projects / linux-2.6-block.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
ASoC: SOF: Drop superfluous snd_pcm_sgbuf_ops_page
[linux-2.6-block.git] / net / ipv6 / ip6mr.c
CommitLineData
2874c5fd 1// SPDX-License-Identifier: GPL-2.0-or-later
7bc570c8
YH		 2/*
3 * Linux IPv6 multicast routing support for BSD pim6sd
4 * Based on net/ipv4/ipmr.c.
5 *
6 * (c) 2004 Mickael Hoerdt, <hoerdt@clarinet.u-strasbg.fr>
7 * LSIIT Laboratory, Strasbourg, France
8 * (c) 2004 Jean-Philippe Andriot, <jean-philippe.andriot@6WIND.com>
9 * 6WIND, Paris, France
10 * Copyright (C)2007,2008 USAGI/WIDE Project
11 * YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
7bc570c8
YH		 12 */
13
7c0f6ba6 14#include <linux/uaccess.h>
7bc570c8
YH		 15#include <linux/types.h>
16#include <linux/sched.h>
17#include <linux/errno.h>
7bc570c8
YH		 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>
7bc570c8
YH		 23#include <linux/inet.h>
24#include <linux/netdevice.h>
25#include <linux/inetdevice.h>
7bc570c8
YH		 26#include <linux/proc_fs.h>
27#include <linux/seq_file.h>
7bc570c8 28#include <linux/init.h>
e2d57766 29#include <linux/compat.h>
0eb71a9d 30#include <linux/rhashtable.h>
7bc570c8
YH		 31#include <net/protocol.h>
32#include <linux/skbuff.h>
7bc570c8 33#include <net/raw.h>
7bc570c8
YH		 34#include <linux/notifier.h>
35#include <linux/if_arp.h>
7bc570c8
YH		 36#include <net/checksum.h>
37#include <net/netlink.h>
d1db275d 38#include <net/fib_rules.h>
7bc570c8
YH		 39
40#include <net/ipv6.h>
41#include <net/ip6_route.h>
42#include <linux/mroute6.h>
14fb64e1 43#include <linux/pim.h>
7bc570c8
YH		 44#include <net/addrconf.h>
45#include <linux/netfilter_ipv6.h>
bc3b2d7f 46#include <linux/export.h>
5d6e430d 47#include <net/ip6_checksum.h>
d67b8c61 48#include <linux/netconf.h>
cb9f1b78 49#include <net/ip_tunnels.h>
7bc570c8 50
69d2c867
GS		 51#include <linux/nospec.h>
52
d1db275d
PM		 53struct ip6mr_rule {
54 struct fib_rule common;
55};
56
57struct ip6mr_result {
b70432f7 58 struct mr_table *mrt;
d1db275d
PM		 59};
60
7bc570c8
YH		 61/* Big lock, protecting vif table, mrt cache and mroute socket state.
62 Note that the changes are semaphored via rtnl_lock.
63 */
64
65static DEFINE_RWLOCK(mrt_lock);
66
b70432f7 67/* Multicast router control variables */
7bc570c8 68
7bc570c8
YH		 69/* Special spinlock for queue of unresolved entries */
70static DEFINE_SPINLOCK(mfc_unres_lock);
71
72/* We return to original Alan's scheme. Hash table of resolved
73 entries is changed only in process context and protected
74 with weak lock mrt_lock. Queue of unresolved entries is protected
75 with strong spinlock mfc_unres_lock.
76
77 In this case data path is free of exclusive locks at all.
78 */
79
80static struct kmem_cache *mrt_cachep __read_mostly;
81
b70432f7
YM		 82static struct mr_table *ip6mr_new_table(struct net *net, u32 id);
83static void ip6mr_free_table(struct mr_table *mrt);
d1db275d 84
b70432f7 85static void ip6_mr_forward(struct net *net, struct mr_table *mrt,
e4a38c0c
PR		 86 struct net_device *dev, struct sk_buff *skb,
87 struct mfc6_cache *cache);
b70432f7 88static int ip6mr_cache_report(struct mr_table *mrt, struct sk_buff *pkt,
8229efda 89 mifi_t mifi, int assert);
b70432f7 90static void mr6_netlink_event(struct mr_table *mrt, struct mfc6_cache *mfc,
812e44dd 91 int cmd);
b70432f7 92static void mrt6msg_netlink_event(struct mr_table *mrt, struct sk_buff *pkt);
5b285cac
PM		 93static int ip6mr_rtm_dumproute(struct sk_buff *skb,
94 struct netlink_callback *cb);
ca8d4794 95static void mroute_clean_tables(struct mr_table *mrt, int flags);
e99e88a9 96static void ipmr_expire_process(struct timer_list *t);
d1db275d
PM		 97
98#ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES
8ffb335e 99#define ip6mr_for_each_table(mrt, net) \
d1db275d
PM		 100 list_for_each_entry_rcu(mrt, &net->ipv6.mr6_tables, list)
101
7b0db857
YM		 102static struct mr_table *ip6mr_mr_table_iter(struct net *net,
103 struct mr_table *mrt)
104{
105 struct mr_table *ret;
106
107 if (!mrt)
108 ret = list_entry_rcu(net->ipv6.mr6_tables.next,
109 struct mr_table, list);
110 else
111 ret = list_entry_rcu(mrt->list.next,
112 struct mr_table, list);
113
114 if (&ret->list == &net->ipv6.mr6_tables)
115 return NULL;
116 return ret;
117}
118
b70432f7 119static struct mr_table *ip6mr_get_table(struct net *net, u32 id)
d1db275d 120{
b70432f7 121 struct mr_table *mrt;
d1db275d
PM		 122
123 ip6mr_for_each_table(mrt, net) {
124 if (mrt->id == id)
125 return mrt;
126 }
127 return NULL;
128}
129
4c9483b2 130static int ip6mr_fib_lookup(struct net *net, struct flowi6 *flp6,
b70432f7 131 struct mr_table **mrt)
d1db275d 132{
d1db275d 133 int err;
95f4a45d
HFS		 134 struct ip6mr_result res;
135 struct fib_lookup_arg arg = {
136 .result = &res,
137 .flags = FIB_LOOKUP_NOREF,
138 };
d1db275d 139
e4a38c0c
PR		 140 /* update flow if oif or iif point to device enslaved to l3mdev */
141 l3mdev_update_flow(net, flowi6_to_flowi(flp6));
142
4c9483b2
DM		 143 err = fib_rules_lookup(net->ipv6.mr6_rules_ops,
144 flowi6_to_flowi(flp6), 0, &arg);
d1db275d
PM		 145 if (err < 0)
146 return err;
147 *mrt = res.mrt;
148 return 0;
149}
150
151static int ip6mr_rule_action(struct fib_rule *rule, struct flowi *flp,
152 int flags, struct fib_lookup_arg *arg)
153{
154 struct ip6mr_result *res = arg->result;
b70432f7 155 struct mr_table *mrt;
d1db275d
PM		 156
157 switch (rule->action) {
158 case FR_ACT_TO_TBL:
159 break;
160 case FR_ACT_UNREACHABLE:
161 return -ENETUNREACH;
162 case FR_ACT_PROHIBIT:
163 return -EACCES;
164 case FR_ACT_BLACKHOLE:
165 default:
166 return -EINVAL;
167 }
168
e4a38c0c
PR		 169 arg->table = fib_rule_get_table(rule, arg);
170
171 mrt = ip6mr_get_table(rule->fr_net, arg->table);
63159f29 172 if (!mrt)
d1db275d
PM		 173 return -EAGAIN;
174 res->mrt = mrt;
175 return 0;
176}
177
178static int ip6mr_rule_match(struct fib_rule *rule, struct flowi *flp, int flags)
179{
180 return 1;
181}
182
183static const struct nla_policy ip6mr_rule_policy[FRA_MAX + 1] = {
184 FRA_GENERIC_POLICY,
185};
186
187static int ip6mr_rule_configure(struct fib_rule *rule, struct sk_buff *skb,
b16fb418
RP		 188 struct fib_rule_hdr *frh, struct nlattr **tb,
189 struct netlink_ext_ack *extack)
d1db275d
PM		 190{
191 return 0;
192}
193
194static int ip6mr_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh,
195 struct nlattr **tb)
196{
197 return 1;
198}
199
200static int ip6mr_rule_fill(struct fib_rule *rule, struct sk_buff *skb,
