool del_rt)
1004 {
1005 struct rt6_info *rt;
1006
1007 rt = addrconf_get_prefix_route(&ifp->addr,
1008 ifp->prefix_len,
1009 ifp->idev->dev,
1010 0, RTF_GATEWAY | RTF_DEFAULT);
1011 if (rt) {
1012 if (del_rt)
1013 ip6_del_rt(rt);
1014 else {
1015 if (!(rt->rt6i_flags & RTF_EXPIRES))
1016 rt6_set_expires(rt, expires);
1017 ip6_rt_put(rt);
1018 }
1019 }
1020 }
1021
1022
1023 /* This function wants to get referenced ifp and releases it before return */
1024
ipv6_del_addr(struct inet6_ifaddr * ifp)1025 static void ipv6_del_addr(struct inet6_ifaddr *ifp)
1026 {
1027 int state;
1028 enum cleanup_prefix_rt_t action = CLEANUP_PREFIX_RT_NOP;
1029 unsigned long expires;
1030
1031 ASSERT_RTNL();
1032
1033 spin_lock_bh(&ifp->state_lock);
1034 state = ifp->state;
1035 ifp->state = INET6_IFADDR_STATE_DEAD;
1036 spin_unlock_bh(&ifp->state_lock);
1037
1038 if (state == INET6_IFADDR_STATE_DEAD)
1039 goto out;
1040
1041 spin_lock_bh(&addrconf_hash_lock);
1042 hlist_del_init_rcu(&ifp->addr_lst);
1043 spin_unlock_bh(&addrconf_hash_lock);
1044
1045 write_lock_bh(&ifp->idev->lock);
1046
1047 if (ifp->flags&IFA_F_TEMPORARY) {
1048 list_del(&ifp->tmp_list);
1049 if (ifp->ifpub) {
1050 in6_ifa_put(ifp->ifpub);
1051 ifp->ifpub = NULL;
1052 }
1053 __in6_ifa_put(ifp);
1054 }
1055
1056 if (ifp->flags & IFA_F_PERMANENT && !(ifp->flags & IFA_F_NOPREFIXROUTE))
1057 action = check_cleanup_prefix_route(ifp, &expires);
1058
1059 list_del_init(&ifp->if_list);
1060 __in6_ifa_put(ifp);
1061
1062 write_unlock_bh(&ifp->idev->lock);
1063
1064 addrconf_del_dad_work(ifp);
1065
1066 ipv6_ifa_notify(RTM_DELADDR, ifp);
1067
1068 inet6addr_notifier_call_chain(NETDEV_DOWN, ifp);
1069
1070 if (action != CLEANUP_PREFIX_RT_NOP) {
1071 cleanup_prefix_route(ifp, expires,
1072 action == CLEANUP_PREFIX_RT_DEL);
1073 }
1074
1075 /* clean up prefsrc entries */
1076 rt6_remove_prefsrc(ifp);
1077 out:
1078 in6_ifa_put(ifp);
1079 }
1080
ipv6_create_tempaddr(struct inet6_ifaddr * ifp,struct inet6_ifaddr * ift)1081 static int ipv6_create_tempaddr(struct inet6_ifaddr *ifp, struct inet6_ifaddr *ift)
1082 {
1083 struct inet6_dev *idev = ifp->idev;
1084 struct in6_addr addr, *tmpaddr;
1085 unsigned long tmp_prefered_lft, tmp_valid_lft, tmp_tstamp, age;
1086 unsigned long regen_advance;
1087 int tmp_plen;
1088 int ret = 0;
1089 u32 addr_flags;
1090 unsigned long now = jiffies;
1091
1092 write_lock_bh(&idev->lock);
1093 if (ift) {
1094 spin_lock_bh(&ift->lock);
1095 memcpy(&addr.s6_addr[8], &ift->addr.s6_addr[8], 8);
1096 spin_unlock_bh(&ift->lock);
1097 tmpaddr = &addr;
1098 } else {
1099 tmpaddr = NULL;
1100 }
1101 retry:
1102 in6_dev_hold(idev);
1103 if (idev->cnf.use_tempaddr <= 0) {
1104 write_unlock_bh(&idev->lock);
1105 pr_info("%s: use_tempaddr is disabled\n", __func__);
1106 in6_dev_put(idev);
1107 ret = -1;
1108 goto out;
1109 }
1110 spin_lock_bh(&ifp->lock);
1111 if (ifp->regen_count++ >= idev->cnf.regen_max_retry) {
1112 idev->cnf.use_tempaddr = -1; /*XXX*/
1113 spin_unlock_bh(&ifp->lock);
1114 write_unlock_bh(&idev->lock);
1115 pr_warn("%s: regeneration time exceeded - disabled temporary address support\n",
1116 __func__);
1117 in6_dev_put(idev);
1118 ret = -1;
1119 goto out;
1120 }
1121 in6_ifa_hold(ifp);
1122 memcpy(addr.s6_addr, ifp->addr.s6_addr, 8);
1123 __ipv6_try_regen_rndid(idev, tmpaddr);
1124 memcpy(&addr.s6_addr[8], idev->rndid, 8);
1125 age = (now - ifp->tstamp) / HZ;
1126 tmp_valid_lft = min_t(__u32,
1127 ifp->valid_lft,
1128 idev->cnf.temp_valid_lft + age);
1129 tmp_prefered_lft = min_t(__u32,
1130 ifp->prefered_lft,
1131 idev->cnf.temp_prefered_lft + age -
1132 idev->cnf.max_desync_factor);
1133 tmp_plen = ifp->prefix_len;
1134 tmp_tstamp = ifp->tstamp;
1135 spin_unlock_bh(&ifp->lock);
1136
1137 regen_advance = idev->cnf.regen_max_retry *
1138 idev->cnf.dad_transmits *
1139 NEIGH_VAR(idev->nd_parms, RETRANS_TIME) / HZ;
1140 write_unlock_bh(&idev->lock);
1141
1142 /* A temporary address is created only if this calculated Preferred
1143 * Lifetime is greater than REGEN_ADVANCE time units. In particular,
1144 * an implementation must not create a temporary address with a zero
1145 * Preferred Lifetime.
1146 * Use age calculation as in addrconf_verify to avoid unnecessary
1147 * temporary addresses being generated.
1148 */
1149 age = (now - tmp_tstamp + ADDRCONF_TIMER_FUZZ_MINUS) / HZ;
1150 if (tmp_prefered_lft <= regen_advance + age) {
1151 in6_ifa_put(ifp);
1152 in6_dev_put(idev);
1153 ret = -1;
1154 goto out;
1155 }
1156
1157 addr_flags = IFA_F_TEMPORARY;
1158 /* set in addrconf_prefix_rcv() */
1159 if (ifp->flags & IFA_F_OPTIMISTIC)
1160 addr_flags |= IFA_F_OPTIMISTIC;
1161
1162 ift = ipv6_add_addr(idev, &addr, NULL, tmp_plen,
1163 ipv6_addr_scope(&addr), addr_flags,
1164 tmp_valid_lft, tmp_prefered_lft);
1165 if (IS_ERR(ift)) {
1166 in6_ifa_put(ifp);
1167 in6_dev_put(idev);
1168 pr_info("%s: retry temporary address regeneration\n", __func__);
1169 tmpaddr = &addr;
1170 write_lock_bh(&idev->lock);
1171 goto retry;
1172 }
1173
1174 spin_lock_bh(&ift->lock);
1175 ift->ifpub = ifp;
1176 ift->cstamp = now;
1177 ift->tstamp = tmp_tstamp;
1178 spin_unlock_bh(&ift->lock);
1179
1180 addrconf_dad_start(ift);
1181 in6_ifa_put(ift);
1182 in6_dev_put(idev);
1183 out:
1184 return ret;
1185 }
1186
1187 /*
1188 * Choose an appropriate source address (RFC3484)
1189 */
1190 enum {
1191 IPV6_SADDR_RULE_INIT = 0,
1192 IPV6_SADDR_RULE_LOCAL,
1193 IPV6_SADDR_RULE_SCOPE,
1194 IPV6_SADDR_RULE_PREFERRED,
1195 #ifdef CONFIG_IPV6_MIP6
1196 IPV6_SADDR_RULE_HOA,
1197 #endif
1198 IPV6_SADDR_RULE_OIF,
1199 IPV6_SADDR_RULE_LABEL,
1200 IPV6_SADDR_RULE_PRIVACY,
1201 IPV6_SADDR_RULE_ORCHID,
1202 IPV6_SADDR_RULE_PREFIX,
1203 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
1204 IPV6_SADDR_RULE_NOT_OPTIMISTIC,
1205 #endif
1206 IPV6_SADDR_RULE_MAX
1207 };
1208
1209 struct ipv6_saddr_score {
1210 int rule;
1211 int addr_type;
1212 struct inet6_ifaddr *ifa;
1213 DECLARE_BITMAP(scorebits, IPV6_SADDR_RULE_MAX);
1214 int scopedist;
1215 int matchlen;
1216 };
1217
1218 struct ipv6_saddr_dst {
1219 const struct in6_addr *addr;
1220 int ifindex;
1221 int scope;
1222 int label;
1223 unsigned int prefs;
1224 };
1225
ipv6_saddr_preferred(int type)1226 static inline int ipv6_saddr_preferred(int type)
1227 {
1228 if (type & (IPV6_ADDR_MAPPED|IPV6_ADDR_COMPATv4|IPV6_ADDR_LOOPBACK))
1229 return 1;
1230 return 0;
1231 }
1232
ipv6_use_optimistic_addr(struct inet6_dev * idev)1233 static inline bool ipv6_use_optimistic_addr(struct inet6_dev *idev)
1234 {
1235 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
1236 return idev && idev->cnf.optimistic_dad && idev->cnf.use_optimistic;
1237 #else
1238 return false;
1239 #endif
1240 }
1241
ipv6_get_saddr_eval(struct net * net,struct ipv6_saddr_score * score,struct ipv6_saddr_dst * dst,int i)1242 static int ipv6_get_saddr_eval(struct net *net,
1243 struct ipv6_saddr_score *score,
1244 struct ipv6_saddr_dst *dst,
1245 int i)
1246 {
1247 int ret;
1248
1249 if (i <= score->rule) {
1250 switch (i) {
1251 case IPV6_SADDR_RULE_SCOPE:
1252 ret = score->scopedist;
1253 break;
1254 case IPV6_SADDR_RULE_PREFIX:
1255 ret = score->matchlen;
1256 break;
1257 default:
1258 ret = !!test_bit(i, score->scorebits);
1259 }
1260 goto out;
1261 }
1262
1263 switch (i) {
1264 case IPV6_SADDR_RULE_INIT:
1265 /* Rule 0: remember if hiscore is not ready yet */
1266 ret = !!score->ifa;
1267 break;
1268 case IPV6_SADDR_RULE_LOCAL:
1269 /* Rule 1: Prefer same address */
1270 ret = ipv6_addr_equal(&score->ifa->addr, dst->addr);
1271 break;
1272 case IPV6_SADDR_RULE_SCOPE:
1273 /* Rule 2: Prefer appropriate scope
1274 *
1275 * ret
1276 * ^
1277 * -1 | d 15
1278 * ---+--+-+---> scope
1279 * |
1280 * | d is scope of the destination.
1281 * B-d | \
1282 * | \ <- smaller scope is better if
1283 * B-15 | \ if scope is enough for destination.
1284 * | ret = B - scope (-1 <= scope >= d <= 15).
1285 * d-C-1 | /
1286 * |/ <- greater is better
1287 * -C / if scope is not enough for destination.
1288 * /| ret = scope - C (-1 <= d < scope <= 15).
1289 *
1290 * d - C - 1 < B -15 (for all -1 <= d <= 15).
1291 * C > d + 14 - B >= 15 + 14 - B = 29 - B.
1292 * Assume B = 0 and we get C > 29.
1293 */
1294 ret = __ipv6_addr_src_scope(score->addr_type);
1295 if (ret >= dst->scope)
1296 ret = -ret;
1297 else
1298 ret -= 128; /* 30 is enough */
1299 score->scopedist = ret;
1300 break;
1301 case IPV6_SADDR_RULE_PREFERRED:
1302 {
1303 /* Rule 3: Avoid deprecated and optimistic addresses */
1304 u8 avoid = IFA_F_DEPRECATED;
1305
1306 if (!ipv6_use_optimistic_addr(score->ifa->idev))
1307 avoid |= IFA_F_OPTIMISTIC;
1308 ret = ipv6_saddr_preferred(score->addr_type) ||
1309 !(score->ifa->flags & avoid);
1310 break;
1311 }
1312 #ifdef CONFIG_IPV6_MIP6
1313 case IPV6_SADDR_RULE_HOA:
1314 {
1315 /* Rule 4: Prefer home address */
1316 int prefhome = !(dst->prefs & IPV6_PREFER_SRC_COA);
1317 ret = !(score->ifa->flags & IFA_F_HOMEADDRESS) ^ prefhome;
1318 break;
1319 }
1320 #endif
1321 case IPV6_SADDR_RULE_OIF:
1322 /* Rule 5: Prefer outgoing interface */
1323 ret = (!dst->ifindex ||
1324 dst->ifindex == score->ifa->idev->dev->ifindex);
1325 break;
1326 case IPV6_SADDR_RULE_LABEL:
1327 /* Rule 6: Prefer matching label */
1328 ret = ipv6_addr_label(net,
1329 &score->ifa->addr, score->addr_type,
1330 score->ifa->idev->dev->ifindex) == dst->label;
1331 break;
1332 case IPV6_SADDR_RULE_PRIVACY:
1333 {
1334 /* Rule 7: Prefer public address
1335 * Note: prefer temporary address if use_tempaddr >= 2
1336 */
1337 int preftmp = dst->prefs & (IPV6_PREFER_SRC_PUBLIC|IPV6_PREFER_SRC_TMP) ?
1338 !!(dst->prefs & IPV6_PREFER_SRC_TMP) :
1339 score->ifa->idev->cnf.use_tempaddr >= 2;
1340 ret = (!(score->ifa->flags & IFA_F_TEMPORARY)) ^ preftmp;
1341 break;
1342 }
1343 case IPV6_SADDR_RULE_ORCHID:
1344 /* Rule 8-: Prefer ORCHID vs ORCHID or
1345 * non-ORCHID vs non-ORCHID
1346 */
1347 ret = !(ipv6_addr_orchid(&score->ifa->addr) ^
1348 ipv6_addr_orchid(dst->addr));
1349 break;
1350 case IPV6_SADDR_RULE_PREFIX:
1351 /* Rule 8: Use longest matching prefix */
1352 ret = ipv6_addr_diff(&score->ifa->addr, dst->addr);
1353 if (ret > score->ifa->prefix_len)
1354 ret = score->ifa->prefix_len;
1355 score->matchlen = ret;
1356 break;
1357 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
1358 case IPV6_SADDR_RULE_NOT_OPTIMISTIC:
1359 /* Optimistic addresses still have lower precedence than other
1360 * preferred addresses.
1361 */
1362 ret = !(score->ifa->flags & IFA_F_OPTIMISTIC);
1363 break;
1364 #endif
1365 default:
1366 ret = 0;
1367 }
1368
1369 if (ret)
1370 __set_bit(i, score->scorebits);
1371 score->rule = i;
1372 out:
1373 return ret;
1374 }
1375
__ipv6_dev_get_saddr(struct net * net,struct ipv6_saddr_dst * dst,struct inet6_dev * idev,struct ipv6_saddr_score * scores,int hiscore_idx)1376 static int __ipv6_dev_get_saddr(struct net *net,
1377 struct ipv6_saddr_dst *dst,
1378 struct inet6_dev *idev,
1379 struct ipv6_saddr_score *scores,
1380 int hiscore_idx)
1381 {
1382 struct ipv6_saddr_score *score = &scores[1 - hiscore_idx], *hiscore = &scores[hiscore_idx];
1383
1384 read_lock_bh(&idev->lock);
1385 list_for_each_entry(score->ifa, &idev->addr_list, if_list) {
1386 int i;
1387
1388 /*
1389 * - Tentative Address (RFC2462 section 5.4)
1390 * - A tentative address is not considered
1391 * "assigned to an interface" in the traditional
1392 * sense, unless it is also flagged as optimistic.
1393 * - Candidate Source Address (section 4)
1394 * - In any case, anycast addresses, multicast
1395 * addresses, and the unspecified address MUST
1396 * NOT be included in a candidate set.
1397 */
1398 if ((score->ifa->flags & IFA_F_TENTATIVE) &&
1399 (!(score->ifa->flags & IFA_F_OPTIMISTIC)))
1400 continue;
1401
1402 score->addr_type = __ipv6_addr_type(&score->ifa->addr);
1403
1404 if (unlikely(score->addr_type == IPV6_ADDR_ANY ||
1405 score->addr_type & IPV6_ADDR_MULTICAST)) {
1406 net_dbg_ratelimited("ADDRCONF: unspecified / multicast address assigned as unicast address on %s",
1407 idev->dev->name);
1408 continue;
1409 }
1410
1411 score->rule = -1;
1412 bitmap_zero(score->scorebits, IPV6_SADDR_RULE_MAX);
1413
1414 for (i = 0; i < IPV6_SADDR_RULE_MAX; i++) {
1415 int minihiscore, miniscore;
1416
1417 minihiscore = ipv6_get_saddr_eval(net, hiscore, dst, i);
1418 miniscore = ipv6_get_saddr_eval(net, score, dst, i);
1419
1420 if (minihiscore > miniscore) {
1421 if (i == IPV6_SADDR_RULE_SCOPE &&
1422 score->scopedist > 0) {
1423 /*
1424 * special case:
1425 * each remaining entry
1426 * has too small (not enough)
1427 * scope, because ifa entries
1428 * are sorted by their scope
1429 * values.
1430 */
1431 goto out;
1432 }
1433 break;
1434 } else if (minihiscore < miniscore) {
1435 if (hiscore->ifa)
1436 in6_ifa_put(hiscore->ifa);
1437
1438 in6_ifa_hold(score->ifa);
1439
1440 swap(hiscore, score);
1441 hiscore_idx = 1 - hiscore_idx;
1442
1443 /* restore our iterator */
1444 score->ifa = hiscore->ifa;
1445
1446 break;
1447 }
1448 }
1449 }
1450 out:
1451 read_unlock_bh(&idev->lock);
1452 return hiscore_idx;
1453 }
1454
ipv6_dev_get_saddr(struct net * net,const struct net_device * dst_dev,const struct in6_addr * daddr,unsigned int prefs,struct in6_addr * saddr)1455 int ipv6_dev_get_saddr(struct net *net, const struct net_device *dst_dev,
1456 const struct in6_addr *daddr, unsigned int prefs,
1457 struct in6_addr *saddr)
1458 {
1459 struct ipv6_saddr_score scores[2], *hiscore;
1460 struct ipv6_saddr_dst dst;
1461 struct inet6_dev *idev;
1462 struct net_device *dev;
1463 int dst_type;
1464 bool use_oif_addr = false;
1465 int hiscore_idx = 0;
1466
1467 dst_type = __ipv6_addr_type(daddr);
1468 dst.addr = daddr;
1469 dst.ifindex = dst_dev ? dst_dev->ifindex : 0;
1470 dst.scope = __ipv6_addr_src_scope(dst_type);
1471 dst.label = ipv6_addr_label(net, daddr, dst_type, dst.ifindex);
1472 dst.prefs = prefs;
1473
1474 scores[hiscore_idx].rule = -1;
1475 scores[hiscore_idx].ifa = NULL;
1476
1477 rcu_read_lock();
1478
1479 /* Candidate Source Address (section 4)
1480 * - multicast and link-local destination address,
1481 * the set of candidate source address MUST only
1482 * include addresses assigned to interfaces
1483 * belonging to the same link as the outgoing
1484 * interface.
1485 * (- For site-local destination addresses, the
1486 * set of candidate source addresses MUST only
1487 * include addresses assigned to interfaces
1488 * belonging to the same site as the outgoing
1489 * interface.)
1490 * - "It is RECOMMENDED that the candidate source addresses
1491 * be the set of unicast addresses assigned to the
1492 * interface that will be used to send to the destination
1493 * (the 'outgoing' interface)." (RFC 6724)
1494 */
1495 if (dst_dev) {
1496 idev = __in6_dev_get(dst_dev);
1497 if ((dst_type & IPV6_ADDR_MULTICAST) ||
1498 dst.scope <= IPV6_ADDR_SCOPE_LINKLOCAL ||
1499 (idev && idev->cnf.use_oif_addrs_only)) {
1500 use_oif_addr = true;
1501 }
1502 }
1503
1504 if (use_oif_addr) {
1505 if (idev)
1506 hiscore_idx = __ipv6_dev_get_saddr(net, &dst, idev, scores, hiscore_idx);
1507 } else {
1508 for_each_netdev_rcu(net, dev) {
1509 idev = __in6_dev_get(dev);
1510 if (!idev)
1511 continue;
1512 hiscore_idx = __ipv6_dev_get_saddr(net, &dst, idev, scores, hiscore_idx);
1513 }
1514 }
1515 rcu_read_unlock();
1516
1517 hiscore = &scores[hiscore_idx];
1518 if (!hiscore->ifa)
1519 return -EADDRNOTAVAIL;
1520
1521 *saddr = hiscore->ifa->addr;
1522 in6_ifa_put(hiscore->ifa);
1523 return 0;
1524 }
1525 EXPORT_SYMBOL(ipv6_dev_get_saddr);
1526
__ipv6_get_lladdr(struct inet6_dev * idev,struct in6_addr * addr,u32 banned_flags)1527 int __ipv6_get_lladdr(struct inet6_dev *idev, struct in6_addr *addr,
1528 u32 banned_flags)
1529 {
1530 struct inet6_ifaddr *ifp;
1531 int err = -EADDRNOTAVAIL;
1532
1533 list_for_each_entry_reverse(ifp, &idev->addr_list, if_list) {
1534 if (ifp->scope > IFA_LINK)
1535 break;
1536 if (ifp->scope == IFA_LINK &&
1537 !(ifp->flags & banned_flags)) {
1538 *addr = ifp->addr;
1539 err = 0;
1540 break;
1541 }
1542 }
1543 return err;
1544 }
1545
ipv6_get_lladdr(struct net_device * dev,struct in6_addr * addr,u32 banned_flags)1546 int ipv6_get_lladdr(struct net_device *dev, struct in6_addr *addr,
1547 u32 banned_flags)
1548 {
1549 struct inet6_dev *idev;
1550 int err = -EADDRNOTAVAIL;
1551
1552 rcu_read_lock();
1553 idev = __in6_dev_get(dev);
1554 if (idev) {
1555 read_lock_bh(&idev->lock);
1556 err = __ipv6_get_lladdr(idev, addr, banned_flags);
1557 read_unlock_bh(&idev->lock);
1558 }
1559 rcu_read_unlock();
1560 return err;
1561 }
1562
ipv6_count_addresses(struct inet6_dev * idev)1563 static int ipv6_count_addresses(struct inet6_dev *idev)
1564 {
1565 int cnt = 0;
1566 struct inet6_ifaddr *ifp;
1567
1568 read_lock_bh(&idev->lock);
1569 list_for_each_entry(ifp, &idev->addr_list, if_list)
1570 cnt++;
1571 read_unlock_bh(&idev->lock);
1572 return cnt;
1573 }
1574
ipv6_chk_addr(struct net * net,const struct in6_addr * addr,const struct net_device * dev,int strict)1575 int ipv6_chk_addr(struct net *net, const struct in6_addr *addr,
1576 const struct net_device *dev, int strict)
1577 {
1578 return ipv6_chk_addr_and_flags(net, addr, dev, strict, IFA_F_TENTATIVE);
1579 }
1580 EXPORT_SYMBOL(ipv6_chk_addr);
1581
ipv6_chk_addr_and_flags(struct net * net,const struct in6_addr * addr,const struct net_device * dev,int strict,u32 banned_flags)1582 int ipv6_chk_addr_and_flags(struct net *net, const struct in6_addr *addr,
1583 const struct net_device *dev, int strict,
1584 u32 banned_flags)
1585 {
1586 struct inet6_ifaddr *ifp;
1587 unsigned int hash = inet6_addr_hash(addr);
1588 u32 ifp_flags;
1589
1590 rcu_read_lock_bh();
1591 hlist_for_each_entry_rcu(ifp, &inet6_addr_lst[hash], addr_lst) {
1592 if (!net_eq(dev_net(ifp->idev->dev), net))
1593 continue;
1594 /* Decouple optimistic from tentative for evaluation here.
