1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19#include <linux/uaccess.h>
20#include <linux/types.h>
21#include <linux/sched.h>
22#include <linux/errno.h>
23#include <linux/mm.h>
24#include <linux/kernel.h>
25#include <linux/fcntl.h>
26#include <linux/stat.h>
27#include <linux/socket.h>
28#include <linux/inet.h>
29#include <linux/netdevice.h>
30#include <linux/inetdevice.h>
31#include <linux/proc_fs.h>
32#include <linux/seq_file.h>
33#include <linux/init.h>
34#include <linux/compat.h>
35#include <linux/rhashtable.h>
36#include <net/protocol.h>
37#include <linux/skbuff.h>
38#include <net/raw.h>
39#include <linux/notifier.h>
40#include <linux/if_arp.h>
41#include <net/checksum.h>
42#include <net/netlink.h>
43#include <net/fib_rules.h>
44
45#include <net/ipv6.h>
46#include <net/ip6_route.h>
47#include <linux/mroute6.h>
48#include <linux/pim.h>
49#include <net/addrconf.h>
50#include <linux/netfilter_ipv6.h>
51#include <linux/export.h>
52#include <net/ip6_checksum.h>
53#include <linux/netconf.h>
54
55struct ip6mr_rule {
56 struct fib_rule common;
57};
58
59struct ip6mr_result {
60 struct mr_table *mrt;
61};
62
63
64
65
66
67static DEFINE_RWLOCK(mrt_lock);
68
69
70
71
72static DEFINE_SPINLOCK(mfc_unres_lock);
73
74
75
76
77
78
79
80
81
82static struct kmem_cache *mrt_cachep __read_mostly;
83
84static struct mr_table *ip6mr_new_table(struct net *net, u32 id);
85static void ip6mr_free_table(struct mr_table *mrt);
86
87static void ip6_mr_forward(struct net *net, struct mr_table *mrt,
88 struct sk_buff *skb, struct mfc6_cache *cache);
89static int ip6mr_cache_report(struct mr_table *mrt, struct sk_buff *pkt,
90 mifi_t mifi, int assert);
91static void mr6_netlink_event(struct mr_table *mrt, struct mfc6_cache *mfc,
92 int cmd);
93static void mrt6msg_netlink_event(struct mr_table *mrt, struct sk_buff *pkt);
94static int ip6mr_rtm_dumproute(struct sk_buff *skb,
95 struct netlink_callback *cb);
96static void mroute_clean_tables(struct mr_table *mrt, bool all);
97static void ipmr_expire_process(struct timer_list *t);
98
99#ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES
100#define ip6mr_for_each_table(mrt, net) \
101 list_for_each_entry_rcu(mrt, &net->ipv6.mr6_tables, list)
102
103static struct mr_table *ip6mr_mr_table_iter(struct net *net,
104 struct mr_table *mrt)
105{
106 struct mr_table *ret;
107
108 if (!mrt)
109 ret = list_entry_rcu(net->ipv6.mr6_tables.next,
110 struct mr_table, list);
111 else
112 ret = list_entry_rcu(mrt->list.next,
113 struct mr_table, list);
114
115 if (&ret->list == &net->ipv6.mr6_tables)
116 return NULL;
117 return ret;
118}
119
120static struct mr_table *ip6mr_get_table(struct net *net, u32 id)
121{
122 struct mr_table *mrt;
123
124 ip6mr_for_each_table(mrt, net) {
125 if (mrt->id == id)
126 return mrt;
127 }
128 return NULL;
129}
130
131static int ip6mr_fib_lookup(struct net *net, struct flowi6 *flp6,
132 struct mr_table **mrt)
133{
134 int err;
135 struct ip6mr_result res;
136 struct fib_lookup_arg arg = {
137 .result = &res,
138 .flags = FIB_LOOKUP_NOREF,
139 };
140
141 err = fib_rules_lookup(net->ipv6.mr6_rules_ops,
142 flowi6_to_flowi(flp6), 0, &arg);
143 if (err < 0)
144 return err;
145 *mrt = res.mrt;
146 return 0;
147}
148
149static int ip6mr_rule_action(struct fib_rule *rule, struct flowi *flp,
150 int flags, struct fib_lookup_arg *arg)
151{
152 struct ip6mr_result *res = arg->result;
153 struct mr_table *mrt;
154
155 switch (rule->action) {
156 case FR_ACT_TO_TBL:
157 break;
158 case FR_ACT_UNREACHABLE:
159 return -ENETUNREACH;
160 case FR_ACT_PROHIBIT:
161 return -EACCES;
162 case FR_ACT_BLACKHOLE:
163 default:
164 return -EINVAL;
165 }
166
167 mrt = ip6mr_get_table(rule->fr_net, rule->table);
168 if (!mrt)
169 return -EAGAIN;
170 res->mrt = mrt;
171 return 0;
172}
173
174static int ip6mr_rule_match(struct fib_rule *rule, struct flowi *flp, int flags)
175{
176 return 1;
177}
178
179static const struct nla_policy ip6mr_rule_policy[FRA_MAX + 1] = {
180 FRA_GENERIC_POLICY,
181};
182
183static int ip6mr_rule_configure(struct fib_rule *rule, struct sk_buff *skb,
184 struct fib_rule_hdr *frh, struct nlattr **tb,
185 struct netlink_ext_ack *extack)
186{
187 return 0;
188}
189
190static int ip6mr_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh,
191 struct nlattr **tb)
192{
193 return 1;
194}
195
196static int ip6mr_rule_fill(struct fib_rule *rule, struct sk_buff *skb,
197 struct fib_rule_hdr *frh)
198{
199 frh->dst_len = 0;
200 frh->src_len = 0;
201 frh->tos = 0;
202 return 0;
203}
204
205static const struct fib_rules_ops __net_initconst ip6mr_rules_ops_template = {
206 .family = RTNL_FAMILY_IP6MR,
207 .rule_size = sizeof(struct ip6mr_rule),
208 .addr_size = sizeof(struct in6_addr),
209 .action = ip6mr_rule_action,
210 .match = ip6mr_rule_match,
211 .configure = ip6mr_rule_configure,
212 .compare = ip6mr_rule_compare,
213 .fill = ip6mr_rule_fill,
214 .nlgroup = RTNLGRP_IPV6_RULE,
215 .policy = ip6mr_rule_policy,
216 .owner = THIS_MODULE,
217};
218
219static int __net_init ip6mr_rules_init(struct net *net)
220{
221 struct fib_rules_ops *ops;
222 struct mr_table *mrt;
223 int err;
224
225 ops = fib_rules_register(&ip6mr_rules_ops_template, net);
226 if (IS_ERR(ops))
227 return PTR_ERR(ops);
228
229 INIT_LIST_HEAD(&net->ipv6.mr6_tables);
230
231 mrt = ip6mr_new_table(net, RT6_TABLE_DFLT);
232 if (IS_ERR(mrt)) {
233 err = PTR_ERR(mrt);
234 goto err1;
235 }
236
237 err = fib_default_rule_add(ops, 0x7fff, RT6_TABLE_DFLT, 0);
238 if (err < 0)
239 goto err2;
240
241 net->ipv6.mr6_rules_ops = ops;
242 return 0;
243
244err2:
245 ip6mr_free_table(mrt);
246err1:
247 fib_rules_unregister(ops);
248 return err;
249}
250
251static void __net_exit ip6mr_rules_exit(struct net *net)
252{
253 struct mr_table *mrt, *next;
254
255 rtnl_lock();
256 list_for_each_entry_safe(mrt, next, &net->ipv6.mr6_tables, list) {
257 list_del(&mrt->list);
258 ip6mr_free_table(mrt);
259 }
260 fib_rules_unregister(net->ipv6.mr6_rules_ops);
261 rtnl_unlock();
262}
263
264static int ip6mr_rules_dump(struct net *net, struct notifier_block *nb)
265{
266 return fib_rules_dump(net, nb, RTNL_FAMILY_IP6MR);
267}
268
269static unsigned int ip6mr_rules_seq_read(struct net *net)
270{
271 return fib_rules_seq_read(net, RTNL_FAMILY_IP6MR);
272}
273
274bool ip6mr_rule_default(const struct fib_rule *rule)
275{
276 return fib_rule_matchall(rule) && rule->action == FR_ACT_TO_TBL &&
277 rule->table == RT6_TABLE_DFLT && !rule->l3mdev;
278}
279EXPORT_SYMBOL(ip6mr_rule_default);
280#else
281#define ip6mr_for_each_table(mrt, net) \
282 for (mrt = net->ipv6.mrt6; mrt; mrt = NULL)
283
284static struct mr_table *ip6mr_mr_table_iter(struct net *net,
285 struct mr_table *mrt)
286{
287 if (!mrt)
288 return net->ipv6.mrt6;
289 return NULL;
290}
291
292static struct mr_table *ip6mr_get_table(struct net *net, u32 id)
293{
294 return net->ipv6.mrt6;
295}
296
297static int ip6mr_fib_lookup(struct net *net, struct flowi6 *flp6,
298 struct mr_table **mrt)
299{
300 *mrt = net->ipv6.mrt6;
301 return 0;
302}
303
304static int __net_init ip6mr_rules_init(struct net *net)
305{
306 struct mr_table *mrt;
307
308 mrt = ip6mr_new_table(net, RT6_TABLE_DFLT);
309 if (IS_ERR(mrt))
310 return PTR_ERR(mrt);
311 net->ipv6.mrt6 = mrt;
312 return 0;
313}
314
315static void __net_exit ip6mr_rules_exit(struct net *net)
316{
317 rtnl_lock();
318 ip6mr_free_table(net->ipv6.mrt6);
319 net->ipv6.mrt6 = NULL;