201 struct fib_rule_hdr *frh)
202{
203 frh->dst_len = 0;
204 frh->src_len = 0;
205 frh->tos = 0;
206 return 0;
207}
208
04a6f82c 209static const struct fib_rules_ops __net_initconst ip6mr_rules_ops_template = {
d1db275d
PM		 210 .family = RTNL_FAMILY_IP6MR,
211 .rule_size = sizeof(struct ip6mr_rule),
212 .addr_size = sizeof(struct in6_addr),
213 .action = ip6mr_rule_action,
214 .match = ip6mr_rule_match,
215 .configure = ip6mr_rule_configure,
216 .compare = ip6mr_rule_compare,
d1db275d
PM		 217 .fill = ip6mr_rule_fill,
218 .nlgroup = RTNLGRP_IPV6_RULE,
219 .policy = ip6mr_rule_policy,
220 .owner = THIS_MODULE,
221};
222
223static int __net_init ip6mr_rules_init(struct net *net)
224{
225 struct fib_rules_ops *ops;
b70432f7 226 struct mr_table *mrt;
d1db275d
PM		 227 int err;
228
229 ops = fib_rules_register(&ip6mr_rules_ops_template, net);
230 if (IS_ERR(ops))
231 return PTR_ERR(ops);
232
233 INIT_LIST_HEAD(&net->ipv6.mr6_tables);
234
235 mrt = ip6mr_new_table(net, RT6_TABLE_DFLT);
e783bb00
SD		 236 if (IS_ERR(mrt)) {
237 err = PTR_ERR(mrt);
d1db275d
PM		 238 goto err1;
239 }
240
241 err = fib_default_rule_add(ops, 0x7fff, RT6_TABLE_DFLT, 0);
242 if (err < 0)
243 goto err2;
244
245 net->ipv6.mr6_rules_ops = ops;
246 return 0;
247
248err2:
f243e5a7 249 ip6mr_free_table(mrt);
d1db275d
PM		 250err1:
251 fib_rules_unregister(ops);
252 return err;
253}
254
255static void __net_exit ip6mr_rules_exit(struct net *net)
256{
b70432f7 257 struct mr_table *mrt, *next;
d1db275d 258
905a6f96 259 rtnl_lock();
035320d5
ED		 260 list_for_each_entry_safe(mrt, next, &net->ipv6.mr6_tables, list) {
261 list_del(&mrt->list);
d1db275d 262 ip6mr_free_table(mrt);
035320d5 263 }
d1db275d 264 fib_rules_unregister(net->ipv6.mr6_rules_ops);
419df12f 265 rtnl_unlock();
d1db275d 266}
088aa3ee
YM		 267
268static int ip6mr_rules_dump(struct net *net, struct notifier_block *nb)
269{
270 return fib_rules_dump(net, nb, RTNL_FAMILY_IP6MR);
271}
272
273static unsigned int ip6mr_rules_seq_read(struct net *net)
274{
275 return fib_rules_seq_read(net, RTNL_FAMILY_IP6MR);
276}
d3c07e5b
YM		 277
278bool ip6mr_rule_default(const struct fib_rule *rule)
279{
280 return fib_rule_matchall(rule) && rule->action == FR_ACT_TO_TBL &&
281 rule->table == RT6_TABLE_DFLT && !rule->l3mdev;
282}
283EXPORT_SYMBOL(ip6mr_rule_default);
d1db275d
PM		 284#else
285#define ip6mr_for_each_table(mrt, net) \
286 for (mrt = net->ipv6.mrt6; mrt; mrt = NULL)
287
7b0db857
YM		 288static struct mr_table *ip6mr_mr_table_iter(struct net *net,
289 struct mr_table *mrt)
290{
291 if (!mrt)
292 return net->ipv6.mrt6;
293 return NULL;
294}
295
b70432f7 296static struct mr_table *ip6mr_get_table(struct net *net, u32 id)
d1db275d
PM		 297{
298 return net->ipv6.mrt6;
299}
300
4c9483b2 301static int ip6mr_fib_lookup(struct net *net, struct flowi6 *flp6,
b70432f7 302 struct mr_table **mrt)
d1db275d
PM		 303{
304 *mrt = net->ipv6.mrt6;
305 return 0;
306}
307
308static int __net_init ip6mr_rules_init(struct net *net)
309{
e783bb00
SD		 310 struct mr_table *mrt;
311
312 mrt = ip6mr_new_table(net, RT6_TABLE_DFLT);
313 if (IS_ERR(mrt))
314 return PTR_ERR(mrt);
315 net->ipv6.mrt6 = mrt;
316 return 0;
d1db275d
PM		 317}
318
319static void __net_exit ip6mr_rules_exit(struct net *net)
320{
905a6f96 321 rtnl_lock();
d1db275d 322 ip6mr_free_table(net->ipv6.mrt6);
905a6f96
HFS		 323 net->ipv6.mrt6 = NULL;
324 rtnl_unlock();
d1db275d 325}
088aa3ee
YM		 326
327static int ip6mr_rules_dump(struct net *net, struct notifier_block *nb)
328{
329 return 0;
330}
331
332static unsigned int ip6mr_rules_seq_read(struct net *net)
333{
334 return 0;
335}
d1db275d
PM		 336#endif
337
87c418bf
YM		 338static int ip6mr_hash_cmp(struct rhashtable_compare_arg *arg,
339 const void *ptr)
340{
341 const struct mfc6_cache_cmp_arg *cmparg = arg->key;
342 struct mfc6_cache *c = (struct mfc6_cache *)ptr;
343
344 return !ipv6_addr_equal(&c->mf6c_mcastgrp, &cmparg->mf6c_mcastgrp) ||
345 !ipv6_addr_equal(&c->mf6c_origin, &cmparg->mf6c_origin);
346}
347
348static const struct rhashtable_params ip6mr_rht_params = {
494fff56 349 .head_offset = offsetof(struct mr_mfc, mnode),
87c418bf
YM		 350 .key_offset = offsetof(struct mfc6_cache, cmparg),
351 .key_len = sizeof(struct mfc6_cache_cmp_arg),
352 .nelem_hint = 3,
87c418bf
YM		 353 .obj_cmpfn = ip6mr_hash_cmp,
354 .automatic_shrinking = true,
355};
356
0bbbf0e7
YM		 357static void ip6mr_new_table_set(struct mr_table *mrt,
358 struct net *net)
359{
360#ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES
361 list_add_tail_rcu(&mrt->list, &net->ipv6.mr6_tables);
362#endif
363}
364
845c9a7a
YM		 365static struct mfc6_cache_cmp_arg ip6mr_mr_table_ops_cmparg_any = {
366 .mf6c_origin = IN6ADDR_ANY_INIT,
367 .mf6c_mcastgrp = IN6ADDR_ANY_INIT,
368};
369
370static struct mr_table_ops ip6mr_mr_table_ops = {
371 .rht_params = &ip6mr_rht_params,
372 .cmparg_any = &ip6mr_mr_table_ops_cmparg_any,
373};
374
b70432f7 375static struct mr_table *ip6mr_new_table(struct net *net, u32 id)
d1db275d 376{
b70432f7 377 struct mr_table *mrt;
d1db275d
PM		 378
379 mrt = ip6mr_get_table(net, id);
53b24b8f 380 if (mrt)
d1db275d
PM		 381 return mrt;
382
845c9a7a 383 return mr_table_alloc(net, id, &ip6mr_mr_table_ops,
0bbbf0e7 384 ipmr_expire_process, ip6mr_new_table_set);
d1db275d 385}
7bc570c8 386
b70432f7 387static void ip6mr_free_table(struct mr_table *mrt)
d1db275d 388{
7ba0c47c 389 del_timer_sync(&mrt->ipmr_expire_timer);
ca8d4794
CS		 390 mroute_clean_tables(mrt, MRT6_FLUSH_MIFS | MRT6_FLUSH_MIFS_STATIC |
391 MRT6_FLUSH_MFC | MRT6_FLUSH_MFC_STATIC);
b70432f7 392 rhltable_destroy(&mrt->mfc_hash);
d1db275d
PM		 393 kfree(mrt);
394}
7bc570c8
YH		 395
396#ifdef CONFIG_PROC_FS
c8d61968
YM		 397/* The /proc interfaces to multicast routing
398 * /proc/ip6_mr_cache /proc/ip6_mr_vif
7bc570c8
YH		 399 */
400
7bc570c8
YH		 401static void *ip6mr_vif_seq_start(struct seq_file *seq, loff_t *pos)
402 __acquires(mrt_lock)
403{
3feda6b4 404 struct mr_vif_iter *iter = seq->private;
8b90fc7e 405 struct net *net = seq_file_net(seq);
b70432f7 406 struct mr_table *mrt;
d1db275d
PM		 407
408 mrt = ip6mr_get_table(net, RT6_TABLE_DFLT);
63159f29 409 if (!mrt)
d1db275d
PM		 410 return ERR_PTR(-ENOENT);
411
412 iter->mrt = mrt;
8b90fc7e 413
7bc570c8 414 read_lock(&mrt_lock);
3feda6b4 415 return mr_vif_seq_start(seq, pos);
7bc570c8
YH		 416}
417
418static void ip6mr_vif_seq_stop(struct seq_file *seq, void *v)
419 __releases(mrt_lock)
420{
421 read_unlock(&mrt_lock);
