1
2
3
4
5
6
7
8
9
10
11
12
13
14#include <linux/uaccess.h>
15#include <linux/types.h>
16#include <linux/sched.h>
17#include <linux/errno.h>
18#include <linux/mm.h>
19#include <linux/kernel.h>
20#include <linux/fcntl.h>
21#include <linux/stat.h>
22#include <linux/socket.h>
23#include <linux/inet.h>
24#include <linux/netdevice.h>
25#include <linux/inetdevice.h>
26#include <linux/proc_fs.h>
27#include <linux/seq_file.h>
28#include <linux/init.h>
29#include <linux/compat.h>
30#include <linux/rhashtable.h>
31#include <net/protocol.h>
32#include <linux/skbuff.h>
33#include <net/raw.h>
34#include <linux/notifier.h>
35#include <linux/if_arp.h>
36#include <net/checksum.h>
37#include <net/netlink.h>
38#include <net/fib_rules.h>
39
40#include <net/ipv6.h>
41#include <net/ip6_route.h>
42#include <linux/mroute6.h>
43#include <linux/pim.h>
44#include <net/addrconf.h>
45#include <linux/netfilter_ipv6.h>
46#include <linux/export.h>
47#include <net/ip6_checksum.h>
48#include <linux/netconf.h>
49#include <net/ip_tunnels.h>
50
51#include <linux/nospec.h>
52
53struct ip6mr_rule {
54 struct fib_rule common;
55};
56
57struct ip6mr_result {
58 struct mr_table *mrt;
59};
60
61
62
63
64
65static DEFINE_RWLOCK(mrt_lock);
66
67
68
69
70static DEFINE_SPINLOCK(mfc_unres_lock);
71
72
73
74
75
76
77
78
79
80static struct kmem_cache *mrt_cachep __read_mostly;
81
82static struct mr_table *ip6mr_new_table(struct net *net, u32 id);
83static void ip6mr_free_table(struct mr_table *mrt);
84
85static void ip6_mr_forward(struct net *net, struct mr_table *mrt,
86 struct net_device *dev, struct sk_buff *skb,
87 struct mfc6_cache *cache);
88static int ip6mr_cache_report(struct mr_table *mrt, struct sk_buff *pkt,
89 mifi_t mifi, int assert);
90static void mr6_netlink_event(struct mr_table *mrt, struct mfc6_cache *mfc,
91 int cmd);
92static void mrt6msg_netlink_event(struct mr_table *mrt, struct sk_buff *pkt);
93static int ip6mr_rtm_dumproute(struct sk_buff *skb,
94 struct netlink_callback *cb);
95static void mroute_clean_tables(struct mr_table *mrt, int flags);
96static void ipmr_expire_process(struct timer_list *t);
97
98#ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES
99#define ip6mr_for_each_table(mrt, net) \
100 list_for_each_entry_rcu(mrt, &net->ipv6.mr6_tables, list, \
101 lockdep_rtnl_is_held() || \
102 list_empty(&net->ipv6.mr6_tables))
103
104static struct mr_table *ip6mr_mr_table_iter(struct net *net,
105 struct mr_table *mrt)
106{
107 struct mr_table *ret;
108
109 if (!mrt)
110 ret = list_entry_rcu(net->ipv6.mr6_tables.next,
111 struct mr_table, list);
112 else
113 ret = list_entry_rcu(mrt->list.next,
114 struct mr_table, list);
115
116 if (&ret->list == &net->ipv6.mr6_tables)
117 return NULL;
118 return ret;
119}
120
121static struct mr_table *ip6mr_get_table(struct net *net, u32 id)
122{
123 struct mr_table *mrt;
124
125 ip6mr_for_each_table(mrt, net) {
126 if (mrt->id == id)
127 return mrt;
128 }
129 return NULL;
130}
131
132static int ip6mr_fib_lookup(struct net *net, struct flowi6 *flp6,
133 struct mr_table **mrt)
134{
135 int err;
136 struct ip6mr_result res;
137 struct fib_lookup_arg arg = {
138 .result = &res,
139 .flags = FIB_LOOKUP_NOREF,
140 };
141
142
143 l3mdev_update_flow(net, flowi6_to_flowi(flp6));
144
145 err = fib_rules_lookup(net->ipv6.mr6_rules_ops,
146 flowi6_to_flowi(flp6), 0, &arg);
147 if (err < 0)
148 return err;
149 *mrt = res.mrt;
150 return 0;
151}
152
153static int ip6mr_rule_action(struct fib_rule *rule, struct flowi *flp,
154 int flags, struct fib_lookup_arg *arg)
155{
156 struct ip6mr_result *res = arg->result;
157 struct mr_table *mrt;
158
159 switch (rule->action) {
160 case FR_ACT_TO_TBL:
161 break;
162 case FR_ACT_UNREACHABLE:
163 return -ENETUNREACH;
164 case FR_ACT_PROHIBIT:
165 return -EACCES;
166 case FR_ACT_BLACKHOLE:
167 default:
168 return -EINVAL;
169 }
170
171 arg->table = fib_rule_get_table(rule, arg);
172
173 mrt = ip6mr_get_table(rule->fr_net, arg->table);
174 if (!mrt)
175 return -EAGAIN;
176 res->mrt = mrt;
177 return 0;
178}
179
180static int ip6mr_rule_match(struct fib_rule *rule, struct flowi *flp, int flags)
181{
182 return 1;
183}
184
185static const struct nla_policy ip6mr_rule_policy[FRA_MAX + 1] = {
186 FRA_GENERIC_POLICY,
187};
188
189static int ip6mr_rule_configure(struct fib_rule *rule, struct sk_buff *skb,
190 struct fib_rule_hdr *frh, struct nlattr **tb,
191 struct netlink_ext_ack *extack)
192{
193 return 0;
194}
195
196static int ip6mr_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh,
197 struct nlattr **tb)
198{
199 return 1;
200}
201
202static int ip6mr_rule_fill(struct fib_rule *rule, struct sk_buff *skb,
203 struct fib_rule_hdr *frh)
204{
205 frh->dst_len = 0;
206 frh->src_len = 0;
207 frh->tos = 0;
208 return 0;
209}
210
211static const struct fib_rules_ops __net_initconst ip6mr_rules_ops_template = {
212 .family = RTNL_FAMILY_IP6MR,
213 .rule_size = sizeof(struct ip6mr_rule),
214 .addr_size = sizeof(struct in6_addr),
215 .action = ip6mr_rule_action,
216 .match = ip6mr_rule_match,
217 .configure = ip6mr_rule_configure,
218 .compare = ip6mr_rule_compare,
219 .fill = ip6mr_rule_fill,
220 .nlgroup = RTNLGRP_IPV6_RULE,
221 .policy = ip6mr_rule_policy,
222 .owner = THIS_MODULE,
223};
224
225static int __net_init ip6mr_rules_init(struct net *net)
226{
227 struct fib_rules_ops *ops;
228 struct mr_table *mrt;
229 int err;
230
231 ops = fib_rules_register(&ip6mr_rules_ops_template, net);
232 if (IS_ERR(ops))
233 return PTR_ERR(ops);
234
235 INIT_LIST_HEAD(&net->ipv6.mr6_tables);
236
237 mrt = ip6mr_new_table(net, RT6_TABLE_DFLT);
238 if (IS_ERR(mrt)) {
239 err = PTR_ERR(mrt);
240 goto err1;
241 }
242
243 err = fib_default_rule_add(ops, 0x7fff, RT6_TABLE_DFLT, 0);
244 if (err < 0)
245 goto err2;
246
247 net->ipv6.mr6_rules_ops = ops;
248 return 0;
249
250err2:
251 ip6mr_free_table(mrt);
252err1:
253 fib_rules_unregister(ops);
254 return err;
255}
256
257static void __net_exit ip6mr_rules_exit(struct net *net)
258{
259 struct mr_table *mrt, *next;
260
261 rtnl_lock();
262 list_for_each_entry_safe(mrt, next, &net->ipv6.mr6_tables, list) {
263 list_del(&mrt->list);
264 ip6mr_free_table(mrt);
265 }
266 fib_rules_unregister(net->ipv6.mr6_rules_ops);
267 rtnl_unlock();
268}
269
270static int ip6mr_rules_dump(struct net *net, struct notifier_block *nb,
271 struct netlink_ext_ack *extack)
272{
273 return fib_rules_dump(net, nb, RTNL_FAMILY_IP6MR, extack);
274}
275
276static unsigned int ip6mr_rules_seq_read(struct net *net)
277{
278 return fib_rules_seq_read(net, RTNL_FAMILY_IP6MR);
279}
280
281bool ip6mr_rule_default(const struct fib_rule *rule)
282{
283 return fib_rule_matchall(rule) && rule->action == FR_ACT_TO_TBL &&
284 rule->table == RT6_TABLE_DFLT && !rule->l3mdev;
285}
286EXPORT_SYMBOL(ip6mr_rule_default);
287#else
288#define ip6mr_for_each_table(mrt, net) \
289 for (mrt = net->ipv6.mrt6; mrt; mrt = NULL)
290
291static struct mr_table *ip6mr_mr_table_iter(struct net *net,
292 struct mr_table *mrt)
293{
294 if (!mrt)
295 return net->ipv6.mrt6;
296 return NULL;
297}
298
299static struct mr_table *ip6mr_get_table(struct net *net, u32 id)
300{
301 return net->ipv6.mrt6;
302}
303
304static int ip6mr_fib_lookup(struct net *net, struct flowi6 *flp6,
305 struct mr_table **mrt)
306{
307 *mrt = net->ipv6.mrt6;
308 return 0;
309}
310
311static int __net_init ip6mr_rules_init(struct net *net)
312{
313 struct mr_table *mrt;
314
315 mrt = ip6mr_new_table(net, RT6_TABLE_DFLT);
316 if (IS_ERR(mrt))
317 return PTR_ERR(mrt);
318 net->ipv6.mrt6 = mrt;
319 return 0;
320}
321
322static void __net_exit ip6mr_rules_exit(struct net *net)
323{
324 rtnl_lock();
325 ip6mr_free_table(net->ipv6.mrt6);
326 net->ipv6.mrt6 = NULL;
