1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19#include <asm/uaccess.h>
20#include <linux/types.h>
21#include <linux/sched.h>
22#include <linux/errno.h>
23#include <linux/timer.h>
24#include <linux/mm.h>
25#include <linux/kernel.h>
26#include <linux/fcntl.h>
27#include <linux/stat.h>
28#include <linux/socket.h>
29#include <linux/inet.h>
30#include <linux/netdevice.h>
31#include <linux/inetdevice.h>
32#include <linux/proc_fs.h>
33#include <linux/seq_file.h>
34#include <linux/init.h>
35#include <linux/slab.h>
36#include <linux/compat.h>
37#include <net/protocol.h>
38#include <linux/skbuff.h>
39#include <net/sock.h>
40#include <net/raw.h>
41#include <linux/notifier.h>
42#include <linux/if_arp.h>
43#include <net/checksum.h>
44#include <net/netlink.h>
45#include <net/fib_rules.h>
46
47#include <net/ipv6.h>
48#include <net/ip6_route.h>
49#include <linux/mroute6.h>
50#include <linux/pim.h>
51#include <net/addrconf.h>
52#include <linux/netfilter_ipv6.h>
53#include <linux/export.h>
54#include <net/ip6_checksum.h>
55#include <linux/netconf.h>
56
57struct mr6_table {
58 struct list_head list;
59 possible_net_t net;
60 u32 id;
61 struct sock *mroute6_sk;
62 struct timer_list ipmr_expire_timer;
63 struct list_head mfc6_unres_queue;
64 struct list_head mfc6_cache_array[MFC6_LINES];
65 struct mif_device vif6_table[MAXMIFS];
66 int maxvif;
67 atomic_t cache_resolve_queue_len;
68 bool mroute_do_assert;
69 bool mroute_do_pim;
70#ifdef CONFIG_IPV6_PIMSM_V2
71 int mroute_reg_vif_num;
72#endif
73};
74
75struct ip6mr_rule {
76 struct fib_rule common;
77};
78
79struct ip6mr_result {
80 struct mr6_table *mrt;
81};
82
83
84
85
86
87static DEFINE_RWLOCK(mrt_lock);
88
89
90
91
92
93#define MIF_EXISTS(_mrt, _idx) ((_mrt)->vif6_table[_idx].dev != NULL)
94
95
96static DEFINE_SPINLOCK(mfc_unres_lock);
97
98
99
100
101
102
103
104
105
106static struct kmem_cache *mrt_cachep __read_mostly;
107
108static struct mr6_table *ip6mr_new_table(struct net *net, u32 id);
109static void ip6mr_free_table(struct mr6_table *mrt);
110
111static void ip6_mr_forward(struct net *net, struct mr6_table *mrt,
112 struct sk_buff *skb, struct mfc6_cache *cache);
113static int ip6mr_cache_report(struct mr6_table *mrt, struct sk_buff *pkt,
114 mifi_t mifi, int assert);
115static int __ip6mr_fill_mroute(struct mr6_table *mrt, struct sk_buff *skb,
116 struct mfc6_cache *c, struct rtmsg *rtm);
117static void mr6_netlink_event(struct mr6_table *mrt, struct mfc6_cache *mfc,
118 int cmd);
119static int ip6mr_rtm_dumproute(struct sk_buff *skb,
120 struct netlink_callback *cb);
121static void mroute_clean_tables(struct mr6_table *mrt);
122static void ipmr_expire_process(unsigned long arg);
123
124#ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES
125#define ip6mr_for_each_table(mrt, net) \
126 list_for_each_entry_rcu(mrt, &net->ipv6.mr6_tables, list)
127
128static struct mr6_table *ip6mr_get_table(struct net *net, u32 id)
129{
130 struct mr6_table *mrt;
131
132 ip6mr_for_each_table(mrt, net) {
133 if (mrt->id == id)
134 return mrt;
135 }
136 return NULL;
137}
138
139static int ip6mr_fib_lookup(struct net *net, struct flowi6 *flp6,
140 struct mr6_table **mrt)
141{
142 int err;
143 struct ip6mr_result res;
144 struct fib_lookup_arg arg = {
145 .result = &res,
146 .flags = FIB_LOOKUP_NOREF,
147 };
148
149 err = fib_rules_lookup(net->ipv6.mr6_rules_ops,
150 flowi6_to_flowi(flp6), 0, &arg);
151 if (err < 0)
152 return err;
153 *mrt = res.mrt;
154 return 0;
155}
156
157static int ip6mr_rule_action(struct fib_rule *rule, struct flowi *flp,
158 int flags, struct fib_lookup_arg *arg)
159{
160 struct ip6mr_result *res = arg->result;
161 struct mr6_table *mrt;
162
163 switch (rule->action) {
164 case FR_ACT_TO_TBL:
165 break;
166 case FR_ACT_UNREACHABLE:
167 return -ENETUNREACH;
168 case FR_ACT_PROHIBIT:
169 return -EACCES;
170 case FR_ACT_BLACKHOLE:
171 default:
172 return -EINVAL;
173 }
174
175 mrt = ip6mr_get_table(rule->fr_net, rule->table);
176 if (!mrt)
177 return -EAGAIN;
178 res->mrt = mrt;
179 return 0;
180}
181
182static int ip6mr_rule_match(struct fib_rule *rule, struct flowi *flp, int flags)
183{
184 return 1;
185}
186
187static const struct nla_policy ip6mr_rule_policy[FRA_MAX + 1] = {
188 FRA_GENERIC_POLICY,
189};
190
191static int ip6mr_rule_configure(struct fib_rule *rule, struct sk_buff *skb,
192 struct fib_rule_hdr *frh, struct nlattr **tb)
193{
194 return 0;
195}
196
197static int ip6mr_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh,
198 struct nlattr **tb)
199{
200 return 1;
201}
202
203static int ip6mr_rule_fill(struct fib_rule *rule, struct sk_buff *skb,
204 struct fib_rule_hdr *frh)
205{
206 frh->dst_len = 0;
207 frh->src_len = 0;
208 frh->tos = 0;
209 return 0;
210}
211
212static const struct fib_rules_ops __net_initconst ip6mr_rules_ops_template = {
213 .family = RTNL_FAMILY_IP6MR,
214 .rule_size = sizeof(struct ip6mr_rule),
215 .addr_size = sizeof(struct in6_addr),
216 .action = ip6mr_rule_action,
217 .match = ip6mr_rule_match,
218 .configure = ip6mr_rule_configure,
219 .compare = ip6mr_rule_compare,
220 .default_pref = fib_default_rule_pref,
221 .fill = ip6mr_rule_fill,
222 .nlgroup = RTNLGRP_IPV6_RULE,
223 .policy = ip6mr_rule_policy,
224 .owner = THIS_MODULE,
225};
226
227static int __net_init ip6mr_rules_init(struct net *net)
228{
229 struct fib_rules_ops *ops;
230 struct mr6_table *mrt;
231 int err;
232
233 ops = fib_rules_register(&ip6mr_rules_ops_template, net);
234 if (IS_ERR(ops))
235 return PTR_ERR(ops);
236
237 INIT_LIST_HEAD(&net->ipv6.mr6_tables);
238
239 mrt = ip6mr_new_table(net, RT6_TABLE_DFLT);
240 if (!mrt) {
241 err = -ENOMEM;
242 goto err1;
243 }
244
245 err = fib_default_rule_add(ops, 0x7fff, RT6_TABLE_DFLT, 0);
246 if (err < 0)
247 goto err2;
248
249 net->ipv6.mr6_rules_ops = ops;
250 return 0;
251
252err2:
253 ip6mr_free_table(mrt);
254err1:
255 fib_rules_unregister(ops);
256 return err;
257}
258
259static void __net_exit ip6mr_rules_exit(struct net *net)
260{
261 struct mr6_table *mrt, *next;
262
263 rtnl_lock();
264 list_for_each_entry_safe(mrt, next, &net->ipv6.mr6_tables, list) {
265 list_del(&mrt->list);
266 ip6mr_free_table(mrt);
267 }
268 fib_rules_unregister(net->ipv6.mr6_rules_ops);
269 rtnl_unlock();
270}
271#else
272#define ip6mr_for_each_table(mrt, net) \
273 for (mrt = net->ipv6.mrt6; mrt; mrt = NULL)
274
275static struct mr6_table *ip6mr_get_table(struct net *net, u32 id)
276{
277 return net->ipv6.mrt6;
278}
279
280static int ip6mr_fib_lookup(struct net *net, struct flowi6 *flp6,
281 struct mr6_table **mrt)
282{
283 *mrt = net->ipv6.mrt6;
284 return 0;
285}
286
287static int __net_init ip6mr_rules_init(struct net *net)
288{
289 net->ipv6.mrt6 = ip6mr_new_table(net, RT6_TABLE_DFLT);
290 return net->ipv6.mrt6 ? 0 : -ENOMEM;
291}
292
293static void __net_exit ip6mr_rules_exit(struct net *net)
294{
295 rtnl_lock();
296 ip6mr_free_table(net->ipv6.mrt6);
297 net->ipv6.mrt6 = NULL;
298 rtnl_unlock();
299}
300#endif
301
302static struct mr6_table *ip6mr_new_table(struct net *net, u32 id)
303{
304 struct mr6_table *mrt;
305 unsigned int i;
306
307 mrt = ip6mr_get_table(net, id);
308 if (mrt)
309 return mrt;
310
311 mrt = kzalloc(sizeof(*mrt), GFP_KERNEL);
312 if (!mrt)
313 return NULL;
314 mrt->id = id;
315 write_pnet(&mrt->net, net);
316
317
318 for (i = 0; i < MFC6_LINES; i++)
319 INIT_LIST_HEAD(&mrt->mfc6_cache_array[i]);
320
321 INIT_LIST_HEAD(&mrt->mfc6_unres_queue);
322
323 setup_timer(&mrt->ipmr_expire_timer, ipmr_expire_process,