1595 * Ban optimistic addresses explicitly, when required.
1596 */
1597 ifp_flags = (ifp->flags&IFA_F_OPTIMISTIC)
1598 ? (ifp->flags&~IFA_F_TENTATIVE)
1599 : ifp->flags;
1600 if (ipv6_addr_equal(&ifp->addr, addr) &&
1601 !(ifp_flags&banned_flags) &&
1602 (dev == NULL || ifp->idev->dev == dev ||
1603 !(ifp->scope&(IFA_LINK|IFA_HOST) || strict))) {
1604 rcu_read_unlock_bh();
1605 return 1;
1606 }
1607 }
1608
1609 rcu_read_unlock_bh();
1610 return 0;
1611 }
1612 EXPORT_SYMBOL(ipv6_chk_addr_and_flags);
1613
ipv6_chk_same_addr(struct net * net,const struct in6_addr * addr,struct net_device * dev)1614 static bool ipv6_chk_same_addr(struct net *net, const struct in6_addr *addr,
1615 struct net_device *dev)
1616 {
1617 unsigned int hash = inet6_addr_hash(addr);
1618 struct inet6_ifaddr *ifp;
1619
1620 hlist_for_each_entry(ifp, &inet6_addr_lst[hash], addr_lst) {
1621 if (!net_eq(dev_net(ifp->idev->dev), net))
1622 continue;
1623 if (ipv6_addr_equal(&ifp->addr, addr)) {
1624 if (dev == NULL || ifp->idev->dev == dev)
1625 return true;
1626 }
1627 }
1628 return false;
1629 }
1630
1631 /* Compares an address/prefix_len with addresses on device @dev.
1632 * If one is found it returns true.
1633 */
ipv6_chk_custom_prefix(const struct in6_addr * addr,const unsigned int prefix_len,struct net_device * dev)1634 bool ipv6_chk_custom_prefix(const struct in6_addr *addr,
1635 const unsigned int prefix_len, struct net_device *dev)
1636 {
1637 struct inet6_dev *idev;
1638 struct inet6_ifaddr *ifa;
1639 bool ret = false;
1640
1641 rcu_read_lock();
1642 idev = __in6_dev_get(dev);
1643 if (idev) {
1644 read_lock_bh(&idev->lock);
1645 list_for_each_entry(ifa, &idev->addr_list, if_list) {
1646 ret = ipv6_prefix_equal(addr, &ifa->addr, prefix_len);
1647 if (ret)
1648 break;
1649 }
1650 read_unlock_bh(&idev->lock);
1651 }
1652 rcu_read_unlock();
1653
1654 return ret;
1655 }
1656 EXPORT_SYMBOL(ipv6_chk_custom_prefix);
1657
ipv6_chk_prefix(const struct in6_addr * addr,struct net_device * dev)1658 int ipv6_chk_prefix(const struct in6_addr *addr, struct net_device *dev)
1659 {
1660 struct inet6_dev *idev;
1661 struct inet6_ifaddr *ifa;
1662 int onlink;
1663
1664 onlink = 0;
1665 rcu_read_lock();
1666 idev = __in6_dev_get(dev);
1667 if (idev) {
1668 read_lock_bh(&idev->lock);
1669 list_for_each_entry(ifa, &idev->addr_list, if_list) {
1670 onlink = ipv6_prefix_equal(addr, &ifa->addr,
1671 ifa->prefix_len);
1672 if (onlink)
1673 break;
1674 }
1675 read_unlock_bh(&idev->lock);
1676 }
1677 rcu_read_unlock();
1678 return onlink;
1679 }
1680 EXPORT_SYMBOL(ipv6_chk_prefix);
1681
ipv6_get_ifaddr(struct net * net,const struct in6_addr * addr,struct net_device * dev,int strict)1682 struct inet6_ifaddr *ipv6_get_ifaddr(struct net *net, const struct in6_addr *addr,
1683 struct net_device *dev, int strict)
1684 {
1685 struct inet6_ifaddr *ifp, *result = NULL;
1686 unsigned int hash = inet6_addr_hash(addr);
1687
1688 rcu_read_lock_bh();
1689 hlist_for_each_entry_rcu_bh(ifp, &inet6_addr_lst[hash], addr_lst) {
1690 if (!net_eq(dev_net(ifp->idev->dev), net))
1691 continue;
1692 if (ipv6_addr_equal(&ifp->addr, addr)) {
1693 if (dev == NULL || ifp->idev->dev == dev ||
1694 !(ifp->scope&(IFA_LINK|IFA_HOST) || strict)) {
1695 result = ifp;
1696 in6_ifa_hold(ifp);
1697 break;
1698 }
1699 }
1700 }
1701 rcu_read_unlock_bh();
1702
1703 return result;
1704 }
1705
1706 /* Gets referenced address, destroys ifaddr */
1707
addrconf_dad_stop(struct inet6_ifaddr * ifp,int dad_failed)1708 static void addrconf_dad_stop(struct inet6_ifaddr *ifp, int dad_failed)
1709 {
1710 if (ifp->flags&IFA_F_TEMPORARY) {
1711 struct inet6_ifaddr *ifpub;
1712 spin_lock_bh(&ifp->lock);
1713 ifpub = ifp->ifpub;
1714 if (ifpub) {
1715 in6_ifa_hold(ifpub);
1716 spin_unlock_bh(&ifp->lock);
1717 ipv6_create_tempaddr(ifpub, ifp);
1718 in6_ifa_put(ifpub);
1719 } else {
1720 spin_unlock_bh(&ifp->lock);
1721 }
1722 ipv6_del_addr(ifp);
1723 } else if (ifp->flags&IFA_F_PERMANENT || !dad_failed) {
1724 spin_lock_bh(&ifp->lock);
1725 addrconf_del_dad_work(ifp);
1726 ifp->flags |= IFA_F_TENTATIVE;
1727 if (dad_failed)
1728 ifp->flags |= IFA_F_DADFAILED;
1729 spin_unlock_bh(&ifp->lock);
1730 if (dad_failed)
1731 ipv6_ifa_notify(0, ifp);
1732 in6_ifa_put(ifp);
1733 } else {
1734 ipv6_del_addr(ifp);
1735 }
1736 }
1737
addrconf_dad_end(struct inet6_ifaddr * ifp)1738 static int addrconf_dad_end(struct inet6_ifaddr *ifp)
1739 {
1740 int err = -ENOENT;
1741
1742 spin_lock_bh(&ifp->state_lock);
1743 if (ifp->state == INET6_IFADDR_STATE_DAD) {
1744 ifp->state = INET6_IFADDR_STATE_POSTDAD;
1745 err = 0;
1746 }
1747 spin_unlock_bh(&ifp->state_lock);
1748
1749 return err;
1750 }
1751
addrconf_dad_failure(struct inet6_ifaddr * ifp)1752 void addrconf_dad_failure(struct inet6_ifaddr *ifp)
1753 {
1754 struct inet6_dev *idev = ifp->idev;
1755
1756 if (addrconf_dad_end(ifp)) {
1757 in6_ifa_put(ifp);
1758 return;
1759 }
1760
1761 net_info_ratelimited("%s: IPv6 duplicate address %pI6c detected!\n",
1762 ifp->idev->dev->name, &ifp->addr);
1763
1764 if (idev->cnf.accept_dad > 1 && !idev->cnf.disable_ipv6) {
1765 struct in6_addr addr;
1766
1767 addr.s6_addr32[0] = htonl(0xfe800000);
1768 addr.s6_addr32[1] = 0;
1769
1770 if (!ipv6_generate_eui64(addr.s6_addr + 8, idev->dev) &&
1771 ipv6_addr_equal(&ifp->addr, &addr)) {
1772 /* DAD failed for link-local based on MAC address */
1773 idev->cnf.disable_ipv6 = 1;
1774
1775 pr_info("%s: IPv6 being disabled!\n",
1776 ifp->idev->dev->name);
1777 }
1778 }
1779
1780 spin_lock_bh(&ifp->state_lock);
1781 /* transition from _POSTDAD to _ERRDAD */
1782 ifp->state = INET6_IFADDR_STATE_ERRDAD;
1783 spin_unlock_bh(&ifp->state_lock);
1784
1785 addrconf_mod_dad_work(ifp, 0);
1786 in6_ifa_put(ifp);
1787 }
1788
1789 /* Join to solicited addr multicast group.
1790 * caller must hold RTNL */
addrconf_join_solict(struct net_device * dev,const struct in6_addr * addr)1791 void addrconf_join_solict(struct net_device *dev, const struct in6_addr *addr)
1792 {
1793 struct in6_addr maddr;
1794
1795 if (dev->flags&(IFF_LOOPBACK|IFF_NOARP))
1796 return;
1797
1798 addrconf_addr_solict_mult(addr, &maddr);
1799 ipv6_dev_mc_inc(dev, &maddr);
1800 }
1801
1802 /* caller must hold RTNL */
addrconf_leave_solict(struct inet6_dev * idev,const struct in6_addr * addr)1803 void addrconf_leave_solict(struct inet6_dev *idev, const struct in6_addr *addr)
1804 {
1805 struct in6_addr maddr;
1806
1807 if (idev->dev->flags&(IFF_LOOPBACK|IFF_NOARP))
1808 return;
1809
1810 addrconf_addr_solict_mult(addr, &maddr);
1811 __ipv6_dev_mc_dec(idev, &maddr);
1812 }
1813
1814 /* caller must hold RTNL */
addrconf_join_anycast(struct inet6_ifaddr * ifp)1815 static void addrconf_join_anycast(struct inet6_ifaddr *ifp)
1816 {
1817 struct in6_addr addr;
1818
1819 if (ifp->prefix_len >= 127) /* RFC 6164 */
1820 return;
1821 ipv6_addr_prefix(&addr, &ifp->addr, ifp->prefix_len);
1822 if (ipv6_addr_any(&addr))
1823 return;
1824 __ipv6_dev_ac_inc(ifp->idev, &addr);
1825 }
1826
1827 /* caller must hold RTNL */
addrconf_leave_anycast(struct inet6_ifaddr * ifp)1828 static void addrconf_leave_anycast(struct inet6_ifaddr *ifp)
1829 {
1830 struct in6_addr addr;
1831
1832 if (ifp->prefix_len >= 127) /* RFC 6164 */
1833 return;
1834 ipv6_addr_prefix(&addr, &ifp->addr, ifp->prefix_len);
1835 if (ipv6_addr_any(&addr))
1836 return;
1837 __ipv6_dev_ac_dec(ifp->idev, &addr);
1838 }
1839
addrconf_ifid_eui48(u8 * eui,struct net_device * dev)1840 static int addrconf_ifid_eui48(u8 *eui, struct net_device *dev)
1841 {
1842 if (dev->addr_len != ETH_ALEN)
1843 return -1;
1844 memcpy(eui, dev->dev_addr, 3);
1845 memcpy(eui + 5, dev->dev_addr + 3, 3);
1846
1847 /*
1848 * The zSeries OSA network cards can be shared among various
1849 * OS instances, but the OSA cards have only one MAC address.
1850 * This leads to duplicate address conflicts in conjunction
1851 * with IPv6 if more than one instance uses the same card.
1852 *
1853 * The driver for these cards can deliver a unique 16-bit
1854 * identifier for each instance sharing the same card. It is
1855 * placed instead of 0xFFFE in the interface identifier. The
1856 * "u" bit of the interface identifier is not inverted in this
1857 * case. Hence the resulting interface identifier has local
1858 * scope according to RFC2373.
1859 */
1860 if (dev->dev_id) {
1861 eui[3] = (dev->dev_id >> 8) & 0xFF;
1862 eui[4] = dev->dev_id & 0xFF;
1863 } else {
1864 eui[3] = 0xFF;
1865 eui[4] = 0xFE;
1866 eui[0] ^= 2;
1867 }
1868 return 0;
1869 }
1870
addrconf_ifid_eui64(u8 * eui,struct net_device * dev)1871 static int addrconf_ifid_eui64(u8 *eui, struct net_device *dev)
1872 {
1873 if (dev->addr_len != IEEE802154_ADDR_LEN)
1874 return -1;
1875 memcpy(eui, dev->dev_addr, 8);
1876 eui[0] ^= 2;
1877 return 0;
1878 }
1879
addrconf_ifid_ieee1394(u8 * eui,struct net_device * dev)1880 static int addrconf_ifid_ieee1394(u8 *eui, struct net_device *dev)
1881 {
1882 union fwnet_hwaddr *ha;
1883
1884 if (dev->addr_len != FWNET_ALEN)
1885 return -1;
1886
1887 ha = (union fwnet_hwaddr *)dev->dev_addr;
1888
1889 memcpy(eui, &ha->uc.uniq_id, sizeof(ha->uc.uniq_id));
1890 eui[0] ^= 2;
1891 return 0;
1892 }
1893
addrconf_ifid_arcnet(u8 * eui,struct net_device * dev)1894 static int addrconf_ifid_arcnet(u8 *eui, struct net_device *dev)
1895 {
1896 /* XXX: inherit EUI-64 from other interface -- yoshfuji */
1897 if (dev->addr_len != ARCNET_ALEN)
1898 return -1;
1899 memset(eui, 0, 7);
1900 eui[7] = *(u8 *)dev->dev_addr;
1901 return 0;
1902 }
1903
addrconf_ifid_infiniband(u8 * eui,struct net_device * dev)1904 static int addrconf_ifid_infiniband(u8 *eui, struct net_device *dev)
1905 {
1906 if (dev->addr_len != INFINIBAND_ALEN)
1907 return -1;
1908 memcpy(eui, dev->dev_addr + 12, 8);
1909 eui[0] |= 2;
1910 return 0;
1911 }
1912
__ipv6_isatap_ifid(u8 * eui,__be32 addr)1913 static int __ipv6_isatap_ifid(u8 *eui, __be32 addr)
1914 {
1915 if (addr == 0)
1916 return -1;
1917 eui[0] = (ipv4_is_zeronet(addr) || ipv4_is_private_10(addr) ||
1918 ipv4_is_loopback(addr) || ipv4_is_linklocal_169(addr) ||
1919 ipv4_is_private_172(addr) || ipv4_is_test_192(addr) ||
1920 ipv4_is_anycast_6to4(addr) || ipv4_is_private_192(addr) ||
1921 ipv4_is_test_198(addr) || ipv4_is_multicast(addr) ||
1922 ipv4_is_lbcast(addr)) ? 0x00 : 0x02;
1923 eui[1] = 0;
1924 eui[2] = 0x5E;
1925 eui[3] = 0xFE;
1926 memcpy(eui + 4, &addr, 4);
1927 return 0;
1928 }
1929
addrconf_ifid_sit(u8 * eui,struct net_device * dev)1930 static int addrconf_ifid_sit(u8 *eui, struct net_device *dev)
1931 {
1932 if (dev->priv_flags & IFF_ISATAP)
1933 return __ipv6_isatap_ifid(eui, *(__be32 *)dev->dev_addr);
1934 return -1;
1935 }
1936
addrconf_ifid_gre(u8 * eui,struct net_device * dev)1937 static int addrconf_ifid_gre(u8 *eui, struct net_device *dev)
1938 {
1939 return __ipv6_isatap_ifid(eui, *(__be32 *)dev->dev_addr);
1940 }
1941
addrconf_ifid_ip6tnl(u8 * eui,struct net_device * dev)1942 static int addrconf_ifid_ip6tnl(u8 *eui, struct net_device *dev)
1943 {
1944 memcpy(eui, dev->perm_addr, 3);
1945 memcpy(eui + 5, dev->perm_addr + 3, 3);
1946 eui[3] = 0xFF;
1947 eui[4] = 0xFE;
1948 eui[0] ^= 2;
1949 return 0;
1950 }
1951
ipv6_generate_eui64(u8 * eui,struct net_device * dev)1952 static int ipv6_generate_eui64(u8 *eui, struct net_device *dev)
1953 {
1954 switch (dev->type) {
1955 case ARPHRD_ETHER:
1956 case ARPHRD_FDDI:
1957 return addrconf_ifid_eui48(eui, dev);
1958 case ARPHRD_ARCNET:
1959 return addrconf_ifid_arcnet(eui, dev);
1960 case ARPHRD_INFINIBAND:
1961 return addrconf_ifid_infiniband(eui, dev);
1962 case ARPHRD_SIT:
1963 return addrconf_ifid_sit(eui, dev);
1964 case ARPHRD_IPGRE:
1965 return addrconf_ifid_gre(eui, dev);
1966 case ARPHRD_6LOWPAN:
1967 case ARPHRD_IEEE802154:
1968 return addrconf_ifid_eui64(eui, dev);
1969 case ARPHRD_IEEE1394:
1970 return addrconf_ifid_ieee1394(eui, dev);
1971 case ARPHRD_TUNNEL6:
1972 return addrconf_ifid_ip6tnl(eui, dev);
1973 }
1974 return -1;
1975 }
1976
ipv6_inherit_eui64(u8 * eui,struct inet6_dev * idev)1977 static int ipv6_inherit_eui64(u8 *eui, struct inet6_dev *idev)
1978 {
1979 int err = -1;
1980 struct inet6_ifaddr *ifp;
1981
1982 read_lock_bh(&idev->lock);
1983 list_for_each_entry_reverse(ifp, &idev->addr_list, if_list) {
1984 if (ifp->scope > IFA_LINK)
1985 break;
1986 if (ifp->scope == IFA_LINK && !(ifp->flags&IFA_F_TENTATIVE)) {
1987 memcpy(eui, ifp->addr.s6_addr+8, 8);
1988 err = 0;
1989 break;
1990 }
1991 }
1992 read_unlock_bh(&idev->lock);
1993 return err;
1994 }
1995
1996 /* (re)generation of randomized interface identifier (RFC 3041 3.2, 3.5) */
__ipv6_regen_rndid(struct inet6_dev * idev)1997 static void __ipv6_regen_rndid(struct inet6_dev *idev)
1998 {
1999 regen:
2000 get_random_bytes(idev->rndid, sizeof(idev->rndid));
2001 idev->rndid[0] &= ~0x02;
2002
2003 /*
2004 * <draft-ietf-ipngwg-temp-addresses-v2-00.txt>:
2005 * check if generated address is not inappropriate
2006 *
2007 * - Reserved subnet anycast (RFC 2526)
2008 * 11111101 11....11 1xxxxxxx
2009 * - ISATAP (RFC4214) 6.1
2010 * 00-00-5E-FE-xx-xx-xx-xx
2011 * - value 0
2012 * - XXX: already assigned to an address on the device
2013 */
2014 if (idev->rndid[0] == 0xfd &&
2015 (idev->rndid[1]&idev->rndid[2]&idev->rndid[3]&idev->rndid[4]&idev->rndid[5]&idev->rndid[6]) == 0xff &&
2016 (idev->rndid[7]&0x80))
2017 goto regen;
2018 if ((idev->rndid[0]|idev->rndid[1]) == 0) {
2019 if (idev->rndid[2] == 0x5e && idev->rndid[3] == 0xfe)
2020 goto regen;
2021 if ((idev->rndid[2]|idev->rndid[3]|idev->rndid[4]|idev->rndid[5]|idev->rndid[6]|idev->rndid[7]) == 0x00)
2022 goto regen;
2023 }
2024 }
2025
ipv6_regen_rndid(unsigned long data)2026 static void ipv6_regen_rndid(unsigned long data)
2027 {
2028 struct inet6_dev *idev = (struct inet6_dev *) data;
2029 unsigned long expires;
2030
2031 rcu_read_lock_bh();
2032 write_lock_bh(&idev->lock);
2033
2034 if (idev->dead)
2035 goto out;
2036
2037 __ipv6_regen_rndid(idev);
2038
2039 expires = jiffies +
2040 idev->cnf.temp_prefered_lft * HZ -
2041 idev->cnf.regen_max_retry * idev->cnf.dad_transmits *
2042 NEIGH_VAR(idev->nd_parms, RETRANS_TIME) -
2043 idev->cnf.max_desync_factor * HZ;
2044 if (time_before(expires, jiffies)) {
2045 pr_warn("%s: too short regeneration interval; timer disabled for %s\n",
2046 __func__, idev->dev->name);
2047 goto out;
2048 }
2049
2050 if (!mod_timer(&idev->regen_timer, expires))
2051 in6_dev_hold(idev);
2052
2053 out:
2054 write_unlock_bh(&idev->lock);
2055 rcu_read_unlock_bh();
2056 in6_dev_put(idev);
2057 }
2058
__ipv6_try_regen_rndid(struct inet6_dev * idev,struct in6_addr * tmpaddr)2059 static void __ipv6_try_regen_rndid(struct inet6_dev *idev, struct in6_addr *tmpaddr)
2060 {
2061 if (tmpaddr && memcmp(idev->rndid, &tmpaddr->s6_addr[8], 8) == 0)
2062 __ipv6_regen_rndid(idev);
2063 }
2064
addrconf_rt_table(const struct net_device * dev,u32 default_table)2065 u32 addrconf_rt_table(const struct net_device *dev, u32 default_table) {
2066 /* Determines into what table to put autoconf PIO/RIO/default routes
2067 * learned on this device.
2068 *
2069 * - If 0, use the same table for every device. This puts routes into
2070 * one of RT_TABLE_{PREFIX,INFO,DFLT} depending on the type of route
2071 * (but note that these three are currently all equal to
2072 * RT6_TABLE_MAIN).
2073 * - If > 0, use the specified table.
2074 * - If < 0, put routes into table dev->ifindex + (-rt_table).
2075 */
2076 struct inet6_dev *idev = in6_dev_get(dev);
2077 u32 table;
2078 int sysctl = idev->cnf.accept_ra_rt_table;
2079 if (sysctl == 0) {
2080 table = default_table;
2081 } else if (sysctl > 0) {
2082 table = (u32) sysctl;
2083 } else {
2084 table = (unsigned) dev->ifindex + (-sysctl);
2085 }
2086 in6_dev_put(idev);
2087 return table;
2088 }
2089
2090 /*
2091 * Add prefix route.
2092 */
2093
2094 static void
addrconf_prefix_route(struct in6_addr * pfx,int plen,struct net_device * dev,unsigned long expires,u32 flags)2095 addrconf_prefix_route(struct in6_addr *pfx, int plen, struct net_device *dev,
2096 unsigned long expires, u32 flags)
2097 {
2098 struct fib6_config cfg = {
2099 .fc_table = addrconf_rt_table(dev, RT6_TABLE_PREFIX),
2100 .fc_metric = IP6_RT_PRIO_ADDRCONF,
2101 .fc_ifindex = dev->ifindex,
2102 .fc_expires = expires,
2103 .fc_dst_len = plen,
2104 .fc_flags = RTF_UP | flags,
2105 .fc_nlinfo.nl_net = dev_net(dev),
2106 .fc_protocol = RTPROT_KERNEL,
2107 };
2108
2109 cfg.fc_dst = *pfx;
2110
2111 /* Prevent useless cloning on PtP SIT.
2112 This thing is done here expecting that the whole
2113 class of non-broadcast devices need not cloning.