422}
423
424static int ip6mr_vif_seq_show(struct seq_file *seq, void *v)
425{
3feda6b4 426 struct mr_vif_iter *iter = seq->private;
b70432f7 427 struct mr_table *mrt = iter->mrt;
8b90fc7e 428
7bc570c8
YH		 429 if (v == SEQ_START_TOKEN) {
430 seq_puts(seq,
431 "Interface BytesIn PktsIn BytesOut PktsOut Flags\n");
432 } else {
6853f21f 433 const struct vif_device *vif = v;
7bc570c8
YH		 434 const char *name = vif->dev ? vif->dev->name : "none";
435
436 seq_printf(seq,
d430a227 437 "%2td %-10s %8ld %7ld %8ld %7ld %05X\n",
b70432f7 438 vif - mrt->vif_table,
7bc570c8
YH		 439 name, vif->bytes_in, vif->pkt_in,
440 vif->bytes_out, vif->pkt_out,
441 vif->flags);
442 }
443 return 0;
444}
445
98147d52 446static const struct seq_operations ip6mr_vif_seq_ops = {
7bc570c8 447 .start = ip6mr_vif_seq_start,
3feda6b4 448 .next = mr_vif_seq_next,
7bc570c8
YH		 449 .stop = ip6mr_vif_seq_stop,
450 .show = ip6mr_vif_seq_show,
451};
452
7bc570c8
YH		 453static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos)
454{
8b90fc7e 455 struct net *net = seq_file_net(seq);
b70432f7 456 struct mr_table *mrt;
8b90fc7e 457
d1db275d 458 mrt = ip6mr_get_table(net, RT6_TABLE_DFLT);
63159f29 459 if (!mrt)
d1db275d
PM		 460 return ERR_PTR(-ENOENT);
461
c8d61968 462 return mr_mfc_seq_start(seq, pos, mrt, &mfc_unres_lock);
7bc570c8
YH		 463}
464
465static int ipmr_mfc_seq_show(struct seq_file *seq, void *v)
466{
467 int n;
468
469 if (v == SEQ_START_TOKEN) {
470 seq_puts(seq,
471 "Group "
472 "Origin "
473 "Iif Pkts Bytes Wrong Oifs\n");
474 } else {
475 const struct mfc6_cache *mfc = v;
c8d61968 476 const struct mr_mfc_iter *it = seq->private;
b70432f7 477 struct mr_table *mrt = it->mrt;
7bc570c8 478
999890b2 479 seq_printf(seq, "%pI6 %pI6 %-3hd",
0c6ce78a 480 &mfc->mf6c_mcastgrp, &mfc->mf6c_origin,
494fff56 481 mfc->_c.mfc_parent);
7bc570c8 482
b70432f7 483 if (it->cache != &mrt->mfc_unres_queue) {
1ea472e2 484 seq_printf(seq, " %8lu %8lu %8lu",
494fff56
YM		 485 mfc->_c.mfc_un.res.pkt,
486 mfc->_c.mfc_un.res.bytes,
487 mfc->_c.mfc_un.res.wrong_if);
488 for (n = mfc->_c.mfc_un.res.minvif;
489 n < mfc->_c.mfc_un.res.maxvif; n++) {
b70432f7 490 if (VIF_EXISTS(mrt, n) &&
494fff56 491 mfc->_c.mfc_un.res.ttls[n] < 255)
7bc570c8 492 seq_printf(seq,
494fff56
YM		 493 " %2d:%-3d", n,
494 mfc->_c.mfc_un.res.ttls[n]);
7bc570c8 495 }
1ea472e2
BT		 496 } else {
497 /* unresolved mfc_caches don't contain
498 * pkt, bytes and wrong_if values
499 */
500 seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul);
7bc570c8
YH		 501 }
502 seq_putc(seq, '\n');
503 }
504 return 0;
505}
506
88e9d34c 507static const struct seq_operations ipmr_mfc_seq_ops = {
7bc570c8 508 .start = ipmr_mfc_seq_start,
c8d61968
YM		 509 .next = mr_mfc_seq_next,
510 .stop = mr_mfc_seq_stop,
7bc570c8
YH		 511 .show = ipmr_mfc_seq_show,
512};
7bc570c8
YH		 513#endif
514
14fb64e1 515#ifdef CONFIG_IPV6_PIMSM_V2
14fb64e1
YH		 516
517static int pim6_rcv(struct sk_buff *skb)
518{
519 struct pimreghdr *pim;
520 struct ipv6hdr *encap;
521 struct net_device *reg_dev = NULL;
8229efda 522 struct net *net = dev_net(skb->dev);
b70432f7 523 struct mr_table *mrt;
4c9483b2
DM		 524 struct flowi6 fl6 = {
525 .flowi6_iif = skb->dev->ifindex,
526 .flowi6_mark = skb->mark,
d1db275d
PM		 527 };
528 int reg_vif_num;
14fb64e1
YH		 529
530 if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(*encap)))
531 goto drop;
532
533 pim = (struct pimreghdr *)skb_transport_header(skb);
56245cae 534 if (pim->type != ((PIM_VERSION << 4) | PIM_TYPE_REGISTER) ||
14fb64e1 535 (pim->flags & PIM_NULL_REGISTER) ||
1d6e55f1
TG		 536 (csum_ipv6_magic(&ipv6_hdr(skb)->saddr, &ipv6_hdr(skb)->daddr,
537 sizeof(*pim), IPPROTO_PIM,
538 csum_partial((void *)pim, sizeof(*pim), 0)) &&
ec6b486f 539 csum_fold(skb_checksum(skb, 0, skb->len, 0))))
14fb64e1
YH		 540 goto drop;
541
542 /* check if the inner packet is destined to mcast group */
543 encap = (struct ipv6hdr *)(skb_transport_header(skb) +
544 sizeof(*pim));
545
546 if (!ipv6_addr_is_multicast(&encap->daddr) ||
547 encap->payload_len == 0 ||
548 ntohs(encap->payload_len) + sizeof(*pim) > skb->len)
549 goto drop;
550
4c9483b2 551 if (ip6mr_fib_lookup(net, &fl6, &mrt) < 0)
d1db275d
PM		 552 goto drop;
553 reg_vif_num = mrt->mroute_reg_vif_num;
554
14fb64e1
YH		 555 read_lock(&mrt_lock);
556 if (reg_vif_num >= 0)
b70432f7 557 reg_dev = mrt->vif_table[reg_vif_num].dev;
14fb64e1
YH		 558 if (reg_dev)
559 dev_hold(reg_dev);
560 read_unlock(&mrt_lock);
561
63159f29 562 if (!reg_dev)
14fb64e1
YH		 563 goto drop;
564
565 skb->mac_header = skb->network_header;
566 skb_pull(skb, (u8 *)encap - skb->data);
567 skb_reset_network_header(skb);
1d6e55f1 568 skb->protocol = htons(ETH_P_IPV6);
3e49e6d5 569 skb->ip_summed = CHECKSUM_NONE;
d19d56dd 570
ea23192e 571 skb_tunnel_rx(skb, reg_dev, dev_net(reg_dev));
d19d56dd 572
caf586e5 573 netif_rx(skb);
8990f468 574
14fb64e1
YH		 575 dev_put(reg_dev);
576 return 0;
577 drop:
578 kfree_skb(skb);
579 return 0;
580}
581
41135cc8 582static const struct inet6_protocol pim6_protocol = {
14fb64e1
YH		 583 .handler = pim6_rcv,
584};
585
586/* Service routines creating virtual interfaces: PIMREG */
587
6fef4c0c
SH		 588static netdev_tx_t reg_vif_xmit(struct sk_buff *skb,
589 struct net_device *dev)
14fb64e1 590{
8229efda 591 struct net *net = dev_net(dev);
b70432f7 592 struct mr_table *mrt;
4c9483b2
DM		 593 struct flowi6 fl6 = {
594 .flowi6_oif = dev->ifindex,
6a662719 595 .flowi6_iif = skb->skb_iif ? : LOOPBACK_IFINDEX,
4c9483b2 596 .flowi6_mark = skb->mark,
d1db275d 597 };
d1db275d 598
cb9f1b78
WB		 599 if (!pskb_inet_may_pull(skb))
600 goto tx_err;
601
602 if (ip6mr_fib_lookup(net, &fl6, &mrt) < 0)
603 goto tx_err;
8229efda 604
14fb64e1 605 read_lock(&mrt_lock);
dc58c78c
PE		 606 dev->stats.tx_bytes += skb->len;
607 dev->stats.tx_packets++;
6bd52143 608 ip6mr_cache_report(mrt, skb, mrt->mroute_reg_vif_num, MRT6MSG_WHOLEPKT);
14fb64e1
YH		 609 read_unlock(&mrt_lock);
610 kfree_skb(skb);
6ed10654 611 return NETDEV_TX_OK;
cb9f1b78
WB		 612
613tx_err:
614 dev->stats.tx_errors++;
615 kfree_skb(skb);
616 return NETDEV_TX_OK;
14fb64e1
YH		 617}
618
ee9b9596
ND		 619static int reg_vif_get_iflink(const struct net_device *dev)
620{
621 return 0;
622}
623
007c3838
SH		 624static const struct net_device_ops reg_vif_netdev_ops = {
625 .ndo_start_xmit = reg_vif_xmit,
ee9b9596 626 .ndo_get_iflink = reg_vif_get_iflink,
007c3838
SH		 627};
628
14fb64e1
YH		 629static void reg_vif_setup(struct net_device *dev)