324 (unsigned long)mrt);
325
326#ifdef CONFIG_IPV6_PIMSM_V2
327 mrt->mroute_reg_vif_num = -1;
328#endif
329#ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES
330 list_add_tail_rcu(&mrt->list, &net->ipv6.mr6_tables);
331#endif
332 return mrt;
333}
334
335static void ip6mr_free_table(struct mr6_table *mrt)
336{
337 del_timer_sync(&mrt->ipmr_expire_timer);
338 mroute_clean_tables(mrt);
339 kfree(mrt);
340}
341
342#ifdef CONFIG_PROC_FS
343
344struct ipmr_mfc_iter {
345 struct seq_net_private p;
346 struct mr6_table *mrt;
347 struct list_head *cache;
348 int ct;
349};
350
351
352static struct mfc6_cache *ipmr_mfc_seq_idx(struct net *net,
353 struct ipmr_mfc_iter *it, loff_t pos)
354{
355 struct mr6_table *mrt = it->mrt;
356 struct mfc6_cache *mfc;
357
358 read_lock(&mrt_lock);
359 for (it->ct = 0; it->ct < MFC6_LINES; it->ct++) {
360 it->cache = &mrt->mfc6_cache_array[it->ct];
361 list_for_each_entry(mfc, it->cache, list)
362 if (pos-- == 0)
363 return mfc;
364 }
365 read_unlock(&mrt_lock);
366
367 spin_lock_bh(&mfc_unres_lock);
368 it->cache = &mrt->mfc6_unres_queue;
369 list_for_each_entry(mfc, it->cache, list)
370 if (pos-- == 0)
371 return mfc;
372 spin_unlock_bh(&mfc_unres_lock);
373
374 it->cache = NULL;
375 return NULL;
376}
377
378
379
380
381
382struct ipmr_vif_iter {
383 struct seq_net_private p;
384 struct mr6_table *mrt;
385 int ct;
386};
387
388static struct mif_device *ip6mr_vif_seq_idx(struct net *net,
389 struct ipmr_vif_iter *iter,
390 loff_t pos)
391{
392 struct mr6_table *mrt = iter->mrt;
393
394 for (iter->ct = 0; iter->ct < mrt->maxvif; ++iter->ct) {
395 if (!MIF_EXISTS(mrt, iter->ct))
396 continue;
397 if (pos-- == 0)
398 return &mrt->vif6_table[iter->ct];
399 }
400 return NULL;
401}
402
403static void *ip6mr_vif_seq_start(struct seq_file *seq, loff_t *pos)
404 __acquires(mrt_lock)
405{
406 struct ipmr_vif_iter *iter = seq->private;
407 struct net *net = seq_file_net(seq);
408 struct mr6_table *mrt;
409
410 mrt = ip6mr_get_table(net, RT6_TABLE_DFLT);
411 if (!mrt)
412 return ERR_PTR(-ENOENT);
413
414 iter->mrt = mrt;
415
416 read_lock(&mrt_lock);
417 return *pos ? ip6mr_vif_seq_idx(net, seq->private, *pos - 1)
418 : SEQ_START_TOKEN;
419}
420
421static void *ip6mr_vif_seq_next(struct seq_file *seq, void *v, loff_t *pos)
422{
423 struct ipmr_vif_iter *iter = seq->private;
424 struct net *net = seq_file_net(seq);
425 struct mr6_table *mrt = iter->mrt;
426
427 ++*pos;
428 if (v == SEQ_START_TOKEN)
429 return ip6mr_vif_seq_idx(net, iter, 0);
430
431 while (++iter->ct < mrt->maxvif) {
432 if (!MIF_EXISTS(mrt, iter->ct))
433 continue;
434 return &mrt->vif6_table[iter->ct];
435 }
436 return NULL;
437}
438
439static void ip6mr_vif_seq_stop(struct seq_file *seq, void *v)
440 __releases(mrt_lock)
441{
442 read_unlock(&mrt_lock);
443}
444
445static int ip6mr_vif_seq_show(struct seq_file *seq, void *v)
446{
447 struct ipmr_vif_iter *iter = seq->private;
448 struct mr6_table *mrt = iter->mrt;
449
450 if (v == SEQ_START_TOKEN) {
451 seq_puts(seq,
452 "Interface BytesIn PktsIn BytesOut PktsOut Flags\n");
453 } else {
454 const struct mif_device *vif = v;
455 const char *name = vif->dev ? vif->dev->name : "none";
456
457 seq_printf(seq,
458 "%2td %-10s %8ld %7ld %8ld %7ld %05X\n",
459 vif - mrt->vif6_table,
460 name, vif->bytes_in, vif->pkt_in,
461 vif->bytes_out, vif->pkt_out,
462 vif->flags);
463 }
464 return 0;
465}
466
467static const struct seq_operations ip6mr_vif_seq_ops = {
468 .start = ip6mr_vif_seq_start,
469 .next = ip6mr_vif_seq_next,
470 .stop = ip6mr_vif_seq_stop,
471 .show = ip6mr_vif_seq_show,
472};
473
474static int ip6mr_vif_open(struct inode *inode, struct file *file)
475{
476 return seq_open_net(inode, file, &ip6mr_vif_seq_ops,
477 sizeof(struct ipmr_vif_iter));
478}
479
480static const struct file_operations ip6mr_vif_fops = {
481 .owner = THIS_MODULE,
482 .open = ip6mr_vif_open,
483 .read = seq_read,
484 .llseek = seq_lseek,
485 .release = seq_release_net,
486};
487
488static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos)
489{
490 struct ipmr_mfc_iter *it = seq->private;
491 struct net *net = seq_file_net(seq);
492 struct mr6_table *mrt;
493
494 mrt = ip6mr_get_table(net, RT6_TABLE_DFLT);
495 if (!mrt)
496 return ERR_PTR(-ENOENT);
497
498 it->mrt = mrt;
499 return *pos ? ipmr_mfc_seq_idx(net, seq->private, *pos - 1)
500 : SEQ_START_TOKEN;
501}
502
503static void *ipmr_mfc_seq_next(struct seq_file *seq, void *v, loff_t *pos)
504{
505 struct mfc6_cache *mfc = v;
506 struct ipmr_mfc_iter *it = seq->private;
507 struct net *net = seq_file_net(seq);
508 struct mr6_table *mrt = it->mrt;
509
510 ++*pos;
511
512 if (v == SEQ_START_TOKEN)
513 return ipmr_mfc_seq_idx(net, seq->private, 0);
514
515 if (mfc->list.next != it->cache)
516 return list_entry(mfc->list.next, struct mfc6_cache, list);
517
518 if (it->cache == &mrt->mfc6_unres_queue)
519 goto end_of_list;
520
521 BUG_ON(it->cache != &mrt->mfc6_cache_array[it->ct]);
522
523 while (++it->ct < MFC6_LINES) {
524 it->cache = &mrt->mfc6_cache_array[it->ct];
525 if (list_empty(it->cache))
526 continue;
527 return list_first_entry(it->cache, struct mfc6_cache, list);
528 }
529
530
531 read_unlock(&mrt_lock);
532 it->cache = &mrt->mfc6_unres_queue;
533 it->ct = 0;
534
535 spin_lock_bh(&mfc_unres_lock);
536 if (!list_empty(it->cache))
537 return list_first_entry(it->cache, struct mfc6_cache, list);
538
539 end_of_list:
540 spin_unlock_bh(&mfc_unres_lock);
541 it->cache = NULL;
542
543 return NULL;
544}
545
546static void ipmr_mfc_seq_stop(struct seq_file *seq, void *v)
547{
548 struct ipmr_mfc_iter *it = seq->private;
549 struct mr6_table *mrt = it->mrt;
550
551 if (it->cache == &mrt->mfc6_unres_queue)
552 spin_unlock_bh(&mfc_unres_lock);
553 else if (it->cache == mrt->mfc6_cache_array)
554 read_unlock(&mrt_lock);
555}
556
557static int ipmr_mfc_seq_show(struct seq_file *seq, void *v)
558{
559 int n;
560
561 if (v == SEQ_START_TOKEN) {
562 seq_puts(seq,
563 "Group "
564 "Origin "
565 "Iif Pkts Bytes Wrong Oifs\n");
566 } else {
567 const struct mfc6_cache *mfc = v;
568 const struct ipmr_mfc_iter *it = seq->private;
569 struct mr6_table *mrt = it->mrt;
570
571 seq_printf(seq, "%pI6 %pI6 %-3hd",
572 &mfc->mf6c_mcastgrp, &mfc->mf6c_origin,
573 mfc->mf6c_parent);
574
575 if (it->cache != &mrt->mfc6_unres_queue) {
576 seq_printf(seq, " %8lu %8lu %8lu",
577 mfc->mfc_un.res.pkt,
578 mfc->mfc_un.res.bytes,
579 mfc->mfc_un.res.wrong_if);
580 for (n = mfc->mfc_un.res.minvif;
581 n < mfc->mfc_un.res.maxvif; n++) {
582 if (MIF_EXISTS(mrt, n) &&
583 mfc->mfc_un.res.ttls[n] < 255)
584 seq_printf(seq,
585 " %2d:%-3d",
586 n, mfc->mfc_un.res.ttls[n]);
587 }
588 } else {
589
590
591
592 seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul);
593 }
594 seq_putc(seq, '\n');
595 }
596 return 0;
597}
598
599static const struct seq_operations ipmr_mfc_seq_ops = {
600 .start = ipmr_mfc_seq_start,
601 .next = ipmr_mfc_seq_next,
602 .stop = ipmr_mfc_seq_stop,
603 .show = ipmr_mfc_seq_show,
604};
605
606static int ipmr_mfc_open(struct inode *inode, struct file *file)
607{
608 return seq_open_net(inode, file, &ipmr_mfc_seq_ops,
609 sizeof(struct ipmr_mfc_iter));
610}
611
612static const struct file_operations ip6mr_mfc_fops = {
613 .owner = THIS_MODULE,
614 .open = ipmr_mfc_open,
615 .read = seq_read,
616 .llseek = seq_lseek,
617 .release = seq_release_net,
618};
619#endif
620
621#ifdef CONFIG_IPV6_PIMSM_V2
622
623static int pim6_rcv(struct sk_buff *skb)
624{