2114 */
2115 #if IS_ENABLED(CONFIG_IPV6_SIT)
2116 if (dev->type == ARPHRD_SIT && (dev->flags & IFF_POINTOPOINT))
2117 cfg.fc_flags |= RTF_NONEXTHOP;
2118 #endif
2119
2120 ip6_route_add(&cfg);
2121 }
2122
2123
addrconf_get_prefix_route(const struct in6_addr * pfx,int plen,const struct net_device * dev,u32 flags,u32 noflags)2124 static struct rt6_info *addrconf_get_prefix_route(const struct in6_addr *pfx,
2125 int plen,
2126 const struct net_device *dev,
2127 u32 flags, u32 noflags)
2128 {
2129 struct fib6_node *fn;
2130 struct rt6_info *rt = NULL;
2131 struct fib6_table *table;
2132
2133 table = fib6_get_table(dev_net(dev),
2134 addrconf_rt_table(dev, RT6_TABLE_PREFIX));
2135 if (table == NULL)
2136 return NULL;
2137
2138 read_lock_bh(&table->tb6_lock);
2139 fn = fib6_locate(&table->tb6_root, pfx, plen, NULL, 0);
2140 if (!fn)
2141 goto out;
2142 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
2143 if (rt->dst.dev->ifindex != dev->ifindex)
2144 continue;
2145 if ((rt->rt6i_flags & flags) != flags)
2146 continue;
2147 if ((rt->rt6i_flags & noflags) != 0)
2148 continue;
2149 dst_hold(&rt->dst);
2150 break;
2151 }
2152 out:
2153 read_unlock_bh(&table->tb6_lock);
2154 return rt;
2155 }
2156
2157
2158 /* Create "default" multicast route to the interface */
2159
addrconf_add_mroute(struct net_device * dev)2160 static void addrconf_add_mroute(struct net_device *dev)
2161 {
2162 struct fib6_config cfg = {
2163 .fc_table = RT6_TABLE_LOCAL,
2164 .fc_metric = IP6_RT_PRIO_ADDRCONF,
2165 .fc_ifindex = dev->ifindex,
2166 .fc_dst_len = 8,
2167 .fc_flags = RTF_UP,
2168 .fc_nlinfo.nl_net = dev_net(dev),
2169 };
2170
2171 ipv6_addr_set(&cfg.fc_dst, htonl(0xFF000000), 0, 0, 0);
2172
2173 ip6_route_add(&cfg);
2174 }
2175
addrconf_add_dev(struct net_device * dev)2176 static struct inet6_dev *addrconf_add_dev(struct net_device *dev)
2177 {
2178 struct inet6_dev *idev;
2179
2180 ASSERT_RTNL();
2181
2182 idev = ipv6_find_idev(dev);
2183 if (!idev)
2184 return ERR_PTR(-ENOBUFS);
2185
2186 if (idev->cnf.disable_ipv6)
2187 return ERR_PTR(-EACCES);
2188
2189 /* Add default multicast route */
2190 if (!(dev->flags & IFF_LOOPBACK))
2191 addrconf_add_mroute(dev);
2192
2193 return idev;
2194 }
2195
manage_tempaddrs(struct inet6_dev * idev,struct inet6_ifaddr * ifp,__u32 valid_lft,__u32 prefered_lft,bool create,unsigned long now)2196 static void manage_tempaddrs(struct inet6_dev *idev,
2197 struct inet6_ifaddr *ifp,
2198 __u32 valid_lft, __u32 prefered_lft,
2199 bool create, unsigned long now)
2200 {
2201 u32 flags;
2202 struct inet6_ifaddr *ift;
2203
2204 read_lock_bh(&idev->lock);
2205 /* update all temporary addresses in the list */
2206 list_for_each_entry(ift, &idev->tempaddr_list, tmp_list) {
2207 int age, max_valid, max_prefered;
2208
2209 if (ifp != ift->ifpub)
2210 continue;
2211
2212 /* RFC 4941 section 3.3:
2213 * If a received option will extend the lifetime of a public
2214 * address, the lifetimes of temporary addresses should
2215 * be extended, subject to the overall constraint that no
2216 * temporary addresses should ever remain "valid" or "preferred"
2217 * for a time longer than (TEMP_VALID_LIFETIME) or
2218 * (TEMP_PREFERRED_LIFETIME - DESYNC_FACTOR), respectively.
2219 */
2220 age = (now - ift->cstamp) / HZ;
2221 max_valid = idev->cnf.temp_valid_lft - age;
2222 if (max_valid < 0)
2223 max_valid = 0;
2224
2225 max_prefered = idev->cnf.temp_prefered_lft -
2226 idev->cnf.max_desync_factor - age;
2227 if (max_prefered < 0)
2228 max_prefered = 0;
2229
2230 if (valid_lft > max_valid)
2231 valid_lft = max_valid;
2232
2233 if (prefered_lft > max_prefered)
2234 prefered_lft = max_prefered;
2235
2236 spin_lock(&ift->lock);
2237 flags = ift->flags;
2238 ift->valid_lft = valid_lft;
2239 ift->prefered_lft = prefered_lft;
2240 ift->tstamp = now;
2241 if (prefered_lft > 0)
2242 ift->flags &= ~IFA_F_DEPRECATED;
2243
2244 spin_unlock(&ift->lock);
2245 if (!(flags&IFA_F_TENTATIVE))
2246 ipv6_ifa_notify(0, ift);
2247 }
2248
2249 if ((create || list_empty(&idev->tempaddr_list)) &&
2250 idev->cnf.use_tempaddr > 0) {
2251 /* When a new public address is created as described
2252 * in [ADDRCONF], also create a new temporary address.
2253 * Also create a temporary address if it's enabled but
2254 * no temporary address currently exists.
2255 */
2256 read_unlock_bh(&idev->lock);
2257 ipv6_create_tempaddr(ifp, NULL);
2258 } else {
2259 read_unlock_bh(&idev->lock);
2260 }
2261 }
2262
addrconf_prefix_rcv(struct net_device * dev,u8 * opt,int len,bool sllao)2263 void addrconf_prefix_rcv(struct net_device *dev, u8 *opt, int len, bool sllao)
2264 {
2265 struct prefix_info *pinfo;
2266 __u32 valid_lft;
2267 __u32 prefered_lft;
2268 int addr_type;
2269 struct inet6_dev *in6_dev;
2270 struct net *net = dev_net(dev);
2271
2272 pinfo = (struct prefix_info *) opt;
2273
2274 if (len < sizeof(struct prefix_info)) {
2275 ADBG("addrconf: prefix option too short\n");
2276 return;
2277 }
2278
2279 /*
2280 * Validation checks ([ADDRCONF], page 19)
2281 */
2282
2283 addr_type = ipv6_addr_type(&pinfo->prefix);
2284
2285 if (addr_type & (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL))
2286 return;
2287
2288 valid_lft = ntohl(pinfo->valid);
2289 prefered_lft = ntohl(pinfo->prefered);
2290
2291 if (prefered_lft > valid_lft) {
2292 net_warn_ratelimited("addrconf: prefix option has invalid lifetime\n");
2293 return;
2294 }
2295
2296 in6_dev = in6_dev_get(dev);
2297
2298 if (in6_dev == NULL) {
2299 net_dbg_ratelimited("addrconf: device %s not configured\n",
2300 dev->name);
2301 return;
2302 }
2303
2304 /*
2305 * Two things going on here:
2306 * 1) Add routes for on-link prefixes
2307 * 2) Configure prefixes with the auto flag set
2308 */
2309
2310 if (pinfo->onlink) {
2311 struct rt6_info *rt;
2312 unsigned long rt_expires;
2313
2314 /* Avoid arithmetic overflow. Really, we could
2315 * save rt_expires in seconds, likely valid_lft,
2316 * but it would require division in fib gc, that it
2317 * not good.
2318 */
2319 if (HZ > USER_HZ)
2320 rt_expires = addrconf_timeout_fixup(valid_lft, HZ);
2321 else
2322 rt_expires = addrconf_timeout_fixup(valid_lft, USER_HZ);
2323
2324 if (addrconf_finite_timeout(rt_expires))
2325 rt_expires *= HZ;
2326
2327 rt = addrconf_get_prefix_route(&pinfo->prefix,
2328 pinfo->prefix_len,
2329 dev,
2330 RTF_ADDRCONF | RTF_PREFIX_RT,
2331 RTF_GATEWAY | RTF_DEFAULT);
2332
2333 if (rt) {
2334 /* Autoconf prefix route */
2335 if (valid_lft == 0) {
2336 ip6_del_rt(rt);
2337 rt = NULL;
2338 } else if (addrconf_finite_timeout(rt_expires)) {
2339 /* not infinity */
2340 rt6_set_expires(rt, jiffies + rt_expires);
2341 } else {
2342 rt6_clean_expires(rt);
2343 }
2344 } else if (valid_lft) {
2345 clock_t expires = 0;
2346 int flags = RTF_ADDRCONF | RTF_PREFIX_RT;
2347 if (addrconf_finite_timeout(rt_expires)) {
2348 /* not infinity */
2349 flags |= RTF_EXPIRES;
2350 expires = jiffies_to_clock_t(rt_expires);
2351 }
2352 addrconf_prefix_route(&pinfo->prefix, pinfo->prefix_len,
2353 dev, expires, flags);
2354 }
2355 ip6_rt_put(rt);
2356 }
2357
2358 /* Try to figure out our local address for this prefix */
2359
2360 if (pinfo->autoconf && in6_dev->cnf.autoconf) {
2361 struct inet6_ifaddr *ifp;
2362 struct in6_addr addr;
2363 int create = 0, update_lft = 0;
2364 bool tokenized = false;
2365
2366 if (pinfo->prefix_len == 64) {
2367 memcpy(&addr, &pinfo->prefix, 8);
2368
2369 if (!ipv6_addr_any(&in6_dev->token)) {
2370 read_lock_bh(&in6_dev->lock);
2371 memcpy(addr.s6_addr + 8,
2372 in6_dev->token.s6_addr + 8, 8);
2373 read_unlock_bh(&in6_dev->lock);
2374 tokenized = true;
2375 } else if (ipv6_generate_eui64(addr.s6_addr + 8, dev) &&
2376 ipv6_inherit_eui64(addr.s6_addr + 8, in6_dev)) {
2377 in6_dev_put(in6_dev);
2378 return;
2379 }
2380 goto ok;
2381 }
2382 net_dbg_ratelimited("IPv6 addrconf: prefix with wrong length %d\n",
2383 pinfo->prefix_len);
2384 in6_dev_put(in6_dev);
2385 return;
2386
2387 ok:
2388
2389 ifp = ipv6_get_ifaddr(net, &addr, dev, 1);
2390
2391 if (ifp == NULL && valid_lft) {
2392 int max_addresses = in6_dev->cnf.max_addresses;
2393 u32 addr_flags = 0;
2394
2395 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
2396 if (in6_dev->cnf.optimistic_dad &&
2397 !net->ipv6.devconf_all->forwarding && sllao)
2398 addr_flags = IFA_F_OPTIMISTIC;
2399 #endif
2400
2401 /* Do not allow to create too much of autoconfigured
2402 * addresses; this would be too easy way to crash kernel.
2403 */
2404 if (!max_addresses ||
2405 ipv6_count_addresses(in6_dev) < max_addresses)
2406 ifp = ipv6_add_addr(in6_dev, &addr, NULL,
2407 pinfo->prefix_len,
2408 addr_type&IPV6_ADDR_SCOPE_MASK,
2409 addr_flags, valid_lft,
2410 prefered_lft);
2411
2412 if (IS_ERR_OR_NULL(ifp)) {
2413 in6_dev_put(in6_dev);
2414 return;
2415 }
2416
2417 update_lft = 0;
2418 create = 1;
2419 spin_lock_bh(&ifp->lock);
2420 ifp->flags |= IFA_F_MANAGETEMPADDR;
2421 ifp->cstamp = jiffies;
2422 ifp->tokenized = tokenized;
2423 spin_unlock_bh(&ifp->lock);
2424 addrconf_dad_start(ifp);
2425 }
2426
2427 if (ifp) {
2428 u32 flags;
2429 unsigned long now;
2430 u32 stored_lft;
2431
2432 /* update lifetime (RFC2462 5.5.3 e) */
2433 spin_lock(&ifp->lock);
2434 now = jiffies;
2435 if (ifp->valid_lft > (now - ifp->tstamp) / HZ)
2436 stored_lft = ifp->valid_lft - (now - ifp->tstamp) / HZ;
2437 else
2438 stored_lft = 0;
2439 if (!update_lft && !create && stored_lft) {
2440 const u32 minimum_lft = min(
2441 stored_lft, (u32)MIN_VALID_LIFETIME);
2442 valid_lft = max(valid_lft, minimum_lft);
2443
2444 /* RFC4862 Section 5.5.3e:
2445 * "Note that the preferred lifetime of the
2446 * corresponding address is always reset to
2447 * the Preferred Lifetime in the received
2448 * Prefix Information option, regardless of
2449 * whether the valid lifetime is also reset or
2450 * ignored."
2451 *
2452 * So we should always update prefered_lft here.
2453 */
2454 update_lft = 1;
2455 }
2456
2457 if (update_lft) {
2458 ifp->valid_lft = valid_lft;
2459 ifp->prefered_lft = prefered_lft;
2460 ifp->tstamp = now;
2461 flags = ifp->flags;
2462 ifp->flags &= ~IFA_F_DEPRECATED;
2463 spin_unlock(&ifp->lock);
2464
2465 if (!(flags&IFA_F_TENTATIVE))
2466 ipv6_ifa_notify(0, ifp);
2467 } else
2468 spin_unlock(&ifp->lock);
2469
2470 manage_tempaddrs(in6_dev, ifp, valid_lft, prefered_lft,
2471 create, now);
2472
2473 in6_ifa_put(ifp);
2474 addrconf_verify();
2475 }
2476 }
2477 inet6_prefix_notify(RTM_NEWPREFIX, in6_dev, pinfo);
2478 in6_dev_put(in6_dev);
2479 }
2480
2481 /*
2482 * Set destination address.
2483 * Special case for SIT interfaces where we create a new "virtual"
2484 * device.
2485 */
addrconf_set_dstaddr(struct net * net,void __user * arg)2486 int addrconf_set_dstaddr(struct net *net, void __user *arg)
2487 {
2488 struct in6_ifreq ireq;
2489 struct net_device *dev;
2490 int err = -EINVAL;
2491
2492 rtnl_lock();
2493
2494 err = -EFAULT;
2495 if (copy_from_user(&ireq, arg, sizeof(struct in6_ifreq)))
2496 goto err_exit;
2497
2498 dev = __dev_get_by_index(net, ireq.ifr6_ifindex);
2499
2500 err = -ENODEV;
2501 if (dev == NULL)
2502 goto err_exit;
2503
2504 #if IS_ENABLED(CONFIG_IPV6_SIT)
2505 if (dev->type == ARPHRD_SIT) {
2506 const struct net_device_ops *ops = dev->netdev_ops;
2507 struct ifreq ifr;
2508 struct ip_tunnel_parm p;
2509
2510 err = -EADDRNOTAVAIL;
2511 if (!(ipv6_addr_type(&ireq.ifr6_addr) & IPV6_ADDR_COMPATv4))
2512 goto err_exit;
2513
2514 memset(&p, 0, sizeof(p));
2515 p.iph.daddr = ireq.ifr6_addr.s6_addr32[3];
2516 p.iph.saddr = 0;
2517 p.iph.version = 4;
2518 p.iph.ihl = 5;
2519 p.iph.protocol = IPPROTO_IPV6;
2520 p.iph.ttl = 64;
2521 ifr.ifr_ifru.ifru_data = (__force void __user *)&p;
2522
2523 if (ops->ndo_do_ioctl) {
2524 mm_segment_t oldfs = get_fs();
2525
2526 set_fs(KERNEL_DS);
2527 err = ops->ndo_do_ioctl(dev, &ifr, SIOCADDTUNNEL);
2528 set_fs(oldfs);
2529 } else
2530 err = -EOPNOTSUPP;
2531
2532 if (err == 0) {
2533 err = -ENOBUFS;
2534 dev = __dev_get_by_name(net, p.name);
2535 if (!dev)
2536 goto err_exit;
2537 err = dev_open(dev);
2538 }
2539 }
2540 #endif
2541
2542 err_exit:
2543 rtnl_unlock();
2544 return err;
2545 }
2546
2547 /*
2548 * Manual configuration of address on an interface
2549 */
inet6_addr_add(struct net * net,int ifindex,const struct in6_addr * pfx,const struct in6_addr * peer_pfx,unsigned int plen,__u32 ifa_flags,__u32 prefered_lft,__u32 valid_lft)2550 static int inet6_addr_add(struct net *net, int ifindex,
2551 const struct in6_addr *pfx,
2552 const struct in6_addr *peer_pfx,
2553 unsigned int plen, __u32 ifa_flags,
2554 __u32 prefered_lft, __u32 valid_lft)
2555 {
2556 struct inet6_ifaddr *ifp;
2557 struct inet6_dev *idev;
2558 struct net_device *dev;
2559 int scope;
2560 u32 flags;
2561 clock_t expires;
2562 unsigned long timeout;
2563
2564 ASSERT_RTNL();
2565
2566 if (plen > 128)
2567 return -EINVAL;
2568
2569 /* check the lifetime */
2570 if (!valid_lft || prefered_lft > valid_lft)
2571 return -EINVAL;
2572
2573 if (ifa_flags & IFA_F_MANAGETEMPADDR && plen != 64)
2574 return -EINVAL;
2575
2576 dev = __dev_get_by_index(net, ifindex);
2577 if (!dev)
2578 return -ENODEV;
2579
2580 idev = addrconf_add_dev(dev);
2581 if (IS_ERR(idev))
2582 return PTR_ERR(idev);
2583
2584 scope = ipv6_addr_scope(pfx);
2585
2586 timeout = addrconf_timeout_fixup(valid_lft, HZ);
2587 if (addrconf_finite_timeout(timeout)) {
2588 expires = jiffies_to_clock_t(timeout * HZ);
2589 valid_lft = timeout;
2590 flags = RTF_EXPIRES;
2591 } else {
2592 expires = 0;
2593 flags = 0;
2594 ifa_flags |= IFA_F_PERMANENT;
2595 }
2596
2597 timeout = addrconf_timeout_fixup(prefered_lft, HZ);
2598 if (addrconf_finite_timeout(timeout)) {
2599 if (timeout == 0)
2600 ifa_flags |= IFA_F_DEPRECATED;
2601 prefered_lft = timeout;
2602 }
2603
2604 ifp = ipv6_add_addr(idev, pfx, peer_pfx, plen, scope, ifa_flags,
2605 valid_lft, prefered_lft);
2606
2607 if (!IS_ERR(ifp)) {
2608 if (!(ifa_flags & IFA_F_NOPREFIXROUTE)) {
2609 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, dev,
2610 expires, flags);
2611 }
2612
2613 /*
2614 * Note that section 3.1 of RFC 4429 indicates
2615 * that the Optimistic flag should not be set for
2616 * manually configured addresses
2617 */
2618 addrconf_dad_start(ifp);
2619 if (ifa_flags & IFA_F_MANAGETEMPADDR)
2620 manage_tempaddrs(idev, ifp, valid_lft, prefered_lft,
2621 true, jiffies);
2622 in6_ifa_put(ifp);
2623 addrconf_verify_rtnl();
2624 return 0;
2625 }
2626
2627 return PTR_ERR(ifp);
2628 }
2629
inet6_addr_del(struct net * net,int ifindex,u32 ifa_flags,const struct in6_addr * pfx,unsigned int plen)2630 static int inet6_addr_del(struct net *net, int ifindex, u32 ifa_flags,
2631 const struct in6_addr *pfx, unsigned int plen)
2632 {
2633 struct inet6_ifaddr *ifp;
2634 struct inet6_dev *idev;
2635 struct net_device *dev;
2636
2637 if (plen > 128)
2638 return -EINVAL;
2639
2640 dev = __dev_get_by_index(net, ifindex);
2641 if (!dev)
2642 return -ENODEV;
2643
2644 if ((idev = __in6_dev_get(dev)) == NULL)
2645 return -ENXIO;
2646
2647 read_lock_bh(&idev->lock);
2648 list_for_each_entry(ifp, &idev->addr_list, if_list) {
2649 if (ifp->prefix_len == plen &&
2650 ipv6_addr_equal(pfx, &ifp->addr)) {
2651 in6_ifa_hold(ifp);
2652 read_unlock_bh(&idev->lock);
2653
2654 if (!(ifp->flags & IFA_F_TEMPORARY) &&
2655 (ifa_flags & IFA_F_MANAGETEMPADDR))
2656 manage_tempaddrs(idev, ifp, 0, 0, false,
2657 jiffies);
2658 ipv6_del_addr(ifp);
2659 addrconf_verify_rtnl();
2660 return 0;
2661 }
2662 }
2663 read_unlock_bh(&idev->lock);
2664 return -EADDRNOTAVAIL;
2665 }
2666
2667
addrconf_add_ifaddr(struct net * net,void __user * arg)2668 int addrconf_add_ifaddr(struct net *net, void __user *arg)
2669 {
2670 struct in6_ifreq ireq;
2671 int err;
2672
2673 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
2674 return -EPERM;
2675
2676 if (copy_from_user(&ireq, arg, sizeof(struct in6_ifreq)))
2677 return -EFAULT;
2678
2679 rtnl_lock();
2680 err = inet6_addr_add(net, ireq.ifr6_ifindex, &ireq.ifr6_addr, NULL,
2681 ireq.ifr6_prefixlen, IFA_F_PERMANENT,
2682 INFINITY_LIFE_TIME, INFINITY_LIFE_TIME);
2683 rtnl_unlock();
2684 return err;
2685 }
2686
addrconf_del_ifaddr(struct net * net,void __user * arg)2687 int addrconf_del_ifaddr(struct net *net, void __user *arg)
2688 {
2689 struct in6_ifreq ireq;
2690 int err;
2691
2692 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
2693 return -EPERM;
2694
2695 if (copy_from_user(&ireq, arg, sizeof(struct in6_ifreq)))
2696 return -EFAULT;
2697
2698 rtnl_lock();
2699 err = inet6_addr_del(net, ireq.ifr6_ifindex, 0, &ireq.ifr6_addr,
2700 ireq.ifr6_prefixlen);
2701 rtnl_unlock();
2702 return err;
2703 }
2704
add_addr(struct inet6_dev * idev,const struct in6_addr * addr,int plen,int scope)2705 static void add_addr(struct inet6_dev *idev, const struct in6_addr *addr,
2706 int plen, int scope)
2707 {
2708 struct inet6_ifaddr *ifp;
2709
2710 ifp = ipv6_add_addr(idev, addr, NULL, plen,
2711 scope, IFA_F_PERMANENT,
2712 INFINITY_LIFE_TIME, INFINITY_LIFE_TIME);
2713 if (!IS_ERR(ifp)) {
2714 spin_lock_bh(&ifp->lock);
2715 ifp->flags &= ~IFA_F_TENTATIVE;
2716 spin_unlock_bh(&ifp->lock);
2717 ipv6_ifa_notify(RTM_NEWADDR, ifp);
2718 in6_ifa_put(ifp);
2719 }
2720 }
2721
2722 #if IS_ENABLED(CONFIG_IPV6_SIT)
sit_add_v4_addrs(struct inet6_dev * idev)2723 static void sit_add_v4_addrs(struct inet6_dev *idev)
2724 {
2725 struct in6_addr addr;
2726 struct net_device *dev;
2727 struct net *net = dev_net(idev->dev);
2728 int scope, plen;
2729 u32 pflags = 0;
2730
2731 ASSERT_RTNL();
2732
2733 memset(&addr, 0, sizeof(struct in6_addr));
2734 memcpy(&addr.s6_addr32[3], idev->dev->dev_addr, 4);
2735
2736 if (idev->dev->flags&IFF_POINTOPOINT) {
2737 addr.s6_addr32[0] = htonl(0xfe800000);
2738 scope = IFA_LINK;
2739 plen = 64;
2740 } else {
2741 scope = IPV6_ADDR_COMPATv4;
2742 plen = 96;
2743 pflags |= RTF_NONEXTHOP;
2744 }
2745
2746 if (addr.s6_addr32[3]) {
2747 add_addr(idev, &addr, plen, scope);
2748 addrconf_prefix_route(&addr, plen, idev->dev, 0, pflags);
2749 return;
2750 }
2751
2752 for_each_netdev(net, dev) {
2753 struct in_device *in_dev = __in_dev_get_rtnl(dev);
2754 if (in_dev && (dev->flags & IFF_UP)) {
2755 struct in_ifaddr *ifa;
2756
2757 int flag = scope;
2758
2759 for (ifa = in_dev->ifa_list; ifa; ifa = ifa->ifa_next) {
2760
2761 addr.s6_addr32[3] = ifa->ifa_local;
2762
2763 if (ifa->ifa_scope == RT_SCOPE_LINK)
2764 continue;
2765 if (ifa->ifa_scope >= RT_SCOPE_HOST) {
2766 if (idev->dev->flags&IFF_POINTOPOINT)
2767 continue;
2768 flag |= IFA_HOST;
2769 }
2770
2771 add_addr(idev, &addr, plen, flag);
2772 addrconf_prefix_route(&addr, plen, idev->dev, 0,
2773 pflags);
2774 }
2775 }
2776 }
2777 }
2778 #endif
2779
init_loopback(struct net_device * dev)2780 static void init_loopback(struct net_device *dev)
2781 {
2782 struct inet6_dev *idev;
2783 struct net_device *sp_dev;
2784 struct inet6_ifaddr *sp_ifa;
2785 struct rt6_info *sp_rt;
2786
2787 /* ::1 */
2788
2789 ASSERT_RTNL();
2790
2791 if ((idev = ipv6_find_idev(dev)) == NULL) {
2792 pr_debug("%s: add_dev failed\n", __func__);
2793 return;
2794 }
2795
2796 add_addr(idev, &in6addr_loopback, 128, IFA_HOST);
2797
2798 /* Add routes to other interface's IPv6 addresses */
2799 for_each_netdev(dev_net(dev), sp_dev) {
2800 if (!strcmp(sp_dev->name, dev->name))
2801 continue;
2802
2803 idev = __in6_dev_get(sp_dev);
2804 if (!idev)
2805 continue;
2806
2807 read_lock_bh(&idev->lock);
2808 list_for_each_entry(sp_ifa, &idev->addr_list, if_list) {
2809
2810 if (sp_ifa->flags & (IFA_F_DADFAILED | IFA_F_TENTATIVE))
2811 continue;
2812
2813 if (sp_ifa->rt) {
2814 /* This dst has been added to garbage list when
2815 * lo device down, release this obsolete dst and
2816 * reallocate a new router for ifa.