320 rtnl_unlock();
321}
322
323static int ip6mr_rules_dump(struct net *net, struct notifier_block *nb)
324{
325 return 0;
326}
327
328static unsigned int ip6mr_rules_seq_read(struct net *net)
329{
330 return 0;
331}
332#endif
333
334static int ip6mr_hash_cmp(struct rhashtable_compare_arg *arg,
335 const void *ptr)
336{
337 const struct mfc6_cache_cmp_arg *cmparg = arg->key;
338 struct mfc6_cache *c = (struct mfc6_cache *)ptr;
339
340 return !ipv6_addr_equal(&c->mf6c_mcastgrp, &cmparg->mf6c_mcastgrp) ||
341 !ipv6_addr_equal(&c->mf6c_origin, &cmparg->mf6c_origin);
342}
343
344static const struct rhashtable_params ip6mr_rht_params = {
345 .head_offset = offsetof(struct mr_mfc, mnode),
346 .key_offset = offsetof(struct mfc6_cache, cmparg),
347 .key_len = sizeof(struct mfc6_cache_cmp_arg),
348 .nelem_hint = 3,
349 .locks_mul = 1,
350 .obj_cmpfn = ip6mr_hash_cmp,
351 .automatic_shrinking = true,
352};
353
354static void ip6mr_new_table_set(struct mr_table *mrt,
355 struct net *net)
356{
357#ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES
358 list_add_tail_rcu(&mrt->list, &net->ipv6.mr6_tables);
359#endif
360}
361
362static struct mfc6_cache_cmp_arg ip6mr_mr_table_ops_cmparg_any = {
363 .mf6c_origin = IN6ADDR_ANY_INIT,
364 .mf6c_mcastgrp = IN6ADDR_ANY_INIT,
365};
366
367static struct mr_table_ops ip6mr_mr_table_ops = {
368 .rht_params = &ip6mr_rht_params,
369 .cmparg_any = &ip6mr_mr_table_ops_cmparg_any,
370};
371
372static struct mr_table *ip6mr_new_table(struct net *net, u32 id)
373{
374 struct mr_table *mrt;
375
376 mrt = ip6mr_get_table(net, id);
377 if (mrt)
378 return mrt;
379
380 return mr_table_alloc(net, id, &ip6mr_mr_table_ops,
381 ipmr_expire_process, ip6mr_new_table_set);
382}
383
384static void ip6mr_free_table(struct mr_table *mrt)
385{
386 del_timer_sync(&mrt->ipmr_expire_timer);
387 mroute_clean_tables(mrt, true);
388 rhltable_destroy(&mrt->mfc_hash);
389 kfree(mrt);
390}
391
392#ifdef CONFIG_PROC_FS
393
394
395
396
397static void *ip6mr_vif_seq_start(struct seq_file *seq, loff_t *pos)
398 __acquires(mrt_lock)
399{
400 struct mr_vif_iter *iter = seq->private;
401 struct net *net = seq_file_net(seq);
402 struct mr_table *mrt;
403
404 mrt = ip6mr_get_table(net, RT6_TABLE_DFLT);
405 if (!mrt)
406 return ERR_PTR(-ENOENT);
407
408 iter->mrt = mrt;
409
410 read_lock(&mrt_lock);
411 return mr_vif_seq_start(seq, pos);
412}
413
414static void ip6mr_vif_seq_stop(struct seq_file *seq, void *v)
415 __releases(mrt_lock)
416{
417 read_unlock(&mrt_lock);
418}
419
420static int ip6mr_vif_seq_show(struct seq_file *seq, void *v)
421{
422 struct mr_vif_iter *iter = seq->private;
423 struct mr_table *mrt = iter->mrt;
424
425 if (v == SEQ_START_TOKEN) {
426 seq_puts(seq,
427 "Interface BytesIn PktsIn BytesOut PktsOut Flags\n");
428 } else {
429 const struct vif_device *vif = v;
430 const char *name = vif->dev ? vif->dev->name : "none";
431
432 seq_printf(seq,
433 "%2td %-10s %8ld %7ld %8ld %7ld %05X\n",
434 vif - mrt->vif_table,
435 name, vif->bytes_in, vif->pkt_in,
436 vif->bytes_out, vif->pkt_out,
437 vif->flags);
438 }
439 return 0;
440}
441
442static const struct seq_operations ip6mr_vif_seq_ops = {
443 .start = ip6mr_vif_seq_start,
444 .next = mr_vif_seq_next,
445 .stop = ip6mr_vif_seq_stop,
446 .show = ip6mr_vif_seq_show,
447};
448
449static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos)
450{
451 struct net *net = seq_file_net(seq);
452 struct mr_table *mrt;
453
454 mrt = ip6mr_get_table(net, RT6_TABLE_DFLT);
455 if (!mrt)
456 return ERR_PTR(-ENOENT);
457
458 return mr_mfc_seq_start(seq, pos, mrt, &mfc_unres_lock);
459}
460
461static int ipmr_mfc_seq_show(struct seq_file *seq, void *v)
462{
463 int n;
464
465 if (v == SEQ_START_TOKEN) {
466 seq_puts(seq,
467 "Group "
468 "Origin "
469 "Iif Pkts Bytes Wrong Oifs\n");
470 } else {
471 const struct mfc6_cache *mfc = v;
472 const struct mr_mfc_iter *it = seq->private;
473 struct mr_table *mrt = it->mrt;
474
475 seq_printf(seq, "%pI6 %pI6 %-3hd",
476 &mfc->mf6c_mcastgrp, &mfc->mf6c_origin,
477 mfc->_c.mfc_parent);
478
479 if (it->cache != &mrt->mfc_unres_queue) {
480 seq_printf(seq, " %8lu %8lu %8lu",
481 mfc->_c.mfc_un.res.pkt,
482 mfc->_c.mfc_un.res.bytes,
483 mfc->_c.mfc_un.res.wrong_if);
484 for (n = mfc->_c.mfc_un.res.minvif;
485 n < mfc->_c.mfc_un.res.maxvif; n++) {
486 if (VIF_EXISTS(mrt, n) &&
487 mfc->_c.mfc_un.res.ttls[n] < 255)
488 seq_printf(seq,
489 " %2d:%-3d", n,
490 mfc->_c.mfc_un.res.ttls[n]);
491 }
492 } else {
493
494
495
496 seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul);
497 }
498 seq_putc(seq, '\n');
499 }
500 return 0;
501}
502
503static const struct seq_operations ipmr_mfc_seq_ops = {
504 .start = ipmr_mfc_seq_start,
505 .next = mr_mfc_seq_next,
506 .stop = mr_mfc_seq_stop,
507 .show = ipmr_mfc_seq_show,
508};
509#endif
510
511#ifdef CONFIG_IPV6_PIMSM_V2
512
513static int pim6_rcv(struct sk_buff *skb)
514{
515 struct pimreghdr *pim;
516 struct ipv6hdr *encap;
517 struct net_device *reg_dev = NULL;
518 struct net *net = dev_net(skb->dev);
519 struct mr_table *mrt;
520 struct flowi6 fl6 = {
521 .flowi6_iif = skb->dev->ifindex,
522 .flowi6_mark = skb->mark,
523 };
524 int reg_vif_num;
525
526 if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(*encap)))
527 goto drop;
528
529 pim = (struct pimreghdr *)skb_transport_header(skb);
530 if (pim->type != ((PIM_VERSION << 4) | PIM_TYPE_REGISTER) ||
531 (pim->flags & PIM_NULL_REGISTER) ||
532 (csum_ipv6_magic(&ipv6_hdr(skb)->saddr, &ipv6_hdr(skb)->daddr,
533 sizeof(*pim), IPPROTO_PIM,
534 csum_partial((void *)pim, sizeof(*pim), 0)) &&
535 csum_fold(skb_checksum(skb, 0, skb->len, 0))))
536 goto drop;
537
538
539 encap = (struct ipv6hdr *)(skb_transport_header(skb) +
540 sizeof(*pim));
541
542 if (!ipv6_addr_is_multicast(&encap->daddr) ||
543 encap->payload_len == 0 ||
544 ntohs(encap->payload_len) + sizeof(*pim) > skb->len)
545 goto drop;
546
547 if (ip6mr_fib_lookup(net, &fl6, &mrt) < 0)
548 goto drop;
549 reg_vif_num = mrt->mroute_reg_vif_num;
550
551 read_lock(&mrt_lock);
552 if (reg_vif_num >= 0)
553 reg_dev = mrt->vif_table[reg_vif_num].dev;
554 if (reg_dev)
555 dev_hold(reg_dev);
556 read_unlock(&mrt_lock);
557
558 if (!reg_dev)
559 goto drop;
560
561 skb->mac_header = skb->network_header;
562 skb_pull(skb, (u8 *)encap - skb->data);
563 skb_reset_network_header(skb);
564 skb->protocol = htons(ETH_P_IPV6);
565 skb->ip_summed = CHECKSUM_NONE;
566
567 skb_tunnel_rx(skb, reg_dev, dev_net(reg_dev));
568
569 netif_rx(skb);
570
571 dev_put(reg_dev);
572 return 0;
573 drop:
574 kfree_skb(skb);
575 return 0;
576}
577
578static const struct inet6_protocol pim6_protocol = {
579 .handler = pim6_rcv,
580};
581
582
583
584static netdev_tx_t reg_vif_xmit(struct sk_buff *skb,
585 struct net_device *dev)
586{
587 struct net *net = dev_net(dev);
588 struct mr_table *mrt;
589 struct flowi6 fl6 = {
590 .flowi6_oif = dev->ifindex,
591 .flowi6_iif = skb->skb_iif ? : LOOPBACK_IFINDEX,
592 .flowi6_mark = skb->mark,
593 };
594 int err;
595
596 err = ip6mr_fib_lookup(net, &fl6, &mrt);
597 if (err < 0) {
598 kfree_skb(skb);
599 return err;
600 }
601
602 read_lock(&mrt_lock);
603 dev->stats.tx_bytes += skb->len;
604 dev->stats.tx_packets++;
605 ip6mr_cache_report(mrt, skb, mrt->mroute_reg_vif_num, MRT6MSG_WHOLEPKT);
606 read_unlock(&mrt_lock);
607 kfree_skb(skb);
608 return NETDEV_TX_OK;
609}
610
611static int reg_vif_get_iflink(const struct net_device *dev)