625 struct pimreghdr *pim;
626 struct ipv6hdr *encap;
627 struct net_device *reg_dev = NULL;
628 struct net *net = dev_net(skb->dev);
629 struct mr6_table *mrt;
630 struct flowi6 fl6 = {
631 .flowi6_iif = skb->dev->ifindex,
632 .flowi6_mark = skb->mark,
633 };
634 int reg_vif_num;
635
636 if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(*encap)))
637 goto drop;
638
639 pim = (struct pimreghdr *)skb_transport_header(skb);
640 if (pim->type != ((PIM_VERSION << 4) | PIM_REGISTER) ||
641 (pim->flags & PIM_NULL_REGISTER) ||
642 (csum_ipv6_magic(&ipv6_hdr(skb)->saddr, &ipv6_hdr(skb)->daddr,
643 sizeof(*pim), IPPROTO_PIM,
644 csum_partial((void *)pim, sizeof(*pim), 0)) &&
645 csum_fold(skb_checksum(skb, 0, skb->len, 0))))
646 goto drop;
647
648
649 encap = (struct ipv6hdr *)(skb_transport_header(skb) +
650 sizeof(*pim));
651
652 if (!ipv6_addr_is_multicast(&encap->daddr) ||
653 encap->payload_len == 0 ||
654 ntohs(encap->payload_len) + sizeof(*pim) > skb->len)
655 goto drop;
656
657 if (ip6mr_fib_lookup(net, &fl6, &mrt) < 0)
658 goto drop;
659 reg_vif_num = mrt->mroute_reg_vif_num;
660
661 read_lock(&mrt_lock);
662 if (reg_vif_num >= 0)
663 reg_dev = mrt->vif6_table[reg_vif_num].dev;
664 if (reg_dev)
665 dev_hold(reg_dev);
666 read_unlock(&mrt_lock);
667
668 if (!reg_dev)
669 goto drop;
670
671 skb->mac_header = skb->network_header;
672 skb_pull(skb, (u8 *)encap - skb->data);
673 skb_reset_network_header(skb);
674 skb->protocol = htons(ETH_P_IPV6);
675 skb->ip_summed = CHECKSUM_NONE;
676
677 skb_tunnel_rx(skb, reg_dev, dev_net(reg_dev));
678
679 netif_rx(skb);
680
681 dev_put(reg_dev);
682 return 0;
683 drop:
684 kfree_skb(skb);
685 return 0;
686}
687
688static const struct inet6_protocol pim6_protocol = {
689 .handler = pim6_rcv,
690};
691
692
693
694static netdev_tx_t reg_vif_xmit(struct sk_buff *skb,
695 struct net_device *dev)
696{
697 struct net *net = dev_net(dev);
698 struct mr6_table *mrt;
699 struct flowi6 fl6 = {
700 .flowi6_oif = dev->ifindex,
701 .flowi6_iif = skb->skb_iif ? : LOOPBACK_IFINDEX,
702 .flowi6_mark = skb->mark,
703 };
704 int err;
705
706 err = ip6mr_fib_lookup(net, &fl6, &mrt);
707 if (err < 0) {
708 kfree_skb(skb);
709 return err;
710 }
711
712 read_lock(&mrt_lock);
713 dev->stats.tx_bytes += skb->len;
714 dev->stats.tx_packets++;
715 ip6mr_cache_report(mrt, skb, mrt->mroute_reg_vif_num, MRT6MSG_WHOLEPKT);
716 read_unlock(&mrt_lock);
717 kfree_skb(skb);
718 return NETDEV_TX_OK;
719}
720
721static int reg_vif_get_iflink(const struct net_device *dev)
722{
723 return 0;
724}
725
726static const struct net_device_ops reg_vif_netdev_ops = {
727 .ndo_start_xmit = reg_vif_xmit,
728 .ndo_get_iflink = reg_vif_get_iflink,
729};
730
731static void reg_vif_setup(struct net_device *dev)
732{
733 dev->type = ARPHRD_PIMREG;
734 dev->mtu = 1500 - sizeof(struct ipv6hdr) - 8;
735 dev->flags = IFF_NOARP;
736 dev->netdev_ops = ®_vif_netdev_ops;
737 dev->destructor = free_netdev;
738 dev->features |= NETIF_F_NETNS_LOCAL;
739}
740
741static struct net_device *ip6mr_reg_vif(struct net *net, struct mr6_table *mrt)
742{
743 struct net_device *dev;
744 char name[IFNAMSIZ];
745
746 if (mrt->id == RT6_TABLE_DFLT)
747 sprintf(name, "pim6reg");
748 else
749 sprintf(name, "pim6reg%u", mrt->id);
750
751 dev = alloc_netdev(0, name, NET_NAME_UNKNOWN, reg_vif_setup);
752 if (!dev)
753 return NULL;
754
755 dev_net_set(dev, net);
756
757 if (register_netdevice(dev)) {
758 free_netdev(dev);
759 return NULL;
760 }
761
762 if (dev_open(dev))
763 goto failure;
764
765 dev_hold(dev);
766 return dev;
767
768failure:
769
770 rtnl_unlock();
771 rtnl_lock();
772
773 unregister_netdevice(dev);
774 return NULL;
775}
776#endif
777
778
779
780
781
782static int mif6_delete(struct mr6_table *mrt, int vifi, struct list_head *head)
783{
784 struct mif_device *v;
785 struct net_device *dev;
786 struct inet6_dev *in6_dev;
787
788 if (vifi < 0 || vifi >= mrt->maxvif)
789 return -EADDRNOTAVAIL;
790
791 v = &mrt->vif6_table[vifi];
792
793 write_lock_bh(&mrt_lock);
794 dev = v->dev;
795 v->dev = NULL;
796
797 if (!dev) {
798 write_unlock_bh(&mrt_lock);
799 return -EADDRNOTAVAIL;
800 }
801
802#ifdef CONFIG_IPV6_PIMSM_V2
803 if (vifi == mrt->mroute_reg_vif_num)
804 mrt->mroute_reg_vif_num = -1;
805#endif
806
807 if (vifi + 1 == mrt->maxvif) {
808 int tmp;
809 for (tmp = vifi - 1; tmp >= 0; tmp--) {
810 if (MIF_EXISTS(mrt, tmp))
811 break;
812 }
813 mrt->maxvif = tmp + 1;
814 }
815
816 write_unlock_bh(&mrt_lock);
817
818 dev_set_allmulti(dev, -1);
819
820 in6_dev = __in6_dev_get(dev);
821 if (in6_dev) {
822 in6_dev->cnf.mc_forwarding--;
823 inet6_netconf_notify_devconf(dev_net(dev),
824 NETCONFA_MC_FORWARDING,
825 dev->ifindex, &in6_dev->cnf);
826 }
827
828 if (v->flags & MIFF_REGISTER)
829 unregister_netdevice_queue(dev, head);
830
831 dev_put(dev);
832 return 0;
833}
834
835static inline void ip6mr_cache_free(struct mfc6_cache *c)
836{
837 kmem_cache_free(mrt_cachep, c);
838}
839
840
841
842
843
844static void ip6mr_destroy_unres(struct mr6_table *mrt, struct mfc6_cache *c)
845{
846 struct net *net = read_pnet(&mrt->net);
847 struct sk_buff *skb;
848
849 atomic_dec(&mrt->cache_resolve_queue_len);
850
851 while ((skb = skb_dequeue(&c->mfc_un.unres.unresolved)) != NULL) {
852 if (ipv6_hdr(skb)->version == 0) {
853 struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct ipv6hdr));
854 nlh->nlmsg_type = NLMSG_ERROR;
855 nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr));
856 skb_trim(skb, nlh->nlmsg_len);
857 ((struct nlmsgerr *)nlmsg_data(nlh))->error = -ETIMEDOUT;
858 rtnl_unicast(skb, net, NETLINK_CB(skb).portid);
859 } else
860 kfree_skb(skb);
861 }
862
863 ip6mr_cache_free(c);
864}
865
866
867
868
869static void ipmr_do_expire_process(struct mr6_table *mrt)
870{
871 unsigned long now = jiffies;
872 unsigned long expires = 10 * HZ;
873 struct mfc6_cache *c, *next;
874
875 list_for_each_entry_safe(c, next, &mrt->mfc6_unres_queue, list) {
876 if (time_after(c->mfc_un.unres.expires, now)) {
877
878 unsigned long interval = c->mfc_un.unres.expires - now;
879 if (interval < expires)
880 expires = interval;
881 continue;
882 }
883
884 list_del(&c->list);
885 mr6_netlink_event(mrt, c, RTM_DELROUTE);
886 ip6mr_destroy_unres(mrt, c);
887 }
888
889 if (!list_empty(&mrt->mfc6_unres_queue))
890 mod_timer(&mrt->ipmr_expire_timer, jiffies + expires);
891}
892
893static void ipmr_expire_process(unsigned long arg)
894{
895 struct mr6_table *mrt = (struct mr6_table *)arg;
896
897 if (!spin_trylock(&mfc_unres_lock)) {
898 mod_timer(&mrt->ipmr_expire_timer, jiffies + 1);
899 return;
900 }
901
902 if (!list_empty(&mrt->mfc6_unres_queue))
903 ipmr_do_expire_process(mrt);
904
905 spin_unlock(&mfc_unres_lock);
906}
907
908
909
910static void ip6mr_update_thresholds(struct mr6_table *mrt, struct mfc6_cache *cache,
911 unsigned char *ttls)
912{
913 int vifi;
914
915 cache->mfc_un.res.minvif = MAXMIFS;
916 cache->mfc_un.res.maxvif = 0;
917 memset(cache->mfc_un.res.ttls, 255, MAXMIFS);
918
919 for (vifi = 0; vifi < mrt->maxvif; vifi++) {
920 if (MIF_EXISTS(mrt, vifi) &&
921 ttls[vifi] && ttls[vifi] < 255) {
922 cache->mfc_un.res.ttls[vifi] = ttls[vifi];
923 if (cache->mfc_un.res.minvif > vifi)
924 cache->mfc_un.res.minvif = vifi;
925 if (cache->mfc_un.res.maxvif <= vifi)
926 cache->mfc_un.res.maxvif = vifi + 1;
927 }
928 }
929}
930
931static int mif6_add(struct net *net, struct mr6_table *mrt,
932 struct mif6ctl *vifc, int mrtsock)
933{
934 int vifi = vifc->mif6c_mifi;