2817 */
2818 if (!atomic_read(&sp_ifa->rt->rt6i_ref)) {
2819 ip6_rt_put(sp_ifa->rt);
2820 sp_ifa->rt = NULL;
2821 } else {
2822 continue;
2823 }
2824 }
2825
2826 sp_rt = addrconf_dst_alloc(idev, &sp_ifa->addr, false);
2827
2828 /* Failure cases are ignored */
2829 if (!IS_ERR(sp_rt)) {
2830 sp_ifa->rt = sp_rt;
2831 ip6_ins_rt(sp_rt);
2832 }
2833 }
2834 read_unlock_bh(&idev->lock);
2835 }
2836 }
2837
addrconf_add_linklocal(struct inet6_dev * idev,const struct in6_addr * addr)2838 static void addrconf_add_linklocal(struct inet6_dev *idev, const struct in6_addr *addr)
2839 {
2840 struct inet6_ifaddr *ifp;
2841 u32 addr_flags = IFA_F_PERMANENT;
2842
2843 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
2844 if (idev->cnf.optimistic_dad &&
2845 !dev_net(idev->dev)->ipv6.devconf_all->forwarding)
2846 addr_flags |= IFA_F_OPTIMISTIC;
2847 #endif
2848
2849
2850 ifp = ipv6_add_addr(idev, addr, NULL, 64, IFA_LINK, addr_flags,
2851 INFINITY_LIFE_TIME, INFINITY_LIFE_TIME);
2852 if (!IS_ERR(ifp)) {
2853 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, idev->dev, 0, 0);
2854 addrconf_dad_start(ifp);
2855 in6_ifa_put(ifp);
2856 }
2857 }
2858
addrconf_addr_gen(struct inet6_dev * idev,bool prefix_route)2859 static void addrconf_addr_gen(struct inet6_dev *idev, bool prefix_route)
2860 {
2861 if (idev->addr_gen_mode == IN6_ADDR_GEN_MODE_EUI64) {
2862 struct in6_addr addr;
2863
2864 ipv6_addr_set(&addr, htonl(0xFE800000), 0, 0, 0);
2865 /* addrconf_add_linklocal also adds a prefix_route and we
2866 * only need to care about prefix routes if ipv6_generate_eui64
2867 * couldn't generate one.
2868 */
2869 if (ipv6_generate_eui64(addr.s6_addr + 8, idev->dev) == 0)
2870 addrconf_add_linklocal(idev, &addr);
2871 else if (prefix_route)
2872 addrconf_prefix_route(&addr, 64, idev->dev, 0, 0);
2873 }
2874 }
2875
addrconf_dev_config(struct net_device * dev)2876 static void addrconf_dev_config(struct net_device *dev)
2877 {
2878 struct inet6_dev *idev;
2879
2880 ASSERT_RTNL();
2881
2882 if ((dev->type != ARPHRD_ETHER) &&
2883 (dev->type != ARPHRD_FDDI) &&
2884 (dev->type != ARPHRD_ARCNET) &&
2885 (dev->type != ARPHRD_INFINIBAND) &&
2886 (dev->type != ARPHRD_IEEE802154) &&
2887 (dev->type != ARPHRD_IEEE1394) &&
2888 (dev->type != ARPHRD_TUNNEL6) &&
2889 (dev->type != ARPHRD_6LOWPAN)) {
2890 /* Alas, we support only Ethernet autoconfiguration. */
2891 return;
2892 }
2893
2894 idev = addrconf_add_dev(dev);
2895 if (IS_ERR(idev))
2896 return;
2897
2898 addrconf_addr_gen(idev, false);
2899 }
2900
2901 #if IS_ENABLED(CONFIG_IPV6_SIT)
addrconf_sit_config(struct net_device * dev)2902 static void addrconf_sit_config(struct net_device *dev)
2903 {
2904 struct inet6_dev *idev;
2905
2906 ASSERT_RTNL();
2907
2908 /*
2909 * Configure the tunnel with one of our IPv4
2910 * addresses... we should configure all of
2911 * our v4 addrs in the tunnel
2912 */
2913
2914 if ((idev = ipv6_find_idev(dev)) == NULL) {
2915 pr_debug("%s: add_dev failed\n", __func__);
2916 return;
2917 }
2918
2919 if (dev->priv_flags & IFF_ISATAP) {
2920 addrconf_addr_gen(idev, false);
2921 return;
2922 }
2923
2924 sit_add_v4_addrs(idev);
2925
2926 if (dev->flags&IFF_POINTOPOINT)
2927 addrconf_add_mroute(dev);
2928 }
2929 #endif
2930
2931 #if IS_ENABLED(CONFIG_NET_IPGRE)
addrconf_gre_config(struct net_device * dev)2932 static void addrconf_gre_config(struct net_device *dev)
2933 {
2934 struct inet6_dev *idev;
2935
2936 ASSERT_RTNL();
2937
2938 if ((idev = ipv6_find_idev(dev)) == NULL) {
2939 pr_debug("%s: add_dev failed\n", __func__);
2940 return;
2941 }
2942
2943 addrconf_addr_gen(idev, true);
2944 }
2945 #endif
2946
addrconf_notify(struct notifier_block * this,unsigned long event,void * ptr)2947 static int addrconf_notify(struct notifier_block *this, unsigned long event,
2948 void *ptr)
2949 {
2950 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
2951 struct inet6_dev *idev = __in6_dev_get(dev);
2952 struct net *net = dev_net(dev);
2953 int run_pending = 0;
2954 int err;
2955
2956 switch (event) {
2957 case NETDEV_REGISTER:
2958 if (!idev && dev->mtu >= IPV6_MIN_MTU) {
2959 idev = ipv6_add_dev(dev);
2960 if (IS_ERR(idev))
2961 return notifier_from_errno(PTR_ERR(idev));
2962 }
2963 break;
2964
2965 case NETDEV_UP:
2966 case NETDEV_CHANGE:
2967 if (dev->flags & IFF_SLAVE)
2968 break;
2969
2970 if (idev && idev->cnf.disable_ipv6)
2971 break;
2972
2973 if (event == NETDEV_UP) {
2974 if (!addrconf_qdisc_ok(dev)) {
2975 /* device is not ready yet. */
2976 pr_info("ADDRCONF(NETDEV_UP): %s: link is not ready\n",
2977 dev->name);
2978 break;
2979 }
2980
2981 if (!idev && dev->mtu >= IPV6_MIN_MTU)
2982 idev = ipv6_add_dev(dev);
2983
2984 if (!IS_ERR_OR_NULL(idev)) {
2985 idev->if_flags |= IF_READY;
2986 run_pending = 1;
2987 }
2988 } else {
2989 if (!addrconf_qdisc_ok(dev)) {
2990 /* device is still not ready. */
2991 break;
2992 }
2993
2994 if (idev) {
2995 if (idev->if_flags & IF_READY)
2996 /* device is already configured. */
2997 break;
2998 idev->if_flags |= IF_READY;
2999 }
3000
3001 pr_info("ADDRCONF(NETDEV_CHANGE): %s: link becomes ready\n",
3002 dev->name);
3003
3004 run_pending = 1;
3005 }
3006
3007 switch (dev->type) {
3008 #if IS_ENABLED(CONFIG_IPV6_SIT)
3009 case ARPHRD_SIT:
3010 addrconf_sit_config(dev);
3011 break;
3012 #endif
3013 #if IS_ENABLED(CONFIG_NET_IPGRE)
3014 case ARPHRD_IPGRE:
3015 addrconf_gre_config(dev);
3016 break;
3017 #endif
3018 case ARPHRD_LOOPBACK:
3019 init_loopback(dev);
3020 break;
3021
3022 default:
3023 addrconf_dev_config(dev);
3024 break;
3025 }
3026
3027 if (!IS_ERR_OR_NULL(idev)) {
3028 if (run_pending)
3029 addrconf_dad_run(idev);
3030
3031 /*
3032 * If the MTU changed during the interface down,
3033 * when the interface up, the changed MTU must be
3034 * reflected in the idev as well as routers.
3035 */
3036 if (idev->cnf.mtu6 != dev->mtu &&
3037 dev->mtu >= IPV6_MIN_MTU) {
3038 rt6_mtu_change(dev, dev->mtu);
3039 idev->cnf.mtu6 = dev->mtu;
3040 }
3041 idev->tstamp = jiffies;
3042 inet6_ifinfo_notify(RTM_NEWLINK, idev);
3043
3044 /*
3045 * If the changed mtu during down is lower than
3046 * IPV6_MIN_MTU stop IPv6 on this interface.
3047 */
3048 if (dev->mtu < IPV6_MIN_MTU)
3049 addrconf_ifdown(dev, dev != net->loopback_dev);
3050 }
3051 break;
3052
3053 case NETDEV_CHANGEMTU:
3054 if (idev && dev->mtu >= IPV6_MIN_MTU) {
3055 rt6_mtu_change(dev, dev->mtu);
3056 idev->cnf.mtu6 = dev->mtu;
3057 break;
3058 }
3059
3060 if (!idev && dev->mtu >= IPV6_MIN_MTU) {
3061 idev = ipv6_add_dev(dev);
3062 if (!IS_ERR(idev))
3063 break;
3064 }
3065
3066 /*
3067 * if MTU under IPV6_MIN_MTU.
3068 * Stop IPv6 on this interface.
3069 */
3070
3071 case NETDEV_DOWN:
3072 case NETDEV_UNREGISTER:
3073 /*
3074 * Remove all addresses from this interface.
3075 */
3076 addrconf_ifdown(dev, event != NETDEV_DOWN);
3077 break;
3078
3079 case NETDEV_CHANGENAME:
3080 if (idev) {
3081 snmp6_unregister_dev(idev);
3082 addrconf_sysctl_unregister(idev);
3083 err = addrconf_sysctl_register(idev);
3084 if (err)
3085 return notifier_from_errno(err);
3086 err = snmp6_register_dev(idev);
3087 if (err) {
3088 addrconf_sysctl_unregister(idev);
3089 return notifier_from_errno(err);
3090 }
3091 }
3092 break;
3093
3094 case NETDEV_PRE_TYPE_CHANGE:
3095 case NETDEV_POST_TYPE_CHANGE:
3096 addrconf_type_change(dev, event);
3097 break;
3098 }
3099
3100 return NOTIFY_OK;
3101 }
3102
3103 /*
3104 * addrconf module should be notified of a device going up
3105 */
3106 static struct notifier_block ipv6_dev_notf = {
3107 .notifier_call = addrconf_notify,
3108 .priority = ADDRCONF_NOTIFY_PRIORITY,
3109 };
3110
addrconf_type_change(struct net_device * dev,unsigned long event)3111 static void addrconf_type_change(struct net_device *dev, unsigned long event)
3112 {
3113 struct inet6_dev *idev;
3114 ASSERT_RTNL();
3115
3116 idev = __in6_dev_get(dev);
3117
3118 if (event == NETDEV_POST_TYPE_CHANGE)
3119 ipv6_mc_remap(idev);
3120 else if (event == NETDEV_PRE_TYPE_CHANGE)
3121 ipv6_mc_unmap(idev);
3122 }
3123
addrconf_ifdown(struct net_device * dev,int how)3124 static int addrconf_ifdown(struct net_device *dev, int how)
3125 {
3126 struct net *net = dev_net(dev);
3127 struct inet6_dev *idev;
3128 struct inet6_ifaddr *ifa;
3129 int state, i;
3130
3131 ASSERT_RTNL();
3132
3133 rt6_ifdown(net, dev);
3134 neigh_ifdown(&nd_tbl, dev);
3135
3136 idev = __in6_dev_get(dev);
3137 if (idev == NULL)
3138 return -ENODEV;
3139
3140 /*
3141 * Step 1: remove reference to ipv6 device from parent device.
3142 * Do not dev_put!
3143 */
3144 if (how) {
3145 idev->dead = 1;
3146
3147 /* protected by rtnl_lock */
3148 RCU_INIT_POINTER(dev->ip6_ptr, NULL);
3149
3150 /* Step 1.5: remove snmp6 entry */
3151 snmp6_unregister_dev(idev);
3152
3153 }
3154
3155 /* Step 2: clear hash table */
3156 for (i = 0; i < IN6_ADDR_HSIZE; i++) {
3157 struct hlist_head *h = &inet6_addr_lst[i];
3158
3159 spin_lock_bh(&addrconf_hash_lock);
3160 restart:
3161 hlist_for_each_entry_rcu(ifa, h, addr_lst) {
3162 if (ifa->idev == idev) {
3163 hlist_del_init_rcu(&ifa->addr_lst);
3164 addrconf_del_dad_work(ifa);
3165 goto restart;
3166 }
3167 }
3168 spin_unlock_bh(&addrconf_hash_lock);
3169 }
3170
3171 write_lock_bh(&idev->lock);
3172
3173 addrconf_del_rs_timer(idev);
3174
3175 /* Step 2: clear flags for stateless addrconf */
3176 if (!how)
3177 idev->if_flags &= ~(IF_RS_SENT|IF_RA_RCVD|IF_READY);
3178
3179 if (how && del_timer(&idev->regen_timer))
3180 in6_dev_put(idev);
3181
3182 /* Step 3: clear tempaddr list */
3183 while (!list_empty(&idev->tempaddr_list)) {
3184 ifa = list_first_entry(&idev->tempaddr_list,
3185 struct inet6_ifaddr, tmp_list);
3186 list_del(&ifa->tmp_list);
3187 write_unlock_bh(&idev->lock);
3188 spin_lock_bh(&ifa->lock);
3189
3190 if (ifa->ifpub) {
3191 in6_ifa_put(ifa->ifpub);
3192 ifa->ifpub = NULL;
3193 }
3194 spin_unlock_bh(&ifa->lock);
3195 in6_ifa_put(ifa);
3196 write_lock_bh(&idev->lock);
3197 }
3198
3199 while (!list_empty(&idev->addr_list)) {
3200 ifa = list_first_entry(&idev->addr_list,
3201 struct inet6_ifaddr, if_list);
3202 addrconf_del_dad_work(ifa);
3203
3204 list_del(&ifa->if_list);
3205
3206 write_unlock_bh(&idev->lock);
3207
3208 spin_lock_bh(&ifa->state_lock);
3209 state = ifa->state;
3210 ifa->state = INET6_IFADDR_STATE_DEAD;
3211 spin_unlock_bh(&ifa->state_lock);
3212
3213 if (state != INET6_IFADDR_STATE_DEAD) {
3214 __ipv6_ifa_notify(RTM_DELADDR, ifa);
3215 inet6addr_notifier_call_chain(NETDEV_DOWN, ifa);
3216 }
3217 in6_ifa_put(ifa);
3218
3219 write_lock_bh(&idev->lock);
3220 }
3221
3222 write_unlock_bh(&idev->lock);
3223
3224 /* Step 5: Discard anycast and multicast list */
3225 if (how) {
3226 ipv6_ac_destroy_dev(idev);
3227 ipv6_mc_destroy_dev(idev);
3228 } else {
3229 ipv6_mc_down(idev);
3230 }
3231
3232 idev->tstamp = jiffies;
3233
3234 /* Last: Shot the device (if unregistered) */
3235 if (how) {
3236 addrconf_sysctl_unregister(idev);
3237 neigh_parms_release(&nd_tbl, idev->nd_parms);
3238 neigh_ifdown(&nd_tbl, dev);
3239 in6_dev_put(idev);
3240 }
3241 return 0;
3242 }
3243
addrconf_rs_timer(unsigned long data)3244 static void addrconf_rs_timer(unsigned long data)
3245 {
3246 struct inet6_dev *idev = (struct inet6_dev *)data;
3247 struct net_device *dev = idev->dev;
3248 struct in6_addr lladdr;
3249
3250 write_lock(&idev->lock);
3251 if (idev->dead || !(idev->if_flags & IF_READY))
3252 goto out;
3253
3254 if (!ipv6_accept_ra(idev))
3255 goto out;
3256
3257 /* Announcement received after solicitation was sent */
3258 if (idev->if_flags & IF_RA_RCVD)
3259 goto out;
3260
3261 if (idev->rs_probes++ < idev->cnf.rtr_solicits || idev->cnf.rtr_solicits < 0) {
3262 write_unlock(&idev->lock);
3263 if (!ipv6_get_lladdr(dev, &lladdr, IFA_F_TENTATIVE))
3264 ndisc_send_rs(dev, &lladdr,
3265 &in6addr_linklocal_allrouters);
3266 else
3267 goto put;
3268
3269 write_lock(&idev->lock);
3270 idev->rs_interval = rfc3315_s14_backoff_update(
3271 idev->rs_interval, idev->cnf.rtr_solicit_max_interval);
3272 /* The wait after the last probe can be shorter */
3273 addrconf_mod_rs_timer(idev, (idev->rs_probes ==
3274 idev->cnf.rtr_solicits) ?
3275 idev->cnf.rtr_solicit_delay :
3276 idev->rs_interval);
3277 } else {
3278 /*
3279 * Note: we do not support deprecated "all on-link"
3280 * assumption any longer.
3281 */
3282 pr_debug("%s: no IPv6 routers present\n", idev->dev->name);
3283 }
3284
3285 out:
3286 write_unlock(&idev->lock);
3287 put:
3288 in6_dev_put(idev);
3289 }
3290
3291 /*
3292 * Duplicate Address Detection
3293 */
addrconf_dad_kick(struct inet6_ifaddr * ifp)3294 static void addrconf_dad_kick(struct inet6_ifaddr *ifp)
3295 {
3296 unsigned long rand_num;
3297 struct inet6_dev *idev = ifp->idev;
3298
3299 if (ifp->flags & IFA_F_OPTIMISTIC)
3300 rand_num = 0;
3301 else
3302 rand_num = prandom_u32() % (idev->cnf.rtr_solicit_delay ? : 1);
3303
3304 ifp->dad_probes = idev->cnf.dad_transmits;
3305 addrconf_mod_dad_work(ifp, rand_num);
3306 }
3307
addrconf_dad_begin(struct inet6_ifaddr * ifp)3308 static void addrconf_dad_begin(struct inet6_ifaddr *ifp)
3309 {
3310 struct inet6_dev *idev = ifp->idev;
3311 struct net_device *dev = idev->dev;
3312
3313 addrconf_join_solict(dev, &ifp->addr);
3314
3315 prandom_seed((__force u32) ifp->addr.s6_addr32[3]);
3316
3317 read_lock_bh(&idev->lock);
3318 spin_lock(&ifp->lock);
3319 if (ifp->state == INET6_IFADDR_STATE_DEAD)
3320 goto out;
3321
3322 if (dev->flags&(IFF_NOARP|IFF_LOOPBACK) ||
3323 idev->cnf.accept_dad < 1 ||
3324 !(ifp->flags&IFA_F_TENTATIVE) ||
3325 ifp->flags & IFA_F_NODAD) {
3326 ifp->flags &= ~(IFA_F_TENTATIVE|IFA_F_OPTIMISTIC|IFA_F_DADFAILED);
3327 spin_unlock(&ifp->lock);
3328 read_unlock_bh(&idev->lock);
3329
3330 addrconf_dad_completed(ifp);
3331 return;
3332 }
3333
3334 if (!(idev->if_flags & IF_READY)) {
3335 spin_unlock(&ifp->lock);
3336 read_unlock_bh(&idev->lock);
3337 /*
3338 * If the device is not ready:
3339 * - keep it tentative if it is a permanent address.
3340 * - otherwise, kill it.
3341 */
3342 in6_ifa_hold(ifp);
3343 addrconf_dad_stop(ifp, 0);
3344 return;
3345 }
3346
3347 /*
3348 * Optimistic nodes can start receiving
3349 * Frames right away
3350 */
3351 if (ifp->flags & IFA_F_OPTIMISTIC) {
3352 ip6_ins_rt(ifp->rt);
3353 if (ipv6_use_optimistic_addr(idev)) {
3354 /* Because optimistic nodes can use this address,
3355 * notify listeners. If DAD fails, RTM_DELADDR is sent.
3356 */
3357 ipv6_ifa_notify(RTM_NEWADDR, ifp);
3358 }
3359 }
3360
3361 addrconf_dad_kick(ifp);
3362 out:
3363 spin_unlock(&ifp->lock);
3364 read_unlock_bh(&idev->lock);
3365 }
3366
addrconf_dad_start(struct inet6_ifaddr * ifp)3367 static void addrconf_dad_start(struct inet6_ifaddr *ifp)
3368 {
3369 bool begin_dad = false;
3370
3371 spin_lock_bh(&ifp->state_lock);
3372 if (ifp->state != INET6_IFADDR_STATE_DEAD) {
3373 ifp->state = INET6_IFADDR_STATE_PREDAD;
3374 begin_dad = true;
3375 }
3376 spin_unlock_bh(&ifp->state_lock);
3377
3378 if (begin_dad)
3379 addrconf_mod_dad_work(ifp, 0);
3380 }
3381
addrconf_dad_work(struct work_struct * w)3382 static void addrconf_dad_work(struct work_struct *w)
3383 {
3384 struct inet6_ifaddr *ifp = container_of(to_delayed_work(w),
3385 struct inet6_ifaddr,
3386 dad_work);
3387 struct inet6_dev *idev = ifp->idev;
3388 struct in6_addr mcaddr;
3389
3390 enum {
3391 DAD_PROCESS,
3392 DAD_BEGIN,
3393 DAD_ABORT,
3394 } action = DAD_PROCESS;
3395
3396 rtnl_lock();
3397
3398 spin_lock_bh(&ifp->state_lock);
3399 if (ifp->state == INET6_IFADDR_STATE_PREDAD) {
3400 action = DAD_BEGIN;
3401 ifp->state = INET6_IFADDR_STATE_DAD;
3402 } else if (ifp->state == INET6_IFADDR_STATE_ERRDAD) {
3403 action = DAD_ABORT;
3404 ifp->state = INET6_IFADDR_STATE_POSTDAD;
3405 }
3406 spin_unlock_bh(&ifp->state_lock);
3407
3408 if (action == DAD_BEGIN) {
3409 addrconf_dad_begin(ifp);
3410 goto out;
3411 } else if (action == DAD_ABORT) {
3412 in6_ifa_hold(ifp);
3413 addrconf_dad_stop(ifp, 1);
3414 goto out;
3415 }
3416
3417 if (!ifp->dad_probes && addrconf_dad_end(ifp))
3418 goto out;
3419
3420 write_lock_bh(&idev->lock);
3421 if (idev->dead || !(idev->if_flags & IF_READY)) {
3422 write_unlock_bh(&idev->lock);
3423 goto out;
3424 }
3425
3426 spin_lock(&ifp->lock);
3427 if (ifp->state == INET6_IFADDR_STATE_DEAD) {
3428 spin_unlock(&ifp->lock);
3429 write_unlock_bh(&idev->lock);
3430 goto out;
3431 }
3432
3433 if (ifp->dad_probes == 0) {
3434 /*
3435 * DAD was successful
3436 */
3437
3438 ifp->flags &= ~(IFA_F_TENTATIVE|IFA_F_OPTIMISTIC|IFA_F_DADFAILED);
3439 spin_unlock(&ifp->lock);
3440 write_unlock_bh(&idev->lock);
3441
3442 addrconf_dad_completed(ifp);
3443
3444 goto out;
3445 }
3446
3447 ifp->dad_probes--;
3448 addrconf_mod_dad_work(ifp,
3449 NEIGH_VAR(ifp->idev->nd_parms, RETRANS_TIME));
3450 spin_unlock(&ifp->lock);
3451 write_unlock_bh(&idev->lock);
3452
3453 /* send a neighbour solicitation for our addr */
3454 addrconf_addr_solict_mult(&ifp->addr, &mcaddr);
3455 ndisc_send_ns(ifp->idev->dev, NULL, &ifp->addr, &mcaddr, &in6addr_any);
3456 out:
3457 in6_ifa_put(ifp);
3458 rtnl_unlock();
3459 }
3460
3461 /* ifp->idev must be at least read locked */
ipv6_lonely_lladdr(struct inet6_ifaddr * ifp)3462 static bool ipv6_lonely_lladdr(struct inet6_ifaddr *ifp)
3463 {
3464 struct inet6_ifaddr *ifpiter;
3465 struct inet6_dev *idev = ifp->idev;
3466
3467 list_for_each_entry_reverse(ifpiter, &idev->addr_list, if_list) {
3468 if (ifpiter->scope > IFA_LINK)
3469 break;
3470 if (ifp != ifpiter && ifpiter->scope == IFA_LINK &&
3471 (ifpiter->flags & (IFA_F_PERMANENT|IFA_F_TENTATIVE|
3472 IFA_F_OPTIMISTIC|IFA_F_DADFAILED)) ==
3473 IFA_F_PERMANENT)
3474 return false;
3475 }
3476 return true;
3477 }
3478
addrconf_dad_completed(struct inet6_ifaddr * ifp)3479 static void addrconf_dad_completed(struct inet6_ifaddr *ifp)
3480 {
3481 struct net_device *dev = ifp->idev->dev;
3482 struct in6_addr lladdr;
3483 bool send_rs, send_mld;
3484
3485 addrconf_del_dad_work(ifp);
3486
3487 /*
3488 * Configure the address for reception. Now it is valid.