612{
613 return 0;
614}
615
616static const struct net_device_ops reg_vif_netdev_ops = {
617 .ndo_start_xmit = reg_vif_xmit,
618 .ndo_get_iflink = reg_vif_get_iflink,
619};
620
621static void reg_vif_setup(struct net_device *dev)
622{
623 dev->type = ARPHRD_PIMREG;
624 dev->mtu = 1500 - sizeof(struct ipv6hdr) - 8;
625 dev->flags = IFF_NOARP;
626 dev->netdev_ops = ®_vif_netdev_ops;
627 dev->needs_free_netdev = true;
628 dev->features |= NETIF_F_NETNS_LOCAL;
629}
630
631static struct net_device *ip6mr_reg_vif(struct net *net, struct mr_table *mrt)
632{
633 struct net_device *dev;
634 char name[IFNAMSIZ];
635
636 if (mrt->id == RT6_TABLE_DFLT)
637 sprintf(name, "pim6reg");
638 else
639 sprintf(name, "pim6reg%u", mrt->id);
640
641 dev = alloc_netdev(0, name, NET_NAME_UNKNOWN, reg_vif_setup);
642 if (!dev)
643 return NULL;
644
645 dev_net_set(dev, net);
646
647 if (register_netdevice(dev)) {
648 free_netdev(dev);
649 return NULL;
650 }
651
652 if (dev_open(dev))
653 goto failure;
654
655 dev_hold(dev);
656 return dev;
657
658failure:
659 unregister_netdevice(dev);
660 return NULL;
661}
662#endif
663
664static int call_ip6mr_vif_entry_notifiers(struct net *net,
665 enum fib_event_type event_type,
666 struct vif_device *vif,
667 mifi_t vif_index, u32 tb_id)
668{
669 return mr_call_vif_notifiers(net, RTNL_FAMILY_IP6MR, event_type,
670 vif, vif_index, tb_id,
671 &net->ipv6.ipmr_seq);
672}
673
674static int call_ip6mr_mfc_entry_notifiers(struct net *net,
675 enum fib_event_type event_type,
676 struct mfc6_cache *mfc, u32 tb_id)
677{
678 return mr_call_mfc_notifiers(net, RTNL_FAMILY_IP6MR, event_type,
679 &mfc->_c, tb_id, &net->ipv6.ipmr_seq);
680}
681
682
683static int mif6_delete(struct mr_table *mrt, int vifi, int notify,
684 struct list_head *head)
685{
686 struct vif_device *v;
687 struct net_device *dev;
688 struct inet6_dev *in6_dev;
689
690 if (vifi < 0 || vifi >= mrt->maxvif)
691 return -EADDRNOTAVAIL;
692
693 v = &mrt->vif_table[vifi];
694
695 if (VIF_EXISTS(mrt, vifi))
696 call_ip6mr_vif_entry_notifiers(read_pnet(&mrt->net),
697 FIB_EVENT_VIF_DEL, v, vifi,
698 mrt->id);
699
700 write_lock_bh(&mrt_lock);
701 dev = v->dev;
702 v->dev = NULL;
703
704 if (!dev) {
705 write_unlock_bh(&mrt_lock);
706 return -EADDRNOTAVAIL;
707 }
708
709#ifdef CONFIG_IPV6_PIMSM_V2
710 if (vifi == mrt->mroute_reg_vif_num)
711 mrt->mroute_reg_vif_num = -1;
712#endif
713
714 if (vifi + 1 == mrt->maxvif) {
715 int tmp;
716 for (tmp = vifi - 1; tmp >= 0; tmp--) {
717 if (VIF_EXISTS(mrt, tmp))
718 break;
719 }
720 mrt->maxvif = tmp + 1;
721 }
722
723 write_unlock_bh(&mrt_lock);
724
725 dev_set_allmulti(dev, -1);
726
727 in6_dev = __in6_dev_get(dev);
728 if (in6_dev) {
729 in6_dev->cnf.mc_forwarding--;
730 inet6_netconf_notify_devconf(dev_net(dev), RTM_NEWNETCONF,
731 NETCONFA_MC_FORWARDING,
732 dev->ifindex, &in6_dev->cnf);
733 }
734
735 if ((v->flags & MIFF_REGISTER) && !notify)
736 unregister_netdevice_queue(dev, head);
737
738 dev_put(dev);
739 return 0;
740}
741
742static inline void ip6mr_cache_free_rcu(struct rcu_head *head)
743{
744 struct mr_mfc *c = container_of(head, struct mr_mfc, rcu);
745
746 kmem_cache_free(mrt_cachep, (struct mfc6_cache *)c);
747}
748
749static inline void ip6mr_cache_free(struct mfc6_cache *c)
750{
751 call_rcu(&c->_c.rcu, ip6mr_cache_free_rcu);
752}
753
754
755
756
757
758static void ip6mr_destroy_unres(struct mr_table *mrt, struct mfc6_cache *c)
759{
760 struct net *net = read_pnet(&mrt->net);
761 struct sk_buff *skb;
762
763 atomic_dec(&mrt->cache_resolve_queue_len);
764
765 while ((skb = skb_dequeue(&c->_c.mfc_un.unres.unresolved)) != NULL) {
766 if (ipv6_hdr(skb)->version == 0) {
767 struct nlmsghdr *nlh = skb_pull(skb,
768 sizeof(struct ipv6hdr));
769 nlh->nlmsg_type = NLMSG_ERROR;
770 nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr));
771 skb_trim(skb, nlh->nlmsg_len);
772 ((struct nlmsgerr *)nlmsg_data(nlh))->error = -ETIMEDOUT;
773 rtnl_unicast(skb, net, NETLINK_CB(skb).portid);
774 } else
775 kfree_skb(skb);
776 }
777
778 ip6mr_cache_free(c);
779}
780
781
782
783
784static void ipmr_do_expire_process(struct mr_table *mrt)
785{
786 unsigned long now = jiffies;
787 unsigned long expires = 10 * HZ;
788 struct mr_mfc *c, *next;
789
790 list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) {
791 if (time_after(c->mfc_un.unres.expires, now)) {
792
793 unsigned long interval = c->mfc_un.unres.expires - now;
794 if (interval < expires)
795 expires = interval;
796 continue;
797 }
798
799 list_del(&c->list);
800 mr6_netlink_event(mrt, (struct mfc6_cache *)c, RTM_DELROUTE);
801 ip6mr_destroy_unres(mrt, (struct mfc6_cache *)c);
802 }
803
804 if (!list_empty(&mrt->mfc_unres_queue))
805 mod_timer(&mrt->ipmr_expire_timer, jiffies + expires);
806}
807
808static void ipmr_expire_process(struct timer_list *t)
809{
810 struct mr_table *mrt = from_timer(mrt, t, ipmr_expire_timer);
811
812 if (!spin_trylock(&mfc_unres_lock)) {
813 mod_timer(&mrt->ipmr_expire_timer, jiffies + 1);
814 return;
815 }
816
817 if (!list_empty(&mrt->mfc_unres_queue))
818 ipmr_do_expire_process(mrt);
819
820 spin_unlock(&mfc_unres_lock);
821}
822
823
824
825static void ip6mr_update_thresholds(struct mr_table *mrt,
826 struct mr_mfc *cache,
827 unsigned char *ttls)
828{
829 int vifi;
830
831 cache->mfc_un.res.minvif = MAXMIFS;
832 cache->mfc_un.res.maxvif = 0;
833 memset(cache->mfc_un.res.ttls, 255, MAXMIFS);
834
835 for (vifi = 0; vifi < mrt->maxvif; vifi++) {
836 if (VIF_EXISTS(mrt, vifi) &&
837 ttls[vifi] && ttls[vifi] < 255) {
838 cache->mfc_un.res.ttls[vifi] = ttls[vifi];
839 if (cache->mfc_un.res.minvif > vifi)
840 cache->mfc_un.res.minvif = vifi;
841 if (cache->mfc_un.res.maxvif <= vifi)
842 cache->mfc_un.res.maxvif = vifi + 1;
843 }
844 }
845 cache->mfc_un.res.lastuse = jiffies;
846}
847
848static int mif6_add(struct net *net, struct mr_table *mrt,
849 struct mif6ctl *vifc, int mrtsock)
850{
851 int vifi = vifc->mif6c_mifi;
852 struct vif_device *v = &mrt->vif_table[vifi];
853 struct net_device *dev;
854 struct inet6_dev *in6_dev;
855 int err;
856
857
858 if (VIF_EXISTS(mrt, vifi))
859 return -EADDRINUSE;
860
861 switch (vifc->mif6c_flags) {
862#ifdef CONFIG_IPV6_PIMSM_V2
863 case MIFF_REGISTER:
864
865
866
867
868 if (mrt->mroute_reg_vif_num >= 0)
869 return -EADDRINUSE;
870 dev = ip6mr_reg_vif(net, mrt);
871 if (!dev)
872 return -ENOBUFS;
873 err = dev_set_allmulti(dev, 1);
874 if (err) {
875 unregister_netdevice(dev);
876 dev_put(dev);
877 return err;
878 }
879 break;
880#endif
881 case 0:
882 dev = dev_get_by_index(net, vifc->mif6c_pifi);
883 if (!dev)
884 return -EADDRNOTAVAIL;
885 err = dev_set_allmulti(dev, 1);
886 if (err) {
887 dev_put(dev);
888 return err;
889 }
890 break;
891 default:
892 return -EINVAL;
893 }
894
895 in6_dev = __in6_dev_get(dev);
896 if (in6_dev) {
897 in6_dev->cnf.mc_forwarding++;
898 inet6_netconf_notify_devconf(dev_net(dev), RTM_NEWNETCONF,
899 NETCONFA_MC_FORWARDING,
900 dev->ifindex, &in6_dev->cnf);
901 }
902
903
904 vif_device_init(v, dev, vifc->vifc_rate_limit, vifc->vifc_threshold,
905 vifc->mif6c_flags | (!mrtsock ? VIFF_STATIC : 0),
906 MIFF_REGISTER);
907
908
909 write_lock_bh(&mrt_lock);
910 v->dev = dev;
911#ifdef CONFIG_IPV6_PIMSM_V2
912 if (v->flags & MIFF_REGISTER)
913 mrt->mroute_reg_vif_num = vifi;
914#endif
915 if (vifi + 1 > mrt->maxvif)
916 mrt->maxvif = vifi + 1;
917 write_unlock_bh(&mrt_lock);