327 rtnl_unlock();
328}
329
330static int ip6mr_rules_dump(struct net *net, struct notifier_block *nb,
331 struct netlink_ext_ack *extack)
332{
333 return 0;
334}
335
336static unsigned int ip6mr_rules_seq_read(struct net *net)
337{
338 return 0;
339}
340#endif
341
342static int ip6mr_hash_cmp(struct rhashtable_compare_arg *arg,
343 const void *ptr)
344{
345 const struct mfc6_cache_cmp_arg *cmparg = arg->key;
346 struct mfc6_cache *c = (struct mfc6_cache *)ptr;
347
348 return !ipv6_addr_equal(&c->mf6c_mcastgrp, &cmparg->mf6c_mcastgrp) ||
349 !ipv6_addr_equal(&c->mf6c_origin, &cmparg->mf6c_origin);
350}
351
352static const struct rhashtable_params ip6mr_rht_params = {
353 .head_offset = offsetof(struct mr_mfc, mnode),
354 .key_offset = offsetof(struct mfc6_cache, cmparg),
355 .key_len = sizeof(struct mfc6_cache_cmp_arg),
356 .nelem_hint = 3,
357 .obj_cmpfn = ip6mr_hash_cmp,
358 .automatic_shrinking = true,
359};
360
361static void ip6mr_new_table_set(struct mr_table *mrt,
362 struct net *net)
363{
364#ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES
365 list_add_tail_rcu(&mrt->list, &net->ipv6.mr6_tables);
366#endif
367}
368
369static struct mfc6_cache_cmp_arg ip6mr_mr_table_ops_cmparg_any = {
370 .mf6c_origin = IN6ADDR_ANY_INIT,
371 .mf6c_mcastgrp = IN6ADDR_ANY_INIT,
372};
373
374static struct mr_table_ops ip6mr_mr_table_ops = {
375 .rht_params = &ip6mr_rht_params,
376 .cmparg_any = &ip6mr_mr_table_ops_cmparg_any,
377};
378
379static struct mr_table *ip6mr_new_table(struct net *net, u32 id)
380{
381 struct mr_table *mrt;
382
383 mrt = ip6mr_get_table(net, id);
384 if (mrt)
385 return mrt;
386
387 return mr_table_alloc(net, id, &ip6mr_mr_table_ops,
388 ipmr_expire_process, ip6mr_new_table_set);
389}
390
391static void ip6mr_free_table(struct mr_table *mrt)
392{
393 del_timer_sync(&mrt->ipmr_expire_timer);
394 mroute_clean_tables(mrt, MRT6_FLUSH_MIFS | MRT6_FLUSH_MIFS_STATIC |
395 MRT6_FLUSH_MFC | MRT6_FLUSH_MFC_STATIC);
396 rhltable_destroy(&mrt->mfc_hash);
397 kfree(mrt);
398}
399
400#ifdef CONFIG_PROC_FS
401
402
403
404
405static void *ip6mr_vif_seq_start(struct seq_file *seq, loff_t *pos)
406 __acquires(mrt_lock)
407{
408 struct mr_vif_iter *iter = seq->private;
409 struct net *net = seq_file_net(seq);
410 struct mr_table *mrt;
411
412 mrt = ip6mr_get_table(net, RT6_TABLE_DFLT);
413 if (!mrt)
414 return ERR_PTR(-ENOENT);
415
416 iter->mrt = mrt;
417
418 read_lock(&mrt_lock);
419 return mr_vif_seq_start(seq, pos);
420}
421
422static void ip6mr_vif_seq_stop(struct seq_file *seq, void *v)
423 __releases(mrt_lock)
424{
425 read_unlock(&mrt_lock);
426}
427
428static int ip6mr_vif_seq_show(struct seq_file *seq, void *v)
429{
430 struct mr_vif_iter *iter = seq->private;
431 struct mr_table *mrt = iter->mrt;
432
433 if (v == SEQ_START_TOKEN) {
434 seq_puts(seq,
435 "Interface BytesIn PktsIn BytesOut PktsOut Flags\n");
436 } else {
437 const struct vif_device *vif = v;
438 const char *name = vif->dev ? vif->dev->name : "none";
439
440 seq_printf(seq,
441 "%2td %-10s %8ld %7ld %8ld %7ld %05X\n",
442 vif - mrt->vif_table,
443 name, vif->bytes_in, vif->pkt_in,
444 vif->bytes_out, vif->pkt_out,
445 vif->flags);
446 }
447 return 0;
448}
449
450static const struct seq_operations ip6mr_vif_seq_ops = {
451 .start = ip6mr_vif_seq_start,
452 .next = mr_vif_seq_next,
453 .stop = ip6mr_vif_seq_stop,
454 .show = ip6mr_vif_seq_show,
455};
456
457static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos)
458{
459 struct net *net = seq_file_net(seq);
460 struct mr_table *mrt;
461
462 mrt = ip6mr_get_table(net, RT6_TABLE_DFLT);
463 if (!mrt)
464 return ERR_PTR(-ENOENT);
465
466 return mr_mfc_seq_start(seq, pos, mrt, &mfc_unres_lock);
467}
468
469static int ipmr_mfc_seq_show(struct seq_file *seq, void *v)
470{
471 int n;
472
473 if (v == SEQ_START_TOKEN) {
474 seq_puts(seq,
475 "Group "
476 "Origin "
477 "Iif Pkts Bytes Wrong Oifs\n");
478 } else {
479 const struct mfc6_cache *mfc = v;
480 const struct mr_mfc_iter *it = seq->private;
481 struct mr_table *mrt = it->mrt;
482
483 seq_printf(seq, "%pI6 %pI6 %-3hd",
484 &mfc->mf6c_mcastgrp, &mfc->mf6c_origin,
485 mfc->_c.mfc_parent);
486
487 if (it->cache != &mrt->mfc_unres_queue) {
488 seq_printf(seq, " %8lu %8lu %8lu",
489 mfc->_c.mfc_un.res.pkt,
490 mfc->_c.mfc_un.res.bytes,
491 mfc->_c.mfc_un.res.wrong_if);
492 for (n = mfc->_c.mfc_un.res.minvif;
493 n < mfc->_c.mfc_un.res.maxvif; n++) {
494 if (VIF_EXISTS(mrt, n) &&
495 mfc->_c.mfc_un.res.ttls[n] < 255)
496 seq_printf(seq,
497 " %2d:%-3d", n,
498 mfc->_c.mfc_un.res.ttls[n]);
499 }
500 } else {
501
502
503
504 seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul);
505 }
506 seq_putc(seq, '\n');
507 }
508 return 0;
509}
510
511static const struct seq_operations ipmr_mfc_seq_ops = {
512 .start = ipmr_mfc_seq_start,
513 .next = mr_mfc_seq_next,
514 .stop = mr_mfc_seq_stop,
515 .show = ipmr_mfc_seq_show,
516};
517#endif
518
519#ifdef CONFIG_IPV6_PIMSM_V2
520
521static int pim6_rcv(struct sk_buff *skb)
522{
523 struct pimreghdr *pim;
524 struct ipv6hdr *encap;
525 struct net_device *reg_dev = NULL;
526 struct net *net = dev_net(skb->dev);
527 struct mr_table *mrt;
528 struct flowi6 fl6 = {
529 .flowi6_iif = skb->dev->ifindex,
530 .flowi6_mark = skb->mark,
531 };
532 int reg_vif_num;
533
534 if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(*encap)))
535 goto drop;
536
537 pim = (struct pimreghdr *)skb_transport_header(skb);
538 if (pim->type != ((PIM_VERSION << 4) | PIM_TYPE_REGISTER) ||
539 (pim->flags & PIM_NULL_REGISTER) ||
540 (csum_ipv6_magic(&ipv6_hdr(skb)->saddr, &ipv6_hdr(skb)->daddr,
541 sizeof(*pim), IPPROTO_PIM,
542 csum_partial((void *)pim, sizeof(*pim), 0)) &&
543 csum_fold(skb_checksum(skb, 0, skb->len, 0))))
544 goto drop;
545
546
547 encap = (struct ipv6hdr *)(skb_transport_header(skb) +
548 sizeof(*pim));
549
550 if (!ipv6_addr_is_multicast(&encap->daddr) ||
551 encap->payload_len == 0 ||
552 ntohs(encap->payload_len) + sizeof(*pim) > skb->len)
553 goto drop;
554
555 if (ip6mr_fib_lookup(net, &fl6, &mrt) < 0)
556 goto drop;
557 reg_vif_num = mrt->mroute_reg_vif_num;
558
559 read_lock(&mrt_lock);
560 if (reg_vif_num >= 0)
561 reg_dev = mrt->vif_table[reg_vif_num].dev;
562 dev_hold(reg_dev);
563 read_unlock(&mrt_lock);
564
565 if (!reg_dev)
566 goto drop;
567
568 skb->mac_header = skb->network_header;
569 skb_pull(skb, (u8 *)encap - skb->data);
570 skb_reset_network_header(skb);
571 skb->protocol = htons(ETH_P_IPV6);
572 skb->ip_summed = CHECKSUM_NONE;
573
574 skb_tunnel_rx(skb, reg_dev, dev_net(reg_dev));
575
576 netif_rx(skb);
577
578 dev_put(reg_dev);
579 return 0;
580 drop:
581 kfree_skb(skb);
582 return 0;
583}
584
585static const struct inet6_protocol pim6_protocol = {
586 .handler = pim6_rcv,
587};
588
589
590
591static netdev_tx_t reg_vif_xmit(struct sk_buff *skb,
592 struct net_device *dev)
593{
594 struct net *net = dev_net(dev);
595 struct mr_table *mrt;
596 struct flowi6 fl6 = {
597 .flowi6_oif = dev->ifindex,
598 .flowi6_iif = skb->skb_iif ? : LOOPBACK_IFINDEX,
599 .flowi6_mark = skb->mark,
600 };
601
602 if (!pskb_inet_may_pull(skb))
603 goto tx_err;
604
605 if (ip6mr_fib_lookup(net, &fl6, &mrt) < 0)
606 goto tx_err;
607
608 read_lock(&mrt_lock);
609 dev->stats.tx_bytes += skb->len;
610 dev->stats.tx_packets++;
611 ip6mr_cache_report(mrt, skb, mrt->mroute_reg_vif_num, MRT6MSG_WHOLEPKT);
612 read_unlock(&mrt_lock);
613 kfree_skb(skb);
614 return NETDEV_TX_OK;
615
616tx_err:
617 dev->stats.tx_errors++;
618 kfree_skb(skb);
619 return NETDEV_TX_OK;
620}
621
622static int reg_vif_get_iflink(const struct net_device *dev)
623{
624 return 0;
625}
626
627static const struct net_device_ops reg_vif_netdev_ops = {
628 .ndo_start_xmit = reg_vif_xmit,