3489 */
3490
3491 ipv6_ifa_notify(RTM_NEWADDR, ifp);
3492
3493 /* If added prefix is link local and we are prepared to process
3494 router advertisements, start sending router solicitations.
3495 */
3496
3497 read_lock_bh(&ifp->idev->lock);
3498 send_mld = ifp->scope == IFA_LINK && ipv6_lonely_lladdr(ifp);
3499 send_rs = send_mld &&
3500 ipv6_accept_ra(ifp->idev) &&
3501 ifp->idev->cnf.rtr_solicits != 0 &&
3502 (dev->flags&IFF_LOOPBACK) == 0;
3503 read_unlock_bh(&ifp->idev->lock);
3504
3505 /* While dad is in progress mld report's source address is in6_addrany.
3506 * Resend with proper ll now.
3507 */
3508 if (send_mld)
3509 ipv6_mc_dad_complete(ifp->idev);
3510
3511 if (send_rs) {
3512 /*
3513 * If a host as already performed a random delay
3514 * [...] as part of DAD [...] there is no need
3515 * to delay again before sending the first RS
3516 */
3517 if (ipv6_get_lladdr(dev, &lladdr, IFA_F_TENTATIVE))
3518 return;
3519 ndisc_send_rs(dev, &lladdr, &in6addr_linklocal_allrouters);
3520
3521 write_lock_bh(&ifp->idev->lock);
3522 spin_lock(&ifp->lock);
3523 ifp->idev->rs_interval = rfc3315_s14_backoff_init(
3524 ifp->idev->cnf.rtr_solicit_interval);
3525 ifp->idev->rs_probes = 1;
3526 ifp->idev->if_flags |= IF_RS_SENT;
3527 addrconf_mod_rs_timer(ifp->idev, ifp->idev->rs_interval);
3528 spin_unlock(&ifp->lock);
3529 write_unlock_bh(&ifp->idev->lock);
3530 }
3531 }
3532
addrconf_dad_run(struct inet6_dev * idev)3533 static void addrconf_dad_run(struct inet6_dev *idev)
3534 {
3535 struct inet6_ifaddr *ifp;
3536
3537 read_lock_bh(&idev->lock);
3538 list_for_each_entry(ifp, &idev->addr_list, if_list) {
3539 spin_lock(&ifp->lock);
3540 if (ifp->flags & IFA_F_TENTATIVE &&
3541 ifp->state == INET6_IFADDR_STATE_DAD)
3542 addrconf_dad_kick(ifp);
3543 spin_unlock(&ifp->lock);
3544 }
3545 read_unlock_bh(&idev->lock);
3546 }
3547
3548 #ifdef CONFIG_PROC_FS
3549 struct if6_iter_state {
3550 struct seq_net_private p;
3551 int bucket;
3552 int offset;
3553 };
3554
if6_get_first(struct seq_file * seq,loff_t pos)3555 static struct inet6_ifaddr *if6_get_first(struct seq_file *seq, loff_t pos)
3556 {
3557 struct inet6_ifaddr *ifa = NULL;
3558 struct if6_iter_state *state = seq->private;
3559 struct net *net = seq_file_net(seq);
3560 int p = 0;
3561
3562 /* initial bucket if pos is 0 */
3563 if (pos == 0) {
3564 state->bucket = 0;
3565 state->offset = 0;
3566 }
3567
3568 for (; state->bucket < IN6_ADDR_HSIZE; ++state->bucket) {
3569 hlist_for_each_entry_rcu_bh(ifa, &inet6_addr_lst[state->bucket],
3570 addr_lst) {
3571 if (!net_eq(dev_net(ifa->idev->dev), net))
3572 continue;
3573 /* sync with offset */
3574 if (p < state->offset) {
3575 p++;
3576 continue;
3577 }
3578 state->offset++;
3579 return ifa;
3580 }
3581
3582 /* prepare for next bucket */
3583 state->offset = 0;
3584 p = 0;
3585 }
3586 return NULL;
3587 }
3588
if6_get_next(struct seq_file * seq,struct inet6_ifaddr * ifa)3589 static struct inet6_ifaddr *if6_get_next(struct seq_file *seq,
3590 struct inet6_ifaddr *ifa)
3591 {
3592 struct if6_iter_state *state = seq->private;
3593 struct net *net = seq_file_net(seq);
3594
3595 hlist_for_each_entry_continue_rcu_bh(ifa, addr_lst) {
3596 if (!net_eq(dev_net(ifa->idev->dev), net))
3597 continue;
3598 state->offset++;
3599 return ifa;
3600 }
3601
3602 while (++state->bucket < IN6_ADDR_HSIZE) {
3603 state->offset = 0;
3604 hlist_for_each_entry_rcu_bh(ifa,
3605 &inet6_addr_lst[state->bucket], addr_lst) {
3606 if (!net_eq(dev_net(ifa->idev->dev), net))
3607 continue;
3608 state->offset++;
3609 return ifa;
3610 }
3611 }
3612
3613 return NULL;
3614 }
3615
if6_seq_start(struct seq_file * seq,loff_t * pos)3616 static void *if6_seq_start(struct seq_file *seq, loff_t *pos)
3617 __acquires(rcu_bh)
3618 {
3619 rcu_read_lock_bh();
3620 return if6_get_first(seq, *pos);
3621 }
3622
if6_seq_next(struct seq_file * seq,void * v,loff_t * pos)3623 static void *if6_seq_next(struct seq_file *seq, void *v, loff_t *pos)
3624 {
3625 struct inet6_ifaddr *ifa;
3626
3627 ifa = if6_get_next(seq, v);
3628 ++*pos;
3629 return ifa;
3630 }
3631
if6_seq_stop(struct seq_file * seq,void * v)3632 static void if6_seq_stop(struct seq_file *seq, void *v)
3633 __releases(rcu_bh)
3634 {
3635 rcu_read_unlock_bh();
3636 }
3637
if6_seq_show(struct seq_file * seq,void * v)3638 static int if6_seq_show(struct seq_file *seq, void *v)
3639 {
3640 struct inet6_ifaddr *ifp = (struct inet6_ifaddr *)v;
3641 seq_printf(seq, "%pi6 %02x %02x %02x %02x %8s\n",
3642 &ifp->addr,
3643 ifp->idev->dev->ifindex,
3644 ifp->prefix_len,
3645 ifp->scope,
3646 (u8) ifp->flags,
3647 ifp->idev->dev->name);
3648 return 0;
3649 }
3650
3651 static const struct seq_operations if6_seq_ops = {
3652 .start = if6_seq_start,
3653 .next = if6_seq_next,
3654 .show = if6_seq_show,
3655 .stop = if6_seq_stop,
3656 };
3657
if6_seq_open(struct inode * inode,struct file * file)3658 static int if6_seq_open(struct inode *inode, struct file *file)
3659 {
3660 return seq_open_net(inode, file, &if6_seq_ops,
3661 sizeof(struct if6_iter_state));
3662 }
3663
3664 static const struct file_operations if6_fops = {
3665 .owner = THIS_MODULE,
3666 .open = if6_seq_open,
3667 .read = seq_read,
3668 .llseek = seq_lseek,
3669 .release = seq_release_net,
3670 };
3671
if6_proc_net_init(struct net * net)3672 static int __net_init if6_proc_net_init(struct net *net)
3673 {
3674 if (!proc_create("if_inet6", S_IRUGO, net->proc_net, &if6_fops))
3675 return -ENOMEM;
3676 return 0;
3677 }
3678
if6_proc_net_exit(struct net * net)3679 static void __net_exit if6_proc_net_exit(struct net *net)
3680 {
3681 remove_proc_entry("if_inet6", net->proc_net);
3682 }
3683
3684 static struct pernet_operations if6_proc_net_ops = {
3685 .init = if6_proc_net_init,
3686 .exit = if6_proc_net_exit,
3687 };
3688
if6_proc_init(void)3689 int __init if6_proc_init(void)
3690 {
3691 return register_pernet_subsys(&if6_proc_net_ops);
3692 }
3693
if6_proc_exit(void)3694 void if6_proc_exit(void)
3695 {
3696 unregister_pernet_subsys(&if6_proc_net_ops);
3697 }
3698 #endif /* CONFIG_PROC_FS */
3699
3700 #if IS_ENABLED(CONFIG_IPV6_MIP6)
3701 /* Check if address is a home address configured on any interface. */
ipv6_chk_home_addr(struct net * net,const struct in6_addr * addr)3702 int ipv6_chk_home_addr(struct net *net, const struct in6_addr *addr)
3703 {
3704 int ret = 0;
3705 struct inet6_ifaddr *ifp = NULL;
3706 unsigned int hash = inet6_addr_hash(addr);
3707
3708 rcu_read_lock_bh();
3709 hlist_for_each_entry_rcu_bh(ifp, &inet6_addr_lst[hash], addr_lst) {
3710 if (!net_eq(dev_net(ifp->idev->dev), net))
3711 continue;
3712 if (ipv6_addr_equal(&ifp->addr, addr) &&
3713 (ifp->flags & IFA_F_HOMEADDRESS)) {
3714 ret = 1;
3715 break;
3716 }
3717 }
3718 rcu_read_unlock_bh();
3719 return ret;
3720 }
3721 #endif
3722
3723 /*
3724 * Periodic address status verification
3725 */
3726
addrconf_verify_rtnl(void)3727 static void addrconf_verify_rtnl(void)
3728 {
3729 unsigned long now, next, next_sec, next_sched;
3730 struct inet6_ifaddr *ifp;
3731 int i;
3732
3733 ASSERT_RTNL();
3734
3735 rcu_read_lock_bh();
3736 now = jiffies;
3737 next = round_jiffies_up(now + ADDR_CHECK_FREQUENCY);
3738
3739 cancel_delayed_work(&addr_chk_work);
3740
3741 for (i = 0; i < IN6_ADDR_HSIZE; i++) {
3742 restart:
3743 hlist_for_each_entry_rcu_bh(ifp, &inet6_addr_lst[i], addr_lst) {
3744 unsigned long age;
3745
3746 /* When setting preferred_lft to a value not zero or
3747 * infinity, while valid_lft is infinity
3748 * IFA_F_PERMANENT has a non-infinity life time.
3749 */
3750 if ((ifp->flags & IFA_F_PERMANENT) &&
3751 (ifp->prefered_lft == INFINITY_LIFE_TIME))
3752 continue;
3753
3754 spin_lock(&ifp->lock);
3755 /* We try to batch several events at once. */
3756 age = (now - ifp->tstamp + ADDRCONF_TIMER_FUZZ_MINUS) / HZ;
3757
3758 if (ifp->valid_lft != INFINITY_LIFE_TIME &&
3759 age >= ifp->valid_lft) {
3760 spin_unlock(&ifp->lock);
3761 in6_ifa_hold(ifp);
3762 ipv6_del_addr(ifp);
3763 goto restart;
3764 } else if (ifp->prefered_lft == INFINITY_LIFE_TIME) {
3765 spin_unlock(&ifp->lock);
3766 continue;
3767 } else if (age >= ifp->prefered_lft) {
3768 /* jiffies - ifp->tstamp > age >= ifp->prefered_lft */
3769 int deprecate = 0;
3770
3771 if (!(ifp->flags&IFA_F_DEPRECATED)) {
3772 deprecate = 1;
3773 ifp->flags |= IFA_F_DEPRECATED;
3774 }
3775
3776 if ((ifp->valid_lft != INFINITY_LIFE_TIME) &&
3777 (time_before(ifp->tstamp + ifp->valid_lft * HZ, next)))
3778 next = ifp->tstamp + ifp->valid_lft * HZ;
3779
3780 spin_unlock(&ifp->lock);
3781
3782 if (deprecate) {
3783 in6_ifa_hold(ifp);
3784
3785 ipv6_ifa_notify(0, ifp);
3786 in6_ifa_put(ifp);
3787 goto restart;
3788 }
3789 } else if ((ifp->flags&IFA_F_TEMPORARY) &&
3790 !(ifp->flags&IFA_F_TENTATIVE)) {
3791 unsigned long regen_advance = ifp->idev->cnf.regen_max_retry *
3792 ifp->idev->cnf.dad_transmits *
3793 NEIGH_VAR(ifp->idev->nd_parms, RETRANS_TIME) / HZ;
3794
3795 if (age >= ifp->prefered_lft - regen_advance) {
3796 struct inet6_ifaddr *ifpub = ifp->ifpub;
3797 if (time_before(ifp->tstamp + ifp->prefered_lft * HZ, next))
3798 next = ifp->tstamp + ifp->prefered_lft * HZ;
3799 if (!ifp->regen_count && ifpub) {
3800 ifp->regen_count++;
3801 in6_ifa_hold(ifp);
3802 in6_ifa_hold(ifpub);
3803 spin_unlock(&ifp->lock);
3804
3805 spin_lock(&ifpub->lock);
3806 ifpub->regen_count = 0;
3807 spin_unlock(&ifpub->lock);
3808 ipv6_create_tempaddr(ifpub, ifp);
3809 in6_ifa_put(ifpub);
3810 in6_ifa_put(ifp);
3811 goto restart;
3812 }
3813 } else if (time_before(ifp->tstamp + ifp->prefered_lft * HZ - regen_advance * HZ, next))
3814 next = ifp->tstamp + ifp->prefered_lft * HZ - regen_advance * HZ;
3815 spin_unlock(&ifp->lock);
3816 } else {
3817 /* ifp->prefered_lft <= ifp->valid_lft */
3818 if (time_before(ifp->tstamp + ifp->prefered_lft * HZ, next))
3819 next = ifp->tstamp + ifp->prefered_lft * HZ;
3820 spin_unlock(&ifp->lock);
3821 }
3822 }
3823 }
3824
3825 next_sec = round_jiffies_up(next);
3826 next_sched = next;
3827
3828 /* If rounded timeout is accurate enough, accept it. */
3829 if (time_before(next_sec, next + ADDRCONF_TIMER_FUZZ))
3830 next_sched = next_sec;
3831
3832 /* And minimum interval is ADDRCONF_TIMER_FUZZ_MAX. */
3833 if (time_before(next_sched, jiffies + ADDRCONF_TIMER_FUZZ_MAX))
3834 next_sched = jiffies + ADDRCONF_TIMER_FUZZ_MAX;
3835
3836 ADBG(KERN_DEBUG "now = %lu, schedule = %lu, rounded schedule = %lu => %lu\n",
3837 now, next, next_sec, next_sched);
3838 mod_delayed_work(addrconf_wq, &addr_chk_work, next_sched - now);
3839 rcu_read_unlock_bh();
3840 }
3841
addrconf_verify_work(struct work_struct * w)3842 static void addrconf_verify_work(struct work_struct *w)
3843 {
3844 rtnl_lock();
3845 addrconf_verify_rtnl();
3846 rtnl_unlock();
3847 }
3848
addrconf_verify(void)3849 static void addrconf_verify(void)
3850 {
3851 mod_delayed_work(addrconf_wq, &addr_chk_work, 0);
3852 }
3853
extract_addr(struct nlattr * addr,struct nlattr * local,struct in6_addr ** peer_pfx)3854 static struct in6_addr *extract_addr(struct nlattr *addr, struct nlattr *local,
3855 struct in6_addr **peer_pfx)
3856 {
3857 struct in6_addr *pfx = NULL;
3858
3859 *peer_pfx = NULL;
3860
3861 if (addr)
3862 pfx = nla_data(addr);
3863
3864 if (local) {
3865 if (pfx && nla_memcmp(local, pfx, sizeof(*pfx)))
3866 *peer_pfx = pfx;
3867 pfx = nla_data(local);
3868 }
3869
3870 return pfx;
3871 }
3872
3873 static const struct nla_policy ifa_ipv6_policy[IFA_MAX+1] = {
3874 [IFA_ADDRESS] = { .len = sizeof(struct in6_addr) },
3875 [IFA_LOCAL] = { .len = sizeof(struct in6_addr) },
3876 [IFA_CACHEINFO] = { .len = sizeof(struct ifa_cacheinfo) },
3877 [IFA_FLAGS] = { .len = sizeof(u32) },
3878 };
3879
3880 static int
inet6_rtm_deladdr(struct sk_buff * skb,struct nlmsghdr * nlh)3881 inet6_rtm_deladdr(struct sk_buff *skb, struct nlmsghdr *nlh)
3882 {
3883 struct net *net = sock_net(skb->sk);
3884 struct ifaddrmsg *ifm;
3885 struct nlattr *tb[IFA_MAX+1];
3886 struct in6_addr *pfx, *peer_pfx;
3887 u32 ifa_flags;
3888 int err;
3889
3890 err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv6_policy);
3891 if (err < 0)
3892 return err;
3893
3894 ifm = nlmsg_data(nlh);
3895 pfx = extract_addr(tb[IFA_ADDRESS], tb[IFA_LOCAL], &peer_pfx);
3896 if (pfx == NULL)
3897 return -EINVAL;
3898
3899 ifa_flags = tb[IFA_FLAGS] ? nla_get_u32(tb[IFA_FLAGS]) : ifm->ifa_flags;
3900
3901 /* We ignore other flags so far. */
3902 ifa_flags &= IFA_F_MANAGETEMPADDR;
3903
3904 return inet6_addr_del(net, ifm->ifa_index, ifa_flags, pfx,
3905 ifm->ifa_prefixlen);
3906 }
3907
inet6_addr_modify(struct inet6_ifaddr * ifp,u32 ifa_flags,u32 prefered_lft,u32 valid_lft)3908 static int inet6_addr_modify(struct inet6_ifaddr *ifp, u32 ifa_flags,
3909 u32 prefered_lft, u32 valid_lft)
3910 {
3911 u32 flags;
3912 clock_t expires;
3913 unsigned long timeout;
3914 bool was_managetempaddr;
3915 bool had_prefixroute;
3916
3917 ASSERT_RTNL();
3918
3919 if (!valid_lft || (prefered_lft > valid_lft))
3920 return -EINVAL;
3921
3922 if (ifa_flags & IFA_F_MANAGETEMPADDR &&
3923 (ifp->flags & IFA_F_TEMPORARY || ifp->prefix_len != 64))
3924 return -EINVAL;
3925
3926 timeout = addrconf_timeout_fixup(valid_lft, HZ);
3927 if (addrconf_finite_timeout(timeout)) {
3928 expires = jiffies_to_clock_t(timeout * HZ);
3929 valid_lft = timeout;
3930 flags = RTF_EXPIRES;
3931 } else {
3932 expires = 0;
3933 flags = 0;
3934 ifa_flags |= IFA_F_PERMANENT;
3935 }
3936
3937 timeout = addrconf_timeout_fixup(prefered_lft, HZ);
3938 if (addrconf_finite_timeout(timeout)) {
3939 if (timeout == 0)
3940 ifa_flags |= IFA_F_DEPRECATED;
3941 prefered_lft = timeout;
3942 }
3943
3944 spin_lock_bh(&ifp->lock);
3945 was_managetempaddr = ifp->flags & IFA_F_MANAGETEMPADDR;
3946 had_prefixroute = ifp->flags & IFA_F_PERMANENT &&
3947 !(ifp->flags & IFA_F_NOPREFIXROUTE);
3948 ifp->flags &= ~(IFA_F_DEPRECATED | IFA_F_PERMANENT | IFA_F_NODAD |
3949 IFA_F_HOMEADDRESS | IFA_F_MANAGETEMPADDR |
3950 IFA_F_NOPREFIXROUTE);
3951 ifp->flags |= ifa_flags;
3952 ifp->tstamp = jiffies;
3953 ifp->valid_lft = valid_lft;
3954 ifp->prefered_lft = prefered_lft;
3955
3956 spin_unlock_bh(&ifp->lock);
3957 if (!(ifp->flags&IFA_F_TENTATIVE))
3958 ipv6_ifa_notify(0, ifp);
3959
3960 if (!(ifa_flags & IFA_F_NOPREFIXROUTE)) {
3961 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, ifp->idev->dev,
3962 expires, flags);
3963 } else if (had_prefixroute) {
3964 enum cleanup_prefix_rt_t action;
3965 unsigned long rt_expires;
3966
3967 write_lock_bh(&ifp->idev->lock);
3968 action = check_cleanup_prefix_route(ifp, &rt_expires);
3969 write_unlock_bh(&ifp->idev->lock);
3970
3971 if (action != CLEANUP_PREFIX_RT_NOP) {
3972 cleanup_prefix_route(ifp, rt_expires,
3973 action == CLEANUP_PREFIX_RT_DEL);
3974 }
3975 }
3976
3977 if (was_managetempaddr || ifp->flags & IFA_F_MANAGETEMPADDR) {
3978 if (was_managetempaddr && !(ifp->flags & IFA_F_MANAGETEMPADDR))
3979 valid_lft = prefered_lft = 0;
3980 manage_tempaddrs(ifp->idev, ifp, valid_lft, prefered_lft,
3981 !was_managetempaddr, jiffies);
3982 }
3983
3984 addrconf_verify_rtnl();
3985
3986 return 0;
3987 }
3988
3989 static int
inet6_rtm_newaddr(struct sk_buff * skb,struct nlmsghdr * nlh)3990 inet6_rtm_newaddr(struct sk_buff *skb, struct nlmsghdr *nlh)
3991 {
3992 struct net *net = sock_net(skb->sk);
3993 struct ifaddrmsg *ifm;
3994 struct nlattr *tb[IFA_MAX+1];
3995 struct in6_addr *pfx, *peer_pfx;
3996 struct inet6_ifaddr *ifa;
3997 struct net_device *dev;
3998 u32 valid_lft = INFINITY_LIFE_TIME, preferred_lft = INFINITY_LIFE_TIME;
3999 u32 ifa_flags;
4000 int err;
4001
4002 err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv6_policy);
4003 if (err < 0)
4004 return err;
4005
4006 ifm = nlmsg_data(nlh);
4007 pfx = extract_addr(tb[IFA_ADDRESS], tb[IFA_LOCAL], &peer_pfx);
4008 if (pfx == NULL)
4009 return -EINVAL;
4010
4011 if (tb[IFA_CACHEINFO]) {
4012 struct ifa_cacheinfo *ci;
4013
4014 ci = nla_data(tb[IFA_CACHEINFO]);
4015 valid_lft = ci->ifa_valid;
4016 preferred_lft = ci->ifa_prefered;
4017 } else {
4018 preferred_lft = INFINITY_LIFE_TIME;
4019 valid_lft = INFINITY_LIFE_TIME;
4020 }
4021
4022 dev = __dev_get_by_index(net, ifm->ifa_index);
4023 if (dev == NULL)
4024 return -ENODEV;
4025
4026 ifa_flags = tb[IFA_FLAGS] ? nla_get_u32(tb[IFA_FLAGS]) : ifm->ifa_flags;
4027
4028 /* We ignore other flags so far. */
4029 ifa_flags &= IFA_F_NODAD | IFA_F_HOMEADDRESS | IFA_F_MANAGETEMPADDR |
4030 IFA_F_NOPREFIXROUTE;
4031
4032 ifa = ipv6_get_ifaddr(net, pfx, dev, 1);
4033 if (ifa == NULL) {
4034 /*
4035 * It would be best to check for !NLM_F_CREATE here but
4036 * userspace already relies on not having to provide this.