918 call_ip6mr_vif_entry_notifiers(net, FIB_EVENT_VIF_ADD,
919 v, vifi, mrt->id);
920 return 0;
921}
922
923static struct mfc6_cache *ip6mr_cache_find(struct mr_table *mrt,
924 const struct in6_addr *origin,
925 const struct in6_addr *mcastgrp)
926{
927 struct mfc6_cache_cmp_arg arg = {
928 .mf6c_origin = *origin,
929 .mf6c_mcastgrp = *mcastgrp,
930 };
931
932 return mr_mfc_find(mrt, &arg);
933}
934
935
936static struct mfc6_cache *ip6mr_cache_find_any(struct mr_table *mrt,
937 struct in6_addr *mcastgrp,
938 mifi_t mifi)
939{
940 struct mfc6_cache_cmp_arg arg = {
941 .mf6c_origin = in6addr_any,
942 .mf6c_mcastgrp = *mcastgrp,
943 };
944
945 if (ipv6_addr_any(mcastgrp))
946 return mr_mfc_find_any_parent(mrt, mifi);
947 return mr_mfc_find_any(mrt, mifi, &arg);
948}
949
950
951static struct mfc6_cache *
952ip6mr_cache_find_parent(struct mr_table *mrt,
953 const struct in6_addr *origin,
954 const struct in6_addr *mcastgrp,
955 int parent)
956{
957 struct mfc6_cache_cmp_arg arg = {
958 .mf6c_origin = *origin,
959 .mf6c_mcastgrp = *mcastgrp,
960 };
961
962 return mr_mfc_find_parent(mrt, &arg, parent);
963}
964
965
966static struct mfc6_cache *ip6mr_cache_alloc(void)
967{
968 struct mfc6_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL);
969 if (!c)
970 return NULL;
971 c->_c.mfc_un.res.last_assert = jiffies - MFC_ASSERT_THRESH - 1;
972 c->_c.mfc_un.res.minvif = MAXMIFS;
973 c->_c.free = ip6mr_cache_free_rcu;
974 refcount_set(&c->_c.mfc_un.res.refcount, 1);
975 return c;
976}
977
978static struct mfc6_cache *ip6mr_cache_alloc_unres(void)
979{
980 struct mfc6_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC);
981 if (!c)
982 return NULL;
983 skb_queue_head_init(&c->_c.mfc_un.unres.unresolved);
984 c->_c.mfc_un.unres.expires = jiffies + 10 * HZ;
985 return c;
986}
987
988
989
990
991
992static void ip6mr_cache_resolve(struct net *net, struct mr_table *mrt,
993 struct mfc6_cache *uc, struct mfc6_cache *c)
994{
995 struct sk_buff *skb;
996
997
998
999
1000
1001 while ((skb = __skb_dequeue(&uc->_c.mfc_un.unres.unresolved))) {
1002 if (ipv6_hdr(skb)->version == 0) {
1003 struct nlmsghdr *nlh = skb_pull(skb,
1004 sizeof(struct ipv6hdr));
1005
1006 if (mr_fill_mroute(mrt, skb, &c->_c,
1007 nlmsg_data(nlh)) > 0) {
1008 nlh->nlmsg_len = skb_tail_pointer(skb) - (u8 *)nlh;
1009 } else {
1010 nlh->nlmsg_type = NLMSG_ERROR;
1011 nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr));
1012 skb_trim(skb, nlh->nlmsg_len);
1013 ((struct nlmsgerr *)nlmsg_data(nlh))->error = -EMSGSIZE;
1014 }
1015 rtnl_unicast(skb, net, NETLINK_CB(skb).portid);
1016 } else
1017 ip6_mr_forward(net, mrt, skb, c);
1018 }
1019}
1020
1021
1022
1023
1024
1025
1026
1027static int ip6mr_cache_report(struct mr_table *mrt, struct sk_buff *pkt,
1028 mifi_t mifi, int assert)
1029{
1030 struct sock *mroute6_sk;
1031 struct sk_buff *skb;
1032 struct mrt6msg *msg;
1033 int ret;
1034
1035#ifdef CONFIG_IPV6_PIMSM_V2
1036 if (assert == MRT6MSG_WHOLEPKT)
1037 skb = skb_realloc_headroom(pkt, -skb_network_offset(pkt)
1038 +sizeof(*msg));
1039 else
1040#endif
1041 skb = alloc_skb(sizeof(struct ipv6hdr) + sizeof(*msg), GFP_ATOMIC);
1042
1043 if (!skb)
1044 return -ENOBUFS;
1045
1046
1047
1048
1049 skb->ip_summed = CHECKSUM_UNNECESSARY;
1050
1051#ifdef CONFIG_IPV6_PIMSM_V2
1052 if (assert == MRT6MSG_WHOLEPKT) {
1053
1054
1055
1056
1057
1058 skb_push(skb, -skb_network_offset(pkt));
1059
1060 skb_push(skb, sizeof(*msg));
1061 skb_reset_transport_header(skb);
1062 msg = (struct mrt6msg *)skb_transport_header(skb);
1063 msg->im6_mbz = 0;
1064 msg->im6_msgtype = MRT6MSG_WHOLEPKT;
1065 msg->im6_mif = mrt->mroute_reg_vif_num;
1066 msg->im6_pad = 0;
1067 msg->im6_src = ipv6_hdr(pkt)->saddr;
1068 msg->im6_dst = ipv6_hdr(pkt)->daddr;
1069
1070 skb->ip_summed = CHECKSUM_UNNECESSARY;
1071 } else
1072#endif
1073 {
1074
1075
1076
1077
1078 skb_put(skb, sizeof(struct ipv6hdr));
1079 skb_reset_network_header(skb);
1080 skb_copy_to_linear_data(skb, ipv6_hdr(pkt), sizeof(struct ipv6hdr));
1081
1082
1083
1084
1085 skb_put(skb, sizeof(*msg));
1086 skb_reset_transport_header(skb);
1087 msg = (struct mrt6msg *)skb_transport_header(skb);
1088
1089 msg->im6_mbz = 0;
1090 msg->im6_msgtype = assert;
1091 msg->im6_mif = mifi;
1092 msg->im6_pad = 0;
1093 msg->im6_src = ipv6_hdr(pkt)->saddr;
1094 msg->im6_dst = ipv6_hdr(pkt)->daddr;
1095
1096 skb_dst_set(skb, dst_clone(skb_dst(pkt)));
1097 skb->ip_summed = CHECKSUM_UNNECESSARY;
1098 }
1099
1100 rcu_read_lock();
1101 mroute6_sk = rcu_dereference(mrt->mroute_sk);
1102 if (!mroute6_sk) {
1103 rcu_read_unlock();
1104 kfree_skb(skb);
1105 return -EINVAL;
1106 }
1107
1108 mrt6msg_netlink_event(mrt, skb);
1109
1110
1111 ret = sock_queue_rcv_skb(mroute6_sk, skb);
1112 rcu_read_unlock();
1113 if (ret < 0) {
1114 net_warn_ratelimited("mroute6: pending queue full, dropping entries\n");
1115 kfree_skb(skb);
1116 }
1117
1118 return ret;
1119}
1120
1121
1122static int ip6mr_cache_unresolved(struct mr_table *mrt, mifi_t mifi,
1123 struct sk_buff *skb)
1124{
1125 struct mfc6_cache *c;
1126 bool found = false;
1127 int err;
1128
1129 spin_lock_bh(&mfc_unres_lock);
1130 list_for_each_entry(c, &mrt->mfc_unres_queue, _c.list) {
1131 if (ipv6_addr_equal(&c->mf6c_mcastgrp, &ipv6_hdr(skb)->daddr) &&
1132 ipv6_addr_equal(&c->mf6c_origin, &ipv6_hdr(skb)->saddr)) {
1133 found = true;
1134 break;
1135 }
1136 }
1137
1138 if (!found) {
1139
1140
1141
1142
1143 if (atomic_read(&mrt->cache_resolve_queue_len) >= 10 ||
1144 (c = ip6mr_cache_alloc_unres()) == NULL) {
1145 spin_unlock_bh(&mfc_unres_lock);
1146
1147 kfree_skb(skb);
1148 return -ENOBUFS;
1149 }
1150
1151
1152 c->_c.mfc_parent = -1;
1153 c->mf6c_origin = ipv6_hdr(skb)->saddr;
1154 c->mf6c_mcastgrp = ipv6_hdr(skb)->daddr;
1155
1156
1157
1158
1159 err = ip6mr_cache_report(mrt, skb, mifi, MRT6MSG_NOCACHE);
1160 if (err < 0) {
1161
1162
1163
1164 spin_unlock_bh(&mfc_unres_lock);
1165
1166 ip6mr_cache_free(c);
1167 kfree_skb(skb);
1168 return err;
1169 }
1170
1171 atomic_inc(&mrt->cache_resolve_queue_len);
1172 list_add(&c->_c.list, &mrt->mfc_unres_queue);
1173 mr6_netlink_event(mrt, c, RTM_NEWROUTE);
1174
1175 ipmr_do_expire_process(mrt);
1176 }
1177
1178
1179 if (c->_c.mfc_un.unres.unresolved.qlen > 3) {
1180 kfree_skb(skb);
1181 err = -ENOBUFS;
1182 } else {
1183 skb_queue_tail(&c->_c.mfc_un.unres.unresolved, skb);
1184 err = 0;
1185 }
1186
1187 spin_unlock_bh(&mfc_unres_lock);
1188 return err;
1189}
1190
1191
1192
1193
1194
1195static int ip6mr_mfc_delete(struct mr_table *mrt, struct mf6cctl *mfc,
1196 int parent)
1197{
1198 struct mfc6_cache *c;
1199
1200
1201 rcu_read_lock();
1202 c = ip6mr_cache_find_parent(mrt, &mfc->mf6cc_origin.sin6_addr,
1203 &mfc->mf6cc_mcastgrp.sin6_addr, parent);
1204 rcu_read_unlock();
1205 if (!c)
1206 return -ENOENT;
1207 rhltable_remove(&mrt->mfc_hash, &c->_c.mnode, ip6mr_rht_params);
1208 list_del_rcu(&c->_c.list);
1209
1210 call_ip6mr_mfc_entry_notifiers(read_pnet(&mrt->net),
1211 FIB_EVENT_ENTRY_DEL, c, mrt->id);
1212 mr6_netlink_event(mrt, c, RTM_DELROUTE);
1213 mr_cache_put(&c->_c);
1214 return 0;
1215}
1216
1217static int ip6mr_device_event(struct notifier_block *this,
1218 unsigned long event, void *ptr)
1219{
1220 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
1221 struct net *net = dev_net(dev);
1222 struct mr_table *mrt;
1223 struct vif_device *v;
1224 int ct;
1225
1226 if (event != NETDEV_UNREGISTER)