630{
631 dev->type = ARPHRD_PIMREG;
632 dev->mtu = 1500 - sizeof(struct ipv6hdr) - 8;
633 dev->flags = IFF_NOARP;
007c3838 634 dev->netdev_ops = &reg_vif_netdev_ops;
cf124db5 635 dev->needs_free_netdev = true;
403dbb97 636 dev->features |= NETIF_F_NETNS_LOCAL;
14fb64e1
YH		 637}
638
b70432f7 639static struct net_device *ip6mr_reg_vif(struct net *net, struct mr_table *mrt)
14fb64e1
YH		 640{
641 struct net_device *dev;
d1db275d
PM		 642 char name[IFNAMSIZ];
643
644 if (mrt->id == RT6_TABLE_DFLT)
645 sprintf(name, "pim6reg");
646 else
647 sprintf(name, "pim6reg%u", mrt->id);
14fb64e1 648
c835a677 649 dev = alloc_netdev(0, name, NET_NAME_UNKNOWN, reg_vif_setup);
63159f29 650 if (!dev)
14fb64e1
YH		 651 return NULL;
652
8229efda
BT		 653 dev_net_set(dev, net);
654
14fb64e1
YH		 655 if (register_netdevice(dev)) {
656 free_netdev(dev);
657 return NULL;
658 }
14fb64e1 659
00f54e68 660 if (dev_open(dev, NULL))
14fb64e1
YH		 661 goto failure;
662
7af3db78 663 dev_hold(dev);
14fb64e1
YH		 664 return dev;
665
666failure:
14fb64e1
YH		 667 unregister_netdevice(dev);
668 return NULL;
669}
670#endif
671
088aa3ee
YM		 672static int call_ip6mr_vif_entry_notifiers(struct net *net,
673 enum fib_event_type event_type,
674 struct vif_device *vif,
675 mifi_t vif_index, u32 tb_id)
676{
677 return mr_call_vif_notifiers(net, RTNL_FAMILY_IP6MR, event_type,
678 vif, vif_index, tb_id,
679 &net->ipv6.ipmr_seq);
680}
7bc570c8 681
088aa3ee
YM		 682static int call_ip6mr_mfc_entry_notifiers(struct net *net,
683 enum fib_event_type event_type,
684 struct mfc6_cache *mfc, u32 tb_id)
685{
686 return mr_call_mfc_notifiers(net, RTNL_FAMILY_IP6MR, event_type,
687 &mfc->_c, tb_id, &net->ipv6.ipmr_seq);
688}
689
690/* Delete a VIF entry */
b70432f7 691static int mif6_delete(struct mr_table *mrt, int vifi, int notify,
723b929c 692 struct list_head *head)
7bc570c8 693{
6853f21f 694 struct vif_device *v;
7bc570c8 695 struct net_device *dev;
1d6e55f1 696 struct inet6_dev *in6_dev;
6bd52143
PM		 697
698 if (vifi < 0 || vifi >= mrt->maxvif)
7bc570c8
YH		 699 return -EADDRNOTAVAIL;
700
b70432f7 701 v = &mrt->vif_table[vifi];
7bc570c8 702
088aa3ee
YM		 703 if (VIF_EXISTS(mrt, vifi))
704 call_ip6mr_vif_entry_notifiers(read_pnet(&mrt->net),
705 FIB_EVENT_VIF_DEL, v, vifi,
706 mrt->id);
707
7bc570c8
YH		 708 write_lock_bh(&mrt_lock);
709 dev = v->dev;
710 v->dev = NULL;
711
712 if (!dev) {
713 write_unlock_bh(&mrt_lock);
714 return -EADDRNOTAVAIL;
715 }
716
14fb64e1 717#ifdef CONFIG_IPV6_PIMSM_V2
6bd52143
PM		 718 if (vifi == mrt->mroute_reg_vif_num)
719 mrt->mroute_reg_vif_num = -1;
14fb64e1
YH		 720#endif
721
6bd52143 722 if (vifi + 1 == mrt->maxvif) {
7bc570c8
YH		 723 int tmp;
724 for (tmp = vifi - 1; tmp >= 0; tmp--) {
b70432f7 725 if (VIF_EXISTS(mrt, tmp))
7bc570c8
YH		 726 break;
727 }
6bd52143 728 mrt->maxvif = tmp + 1;
7bc570c8
YH		 729 }
730
731 write_unlock_bh(&mrt_lock);
732
733 dev_set_allmulti(dev, -1);
734
1d6e55f1 735 in6_dev = __in6_dev_get(dev);
d67b8c61 736 if (in6_dev) {
1d6e55f1 737 in6_dev->cnf.mc_forwarding--;
85b3daad 738 inet6_netconf_notify_devconf(dev_net(dev), RTM_NEWNETCONF,
d67b8c61
ND		 739 NETCONFA_MC_FORWARDING,
740 dev->ifindex, &in6_dev->cnf);
741 }
1d6e55f1 742
723b929c 743 if ((v->flags & MIFF_REGISTER) && !notify)
c871e664 744 unregister_netdevice_queue(dev, head);
7bc570c8
YH		 745
746 dev_put(dev);
747 return 0;
748}
749
87c418bf 750static inline void ip6mr_cache_free_rcu(struct rcu_head *head)
58701ad4 751{
494fff56 752 struct mr_mfc *c = container_of(head, struct mr_mfc, rcu);
87c418bf 753
494fff56 754 kmem_cache_free(mrt_cachep, (struct mfc6_cache *)c);
58701ad4
BT		 755}
756
87c418bf
YM		 757static inline void ip6mr_cache_free(struct mfc6_cache *c)
758{
494fff56 759 call_rcu(&c->_c.rcu, ip6mr_cache_free_rcu);
87c418bf
YM		 760}
761
7bc570c8
YH		 762/* Destroy an unresolved cache entry, killing queued skbs
763 and reporting error to netlink readers.
764 */
765
b70432f7 766static void ip6mr_destroy_unres(struct mr_table *mrt, struct mfc6_cache *c)
7bc570c8 767{
6bd52143 768 struct net *net = read_pnet(&mrt->net);
7bc570c8
YH		 769 struct sk_buff *skb;
770
6bd52143 771 atomic_dec(&mrt->cache_resolve_queue_len);
7bc570c8 772
494fff56 773 while ((skb = skb_dequeue(&c->_c.mfc_un.unres.unresolved)) != NULL) {
7bc570c8 774 if (ipv6_hdr(skb)->version == 0) {
af72868b
JB		 775 struct nlmsghdr *nlh = skb_pull(skb,
776 sizeof(struct ipv6hdr));
7bc570c8 777 nlh->nlmsg_type = NLMSG_ERROR;
573ce260 778 nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr));
7bc570c8 779 skb_trim(skb, nlh->nlmsg_len);
573ce260 780 ((struct nlmsgerr *)nlmsg_data(nlh))->error = -ETIMEDOUT;
15e47304 781 rtnl_unicast(skb, net, NETLINK_CB(skb).portid);
7bc570c8
YH		 782 } else
783 kfree_skb(skb);
784 }
785
58701ad4 786 ip6mr_cache_free(c);
7bc570c8
YH		 787}
788
789
c476efbc 790/* Timer process for all the unresolved queue. */
7bc570c8 791
b70432f7 792static void ipmr_do_expire_process(struct mr_table *mrt)
7bc570c8
YH		 793{
794 unsigned long now = jiffies;
795 unsigned long expires = 10 * HZ;
494fff56 796 struct mr_mfc *c, *next;
7bc570c8 797
b70432f7 798 list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) {
7bc570c8
YH		 799 if (time_after(c->mfc_un.unres.expires, now)) {
800 /* not yet... */
801 unsigned long interval = c->mfc_un.unres.expires - now;
802 if (interval < expires)
803 expires = interval;
7bc570c8
YH		 804 continue;
805 }
806
f30a7784 807 list_del(&c->list);
494fff56
YM		 808 mr6_netlink_event(mrt, (struct mfc6_cache *)c, RTM_DELROUTE);
809 ip6mr_destroy_unres(mrt, (struct mfc6_cache *)c);
7bc570c8
YH		 810 }
811
b70432f7 812 if (!list_empty(&mrt->mfc_unres_queue))
6bd52143 813 mod_timer(&mrt->ipmr_expire_timer, jiffies + expires);
7bc570c8
YH		 814}
815
e99e88a9 816static void ipmr_expire_process(struct timer_list *t)
7bc570c8 817{
b70432f7 818 struct mr_table *mrt = from_timer(mrt, t, ipmr_expire_timer);
c476efbc 819
7bc570c8 820 if (!spin_trylock(&mfc_unres_lock)) {
6bd52143 821 mod_timer(&mrt->ipmr_expire_timer, jiffies + 1);
7bc570c8
YH		 822 return;
823 }
824
b70432f7 825 if (!list_empty(&mrt->mfc_unres_queue))
6bd52143 826 ipmr_do_expire_process(mrt);
7bc570c8
YH		 827
828 spin_unlock(&mfc_unres_lock);
829}
830
831/* Fill oifs list. It is called under write locked mrt_lock. */
832
b70432f7 833static void ip6mr_update_thresholds(struct mr_table *mrt,
494fff56 834 struct mr_mfc *cache,
b5aa30b1 835 unsigned char *ttls)