4037 */
4038 return inet6_addr_add(net, ifm->ifa_index, pfx, peer_pfx,
4039 ifm->ifa_prefixlen, ifa_flags,
4040 preferred_lft, valid_lft);
4041 }
4042
4043 if (nlh->nlmsg_flags & NLM_F_EXCL ||
4044 !(nlh->nlmsg_flags & NLM_F_REPLACE))
4045 err = -EEXIST;
4046 else
4047 err = inet6_addr_modify(ifa, ifa_flags, preferred_lft, valid_lft);
4048
4049 in6_ifa_put(ifa);
4050
4051 return err;
4052 }
4053
put_ifaddrmsg(struct nlmsghdr * nlh,u8 prefixlen,u32 flags,u8 scope,int ifindex)4054 static void put_ifaddrmsg(struct nlmsghdr *nlh, u8 prefixlen, u32 flags,
4055 u8 scope, int ifindex)
4056 {
4057 struct ifaddrmsg *ifm;
4058
4059 ifm = nlmsg_data(nlh);
4060 ifm->ifa_family = AF_INET6;
4061 ifm->ifa_prefixlen = prefixlen;
4062 ifm->ifa_flags = flags;
4063 ifm->ifa_scope = scope;
4064 ifm->ifa_index = ifindex;
4065 }
4066
put_cacheinfo(struct sk_buff * skb,unsigned long cstamp,unsigned long tstamp,u32 preferred,u32 valid)4067 static int put_cacheinfo(struct sk_buff *skb, unsigned long cstamp,
4068 unsigned long tstamp, u32 preferred, u32 valid)
4069 {
4070 struct ifa_cacheinfo ci;
4071
4072 ci.cstamp = cstamp_delta(cstamp);
4073 ci.tstamp = cstamp_delta(tstamp);
4074 ci.ifa_prefered = preferred;
4075 ci.ifa_valid = valid;
4076
4077 return nla_put(skb, IFA_CACHEINFO, sizeof(ci), &ci);
4078 }
4079
rt_scope(int ifa_scope)4080 static inline int rt_scope(int ifa_scope)
4081 {
4082 if (ifa_scope & IFA_HOST)
4083 return RT_SCOPE_HOST;
4084 else if (ifa_scope & IFA_LINK)
4085 return RT_SCOPE_LINK;
4086 else if (ifa_scope & IFA_SITE)
4087 return RT_SCOPE_SITE;
4088 else
4089 return RT_SCOPE_UNIVERSE;
4090 }
4091
inet6_ifaddr_msgsize(void)4092 static inline int inet6_ifaddr_msgsize(void)
4093 {
4094 return NLMSG_ALIGN(sizeof(struct ifaddrmsg))
4095 + nla_total_size(16) /* IFA_LOCAL */
4096 + nla_total_size(16) /* IFA_ADDRESS */
4097 + nla_total_size(sizeof(struct ifa_cacheinfo))
4098 + nla_total_size(4) /* IFA_FLAGS */;
4099 }
4100
inet6_fill_ifaddr(struct sk_buff * skb,struct inet6_ifaddr * ifa,u32 portid,u32 seq,int event,unsigned int flags)4101 static int inet6_fill_ifaddr(struct sk_buff *skb, struct inet6_ifaddr *ifa,
4102 u32 portid, u32 seq, int event, unsigned int flags)
4103 {
4104 struct nlmsghdr *nlh;
4105 u32 preferred, valid;
4106
4107 nlh = nlmsg_put(skb, portid, seq, event, sizeof(struct ifaddrmsg), flags);
4108 if (nlh == NULL)
4109 return -EMSGSIZE;
4110
4111 put_ifaddrmsg(nlh, ifa->prefix_len, ifa->flags, rt_scope(ifa->scope),
4112 ifa->idev->dev->ifindex);
4113
4114 if (!((ifa->flags&IFA_F_PERMANENT) &&
4115 (ifa->prefered_lft == INFINITY_LIFE_TIME))) {
4116 preferred = ifa->prefered_lft;
4117 valid = ifa->valid_lft;
4118 if (preferred != INFINITY_LIFE_TIME) {
4119 long tval = (jiffies - ifa->tstamp)/HZ;
4120 if (preferred > tval)
4121 preferred -= tval;
4122 else
4123 preferred = 0;
4124 if (valid != INFINITY_LIFE_TIME) {
4125 if (valid > tval)
4126 valid -= tval;
4127 else
4128 valid = 0;
4129 }
4130 }
4131 } else {
4132 preferred = INFINITY_LIFE_TIME;
4133 valid = INFINITY_LIFE_TIME;
4134 }
4135
4136 if (!ipv6_addr_any(&ifa->peer_addr)) {
4137 if (nla_put(skb, IFA_LOCAL, 16, &ifa->addr) < 0 ||
4138 nla_put(skb, IFA_ADDRESS, 16, &ifa->peer_addr) < 0)
4139 goto error;
4140 } else
4141 if (nla_put(skb, IFA_ADDRESS, 16, &ifa->addr) < 0)
4142 goto error;
4143
4144 if (put_cacheinfo(skb, ifa->cstamp, ifa->tstamp, preferred, valid) < 0)
4145 goto error;
4146
4147 if (nla_put_u32(skb, IFA_FLAGS, ifa->flags) < 0)
4148 goto error;
4149
4150 return nlmsg_end(skb, nlh);
4151
4152 error:
4153 nlmsg_cancel(skb, nlh);
4154 return -EMSGSIZE;
4155 }
4156
inet6_fill_ifmcaddr(struct sk_buff * skb,struct ifmcaddr6 * ifmca,u32 portid,u32 seq,int event,u16 flags)4157 static int inet6_fill_ifmcaddr(struct sk_buff *skb, struct ifmcaddr6 *ifmca,
4158 u32 portid, u32 seq, int event, u16 flags)
4159 {
4160 struct nlmsghdr *nlh;
4161 u8 scope = RT_SCOPE_UNIVERSE;
4162 int ifindex = ifmca->idev->dev->ifindex;
4163
4164 if (ipv6_addr_scope(&ifmca->mca_addr) & IFA_SITE)
4165 scope = RT_SCOPE_SITE;
4166
4167 nlh = nlmsg_put(skb, portid, seq, event, sizeof(struct ifaddrmsg), flags);
4168 if (nlh == NULL)
4169 return -EMSGSIZE;
4170
4171 put_ifaddrmsg(nlh, 128, IFA_F_PERMANENT, scope, ifindex);
4172 if (nla_put(skb, IFA_MULTICAST, 16, &ifmca->mca_addr) < 0 ||
4173 put_cacheinfo(skb, ifmca->mca_cstamp, ifmca->mca_tstamp,
4174 INFINITY_LIFE_TIME, INFINITY_LIFE_TIME) < 0) {
4175 nlmsg_cancel(skb, nlh);
4176 return -EMSGSIZE;
4177 }
4178
4179 return nlmsg_end(skb, nlh);
4180 }
4181
inet6_fill_ifacaddr(struct sk_buff * skb,struct ifacaddr6 * ifaca,u32 portid,u32 seq,int event,unsigned int flags)4182 static int inet6_fill_ifacaddr(struct sk_buff *skb, struct ifacaddr6 *ifaca,
4183 u32 portid, u32 seq, int event, unsigned int flags)
4184 {
4185 struct nlmsghdr *nlh;
4186 u8 scope = RT_SCOPE_UNIVERSE;
4187 int ifindex = ifaca->aca_idev->dev->ifindex;
4188
4189 if (ipv6_addr_scope(&ifaca->aca_addr) & IFA_SITE)
4190 scope = RT_SCOPE_SITE;
4191
4192 nlh = nlmsg_put(skb, portid, seq, event, sizeof(struct ifaddrmsg), flags);
4193 if (nlh == NULL)
4194 return -EMSGSIZE;
4195
4196 put_ifaddrmsg(nlh, 128, IFA_F_PERMANENT, scope, ifindex);
4197 if (nla_put(skb, IFA_ANYCAST, 16, &ifaca->aca_addr) < 0 ||
4198 put_cacheinfo(skb, ifaca->aca_cstamp, ifaca->aca_tstamp,
4199 INFINITY_LIFE_TIME, INFINITY_LIFE_TIME) < 0) {
4200 nlmsg_cancel(skb, nlh);
4201 return -EMSGSIZE;
4202 }
4203
4204 return nlmsg_end(skb, nlh);
4205 }
4206
4207 enum addr_type_t {
4208 UNICAST_ADDR,
4209 MULTICAST_ADDR,
4210 ANYCAST_ADDR,
4211 };
4212
4213 /* called with rcu_read_lock() */
in6_dump_addrs(struct inet6_dev * idev,struct sk_buff * skb,struct netlink_callback * cb,enum addr_type_t type,int s_ip_idx,int * p_ip_idx)4214 static int in6_dump_addrs(struct inet6_dev *idev, struct sk_buff *skb,
4215 struct netlink_callback *cb, enum addr_type_t type,
4216 int s_ip_idx, int *p_ip_idx)
4217 {
4218 struct ifmcaddr6 *ifmca;
4219 struct ifacaddr6 *ifaca;
4220 int err = 1;
4221 int ip_idx = *p_ip_idx;
4222
4223 read_lock_bh(&idev->lock);
4224 switch (type) {
4225 case UNICAST_ADDR: {
4226 struct inet6_ifaddr *ifa;
4227
4228 /* unicast address incl. temp addr */
4229 list_for_each_entry(ifa, &idev->addr_list, if_list) {
4230 if (++ip_idx < s_ip_idx)
4231 continue;
4232 err = inet6_fill_ifaddr(skb, ifa,
4233 NETLINK_CB(cb->skb).portid,
4234 cb->nlh->nlmsg_seq,
4235 RTM_NEWADDR,
4236 NLM_F_MULTI);
4237 if (err <= 0)
4238 break;
4239 nl_dump_check_consistent(cb, nlmsg_hdr(skb));
4240 }
4241 break;
4242 }
4243 case MULTICAST_ADDR:
4244 /* multicast address */
4245 for (ifmca = idev->mc_list; ifmca;
4246 ifmca = ifmca->next, ip_idx++) {
4247 if (ip_idx < s_ip_idx)
4248 continue;
4249 err = inet6_fill_ifmcaddr(skb, ifmca,
4250 NETLINK_CB(cb->skb).portid,
4251 cb->nlh->nlmsg_seq,
4252 RTM_GETMULTICAST,
4253 NLM_F_MULTI);
4254 if (err <= 0)
4255 break;
4256 }
4257 break;
4258 case ANYCAST_ADDR:
4259 /* anycast address */
4260 for (ifaca = idev->ac_list; ifaca;
4261 ifaca = ifaca->aca_next, ip_idx++) {
4262 if (ip_idx < s_ip_idx)
4263 continue;
4264 err = inet6_fill_ifacaddr(skb, ifaca,
4265 NETLINK_CB(cb->skb).portid,
4266 cb->nlh->nlmsg_seq,
4267 RTM_GETANYCAST,
4268 NLM_F_MULTI);
4269 if (err <= 0)
4270 break;
4271 }
4272 break;
4273 default:
4274 break;
4275 }
4276 read_unlock_bh(&idev->lock);
4277 *p_ip_idx = ip_idx;
4278 return err;
4279 }
4280
inet6_dump_addr(struct sk_buff * skb,struct netlink_callback * cb,enum addr_type_t type)4281 static int inet6_dump_addr(struct sk_buff *skb, struct netlink_callback *cb,
4282 enum addr_type_t type)
4283 {
4284 struct net *net = sock_net(skb->sk);
4285 int h, s_h;
4286 int idx, ip_idx;
4287 int s_idx, s_ip_idx;
4288 struct net_device *dev;
4289 struct inet6_dev *idev;
4290 struct hlist_head *head;
4291
4292 s_h = cb->args[0];
4293 s_idx = idx = cb->args[1];
4294 s_ip_idx = ip_idx = cb->args[2];
4295
4296 rcu_read_lock();
4297 cb->seq = atomic_read(&net->ipv6.dev_addr_genid) ^ net->dev_base_seq;
4298 for (h = s_h; h < NETDEV_HASHENTRIES; h++, s_idx = 0) {
4299 idx = 0;
4300 head = &net->dev_index_head[h];
4301 hlist_for_each_entry_rcu(dev, head, index_hlist) {
4302 if (idx < s_idx)
4303 goto cont;
4304 if (h > s_h || idx > s_idx)
4305 s_ip_idx = 0;
4306 ip_idx = 0;
4307 idev = __in6_dev_get(dev);
4308 if (!idev)
4309 goto cont;
4310
4311 if (in6_dump_addrs(idev, skb, cb, type,
4312 s_ip_idx, &ip_idx) <= 0)
4313 goto done;
4314 cont:
4315 idx++;
4316 }
4317 }
4318 done:
4319 rcu_read_unlock();
4320 cb->args[0] = h;
4321 cb->args[1] = idx;
4322 cb->args[2] = ip_idx;
4323
4324 return skb->len;
4325 }
4326
inet6_dump_ifaddr(struct sk_buff * skb,struct netlink_callback * cb)4327 static int inet6_dump_ifaddr(struct sk_buff *skb, struct netlink_callback *cb)
4328 {
4329 enum addr_type_t type = UNICAST_ADDR;
4330
4331 return inet6_dump_addr(skb, cb, type);
4332 }
4333
inet6_dump_ifmcaddr(struct sk_buff * skb,struct netlink_callback * cb)4334 static int inet6_dump_ifmcaddr(struct sk_buff *skb, struct netlink_callback *cb)
4335 {
4336 enum addr_type_t type = MULTICAST_ADDR;
4337
4338 return inet6_dump_addr(skb, cb, type);
4339 }
4340
4341
inet6_dump_ifacaddr(struct sk_buff * skb,struct netlink_callback * cb)4342 static int inet6_dump_ifacaddr(struct sk_buff *skb, struct netlink_callback *cb)
4343 {
4344 enum addr_type_t type = ANYCAST_ADDR;
4345
4346 return inet6_dump_addr(skb, cb, type);
4347 }
4348
inet6_rtm_getaddr(struct sk_buff * in_skb,struct nlmsghdr * nlh)4349 static int inet6_rtm_getaddr(struct sk_buff *in_skb, struct nlmsghdr *nlh)
4350 {
4351 struct net *net = sock_net(in_skb->sk);
4352 struct ifaddrmsg *ifm;
4353 struct nlattr *tb[IFA_MAX+1];
4354 struct in6_addr *addr = NULL, *peer;
4355 struct net_device *dev = NULL;
4356 struct inet6_ifaddr *ifa;
4357 struct sk_buff *skb;
4358 int err;
4359
4360 err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv6_policy);
4361 if (err < 0)
4362 goto errout;
4363
4364 addr = extract_addr(tb[IFA_ADDRESS], tb[IFA_LOCAL], &peer);
4365 if (addr == NULL) {
4366 err = -EINVAL;
4367 goto errout;
4368 }
4369
4370 ifm = nlmsg_data(nlh);
4371 if (ifm->ifa_index)
4372 dev = __dev_get_by_index(net, ifm->ifa_index);
4373
4374 ifa = ipv6_get_ifaddr(net, addr, dev, 1);
4375 if (!ifa) {
4376 err = -EADDRNOTAVAIL;
4377 goto errout;
4378 }
4379
4380 skb = nlmsg_new(inet6_ifaddr_msgsize(), GFP_KERNEL);
4381 if (!skb) {
4382 err = -ENOBUFS;
4383 goto errout_ifa;
4384 }
4385
4386 err = inet6_fill_ifaddr(skb, ifa, NETLINK_CB(in_skb).portid,
4387 nlh->nlmsg_seq, RTM_NEWADDR, 0);
4388 if (err < 0) {
4389 /* -EMSGSIZE implies BUG in inet6_ifaddr_msgsize() */
4390 WARN_ON(err == -EMSGSIZE);
4391 kfree_skb(skb);
4392 goto errout_ifa;
4393 }
4394 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
4395 errout_ifa:
4396 in6_ifa_put(ifa);
4397 errout:
4398 return err;
4399 }
4400
inet6_ifa_notify(int event,struct inet6_ifaddr * ifa)4401 static void inet6_ifa_notify(int event, struct inet6_ifaddr *ifa)
4402 {
4403 struct sk_buff *skb;
4404 struct net *net = dev_net(ifa->idev->dev);
4405 int err = -ENOBUFS;
4406
4407 skb = nlmsg_new(inet6_ifaddr_msgsize(), GFP_ATOMIC);
4408 if (skb == NULL)
4409 goto errout;
4410
4411 err = inet6_fill_ifaddr(skb, ifa, 0, 0, event, 0);
4412 if (err < 0) {
4413 /* -EMSGSIZE implies BUG in inet6_ifaddr_msgsize() */
4414 WARN_ON(err == -EMSGSIZE);
4415 kfree_skb(skb);
4416 goto errout;
4417 }
4418 rtnl_notify(skb, net, 0, RTNLGRP_IPV6_IFADDR, NULL, GFP_ATOMIC);
4419 return;
4420 errout:
4421 if (err < 0)
4422 rtnl_set_sk_err(net, RTNLGRP_IPV6_IFADDR, err);
4423 }
4424
ipv6_store_devconf(struct ipv6_devconf * cnf,__s32 * array,int bytes)4425 static inline void ipv6_store_devconf(struct ipv6_devconf *cnf,
4426 __s32 *array, int bytes)
4427 {
4428 BUG_ON(bytes < (DEVCONF_MAX * 4));
4429
4430 memset(array, 0, bytes);
4431 array[DEVCONF_FORWARDING] = cnf->forwarding;
4432 array[DEVCONF_HOPLIMIT] = cnf->hop_limit;
4433 array[DEVCONF_MTU6] = cnf->mtu6;
4434 array[DEVCONF_ACCEPT_RA] = cnf->accept_ra;
4435 array[DEVCONF_ACCEPT_REDIRECTS] = cnf->accept_redirects;
4436 array[DEVCONF_AUTOCONF] = cnf->autoconf;
4437 array[DEVCONF_DAD_TRANSMITS] = cnf->dad_transmits;
4438 array[DEVCONF_RTR_SOLICITS] = cnf->rtr_solicits;
4439 array[DEVCONF_RTR_SOLICIT_INTERVAL] =
4440 jiffies_to_msecs(cnf->rtr_solicit_interval);
4441 array[DEVCONF_RTR_SOLICIT_MAX_INTERVAL] =
4442 jiffies_to_msecs(cnf->rtr_solicit_max_interval);
4443 array[DEVCONF_RTR_SOLICIT_DELAY] =
4444 jiffies_to_msecs(cnf->rtr_solicit_delay);
4445 array[DEVCONF_FORCE_MLD_VERSION] = cnf->force_mld_version;
4446 array[DEVCONF_MLDV1_UNSOLICITED_REPORT_INTERVAL] =
4447 jiffies_to_msecs(cnf->mldv1_unsolicited_report_interval);
4448 array[DEVCONF_MLDV2_UNSOLICITED_REPORT_INTERVAL] =
4449 jiffies_to_msecs(cnf->mldv2_unsolicited_report_interval);
4450 array[DEVCONF_USE_TEMPADDR] = cnf->use_tempaddr;
4451 array[DEVCONF_TEMP_VALID_LFT] = cnf->temp_valid_lft;
4452 array[DEVCONF_TEMP_PREFERED_LFT] = cnf->temp_prefered_lft;
4453 array[DEVCONF_REGEN_MAX_RETRY] = cnf->regen_max_retry;
4454 array[DEVCONF_MAX_DESYNC_FACTOR] = cnf->max_desync_factor;
4455 array[DEVCONF_MAX_ADDRESSES] = cnf->max_addresses;
4456 array[DEVCONF_ACCEPT_RA_DEFRTR] = cnf->accept_ra_defrtr;
4457 array[DEVCONF_ACCEPT_RA_PINFO] = cnf->accept_ra_pinfo;
4458 #ifdef CONFIG_IPV6_ROUTER_PREF
4459 array[DEVCONF_ACCEPT_RA_RTR_PREF] = cnf->accept_ra_rtr_pref;
4460 array[DEVCONF_RTR_PROBE_INTERVAL] =
4461 jiffies_to_msecs(cnf->rtr_probe_interval);
4462 #ifdef CONFIG_IPV6_ROUTE_INFO
4463 array[DEVCONF_ACCEPT_RA_RT_INFO_MIN_PLEN] = cnf->accept_ra_rt_info_min_plen;
4464 array[DEVCONF_ACCEPT_RA_RT_INFO_MAX_PLEN] = cnf->accept_ra_rt_info_max_plen;
4465 #endif
4466 #endif
4467 array[DEVCONF_ACCEPT_RA_RT_TABLE] = cnf->accept_ra_rt_table;
4468 array[DEVCONF_PROXY_NDP] = cnf->proxy_ndp;
4469 array[DEVCONF_ACCEPT_SOURCE_ROUTE] = cnf->accept_source_route;
4470 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
4471 array[DEVCONF_OPTIMISTIC_DAD] = cnf->optimistic_dad;
4472 array[DEVCONF_USE_OPTIMISTIC] = cnf->use_optimistic;
4473 #endif
4474 #ifdef CONFIG_IPV6_MROUTE
4475 array[DEVCONF_MC_FORWARDING] = cnf->mc_forwarding;
4476 #endif
4477 array[DEVCONF_DISABLE_IPV6] = cnf->disable_ipv6;
4478 array[DEVCONF_ACCEPT_DAD] = cnf->accept_dad;
4479 array[DEVCONF_FORCE_TLLAO] = cnf->force_tllao;
4480 array[DEVCONF_NDISC_NOTIFY] = cnf->ndisc_notify;
4481 array[DEVCONF_SUPPRESS_FRAG_NDISC] = cnf->suppress_frag_ndisc;
4482 array[DEVCONF_ACCEPT_RA_FROM_LOCAL] = cnf->accept_ra_from_local;
4483 array[DEVCONF_USE_OIF_ADDRS_ONLY] = cnf->use_oif_addrs_only;
4484 }
4485
inet6_ifla6_size(void)4486 static inline size_t inet6_ifla6_size(void)
4487 {
4488 return nla_total_size(4) /* IFLA_INET6_FLAGS */
4489 + nla_total_size(sizeof(struct ifla_cacheinfo))
4490 + nla_total_size(DEVCONF_MAX * 4) /* IFLA_INET6_CONF */
4491 + nla_total_size(IPSTATS_MIB_MAX * 8) /* IFLA_INET6_STATS */
4492 + nla_total_size(ICMP6_MIB_MAX * 8) /* IFLA_INET6_ICMP6STATS */
4493 + nla_total_size(sizeof(struct in6_addr)); /* IFLA_INET6_TOKEN */
4494 }
4495
inet6_if_nlmsg_size(void)4496 static inline size_t inet6_if_nlmsg_size(void)
4497 {
4498 return NLMSG_ALIGN(sizeof(struct ifinfomsg))
4499 + nla_total_size(IFNAMSIZ) /* IFLA_IFNAME */
4500 + nla_total_size(MAX_ADDR_LEN) /* IFLA_ADDRESS */
4501 + nla_total_size(4) /* IFLA_MTU */
4502 + nla_total_size(4) /* IFLA_LINK */
4503 + nla_total_size(inet6_ifla6_size()); /* IFLA_PROTINFO */
4504 }
4505
__snmp6_fill_statsdev(u64 * stats,atomic_long_t * mib,int items,int bytes)4506 static inline void __snmp6_fill_statsdev(u64 *stats, atomic_long_t *mib,
4507 int items, int bytes)
4508 {
4509 int i;
4510 int pad = bytes - sizeof(u64) * items;
4511 BUG_ON(pad < 0);
4512
4513 /* Use put_unaligned() because stats may not be aligned for u64. */
4514 put_unaligned(items, &stats[0]);
4515 for (i = 1; i < items; i++)
4516 put_unaligned(atomic_long_read(&mib[i]), &stats[i]);
4517
4518 memset(&stats[items], 0, pad);
4519 }
4520
__snmp6_fill_stats64(u64 * stats,void __percpu * mib,int items,int bytes,size_t syncpoff)4521 static inline void __snmp6_fill_stats64(u64 *stats, void __percpu *mib,
4522 int items, int bytes, size_t syncpoff)
4523 {
4524 int i;
4525 int pad = bytes - sizeof(u64) * items;
4526 BUG_ON(pad < 0);
4527
4528 /* Use put_unaligned() because stats may not be aligned for u64. */
4529 put_unaligned(items, &stats[0]);
4530 for (i = 1; i < items; i++)
4531 put_unaligned(snmp_fold_field64(mib, i, syncpoff), &stats[i]);
4532
4533 memset(&stats[items], 0, pad);
4534 }
4535