629 .ndo_get_iflink = reg_vif_get_iflink,
630};
631
632static void reg_vif_setup(struct net_device *dev)
633{
634 dev->type = ARPHRD_PIMREG;
635 dev->mtu = 1500 - sizeof(struct ipv6hdr) - 8;
636 dev->flags = IFF_NOARP;
637 dev->netdev_ops = ®_vif_netdev_ops;
638 dev->needs_free_netdev = true;
639 dev->features |= NETIF_F_NETNS_LOCAL;
640}
641
642static struct net_device *ip6mr_reg_vif(struct net *net, struct mr_table *mrt)
643{
644 struct net_device *dev;
645 char name[IFNAMSIZ];
646
647 if (mrt->id == RT6_TABLE_DFLT)
648 sprintf(name, "pim6reg");
649 else
650 sprintf(name, "pim6reg%u", mrt->id);
651
652 dev = alloc_netdev(0, name, NET_NAME_UNKNOWN, reg_vif_setup);
653 if (!dev)
654 return NULL;
655
656 dev_net_set(dev, net);
657
658 if (register_netdevice(dev)) {
659 free_netdev(dev);
660 return NULL;
661 }
662
663 if (dev_open(dev, NULL))
664 goto failure;
665
666 dev_hold(dev);
667 return dev;
668
669failure:
670 unregister_netdevice(dev);
671 return NULL;
672}
673#endif
674
675static int call_ip6mr_vif_entry_notifiers(struct net *net,
676 enum fib_event_type event_type,
677 struct vif_device *vif,
678 mifi_t vif_index, u32 tb_id)
679{
680 return mr_call_vif_notifiers(net, RTNL_FAMILY_IP6MR, event_type,
681 vif, vif_index, tb_id,
682 &net->ipv6.ipmr_seq);
683}
684
685static int call_ip6mr_mfc_entry_notifiers(struct net *net,
686 enum fib_event_type event_type,
687 struct mfc6_cache *mfc, u32 tb_id)
688{
689 return mr_call_mfc_notifiers(net, RTNL_FAMILY_IP6MR, event_type,
690 &mfc->_c, tb_id, &net->ipv6.ipmr_seq);
691}
692
693
694static int mif6_delete(struct mr_table *mrt, int vifi, int notify,
695 struct list_head *head)
696{
697 struct vif_device *v;
698 struct net_device *dev;
699 struct inet6_dev *in6_dev;
700
701 if (vifi < 0 || vifi >= mrt->maxvif)
702 return -EADDRNOTAVAIL;
703
704 v = &mrt->vif_table[vifi];
705
706 if (VIF_EXISTS(mrt, vifi))
707 call_ip6mr_vif_entry_notifiers(read_pnet(&mrt->net),
708 FIB_EVENT_VIF_DEL, v, vifi,
709 mrt->id);
710
711 write_lock_bh(&mrt_lock);
712 dev = v->dev;
713 v->dev = NULL;
714
715 if (!dev) {
716 write_unlock_bh(&mrt_lock);
717 return -EADDRNOTAVAIL;
718 }
719
720#ifdef CONFIG_IPV6_PIMSM_V2
721 if (vifi == mrt->mroute_reg_vif_num)
722 mrt->mroute_reg_vif_num = -1;
723#endif
724
725 if (vifi + 1 == mrt->maxvif) {
726 int tmp;
727 for (tmp = vifi - 1; tmp >= 0; tmp--) {
728 if (VIF_EXISTS(mrt, tmp))
729 break;
730 }
731 mrt->maxvif = tmp + 1;
732 }
733
734 write_unlock_bh(&mrt_lock);
735
736 dev_set_allmulti(dev, -1);
737
738 in6_dev = __in6_dev_get(dev);
739 if (in6_dev) {
740 in6_dev->cnf.mc_forwarding--;
741 inet6_netconf_notify_devconf(dev_net(dev), RTM_NEWNETCONF,
742 NETCONFA_MC_FORWARDING,
743 dev->ifindex, &in6_dev->cnf);
744 }
745
746 if ((v->flags & MIFF_REGISTER) && !notify)
747 unregister_netdevice_queue(dev, head);
748
749 dev_put(dev);
750 return 0;
751}
752
753static inline void ip6mr_cache_free_rcu(struct rcu_head *head)
754{
755 struct mr_mfc *c = container_of(head, struct mr_mfc, rcu);
756
757 kmem_cache_free(mrt_cachep, (struct mfc6_cache *)c);
758}
759
760static inline void ip6mr_cache_free(struct mfc6_cache *c)
761{
762 call_rcu(&c->_c.rcu, ip6mr_cache_free_rcu);
763}
764
765
766
767
768
769static void ip6mr_destroy_unres(struct mr_table *mrt, struct mfc6_cache *c)
770{
771 struct net *net = read_pnet(&mrt->net);
772 struct sk_buff *skb;
773
774 atomic_dec(&mrt->cache_resolve_queue_len);
775
776 while ((skb = skb_dequeue(&c->_c.mfc_un.unres.unresolved)) != NULL) {
777 if (ipv6_hdr(skb)->version == 0) {
778 struct nlmsghdr *nlh = skb_pull(skb,
779 sizeof(struct ipv6hdr));
780 nlh->nlmsg_type = NLMSG_ERROR;
781 nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr));
782 skb_trim(skb, nlh->nlmsg_len);
783 ((struct nlmsgerr *)nlmsg_data(nlh))->error = -ETIMEDOUT;
784 rtnl_unicast(skb, net, NETLINK_CB(skb).portid);
785 } else
786 kfree_skb(skb);
787 }
788
789 ip6mr_cache_free(c);
790}
791
792
793
794
795static void ipmr_do_expire_process(struct mr_table *mrt)
796{
797 unsigned long now = jiffies;
798 unsigned long expires = 10 * HZ;
799 struct mr_mfc *c, *next;
800
801 list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) {
802 if (time_after(c->mfc_un.unres.expires, now)) {
803
804 unsigned long interval = c->mfc_un.unres.expires - now;
805 if (interval < expires)
806 expires = interval;
807 continue;
808 }
809
810 list_del(&c->list);
811 mr6_netlink_event(mrt, (struct mfc6_cache *)c, RTM_DELROUTE);
812 ip6mr_destroy_unres(mrt, (struct mfc6_cache *)c);
813 }
814
815 if (!list_empty(&mrt->mfc_unres_queue))
816 mod_timer(&mrt->ipmr_expire_timer, jiffies + expires);
817}
818
819static void ipmr_expire_process(struct timer_list *t)
820{
821 struct mr_table *mrt = from_timer(mrt, t, ipmr_expire_timer);
822
823 if (!spin_trylock(&mfc_unres_lock)) {
824 mod_timer(&mrt->ipmr_expire_timer, jiffies + 1);
825 return;
826 }
827
828 if (!list_empty(&mrt->mfc_unres_queue))
829 ipmr_do_expire_process(mrt);
830
831 spin_unlock(&mfc_unres_lock);
832}
833
834
835
836static void ip6mr_update_thresholds(struct mr_table *mrt,
837 struct mr_mfc *cache,
838 unsigned char *ttls)
839{
840 int vifi;
841
842 cache->mfc_un.res.minvif = MAXMIFS;
843 cache->mfc_un.res.maxvif = 0;
844 memset(cache->mfc_un.res.ttls, 255, MAXMIFS);
845
846 for (vifi = 0; vifi < mrt->maxvif; vifi++) {
847 if (VIF_EXISTS(mrt, vifi) &&
848 ttls[vifi] && ttls[vifi] < 255) {
849 cache->mfc_un.res.ttls[vifi] = ttls[vifi];
850 if (cache->mfc_un.res.minvif > vifi)
851 cache->mfc_un.res.minvif = vifi;
852 if (cache->mfc_un.res.maxvif <= vifi)
853 cache->mfc_un.res.maxvif = vifi + 1;
854 }
855 }
856 cache->mfc_un.res.lastuse = jiffies;
857}
858
859static int mif6_add(struct net *net, struct mr_table *mrt,
860 struct mif6ctl *vifc, int mrtsock)
861{
862 int vifi = vifc->mif6c_mifi;
863 struct vif_device *v = &mrt->vif_table[vifi];
864 struct net_device *dev;
865 struct inet6_dev *in6_dev;
866 int err;
867
868
869 if (VIF_EXISTS(mrt, vifi))
870 return -EADDRINUSE;
871
872 switch (vifc->mif6c_flags) {
873#ifdef CONFIG_IPV6_PIMSM_V2
874 case MIFF_REGISTER:
875
876
877
878
879 if (mrt->mroute_reg_vif_num >= 0)
880 return -EADDRINUSE;
881 dev = ip6mr_reg_vif(net, mrt);
882 if (!dev)
883 return -ENOBUFS;
884 err = dev_set_allmulti(dev, 1);
885 if (err) {
886 unregister_netdevice(dev);
887 dev_put(dev);
888 return err;
889 }
890 break;
891#endif
892 case 0:
893 dev = dev_get_by_index(net, vifc->mif6c_pifi);
894 if (!dev)
895 return -EADDRNOTAVAIL;
896 err = dev_set_allmulti(dev, 1);
897 if (err) {
898 dev_put(dev);
899 return err;
900 }
901 break;
902 default:
903 return -EINVAL;
904 }
905
906 in6_dev = __in6_dev_get(dev);
907 if (in6_dev) {
908 in6_dev->cnf.mc_forwarding++;
909 inet6_netconf_notify_devconf(dev_net(dev), RTM_NEWNETCONF,
910 NETCONFA_MC_FORWARDING,
911 dev->ifindex, &in6_dev->cnf);
912 }
913
914
915 vif_device_init(v, dev, vifc->vifc_rate_limit, vifc->vifc_threshold,
916 vifc->mif6c_flags | (!mrtsock ? VIFF_STATIC : 0),
917 MIFF_REGISTER);
918
919
920 write_lock_bh(&mrt_lock);
921 v->dev = dev;
922#ifdef CONFIG_IPV6_PIMSM_V2
923 if (v->flags & MIFF_REGISTER)
924 mrt->mroute_reg_vif_num = vifi;
925#endif
926 if (vifi + 1 > mrt->maxvif)
927 mrt->maxvif = vifi + 1;
928 write_unlock_bh(&mrt_lock);
929 call_ip6mr_vif_entry_notifiers(net, FIB_EVENT_VIF_ADD,
930 v, vifi, mrt->id);
931 return 0;
932}
933
934static struct mfc6_cache *ip6mr_cache_find(struct mr_table *mrt,
935 const struct in6_addr *origin,
936 const struct in6_addr *mcastgrp)
937{
938 struct mfc6_cache_cmp_arg arg = {
939 .mf6c_origin = *origin,
940 .mf6c_mcastgrp = *mcastgrp,
941 };
942
943 return mr_mfc_find(mrt, &arg);
944}
945
946