7bc570c8
YH		 836{
837 int vifi;
838
6ac7eb08 839 cache->mfc_un.res.minvif = MAXMIFS;
7bc570c8 840 cache->mfc_un.res.maxvif = 0;
6ac7eb08 841 memset(cache->mfc_un.res.ttls, 255, MAXMIFS);
7bc570c8 842
6bd52143 843 for (vifi = 0; vifi < mrt->maxvif; vifi++) {
b70432f7 844 if (VIF_EXISTS(mrt, vifi) &&
4e16880c 845 ttls[vifi] && ttls[vifi] < 255) {
7bc570c8
YH		 846 cache->mfc_un.res.ttls[vifi] = ttls[vifi];
847 if (cache->mfc_un.res.minvif > vifi)
848 cache->mfc_un.res.minvif = vifi;
849 if (cache->mfc_un.res.maxvif <= vifi)
850 cache->mfc_un.res.maxvif = vifi + 1;
851 }
852 }
90b5ca17 853 cache->mfc_un.res.lastuse = jiffies;
7bc570c8
YH		 854}
855
b70432f7 856static int mif6_add(struct net *net, struct mr_table *mrt,
6bd52143 857 struct mif6ctl *vifc, int mrtsock)
7bc570c8
YH		 858{
859 int vifi = vifc->mif6c_mifi;
b70432f7 860 struct vif_device *v = &mrt->vif_table[vifi];
7bc570c8 861 struct net_device *dev;
1d6e55f1 862 struct inet6_dev *in6_dev;
5ae7b444 863 int err;
7bc570c8
YH		 864
865 /* Is vif busy ? */
b70432f7 866 if (VIF_EXISTS(mrt, vifi))
7bc570c8
YH		 867 return -EADDRINUSE;
868
869 switch (vifc->mif6c_flags) {
14fb64e1
YH		 870#ifdef CONFIG_IPV6_PIMSM_V2
871 case MIFF_REGISTER:
872 /*
873 * Special Purpose VIF in PIM
874 * All the packets will be sent to the daemon
875 */
6bd52143 876 if (mrt->mroute_reg_vif_num >= 0)
14fb64e1 877 return -EADDRINUSE;
d1db275d 878 dev = ip6mr_reg_vif(net, mrt);
14fb64e1
YH		 879 if (!dev)
880 return -ENOBUFS;
5ae7b444
WC		 881 err = dev_set_allmulti(dev, 1);
882 if (err) {
883 unregister_netdevice(dev);
7af3db78 884 dev_put(dev);
5ae7b444
WC		 885 return err;
886 }
14fb64e1
YH		 887 break;
888#endif
7bc570c8 889 case 0:
8229efda 890 dev = dev_get_by_index(net, vifc->mif6c_pifi);
7bc570c8
YH		 891 if (!dev)
892 return -EADDRNOTAVAIL;
5ae7b444 893 err = dev_set_allmulti(dev, 1);
7af3db78
WC		 894 if (err) {
895 dev_put(dev);
5ae7b444 896 return err;
7af3db78 897 }
7bc570c8
YH		 898 break;
899 default:
900 return -EINVAL;
901 }
902
1d6e55f1 903 in6_dev = __in6_dev_get(dev);
d67b8c61 904 if (in6_dev) {
1d6e55f1 905 in6_dev->cnf.mc_forwarding++;
85b3daad 906 inet6_netconf_notify_devconf(dev_net(dev), RTM_NEWNETCONF,
d67b8c61
ND		 907 NETCONFA_MC_FORWARDING,
908 dev->ifindex, &in6_dev->cnf);
909 }
1d6e55f1 910
6853f21f
YM		 911 /* Fill in the VIF structures */
912 vif_device_init(v, dev, vifc->vifc_rate_limit, vifc->vifc_threshold,
913 vifc->mif6c_flags | (!mrtsock ? VIFF_STATIC : 0),
914 MIFF_REGISTER);
7bc570c8
YH		 915
916 /* And finish update writing critical data */
917 write_lock_bh(&mrt_lock);
7bc570c8 918 v->dev = dev;
14fb64e1
YH		 919#ifdef CONFIG_IPV6_PIMSM_V2
920 if (v->flags & MIFF_REGISTER)
6bd52143 921 mrt->mroute_reg_vif_num = vifi;
14fb64e1 922#endif
6bd52143
PM		 923 if (vifi + 1 > mrt->maxvif)
924 mrt->maxvif = vifi + 1;
7bc570c8 925 write_unlock_bh(&mrt_lock);
088aa3ee
YM		 926 call_ip6mr_vif_entry_notifiers(net, FIB_EVENT_VIF_ADD,
927 v, vifi, mrt->id);
7bc570c8
YH		 928 return 0;
929}
930
b70432f7 931static struct mfc6_cache *ip6mr_cache_find(struct mr_table *mrt,
b71d1d42
ED		 932 const struct in6_addr *origin,
933 const struct in6_addr *mcastgrp)
7bc570c8 934{
87c418bf
YM		 935 struct mfc6_cache_cmp_arg arg = {
936 .mf6c_origin = *origin,
937 .mf6c_mcastgrp = *mcastgrp,
938 };
87c418bf 939
845c9a7a 940 return mr_mfc_find(mrt, &arg);
660b26dc
ND		 941}
942
943/* Look for a (*,G) entry */
b70432f7 944static struct mfc6_cache *ip6mr_cache_find_any(struct mr_table *mrt,
660b26dc
ND		 945 struct in6_addr *mcastgrp,
946 mifi_t mifi)
947{
87c418bf
YM		 948 struct mfc6_cache_cmp_arg arg = {
949 .mf6c_origin = in6addr_any,
950 .mf6c_mcastgrp = *mcastgrp,
951 };
660b26dc
ND		 952
953 if (ipv6_addr_any(mcastgrp))
845c9a7a
YM		 954 return mr_mfc_find_any_parent(mrt, mifi);
955 return mr_mfc_find_any(mrt, mifi, &arg);
660b26dc
ND		 956}
957
87c418bf
YM		 958/* Look for a (S,G,iif) entry if parent != -1 */
959static struct mfc6_cache *
b70432f7 960ip6mr_cache_find_parent(struct mr_table *mrt,
87c418bf
YM		 961 const struct in6_addr *origin,
962 const struct in6_addr *mcastgrp,
963 int parent)
964{
965 struct mfc6_cache_cmp_arg arg = {
966 .mf6c_origin = *origin,
967 .mf6c_mcastgrp = *mcastgrp,
968 };
87c418bf 969
845c9a7a 970 return mr_mfc_find_parent(mrt, &arg, parent);
87c418bf
YM		 971}
972
845c9a7a 973/* Allocate a multicast cache entry */
b5aa30b1 974static struct mfc6_cache *ip6mr_cache_alloc(void)
7bc570c8 975{
36cbac59 976 struct mfc6_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL);
63159f29 977 if (!c)
7bc570c8 978 return NULL;
494fff56
YM		 979 c->_c.mfc_un.res.last_assert = jiffies - MFC_ASSERT_THRESH - 1;
980 c->_c.mfc_un.res.minvif = MAXMIFS;
8c13af2a
YM		 981 c->_c.free = ip6mr_cache_free_rcu;
982 refcount_set(&c->_c.mfc_un.res.refcount, 1);
7bc570c8
YH		 983 return c;
984}
985
b5aa30b1 986static struct mfc6_cache *ip6mr_cache_alloc_unres(void)
7bc570c8 987{
36cbac59 988 struct mfc6_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC);
63159f29 989 if (!c)
7bc570c8 990 return NULL;
494fff56
YM		 991 skb_queue_head_init(&c->_c.mfc_un.unres.unresolved);
992 c->_c.mfc_un.unres.expires = jiffies + 10 * HZ;
7bc570c8
YH		 993 return c;
994}
995
996/*
997 * A cache entry has gone into a resolved state from queued
998 */
999
b70432f7 1000static void ip6mr_cache_resolve(struct net *net, struct mr_table *mrt,
6bd52143 1001 struct mfc6_cache *uc, struct mfc6_cache *c)
7bc570c8
YH		 1002{
1003 struct sk_buff *skb;
1004
1005 /*
1006 * Play the pending entries through our router
1007 */
1008
494fff56 1009 while ((skb = __skb_dequeue(&uc->_c.mfc_un.unres.unresolved))) {
7bc570c8 1010 if (ipv6_hdr(skb)->version == 0) {
af72868b
JB		 1011 struct nlmsghdr *nlh = skb_pull(skb,
1012 sizeof(struct ipv6hdr));
7bc570c8 1013
7b0db857
YM		 1014 if (mr_fill_mroute(mrt, skb, &c->_c,
1015 nlmsg_data(nlh)) > 0) {
549e028d 1016 nlh->nlmsg_len = skb_tail_pointer(skb) - (u8 *)nlh;
7bc570c8
YH		 1017 } else {
1018 nlh->nlmsg_type = NLMSG_ERROR;
573ce260 1019 nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr));
7bc570c8 1020 skb_trim(skb, nlh->nlmsg_len);
573ce260 1021 ((struct nlmsgerr *)nlmsg_data(nlh))->error = -EMSGSIZE;
7bc570c8 1022 }
15e47304 1023 rtnl_unicast(skb, net, NETLINK_CB(skb).portid);
7bc570c8 1024 } else
e4a38c0c 1025 ip6_mr_forward(net, mrt, skb->dev, skb, c);
7bc570c8
YH		 1026 }
1027}
1028
1029/*
dd12d15c 1030 * Bounce a cache query up to pim6sd and netlink.