935 struct mif_device *v = &mrt->vif6_table[vifi];
936 struct net_device *dev;
937 struct inet6_dev *in6_dev;
938 int err;
939
940
941 if (MIF_EXISTS(mrt, vifi))
942 return -EADDRINUSE;
943
944 switch (vifc->mif6c_flags) {
945#ifdef CONFIG_IPV6_PIMSM_V2
946 case MIFF_REGISTER:
947
948
949
950
951 if (mrt->mroute_reg_vif_num >= 0)
952 return -EADDRINUSE;
953 dev = ip6mr_reg_vif(net, mrt);
954 if (!dev)
955 return -ENOBUFS;
956 err = dev_set_allmulti(dev, 1);
957 if (err) {
958 unregister_netdevice(dev);
959 dev_put(dev);
960 return err;
961 }
962 break;
963#endif
964 case 0:
965 dev = dev_get_by_index(net, vifc->mif6c_pifi);
966 if (!dev)
967 return -EADDRNOTAVAIL;
968 err = dev_set_allmulti(dev, 1);
969 if (err) {
970 dev_put(dev);
971 return err;
972 }
973 break;
974 default:
975 return -EINVAL;
976 }
977
978 in6_dev = __in6_dev_get(dev);
979 if (in6_dev) {
980 in6_dev->cnf.mc_forwarding++;
981 inet6_netconf_notify_devconf(dev_net(dev),
982 NETCONFA_MC_FORWARDING,
983 dev->ifindex, &in6_dev->cnf);
984 }
985
986
987
988
989 v->rate_limit = vifc->vifc_rate_limit;
990 v->flags = vifc->mif6c_flags;
991 if (!mrtsock)
992 v->flags |= VIFF_STATIC;
993 v->threshold = vifc->vifc_threshold;
994 v->bytes_in = 0;
995 v->bytes_out = 0;
996 v->pkt_in = 0;
997 v->pkt_out = 0;
998 v->link = dev->ifindex;
999 if (v->flags & MIFF_REGISTER)
1000 v->link = dev_get_iflink(dev);
1001
1002
1003 write_lock_bh(&mrt_lock);
1004 v->dev = dev;
1005#ifdef CONFIG_IPV6_PIMSM_V2
1006 if (v->flags & MIFF_REGISTER)
1007 mrt->mroute_reg_vif_num = vifi;
1008#endif
1009 if (vifi + 1 > mrt->maxvif)
1010 mrt->maxvif = vifi + 1;
1011 write_unlock_bh(&mrt_lock);
1012 return 0;
1013}
1014
1015static struct mfc6_cache *ip6mr_cache_find(struct mr6_table *mrt,
1016 const struct in6_addr *origin,
1017 const struct in6_addr *mcastgrp)
1018{
1019 int line = MFC6_HASH(mcastgrp, origin);
1020 struct mfc6_cache *c;
1021
1022 list_for_each_entry(c, &mrt->mfc6_cache_array[line], list) {
1023 if (ipv6_addr_equal(&c->mf6c_origin, origin) &&
1024 ipv6_addr_equal(&c->mf6c_mcastgrp, mcastgrp))
1025 return c;
1026 }
1027 return NULL;
1028}
1029
1030
1031static struct mfc6_cache *ip6mr_cache_find_any_parent(struct mr6_table *mrt,
1032 mifi_t mifi)
1033{
1034 int line = MFC6_HASH(&in6addr_any, &in6addr_any);
1035 struct mfc6_cache *c;
1036
1037 list_for_each_entry(c, &mrt->mfc6_cache_array[line], list)
1038 if (ipv6_addr_any(&c->mf6c_origin) &&
1039 ipv6_addr_any(&c->mf6c_mcastgrp) &&
1040 (c->mfc_un.res.ttls[mifi] < 255))
1041 return c;
1042
1043 return NULL;
1044}
1045
1046
1047static struct mfc6_cache *ip6mr_cache_find_any(struct mr6_table *mrt,
1048 struct in6_addr *mcastgrp,
1049 mifi_t mifi)
1050{
1051 int line = MFC6_HASH(mcastgrp, &in6addr_any);
1052 struct mfc6_cache *c, *proxy;
1053
1054 if (ipv6_addr_any(mcastgrp))
1055 goto skip;
1056
1057 list_for_each_entry(c, &mrt->mfc6_cache_array[line], list)
1058 if (ipv6_addr_any(&c->mf6c_origin) &&
1059 ipv6_addr_equal(&c->mf6c_mcastgrp, mcastgrp)) {
1060 if (c->mfc_un.res.ttls[mifi] < 255)
1061 return c;
1062
1063
1064 proxy = ip6mr_cache_find_any_parent(mrt,
1065 c->mf6c_parent);
1066 if (proxy && proxy->mfc_un.res.ttls[mifi] < 255)
1067 return c;
1068 }
1069
1070skip:
1071 return ip6mr_cache_find_any_parent(mrt, mifi);
1072}
1073
1074
1075
1076
1077static struct mfc6_cache *ip6mr_cache_alloc(void)
1078{
1079 struct mfc6_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL);
1080 if (!c)
1081 return NULL;
1082 c->mfc_un.res.minvif = MAXMIFS;
1083 return c;
1084}
1085
1086static struct mfc6_cache *ip6mr_cache_alloc_unres(void)
1087{
1088 struct mfc6_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC);
1089 if (!c)
1090 return NULL;
1091 skb_queue_head_init(&c->mfc_un.unres.unresolved);
1092 c->mfc_un.unres.expires = jiffies + 10 * HZ;
1093 return c;
1094}
1095
1096
1097
1098
1099
1100static void ip6mr_cache_resolve(struct net *net, struct mr6_table *mrt,
1101 struct mfc6_cache *uc, struct mfc6_cache *c)
1102{
1103 struct sk_buff *skb;
1104
1105
1106
1107
1108
1109 while ((skb = __skb_dequeue(&uc->mfc_un.unres.unresolved))) {
1110 if (ipv6_hdr(skb)->version == 0) {
1111 struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct ipv6hdr));
1112
1113 if (__ip6mr_fill_mroute(mrt, skb, c, nlmsg_data(nlh)) > 0) {
1114 nlh->nlmsg_len = skb_tail_pointer(skb) - (u8 *)nlh;
1115 } else {
1116 nlh->nlmsg_type = NLMSG_ERROR;
1117 nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr));
1118 skb_trim(skb, nlh->nlmsg_len);
1119 ((struct nlmsgerr *)nlmsg_data(nlh))->error = -EMSGSIZE;
1120 }
1121 rtnl_unicast(skb, net, NETLINK_CB(skb).portid);
1122 } else
1123 ip6_mr_forward(net, mrt, skb, c);
1124 }
1125}
1126
1127
1128
1129
1130
1131
1132
1133
1134static int ip6mr_cache_report(struct mr6_table *mrt, struct sk_buff *pkt,
1135 mifi_t mifi, int assert)
1136{
1137 struct sk_buff *skb;
1138 struct mrt6msg *msg;
1139 int ret;
1140
1141#ifdef CONFIG_IPV6_PIMSM_V2
1142 if (assert == MRT6MSG_WHOLEPKT)
1143 skb = skb_realloc_headroom(pkt, -skb_network_offset(pkt)
1144 +sizeof(*msg));
1145 else
1146#endif
1147 skb = alloc_skb(sizeof(struct ipv6hdr) + sizeof(*msg), GFP_ATOMIC);
1148
1149 if (!skb)
1150 return -ENOBUFS;
1151
1152
1153
1154
1155 skb->ip_summed = CHECKSUM_UNNECESSARY;
1156
1157#ifdef CONFIG_IPV6_PIMSM_V2
1158 if (assert == MRT6MSG_WHOLEPKT) {
1159
1160
1161
1162
1163
1164 skb_push(skb, -skb_network_offset(pkt));
1165
1166 skb_push(skb, sizeof(*msg));
1167 skb_reset_transport_header(skb);
1168 msg = (struct mrt6msg *)skb_transport_header(skb);
1169 msg->im6_mbz = 0;
1170 msg->im6_msgtype = MRT6MSG_WHOLEPKT;
1171 msg->im6_mif = mrt->mroute_reg_vif_num;
1172 msg->im6_pad = 0;
1173 msg->im6_src = ipv6_hdr(pkt)->saddr;
1174 msg->im6_dst = ipv6_hdr(pkt)->daddr;
1175
1176 skb->ip_summed = CHECKSUM_UNNECESSARY;
1177 } else
1178#endif
1179 {
1180
1181
1182
1183
1184 skb_put(skb, sizeof(struct ipv6hdr));
1185 skb_reset_network_header(skb);
1186 skb_copy_to_linear_data(skb, ipv6_hdr(pkt), sizeof(struct ipv6hdr));
1187
1188
1189
1190
1191 skb_put(skb, sizeof(*msg));
1192 skb_reset_transport_header(skb);
1193 msg = (struct mrt6msg *)skb_transport_header(skb);
1194
1195 msg->im6_mbz = 0;
1196 msg->im6_msgtype = assert;
1197 msg->im6_mif = mifi;
1198 msg->im6_pad = 0;
1199 msg->im6_src = ipv6_hdr(pkt)->saddr;
1200 msg->im6_dst = ipv6_hdr(pkt)->daddr;
1201
1202 skb_dst_set(skb, dst_clone(skb_dst(pkt)));
1203 skb->ip_summed = CHECKSUM_UNNECESSARY;
1204 }
1205
1206 if (!mrt->mroute6_sk) {
1207 kfree_skb(skb);
1208 return -EINVAL;
1209 }
1210
1211
1212
1213
1214 ret = sock_queue_rcv_skb(mrt->mroute6_sk, skb);
1215 if (ret < 0) {
1216 net_warn_ratelimited("mroute6: pending queue full, dropping entries\n");
1217 kfree_skb(skb);
1218 }
1219
1220 return ret;
1221}
1222
1223
1224
1225
1226
1227static int
1228ip6mr_cache_unresolved(struct mr6_table *mrt, mifi_t mifi, struct sk_buff *skb)
1229{
1230 bool found = false;
1231 int err;
1232 struct mfc6_cache *c;
1233
1234 spin_lock_bh(&mfc_unres_lock);
1235 list_for_each_entry(c, &mrt->mfc6_unres_queue, list) {
1236 if (ipv6_addr_equal(&c->mf6c_mcastgrp, &ipv6_hdr(skb)->daddr) &&
1237 ipv6_addr_equal(&c->mf6c_origin, &ipv6_hdr(skb)->saddr)) {
1238 found = true;
1239 break;
1240 }
1241 }
1242
1243 if (!found) {
1244
1245
1246
1247
1248 if (atomic_read(&mrt->cache_resolve_queue_len) >= 10 ||
1249 (c = ip6mr_cache_alloc_unres()) == NULL) {
1250 spin_unlock_bh(&mfc_unres_lock);
1251
1252 kfree_skb(skb);
1253 return -ENOBUFS;
1254 }