947static struct mfc6_cache *ip6mr_cache_find_any(struct mr_table *mrt,
948 struct in6_addr *mcastgrp,
949 mifi_t mifi)
950{
951 struct mfc6_cache_cmp_arg arg = {
952 .mf6c_origin = in6addr_any,
953 .mf6c_mcastgrp = *mcastgrp,
954 };
955
956 if (ipv6_addr_any(mcastgrp))
957 return mr_mfc_find_any_parent(mrt, mifi);
958 return mr_mfc_find_any(mrt, mifi, &arg);
959}
960
961
962static struct mfc6_cache *
963ip6mr_cache_find_parent(struct mr_table *mrt,
964 const struct in6_addr *origin,
965 const struct in6_addr *mcastgrp,
966 int parent)
967{
968 struct mfc6_cache_cmp_arg arg = {
969 .mf6c_origin = *origin,
970 .mf6c_mcastgrp = *mcastgrp,
971 };
972
973 return mr_mfc_find_parent(mrt, &arg, parent);
974}
975
976
977static struct mfc6_cache *ip6mr_cache_alloc(void)
978{
979 struct mfc6_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL);
980 if (!c)
981 return NULL;
982 c->_c.mfc_un.res.last_assert = jiffies - MFC_ASSERT_THRESH - 1;
983 c->_c.mfc_un.res.minvif = MAXMIFS;
984 c->_c.free = ip6mr_cache_free_rcu;
985 refcount_set(&c->_c.mfc_un.res.refcount, 1);
986 return c;
987}
988
989static struct mfc6_cache *ip6mr_cache_alloc_unres(void)
990{
991 struct mfc6_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC);
992 if (!c)
993 return NULL;
994 skb_queue_head_init(&c->_c.mfc_un.unres.unresolved);
995 c->_c.mfc_un.unres.expires = jiffies + 10 * HZ;
996 return c;
997}
998
999
1000
1001
1002
1003static void ip6mr_cache_resolve(struct net *net, struct mr_table *mrt,
1004 struct mfc6_cache *uc, struct mfc6_cache *c)
1005{
1006 struct sk_buff *skb;
1007
1008
1009
1010
1011
1012 while ((skb = __skb_dequeue(&uc->_c.mfc_un.unres.unresolved))) {
1013 if (ipv6_hdr(skb)->version == 0) {
1014 struct nlmsghdr *nlh = skb_pull(skb,
1015 sizeof(struct ipv6hdr));
1016
1017 if (mr_fill_mroute(mrt, skb, &c->_c,
1018 nlmsg_data(nlh)) > 0) {
1019 nlh->nlmsg_len = skb_tail_pointer(skb) - (u8 *)nlh;
1020 } else {
1021 nlh->nlmsg_type = NLMSG_ERROR;
1022 nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr));
1023 skb_trim(skb, nlh->nlmsg_len);
1024 ((struct nlmsgerr *)nlmsg_data(nlh))->error = -EMSGSIZE;
1025 }
1026 rtnl_unicast(skb, net, NETLINK_CB(skb).portid);
1027 } else
1028 ip6_mr_forward(net, mrt, skb->dev, skb, c);
1029 }
1030}
1031
1032
1033
1034
1035
1036
1037
1038static int ip6mr_cache_report(struct mr_table *mrt, struct sk_buff *pkt,
1039 mifi_t mifi, int assert)
1040{
1041 struct sock *mroute6_sk;
1042 struct sk_buff *skb;
1043 struct mrt6msg *msg;
1044 int ret;
1045
1046#ifdef CONFIG_IPV6_PIMSM_V2
1047 if (assert == MRT6MSG_WHOLEPKT)
1048 skb = skb_realloc_headroom(pkt, -skb_network_offset(pkt)
1049 +sizeof(*msg));
1050 else
1051#endif
1052 skb = alloc_skb(sizeof(struct ipv6hdr) + sizeof(*msg), GFP_ATOMIC);
1053
1054 if (!skb)
1055 return -ENOBUFS;
1056
1057
1058
1059
1060 skb->ip_summed = CHECKSUM_UNNECESSARY;
1061
1062#ifdef CONFIG_IPV6_PIMSM_V2
1063 if (assert == MRT6MSG_WHOLEPKT) {
1064
1065
1066
1067
1068
1069 skb_push(skb, -skb_network_offset(pkt));
1070
1071 skb_push(skb, sizeof(*msg));
1072 skb_reset_transport_header(skb);
1073 msg = (struct mrt6msg *)skb_transport_header(skb);
1074 msg->im6_mbz = 0;
1075 msg->im6_msgtype = MRT6MSG_WHOLEPKT;
1076 msg->im6_mif = mrt->mroute_reg_vif_num;
1077 msg->im6_pad = 0;
1078 msg->im6_src = ipv6_hdr(pkt)->saddr;
1079 msg->im6_dst = ipv6_hdr(pkt)->daddr;
1080
1081 skb->ip_summed = CHECKSUM_UNNECESSARY;
1082 } else
1083#endif
1084 {
1085
1086
1087
1088
1089 skb_put(skb, sizeof(struct ipv6hdr));
1090 skb_reset_network_header(skb);
1091 skb_copy_to_linear_data(skb, ipv6_hdr(pkt), sizeof(struct ipv6hdr));
1092
1093
1094
1095
1096 skb_put(skb, sizeof(*msg));
1097 skb_reset_transport_header(skb);
1098 msg = (struct mrt6msg *)skb_transport_header(skb);
1099
1100 msg->im6_mbz = 0;
1101 msg->im6_msgtype = assert;
1102 msg->im6_mif = mifi;
1103 msg->im6_pad = 0;
1104 msg->im6_src = ipv6_hdr(pkt)->saddr;
1105 msg->im6_dst = ipv6_hdr(pkt)->daddr;
1106
1107 skb_dst_set(skb, dst_clone(skb_dst(pkt)));
1108 skb->ip_summed = CHECKSUM_UNNECESSARY;
1109 }
1110
1111 rcu_read_lock();
1112 mroute6_sk = rcu_dereference(mrt->mroute_sk);
1113 if (!mroute6_sk) {
1114 rcu_read_unlock();
1115 kfree_skb(skb);
1116 return -EINVAL;
1117 }
1118
1119 mrt6msg_netlink_event(mrt, skb);
1120
1121
1122 ret = sock_queue_rcv_skb(mroute6_sk, skb);
1123 rcu_read_unlock();
1124 if (ret < 0) {
1125 net_warn_ratelimited("mroute6: pending queue full, dropping entries\n");
1126 kfree_skb(skb);
1127 }
1128
1129 return ret;
1130}
1131
1132
1133static int ip6mr_cache_unresolved(struct mr_table *mrt, mifi_t mifi,
1134 struct sk_buff *skb, struct net_device *dev)
1135{
1136 struct mfc6_cache *c;
1137 bool found = false;
1138 int err;
1139
1140 spin_lock_bh(&mfc_unres_lock);
1141 list_for_each_entry(c, &mrt->mfc_unres_queue, _c.list) {
1142 if (ipv6_addr_equal(&c->mf6c_mcastgrp, &ipv6_hdr(skb)->daddr) &&
1143 ipv6_addr_equal(&c->mf6c_origin, &ipv6_hdr(skb)->saddr)) {
1144 found = true;
1145 break;
1146 }
1147 }
1148
1149 if (!found) {
1150
1151
1152
1153
1154 c = ip6mr_cache_alloc_unres();
1155 if (!c) {
1156 spin_unlock_bh(&mfc_unres_lock);
1157
1158 kfree_skb(skb);
1159 return -ENOBUFS;
1160 }
1161
1162
1163 c->_c.mfc_parent = -1;
1164 c->mf6c_origin = ipv6_hdr(skb)->saddr;
1165 c->mf6c_mcastgrp = ipv6_hdr(skb)->daddr;
1166
1167
1168
1169
1170 err = ip6mr_cache_report(mrt, skb, mifi, MRT6MSG_NOCACHE);
1171 if (err < 0) {
1172
1173
1174
1175 spin_unlock_bh(&mfc_unres_lock);
1176
1177 ip6mr_cache_free(c);
1178 kfree_skb(skb);
1179 return err;
1180 }
1181
1182 atomic_inc(&mrt->cache_resolve_queue_len);
1183 list_add(&c->_c.list, &mrt->mfc_unres_queue);
1184 mr6_netlink_event(mrt, c, RTM_NEWROUTE);
1185
1186 ipmr_do_expire_process(mrt);
1187 }
1188
1189
1190 if (c->_c.mfc_un.unres.unresolved.qlen > 3) {
1191 kfree_skb(skb);
1192 err = -ENOBUFS;
1193 } else {
1194 if (dev) {
1195 skb->dev = dev;
1196 skb->skb_iif = dev->ifindex;
1197 }
1198 skb_queue_tail(&c->_c.mfc_un.unres.unresolved, skb);
1199 err = 0;
1200 }
1201
1202 spin_unlock_bh(&mfc_unres_lock);
1203 return err;
1204}
1205
1206
1207
1208
1209
1210static int ip6mr_mfc_delete(struct mr_table *mrt, struct mf6cctl *mfc,
1211 int parent)
1212{
1213 struct mfc6_cache *c;
1214
1215
1216 rcu_read_lock();
1217 c = ip6mr_cache_find_parent(mrt, &mfc->mf6cc_origin.sin6_addr,
1218 &mfc->mf6cc_mcastgrp.sin6_addr, parent);
1219 rcu_read_unlock();
1220 if (!c)
1221 return -ENOENT;
1222 rhltable_remove(&mrt->mfc_hash, &c->_c.mnode, ip6mr_rht_params);
1223 list_del_rcu(&c->_c.list);
1224
1225 call_ip6mr_mfc_entry_notifiers(read_pnet(&mrt->net),
1226 FIB_EVENT_ENTRY_DEL, c, mrt->id);
1227 mr6_netlink_event(mrt, c, RTM_DELROUTE);
1228 mr_cache_put(&c->_c);
1229 return 0;
1230}
1231
1232static int ip6mr_device_event(struct notifier_block *this,
1233 unsigned long event, void *ptr)
1234{
1235 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
1236 struct net *net = dev_net(dev);
1237 struct mr_table *mrt;
1238 struct vif_device *v;
1239 int ct;
1240
1241 if (event != NETDEV_UNREGISTER)
1242 return NOTIFY_DONE;
1243
1244 ip6mr_for_each_table(mrt, net) {
1245 v = &mrt->vif_table[0];
1246 for (ct = 0; ct < mrt->maxvif; ct++, v++) {
1247 if (v->dev == dev)
1248 mif6_delete(mrt, ct, 1, NULL);
1249 }
1250 }
1251
1252 return NOTIFY_DONE;
1253}
1254
1255static unsigned int ip6mr_seq_read(struct net *net)
1256{
1257 ASSERT_RTNL();
1258
1259 return net->ipv6.ipmr_seq + ip6mr_rules_seq_read(net);
1260}
1261
1262static int ip6mr_dump(struct net *net, struct notifier_block *nb,
1263 struct netlink_ext_ack *extack)
1264{
1265 return mr_dump(net, nb, RTNL_FAMILY_IP6MR, ip6mr_rules_dump,
1266 ip6mr_mr_table_iter, &mrt_lock, extack);
1267}
1268