7bc570c8
YH		 1031 *
1032 * Called under mrt_lock.
1033 */
1034
b70432f7 1035static int ip6mr_cache_report(struct mr_table *mrt, struct sk_buff *pkt,
6bd52143 1036 mifi_t mifi, int assert)
7bc570c8 1037{
8571ab47 1038 struct sock *mroute6_sk;
7bc570c8
YH		 1039 struct sk_buff *skb;
1040 struct mrt6msg *msg;
1041 int ret;
1042
14fb64e1
YH		 1043#ifdef CONFIG_IPV6_PIMSM_V2
1044 if (assert == MRT6MSG_WHOLEPKT)
1045 skb = skb_realloc_headroom(pkt, -skb_network_offset(pkt)
1046 +sizeof(*msg));
1047 else
1048#endif
1049 skb = alloc_skb(sizeof(struct ipv6hdr) + sizeof(*msg), GFP_ATOMIC);
7bc570c8
YH		 1050
1051 if (!skb)
1052 return -ENOBUFS;
1053
1054 /* I suppose that internal messages
1055 * do not require checksums */
1056
1057 skb->ip_summed = CHECKSUM_UNNECESSARY;
1058
14fb64e1
YH		 1059#ifdef CONFIG_IPV6_PIMSM_V2
1060 if (assert == MRT6MSG_WHOLEPKT) {
1061 /* Ugly, but we have no choice with this interface.
1062 Duplicate old header, fix length etc.
1063 And all this only to mangle msg->im6_msgtype and
1064 to set msg->im6_mbz to "mbz" :-)
1065 */
1066 skb_push(skb, -skb_network_offset(pkt));
1067
1068 skb_push(skb, sizeof(*msg));
1069 skb_reset_transport_header(skb);
1070 msg = (struct mrt6msg *)skb_transport_header(skb);
1071 msg->im6_mbz = 0;
1072 msg->im6_msgtype = MRT6MSG_WHOLEPKT;
6bd52143 1073 msg->im6_mif = mrt->mroute_reg_vif_num;
14fb64e1 1074 msg->im6_pad = 0;
4e3fd7a0
AD		 1075 msg->im6_src = ipv6_hdr(pkt)->saddr;
1076 msg->im6_dst = ipv6_hdr(pkt)->daddr;
14fb64e1
YH		 1077
1078 skb->ip_summed = CHECKSUM_UNNECESSARY;
1079 } else
1080#endif
1081 {
7bc570c8
YH		 1082 /*
1083 * Copy the IP header
1084 */
1085
1086 skb_put(skb, sizeof(struct ipv6hdr));
1087 skb_reset_network_header(skb);
1088 skb_copy_to_linear_data(skb, ipv6_hdr(pkt), sizeof(struct ipv6hdr));
1089
1090 /*
1091 * Add our header
1092 */
1093 skb_put(skb, sizeof(*msg));
1094 skb_reset_transport_header(skb);
1095 msg = (struct mrt6msg *)skb_transport_header(skb);
1096
1097 msg->im6_mbz = 0;
1098 msg->im6_msgtype = assert;
6ac7eb08 1099 msg->im6_mif = mifi;
7bc570c8 1100 msg->im6_pad = 0;
4e3fd7a0
AD		 1101 msg->im6_src = ipv6_hdr(pkt)->saddr;
1102 msg->im6_dst = ipv6_hdr(pkt)->daddr;
7bc570c8 1103
adf30907 1104 skb_dst_set(skb, dst_clone(skb_dst(pkt)));
7bc570c8 1105 skb->ip_summed = CHECKSUM_UNNECESSARY;
14fb64e1 1106 }
7bc570c8 1107
8571ab47 1108 rcu_read_lock();
b70432f7 1109 mroute6_sk = rcu_dereference(mrt->mroute_sk);
8571ab47
YM		 1110 if (!mroute6_sk) {
1111 rcu_read_unlock();
7bc570c8
YH		 1112 kfree_skb(skb);
1113 return -EINVAL;
1114 }
1115
dd12d15c
JG		 1116 mrt6msg_netlink_event(mrt, skb);
1117
8571ab47
YM		 1118 /* Deliver to user space multicast routing algorithms */
1119 ret = sock_queue_rcv_skb(mroute6_sk, skb);
1120 rcu_read_unlock();
bd91b8bf 1121 if (ret < 0) {
e87cc472 1122 net_warn_ratelimited("mroute6: pending queue full, dropping entries\n");
7bc570c8
YH		 1123 kfree_skb(skb);
1124 }
1125
1126 return ret;
1127}
1128
494fff56
YM		 1129/* Queue a packet for resolution. It gets locked cache entry! */
1130static int ip6mr_cache_unresolved(struct mr_table *mrt, mifi_t mifi,
e4a38c0c 1131 struct sk_buff *skb, struct net_device *dev)
7bc570c8 1132{
494fff56 1133 struct mfc6_cache *c;
f30a7784 1134 bool found = false;
7bc570c8 1135 int err;
7bc570c8
YH		 1136
1137 spin_lock_bh(&mfc_unres_lock);
494fff56 1138 list_for_each_entry(c, &mrt->mfc_unres_queue, _c.list) {
c476efbc 1139 if (ipv6_addr_equal(&c->mf6c_mcastgrp, &ipv6_hdr(skb)->daddr) &&
f30a7784
PM		 1140 ipv6_addr_equal(&c->mf6c_origin, &ipv6_hdr(skb)->saddr)) {
1141 found = true;
7bc570c8 1142 break;
f30a7784 1143 }
7bc570c8
YH		 1144 }
1145
f30a7784 1146 if (!found) {
7bc570c8
YH		 1147 /*
1148 * Create a new entry if allowable
1149 */
1150
0079ad8e
HL		 1151 c = ip6mr_cache_alloc_unres();
1152 if (!c) {
7bc570c8
YH		 1153 spin_unlock_bh(&mfc_unres_lock);
1154
1155 kfree_skb(skb);
1156 return -ENOBUFS;
1157 }
1158
494fff56
YM		 1159 /* Fill in the new cache entry */
1160 c->_c.mfc_parent = -1;
7bc570c8
YH		 1161 c->mf6c_origin = ipv6_hdr(skb)->saddr;
1162 c->mf6c_mcastgrp = ipv6_hdr(skb)->daddr;
1163
1164 /*
1165 * Reflect first query at pim6sd
1166 */
6bd52143 1167 err = ip6mr_cache_report(mrt, skb, mifi, MRT6MSG_NOCACHE);
8229efda 1168 if (err < 0) {
7bc570c8
YH		 1169 /* If the report failed throw the cache entry
1170 out - Brad Parker
1171 */
1172 spin_unlock_bh(&mfc_unres_lock);
1173
58701ad4 1174 ip6mr_cache_free(c);
7bc570c8
YH		 1175 kfree_skb(skb);
1176 return err;
1177 }
1178
6bd52143 1179 atomic_inc(&mrt->cache_resolve_queue_len);
494fff56 1180 list_add(&c->_c.list, &mrt->mfc_unres_queue);
812e44dd 1181 mr6_netlink_event(mrt, c, RTM_NEWROUTE);
7bc570c8 1182
6bd52143 1183 ipmr_do_expire_process(mrt);
7bc570c8
YH		 1184 }
1185
494fff56
YM		 1186 /* See if we can append the packet */
1187 if (c->_c.mfc_un.unres.unresolved.qlen > 3) {
7bc570c8
YH		 1188 kfree_skb(skb);
1189 err = -ENOBUFS;
1190 } else {
e4a38c0c
PR		 1191 if (dev) {
1192 skb->dev = dev;
1193 skb->skb_iif = dev->ifindex;
1194 }
494fff56 1195 skb_queue_tail(&c->_c.mfc_un.unres.unresolved, skb);
7bc570c8
YH		 1196 err = 0;
1197 }
1198
1199 spin_unlock_bh(&mfc_unres_lock);
1200 return err;
1201}
1202
1203/*
1204 * MFC6 cache manipulation by user space
1205 */
1206
b70432f7 1207static int ip6mr_mfc_delete(struct mr_table *mrt, struct mf6cctl *mfc,
660b26dc 1208 int parent)
7bc570c8 1209{
87c418bf 1210 struct mfc6_cache *c;
7bc570c8 1211
87c418bf
YM		 1212 /* The entries are added/deleted only under RTNL */
1213 rcu_read_lock();
1214 c = ip6mr_cache_find_parent(mrt, &mfc->mf6cc_origin.sin6_addr,
1215 &mfc->mf6cc_mcastgrp.sin6_addr, parent);
1216 rcu_read_unlock();
1217 if (!c)
1218 return -ENOENT;
494fff56
YM		 1219 rhltable_remove(&mrt->mfc_hash, &c->_c.mnode, ip6mr_rht_params);
1220 list_del_rcu(&c->_c.list);
7bc570c8 1221
088aa3ee
YM		 1222 call_ip6mr_mfc_entry_notifiers(read_pnet(&mrt->net),
1223 FIB_EVENT_ENTRY_DEL, c, mrt->id);
87c418bf 1224 mr6_netlink_event(mrt, c, RTM_DELROUTE);
8c13af2a 1225 mr_cache_put(&c->_c);
87c418bf 1226 return 0;
7bc570c8
YH		 1227}
1228
1229static int ip6mr_device_event(struct notifier_block *this,
1230 unsigned long event, void *ptr)
1231{
351638e7 1232 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
8229efda 1233 struct net *net = dev_net(dev);
b70432f7 1234 struct mr_table *mrt;
6853f21f 1235 struct vif_device *v;
7bc570c8
YH		 1236 int ct;
1237
7bc570c8
YH		 1238 if (event != NETDEV_UNREGISTER)
1239 return NOTIFY_DONE;
1240
d1db275d 1241 ip6mr_for_each_table(mrt, net) {
b70432f7 1242 v = &mrt->vif_table[0];
d1db275d
PM		 1243 for (ct = 0; ct < mrt->maxvif; ct++, v++) {
1244 if (v->dev == dev)
723b929c 1245 mif6_delete(mrt, ct, 1, NULL);
d1db275d 1246 }
7bc570c8 1247 }
c871e664 1248
7bc570c8
YH		 1249 return NOTIFY_DONE;
1250}
1251
088aa3ee
YM		 1252static unsigned int ip6mr_seq_read(struct net *net)