snmp6_fill_stats(u64 * stats,struct inet6_dev * idev,int attrtype,int bytes)4536 static void snmp6_fill_stats(u64 *stats, struct inet6_dev *idev, int attrtype,
4537 int bytes)
4538 {
4539 switch (attrtype) {
4540 case IFLA_INET6_STATS:
4541 __snmp6_fill_stats64(stats, idev->stats.ipv6,
4542 IPSTATS_MIB_MAX, bytes, offsetof(struct ipstats_mib, syncp));
4543 break;
4544 case IFLA_INET6_ICMP6STATS:
4545 __snmp6_fill_statsdev(stats, idev->stats.icmpv6dev->mibs, ICMP6_MIB_MAX, bytes);
4546 break;
4547 }
4548 }
4549
inet6_fill_ifla6_attrs(struct sk_buff * skb,struct inet6_dev * idev)4550 static int inet6_fill_ifla6_attrs(struct sk_buff *skb, struct inet6_dev *idev)
4551 {
4552 struct nlattr *nla;
4553 struct ifla_cacheinfo ci;
4554
4555 if (nla_put_u32(skb, IFLA_INET6_FLAGS, idev->if_flags))
4556 goto nla_put_failure;
4557 ci.max_reasm_len = IPV6_MAXPLEN;
4558 ci.tstamp = cstamp_delta(idev->tstamp);
4559 ci.reachable_time = jiffies_to_msecs(idev->nd_parms->reachable_time);
4560 ci.retrans_time = jiffies_to_msecs(NEIGH_VAR(idev->nd_parms, RETRANS_TIME));
4561 if (nla_put(skb, IFLA_INET6_CACHEINFO, sizeof(ci), &ci))
4562 goto nla_put_failure;
4563 nla = nla_reserve(skb, IFLA_INET6_CONF, DEVCONF_MAX * sizeof(s32));
4564 if (nla == NULL)
4565 goto nla_put_failure;
4566 ipv6_store_devconf(&idev->cnf, nla_data(nla), nla_len(nla));
4567
4568 /* XXX - MC not implemented */
4569
4570 nla = nla_reserve(skb, IFLA_INET6_STATS, IPSTATS_MIB_MAX * sizeof(u64));
4571 if (nla == NULL)
4572 goto nla_put_failure;
4573 snmp6_fill_stats(nla_data(nla), idev, IFLA_INET6_STATS, nla_len(nla));
4574
4575 nla = nla_reserve(skb, IFLA_INET6_ICMP6STATS, ICMP6_MIB_MAX * sizeof(u64));
4576 if (nla == NULL)
4577 goto nla_put_failure;
4578 snmp6_fill_stats(nla_data(nla), idev, IFLA_INET6_ICMP6STATS, nla_len(nla));
4579
4580 nla = nla_reserve(skb, IFLA_INET6_TOKEN, sizeof(struct in6_addr));
4581 if (nla == NULL)
4582 goto nla_put_failure;
4583
4584 if (nla_put_u8(skb, IFLA_INET6_ADDR_GEN_MODE, idev->addr_gen_mode))
4585 goto nla_put_failure;
4586
4587 read_lock_bh(&idev->lock);
4588 memcpy(nla_data(nla), idev->token.s6_addr, nla_len(nla));
4589 read_unlock_bh(&idev->lock);
4590
4591 return 0;
4592
4593 nla_put_failure:
4594 return -EMSGSIZE;
4595 }
4596
inet6_get_link_af_size(const struct net_device * dev)4597 static size_t inet6_get_link_af_size(const struct net_device *dev)
4598 {
4599 if (!__in6_dev_get(dev))
4600 return 0;
4601
4602 return inet6_ifla6_size();
4603 }
4604
inet6_fill_link_af(struct sk_buff * skb,const struct net_device * dev)4605 static int inet6_fill_link_af(struct sk_buff *skb, const struct net_device *dev)
4606 {
4607 struct inet6_dev *idev = __in6_dev_get(dev);
4608
4609 if (!idev)
4610 return -ENODATA;
4611
4612 if (inet6_fill_ifla6_attrs(skb, idev) < 0)
4613 return -EMSGSIZE;
4614
4615 return 0;
4616 }
4617
inet6_set_iftoken(struct inet6_dev * idev,struct in6_addr * token)4618 static int inet6_set_iftoken(struct inet6_dev *idev, struct in6_addr *token)
4619 {
4620 struct inet6_ifaddr *ifp;
4621 struct net_device *dev = idev->dev;
4622 bool update_rs = false;
4623 struct in6_addr ll_addr;
4624
4625 ASSERT_RTNL();
4626
4627 if (token == NULL)
4628 return -EINVAL;
4629 if (ipv6_addr_any(token))
4630 return -EINVAL;
4631 if (dev->flags & (IFF_LOOPBACK | IFF_NOARP))
4632 return -EINVAL;
4633 if (!ipv6_accept_ra(idev))
4634 return -EINVAL;
4635 if (idev->cnf.rtr_solicits == 0)
4636 return -EINVAL;
4637
4638 write_lock_bh(&idev->lock);
4639
4640 BUILD_BUG_ON(sizeof(token->s6_addr) != 16);
4641 memcpy(idev->token.s6_addr + 8, token->s6_addr + 8, 8);
4642
4643 write_unlock_bh(&idev->lock);
4644
4645 if (!idev->dead && (idev->if_flags & IF_READY) &&
4646 !ipv6_get_lladdr(dev, &ll_addr, IFA_F_TENTATIVE |
4647 IFA_F_OPTIMISTIC)) {
4648
4649 /* If we're not ready, then normal ifup will take care
4650 * of this. Otherwise, we need to request our rs here.
4651 */
4652 ndisc_send_rs(dev, &ll_addr, &in6addr_linklocal_allrouters);
4653 update_rs = true;
4654 }
4655
4656 write_lock_bh(&idev->lock);
4657
4658 if (update_rs) {
4659 idev->if_flags |= IF_RS_SENT;
4660 idev->rs_interval = rfc3315_s14_backoff_init(
4661 idev->cnf.rtr_solicit_interval);
4662 idev->rs_probes = 1;
4663 addrconf_mod_rs_timer(idev, idev->rs_interval);
4664 }
4665
4666 /* Well, that's kinda nasty ... */
4667 list_for_each_entry(ifp, &idev->addr_list, if_list) {
4668 spin_lock(&ifp->lock);
4669 if (ifp->tokenized) {
4670 ifp->valid_lft = 0;
4671 ifp->prefered_lft = 0;
4672 }
4673 spin_unlock(&ifp->lock);
4674 }
4675
4676 write_unlock_bh(&idev->lock);
4677 inet6_ifinfo_notify(RTM_NEWLINK, idev);
4678 addrconf_verify_rtnl();
4679 return 0;
4680 }
4681
4682 static const struct nla_policy inet6_af_policy[IFLA_INET6_MAX + 1] = {
4683 [IFLA_INET6_ADDR_GEN_MODE] = { .type = NLA_U8 },
4684 [IFLA_INET6_TOKEN] = { .len = sizeof(struct in6_addr) },
4685 };
4686
inet6_validate_link_af(const struct net_device * dev,const struct nlattr * nla)4687 static int inet6_validate_link_af(const struct net_device *dev,
4688 const struct nlattr *nla)
4689 {
4690 struct nlattr *tb[IFLA_INET6_MAX + 1];
4691
4692 if (dev && !__in6_dev_get(dev))
4693 return -EAFNOSUPPORT;
4694
4695 return nla_parse_nested(tb, IFLA_INET6_MAX, nla, inet6_af_policy);
4696 }
4697
inet6_set_link_af(struct net_device * dev,const struct nlattr * nla)4698 static int inet6_set_link_af(struct net_device *dev, const struct nlattr *nla)
4699 {
4700 int err = -EINVAL;
4701 struct inet6_dev *idev = __in6_dev_get(dev);
4702 struct nlattr *tb[IFLA_INET6_MAX + 1];
4703
4704 if (!idev)
4705 return -EAFNOSUPPORT;
4706
4707 if (nla_parse_nested(tb, IFLA_INET6_MAX, nla, NULL) < 0)
4708 BUG();
4709
4710 if (tb[IFLA_INET6_TOKEN]) {
4711 err = inet6_set_iftoken(idev, nla_data(tb[IFLA_INET6_TOKEN]));
4712 if (err)
4713 return err;
4714 }
4715
4716 if (tb[IFLA_INET6_ADDR_GEN_MODE]) {
4717 u8 mode = nla_get_u8(tb[IFLA_INET6_ADDR_GEN_MODE]);
4718
4719 if (mode != IN6_ADDR_GEN_MODE_EUI64 &&
4720 mode != IN6_ADDR_GEN_MODE_NONE)
4721 return -EINVAL;
4722 idev->addr_gen_mode = mode;
4723 err = 0;
4724 }
4725
4726 return err;
4727 }
4728
inet6_fill_ifinfo(struct sk_buff * skb,struct inet6_dev * idev,u32 portid,u32 seq,int event,unsigned int flags)4729 static int inet6_fill_ifinfo(struct sk_buff *skb, struct inet6_dev *idev,
4730 u32 portid, u32 seq, int event, unsigned int flags)
4731 {
4732 struct net_device *dev = idev->dev;
4733 struct ifinfomsg *hdr;
4734 struct nlmsghdr *nlh;
4735 void *protoinfo;
4736
4737 nlh = nlmsg_put(skb, portid, seq, event, sizeof(*hdr), flags);
4738 if (nlh == NULL)
4739 return -EMSGSIZE;
4740
4741 hdr = nlmsg_data(nlh);
4742 hdr->ifi_family = AF_INET6;
4743 hdr->__ifi_pad = 0;
4744 hdr->ifi_type = dev->type;
4745 hdr->ifi_index = dev->ifindex;
4746 hdr->ifi_flags = dev_get_flags(dev);
4747 hdr->ifi_change = 0;
4748
4749 if (nla_put_string(skb, IFLA_IFNAME, dev->name) ||
4750 (dev->addr_len &&
4751 nla_put(skb, IFLA_ADDRESS, dev->addr_len, dev->dev_addr)) ||
4752 nla_put_u32(skb, IFLA_MTU, dev->mtu) ||
4753 (dev->ifindex != dev->iflink &&
4754 nla_put_u32(skb, IFLA_LINK, dev->iflink)))
4755 goto nla_put_failure;
4756 protoinfo = nla_nest_start(skb, IFLA_PROTINFO);
4757 if (protoinfo == NULL)
4758 goto nla_put_failure;
4759
4760 if (inet6_fill_ifla6_attrs(skb, idev) < 0)
4761 goto nla_put_failure;
4762
4763 nla_nest_end(skb, protoinfo);
4764 return nlmsg_end(skb, nlh);
4765
4766 nla_put_failure:
4767 nlmsg_cancel(skb, nlh);
4768 return -EMSGSIZE;
4769 }
4770
inet6_dump_ifinfo(struct sk_buff * skb,struct netlink_callback * cb)4771 static int inet6_dump_ifinfo(struct sk_buff *skb, struct netlink_callback *cb)
4772 {
4773 struct net *net = sock_net(skb->sk);
4774 int h, s_h;
4775 int idx = 0, s_idx;
4776 struct net_device *dev;
4777 struct inet6_dev *idev;
4778 struct hlist_head *head;
4779
4780 s_h = cb->args[0];
4781 s_idx = cb->args[1];
4782
4783 rcu_read_lock();
4784 for (h = s_h; h < NETDEV_HASHENTRIES; h++, s_idx = 0) {
4785 idx = 0;
4786 head = &net->dev_index_head[h];
4787 hlist_for_each_entry_rcu(dev, head, index_hlist) {
4788 if (idx < s_idx)
4789 goto cont;
4790 idev = __in6_dev_get(dev);
4791 if (!idev)
4792 goto cont;
4793 if (inet6_fill_ifinfo(skb, idev,
4794 NETLINK_CB(cb->skb).portid,
4795 cb->nlh->nlmsg_seq,
4796 RTM_NEWLINK, NLM_F_MULTI) <= 0)
4797 goto out;
4798 cont:
4799 idx++;
4800 }
4801 }
4802 out:
4803 rcu_read_unlock();
4804 cb->args[1] = idx;
4805 cb->args[0] = h;
4806
4807 return skb->len;
4808 }
4809
inet6_ifinfo_notify(int event,struct inet6_dev * idev)4810 void inet6_ifinfo_notify(int event, struct inet6_dev *idev)
4811 {
4812 struct sk_buff *skb;
4813 struct net *net = dev_net(idev->dev);
4814 int err = -ENOBUFS;
4815
4816 skb = nlmsg_new(inet6_if_nlmsg_size(), GFP_ATOMIC);
4817 if (skb == NULL)
4818 goto errout;
4819
4820 err = inet6_fill_ifinfo(skb, idev, 0, 0, event, 0);
4821 if (err < 0) {
4822 /* -EMSGSIZE implies BUG in inet6_if_nlmsg_size() */
4823 WARN_ON(err == -EMSGSIZE);
4824 kfree_skb(skb);
4825 goto errout;
4826 }
4827 rtnl_notify(skb, net, 0, RTNLGRP_IPV6_IFINFO, NULL, GFP_ATOMIC);
4828 return;
4829 errout:
4830 if (err < 0)
4831 rtnl_set_sk_err(net, RTNLGRP_IPV6_IFINFO, err);
4832 }
4833
inet6_prefix_nlmsg_size(void)4834 static inline size_t inet6_prefix_nlmsg_size(void)
4835 {
4836 return NLMSG_ALIGN(sizeof(struct prefixmsg))
4837 + nla_total_size(sizeof(struct in6_addr))
4838 + nla_total_size(sizeof(struct prefix_cacheinfo));
4839 }
4840
inet6_fill_prefix(struct sk_buff * skb,struct inet6_dev * idev,struct prefix_info * pinfo,u32 portid,u32 seq,int event,unsigned int flags)4841 static int inet6_fill_prefix(struct sk_buff *skb, struct inet6_dev *idev,
4842 struct prefix_info *pinfo, u32 portid, u32 seq,
4843 int event, unsigned int flags)
4844 {
4845 struct prefixmsg *pmsg;
4846 struct nlmsghdr *nlh;
4847 struct prefix_cacheinfo ci;
4848
4849 nlh = nlmsg_put(skb, portid, seq, event, sizeof(*pmsg), flags);
4850 if (nlh == NULL)
4851 return -EMSGSIZE;
4852
4853 pmsg = nlmsg_data(nlh);
4854 pmsg->prefix_family = AF_INET6;
4855 pmsg->prefix_pad1 = 0;
4856 pmsg->prefix_pad2 = 0;
4857 pmsg->prefix_ifindex = idev->dev->ifindex;
4858 pmsg->prefix_len = pinfo->prefix_len;
4859 pmsg->prefix_type = pinfo->type;
4860 pmsg->prefix_pad3 = 0;
4861 pmsg->prefix_flags = 0;
4862 if (pinfo->onlink)
4863 pmsg->prefix_flags |= IF_PREFIX_ONLINK;
4864 if (pinfo->autoconf)
4865 pmsg->prefix_flags |= IF_PREFIX_AUTOCONF;
4866
4867 if (nla_put(skb, PREFIX_ADDRESS, sizeof(pinfo->prefix), &pinfo->prefix))
4868 goto nla_put_failure;
4869 ci.preferred_time = ntohl(pinfo->prefered);
4870 ci.valid_time = ntohl(pinfo->valid);
4871 if (nla_put(skb, PREFIX_CACHEINFO, sizeof(ci), &ci))
4872 goto nla_put_failure;
4873 return nlmsg_end(skb, nlh);
4874
4875 nla_put_failure:
4876 nlmsg_cancel(skb, nlh);
4877 return -EMSGSIZE;
4878 }
4879
inet6_prefix_notify(int event,struct inet6_dev * idev,struct prefix_info * pinfo)4880 static void inet6_prefix_notify(int event, struct inet6_dev *idev,
4881 struct prefix_info *pinfo)
4882 {
4883 struct sk_buff *skb;
4884 struct net *net = dev_net(idev->dev);
4885 int err = -ENOBUFS;
4886
4887 skb = nlmsg_new(inet6_prefix_nlmsg_size(), GFP_ATOMIC);
4888 if (skb == NULL)
4889 goto errout;
4890
4891 err = inet6_fill_prefix(skb, idev, pinfo, 0, 0, event, 0);
4892 if (err < 0) {
4893 /* -EMSGSIZE implies BUG in inet6_prefix_nlmsg_size() */
4894 WARN_ON(err == -EMSGSIZE);
4895 kfree_skb(skb);
4896 goto errout;
4897 }
4898 rtnl_notify(skb, net, 0, RTNLGRP_IPV6_PREFIX, NULL, GFP_ATOMIC);
4899 return;
4900 errout:
4901 if (err < 0)
4902 rtnl_set_sk_err(net, RTNLGRP_IPV6_PREFIX, err);
4903 }
4904
__ipv6_ifa_notify(int event,struct inet6_ifaddr * ifp)4905 static void __ipv6_ifa_notify(int event, struct inet6_ifaddr *ifp)
4906 {
4907 struct net *net = dev_net(ifp->idev->dev);
4908
4909 if (event)
4910 ASSERT_RTNL();
4911
4912 inet6_ifa_notify(event ? : RTM_NEWADDR, ifp);
4913
4914 switch (event) {
4915 case RTM_NEWADDR:
4916 /*
4917 * If the address was optimistic
4918 * we inserted the route at the start of
4919 * our DAD process, so we don't need
4920 * to do it again
4921 */
4922 if (!(ifp->rt->rt6i_node))
4923 ip6_ins_rt(ifp->rt);
4924 if (ifp->idev->cnf.forwarding)
4925 addrconf_join_anycast(ifp);
4926 if (!ipv6_addr_any(&ifp->peer_addr))
4927 addrconf_prefix_route(&ifp->peer_addr, 128,
4928 ifp->idev->dev, 0, 0);
4929 break;
4930 case RTM_DELADDR:
4931 if (ifp->idev->cnf.forwarding)
4932 addrconf_leave_anycast(ifp);
4933 addrconf_leave_solict(ifp->idev, &ifp->addr);
4934 if (!ipv6_addr_any(&ifp->peer_addr)) {
4935 struct rt6_info *rt;
4936
4937 rt = addrconf_get_prefix_route(&ifp->peer_addr, 128,
4938 ifp->idev->dev, 0, 0);
4939 if (rt && ip6_del_rt(rt))
4940 dst_free(&rt->dst);
4941 }
4942 dst_hold(&ifp->rt->dst);
4943
4944 if (ip6_del_rt(ifp->rt))
4945 dst_free(&ifp->rt->dst);
4946
4947 rt_genid_bump_ipv6(net);
4948 break;
4949 }
4950 atomic_inc(&net->ipv6.dev_addr_genid);
4951 }
4952
ipv6_ifa_notify(int event,struct inet6_ifaddr * ifp)4953 static void ipv6_ifa_notify(int event, struct inet6_ifaddr *ifp)
4954 {
4955 rcu_read_lock_bh();
4956 if (likely(ifp->idev->dead == 0))
4957 __ipv6_ifa_notify(event, ifp);
4958 rcu_read_unlock_bh();
4959 }
4960
4961 #ifdef CONFIG_SYSCTL
4962
4963 static
addrconf_sysctl_forward(struct ctl_table * ctl,int write,void __user * buffer,size_t * lenp,loff_t * ppos)4964 int addrconf_sysctl_forward(struct ctl_table *ctl, int write,
4965 void __user *buffer, size_t *lenp, loff_t *ppos)
4966 {
4967 int *valp = ctl->data;
4968 int val = *valp;
4969 loff_t pos = *ppos;
4970 struct ctl_table lctl;
4971 int ret;
4972
4973 /*
4974 * ctl->data points to idev->cnf.forwarding, we should
4975 * not modify it until we get the rtnl lock.
4976 */
4977 lctl = *ctl;
4978 lctl.data = &val;
4979
4980 ret = proc_dointvec(&lctl, write, buffer, lenp, ppos);
4981
4982 if (write)
4983 ret = addrconf_fixup_forwarding(ctl, valp, val);
4984 if (ret)
4985 *ppos = pos;
4986 return ret;
4987 }
4988
4989 static
addrconf_sysctl_mtu(struct ctl_table * ctl,int write,void __user * buffer,size_t * lenp,loff_t * ppos)4990 int addrconf_sysctl_mtu(struct ctl_table *ctl, int write,
4991 void __user *buffer, size_t *lenp, loff_t *ppos)
4992 {
4993 struct inet6_dev *idev = ctl->extra1;
4994 int min_mtu = IPV6_MIN_MTU;
4995 struct ctl_table lctl;
4996
4997 lctl = *ctl;
4998 lctl.extra1 = &min_mtu;
4999 lctl.extra2 = idev ? &idev->dev->mtu : NULL;
5000
5001 return proc_dointvec_minmax(&lctl, write, buffer, lenp, ppos);
5002 }
5003
dev_disable_change(struct inet6_dev * idev)5004 static void dev_disable_change(struct inet6_dev *idev)
5005 {
5006 struct netdev_notifier_info info;
5007
5008 if (!idev || !idev->dev)
5009 return;
5010
5011 netdev_notifier_info_init(&info, idev->dev);
5012 if (idev->cnf.disable_ipv6)
5013 addrconf_notify(NULL, NETDEV_DOWN, &info);
5014 else
5015 addrconf_notify(NULL, NETDEV_UP, &info);
5016 }
5017
addrconf_disable_change(struct net * net,__s32 newf)5018 static void addrconf_disable_change(struct net *net, __s32 newf)
5019 {
5020 struct net_device *dev;
5021 struct inet6_dev *idev;
5022
5023 rcu_read_lock();
5024 for_each_netdev_rcu(net, dev) {
5025 idev = __in6_dev_get(dev);
5026 if (idev) {
5027 int changed = (!idev->cnf.disable_ipv6) ^ (!newf);
5028 idev->cnf.disable_ipv6 = newf;
5029 if (changed)
5030 dev_disable_change(idev);
5031 }
5032 }
5033 rcu_read_unlock();
5034 }
5035
addrconf_disable_ipv6(struct ctl_table * table,int * p,int newf)5036 static int addrconf_disable_ipv6(struct ctl_table *table, int *p, int newf)
5037 {
5038 struct net *net;
5039 int old;
5040
5041 if (!rtnl_trylock())
5042 return restart_syscall();
5043
5044 net = (struct net *)table->extra2;
5045 old = *p;
5046 *p = newf;
5047
5048 if (p == &net->ipv6.devconf_dflt->disable_ipv6) {
5049 rtnl_unlock();
5050 return 0;
5051 }
5052
5053 if (p == &net->ipv6.devconf_all->disable_ipv6) {
5054 net->ipv6.devconf_dflt->disable_ipv6 = newf;
5055 addrconf_disable_change(net, newf);
5056 } else if ((!newf) ^ (!old))
5057 dev_disable_change((struct inet6_dev *)table->extra1);
5058
5059 rtnl_unlock();
5060 return 0;
5061 }
5062
5063 static
addrconf_sysctl_disable(struct ctl_table * ctl,int write,void __user * buffer,size_t * lenp,loff_t * ppos)5064 int addrconf_sysctl_disable(struct ctl_table *ctl, int write,
5065 void __user *buffer, size_t *lenp, loff_t *ppos)
5066 {
5067 int *valp = ctl->data;
5068 int val = *valp;
5069 loff_t pos = *ppos;
5070 struct ctl_table lctl;
5071 int ret;
5072
5073 /*
5074 * ctl->data points to idev->cnf.disable_ipv6, we should
5075 * not modify it until we get the rtnl lock.