1269static struct notifier_block ip6_mr_notifier = {
1270 .notifier_call = ip6mr_device_event
1271};
1272
1273static const struct fib_notifier_ops ip6mr_notifier_ops_template = {
1274 .family = RTNL_FAMILY_IP6MR,
1275 .fib_seq_read = ip6mr_seq_read,
1276 .fib_dump = ip6mr_dump,
1277 .owner = THIS_MODULE,
1278};
1279
1280static int __net_init ip6mr_notifier_init(struct net *net)
1281{
1282 struct fib_notifier_ops *ops;
1283
1284 net->ipv6.ipmr_seq = 0;
1285
1286 ops = fib_notifier_ops_register(&ip6mr_notifier_ops_template, net);
1287 if (IS_ERR(ops))
1288 return PTR_ERR(ops);
1289
1290 net->ipv6.ip6mr_notifier_ops = ops;
1291
1292 return 0;
1293}
1294
1295static void __net_exit ip6mr_notifier_exit(struct net *net)
1296{
1297 fib_notifier_ops_unregister(net->ipv6.ip6mr_notifier_ops);
1298 net->ipv6.ip6mr_notifier_ops = NULL;
1299}
1300
1301
1302static int __net_init ip6mr_net_init(struct net *net)
1303{
1304 int err;
1305
1306 err = ip6mr_notifier_init(net);
1307 if (err)
1308 return err;
1309
1310 err = ip6mr_rules_init(net);
1311 if (err < 0)
1312 goto ip6mr_rules_fail;
1313
1314#ifdef CONFIG_PROC_FS
1315 err = -ENOMEM;
1316 if (!proc_create_net("ip6_mr_vif", 0, net->proc_net, &ip6mr_vif_seq_ops,
1317 sizeof(struct mr_vif_iter)))
1318 goto proc_vif_fail;
1319 if (!proc_create_net("ip6_mr_cache", 0, net->proc_net, &ipmr_mfc_seq_ops,
1320 sizeof(struct mr_mfc_iter)))
1321 goto proc_cache_fail;
1322#endif
1323
1324 return 0;
1325
1326#ifdef CONFIG_PROC_FS
1327proc_cache_fail:
1328 remove_proc_entry("ip6_mr_vif", net->proc_net);
1329proc_vif_fail:
1330 ip6mr_rules_exit(net);
1331#endif
1332ip6mr_rules_fail:
1333 ip6mr_notifier_exit(net);
1334 return err;
1335}
1336
1337static void __net_exit ip6mr_net_exit(struct net *net)
1338{
1339#ifdef CONFIG_PROC_FS
1340 remove_proc_entry("ip6_mr_cache", net->proc_net);
1341 remove_proc_entry("ip6_mr_vif", net->proc_net);
1342#endif
1343 ip6mr_rules_exit(net);
1344 ip6mr_notifier_exit(net);
1345}
1346
1347static struct pernet_operations ip6mr_net_ops = {
1348 .init = ip6mr_net_init,
1349 .exit = ip6mr_net_exit,
1350};
1351
1352int __init ip6_mr_init(void)
1353{
1354 int err;
1355
1356 mrt_cachep = kmem_cache_create("ip6_mrt_cache",
1357 sizeof(struct mfc6_cache),
1358 0, SLAB_HWCACHE_ALIGN,
1359 NULL);
1360 if (!mrt_cachep)
1361 return -ENOMEM;
1362
1363 err = register_pernet_subsys(&ip6mr_net_ops);
1364 if (err)
1365 goto reg_pernet_fail;
1366
1367 err = register_netdevice_notifier(&ip6_mr_notifier);
1368 if (err)
1369 goto reg_notif_fail;
1370#ifdef CONFIG_IPV6_PIMSM_V2
1371 if (inet6_add_protocol(&pim6_protocol, IPPROTO_PIM) < 0) {
1372 pr_err("%s: can't add PIM protocol\n", __func__);
1373 err = -EAGAIN;
1374 goto add_proto_fail;
1375 }
1376#endif
1377 err = rtnl_register_module(THIS_MODULE, RTNL_FAMILY_IP6MR, RTM_GETROUTE,
1378 NULL, ip6mr_rtm_dumproute, 0);
1379 if (err == 0)
1380 return 0;
1381
1382#ifdef CONFIG_IPV6_PIMSM_V2
1383 inet6_del_protocol(&pim6_protocol, IPPROTO_PIM);
1384add_proto_fail:
1385 unregister_netdevice_notifier(&ip6_mr_notifier);
1386#endif
1387reg_notif_fail:
1388 unregister_pernet_subsys(&ip6mr_net_ops);
1389reg_pernet_fail:
1390 kmem_cache_destroy(mrt_cachep);
1391 return err;
1392}
1393
1394void ip6_mr_cleanup(void)
1395{
1396 rtnl_unregister(RTNL_FAMILY_IP6MR, RTM_GETROUTE);
1397#ifdef CONFIG_IPV6_PIMSM_V2
1398 inet6_del_protocol(&pim6_protocol, IPPROTO_PIM);
1399#endif
1400 unregister_netdevice_notifier(&ip6_mr_notifier);
1401 unregister_pernet_subsys(&ip6mr_net_ops);
1402 kmem_cache_destroy(mrt_cachep);
1403}
1404
1405static int ip6mr_mfc_add(struct net *net, struct mr_table *mrt,
1406 struct mf6cctl *mfc, int mrtsock, int parent)
1407{
1408 unsigned char ttls[MAXMIFS];
1409 struct mfc6_cache *uc, *c;
1410 struct mr_mfc *_uc;
1411 bool found;
1412 int i, err;
1413
1414 if (mfc->mf6cc_parent >= MAXMIFS)
1415 return -ENFILE;
1416
1417 memset(ttls, 255, MAXMIFS);
1418 for (i = 0; i < MAXMIFS; i++) {
1419 if (IF_ISSET(i, &mfc->mf6cc_ifset))
1420 ttls[i] = 1;
1421 }
1422
1423
1424 rcu_read_lock();
1425 c = ip6mr_cache_find_parent(mrt, &mfc->mf6cc_origin.sin6_addr,
1426 &mfc->mf6cc_mcastgrp.sin6_addr, parent);
1427 rcu_read_unlock();
1428 if (c) {
1429 write_lock_bh(&mrt_lock);
1430 c->_c.mfc_parent = mfc->mf6cc_parent;
1431 ip6mr_update_thresholds(mrt, &c->_c, ttls);
1432 if (!mrtsock)
1433 c->_c.mfc_flags |= MFC_STATIC;
1434 write_unlock_bh(&mrt_lock);
1435 call_ip6mr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_REPLACE,
1436 c, mrt->id);
1437 mr6_netlink_event(mrt, c, RTM_NEWROUTE);
1438 return 0;
1439 }
1440
1441 if (!ipv6_addr_any(&mfc->mf6cc_mcastgrp.sin6_addr) &&
1442 !ipv6_addr_is_multicast(&mfc->mf6cc_mcastgrp.sin6_addr))
1443 return -EINVAL;
1444
1445 c = ip6mr_cache_alloc();
1446 if (!c)
1447 return -ENOMEM;
1448
1449 c->mf6c_origin = mfc->mf6cc_origin.sin6_addr;
1450 c->mf6c_mcastgrp = mfc->mf6cc_mcastgrp.sin6_addr;
1451 c->_c.mfc_parent = mfc->mf6cc_parent;
1452 ip6mr_update_thresholds(mrt, &c->_c, ttls);
1453 if (!mrtsock)
1454 c->_c.mfc_flags |= MFC_STATIC;
1455
1456 err = rhltable_insert_key(&mrt->mfc_hash, &c->cmparg, &c->_c.mnode,
1457 ip6mr_rht_params);
1458 if (err) {
1459 pr_err("ip6mr: rhtable insert error %d\n", err);
1460 ip6mr_cache_free(c);
1461 return err;
1462 }
1463 list_add_tail_rcu(&c->_c.list, &mrt->mfc_cache_list);
1464
1465
1466
1467
1468 found = false;
1469 spin_lock_bh(&mfc_unres_lock);
1470 list_for_each_entry(_uc, &mrt->mfc_unres_queue, list) {
1471 uc = (struct mfc6_cache *)_uc;
1472 if (ipv6_addr_equal(&uc->mf6c_origin, &c->mf6c_origin) &&
1473 ipv6_addr_equal(&uc->mf6c_mcastgrp, &c->mf6c_mcastgrp)) {
1474 list_del(&_uc->list);
1475 atomic_dec(&mrt->cache_resolve_queue_len);
1476 found = true;
1477 break;
1478 }
1479 }
1480 if (list_empty(&mrt->mfc_unres_queue))
1481 del_timer(&mrt->ipmr_expire_timer);
1482 spin_unlock_bh(&mfc_unres_lock);
1483
1484 if (found) {
1485 ip6mr_cache_resolve(net, mrt, uc, c);
1486 ip6mr_cache_free(uc);
1487 }
1488 call_ip6mr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_ADD,
1489 c, mrt->id);
1490 mr6_netlink_event(mrt, c, RTM_NEWROUTE);
1491 return 0;
1492}
1493
1494
1495
1496
1497
1498static void mroute_clean_tables(struct mr_table *mrt, int flags)
1499{
1500 struct mr_mfc *c, *tmp;
1501 LIST_HEAD(list);
1502 int i;
1503
1504
1505 if (flags & (MRT6_FLUSH_MIFS | MRT6_FLUSH_MIFS_STATIC)) {
1506 for (i = 0; i < mrt->maxvif; i++) {
1507 if (((mrt->vif_table[i].flags & VIFF_STATIC) &&
1508 !(flags & MRT6_FLUSH_MIFS_STATIC)) ||
1509 (!(mrt->vif_table[i].flags & VIFF_STATIC) && !(flags & MRT6_FLUSH_MIFS)))
1510 continue;
1511 mif6_delete(mrt, i, 0, &list);
1512 }
1513 unregister_netdevice_many(&list);
1514 }
1515
1516
1517 if (flags & (MRT6_FLUSH_MFC | MRT6_FLUSH_MFC_STATIC)) {
1518 list_for_each_entry_safe(c, tmp, &mrt->mfc_cache_list, list) {
1519 if (((c->mfc_flags & MFC_STATIC) && !(flags & MRT6_FLUSH_MFC_STATIC)) ||
1520 (!(c->mfc_flags & MFC_STATIC) && !(flags & MRT6_FLUSH_MFC)))
1521 continue;
1522 rhltable_remove(&mrt->mfc_hash, &c->mnode, ip6mr_rht_params);
1523 list_del_rcu(&c->list);
1524 call_ip6mr_mfc_entry_notifiers(read_pnet(&mrt->net),
1525 FIB_EVENT_ENTRY_DEL,
1526 (struct mfc6_cache *)c, mrt->id);
1527 mr6_netlink_event(mrt, (struct mfc6_cache *)c, RTM_DELROUTE);
1528 mr_cache_put(c);
1529 }
1530 }
1531
1532 if (flags & MRT6_FLUSH_MFC) {
1533 if (atomic_read(&mrt->cache_resolve_queue_len) != 0) {
1534 spin_lock_bh(&mfc_unres_lock);
1535 list_for_each_entry_safe(c, tmp, &mrt->mfc_unres_queue, list) {
1536 list_del(&c->list);
1537 mr6_netlink_event(mrt, (struct mfc6_cache *)c,
1538 RTM_DELROUTE);
1539 ip6mr_destroy_unres(mrt, (struct mfc6_cache *)c);
1540 }
1541 spin_unlock_bh(&mfc_unres_lock);
1542 }