1227 return NOTIFY_DONE;
1228
1229 ip6mr_for_each_table(mrt, net) {
1230 v = &mrt->vif_table[0];
1231 for (ct = 0; ct < mrt->maxvif; ct++, v++) {
1232 if (v->dev == dev)
1233 mif6_delete(mrt, ct, 1, NULL);
1234 }
1235 }
1236
1237 return NOTIFY_DONE;
1238}
1239
1240static unsigned int ip6mr_seq_read(struct net *net)
1241{
1242 ASSERT_RTNL();
1243
1244 return net->ipv6.ipmr_seq + ip6mr_rules_seq_read(net);
1245}
1246
1247static int ip6mr_dump(struct net *net, struct notifier_block *nb)
1248{
1249 return mr_dump(net, nb, RTNL_FAMILY_IP6MR, ip6mr_rules_dump,
1250 ip6mr_mr_table_iter, &mrt_lock);
1251}
1252
1253static struct notifier_block ip6_mr_notifier = {
1254 .notifier_call = ip6mr_device_event
1255};
1256
1257static const struct fib_notifier_ops ip6mr_notifier_ops_template = {
1258 .family = RTNL_FAMILY_IP6MR,
1259 .fib_seq_read = ip6mr_seq_read,
1260 .fib_dump = ip6mr_dump,
1261 .owner = THIS_MODULE,
1262};
1263
1264static int __net_init ip6mr_notifier_init(struct net *net)
1265{
1266 struct fib_notifier_ops *ops;
1267
1268 net->ipv6.ipmr_seq = 0;
1269
1270 ops = fib_notifier_ops_register(&ip6mr_notifier_ops_template, net);
1271 if (IS_ERR(ops))
1272 return PTR_ERR(ops);
1273
1274 net->ipv6.ip6mr_notifier_ops = ops;
1275
1276 return 0;
1277}
1278
1279static void __net_exit ip6mr_notifier_exit(struct net *net)
1280{
1281 fib_notifier_ops_unregister(net->ipv6.ip6mr_notifier_ops);
1282 net->ipv6.ip6mr_notifier_ops = NULL;
1283}
1284
1285
1286static int __net_init ip6mr_net_init(struct net *net)
1287{
1288 int err;
1289
1290 err = ip6mr_notifier_init(net);
1291 if (err)
1292 return err;
1293
1294 err = ip6mr_rules_init(net);
1295 if (err < 0)
1296 goto ip6mr_rules_fail;
1297
1298#ifdef CONFIG_PROC_FS
1299 err = -ENOMEM;
1300 if (!proc_create_net("ip6_mr_vif", 0, net->proc_net, &ip6mr_vif_seq_ops,
1301 sizeof(struct mr_vif_iter)))
1302 goto proc_vif_fail;
1303 if (!proc_create_net("ip6_mr_cache", 0, net->proc_net, &ipmr_mfc_seq_ops,
1304 sizeof(struct mr_mfc_iter)))
1305 goto proc_cache_fail;
1306#endif
1307
1308 return 0;
1309
1310#ifdef CONFIG_PROC_FS
1311proc_cache_fail:
1312 remove_proc_entry("ip6_mr_vif", net->proc_net);
1313proc_vif_fail:
1314 ip6mr_rules_exit(net);
1315#endif
1316ip6mr_rules_fail:
1317 ip6mr_notifier_exit(net);
1318 return err;
1319}
1320
1321static void __net_exit ip6mr_net_exit(struct net *net)
1322{
1323#ifdef CONFIG_PROC_FS
1324 remove_proc_entry("ip6_mr_cache", net->proc_net);
1325 remove_proc_entry("ip6_mr_vif", net->proc_net);
1326#endif
1327 ip6mr_rules_exit(net);
1328 ip6mr_notifier_exit(net);
1329}
1330
1331static struct pernet_operations ip6mr_net_ops = {
1332 .init = ip6mr_net_init,
1333 .exit = ip6mr_net_exit,
1334};
1335
1336int __init ip6_mr_init(void)
1337{
1338 int err;
1339
1340 mrt_cachep = kmem_cache_create("ip6_mrt_cache",
1341 sizeof(struct mfc6_cache),
1342 0, SLAB_HWCACHE_ALIGN,
1343 NULL);
1344 if (!mrt_cachep)
1345 return -ENOMEM;
1346
1347 err = register_pernet_subsys(&ip6mr_net_ops);
1348 if (err)
1349 goto reg_pernet_fail;
1350
1351 err = register_netdevice_notifier(&ip6_mr_notifier);
1352 if (err)
1353 goto reg_notif_fail;
1354#ifdef CONFIG_IPV6_PIMSM_V2
1355 if (inet6_add_protocol(&pim6_protocol, IPPROTO_PIM) < 0) {
1356 pr_err("%s: can't add PIM protocol\n", __func__);
1357 err = -EAGAIN;
1358 goto add_proto_fail;
1359 }
1360#endif
1361 err = rtnl_register_module(THIS_MODULE, RTNL_FAMILY_IP6MR, RTM_GETROUTE,
1362 NULL, ip6mr_rtm_dumproute, 0);
1363 if (err == 0)
1364 return 0;
1365
1366#ifdef CONFIG_IPV6_PIMSM_V2
1367 inet6_del_protocol(&pim6_protocol, IPPROTO_PIM);
1368add_proto_fail:
1369 unregister_netdevice_notifier(&ip6_mr_notifier);
1370#endif
1371reg_notif_fail:
1372 unregister_pernet_subsys(&ip6mr_net_ops);
1373reg_pernet_fail:
1374 kmem_cache_destroy(mrt_cachep);
1375 return err;
1376}
1377
1378void ip6_mr_cleanup(void)
1379{
1380 rtnl_unregister(RTNL_FAMILY_IP6MR, RTM_GETROUTE);
1381#ifdef CONFIG_IPV6_PIMSM_V2
1382 inet6_del_protocol(&pim6_protocol, IPPROTO_PIM);
1383#endif
1384 unregister_netdevice_notifier(&ip6_mr_notifier);
1385 unregister_pernet_subsys(&ip6mr_net_ops);
1386 kmem_cache_destroy(mrt_cachep);
1387}
1388
1389static int ip6mr_mfc_add(struct net *net, struct mr_table *mrt,
1390 struct mf6cctl *mfc, int mrtsock, int parent)
1391{
1392 unsigned char ttls[MAXMIFS];
1393 struct mfc6_cache *uc, *c;
1394 struct mr_mfc *_uc;
1395 bool found;
1396 int i, err;
1397
1398 if (mfc->mf6cc_parent >= MAXMIFS)
1399 return -ENFILE;
1400
1401 memset(ttls, 255, MAXMIFS);
1402 for (i = 0; i < MAXMIFS; i++) {
1403 if (IF_ISSET(i, &mfc->mf6cc_ifset))
1404 ttls[i] = 1;
1405 }
1406
1407
1408 rcu_read_lock();
1409 c = ip6mr_cache_find_parent(mrt, &mfc->mf6cc_origin.sin6_addr,
1410 &mfc->mf6cc_mcastgrp.sin6_addr, parent);
1411 rcu_read_unlock();
1412 if (c) {
1413 write_lock_bh(&mrt_lock);
1414 c->_c.mfc_parent = mfc->mf6cc_parent;
1415 ip6mr_update_thresholds(mrt, &c->_c, ttls);
1416 if (!mrtsock)
1417 c->_c.mfc_flags |= MFC_STATIC;
1418 write_unlock_bh(&mrt_lock);
1419 call_ip6mr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_REPLACE,
1420 c, mrt->id);
1421 mr6_netlink_event(mrt, c, RTM_NEWROUTE);
1422 return 0;
1423 }
1424
1425 if (!ipv6_addr_any(&mfc->mf6cc_mcastgrp.sin6_addr) &&
1426 !ipv6_addr_is_multicast(&mfc->mf6cc_mcastgrp.sin6_addr))
1427 return -EINVAL;
1428
1429 c = ip6mr_cache_alloc();
1430 if (!c)
1431 return -ENOMEM;
1432
1433 c->mf6c_origin = mfc->mf6cc_origin.sin6_addr;
1434 c->mf6c_mcastgrp = mfc->mf6cc_mcastgrp.sin6_addr;
1435 c->_c.mfc_parent = mfc->mf6cc_parent;
1436 ip6mr_update_thresholds(mrt, &c->_c, ttls);
1437 if (!mrtsock)
1438 c->_c.mfc_flags |= MFC_STATIC;
1439
1440 err = rhltable_insert_key(&mrt->mfc_hash, &c->cmparg, &c->_c.mnode,
1441 ip6mr_rht_params);
1442 if (err) {
1443 pr_err("ip6mr: rhtable insert error %d\n", err);
1444 ip6mr_cache_free(c);
1445 return err;
1446 }
1447 list_add_tail_rcu(&c->_c.list, &mrt->mfc_cache_list);
1448
1449
1450
1451
1452 found = false;
1453 spin_lock_bh(&mfc_unres_lock);
1454 list_for_each_entry(_uc, &mrt->mfc_unres_queue, list) {
1455 uc = (struct mfc6_cache *)_uc;
1456 if (ipv6_addr_equal(&uc->mf6c_origin, &c->mf6c_origin) &&
1457 ipv6_addr_equal(&uc->mf6c_mcastgrp, &c->mf6c_mcastgrp)) {
1458 list_del(&_uc->list);
1459 atomic_dec(&mrt->cache_resolve_queue_len);
1460 found = true;
1461 break;
1462 }
1463 }
1464 if (list_empty(&mrt->mfc_unres_queue))
1465 del_timer(&mrt->ipmr_expire_timer);
1466 spin_unlock_bh(&mfc_unres_lock);
1467
1468 if (found) {
1469 ip6mr_cache_resolve(net, mrt, uc, c);
1470 ip6mr_cache_free(uc);
1471 }
1472 call_ip6mr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_ADD,
1473 c, mrt->id);
1474 mr6_netlink_event(mrt, c, RTM_NEWROUTE);
1475 return 0;
1476}
1477
1478
1479
1480
1481
1482static void mroute_clean_tables(struct mr_table *mrt, bool all)
1483{
1484 struct mr_mfc *c, *tmp;
1485 LIST_HEAD(list);
1486 int i;
1487
1488
1489 for (i = 0; i < mrt->maxvif; i++) {
1490 if (!all && (mrt->vif_table[i].flags & VIFF_STATIC))
1491 continue;
1492 mif6_delete(mrt, i, 0, &list);
1493 }
1494 unregister_netdevice_many(&list);
1495
1496
1497 list_for_each_entry_safe(c, tmp, &mrt->mfc_cache_list, list) {
1498 if (!all && (c->mfc_flags & MFC_STATIC))
1499 continue;
1500 rhltable_remove(&mrt->mfc_hash, &c->mnode, ip6mr_rht_params);
1501 list_del_rcu(&c->list);
1502 mr6_netlink_event(mrt, (struct mfc6_cache *)c, RTM_DELROUTE);