1253{
1254 ASSERT_RTNL();
1255
1256 return net->ipv6.ipmr_seq + ip6mr_rules_seq_read(net);
1257}
1258
1259static int ip6mr_dump(struct net *net, struct notifier_block *nb)
1260{
1261 return mr_dump(net, nb, RTNL_FAMILY_IP6MR, ip6mr_rules_dump,
1262 ip6mr_mr_table_iter, &mrt_lock);
1263}
1264
7bc570c8
YH		 1265static struct notifier_block ip6_mr_notifier = {
1266 .notifier_call = ip6mr_device_event
1267};
1268
088aa3ee
YM		 1269static const struct fib_notifier_ops ip6mr_notifier_ops_template = {
1270 .family = RTNL_FAMILY_IP6MR,
1271 .fib_seq_read = ip6mr_seq_read,
1272 .fib_dump = ip6mr_dump,
1273 .owner = THIS_MODULE,
1274};
1275
1276static int __net_init ip6mr_notifier_init(struct net *net)
1277{
1278 struct fib_notifier_ops *ops;
1279
1280 net->ipv6.ipmr_seq = 0;
7bc570c8 1281
088aa3ee
YM		 1282 ops = fib_notifier_ops_register(&ip6mr_notifier_ops_template, net);
1283 if (IS_ERR(ops))
1284 return PTR_ERR(ops);
1285
1286 net->ipv6.ip6mr_notifier_ops = ops;
1287
1288 return 0;
1289}
1290
1291static void __net_exit ip6mr_notifier_exit(struct net *net)
1292{
1293 fib_notifier_ops_unregister(net->ipv6.ip6mr_notifier_ops);
1294 net->ipv6.ip6mr_notifier_ops = NULL;
1295}
1296
1297/* Setup for IP multicast routing */
4e16880c
BT		 1298static int __net_init ip6mr_net_init(struct net *net)
1299{
d1db275d 1300 int err;
f30a7784 1301
088aa3ee
YM		 1302 err = ip6mr_notifier_init(net);
1303 if (err)
1304 return err;
1305
d1db275d
PM		 1306 err = ip6mr_rules_init(net);
1307 if (err < 0)
088aa3ee 1308 goto ip6mr_rules_fail;
8b90fc7e
BT		 1309
1310#ifdef CONFIG_PROC_FS
1311 err = -ENOMEM;
c3506372
CH		 1312 if (!proc_create_net("ip6_mr_vif", 0, net->proc_net, &ip6mr_vif_seq_ops,
1313 sizeof(struct mr_vif_iter)))
8b90fc7e 1314 goto proc_vif_fail;
c3506372
CH		 1315 if (!proc_create_net("ip6_mr_cache", 0, net->proc_net, &ipmr_mfc_seq_ops,
1316 sizeof(struct mr_mfc_iter)))
8b90fc7e
BT		 1317 goto proc_cache_fail;
1318#endif
6bd52143 1319
4a6258a0
BT		 1320 return 0;
1321
8b90fc7e
BT		 1322#ifdef CONFIG_PROC_FS
1323proc_cache_fail:
ece31ffd 1324 remove_proc_entry("ip6_mr_vif", net->proc_net);
8b90fc7e 1325proc_vif_fail:
d1db275d 1326 ip6mr_rules_exit(net);
8b90fc7e 1327#endif
088aa3ee
YM		 1328ip6mr_rules_fail:
1329 ip6mr_notifier_exit(net);
4e16880c
BT		 1330 return err;
1331}
1332
1333static void __net_exit ip6mr_net_exit(struct net *net)
1334{
8b90fc7e 1335#ifdef CONFIG_PROC_FS
ece31ffd
G		 1336 remove_proc_entry("ip6_mr_cache", net->proc_net);
1337 remove_proc_entry("ip6_mr_vif", net->proc_net);
8b90fc7e 1338#endif
d1db275d 1339 ip6mr_rules_exit(net);
088aa3ee 1340 ip6mr_notifier_exit(net);
4e16880c
BT		 1341}
1342
1343static struct pernet_operations ip6mr_net_ops = {
1344 .init = ip6mr_net_init,
1345 .exit = ip6mr_net_exit,
1346};
1347
623d1a1a 1348int __init ip6_mr_init(void)
7bc570c8 1349{
623d1a1a
WC		 1350 int err;
1351
7bc570c8
YH		 1352 mrt_cachep = kmem_cache_create("ip6_mrt_cache",
1353 sizeof(struct mfc6_cache),
1354 0, SLAB_HWCACHE_ALIGN,
1355 NULL);
1356 if (!mrt_cachep)
623d1a1a 1357 return -ENOMEM;
7bc570c8 1358
4e16880c
BT		 1359 err = register_pernet_subsys(&ip6mr_net_ops);
1360 if (err)
1361 goto reg_pernet_fail;
1362
623d1a1a
WC		 1363 err = register_netdevice_notifier(&ip6_mr_notifier);
1364 if (err)
1365 goto reg_notif_fail;
403dbb97
TG		 1366#ifdef CONFIG_IPV6_PIMSM_V2
1367 if (inet6_add_protocol(&pim6_protocol, IPPROTO_PIM) < 0) {
f3213831 1368 pr_err("%s: can't add PIM protocol\n", __func__);
403dbb97
TG		 1369 err = -EAGAIN;
1370 goto add_proto_fail;
1371 }
1372#endif
a3fde2ad
FW		 1373 err = rtnl_register_module(THIS_MODULE, RTNL_FAMILY_IP6MR, RTM_GETROUTE,
1374 NULL, ip6mr_rtm_dumproute, 0);
1375 if (err == 0)
1376 return 0;
1377
403dbb97 1378#ifdef CONFIG_IPV6_PIMSM_V2
a3fde2ad 1379 inet6_del_protocol(&pim6_protocol, IPPROTO_PIM);
403dbb97
TG		 1380add_proto_fail:
1381 unregister_netdevice_notifier(&ip6_mr_notifier);
1382#endif
87b30a65 1383reg_notif_fail:
4e16880c
BT		 1384 unregister_pernet_subsys(&ip6mr_net_ops);
1385reg_pernet_fail:
87b30a65 1386 kmem_cache_destroy(mrt_cachep);
623d1a1a 1387 return err;
7bc570c8
YH		 1388}
1389
623d1a1a
WC		 1390void ip6_mr_cleanup(void)
1391{
ffb1388a
DJ		 1392 rtnl_unregister(RTNL_FAMILY_IP6MR, RTM_GETROUTE);
1393#ifdef CONFIG_IPV6_PIMSM_V2
1394 inet6_del_protocol(&pim6_protocol, IPPROTO_PIM);
1395#endif
623d1a1a 1396 unregister_netdevice_notifier(&ip6_mr_notifier);
4e16880c 1397 unregister_pernet_subsys(&ip6mr_net_ops);
623d1a1a
WC		 1398 kmem_cache_destroy(mrt_cachep);
1399}
7bc570c8 1400
b70432f7 1401static int ip6mr_mfc_add(struct net *net, struct mr_table *mrt,
660b26dc 1402 struct mf6cctl *mfc, int mrtsock, int parent)
7bc570c8 1403{
6ac7eb08 1404 unsigned char ttls[MAXMIFS];
87c418bf 1405 struct mfc6_cache *uc, *c;
494fff56 1406 struct mr_mfc *_uc;
87c418bf
YM		 1407 bool found;
1408 int i, err;
7bc570c8 1409
a50436f2
PM		 1410 if (mfc->mf6cc_parent >= MAXMIFS)
1411 return -ENFILE;
1412
6ac7eb08
RR		 1413 memset(ttls, 255, MAXMIFS);
1414 for (i = 0; i < MAXMIFS; i++) {
7bc570c8
YH		 1415 if (IF_ISSET(i, &mfc->mf6cc_ifset))
1416 ttls[i] = 1;
7bc570c8
YH		 1417 }
1418
87c418bf
YM		 1419 /* The entries are added/deleted only under RTNL */
1420 rcu_read_lock();
1421 c = ip6mr_cache_find_parent(mrt, &mfc->mf6cc_origin.sin6_addr,
1422 &mfc->mf6cc_mcastgrp.sin6_addr, parent);
1423 rcu_read_unlock();
1424 if (c) {
7bc570c8 1425 write_lock_bh(&mrt_lock);
494fff56
YM		 1426 c->_c.mfc_parent = mfc->mf6cc_parent;
1427 ip6mr_update_thresholds(mrt, &c->_c, ttls);
7bc570c8 1428 if (!mrtsock)
494fff56 1429 c->_c.mfc_flags |= MFC_STATIC;
7bc570c8 1430 write_unlock_bh(&mrt_lock);
088aa3ee
YM		 1431 call_ip6mr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_REPLACE,
1432 c, mrt->id);
812e44dd 1433 mr6_netlink_event(mrt, c, RTM_NEWROUTE);
7bc570c8
YH		 1434 return 0;
1435 }
1436
660b26dc
ND		 1437 if (!ipv6_addr_any(&mfc->mf6cc_mcastgrp.sin6_addr) &&
1438 !ipv6_addr_is_multicast(&mfc->mf6cc_mcastgrp.sin6_addr))
7bc570c8
YH		 1439 return -EINVAL;
1440
b5aa30b1 1441 c = ip6mr_cache_alloc();
63159f29 1442 if (!c)
7bc570c8
YH		 1443 return -ENOMEM;
1444
1445 c->mf6c_origin = mfc->mf6cc_origin.sin6_addr;
1446 c->mf6c_mcastgrp = mfc->mf6cc_mcastgrp.sin6_addr;
494fff56
YM		 1447 c->_c.mfc_parent = mfc->mf6cc_parent;
1448 ip6mr_update_thresholds(mrt, &c->_c, ttls);
7bc570c8 1449 if (!mrtsock)
494fff56 1450 c->_c.mfc_flags |= MFC_STATIC;
7bc570c8 1451
494fff56 1452 err = rhltable_insert_key(&mrt->mfc_hash, &c->cmparg, &c->_c.mnode,
87c418bf
YM		 1453 ip6mr_rht_params);
1454 if (err) {
1455 pr_err("ip6mr: rhtable insert error %d\n", err);
1456 ip6mr_cache_free(c);
1457 return err;
1458 }
494fff56 1459 list_add_tail_rcu(&c->_c.list, &mrt->mfc_cache_list);
7bc570c8 1460
87c418bf
YM		 1461 /* Check to see if we resolved a queued list. If so we
1462 * need to send on the frames and tidy up.