1543 }
1544}
1545
1546static int ip6mr_sk_init(struct mr_table *mrt, struct sock *sk)
1547{
1548 int err = 0;
1549 struct net *net = sock_net(sk);
1550
1551 rtnl_lock();
1552 write_lock_bh(&mrt_lock);
1553 if (rtnl_dereference(mrt->mroute_sk)) {
1554 err = -EADDRINUSE;
1555 } else {
1556 rcu_assign_pointer(mrt->mroute_sk, sk);
1557 sock_set_flag(sk, SOCK_RCU_FREE);
1558 net->ipv6.devconf_all->mc_forwarding++;
1559 }
1560 write_unlock_bh(&mrt_lock);
1561
1562 if (!err)
1563 inet6_netconf_notify_devconf(net, RTM_NEWNETCONF,
1564 NETCONFA_MC_FORWARDING,
1565 NETCONFA_IFINDEX_ALL,
1566 net->ipv6.devconf_all);
1567 rtnl_unlock();
1568
1569 return err;
1570}
1571
1572int ip6mr_sk_done(struct sock *sk)
1573{
1574 int err = -EACCES;
1575 struct net *net = sock_net(sk);
1576 struct mr_table *mrt;
1577
1578 if (sk->sk_type != SOCK_RAW ||
1579 inet_sk(sk)->inet_num != IPPROTO_ICMPV6)
1580 return err;
1581
1582 rtnl_lock();
1583 ip6mr_for_each_table(mrt, net) {
1584 if (sk == rtnl_dereference(mrt->mroute_sk)) {
1585 write_lock_bh(&mrt_lock);
1586 RCU_INIT_POINTER(mrt->mroute_sk, NULL);
1587
1588
1589
1590
1591 net->ipv6.devconf_all->mc_forwarding--;
1592 write_unlock_bh(&mrt_lock);
1593 inet6_netconf_notify_devconf(net, RTM_NEWNETCONF,
1594 NETCONFA_MC_FORWARDING,
1595 NETCONFA_IFINDEX_ALL,
1596 net->ipv6.devconf_all);
1597
1598 mroute_clean_tables(mrt, MRT6_FLUSH_MIFS | MRT6_FLUSH_MFC);
1599 err = 0;
1600 break;
1601 }
1602 }
1603 rtnl_unlock();
1604
1605 return err;
1606}
1607
1608bool mroute6_is_socket(struct net *net, struct sk_buff *skb)
1609{
1610 struct mr_table *mrt;
1611 struct flowi6 fl6 = {
1612 .flowi6_iif = skb->skb_iif ? : LOOPBACK_IFINDEX,
1613 .flowi6_oif = skb->dev->ifindex,
1614 .flowi6_mark = skb->mark,
1615 };
1616
1617 if (ip6mr_fib_lookup(net, &fl6, &mrt) < 0)
1618 return NULL;
1619
1620 return rcu_access_pointer(mrt->mroute_sk);
1621}
1622EXPORT_SYMBOL(mroute6_is_socket);
1623
1624
1625
1626
1627
1628
1629
1630
1631int ip6_mroute_setsockopt(struct sock *sk, int optname, sockptr_t optval,
1632 unsigned int optlen)
1633{
1634 int ret, parent = 0;
1635 struct mif6ctl vif;
1636 struct mf6cctl mfc;
1637 mifi_t mifi;
1638 struct net *net = sock_net(sk);
1639 struct mr_table *mrt;
1640
1641 if (sk->sk_type != SOCK_RAW ||
1642 inet_sk(sk)->inet_num != IPPROTO_ICMPV6)
1643 return -EOPNOTSUPP;
1644
1645 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
1646 if (!mrt)
1647 return -ENOENT;
1648
1649 if (optname != MRT6_INIT) {
1650 if (sk != rcu_access_pointer(mrt->mroute_sk) &&
1651 !ns_capable(net->user_ns, CAP_NET_ADMIN))
1652 return -EACCES;
1653 }
1654
1655 switch (optname) {
1656 case MRT6_INIT:
1657 if (optlen < sizeof(int))
1658 return -EINVAL;
1659
1660 return ip6mr_sk_init(mrt, sk);
1661
1662 case MRT6_DONE:
1663 return ip6mr_sk_done(sk);
1664
1665 case MRT6_ADD_MIF:
1666 if (optlen < sizeof(vif))
1667 return -EINVAL;
1668 if (copy_from_sockptr(&vif, optval, sizeof(vif)))
1669 return -EFAULT;
1670 if (vif.mif6c_mifi >= MAXMIFS)
1671 return -ENFILE;
1672 rtnl_lock();
1673 ret = mif6_add(net, mrt, &vif,
1674 sk == rtnl_dereference(mrt->mroute_sk));
1675 rtnl_unlock();
1676 return ret;
1677
1678 case MRT6_DEL_MIF:
1679 if (optlen < sizeof(mifi_t))
1680 return -EINVAL;
1681 if (copy_from_sockptr(&mifi, optval, sizeof(mifi_t)))
1682 return -EFAULT;
1683 rtnl_lock();
1684 ret = mif6_delete(mrt, mifi, 0, NULL);
1685 rtnl_unlock();
1686 return ret;
1687
1688
1689
1690
1691
1692 case MRT6_ADD_MFC:
1693 case MRT6_DEL_MFC:
1694 parent = -1;
1695 fallthrough;
1696 case MRT6_ADD_MFC_PROXY:
1697 case MRT6_DEL_MFC_PROXY:
1698 if (optlen < sizeof(mfc))
1699 return -EINVAL;
1700 if (copy_from_sockptr(&mfc, optval, sizeof(mfc)))
1701 return -EFAULT;
1702 if (parent == 0)
1703 parent = mfc.mf6cc_parent;
1704 rtnl_lock();
1705 if (optname == MRT6_DEL_MFC || optname == MRT6_DEL_MFC_PROXY)
1706 ret = ip6mr_mfc_delete(mrt, &mfc, parent);
1707 else
1708 ret = ip6mr_mfc_add(net, mrt, &mfc,
1709 sk ==
1710 rtnl_dereference(mrt->mroute_sk),
1711 parent);
1712 rtnl_unlock();
1713 return ret;
1714
1715 case MRT6_FLUSH:
1716 {
1717 int flags;
1718
1719 if (optlen != sizeof(flags))
1720 return -EINVAL;
1721 if (copy_from_sockptr(&flags, optval, sizeof(flags)))
1722 return -EFAULT;
1723 rtnl_lock();
1724 mroute_clean_tables(mrt, flags);
1725 rtnl_unlock();
1726 return 0;
1727 }
1728
1729
1730
1731
1732 case MRT6_ASSERT:
1733 {
1734 int v;
1735
1736 if (optlen != sizeof(v))
1737 return -EINVAL;
1738 if (copy_from_sockptr(&v, optval, sizeof(v)))
1739 return -EFAULT;
1740 mrt->mroute_do_assert = v;
1741 return 0;
1742 }
1743
1744#ifdef CONFIG_IPV6_PIMSM_V2
1745 case MRT6_PIM:
1746 {
1747 int v;
1748
1749 if (optlen != sizeof(v))
1750 return -EINVAL;
1751 if (copy_from_sockptr(&v, optval, sizeof(v)))
1752 return -EFAULT;
1753 v = !!v;
1754 rtnl_lock();
1755 ret = 0;
1756 if (v != mrt->mroute_do_pim) {
1757 mrt->mroute_do_pim = v;
1758 mrt->mroute_do_assert = v;
1759 }
1760 rtnl_unlock();
1761 return ret;
1762 }
1763
1764#endif
1765#ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES
1766 case MRT6_TABLE:
1767 {
1768 u32 v;
1769
1770 if (optlen != sizeof(u32))
1771 return -EINVAL;
1772 if (copy_from_sockptr(&v, optval, sizeof(v)))
1773 return -EFAULT;
1774
1775 if (v != RT_TABLE_DEFAULT && v >= 100000000)
1776 return -EINVAL;
1777 if (sk == rcu_access_pointer(mrt->mroute_sk))
1778 return -EBUSY;
1779
1780 rtnl_lock();
1781 ret = 0;
1782 mrt = ip6mr_new_table(net, v);
1783 if (IS_ERR(mrt))
1784 ret = PTR_ERR(mrt);
1785 else
1786 raw6_sk(sk)->ip6mr_table = v;
1787 rtnl_unlock();
1788 return ret;
1789 }
1790#endif
1791
1792
1793
1794
1795 default:
1796 return -ENOPROTOOPT;
1797 }
1798}
1799
1800
1801
1802
1803
1804int ip6_mroute_getsockopt(struct sock *sk, int optname, char __user *optval,
1805 int __user *optlen)
1806{
1807 int olr;
1808 int val;
1809 struct net *net = sock_net(sk);
1810 struct mr_table *mrt;
1811
1812 if (sk->sk_type != SOCK_RAW ||
1813 inet_sk(sk)->inet_num != IPPROTO_ICMPV6)
1814 return -EOPNOTSUPP;
1815
1816 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
1817 if (!mrt)
1818 return -ENOENT;
1819
1820 switch (optname) {
1821 case MRT6_VERSION:
1822 val = 0x0305;
1823 break;
1824#ifdef CONFIG_IPV6_PIMSM_V2
1825 case MRT6_PIM:
1826 val = mrt->mroute_do_pim;
1827 break;
1828#endif
1829 case MRT6_ASSERT:
1830 val = mrt->mroute_do_assert;
1831 break;
1832 default:
1833 return -ENOPROTOOPT;
1834 }
1835
1836 if (get_user(olr, optlen))
1837 return -EFAULT;
1838
1839 olr = min_t(int, olr, sizeof(int));
1840 if (olr < 0)
1841 return -EINVAL;
1842
1843 if (put_user(olr, optlen))
1844 return -EFAULT;
1845 if (copy_to_user(optval, &val, olr))
1846 return -EFAULT;
1847 return 0;
1848}
1849
1850
1851
1852
1853
1854int ip6mr_ioctl(struct sock *sk, int cmd, void __user *arg)
1855{
1856 struct sioc_sg_req6 sr;
1857 struct sioc_mif_req6 vr;
1858 struct vif_device *vif;
1859 struct mfc6_cache *c;
1860 struct net *net = sock_net(sk);
1861 struct mr_table *mrt;
1862
1863 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
1864 if (!mrt)
1865 return -ENOENT;
1866
1867 switch (cmd) {
1868 case SIOCGETMIFCNT_IN6:
1869 if (copy_from_user(&vr, arg, sizeof(vr)))
1870 return -EFAULT;
1871 if (vr.mifi >= mrt->maxvif)
1872 return -EINVAL;
1873 vr.mifi = array_index_nospec(vr.mifi, mrt->maxvif);
1874 read_lock(&mrt_lock);
1875 vif = &mrt->vif_table[vr.mifi];
1876 if (VIF_EXISTS(mrt, vr.mifi)) {
1877 vr.icount = vif->pkt_in;
1878 vr.ocount = vif->pkt_out;
1879 vr.ibytes = vif->bytes_in;
1880 vr.obytes = vif->bytes_out;
1881 read_unlock(&mrt_lock);
1882
1883 if (copy_to_user(arg, &vr, sizeof(vr)))
1884 return -EFAULT;
1885 return 0;
1886 }
1887 read_unlock(&mrt_lock);