1503 mr_cache_put(c);
1504 }
1505
1506 if (atomic_read(&mrt->cache_resolve_queue_len) != 0) {
1507 spin_lock_bh(&mfc_unres_lock);
1508 list_for_each_entry_safe(c, tmp, &mrt->mfc_unres_queue, list) {
1509 list_del(&c->list);
1510 call_ip6mr_mfc_entry_notifiers(read_pnet(&mrt->net),
1511 FIB_EVENT_ENTRY_DEL,
1512 (struct mfc6_cache *)c,
1513 mrt->id);
1514 mr6_netlink_event(mrt, (struct mfc6_cache *)c,
1515 RTM_DELROUTE);
1516 ip6mr_destroy_unres(mrt, (struct mfc6_cache *)c);
1517 }
1518 spin_unlock_bh(&mfc_unres_lock);
1519 }
1520}
1521
1522static int ip6mr_sk_init(struct mr_table *mrt, struct sock *sk)
1523{
1524 int err = 0;
1525 struct net *net = sock_net(sk);
1526
1527 rtnl_lock();
1528 write_lock_bh(&mrt_lock);
1529 if (rtnl_dereference(mrt->mroute_sk)) {
1530 err = -EADDRINUSE;
1531 } else {
1532 rcu_assign_pointer(mrt->mroute_sk, sk);
1533 sock_set_flag(sk, SOCK_RCU_FREE);
1534 net->ipv6.devconf_all->mc_forwarding++;
1535 }
1536 write_unlock_bh(&mrt_lock);
1537
1538 if (!err)
1539 inet6_netconf_notify_devconf(net, RTM_NEWNETCONF,
1540 NETCONFA_MC_FORWARDING,
1541 NETCONFA_IFINDEX_ALL,
1542 net->ipv6.devconf_all);
1543 rtnl_unlock();
1544
1545 return err;
1546}
1547
1548int ip6mr_sk_done(struct sock *sk)
1549{
1550 int err = -EACCES;
1551 struct net *net = sock_net(sk);
1552 struct mr_table *mrt;
1553
1554 if (sk->sk_type != SOCK_RAW ||
1555 inet_sk(sk)->inet_num != IPPROTO_ICMPV6)
1556 return err;
1557
1558 rtnl_lock();
1559 ip6mr_for_each_table(mrt, net) {
1560 if (sk == rtnl_dereference(mrt->mroute_sk)) {
1561 write_lock_bh(&mrt_lock);
1562 RCU_INIT_POINTER(mrt->mroute_sk, NULL);
1563
1564
1565
1566
1567 net->ipv6.devconf_all->mc_forwarding--;
1568 write_unlock_bh(&mrt_lock);
1569 inet6_netconf_notify_devconf(net, RTM_NEWNETCONF,
1570 NETCONFA_MC_FORWARDING,
1571 NETCONFA_IFINDEX_ALL,
1572 net->ipv6.devconf_all);
1573
1574 mroute_clean_tables(mrt, false);
1575 err = 0;
1576 break;
1577 }
1578 }
1579 rtnl_unlock();
1580
1581 return err;
1582}
1583
1584bool mroute6_is_socket(struct net *net, struct sk_buff *skb)
1585{
1586 struct mr_table *mrt;
1587 struct flowi6 fl6 = {
1588 .flowi6_iif = skb->skb_iif ? : LOOPBACK_IFINDEX,
1589 .flowi6_oif = skb->dev->ifindex,
1590 .flowi6_mark = skb->mark,
1591 };
1592
1593 if (ip6mr_fib_lookup(net, &fl6, &mrt) < 0)
1594 return NULL;
1595
1596 return rcu_access_pointer(mrt->mroute_sk);
1597}
1598EXPORT_SYMBOL(mroute6_is_socket);
1599
1600
1601
1602
1603
1604
1605
1606
1607int ip6_mroute_setsockopt(struct sock *sk, int optname, char __user *optval, unsigned int optlen)
1608{
1609 int ret, parent = 0;
1610 struct mif6ctl vif;
1611 struct mf6cctl mfc;
1612 mifi_t mifi;
1613 struct net *net = sock_net(sk);
1614 struct mr_table *mrt;
1615
1616 if (sk->sk_type != SOCK_RAW ||
1617 inet_sk(sk)->inet_num != IPPROTO_ICMPV6)
1618 return -EOPNOTSUPP;
1619
1620 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
1621 if (!mrt)
1622 return -ENOENT;
1623
1624 if (optname != MRT6_INIT) {
1625 if (sk != rcu_access_pointer(mrt->mroute_sk) &&
1626 !ns_capable(net->user_ns, CAP_NET_ADMIN))
1627 return -EACCES;
1628 }
1629
1630 switch (optname) {
1631 case MRT6_INIT:
1632 if (optlen < sizeof(int))
1633 return -EINVAL;
1634
1635 return ip6mr_sk_init(mrt, sk);
1636
1637 case MRT6_DONE:
1638 return ip6mr_sk_done(sk);
1639
1640 case MRT6_ADD_MIF:
1641 if (optlen < sizeof(vif))
1642 return -EINVAL;
1643 if (copy_from_user(&vif, optval, sizeof(vif)))
1644 return -EFAULT;
1645 if (vif.mif6c_mifi >= MAXMIFS)
1646 return -ENFILE;
1647 rtnl_lock();
1648 ret = mif6_add(net, mrt, &vif,
1649 sk == rtnl_dereference(mrt->mroute_sk));
1650 rtnl_unlock();
1651 return ret;
1652
1653 case MRT6_DEL_MIF:
1654 if (optlen < sizeof(mifi_t))
1655 return -EINVAL;
1656 if (copy_from_user(&mifi, optval, sizeof(mifi_t)))
1657 return -EFAULT;
1658 rtnl_lock();
1659 ret = mif6_delete(mrt, mifi, 0, NULL);
1660 rtnl_unlock();
1661 return ret;
1662
1663
1664
1665
1666
1667 case MRT6_ADD_MFC:
1668 case MRT6_DEL_MFC:
1669 parent = -1;
1670
1671 case MRT6_ADD_MFC_PROXY:
1672 case MRT6_DEL_MFC_PROXY:
1673 if (optlen < sizeof(mfc))
1674 return -EINVAL;
1675 if (copy_from_user(&mfc, optval, sizeof(mfc)))
1676 return -EFAULT;
1677 if (parent == 0)
1678 parent = mfc.mf6cc_parent;
1679 rtnl_lock();
1680 if (optname == MRT6_DEL_MFC || optname == MRT6_DEL_MFC_PROXY)
1681 ret = ip6mr_mfc_delete(mrt, &mfc, parent);
1682 else
1683 ret = ip6mr_mfc_add(net, mrt, &mfc,
1684 sk ==
1685 rtnl_dereference(mrt->mroute_sk),
1686 parent);
1687 rtnl_unlock();
1688 return ret;
1689
1690
1691
1692
1693 case MRT6_ASSERT:
1694 {
1695 int v;
1696
1697 if (optlen != sizeof(v))
1698 return -EINVAL;
1699 if (get_user(v, (int __user *)optval))
1700 return -EFAULT;
1701 mrt->mroute_do_assert = v;
1702 return 0;
1703 }
1704
1705#ifdef CONFIG_IPV6_PIMSM_V2
1706 case MRT6_PIM:
1707 {
1708 int v;
1709
1710 if (optlen != sizeof(v))
1711 return -EINVAL;
1712 if (get_user(v, (int __user *)optval))
1713 return -EFAULT;
1714 v = !!v;
1715 rtnl_lock();
1716 ret = 0;
1717 if (v != mrt->mroute_do_pim) {
1718 mrt->mroute_do_pim = v;
1719 mrt->mroute_do_assert = v;
1720 }
1721 rtnl_unlock();
1722 return ret;
1723 }
1724
1725#endif
1726#ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES
1727 case MRT6_TABLE:
1728 {
1729 u32 v;
1730
1731 if (optlen != sizeof(u32))
1732 return -EINVAL;
1733 if (get_user(v, (u32 __user *)optval))
1734 return -EFAULT;
1735
1736 if (v != RT_TABLE_DEFAULT && v >= 100000000)
1737 return -EINVAL;
1738 if (sk == rcu_access_pointer(mrt->mroute_sk))
1739 return -EBUSY;
1740
1741 rtnl_lock();
1742 ret = 0;
1743 mrt = ip6mr_new_table(net, v);
1744 if (IS_ERR(mrt))
1745 ret = PTR_ERR(mrt);
1746 else
1747 raw6_sk(sk)->ip6mr_table = v;
1748 rtnl_unlock();
1749 return ret;
1750 }
1751#endif
1752
1753
1754
1755
1756 default:
1757 return -ENOPROTOOPT;
1758 }
1759}
1760
1761
1762
1763
1764
1765int ip6_mroute_getsockopt(struct sock *sk, int optname, char __user *optval,
1766 int __user *optlen)
1767{
1768 int olr;
1769 int val;
1770 struct net *net = sock_net(sk);
1771 struct mr_table *mrt;
1772
1773 if (sk->sk_type != SOCK_RAW ||
1774 inet_sk(sk)->inet_num != IPPROTO_ICMPV6)
1775 return -EOPNOTSUPP;
1776
1777 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
1778 if (!mrt)
1779 return -ENOENT;
1780
1781 switch (optname) {
1782 case MRT6_VERSION:
1783 val = 0x0305;
1784 break;
1785#ifdef CONFIG_IPV6_PIMSM_V2
1786 case MRT6_PIM:
1787 val = mrt->mroute_do_pim;
1788 break;
1789#endif
1790 case MRT6_ASSERT:
1791 val = mrt->mroute_do_assert;
1792 break;
1793 default:
1794 return -ENOPROTOOPT;
1795 }
1796
1797 if (get_user(olr, optlen))
1798 return -EFAULT;
1799
1800 olr = min_t(int, olr, sizeof(int));
1801 if (olr < 0)
1802 return -EINVAL;
1803
1804 if (put_user(olr, optlen))
1805 return -EFAULT;
1806 if (copy_to_user(optval, &val, olr))
1807 return -EFAULT;
1808 return 0;
1809}
1810
1811
1812
1813
1814
1815int ip6mr_ioctl(struct sock *sk, int cmd, void __user *arg)
1816{
1817 struct sioc_sg_req6 sr;
1818 struct sioc_mif_req6 vr;
1819 struct vif_device *vif;
1820 struct mfc6_cache *c;
1821 struct net *net = sock_net(sk);
1822 struct mr_table *mrt;
1823
1824 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
1825 if (!mrt)
1826 return -ENOENT;
1827
1828 switch (cmd) {
1829 case SIOCGETMIFCNT_IN6:
1830 if (copy_from_user(&vr, arg, sizeof(vr)))
1831 return -EFAULT;
1832 if (vr.mifi >= mrt->maxvif)