5076 */
5077 lctl = *ctl;
5078 lctl.data = &val;
5079
5080 ret = proc_dointvec(&lctl, write, buffer, lenp, ppos);
5081
5082 if (write)
5083 ret = addrconf_disable_ipv6(ctl, valp, val);
5084 if (ret)
5085 *ppos = pos;
5086 return ret;
5087 }
5088
5089 static
addrconf_sysctl_proxy_ndp(struct ctl_table * ctl,int write,void __user * buffer,size_t * lenp,loff_t * ppos)5090 int addrconf_sysctl_proxy_ndp(struct ctl_table *ctl, int write,
5091 void __user *buffer, size_t *lenp, loff_t *ppos)
5092 {
5093 int *valp = ctl->data;
5094 int ret;
5095 int old, new;
5096
5097 old = *valp;
5098 ret = proc_dointvec(ctl, write, buffer, lenp, ppos);
5099 new = *valp;
5100
5101 if (write && old != new) {
5102 struct net *net = ctl->extra2;
5103
5104 if (!rtnl_trylock())
5105 return restart_syscall();
5106
5107 if (valp == &net->ipv6.devconf_dflt->proxy_ndp)
5108 inet6_netconf_notify_devconf(net, NETCONFA_PROXY_NEIGH,
5109 NETCONFA_IFINDEX_DEFAULT,
5110 net->ipv6.devconf_dflt);
5111 else if (valp == &net->ipv6.devconf_all->proxy_ndp)
5112 inet6_netconf_notify_devconf(net, NETCONFA_PROXY_NEIGH,
5113 NETCONFA_IFINDEX_ALL,
5114 net->ipv6.devconf_all);
5115 else {
5116 struct inet6_dev *idev = ctl->extra1;
5117
5118 inet6_netconf_notify_devconf(net, NETCONFA_PROXY_NEIGH,
5119 idev->dev->ifindex,
5120 &idev->cnf);
5121 }
5122 rtnl_unlock();
5123 }
5124
5125 return ret;
5126 }
5127
5128
5129 static struct addrconf_sysctl_table
5130 {
5131 struct ctl_table_header *sysctl_header;
5132 struct ctl_table addrconf_vars[DEVCONF_MAX+1];
5133 } addrconf_sysctl __read_mostly = {
5134 .sysctl_header = NULL,
5135 .addrconf_vars = {
5136 {
5137 .procname = "forwarding",
5138 .data = &ipv6_devconf.forwarding,
5139 .maxlen = sizeof(int),
5140 .mode = 0644,
5141 .proc_handler = addrconf_sysctl_forward,
5142 },
5143 {
5144 .procname = "hop_limit",
5145 .data = &ipv6_devconf.hop_limit,
5146 .maxlen = sizeof(int),
5147 .mode = 0644,
5148 .proc_handler = proc_dointvec,
5149 },
5150 {
5151 .procname = "mtu",
5152 .data = &ipv6_devconf.mtu6,
5153 .maxlen = sizeof(int),
5154 .mode = 0644,
5155 .proc_handler = addrconf_sysctl_mtu,
5156 },
5157 {
5158 .procname = "accept_ra",
5159 .data = &ipv6_devconf.accept_ra,
5160 .maxlen = sizeof(int),
5161 .mode = 0644,
5162 .proc_handler = proc_dointvec,
5163 },
5164 {
5165 .procname = "accept_redirects",
5166 .data = &ipv6_devconf.accept_redirects,
5167 .maxlen = sizeof(int),
5168 .mode = 0644,
5169 .proc_handler = proc_dointvec,
5170 },
5171 {
5172 .procname = "autoconf",
5173 .data = &ipv6_devconf.autoconf,
5174 .maxlen = sizeof(int),
5175 .mode = 0644,
5176 .proc_handler = proc_dointvec,
5177 },
5178 {
5179 .procname = "dad_transmits",
5180 .data = &ipv6_devconf.dad_transmits,
5181 .maxlen = sizeof(int),
5182 .mode = 0644,
5183 .proc_handler = proc_dointvec,
5184 },
5185 {
5186 .procname = "router_solicitations",
5187 .data = &ipv6_devconf.rtr_solicits,
5188 .maxlen = sizeof(int),
5189 .mode = 0644,
5190 .proc_handler = proc_dointvec,
5191 },
5192 {
5193 .procname = "router_solicitation_interval",
5194 .data = &ipv6_devconf.rtr_solicit_interval,
5195 .maxlen = sizeof(int),
5196 .mode = 0644,
5197 .proc_handler = proc_dointvec_jiffies,
5198 },
5199 {
5200 .procname = "router_solicitation_max_interval",
5201 .data = &ipv6_devconf.rtr_solicit_max_interval,
5202 .maxlen = sizeof(int),
5203 .mode = 0644,
5204 .proc_handler = proc_dointvec_jiffies,
5205 },
5206 {
5207 .procname = "router_solicitation_delay",
5208 .data = &ipv6_devconf.rtr_solicit_delay,
5209 .maxlen = sizeof(int),
5210 .mode = 0644,
5211 .proc_handler = proc_dointvec_jiffies,
5212 },
5213 {
5214 .procname = "force_mld_version",
5215 .data = &ipv6_devconf.force_mld_version,
5216 .maxlen = sizeof(int),
5217 .mode = 0644,
5218 .proc_handler = proc_dointvec,
5219 },
5220 {
5221 .procname = "mldv1_unsolicited_report_interval",
5222 .data =
5223 &ipv6_devconf.mldv1_unsolicited_report_interval,
5224 .maxlen = sizeof(int),
5225 .mode = 0644,
5226 .proc_handler = proc_dointvec_ms_jiffies,
5227 },
5228 {
5229 .procname = "mldv2_unsolicited_report_interval",
5230 .data =
5231 &ipv6_devconf.mldv2_unsolicited_report_interval,
5232 .maxlen = sizeof(int),
5233 .mode = 0644,
5234 .proc_handler = proc_dointvec_ms_jiffies,
5235 },
5236 {
5237 .procname = "use_tempaddr",
5238 .data = &ipv6_devconf.use_tempaddr,
5239 .maxlen = sizeof(int),
5240 .mode = 0644,
5241 .proc_handler = proc_dointvec,
5242 },
5243 {
5244 .procname = "temp_valid_lft",
5245 .data = &ipv6_devconf.temp_valid_lft,
5246 .maxlen = sizeof(int),
5247 .mode = 0644,
5248 .proc_handler = proc_dointvec,
5249 },
5250 {
5251 .procname = "temp_prefered_lft",
5252 .data = &ipv6_devconf.temp_prefered_lft,
5253 .maxlen = sizeof(int),
5254 .mode = 0644,
5255 .proc_handler = proc_dointvec,
5256 },
5257 {
5258 .procname = "regen_max_retry",
5259 .data = &ipv6_devconf.regen_max_retry,
5260 .maxlen = sizeof(int),
5261 .mode = 0644,
5262 .proc_handler = proc_dointvec,
5263 },
5264 {
5265 .procname = "max_desync_factor",
5266 .data = &ipv6_devconf.max_desync_factor,
5267 .maxlen = sizeof(int),
5268 .mode = 0644,
5269 .proc_handler = proc_dointvec,
5270 },
5271 {
5272 .procname = "max_addresses",
5273 .data = &ipv6_devconf.max_addresses,
5274 .maxlen = sizeof(int),
5275 .mode = 0644,
5276 .proc_handler = proc_dointvec,
5277 },
5278 {
5279 .procname = "accept_ra_defrtr",
5280 .data = &ipv6_devconf.accept_ra_defrtr,
5281 .maxlen = sizeof(int),
5282 .mode = 0644,
5283 .proc_handler = proc_dointvec,
5284 },
5285 {
5286 .procname = "accept_ra_pinfo",
5287 .data = &ipv6_devconf.accept_ra_pinfo,
5288 .maxlen = sizeof(int),
5289 .mode = 0644,
5290 .proc_handler = proc_dointvec,
5291 },
5292 #ifdef CONFIG_IPV6_ROUTER_PREF
5293 {
5294 .procname = "accept_ra_rtr_pref",
5295 .data = &ipv6_devconf.accept_ra_rtr_pref,
5296 .maxlen = sizeof(int),
5297 .mode = 0644,
5298 .proc_handler = proc_dointvec,
5299 },
5300 {
5301 .procname = "router_probe_interval",
5302 .data = &ipv6_devconf.rtr_probe_interval,
5303 .maxlen = sizeof(int),
5304 .mode = 0644,
5305 .proc_handler = proc_dointvec_jiffies,
5306 },
5307 #ifdef CONFIG_IPV6_ROUTE_INFO
5308 {
5309 .procname = "accept_ra_rt_info_min_plen",
5310 .data = &ipv6_devconf.accept_ra_rt_info_min_plen,
5311 .maxlen = sizeof(int),
5312 .mode = 0644,
5313 .proc_handler = proc_dointvec,
5314 },
5315 {
5316 .procname = "accept_ra_rt_info_max_plen",
5317 .data = &ipv6_devconf.accept_ra_rt_info_max_plen,
5318 .maxlen = sizeof(int),
5319 .mode = 0644,
5320 .proc_handler = proc_dointvec,
5321 },
5322 #endif
5323 #endif
5324 {
5325 .procname = "accept_ra_rt_table",
5326 .data = &ipv6_devconf.accept_ra_rt_table,
5327 .maxlen = sizeof(int),
5328 .mode = 0644,
5329 .proc_handler = proc_dointvec,
5330 },
5331 {
5332 .procname = "proxy_ndp",
5333 .data = &ipv6_devconf.proxy_ndp,
5334 .maxlen = sizeof(int),
5335 .mode = 0644,
5336 .proc_handler = addrconf_sysctl_proxy_ndp,
5337 },
5338 {
5339 .procname = "accept_source_route",
5340 .data = &ipv6_devconf.accept_source_route,
5341 .maxlen = sizeof(int),
5342 .mode = 0644,
5343 .proc_handler = proc_dointvec,
5344 },
5345 #ifdef CONFIG_IPV6_OPTIMISTIC_DAD
5346 {
5347 .procname = "optimistic_dad",
5348 .data = &ipv6_devconf.optimistic_dad,
5349 .maxlen = sizeof(int),
5350 .mode = 0644,
5351 .proc_handler = proc_dointvec,
5352
5353 },
5354 {
5355 .procname = "use_optimistic",
5356 .data = &ipv6_devconf.use_optimistic,
5357 .maxlen = sizeof(int),
5358 .mode = 0644,
5359 .proc_handler = proc_dointvec,
5360
5361 },
5362 #endif
5363 #ifdef CONFIG_IPV6_MROUTE
5364 {
5365 .procname = "mc_forwarding",
5366 .data = &ipv6_devconf.mc_forwarding,
5367 .maxlen = sizeof(int),
5368 .mode = 0444,
5369 .proc_handler = proc_dointvec,
5370 },
5371 #endif
5372 {
5373 .procname = "disable_ipv6",
5374 .data = &ipv6_devconf.disable_ipv6,
5375 .maxlen = sizeof(int),
5376 .mode = 0644,
5377 .proc_handler = addrconf_sysctl_disable,
5378 },
5379 {
5380 .procname = "accept_dad",
5381 .data = &ipv6_devconf.accept_dad,
5382 .maxlen = sizeof(int),
5383 .mode = 0644,
5384 .proc_handler = proc_dointvec,
5385 },
5386 {
5387 .procname = "force_tllao",
5388 .data = &ipv6_devconf.force_tllao,
5389 .maxlen = sizeof(int),
5390 .mode = 0644,
5391 .proc_handler = proc_dointvec
5392 },
5393 {
5394 .procname = "ndisc_notify",
5395 .data = &ipv6_devconf.ndisc_notify,
5396 .maxlen = sizeof(int),
5397 .mode = 0644,
5398 .proc_handler = proc_dointvec
5399 },
5400 {
5401 .procname = "suppress_frag_ndisc",
5402 .data = &ipv6_devconf.suppress_frag_ndisc,
5403 .maxlen = sizeof(int),
5404 .mode = 0644,
5405 .proc_handler = proc_dointvec
5406 },
5407 {
5408 .procname = "accept_ra_from_local",
5409 .data = &ipv6_devconf.accept_ra_from_local,
5410 .maxlen = sizeof(int),
5411 .mode = 0644,
5412 .proc_handler = proc_dointvec,
5413 },
5414 {
5415 .procname = "use_oif_addrs_only",
5416 .data = &ipv6_devconf.use_oif_addrs_only,
5417 .maxlen = sizeof(int),
5418 .mode = 0644,
5419 .proc_handler = proc_dointvec,
5420
5421 },
5422 {
5423 /* sentinel */
5424 }
5425 },
5426 };
5427
__addrconf_sysctl_register(struct net * net,char * dev_name,struct inet6_dev * idev,struct ipv6_devconf * p)5428 static int __addrconf_sysctl_register(struct net *net, char *dev_name,
5429 struct inet6_dev *idev, struct ipv6_devconf *p)
5430 {
5431 int i;
5432 struct addrconf_sysctl_table *t;
5433 char path[sizeof("net/ipv6/conf/") + IFNAMSIZ];
5434
5435 t = kmemdup(&addrconf_sysctl, sizeof(*t), GFP_KERNEL);
5436 if (t == NULL)
5437 goto out;
5438
5439 for (i = 0; t->addrconf_vars[i].data; i++) {
5440 t->addrconf_vars[i].data += (char *)p - (char *)&ipv6_devconf;
5441 t->addrconf_vars[i].extra1 = idev; /* embedded; no ref */
5442 t->addrconf_vars[i].extra2 = net;
5443 }
5444
5445 snprintf(path, sizeof(path), "net/ipv6/conf/%s", dev_name);
5446
5447 t->sysctl_header = register_net_sysctl(net, path, t->addrconf_vars);
5448 if (t->sysctl_header == NULL)
5449 goto free;
5450
5451 p->sysctl = t;
5452 return 0;
5453
5454 free:
5455 kfree(t);
5456 out:
5457 return -ENOBUFS;
5458 }
5459
__addrconf_sysctl_unregister(struct ipv6_devconf * p)5460 static void __addrconf_sysctl_unregister(struct ipv6_devconf *p)
5461 {
5462 struct addrconf_sysctl_table *t;
5463
5464 if (p->sysctl == NULL)
5465 return;
5466
5467 t = p->sysctl;
5468 p->sysctl = NULL;
5469 unregister_net_sysctl_table(t->sysctl_header);
5470 kfree(t);
5471 }
5472
addrconf_sysctl_register(struct inet6_dev * idev)5473 static int addrconf_sysctl_register(struct inet6_dev *idev)
5474 {
5475 int err;
5476
5477 if (!sysctl_dev_name_is_allowed(idev->dev->name))
5478 return -EINVAL;
5479
5480 err = neigh_sysctl_register(idev->dev, idev->nd_parms,
5481 &ndisc_ifinfo_sysctl_change);
5482 if (err)
5483 return err;
5484 err = __addrconf_sysctl_register(dev_net(idev->dev), idev->dev->name,
5485 idev, &idev->cnf);
5486 if (err)
5487 neigh_sysctl_unregister(idev->nd_parms);
5488
5489 return err;
5490 }
5491
addrconf_sysctl_unregister(struct inet6_dev * idev)5492 static void addrconf_sysctl_unregister(struct inet6_dev *idev)
5493 {
5494 __addrconf_sysctl_unregister(&idev->cnf);
5495 neigh_sysctl_unregister(idev->nd_parms);
5496 }
5497
5498
5499 #endif
5500
addrconf_init_net(struct net * net)5501 static int __net_init addrconf_init_net(struct net *net)
5502 {
5503 int err = -ENOMEM;
5504 struct ipv6_devconf *all, *dflt;
5505
5506 all = kmemdup(&ipv6_devconf, sizeof(ipv6_devconf), GFP_KERNEL);
5507 if (all == NULL)
5508 goto err_alloc_all;
5509
5510 dflt = kmemdup(&ipv6_devconf_dflt, sizeof(ipv6_devconf_dflt), GFP_KERNEL);
5511 if (dflt == NULL)
5512 goto err_alloc_dflt;
5513
5514 /* these will be inherited by all namespaces */
5515 dflt->autoconf = ipv6_defaults.autoconf;
5516 dflt->disable_ipv6 = ipv6_defaults.disable_ipv6;
5517
5518 net->ipv6.devconf_all = all;
5519 net->ipv6.devconf_dflt = dflt;
5520
5521 #ifdef CONFIG_SYSCTL
5522 err = __addrconf_sysctl_register(net, "all", NULL, all);
5523 if (err < 0)
5524 goto err_reg_all;
5525
5526 err = __addrconf_sysctl_register(net, "default", NULL, dflt);
5527 if (err < 0)
5528 goto err_reg_dflt;
5529 #endif
5530 return 0;
5531
5532 #ifdef CONFIG_SYSCTL
5533 err_reg_dflt:
5534 __addrconf_sysctl_unregister(all);
5535 err_reg_all:
5536 kfree(dflt);
5537 #endif
5538 err_alloc_dflt:
5539 kfree(all);
5540 err_alloc_all:
5541 return err;
5542 }
5543
addrconf_exit_net(struct net * net)5544 static void __net_exit addrconf_exit_net(struct net *net)
5545 {
5546 #ifdef CONFIG_SYSCTL
5547 __addrconf_sysctl_unregister(net->ipv6.devconf_dflt);
5548 __addrconf_sysctl_unregister(net->ipv6.devconf_all);
5549 #endif
5550 if (!net_eq(net, &init_net)) {
5551 kfree(net->ipv6.devconf_dflt);
5552 kfree(net->ipv6.devconf_all);
5553 }
5554 }
5555
5556 static struct pernet_operations addrconf_ops = {
5557 .init = addrconf_init_net,
5558 .exit = addrconf_exit_net,
5559 };
5560
5561 static struct rtnl_af_ops inet6_ops = {
5562 .family = AF_INET6,
5563 .fill_link_af = inet6_fill_link_af,
5564 .get_link_af_size = inet6_get_link_af_size,
5565 .validate_link_af = inet6_validate_link_af,
5566 .set_link_af = inet6_set_link_af,
5567 };
5568
5569 /*
5570 * Init / cleanup code
5571 */
5572
addrconf_init(void)5573 int __init addrconf_init(void)
5574 {
5575 struct inet6_dev *idev;
5576 int i, err;
5577
5578 err = ipv6_addr_label_init();
5579 if (err < 0) {
5580 pr_crit("%s: cannot initialize default policy table: %d\n",
5581 __func__, err);
5582 goto out;
5583 }
5584
5585 err = register_pernet_subsys(&addrconf_ops);
5586 if (err < 0)
5587 goto out_addrlabel;
5588
5589 addrconf_wq = create_workqueue("ipv6_addrconf");
5590 if (!addrconf_wq) {
5591 err = -ENOMEM;
5592 goto out_nowq;
5593 }
5594
5595 /* The addrconf netdev notifier requires that loopback_dev
5596 * has it's ipv6 private information allocated and setup
5597 * before it can bring up and give link-local addresses
5598 * to other devices which are up.
5599 *
5600 * Unfortunately, loopback_dev is not necessarily the first
5601 * entry in the global dev_base list of net devices. In fact,
5602 * it is likely to be the very last entry on that list.
5603 * So this causes the notifier registry below to try and
5604 * give link-local addresses to all devices besides loopback_dev
5605 * first, then loopback_dev, which cases all the non-loopback_dev
5606 * devices to fail to get a link-local address.
5607 *
5608 * So, as a temporary fix, allocate the ipv6 structure for
5609 * loopback_dev first by hand.
5610 * Longer term, all of the dependencies ipv6 has upon the loopback
5611 * device and it being up should be removed.
5612 */
5613 rtnl_lock();
5614 idev = ipv6_add_dev(init_net.loopback_dev);
5615 rtnl_unlock();
5616 if (IS_ERR(idev)) {
5617 err = PTR_ERR(idev);
5618 goto errlo;
5619 }
5620
5621 ip6_route_init_special_entries();
5622
5623 for (i = 0; i < IN6_ADDR_HSIZE; i++)
5624 INIT_HLIST_HEAD(&inet6_addr_lst[i]);
5625
5626 register_netdevice_notifier(&ipv6_dev_notf);
5627
5628 addrconf_verify();
5629
5630 rtnl_af_register(&inet6_ops);
5631
5632 err = __rtnl_register(PF_INET6, RTM_GETLINK, NULL, inet6_dump_ifinfo,
5633 NULL);
5634 if (err < 0)
5635 goto errout;
5636
5637 /* Only the first call to __rtnl_register can fail */
5638 __rtnl_register(PF_INET6, RTM_NEWADDR, inet6_rtm_newaddr, NULL, NULL);
5639 __rtnl_register(PF_INET6, RTM_DELADDR, inet6_rtm_deladdr, NULL, NULL);
5640 __rtnl_register(PF_INET6, RTM_GETADDR, inet6_rtm_getaddr,
5641 inet6_dump_ifaddr, NULL);
5642 __rtnl_register(PF_INET6, RTM_GETMULTICAST, NULL,
5643 inet6_dump_ifmcaddr, NULL);
5644 __rtnl_register(PF_INET6, RTM_GETANYCAST, NULL,
5645 inet6_dump_ifacaddr, NULL);
5646 __rtnl_register(PF_INET6, RTM_GETNETCONF, inet6_netconf_get_devconf,
5647 inet6_netconf_dump_devconf, NULL);
5648
5649 ipv6_addr_label_rtnl_register();
5650
5651 return 0;
5652 errout:
5653 rtnl_af_unregister(&inet6_ops);
5654 unregister_netdevice_notifier(&ipv6_dev_notf);
5655 errlo:
5656 destroy_workqueue(addrconf_wq);
5657 out_nowq:
5658 unregister_pernet_subsys(&addrconf_ops);
5659 out_addrlabel:
5660 ipv6_addr_label_cleanup();
5661 out:
5662 return err;
5663 }
5664
addrconf_cleanup(void)5665 void addrconf_cleanup(void)
5666 {
5667 struct net_device *dev;
5668 int i;
5669
5670 unregister_netdevice_notifier(&ipv6_dev_notf);
5671 unregister_pernet_subsys(&addrconf_ops);
5672 ipv6_addr_label_cleanup();
5673
5674 rtnl_lock();
5675
5676 __rtnl_af_unregister(&inet6_ops);
5677
5678 /* clean dev list */
5679 for_each_netdev(&init_net, dev) {
5680 if (__in6_dev_get(dev) == NULL)
5681 continue;
5682 addrconf_ifdown(dev, 1);
5683 }
5684 addrconf_ifdown(init_net.loopback_dev, 2);
5685
5686 /*
5687 * Check hash table.
5688 */
5689 spin_lock_bh(&addrconf_hash_lock);
5690 for (i = 0; i < IN6_ADDR_HSIZE; i++)
5691 WARN_ON(!hlist_empty(&inet6_addr_lst[i]));
5692 spin_unlock_bh(&addrconf_hash_lock);
5693 cancel_delayed_work(&addr_chk_work);
5694 rtnl_unlock();
5695
5696 destroy_workqueue(addrconf_wq);
5697 }
5698