1255
1256
1257
1258
1259 c->mf6c_parent = -1;
1260 c->mf6c_origin = ipv6_hdr(skb)->saddr;
1261 c->mf6c_mcastgrp = ipv6_hdr(skb)->daddr;
1262
1263
1264
1265
1266 err = ip6mr_cache_report(mrt, skb, mifi, MRT6MSG_NOCACHE);
1267 if (err < 0) {
1268
1269
1270
1271 spin_unlock_bh(&mfc_unres_lock);
1272
1273 ip6mr_cache_free(c);
1274 kfree_skb(skb);
1275 return err;
1276 }
1277
1278 atomic_inc(&mrt->cache_resolve_queue_len);
1279 list_add(&c->list, &mrt->mfc6_unres_queue);
1280 mr6_netlink_event(mrt, c, RTM_NEWROUTE);
1281
1282 ipmr_do_expire_process(mrt);
1283 }
1284
1285
1286
1287
1288 if (c->mfc_un.unres.unresolved.qlen > 3) {
1289 kfree_skb(skb);
1290 err = -ENOBUFS;
1291 } else {
1292 skb_queue_tail(&c->mfc_un.unres.unresolved, skb);
1293 err = 0;
1294 }
1295
1296 spin_unlock_bh(&mfc_unres_lock);
1297 return err;
1298}
1299
1300
1301
1302
1303
1304static int ip6mr_mfc_delete(struct mr6_table *mrt, struct mf6cctl *mfc,
1305 int parent)
1306{
1307 int line;
1308 struct mfc6_cache *c, *next;
1309
1310 line = MFC6_HASH(&mfc->mf6cc_mcastgrp.sin6_addr, &mfc->mf6cc_origin.sin6_addr);
1311
1312 list_for_each_entry_safe(c, next, &mrt->mfc6_cache_array[line], list) {
1313 if (ipv6_addr_equal(&c->mf6c_origin, &mfc->mf6cc_origin.sin6_addr) &&
1314 ipv6_addr_equal(&c->mf6c_mcastgrp,
1315 &mfc->mf6cc_mcastgrp.sin6_addr) &&
1316 (parent == -1 || parent == c->mf6c_parent)) {
1317 write_lock_bh(&mrt_lock);
1318 list_del(&c->list);
1319 write_unlock_bh(&mrt_lock);
1320
1321 mr6_netlink_event(mrt, c, RTM_DELROUTE);
1322 ip6mr_cache_free(c);
1323 return 0;
1324 }
1325 }
1326 return -ENOENT;
1327}
1328
1329static int ip6mr_device_event(struct notifier_block *this,
1330 unsigned long event, void *ptr)
1331{
1332 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
1333 struct net *net = dev_net(dev);
1334 struct mr6_table *mrt;
1335 struct mif_device *v;
1336 int ct;
1337 LIST_HEAD(list);
1338
1339 if (event != NETDEV_UNREGISTER)
1340 return NOTIFY_DONE;
1341
1342 ip6mr_for_each_table(mrt, net) {
1343 v = &mrt->vif6_table[0];
1344 for (ct = 0; ct < mrt->maxvif; ct++, v++) {
1345 if (v->dev == dev)
1346 mif6_delete(mrt, ct, &list);
1347 }
1348 }
1349 unregister_netdevice_many(&list);
1350
1351 return NOTIFY_DONE;
1352}
1353
1354static struct notifier_block ip6_mr_notifier = {
1355 .notifier_call = ip6mr_device_event
1356};
1357
1358
1359
1360
1361
1362static int __net_init ip6mr_net_init(struct net *net)
1363{
1364 int err;
1365
1366 err = ip6mr_rules_init(net);
1367 if (err < 0)
1368 goto fail;
1369
1370#ifdef CONFIG_PROC_FS
1371 err = -ENOMEM;
1372 if (!proc_create("ip6_mr_vif", 0, net->proc_net, &ip6mr_vif_fops))
1373 goto proc_vif_fail;
1374 if (!proc_create("ip6_mr_cache", 0, net->proc_net, &ip6mr_mfc_fops))
1375 goto proc_cache_fail;
1376#endif
1377
1378 return 0;
1379
1380#ifdef CONFIG_PROC_FS
1381proc_cache_fail:
1382 remove_proc_entry("ip6_mr_vif", net->proc_net);
1383proc_vif_fail:
1384 ip6mr_rules_exit(net);
1385#endif
1386fail:
1387 return err;
1388}
1389
1390static void __net_exit ip6mr_net_exit(struct net *net)
1391{
1392#ifdef CONFIG_PROC_FS
1393 remove_proc_entry("ip6_mr_cache", net->proc_net);
1394 remove_proc_entry("ip6_mr_vif", net->proc_net);
1395#endif
1396 ip6mr_rules_exit(net);
1397}
1398
1399static struct pernet_operations ip6mr_net_ops = {
1400 .init = ip6mr_net_init,
1401 .exit = ip6mr_net_exit,
1402};
1403
1404int __init ip6_mr_init(void)
1405{
1406 int err;
1407
1408 mrt_cachep = kmem_cache_create("ip6_mrt_cache",
1409 sizeof(struct mfc6_cache),
1410 0, SLAB_HWCACHE_ALIGN,
1411 NULL);
1412 if (!mrt_cachep)
1413 return -ENOMEM;
1414
1415 err = register_pernet_subsys(&ip6mr_net_ops);
1416 if (err)
1417 goto reg_pernet_fail;
1418
1419 err = register_netdevice_notifier(&ip6_mr_notifier);
1420 if (err)
1421 goto reg_notif_fail;
1422#ifdef CONFIG_IPV6_PIMSM_V2
1423 if (inet6_add_protocol(&pim6_protocol, IPPROTO_PIM) < 0) {
1424 pr_err("%s: can't add PIM protocol\n", __func__);
1425 err = -EAGAIN;
1426 goto add_proto_fail;
1427 }
1428#endif
1429 rtnl_register(RTNL_FAMILY_IP6MR, RTM_GETROUTE, NULL,
1430 ip6mr_rtm_dumproute, NULL);
1431 return 0;
1432#ifdef CONFIG_IPV6_PIMSM_V2
1433add_proto_fail:
1434 unregister_netdevice_notifier(&ip6_mr_notifier);
1435#endif
1436reg_notif_fail:
1437 unregister_pernet_subsys(&ip6mr_net_ops);
1438reg_pernet_fail:
1439 kmem_cache_destroy(mrt_cachep);
1440 return err;
1441}
1442
1443void ip6_mr_cleanup(void)
1444{
1445 rtnl_unregister(RTNL_FAMILY_IP6MR, RTM_GETROUTE);
1446#ifdef CONFIG_IPV6_PIMSM_V2
1447 inet6_del_protocol(&pim6_protocol, IPPROTO_PIM);
1448#endif
1449 unregister_netdevice_notifier(&ip6_mr_notifier);
1450 unregister_pernet_subsys(&ip6mr_net_ops);
1451 kmem_cache_destroy(mrt_cachep);
1452}
1453
1454static int ip6mr_mfc_add(struct net *net, struct mr6_table *mrt,
1455 struct mf6cctl *mfc, int mrtsock, int parent)
1456{
1457 bool found = false;
1458 int line;
1459 struct mfc6_cache *uc, *c;
1460 unsigned char ttls[MAXMIFS];
1461 int i;
1462
1463 if (mfc->mf6cc_parent >= MAXMIFS)
1464 return -ENFILE;
1465
1466 memset(ttls, 255, MAXMIFS);
1467 for (i = 0; i < MAXMIFS; i++) {
1468 if (IF_ISSET(i, &mfc->mf6cc_ifset))
1469 ttls[i] = 1;
1470
1471 }
1472
1473 line = MFC6_HASH(&mfc->mf6cc_mcastgrp.sin6_addr, &mfc->mf6cc_origin.sin6_addr);
1474
1475 list_for_each_entry(c, &mrt->mfc6_cache_array[line], list) {
1476 if (ipv6_addr_equal(&c->mf6c_origin, &mfc->mf6cc_origin.sin6_addr) &&
1477 ipv6_addr_equal(&c->mf6c_mcastgrp,
1478 &mfc->mf6cc_mcastgrp.sin6_addr) &&
1479 (parent == -1 || parent == mfc->mf6cc_parent)) {
1480 found = true;
1481 break;
1482 }
1483 }
1484
1485 if (found) {
1486 write_lock_bh(&mrt_lock);
1487 c->mf6c_parent = mfc->mf6cc_parent;
1488 ip6mr_update_thresholds(mrt, c, ttls);
1489 if (!mrtsock)
1490 c->mfc_flags |= MFC_STATIC;
1491 write_unlock_bh(&mrt_lock);
1492 mr6_netlink_event(mrt, c, RTM_NEWROUTE);
1493 return 0;
1494 }
1495
1496 if (!ipv6_addr_any(&mfc->mf6cc_mcastgrp.sin6_addr) &&
1497 !ipv6_addr_is_multicast(&mfc->mf6cc_mcastgrp.sin6_addr))
1498 return -EINVAL;
1499
1500 c = ip6mr_cache_alloc();
1501 if (!c)
1502 return -ENOMEM;
1503
1504 c->mf6c_origin = mfc->mf6cc_origin.sin6_addr;
1505 c->mf6c_mcastgrp = mfc->mf6cc_mcastgrp.sin6_addr;
1506 c->mf6c_parent = mfc->mf6cc_parent;
1507 ip6mr_update_thresholds(mrt, c, ttls);
1508 if (!mrtsock)
1509 c->mfc_flags |= MFC_STATIC;
1510
1511 write_lock_bh(&mrt_lock);
1512 list_add(&c->list, &mrt->mfc6_cache_array[line]);
1513 write_unlock_bh(&mrt_lock);
1514
1515
1516
1517
1518
1519 found = false;
1520 spin_lock_bh(&mfc_unres_lock);
1521 list_for_each_entry(uc, &mrt->mfc6_unres_queue, list) {
1522 if (ipv6_addr_equal(&uc->mf6c_origin, &c->mf6c_origin) &&
1523 ipv6_addr_equal(&uc->mf6c_mcastgrp, &c->mf6c_mcastgrp)) {
1524 list_del(&uc->list);
1525 atomic_dec(&mrt->cache_resolve_queue_len);
1526 found = true;
1527 break;
1528 }
1529 }
1530 if (list_empty(&mrt->mfc6_unres_queue))
1531 del_timer(&mrt->ipmr_expire_timer);
1532 spin_unlock_bh(&mfc_unres_lock);
1533
1534 if (found) {
1535 ip6mr_cache_resolve(net, mrt, uc, c);
1536 ip6mr_cache_free(uc);
1537 }
1538 mr6_netlink_event(mrt, c, RTM_NEWROUTE);
1539 return 0;
1540}
1541
1542
1543
1544
1545
1546static void mroute_clean_tables(struct mr6_table *mrt)
1547{
1548 int i;
1549 LIST_HEAD(list);
1550 struct mfc6_cache *c, *next;
1551
1552
1553
1554
1555 for (i = 0; i < mrt->maxvif; i++) {
1556 if (!(mrt->vif6_table[i].flags & VIFF_STATIC))
1557 mif6_delete(mrt, i, &list);
1558 }