1833 return -EINVAL;
1834 read_lock(&mrt_lock);
1835 vif = &mrt->vif_table[vr.mifi];
1836 if (VIF_EXISTS(mrt, vr.mifi)) {
1837 vr.icount = vif->pkt_in;
1838 vr.ocount = vif->pkt_out;
1839 vr.ibytes = vif->bytes_in;
1840 vr.obytes = vif->bytes_out;
1841 read_unlock(&mrt_lock);
1842
1843 if (copy_to_user(arg, &vr, sizeof(vr)))
1844 return -EFAULT;
1845 return 0;
1846 }
1847 read_unlock(&mrt_lock);
1848 return -EADDRNOTAVAIL;
1849 case SIOCGETSGCNT_IN6:
1850 if (copy_from_user(&sr, arg, sizeof(sr)))
1851 return -EFAULT;
1852
1853 rcu_read_lock();
1854 c = ip6mr_cache_find(mrt, &sr.src.sin6_addr, &sr.grp.sin6_addr);
1855 if (c) {
1856 sr.pktcnt = c->_c.mfc_un.res.pkt;
1857 sr.bytecnt = c->_c.mfc_un.res.bytes;
1858 sr.wrong_if = c->_c.mfc_un.res.wrong_if;
1859 rcu_read_unlock();
1860
1861 if (copy_to_user(arg, &sr, sizeof(sr)))
1862 return -EFAULT;
1863 return 0;
1864 }
1865 rcu_read_unlock();
1866 return -EADDRNOTAVAIL;
1867 default:
1868 return -ENOIOCTLCMD;
1869 }
1870}
1871
1872#ifdef CONFIG_COMPAT
1873struct compat_sioc_sg_req6 {
1874 struct sockaddr_in6 src;
1875 struct sockaddr_in6 grp;
1876 compat_ulong_t pktcnt;
1877 compat_ulong_t bytecnt;
1878 compat_ulong_t wrong_if;
1879};
1880
1881struct compat_sioc_mif_req6 {
1882 mifi_t mifi;
1883 compat_ulong_t icount;
1884 compat_ulong_t ocount;
1885 compat_ulong_t ibytes;
1886 compat_ulong_t obytes;
1887};
1888
1889int ip6mr_compat_ioctl(struct sock *sk, unsigned int cmd, void __user *arg)
1890{
1891 struct compat_sioc_sg_req6 sr;
1892 struct compat_sioc_mif_req6 vr;
1893 struct vif_device *vif;
1894 struct mfc6_cache *c;
1895 struct net *net = sock_net(sk);
1896 struct mr_table *mrt;
1897
1898 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
1899 if (!mrt)
1900 return -ENOENT;
1901
1902 switch (cmd) {
1903 case SIOCGETMIFCNT_IN6:
1904 if (copy_from_user(&vr, arg, sizeof(vr)))
1905 return -EFAULT;
1906 if (vr.mifi >= mrt->maxvif)
1907 return -EINVAL;
1908 read_lock(&mrt_lock);
1909 vif = &mrt->vif_table[vr.mifi];
1910 if (VIF_EXISTS(mrt, vr.mifi)) {
1911 vr.icount = vif->pkt_in;
1912 vr.ocount = vif->pkt_out;
1913 vr.ibytes = vif->bytes_in;
1914 vr.obytes = vif->bytes_out;
1915 read_unlock(&mrt_lock);
1916
1917 if (copy_to_user(arg, &vr, sizeof(vr)))
1918 return -EFAULT;
1919 return 0;
1920 }
1921 read_unlock(&mrt_lock);
1922 return -EADDRNOTAVAIL;
1923 case SIOCGETSGCNT_IN6:
1924 if (copy_from_user(&sr, arg, sizeof(sr)))
1925 return -EFAULT;
1926
1927 rcu_read_lock();
1928 c = ip6mr_cache_find(mrt, &sr.src.sin6_addr, &sr.grp.sin6_addr);
1929 if (c) {
1930 sr.pktcnt = c->_c.mfc_un.res.pkt;
1931 sr.bytecnt = c->_c.mfc_un.res.bytes;
1932 sr.wrong_if = c->_c.mfc_un.res.wrong_if;
1933 rcu_read_unlock();
1934
1935 if (copy_to_user(arg, &sr, sizeof(sr)))
1936 return -EFAULT;
1937 return 0;
1938 }
1939 rcu_read_unlock();
1940 return -EADDRNOTAVAIL;
1941 default:
1942 return -ENOIOCTLCMD;
1943 }
1944}
1945#endif
1946
1947static inline int ip6mr_forward2_finish(struct net *net, struct sock *sk, struct sk_buff *skb)
1948{
1949 __IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
1950 IPSTATS_MIB_OUTFORWDATAGRAMS);
1951 __IP6_ADD_STATS(net, ip6_dst_idev(skb_dst(skb)),
1952 IPSTATS_MIB_OUTOCTETS, skb->len);
1953 return dst_output(net, sk, skb);
1954}
1955
1956
1957
1958
1959
1960static int ip6mr_forward2(struct net *net, struct mr_table *mrt,
1961 struct sk_buff *skb, struct mfc6_cache *c, int vifi)
1962{
1963 struct ipv6hdr *ipv6h;
1964 struct vif_device *vif = &mrt->vif_table[vifi];
1965 struct net_device *dev;
1966 struct dst_entry *dst;
1967 struct flowi6 fl6;
1968
1969 if (!vif->dev)
1970 goto out_free;
1971
1972#ifdef CONFIG_IPV6_PIMSM_V2
1973 if (vif->flags & MIFF_REGISTER) {
1974 vif->pkt_out++;
1975 vif->bytes_out += skb->len;
1976 vif->dev->stats.tx_bytes += skb->len;
1977 vif->dev->stats.tx_packets++;
1978 ip6mr_cache_report(mrt, skb, vifi, MRT6MSG_WHOLEPKT);
1979 goto out_free;
1980 }
1981#endif
1982
1983 ipv6h = ipv6_hdr(skb);
1984
1985 fl6 = (struct flowi6) {
1986 .flowi6_oif = vif->link,
1987 .daddr = ipv6h->daddr,
1988 };
1989
1990 dst = ip6_route_output(net, NULL, &fl6);
1991 if (dst->error) {
1992 dst_release(dst);
1993 goto out_free;
1994 }
1995
1996 skb_dst_drop(skb);
1997 skb_dst_set(skb, dst);
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010 dev = vif->dev;
2011 skb->dev = dev;
2012 vif->pkt_out++;
2013 vif->bytes_out += skb->len;
2014
2015
2016
2017 if (skb_cow(skb, sizeof(*ipv6h) + LL_RESERVED_SPACE(dev)))
2018 goto out_free;
2019
2020 ipv6h = ipv6_hdr(skb);
2021 ipv6h->hop_limit--;
2022
2023 IP6CB(skb)->flags |= IP6SKB_FORWARDED;
2024
2025 return NF_HOOK(NFPROTO_IPV6, NF_INET_FORWARD,
2026 net, NULL, skb, skb->dev, dev,
2027 ip6mr_forward2_finish);
2028
2029out_free:
2030 kfree_skb(skb);
2031 return 0;
2032}
2033
2034static int ip6mr_find_vif(struct mr_table *mrt, struct net_device *dev)
2035{
2036 int ct;
2037
2038 for (ct = mrt->maxvif - 1; ct >= 0; ct--) {
2039 if (mrt->vif_table[ct].dev == dev)
2040 break;
2041 }
2042 return ct;
2043}
2044
2045static void ip6_mr_forward(struct net *net, struct mr_table *mrt,
2046 struct sk_buff *skb, struct mfc6_cache *c)
2047{
2048 int psend = -1;
2049 int vif, ct;
2050 int true_vifi = ip6mr_find_vif(mrt, skb->dev);
2051
2052 vif = c->_c.mfc_parent;
2053 c->_c.mfc_un.res.pkt++;
2054 c->_c.mfc_un.res.bytes += skb->len;
2055 c->_c.mfc_un.res.lastuse = jiffies;
2056
2057 if (ipv6_addr_any(&c->mf6c_origin) && true_vifi >= 0) {
2058 struct mfc6_cache *cache_proxy;
2059
2060
2061
2062
2063 rcu_read_lock();
2064 cache_proxy = mr_mfc_find_any_parent(mrt, vif);
2065 if (cache_proxy &&
2066 cache_proxy->_c.mfc_un.res.ttls[true_vifi] < 255) {
2067 rcu_read_unlock();
2068 goto forward;
2069 }
2070 rcu_read_unlock();
2071 }
2072
2073
2074
2075
2076 if (mrt->vif_table[vif].dev != skb->dev) {
2077 c->_c.mfc_un.res.wrong_if++;
2078
2079 if (true_vifi >= 0 && mrt->mroute_do_assert &&
2080
2081
2082
2083
2084
2085 (mrt->mroute_do_pim ||
2086 c->_c.mfc_un.res.ttls[true_vifi] < 255) &&
2087 time_after(jiffies,
2088 c->_c.mfc_un.res.last_assert +
2089 MFC_ASSERT_THRESH)) {
2090 c->_c.mfc_un.res.last_assert = jiffies;
2091 ip6mr_cache_report(mrt, skb, true_vifi, MRT6MSG_WRONGMIF);
2092 }
2093 goto dont_forward;
2094 }
2095
2096forward:
2097 mrt->vif_table[vif].pkt_in++;
2098 mrt->vif_table[vif].bytes_in += skb->len;
2099
2100
2101
2102
2103 if (ipv6_addr_any(&c->mf6c_origin) &&
2104 ipv6_addr_any(&c->mf6c_mcastgrp)) {
2105 if (true_vifi >= 0 &&
2106 true_vifi != c->_c.mfc_parent &&
2107 ipv6_hdr(skb)->hop_limit >
2108 c->_c.mfc_un.res.ttls[c->_c.mfc_parent]) {
2109
2110
2111
2112
2113 psend = c->_c.mfc_parent;
2114 goto last_forward;
2115 }
2116 goto dont_forward;
2117 }
2118 for (ct = c->_c.mfc_un.res.maxvif - 1;
2119 ct >= c->_c.mfc_un.res.minvif; ct--) {
2120
2121 if ((!ipv6_addr_any(&c->mf6c_origin) || ct != true_vifi) &&
2122 ipv6_hdr(skb)->hop_limit > c->_c.mfc_un.res.ttls[ct]) {
2123 if (psend != -1) {
2124 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
2125 if (skb2)
2126 ip6mr_forward2(net, mrt, skb2,
2127 c, psend);
2128 }
2129 psend = ct;
2130 }
2131 }
2132last_forward:
2133 if (psend != -1) {
2134 ip6mr_forward2(net, mrt, skb, c, psend);
2135 return;
2136 }
2137
2138dont_forward:
2139 kfree_skb(skb);
2140}
2141
2142
2143
2144
2145
2146
2147int ip6_mr_input(struct sk_buff *skb)
2148{
2149 struct mfc6_cache *cache;
2150 struct net *net = dev_net(skb->dev);