1888 return -EADDRNOTAVAIL;
1889 case SIOCGETSGCNT_IN6:
1890 if (copy_from_user(&sr, arg, sizeof(sr)))
1891 return -EFAULT;
1892
1893 rcu_read_lock();
1894 c = ip6mr_cache_find(mrt, &sr.src.sin6_addr, &sr.grp.sin6_addr);
1895 if (c) {
1896 sr.pktcnt = c->_c.mfc_un.res.pkt;
1897 sr.bytecnt = c->_c.mfc_un.res.bytes;
1898 sr.wrong_if = c->_c.mfc_un.res.wrong_if;
1899 rcu_read_unlock();
1900
1901 if (copy_to_user(arg, &sr, sizeof(sr)))
1902 return -EFAULT;
1903 return 0;
1904 }
1905 rcu_read_unlock();
1906 return -EADDRNOTAVAIL;
1907 default:
1908 return -ENOIOCTLCMD;
1909 }
1910}
1911
1912#ifdef CONFIG_COMPAT
1913struct compat_sioc_sg_req6 {
1914 struct sockaddr_in6 src;
1915 struct sockaddr_in6 grp;
1916 compat_ulong_t pktcnt;
1917 compat_ulong_t bytecnt;
1918 compat_ulong_t wrong_if;
1919};
1920
1921struct compat_sioc_mif_req6 {
1922 mifi_t mifi;
1923 compat_ulong_t icount;
1924 compat_ulong_t ocount;
1925 compat_ulong_t ibytes;
1926 compat_ulong_t obytes;
1927};
1928
1929int ip6mr_compat_ioctl(struct sock *sk, unsigned int cmd, void __user *arg)
1930{
1931 struct compat_sioc_sg_req6 sr;
1932 struct compat_sioc_mif_req6 vr;
1933 struct vif_device *vif;
1934 struct mfc6_cache *c;
1935 struct net *net = sock_net(sk);
1936 struct mr_table *mrt;
1937
1938 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
1939 if (!mrt)
1940 return -ENOENT;
1941
1942 switch (cmd) {
1943 case SIOCGETMIFCNT_IN6:
1944 if (copy_from_user(&vr, arg, sizeof(vr)))
1945 return -EFAULT;
1946 if (vr.mifi >= mrt->maxvif)
1947 return -EINVAL;
1948 vr.mifi = array_index_nospec(vr.mifi, mrt->maxvif);
1949 read_lock(&mrt_lock);
1950 vif = &mrt->vif_table[vr.mifi];
1951 if (VIF_EXISTS(mrt, vr.mifi)) {
1952 vr.icount = vif->pkt_in;
1953 vr.ocount = vif->pkt_out;
1954 vr.ibytes = vif->bytes_in;
1955 vr.obytes = vif->bytes_out;
1956 read_unlock(&mrt_lock);
1957
1958 if (copy_to_user(arg, &vr, sizeof(vr)))
1959 return -EFAULT;
1960 return 0;
1961 }
1962 read_unlock(&mrt_lock);
1963 return -EADDRNOTAVAIL;
1964 case SIOCGETSGCNT_IN6:
1965 if (copy_from_user(&sr, arg, sizeof(sr)))
1966 return -EFAULT;
1967
1968 rcu_read_lock();
1969 c = ip6mr_cache_find(mrt, &sr.src.sin6_addr, &sr.grp.sin6_addr);
1970 if (c) {
1971 sr.pktcnt = c->_c.mfc_un.res.pkt;
1972 sr.bytecnt = c->_c.mfc_un.res.bytes;
1973 sr.wrong_if = c->_c.mfc_un.res.wrong_if;
1974 rcu_read_unlock();
1975
1976 if (copy_to_user(arg, &sr, sizeof(sr)))
1977 return -EFAULT;
1978 return 0;
1979 }
1980 rcu_read_unlock();
1981 return -EADDRNOTAVAIL;
1982 default:
1983 return -ENOIOCTLCMD;
1984 }
1985}
1986#endif
1987
1988static inline int ip6mr_forward2_finish(struct net *net, struct sock *sk, struct sk_buff *skb)
1989{
1990 IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
1991 IPSTATS_MIB_OUTFORWDATAGRAMS);
1992 IP6_ADD_STATS(net, ip6_dst_idev(skb_dst(skb)),
1993 IPSTATS_MIB_OUTOCTETS, skb->len);
1994 return dst_output(net, sk, skb);
1995}
1996
1997
1998
1999
2000
2001static int ip6mr_forward2(struct net *net, struct mr_table *mrt,
2002 struct sk_buff *skb, int vifi)
2003{
2004 struct ipv6hdr *ipv6h;
2005 struct vif_device *vif = &mrt->vif_table[vifi];
2006 struct net_device *dev;
2007 struct dst_entry *dst;
2008 struct flowi6 fl6;
2009
2010 if (!vif->dev)
2011 goto out_free;
2012
2013#ifdef CONFIG_IPV6_PIMSM_V2
2014 if (vif->flags & MIFF_REGISTER) {
2015 vif->pkt_out++;
2016 vif->bytes_out += skb->len;
2017 vif->dev->stats.tx_bytes += skb->len;
2018 vif->dev->stats.tx_packets++;
2019 ip6mr_cache_report(mrt, skb, vifi, MRT6MSG_WHOLEPKT);
2020 goto out_free;
2021 }
2022#endif
2023
2024 ipv6h = ipv6_hdr(skb);
2025
2026 fl6 = (struct flowi6) {
2027 .flowi6_oif = vif->link,
2028 .daddr = ipv6h->daddr,
2029 };
2030
2031 dst = ip6_route_output(net, NULL, &fl6);
2032 if (dst->error) {
2033 dst_release(dst);
2034 goto out_free;
2035 }
2036
2037 skb_dst_drop(skb);
2038 skb_dst_set(skb, dst);
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051 dev = vif->dev;
2052 skb->dev = dev;
2053 vif->pkt_out++;
2054 vif->bytes_out += skb->len;
2055
2056
2057
2058 if (skb_cow(skb, sizeof(*ipv6h) + LL_RESERVED_SPACE(dev)))
2059 goto out_free;
2060
2061 ipv6h = ipv6_hdr(skb);
2062 ipv6h->hop_limit--;
2063
2064 IP6CB(skb)->flags |= IP6SKB_FORWARDED;
2065
2066 return NF_HOOK(NFPROTO_IPV6, NF_INET_FORWARD,
2067 net, NULL, skb, skb->dev, dev,
2068 ip6mr_forward2_finish);
2069
2070out_free:
2071 kfree_skb(skb);
2072 return 0;
2073}
2074
2075static int ip6mr_find_vif(struct mr_table *mrt, struct net_device *dev)
2076{
2077 int ct;
2078
2079 for (ct = mrt->maxvif - 1; ct >= 0; ct--) {
2080 if (mrt->vif_table[ct].dev == dev)
2081 break;
2082 }
2083 return ct;
2084}
2085
2086static void ip6_mr_forward(struct net *net, struct mr_table *mrt,
2087 struct net_device *dev, struct sk_buff *skb,
2088 struct mfc6_cache *c)
2089{
2090 int psend = -1;
2091 int vif, ct;
2092 int true_vifi = ip6mr_find_vif(mrt, dev);
2093
2094 vif = c->_c.mfc_parent;
2095 c->_c.mfc_un.res.pkt++;
2096 c->_c.mfc_un.res.bytes += skb->len;
2097 c->_c.mfc_un.res.lastuse = jiffies;
2098
2099 if (ipv6_addr_any(&c->mf6c_origin) && true_vifi >= 0) {
2100 struct mfc6_cache *cache_proxy;
2101
2102
2103
2104
2105 rcu_read_lock();
2106 cache_proxy = mr_mfc_find_any_parent(mrt, vif);
2107 if (cache_proxy &&
2108 cache_proxy->_c.mfc_un.res.ttls[true_vifi] < 255) {
2109 rcu_read_unlock();
2110 goto forward;
2111 }
2112 rcu_read_unlock();
2113 }
2114
2115
2116
2117
2118 if (mrt->vif_table[vif].dev != dev) {
2119 c->_c.mfc_un.res.wrong_if++;
2120
2121 if (true_vifi >= 0 && mrt->mroute_do_assert &&
2122
2123
2124
2125
2126
2127 (mrt->mroute_do_pim ||
2128 c->_c.mfc_un.res.ttls[true_vifi] < 255) &&
2129 time_after(jiffies,
2130 c->_c.mfc_un.res.last_assert +
2131 MFC_ASSERT_THRESH)) {
2132 c->_c.mfc_un.res.last_assert = jiffies;
2133 ip6mr_cache_report(mrt, skb, true_vifi, MRT6MSG_WRONGMIF);
2134 }
2135 goto dont_forward;
2136 }
2137
2138forward:
2139 mrt->vif_table[vif].pkt_in++;
2140 mrt->vif_table[vif].bytes_in += skb->len;
2141
2142
2143
2144
2145 if (ipv6_addr_any(&c->mf6c_origin) &&
2146 ipv6_addr_any(&c->mf6c_mcastgrp)) {
2147 if (true_vifi >= 0 &&
2148 true_vifi != c->_c.mfc_parent &&
2149 ipv6_hdr(skb)->hop_limit >
2150 c->_c.mfc_un.res.ttls[c->_c.mfc_parent]) {
2151
2152
2153
2154
2155 psend = c->_c.mfc_parent;
2156 goto last_forward;
2157 }
2158 goto dont_forward;
2159 }
2160 for (ct = c->_c.mfc_un.res.maxvif - 1;
2161 ct >= c->_c.mfc_un.res.minvif; ct--) {
2162
2163 if ((!ipv6_addr_any(&c->mf6c_origin) || ct != true_vifi) &&
2164 ipv6_hdr(skb)->hop_limit > c->_c.mfc_un.res.ttls[ct]) {
2165 if (psend != -1) {
2166 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
2167 if (skb2)
2168 ip6mr_forward2(net, mrt, skb2, psend);
2169 }
2170 psend = ct;
2171 }
2172 }
2173last_forward:
2174 if (psend != -1) {
2175 ip6mr_forward2(net, mrt, skb, psend);
2176 return;
2177 }
2178
2179dont_forward:
2180 kfree_skb(skb);
2181}
2182
2183
2184
2185
2186
2187
2188int ip6_mr_input(struct sk_buff *skb)
2189{
2190 struct mfc6_cache *cache;
2191 struct net *net = dev_net(skb->dev);
2192 struct mr_table *mrt;
2193 struct flowi6 fl6 = {
2194 .flowi6_iif = skb->dev->ifindex,
2195 .flowi6_mark = skb->mark,
2196 };
2197 int err;
2198 struct net_device *dev;
2199
2200
2201
2202
2203 dev = skb->dev;
2204 if (netif_is_l3_master(skb->dev)) {
2205 dev = dev_get_by_index_rcu(net, IPCB(skb)->iif);
2206 if (!dev) {
2207 kfree_skb(skb);
2208 return -ENODEV;
2209 }
2210 }
2211
2212 err = ip6mr_fib_lookup(net, &fl6, &mrt);
2213 if (err < 0) {
2214 kfree_skb(skb);
2215 return err;
2216 }
2217
2218 read_lock(&mrt_lock);
2219 cache = ip6mr_cache_find(mrt,
2220 &ipv6_hdr(skb)->saddr, &ipv6_hdr(skb)->daddr);
2221 if (!cache) {