1559 unregister_netdevice_many(&list);
1560
1561
1562
1563
1564 for (i = 0; i < MFC6_LINES; i++) {
1565 list_for_each_entry_safe(c, next, &mrt->mfc6_cache_array[i], list) {
1566 if (c->mfc_flags & MFC_STATIC)
1567 continue;
1568 write_lock_bh(&mrt_lock);
1569 list_del(&c->list);
1570 write_unlock_bh(&mrt_lock);
1571
1572 mr6_netlink_event(mrt, c, RTM_DELROUTE);
1573 ip6mr_cache_free(c);
1574 }
1575 }
1576
1577 if (atomic_read(&mrt->cache_resolve_queue_len) != 0) {
1578 spin_lock_bh(&mfc_unres_lock);
1579 list_for_each_entry_safe(c, next, &mrt->mfc6_unres_queue, list) {
1580 list_del(&c->list);
1581 mr6_netlink_event(mrt, c, RTM_DELROUTE);
1582 ip6mr_destroy_unres(mrt, c);
1583 }
1584 spin_unlock_bh(&mfc_unres_lock);
1585 }
1586}
1587
1588static int ip6mr_sk_init(struct mr6_table *mrt, struct sock *sk)
1589{
1590 int err = 0;
1591 struct net *net = sock_net(sk);
1592
1593 rtnl_lock();
1594 write_lock_bh(&mrt_lock);
1595 if (likely(mrt->mroute6_sk == NULL)) {
1596 mrt->mroute6_sk = sk;
1597 net->ipv6.devconf_all->mc_forwarding++;
1598 inet6_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING,
1599 NETCONFA_IFINDEX_ALL,
1600 net->ipv6.devconf_all);
1601 }
1602 else
1603 err = -EADDRINUSE;
1604 write_unlock_bh(&mrt_lock);
1605
1606 rtnl_unlock();
1607
1608 return err;
1609}
1610
1611int ip6mr_sk_done(struct sock *sk)
1612{
1613 int err = -EACCES;
1614 struct net *net = sock_net(sk);
1615 struct mr6_table *mrt;
1616
1617 rtnl_lock();
1618 ip6mr_for_each_table(mrt, net) {
1619 if (sk == mrt->mroute6_sk) {
1620 write_lock_bh(&mrt_lock);
1621 mrt->mroute6_sk = NULL;
1622 net->ipv6.devconf_all->mc_forwarding--;
1623 inet6_netconf_notify_devconf(net,
1624 NETCONFA_MC_FORWARDING,
1625 NETCONFA_IFINDEX_ALL,
1626 net->ipv6.devconf_all);
1627 write_unlock_bh(&mrt_lock);
1628
1629 mroute_clean_tables(mrt);
1630 err = 0;
1631 break;
1632 }
1633 }
1634 rtnl_unlock();
1635
1636 return err;
1637}
1638
1639struct sock *mroute6_socket(struct net *net, struct sk_buff *skb)
1640{
1641 struct mr6_table *mrt;
1642 struct flowi6 fl6 = {
1643 .flowi6_iif = skb->skb_iif ? : LOOPBACK_IFINDEX,
1644 .flowi6_oif = skb->dev->ifindex,
1645 .flowi6_mark = skb->mark,
1646 };
1647
1648 if (ip6mr_fib_lookup(net, &fl6, &mrt) < 0)
1649 return NULL;
1650
1651 return mrt->mroute6_sk;
1652}
1653
1654
1655
1656
1657
1658
1659
1660
1661int ip6_mroute_setsockopt(struct sock *sk, int optname, char __user *optval, unsigned int optlen)
1662{
1663 int ret, parent = 0;
1664 struct mif6ctl vif;
1665 struct mf6cctl mfc;
1666 mifi_t mifi;
1667 struct net *net = sock_net(sk);
1668 struct mr6_table *mrt;
1669
1670 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
1671 if (!mrt)
1672 return -ENOENT;
1673
1674 if (optname != MRT6_INIT) {
1675 if (sk != mrt->mroute6_sk && !ns_capable(net->user_ns, CAP_NET_ADMIN))
1676 return -EACCES;
1677 }
1678
1679 switch (optname) {
1680 case MRT6_INIT:
1681 if (sk->sk_type != SOCK_RAW ||
1682 inet_sk(sk)->inet_num != IPPROTO_ICMPV6)
1683 return -EOPNOTSUPP;
1684 if (optlen < sizeof(int))
1685 return -EINVAL;
1686
1687 return ip6mr_sk_init(mrt, sk);
1688
1689 case MRT6_DONE:
1690 return ip6mr_sk_done(sk);
1691
1692 case MRT6_ADD_MIF:
1693 if (optlen < sizeof(vif))
1694 return -EINVAL;
1695 if (copy_from_user(&vif, optval, sizeof(vif)))
1696 return -EFAULT;
1697 if (vif.mif6c_mifi >= MAXMIFS)
1698 return -ENFILE;
1699 rtnl_lock();
1700 ret = mif6_add(net, mrt, &vif, sk == mrt->mroute6_sk);
1701 rtnl_unlock();
1702 return ret;
1703
1704 case MRT6_DEL_MIF:
1705 if (optlen < sizeof(mifi_t))
1706 return -EINVAL;
1707 if (copy_from_user(&mifi, optval, sizeof(mifi_t)))
1708 return -EFAULT;
1709 rtnl_lock();
1710 ret = mif6_delete(mrt, mifi, NULL);
1711 rtnl_unlock();
1712 return ret;
1713
1714
1715
1716
1717
1718 case MRT6_ADD_MFC:
1719 case MRT6_DEL_MFC:
1720 parent = -1;
1721 case MRT6_ADD_MFC_PROXY:
1722 case MRT6_DEL_MFC_PROXY:
1723 if (optlen < sizeof(mfc))
1724 return -EINVAL;
1725 if (copy_from_user(&mfc, optval, sizeof(mfc)))
1726 return -EFAULT;
1727 if (parent == 0)
1728 parent = mfc.mf6cc_parent;
1729 rtnl_lock();
1730 if (optname == MRT6_DEL_MFC || optname == MRT6_DEL_MFC_PROXY)
1731 ret = ip6mr_mfc_delete(mrt, &mfc, parent);
1732 else
1733 ret = ip6mr_mfc_add(net, mrt, &mfc,
1734 sk == mrt->mroute6_sk, parent);
1735 rtnl_unlock();
1736 return ret;
1737
1738
1739
1740
1741 case MRT6_ASSERT:
1742 {
1743 int v;
1744
1745 if (optlen != sizeof(v))
1746 return -EINVAL;
1747 if (get_user(v, (int __user *)optval))
1748 return -EFAULT;
1749 mrt->mroute_do_assert = v;
1750 return 0;
1751 }
1752
1753#ifdef CONFIG_IPV6_PIMSM_V2
1754 case MRT6_PIM:
1755 {
1756 int v;
1757
1758 if (optlen != sizeof(v))
1759 return -EINVAL;
1760 if (get_user(v, (int __user *)optval))
1761 return -EFAULT;
1762 v = !!v;
1763 rtnl_lock();
1764 ret = 0;
1765 if (v != mrt->mroute_do_pim) {
1766 mrt->mroute_do_pim = v;
1767 mrt->mroute_do_assert = v;
1768 }
1769 rtnl_unlock();
1770 return ret;
1771 }
1772
1773#endif
1774#ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES
1775 case MRT6_TABLE:
1776 {
1777 u32 v;
1778
1779 if (optlen != sizeof(u32))
1780 return -EINVAL;
1781 if (get_user(v, (u32 __user *)optval))
1782 return -EFAULT;
1783
1784 if (v != RT_TABLE_DEFAULT && v >= 100000000)
1785 return -EINVAL;
1786 if (sk == mrt->mroute6_sk)
1787 return -EBUSY;
1788
1789 rtnl_lock();
1790 ret = 0;
1791 if (!ip6mr_new_table(net, v))
1792 ret = -ENOMEM;
1793 raw6_sk(sk)->ip6mr_table = v;
1794 rtnl_unlock();
1795 return ret;
1796 }
1797#endif
1798
1799
1800
1801
1802 default:
1803 return -ENOPROTOOPT;
1804 }
1805}
1806
1807
1808
1809
1810
1811int ip6_mroute_getsockopt(struct sock *sk, int optname, char __user *optval,
1812 int __user *optlen)
1813{
1814 int olr;
1815 int val;
1816 struct net *net = sock_net(sk);
1817 struct mr6_table *mrt;
1818
1819 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
1820 if (!mrt)
1821 return -ENOENT;
1822
1823 switch (optname) {
1824 case MRT6_VERSION:
1825 val = 0x0305;
1826 break;
1827#ifdef CONFIG_IPV6_PIMSM_V2
1828 case MRT6_PIM:
1829 val = mrt->mroute_do_pim;
1830 break;
1831#endif
1832 case MRT6_ASSERT:
1833 val = mrt->mroute_do_assert;
1834 break;
1835 default:
1836 return -ENOPROTOOPT;
1837 }
1838
1839 if (get_user(olr, optlen))
1840 return -EFAULT;
1841
1842 olr = min_t(int, olr, sizeof(int));
1843 if (olr < 0)
1844 return -EINVAL;
1845
1846 if (put_user(olr, optlen))
1847 return -EFAULT;
1848 if (copy_to_user(optval, &val, olr))
1849 return -EFAULT;
1850 return 0;
1851}
1852
1853
1854
1855
1856
1857int ip6mr_ioctl(struct sock *sk, int cmd, void __user *arg)
1858{
1859 struct sioc_sg_req6 sr;
1860 struct sioc_mif_req6 vr;
1861 struct mif_device *vif;
1862 struct mfc6_cache *c;
1863 struct net *net = sock_net(sk);
1864 struct mr6_table *mrt;
1865
1866 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
1867 if (!mrt)
1868 return -ENOENT;
1869
1870 switch (cmd) {
1871 case SIOCGETMIFCNT_IN6:
1872 if (copy_from_user(&vr, arg, sizeof(vr)))
1873 return -EFAULT;
1874 if (vr.mifi >= mrt->maxvif)
1875 return -EINVAL;
1876 read_lock(&mrt_lock);
1877 vif = &mrt->vif6_table[vr.mifi];
1878 if (MIF_EXISTS(mrt, vr.mifi)) {
1879 vr.icount = vif->pkt_in;
1880 vr.ocount = vif->pkt_out;
1881 vr.ibytes = vif->bytes_in;
1882 vr.obytes = vif->bytes_out;
1883 read_unlock(&mrt_lock);
1884
1885 if (copy_to_user(arg, &vr, sizeof(vr)))
1886 return -EFAULT;
1887 return 0;
1888 }
1889 read_unlock(&mrt_lock);
1890 return -EADDRNOTAVAIL;
1891 case SIOCGETSGCNT_IN6:
1892 if (copy_from_user(&sr, arg, sizeof(sr)))
1893 return -EFAULT;
1894
1895 read_lock(&mrt_lock);
1896 c = ip6mr_cache_find(mrt, &sr.src.sin6_addr, &sr.grp.sin6_addr);
1897 if (c) {
1898 sr.pktcnt = c->mfc_un.res.pkt;
1899 sr.bytecnt = c->mfc_un.res.bytes;
1900 sr.wrong_if = c->mfc_un.res.wrong_if;
1901 read_unlock(&mrt_lock);
1902
1903 if (copy_to_user(arg, &sr, sizeof(sr)))
1904 return -EFAULT;
1905 return 0;
1906 }
1907 read_unlock(&mrt_lock);
1908 return -EADDRNOTAVAIL;
1909 default:
1910 return -ENOIOCTLCMD;
1911 }
1912}
1913
1914#ifdef CONFIG_COMPAT
1915struct compat_sioc_sg_req6 {
1916 struct sockaddr_in6 src;
1917 struct sockaddr_in6 grp;
1918 compat_ulong_t pktcnt;
1919 compat_ulong_t bytecnt;
1920 compat_ulong_t wrong_if;
1921};
1922
1923struct compat_sioc_mif_req6 {
1924 mifi_t mifi;
1925 compat_ulong_t icount;
1926 compat_ulong_t ocount;
1927 compat_ulong_t ibytes;
1928 compat_ulong_t obytes;
1929};
1930
1931int ip6mr_compat_ioctl(struct sock *sk, unsigned int cmd, void __user *arg)
1932{
1933 struct compat_sioc_sg_req6 sr;
1934 struct compat_sioc_mif_req6 vr;
1935 struct mif_device *vif;
1936 struct mfc6_cache *c;
1937 struct net *net = sock_net(sk);
1938 struct mr6_table *mrt;
1939
1940 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
1941 if (!mrt)
1942 return -ENOENT;
1943
1944 switch (cmd) {
1945 case SIOCGETMIFCNT_IN6:
1946 if (copy_from_user(&vr, arg, sizeof(vr)))
1947 return -EFAULT;
1948 if (vr.mifi >= mrt->maxvif)
1949 return -EINVAL;
1950 read_lock(&mrt_lock);
1951 vif = &mrt->vif6_table[vr.mifi];
1952 if (MIF_EXISTS(mrt, vr.mifi)) {
1953 vr.icount = vif->pkt_in;
1954 vr.ocount = vif->pkt_out;
1955 vr.ibytes = vif->bytes_in;
1956 vr.obytes = vif->bytes_out;
1957 read_unlock(&mrt_lock);
1958
1959 if (copy_to_user(arg, &vr, sizeof(vr)))
1960 return -EFAULT;
1961 return 0;
1962 }
1963 read_unlock(&mrt_lock);
1964 return -EADDRNOTAVAIL;
1965 case SIOCGETSGCNT_IN6:
1966 if (copy_from_user(&sr, arg, sizeof(sr)))
1967 return -EFAULT;
1968
1969 read_lock(&mrt_lock);
1970 c = ip6mr_cache_find(mrt, &sr.src.sin6_addr, &sr.grp.sin6_addr);
1971 if (c) {
1972 sr.pktcnt = c->mfc_un.res.pkt;
1973 sr.bytecnt = c->mfc_un.res.bytes;
1974 sr.wrong_if = c->mfc_un.res.wrong_if;
1975 read_unlock(&mrt_lock);
1976
1977 if (copy_to_user(arg, &sr, sizeof(sr)))
1978 return -EFAULT;
1979 return 0;
1980 }
1981 read_unlock(&mrt_lock);
1982 return -EADDRNOTAVAIL;
1983 default:
1984 return -ENOIOCTLCMD;
1985 }
1986}
1987#endif
1988
1989static inline int ip6mr_forward2_finish(struct sock *sk, struct sk_buff *skb)
1990{
1991 IP6_INC_STATS_BH(dev_net(skb_dst(skb)->dev), ip6_dst_idev(skb_dst(skb)),
1992 IPSTATS_MIB_OUTFORWDATAGRAMS);
1993 IP6_ADD_STATS_BH(dev_net(skb_dst(skb)->dev), ip6_dst_idev(skb_dst(skb)),
1994 IPSTATS_MIB_OUTOCTETS, skb->len);
1995 return dst_output_sk(sk, skb);
1996}
1997
1998
1999
2000
2001
2002static int ip6mr_forward2(struct net *net, struct mr6_table *mrt,
2003 struct sk_buff *skb, struct mfc6_cache *c, int vifi)
2004{
2005 struct ipv6hdr *ipv6h;
2006 struct mif_device *vif = &mrt->vif6_table[vifi];
2007 struct net_device *dev;
2008 struct dst_entry *dst;
2009 struct flowi6 fl6;
2010
2011 if (!vif->dev)
2012 goto out_free;
2013
2014#ifdef CONFIG_IPV6_PIMSM_V2
2015 if (vif->flags & MIFF_REGISTER) {
2016 vif->pkt_out++;
2017 vif->bytes_out += skb->len;
2018 vif->dev->stats.tx_bytes += skb->len;
2019 vif->dev->stats.tx_packets++;
2020 ip6mr_cache_report(mrt, skb, vifi, MRT6MSG_WHOLEPKT);
2021 goto out_free;
2022 }
2023#endif
2024
2025 ipv6h = ipv6_hdr(skb);
2026
2027 fl6 = (struct flowi6) {
2028 .flowi6_oif = vif->link,
2029 .daddr = ipv6h->daddr,
2030 };
2031
2032 dst = ip6_route_output(net, NULL, &fl6);
2033 if (dst->error) {
2034 dst_release(dst);
2035 goto out_free;
2036 }
2037
2038 skb_dst_drop(skb);
2039 skb_dst_set(skb, dst);
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052 dev = vif->dev;
2053 skb->dev = dev;
2054 vif->pkt_out++;
2055 vif->bytes_out += skb->len;
2056
2057
2058
2059 if (skb_cow(skb, sizeof(*ipv6h) + LL_RESERVED_SPACE(dev)))
2060 goto out_free;
2061
2062 ipv6h = ipv6_hdr(skb);
2063 ipv6h->hop_limit--;
2064
2065 IP6CB(skb)->flags |= IP6SKB_FORWARDED;
2066
2067 return NF_HOOK(NFPROTO_IPV6, NF_INET_FORWARD, NULL, skb,
2068 skb->dev, dev,
2069 ip6mr_forward2_finish);
2070
2071out_free:
2072 kfree_skb(skb);
2073 return 0;
2074}
2075
2076static int ip6mr_find_vif(struct mr6_table *mrt, struct net_device *dev)
2077{
2078 int ct;
2079
2080 for (ct = mrt->maxvif - 1; ct >= 0; ct--) {
2081 if (mrt->vif6_table[ct].dev == dev)
2082 break;
2083 }
2084 return ct;
2085}
2086
2087static void ip6_mr_forward(struct net *net, struct mr6_table *mrt,
2088 struct sk_buff *skb, struct mfc6_cache *cache)
2089{
2090 int psend = -1;
2091 int vif, ct;
2092 int true_vifi = ip6mr_find_vif(mrt, skb->dev);
2093
2094 vif = cache->mf6c_parent;
2095 cache->mfc_un.res.pkt++;
2096 cache->mfc_un.res.bytes += skb->len;
2097
2098 if (ipv6_addr_any(&cache->mf6c_origin) && true_vifi >= 0) {
2099 struct mfc6_cache *cache_proxy;
2100
2101
2102
2103
2104 cache_proxy = ip6mr_cache_find_any_parent(mrt, vif);
2105 if (cache_proxy &&
2106 cache_proxy->mfc_un.res.ttls[true_vifi] < 255)
2107 goto forward;
2108 }
2109
2110
2111
2112
2113 if (mrt->vif6_table[vif].dev != skb->dev) {
2114 cache->mfc_un.res.wrong_if++;
2115
2116 if (true_vifi >= 0 && mrt->mroute_do_assert &&
2117
2118
2119
2120
2121
2122 (mrt->mroute_do_pim ||
2123 cache->mfc_un.res.ttls[true_vifi] < 255) &&
2124 time_after(jiffies,
2125 cache->mfc_un.res.last_assert + MFC_ASSERT_THRESH)) {
2126 cache->mfc_un.res.last_assert = jiffies;
2127 ip6mr_cache_report(mrt, skb, true_vifi, MRT6MSG_WRONGMIF);
2128 }
2129 goto dont_forward;
2130 }
2131
2132forward:
2133 mrt->vif6_table[vif].pkt_in++;
2134 mrt->vif6_table[vif].bytes_in += skb->len;
2135
2136
2137
2138
2139 if (ipv6_addr_any(&cache->mf6c_origin) &&
2140 ipv6_addr_any(&cache->mf6c_mcastgrp)) {
2141 if (true_vifi >= 0 &&
2142 true_vifi != cache->mf6c_parent &&
2143 ipv6_hdr(skb)->hop_limit >
2144 cache->mfc_un.res.ttls[cache->mf6c_parent]) {
2145
2146
2147
2148
2149 psend = cache->mf6c_parent;
2150 goto last_forward;
2151 }
2152 goto dont_forward;
2153 }
2154 for (ct = cache->mfc_un.res.maxvif - 1; ct >= cache->mfc_un.res.minvif; ct--) {
2155
2156 if ((!ipv6_addr_any(&cache->mf6c_origin) || ct != true_vifi) &&
2157 ipv6_hdr(skb)->hop_limit > cache->mfc_un.res.ttls[ct]) {
2158 if (psend != -1) {
2159 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
2160 if (skb2)
2161 ip6mr_forward2(net, mrt, skb2, cache, psend);
2162 }
2163 psend = ct;
2164 }
2165 }
2166last_forward:
2167 if (psend != -1) {
2168 ip6mr_forward2(net, mrt, skb, cache, psend);
2169 return;
2170 }
2171
2172dont_forward:
2173 kfree_skb(skb);
2174}
2175
2176
2177
2178
2179
2180
2181int ip6_mr_input(struct sk_buff *skb)
2182{
2183 struct mfc6_cache *cache;
2184 struct net *net = dev_net(skb->dev);
2185 struct mr6_table *mrt;
2186 struct flowi6 fl6 = {
2187 .flowi6_iif = skb->dev->ifindex,
2188 .flowi6_mark = skb->mark,
2189 };
2190 int err;
2191
2192 err = ip6mr_fib_lookup(net, &fl6, &mrt);
2193 if (err < 0) {
2194 kfree_skb(skb);
2195 return err;
2196 }
2197
2198 read_lock(&mrt_lock);
2199 cache = ip6mr_cache_find(mrt,
2200 &ipv6_hdr(skb)->saddr, &ipv6_hdr(skb)->daddr);
2201 if (!cache) {
2202 int vif = ip6mr_find_vif(mrt, skb->dev);
2203
2204 if (vif >= 0)
2205 cache = ip6mr_cache_find_any(mrt,
2206 &ipv6_hdr(skb)->daddr,
2207 vif);
2208 }
2209
2210