2151 struct mr_table *mrt;
2152 struct flowi6 fl6 = {
2153 .flowi6_iif = skb->dev->ifindex,
2154 .flowi6_mark = skb->mark,
2155 };
2156 int err;
2157
2158 err = ip6mr_fib_lookup(net, &fl6, &mrt);
2159 if (err < 0) {
2160 kfree_skb(skb);
2161 return err;
2162 }
2163
2164 read_lock(&mrt_lock);
2165 cache = ip6mr_cache_find(mrt,
2166 &ipv6_hdr(skb)->saddr, &ipv6_hdr(skb)->daddr);
2167 if (!cache) {
2168 int vif = ip6mr_find_vif(mrt, skb->dev);
2169
2170 if (vif >= 0)
2171 cache = ip6mr_cache_find_any(mrt,
2172 &ipv6_hdr(skb)->daddr,
2173 vif);
2174 }
2175
2176
2177
2178
2179 if (!cache) {
2180 int vif;
2181
2182 vif = ip6mr_find_vif(mrt, skb->dev);
2183 if (vif >= 0) {
2184 int err = ip6mr_cache_unresolved(mrt, vif, skb);
2185 read_unlock(&mrt_lock);
2186
2187 return err;
2188 }
2189 read_unlock(&mrt_lock);
2190 kfree_skb(skb);
2191 return -ENODEV;
2192 }
2193
2194 ip6_mr_forward(net, mrt, skb, cache);
2195
2196 read_unlock(&mrt_lock);
2197
2198 return 0;
2199}
2200
2201int ip6mr_get_route(struct net *net, struct sk_buff *skb, struct rtmsg *rtm,
2202 u32 portid)
2203{
2204 int err;
2205 struct mr_table *mrt;
2206 struct mfc6_cache *cache;
2207 struct rt6_info *rt = (struct rt6_info *)skb_dst(skb);
2208
2209 mrt = ip6mr_get_table(net, RT6_TABLE_DFLT);
2210 if (!mrt)
2211 return -ENOENT;
2212
2213 read_lock(&mrt_lock);
2214 cache = ip6mr_cache_find(mrt, &rt->rt6i_src.addr, &rt->rt6i_dst.addr);
2215 if (!cache && skb->dev) {
2216 int vif = ip6mr_find_vif(mrt, skb->dev);
2217
2218 if (vif >= 0)
2219 cache = ip6mr_cache_find_any(mrt, &rt->rt6i_dst.addr,
2220 vif);
2221 }
2222
2223 if (!cache) {
2224 struct sk_buff *skb2;
2225 struct ipv6hdr *iph;
2226 struct net_device *dev;
2227 int vif;
2228
2229 dev = skb->dev;
2230 if (!dev || (vif = ip6mr_find_vif(mrt, dev)) < 0) {
2231 read_unlock(&mrt_lock);
2232 return -ENODEV;
2233 }
2234
2235
2236 skb2 = alloc_skb(sizeof(struct ipv6hdr), GFP_ATOMIC);
2237 if (!skb2) {
2238 read_unlock(&mrt_lock);
2239 return -ENOMEM;
2240 }
2241
2242 NETLINK_CB(skb2).portid = portid;
2243 skb_reset_transport_header(skb2);
2244
2245 skb_put(skb2, sizeof(struct ipv6hdr));
2246 skb_reset_network_header(skb2);
2247
2248 iph = ipv6_hdr(skb2);
2249 iph->version = 0;
2250 iph->priority = 0;
2251 iph->flow_lbl[0] = 0;
2252 iph->flow_lbl[1] = 0;
2253 iph->flow_lbl[2] = 0;
2254 iph->payload_len = 0;
2255 iph->nexthdr = IPPROTO_NONE;
2256 iph->hop_limit = 0;
2257 iph->saddr = rt->rt6i_src.addr;
2258 iph->daddr = rt->rt6i_dst.addr;
2259
2260 err = ip6mr_cache_unresolved(mrt, vif, skb2);
2261 read_unlock(&mrt_lock);
2262
2263 return err;
2264 }
2265
2266 err = mr_fill_mroute(mrt, skb, &cache->_c, rtm);
2267 read_unlock(&mrt_lock);
2268 return err;
2269}
2270
2271static int ip6mr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb,
2272 u32 portid, u32 seq, struct mfc6_cache *c, int cmd,
2273 int flags)
2274{
2275 struct nlmsghdr *nlh;
2276 struct rtmsg *rtm;
2277 int err;
2278
2279 nlh = nlmsg_put(skb, portid, seq, cmd, sizeof(*rtm), flags);
2280 if (!nlh)
2281 return -EMSGSIZE;
2282
2283 rtm = nlmsg_data(nlh);
2284 rtm->rtm_family = RTNL_FAMILY_IP6MR;
2285 rtm->rtm_dst_len = 128;
2286 rtm->rtm_src_len = 128;
2287 rtm->rtm_tos = 0;
2288 rtm->rtm_table = mrt->id;
2289 if (nla_put_u32(skb, RTA_TABLE, mrt->id))
2290 goto nla_put_failure;
2291 rtm->rtm_type = RTN_MULTICAST;
2292 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
2293 if (c->_c.mfc_flags & MFC_STATIC)
2294 rtm->rtm_protocol = RTPROT_STATIC;
2295 else
2296 rtm->rtm_protocol = RTPROT_MROUTED;
2297 rtm->rtm_flags = 0;
2298
2299 if (nla_put_in6_addr(skb, RTA_SRC, &c->mf6c_origin) ||
2300 nla_put_in6_addr(skb, RTA_DST, &c->mf6c_mcastgrp))
2301 goto nla_put_failure;
2302 err = mr_fill_mroute(mrt, skb, &c->_c, rtm);
2303
2304 if (err < 0 && err != -ENOENT)
2305 goto nla_put_failure;
2306
2307 nlmsg_end(skb, nlh);
2308 return 0;
2309
2310nla_put_failure:
2311 nlmsg_cancel(skb, nlh);
2312 return -EMSGSIZE;
2313}
2314
2315static int _ip6mr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb,
2316 u32 portid, u32 seq, struct mr_mfc *c,
2317 int cmd, int flags)
2318{
2319 return ip6mr_fill_mroute(mrt, skb, portid, seq, (struct mfc6_cache *)c,
2320 cmd, flags);
2321}
2322
2323static int mr6_msgsize(bool unresolved, int maxvif)
2324{
2325 size_t len =
2326 NLMSG_ALIGN(sizeof(struct rtmsg))
2327 + nla_total_size(4)
2328 + nla_total_size(sizeof(struct in6_addr))
2329 + nla_total_size(sizeof(struct in6_addr))
2330 ;
2331
2332 if (!unresolved)
2333 len = len
2334 + nla_total_size(4)
2335 + nla_total_size(0)
2336 + maxvif * NLA_ALIGN(sizeof(struct rtnexthop))
2337
2338 + nla_total_size_64bit(sizeof(struct rta_mfc_stats))
2339 ;
2340
2341 return len;
2342}
2343
2344static void mr6_netlink_event(struct mr_table *mrt, struct mfc6_cache *mfc,
2345 int cmd)
2346{
2347 struct net *net = read_pnet(&mrt->net);
2348 struct sk_buff *skb;
2349 int err = -ENOBUFS;
2350
2351 skb = nlmsg_new(mr6_msgsize(mfc->_c.mfc_parent >= MAXMIFS, mrt->maxvif),
2352 GFP_ATOMIC);
2353 if (!skb)
2354 goto errout;
2355
2356 err = ip6mr_fill_mroute(mrt, skb, 0, 0, mfc, cmd, 0);
2357 if (err < 0)
2358 goto errout;
2359
2360 rtnl_notify(skb, net, 0, RTNLGRP_IPV6_MROUTE, NULL, GFP_ATOMIC);
2361 return;
2362
2363errout:
2364 kfree_skb(skb);
2365 if (err < 0)
2366 rtnl_set_sk_err(net, RTNLGRP_IPV6_MROUTE, err);
2367}
2368
2369static size_t mrt6msg_netlink_msgsize(size_t payloadlen)
2370{
2371 size_t len =
2372 NLMSG_ALIGN(sizeof(struct rtgenmsg))
2373 + nla_total_size(1)
2374 + nla_total_size(4)
2375
2376 + nla_total_size(sizeof(struct in6_addr))
2377
2378 + nla_total_size(sizeof(struct in6_addr))
2379
2380 + nla_total_size(payloadlen)
2381 ;
2382
2383 return len;
2384}
2385
2386static void mrt6msg_netlink_event(struct mr_table *mrt, struct sk_buff *pkt)
2387{
2388 struct net *net = read_pnet(&mrt->net);
2389 struct nlmsghdr *nlh;
2390 struct rtgenmsg *rtgenm;
2391 struct mrt6msg *msg;
2392 struct sk_buff *skb;
2393 struct nlattr *nla;
2394 int payloadlen;
2395
2396 payloadlen = pkt->len - sizeof(struct mrt6msg);
2397 msg = (struct mrt6msg *)skb_transport_header(pkt);
2398
2399 skb = nlmsg_new(mrt6msg_netlink_msgsize(payloadlen), GFP_ATOMIC);
2400 if (!skb)
2401 goto errout;
2402
2403 nlh = nlmsg_put(skb, 0, 0, RTM_NEWCACHEREPORT,
2404 sizeof(struct rtgenmsg), 0);
2405 if (!nlh)
2406 goto errout;
2407 rtgenm = nlmsg_data(nlh);
2408 rtgenm->rtgen_family = RTNL_FAMILY_IP6MR;
2409 if (nla_put_u8(skb, IP6MRA_CREPORT_MSGTYPE, msg->im6_msgtype) ||
2410 nla_put_u32(skb, IP6MRA_CREPORT_MIF_ID, msg->im6_mif) ||
2411 nla_put_in6_addr(skb, IP6MRA_CREPORT_SRC_ADDR,
2412 &msg->im6_src) ||
2413 nla_put_in6_addr(skb, IP6MRA_CREPORT_DST_ADDR,
2414 &msg->im6_dst))
2415 goto nla_put_failure;
2416
2417 nla = nla_reserve(skb, IP6MRA_CREPORT_PKT, payloadlen);
2418 if (!nla || skb_copy_bits(pkt, sizeof(struct mrt6msg),
2419 nla_data(nla), payloadlen))
2420 goto nla_put_failure;
2421
2422 nlmsg_end(skb, nlh);
2423
2424 rtnl_notify(skb, net, 0, RTNLGRP_IPV6_MROUTE_R, NULL, GFP_ATOMIC);
2425 return;
2426
2427nla_put_failure:
2428 nlmsg_cancel(skb, nlh);
2429errout:
2430 kfree_skb(skb);
2431 rtnl_set_sk_err(net, RTNLGRP_IPV6_MROUTE_R, -ENOBUFS);
2432}
2433
2434static int ip6mr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb)
2435{
2436 return mr_rtm_dumproute(skb, cb, ip6mr_mr_table_iter,
2437 _ip6mr_fill_mroute, &mfc_unres_lock);
2438}
2439