2222 int vif = ip6mr_find_vif(mrt, dev);
2223
2224 if (vif >= 0)
2225 cache = ip6mr_cache_find_any(mrt,
2226 &ipv6_hdr(skb)->daddr,
2227 vif);
2228 }
2229
2230
2231
2232
2233 if (!cache) {
2234 int vif;
2235
2236 vif = ip6mr_find_vif(mrt, dev);
2237 if (vif >= 0) {
2238 int err = ip6mr_cache_unresolved(mrt, vif, skb, dev);
2239 read_unlock(&mrt_lock);
2240
2241 return err;
2242 }
2243 read_unlock(&mrt_lock);
2244 kfree_skb(skb);
2245 return -ENODEV;
2246 }
2247
2248 ip6_mr_forward(net, mrt, dev, skb, cache);
2249
2250 read_unlock(&mrt_lock);
2251
2252 return 0;
2253}
2254
2255int ip6mr_get_route(struct net *net, struct sk_buff *skb, struct rtmsg *rtm,
2256 u32 portid)
2257{
2258 int err;
2259 struct mr_table *mrt;
2260 struct mfc6_cache *cache;
2261 struct rt6_info *rt = (struct rt6_info *)skb_dst(skb);
2262
2263 mrt = ip6mr_get_table(net, RT6_TABLE_DFLT);
2264 if (!mrt)
2265 return -ENOENT;
2266
2267 read_lock(&mrt_lock);
2268 cache = ip6mr_cache_find(mrt, &rt->rt6i_src.addr, &rt->rt6i_dst.addr);
2269 if (!cache && skb->dev) {
2270 int vif = ip6mr_find_vif(mrt, skb->dev);
2271
2272 if (vif >= 0)
2273 cache = ip6mr_cache_find_any(mrt, &rt->rt6i_dst.addr,
2274 vif);
2275 }
2276
2277 if (!cache) {
2278 struct sk_buff *skb2;
2279 struct ipv6hdr *iph;
2280 struct net_device *dev;
2281 int vif;
2282
2283 dev = skb->dev;
2284 if (!dev || (vif = ip6mr_find_vif(mrt, dev)) < 0) {
2285 read_unlock(&mrt_lock);
2286 return -ENODEV;
2287 }
2288
2289
2290 skb2 = alloc_skb(sizeof(struct ipv6hdr), GFP_ATOMIC);
2291 if (!skb2) {
2292 read_unlock(&mrt_lock);
2293 return -ENOMEM;
2294 }
2295
2296 NETLINK_CB(skb2).portid = portid;
2297 skb_reset_transport_header(skb2);
2298
2299 skb_put(skb2, sizeof(struct ipv6hdr));
2300 skb_reset_network_header(skb2);
2301
2302 iph = ipv6_hdr(skb2);
2303 iph->version = 0;
2304 iph->priority = 0;
2305 iph->flow_lbl[0] = 0;
2306 iph->flow_lbl[1] = 0;
2307 iph->flow_lbl[2] = 0;
2308 iph->payload_len = 0;
2309 iph->nexthdr = IPPROTO_NONE;
2310 iph->hop_limit = 0;
2311 iph->saddr = rt->rt6i_src.addr;
2312 iph->daddr = rt->rt6i_dst.addr;
2313
2314 err = ip6mr_cache_unresolved(mrt, vif, skb2, dev);
2315 read_unlock(&mrt_lock);
2316
2317 return err;
2318 }
2319
2320 err = mr_fill_mroute(mrt, skb, &cache->_c, rtm);
2321 read_unlock(&mrt_lock);
2322 return err;
2323}
2324
2325static int ip6mr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb,
2326 u32 portid, u32 seq, struct mfc6_cache *c, int cmd,
2327 int flags)
2328{
2329 struct nlmsghdr *nlh;
2330 struct rtmsg *rtm;
2331 int err;
2332
2333 nlh = nlmsg_put(skb, portid, seq, cmd, sizeof(*rtm), flags);
2334 if (!nlh)
2335 return -EMSGSIZE;
2336
2337 rtm = nlmsg_data(nlh);
2338 rtm->rtm_family = RTNL_FAMILY_IP6MR;
2339 rtm->rtm_dst_len = 128;
2340 rtm->rtm_src_len = 128;
2341 rtm->rtm_tos = 0;
2342 rtm->rtm_table = mrt->id;
2343 if (nla_put_u32(skb, RTA_TABLE, mrt->id))
2344 goto nla_put_failure;
2345 rtm->rtm_type = RTN_MULTICAST;
2346 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
2347 if (c->_c.mfc_flags & MFC_STATIC)
2348 rtm->rtm_protocol = RTPROT_STATIC;
2349 else
2350 rtm->rtm_protocol = RTPROT_MROUTED;
2351 rtm->rtm_flags = 0;
2352
2353 if (nla_put_in6_addr(skb, RTA_SRC, &c->mf6c_origin) ||
2354 nla_put_in6_addr(skb, RTA_DST, &c->mf6c_mcastgrp))
2355 goto nla_put_failure;
2356 err = mr_fill_mroute(mrt, skb, &c->_c, rtm);
2357
2358 if (err < 0 && err != -ENOENT)
2359 goto nla_put_failure;
2360
2361 nlmsg_end(skb, nlh);
2362 return 0;
2363
2364nla_put_failure:
2365 nlmsg_cancel(skb, nlh);
2366 return -EMSGSIZE;
2367}
2368
2369static int _ip6mr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb,
2370 u32 portid, u32 seq, struct mr_mfc *c,
2371 int cmd, int flags)
2372{
2373 return ip6mr_fill_mroute(mrt, skb, portid, seq, (struct mfc6_cache *)c,
2374 cmd, flags);
2375}
2376
2377static int mr6_msgsize(bool unresolved, int maxvif)
2378{
2379 size_t len =
2380 NLMSG_ALIGN(sizeof(struct rtmsg))
2381 + nla_total_size(4)
2382 + nla_total_size(sizeof(struct in6_addr))
2383 + nla_total_size(sizeof(struct in6_addr))
2384 ;
2385
2386 if (!unresolved)
2387 len = len
2388 + nla_total_size(4)
2389 + nla_total_size(0)
2390 + maxvif * NLA_ALIGN(sizeof(struct rtnexthop))
2391
2392 + nla_total_size_64bit(sizeof(struct rta_mfc_stats))
2393 ;
2394
2395 return len;
2396}
2397
2398static void mr6_netlink_event(struct mr_table *mrt, struct mfc6_cache *mfc,
2399 int cmd)
2400{
2401 struct net *net = read_pnet(&mrt->net);
2402 struct sk_buff *skb;
2403 int err = -ENOBUFS;
2404
2405 skb = nlmsg_new(mr6_msgsize(mfc->_c.mfc_parent >= MAXMIFS, mrt->maxvif),
2406 GFP_ATOMIC);
2407 if (!skb)
2408 goto errout;
2409
2410 err = ip6mr_fill_mroute(mrt, skb, 0, 0, mfc, cmd, 0);
2411 if (err < 0)
2412 goto errout;
2413
2414 rtnl_notify(skb, net, 0, RTNLGRP_IPV6_MROUTE, NULL, GFP_ATOMIC);
2415 return;
2416
2417errout:
2418 kfree_skb(skb);
2419 if (err < 0)
2420 rtnl_set_sk_err(net, RTNLGRP_IPV6_MROUTE, err);
2421}
2422
2423static size_t mrt6msg_netlink_msgsize(size_t payloadlen)
2424{
2425 size_t len =
2426 NLMSG_ALIGN(sizeof(struct rtgenmsg))
2427 + nla_total_size(1)
2428 + nla_total_size(4)
2429
2430 + nla_total_size(sizeof(struct in6_addr))
2431
2432 + nla_total_size(sizeof(struct in6_addr))
2433
2434 + nla_total_size(payloadlen)
2435 ;
2436
2437 return len;
2438}
2439
2440static void mrt6msg_netlink_event(struct mr_table *mrt, struct sk_buff *pkt)
2441{
2442 struct net *net = read_pnet(&mrt->net);
2443 struct nlmsghdr *nlh;
2444 struct rtgenmsg *rtgenm;
2445 struct mrt6msg *msg;
2446 struct sk_buff *skb;
2447 struct nlattr *nla;
2448 int payloadlen;
2449
2450 payloadlen = pkt->len - sizeof(struct mrt6msg);
2451 msg = (struct mrt6msg *)skb_transport_header(pkt);
2452
2453 skb = nlmsg_new(mrt6msg_netlink_msgsize(payloadlen), GFP_ATOMIC);
2454 if (!skb)
2455 goto errout;
2456
2457 nlh = nlmsg_put(skb, 0, 0, RTM_NEWCACHEREPORT,
2458 sizeof(struct rtgenmsg), 0);
2459 if (!nlh)
2460 goto errout;
2461 rtgenm = nlmsg_data(nlh);
2462 rtgenm->rtgen_family = RTNL_FAMILY_IP6MR;
2463 if (nla_put_u8(skb, IP6MRA_CREPORT_MSGTYPE, msg->im6_msgtype) ||
2464 nla_put_u32(skb, IP6MRA_CREPORT_MIF_ID, msg->im6_mif) ||
2465 nla_put_in6_addr(skb, IP6MRA_CREPORT_SRC_ADDR,
2466 &msg->im6_src) ||
2467 nla_put_in6_addr(skb, IP6MRA_CREPORT_DST_ADDR,
2468 &msg->im6_dst))
2469 goto nla_put_failure;
2470
2471 nla = nla_reserve(skb, IP6MRA_CREPORT_PKT, payloadlen);
2472 if (!nla || skb_copy_bits(pkt, sizeof(struct mrt6msg),
2473 nla_data(nla), payloadlen))
2474 goto nla_put_failure;
2475
2476 nlmsg_end(skb, nlh);
2477
2478 rtnl_notify(skb, net, 0, RTNLGRP_IPV6_MROUTE_R, NULL, GFP_ATOMIC);
2479 return;
2480
2481nla_put_failure:
2482 nlmsg_cancel(skb, nlh);
2483errout:
2484 kfree_skb(skb);
2485 rtnl_set_sk_err(net, RTNLGRP_IPV6_MROUTE_R, -ENOBUFS);
2486}
2487
2488static int ip6mr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb)
2489{
2490 const struct nlmsghdr *nlh = cb->nlh;
2491 struct fib_dump_filter filter = {};
2492 int err;
2493
2494 if (cb->strict_check) {
2495 err = ip_valid_fib_dump_req(sock_net(skb->sk), nlh,
2496 &filter, cb);
2497 if (err < 0)
2498 return err;
2499 }
2500
2501 if (filter.table_id) {
2502 struct mr_table *mrt;
2503
2504 mrt = ip6mr_get_table(sock_net(skb->sk), filter.table_id);
2505 if (!mrt) {
2506 if (rtnl_msg_family(cb->nlh) != RTNL_FAMILY_IP6MR)
2507 return skb->len;
2508
2509 NL_SET_ERR_MSG_MOD(cb->extack, "MR table does not exist");
2510 return -ENOENT;
2511 }
2512 err = mr_table_dump(mrt, skb, cb, _ip6mr_fill_mroute,
2513 &mfc_unres_lock, &filter);
2514 return skb->len ? : err;
2515 }
2516
2517 return mr_rtm_dumproute(skb, cb, ip6mr_mr_table_iter,
2518 _ip6mr_fill_mroute, &mfc_unres_lock, &filter);
2519}
2520