2211
2212
2213 if (!cache) {
2214 int vif;
2215
2216 vif = ip6mr_find_vif(mrt, skb->dev);
2217 if (vif >= 0) {
2218 int err = ip6mr_cache_unresolved(mrt, vif, skb);
2219 read_unlock(&mrt_lock);
2220
2221 return err;
2222 }
2223 read_unlock(&mrt_lock);
2224 kfree_skb(skb);
2225 return -ENODEV;
2226 }
2227
2228 ip6_mr_forward(net, mrt, skb, cache);
2229
2230 read_unlock(&mrt_lock);
2231
2232 return 0;
2233}
2234
2235
2236static int __ip6mr_fill_mroute(struct mr6_table *mrt, struct sk_buff *skb,
2237 struct mfc6_cache *c, struct rtmsg *rtm)
2238{
2239 int ct;
2240 struct rtnexthop *nhp;
2241 struct nlattr *mp_attr;
2242 struct rta_mfc_stats mfcs;
2243
2244
2245 if (c->mf6c_parent >= MAXMIFS)
2246 return -ENOENT;
2247
2248 if (MIF_EXISTS(mrt, c->mf6c_parent) &&
2249 nla_put_u32(skb, RTA_IIF, mrt->vif6_table[c->mf6c_parent].dev->ifindex) < 0)
2250 return -EMSGSIZE;
2251 mp_attr = nla_nest_start(skb, RTA_MULTIPATH);
2252 if (!mp_attr)
2253 return -EMSGSIZE;
2254
2255 for (ct = c->mfc_un.res.minvif; ct < c->mfc_un.res.maxvif; ct++) {
2256 if (MIF_EXISTS(mrt, ct) && c->mfc_un.res.ttls[ct] < 255) {
2257 nhp = nla_reserve_nohdr(skb, sizeof(*nhp));
2258 if (!nhp) {
2259 nla_nest_cancel(skb, mp_attr);
2260 return -EMSGSIZE;
2261 }
2262
2263 nhp->rtnh_flags = 0;
2264 nhp->rtnh_hops = c->mfc_un.res.ttls[ct];
2265 nhp->rtnh_ifindex = mrt->vif6_table[ct].dev->ifindex;
2266 nhp->rtnh_len = sizeof(*nhp);
2267 }
2268 }
2269
2270 nla_nest_end(skb, mp_attr);
2271
2272 mfcs.mfcs_packets = c->mfc_un.res.pkt;
2273 mfcs.mfcs_bytes = c->mfc_un.res.bytes;
2274 mfcs.mfcs_wrong_if = c->mfc_un.res.wrong_if;
2275 if (nla_put(skb, RTA_MFC_STATS, sizeof(mfcs), &mfcs) < 0)
2276 return -EMSGSIZE;
2277
2278 rtm->rtm_type = RTN_MULTICAST;
2279 return 1;
2280}
2281
2282int ip6mr_get_route(struct net *net,
2283 struct sk_buff *skb, struct rtmsg *rtm, int nowait)
2284{
2285 int err;
2286 struct mr6_table *mrt;
2287 struct mfc6_cache *cache;
2288 struct rt6_info *rt = (struct rt6_info *)skb_dst(skb);
2289
2290 mrt = ip6mr_get_table(net, RT6_TABLE_DFLT);
2291 if (!mrt)
2292 return -ENOENT;
2293
2294 read_lock(&mrt_lock);
2295 cache = ip6mr_cache_find(mrt, &rt->rt6i_src.addr, &rt->rt6i_dst.addr);
2296 if (!cache && skb->dev) {
2297 int vif = ip6mr_find_vif(mrt, skb->dev);
2298
2299 if (vif >= 0)
2300 cache = ip6mr_cache_find_any(mrt, &rt->rt6i_dst.addr,
2301 vif);
2302 }
2303
2304 if (!cache) {
2305 struct sk_buff *skb2;
2306 struct ipv6hdr *iph;
2307 struct net_device *dev;
2308 int vif;
2309
2310 if (nowait) {
2311 read_unlock(&mrt_lock);
2312 return -EAGAIN;
2313 }
2314
2315 dev = skb->dev;
2316 if (!dev || (vif = ip6mr_find_vif(mrt, dev)) < 0) {
2317 read_unlock(&mrt_lock);
2318 return -ENODEV;
2319 }
2320
2321
2322 skb2 = alloc_skb(sizeof(struct ipv6hdr), GFP_ATOMIC);
2323 if (!skb2) {
2324 read_unlock(&mrt_lock);
2325 return -ENOMEM;
2326 }
2327
2328 skb_reset_transport_header(skb2);
2329
2330 skb_put(skb2, sizeof(struct ipv6hdr));
2331 skb_reset_network_header(skb2);
2332
2333 iph = ipv6_hdr(skb2);
2334 iph->version = 0;
2335 iph->priority = 0;
2336 iph->flow_lbl[0] = 0;
2337 iph->flow_lbl[1] = 0;
2338 iph->flow_lbl[2] = 0;
2339 iph->payload_len = 0;
2340 iph->nexthdr = IPPROTO_NONE;
2341 iph->hop_limit = 0;
2342 iph->saddr = rt->rt6i_src.addr;
2343 iph->daddr = rt->rt6i_dst.addr;
2344
2345 err = ip6mr_cache_unresolved(mrt, vif, skb2);
2346 read_unlock(&mrt_lock);
2347
2348 return err;
2349 }
2350
2351 if (!nowait && (rtm->rtm_flags&RTM_F_NOTIFY))
2352 cache->mfc_flags |= MFC_NOTIFY;
2353
2354 err = __ip6mr_fill_mroute(mrt, skb, cache, rtm);
2355 read_unlock(&mrt_lock);
2356 return err;
2357}
2358
2359static int ip6mr_fill_mroute(struct mr6_table *mrt, struct sk_buff *skb,
2360 u32 portid, u32 seq, struct mfc6_cache *c, int cmd,
2361 int flags)
2362{
2363 struct nlmsghdr *nlh;
2364 struct rtmsg *rtm;
2365 int err;
2366
2367 nlh = nlmsg_put(skb, portid, seq, cmd, sizeof(*rtm), flags);
2368 if (!nlh)
2369 return -EMSGSIZE;
2370
2371 rtm = nlmsg_data(nlh);
2372 rtm->rtm_family = RTNL_FAMILY_IP6MR;
2373 rtm->rtm_dst_len = 128;
2374 rtm->rtm_src_len = 128;
2375 rtm->rtm_tos = 0;
2376 rtm->rtm_table = mrt->id;
2377 if (nla_put_u32(skb, RTA_TABLE, mrt->id))
2378 goto nla_put_failure;
2379 rtm->rtm_type = RTN_MULTICAST;
2380 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
2381 if (c->mfc_flags & MFC_STATIC)
2382 rtm->rtm_protocol = RTPROT_STATIC;
2383 else
2384 rtm->rtm_protocol = RTPROT_MROUTED;
2385 rtm->rtm_flags = 0;
2386
2387 if (nla_put_in6_addr(skb, RTA_SRC, &c->mf6c_origin) ||
2388 nla_put_in6_addr(skb, RTA_DST, &c->mf6c_mcastgrp))
2389 goto nla_put_failure;
2390 err = __ip6mr_fill_mroute(mrt, skb, c, rtm);
2391
2392 if (err < 0 && err != -ENOENT)
2393 goto nla_put_failure;
2394
2395 nlmsg_end(skb, nlh);
2396 return 0;
2397
2398nla_put_failure:
2399 nlmsg_cancel(skb, nlh);
2400 return -EMSGSIZE;
2401}
2402
2403static int mr6_msgsize(bool unresolved, int maxvif)
2404{
2405 size_t len =
2406 NLMSG_ALIGN(sizeof(struct rtmsg))
2407 + nla_total_size(4)
2408 + nla_total_size(sizeof(struct in6_addr))
2409 + nla_total_size(sizeof(struct in6_addr))
2410 ;
2411
2412 if (!unresolved)
2413 len = len
2414 + nla_total_size(4)
2415 + nla_total_size(0)
2416 + maxvif * NLA_ALIGN(sizeof(struct rtnexthop))
2417
2418 + nla_total_size(sizeof(struct rta_mfc_stats))
2419 ;
2420
2421 return len;
2422}
2423
2424static void mr6_netlink_event(struct mr6_table *mrt, struct mfc6_cache *mfc,
2425 int cmd)
2426{
2427 struct net *net = read_pnet(&mrt->net);
2428 struct sk_buff *skb;
2429 int err = -ENOBUFS;
2430
2431 skb = nlmsg_new(mr6_msgsize(mfc->mf6c_parent >= MAXMIFS, mrt->maxvif),
2432 GFP_ATOMIC);
2433 if (!skb)
2434 goto errout;
2435
2436 err = ip6mr_fill_mroute(mrt, skb, 0, 0, mfc, cmd, 0);
2437 if (err < 0)
2438 goto errout;
2439
2440 rtnl_notify(skb, net, 0, RTNLGRP_IPV6_MROUTE, NULL, GFP_ATOMIC);
2441 return;
2442
2443errout:
2444 kfree_skb(skb);
2445 if (err < 0)
2446 rtnl_set_sk_err(net, RTNLGRP_IPV6_MROUTE, err);
2447}
2448
2449static int ip6mr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb)
2450{
2451 struct net *net = sock_net(skb->sk);
2452 struct mr6_table *mrt;
2453 struct mfc6_cache *mfc;
2454 unsigned int t = 0, s_t;
2455 unsigned int h = 0, s_h;
2456 unsigned int e = 0, s_e;
2457
2458 s_t = cb->args[0];
2459 s_h = cb->args[1];
2460 s_e = cb->args[2];
2461
2462 read_lock(&mrt_lock);
2463 ip6mr_for_each_table(mrt, net) {
2464 if (t < s_t)
2465 goto next_table;
2466 if (t > s_t)
2467 s_h = 0;
2468 for (h = s_h; h < MFC6_LINES; h++) {
2469 list_for_each_entry(mfc, &mrt->mfc6_cache_array[h], list) {
2470 if (e < s_e)
2471 goto next_entry;
2472 if (ip6mr_fill_mroute(mrt, skb,
2473 NETLINK_CB(cb->skb).portid,
2474 cb->nlh->nlmsg_seq,
2475 mfc, RTM_NEWROUTE,
2476 NLM_F_MULTI) < 0)
2477 goto done;
2478next_entry:
2479 e++;
2480 }
2481 e = s_e = 0;
2482 }
2483 spin_lock_bh(&mfc_unres_lock);
2484 list_for_each_entry(mfc, &mrt->mfc6_unres_queue, list) {
2485 if (e < s_e)
2486 goto next_entry2;
2487 if (ip6mr_fill_mroute(mrt, skb,
2488 NETLINK_CB(cb->skb).portid,
2489 cb->nlh->nlmsg_seq,
2490 mfc, RTM_NEWROUTE,
2491 NLM_F_MULTI) < 0) {
2492 spin_unlock_bh(&mfc_unres_lock);
2493 goto done;
2494 }
2495next_entry2:
2496 e++;
2497 }
2498 spin_unlock_bh(&mfc_unres_lock);
2499 e = s_e = 0;
2500 s_h = 0;
2501next_table:
2502 t++;
2503 }
2504done:
2505 read_unlock(&mrt_lock);
2506
2507 cb->args[2] = e;
2508 cb->args[1] = h;
2509 cb->args[0] = t;
2510
2511 return skb->len;
2512}
2513