7bc570c8 1463 */
f30a7784 1464 found = false;
7bc570c8 1465 spin_lock_bh(&mfc_unres_lock);
494fff56
YM		 1466 list_for_each_entry(_uc, &mrt->mfc_unres_queue, list) {
1467 uc = (struct mfc6_cache *)_uc;
c476efbc 1468 if (ipv6_addr_equal(&uc->mf6c_origin, &c->mf6c_origin) &&
7bc570c8 1469 ipv6_addr_equal(&uc->mf6c_mcastgrp, &c->mf6c_mcastgrp)) {
494fff56 1470 list_del(&_uc->list);
6bd52143 1471 atomic_dec(&mrt->cache_resolve_queue_len);
f30a7784 1472 found = true;
7bc570c8
YH		 1473 break;
1474 }
1475 }
b70432f7 1476 if (list_empty(&mrt->mfc_unres_queue))
6bd52143 1477 del_timer(&mrt->ipmr_expire_timer);
7bc570c8
YH		 1478 spin_unlock_bh(&mfc_unres_lock);
1479
f30a7784 1480 if (found) {
6bd52143 1481 ip6mr_cache_resolve(net, mrt, uc, c);
58701ad4 1482 ip6mr_cache_free(uc);
7bc570c8 1483 }
088aa3ee
YM		 1484 call_ip6mr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_ADD,
1485 c, mrt->id);
812e44dd 1486 mr6_netlink_event(mrt, c, RTM_NEWROUTE);
7bc570c8
YH		 1487 return 0;
1488}
1489
1490/*
1491 * Close the multicast socket, and clear the vif tables etc
1492 */
1493
ca8d4794 1494static void mroute_clean_tables(struct mr_table *mrt, int flags)
7bc570c8 1495{
494fff56 1496 struct mr_mfc *c, *tmp;
c871e664 1497 LIST_HEAD(list);
87c418bf 1498 int i;
7bc570c8 1499
87c418bf 1500 /* Shut down all active vif entries */
ca8d4794
CS		 1501 if (flags & (MRT6_FLUSH_MIFS | MRT6_FLUSH_MIFS_STATIC)) {
1502 for (i = 0; i < mrt->maxvif; i++) {
1503 if (((mrt->vif_table[i].flags & VIFF_STATIC) &&
1504 !(flags & MRT6_FLUSH_MIFS_STATIC)) ||
1505 (!(mrt->vif_table[i].flags & VIFF_STATIC) && !(flags & MRT6_FLUSH_MIFS)))
1506 continue;
1507 mif6_delete(mrt, i, 0, &list);
1508 }
1509 unregister_netdevice_many(&list);
7bc570c8
YH		 1510 }
1511
87c418bf 1512 /* Wipe the cache */
ca8d4794
CS		 1513 if (flags & (MRT6_FLUSH_MFC | MRT6_FLUSH_MFC_STATIC)) {
1514 list_for_each_entry_safe(c, tmp, &mrt->mfc_cache_list, list) {
1515 if (((c->mfc_flags & MFC_STATIC) && !(flags & MRT6_FLUSH_MFC_STATIC)) ||
1516 (!(c->mfc_flags & MFC_STATIC) && !(flags & MRT6_FLUSH_MFC)))
1517 continue;
1518 rhltable_remove(&mrt->mfc_hash, &c->mnode, ip6mr_rht_params);
1519 list_del_rcu(&c->list);
1520 call_ip6mr_mfc_entry_notifiers(read_pnet(&mrt->net),
1521 FIB_EVENT_ENTRY_DEL,
1522 (struct mfc6_cache *)c, mrt->id);
1523 mr6_netlink_event(mrt, (struct mfc6_cache *)c, RTM_DELROUTE);
1524 mr_cache_put(c);
1525 }
7bc570c8
YH		 1526 }
1527
ca8d4794
CS		 1528 if (flags & MRT6_FLUSH_MFC) {
1529 if (atomic_read(&mrt->cache_resolve_queue_len) != 0) {
1530 spin_lock_bh(&mfc_unres_lock);
1531 list_for_each_entry_safe(c, tmp, &mrt->mfc_unres_queue, list) {
1532 list_del(&c->list);
1533 mr6_netlink_event(mrt, (struct mfc6_cache *)c,
1534 RTM_DELROUTE);
1535 ip6mr_destroy_unres(mrt, (struct mfc6_cache *)c);
1536 }
1537 spin_unlock_bh(&mfc_unres_lock);
7bc570c8 1538 }
7bc570c8
YH		 1539 }
1540}
1541
b70432f7 1542static int ip6mr_sk_init(struct mr_table *mrt, struct sock *sk)
7bc570c8
YH		 1543{
1544 int err = 0;
8229efda 1545 struct net *net = sock_net(sk);
7bc570c8
YH		 1546
1547 rtnl_lock();
1548 write_lock_bh(&mrt_lock);
b70432f7 1549 if (rtnl_dereference(mrt->mroute_sk)) {
7bc570c8 1550 err = -EADDRINUSE;
8571ab47 1551 } else {
b70432f7 1552 rcu_assign_pointer(mrt->mroute_sk, sk);
a366e300 1553 sock_set_flag(sk, SOCK_RCU_FREE);
8571ab47 1554 net->ipv6.devconf_all->mc_forwarding++;
927265bc 1555 }
7bc570c8
YH		 1556 write_unlock_bh(&mrt_lock);
1557
927265bc 1558 if (!err)
85b3daad
DA		 1559 inet6_netconf_notify_devconf(net, RTM_NEWNETCONF,
1560 NETCONFA_MC_FORWARDING,
927265bc
ED		 1561 NETCONFA_IFINDEX_ALL,
1562 net->ipv6.devconf_all);
7bc570c8
YH		 1563 rtnl_unlock();
1564
1565 return err;
1566}
1567
1568int ip6mr_sk_done(struct sock *sk)
1569{
d1db275d 1570 int err = -EACCES;
8229efda 1571 struct net *net = sock_net(sk);
b70432f7 1572 struct mr_table *mrt;
7bc570c8 1573
338d182f
FR		 1574 if (sk->sk_type != SOCK_RAW ||
1575 inet_sk(sk)->inet_num != IPPROTO_ICMPV6)
1576 return err;
1577
7bc570c8 1578 rtnl_lock();
d1db275d 1579 ip6mr_for_each_table(mrt, net) {
b70432f7 1580 if (sk == rtnl_dereference(mrt->mroute_sk)) {
d1db275d 1581 write_lock_bh(&mrt_lock);
b70432f7 1582 RCU_INIT_POINTER(mrt->mroute_sk, NULL);
a366e300
ED		 1583 /* Note that mroute_sk had SOCK_RCU_FREE set,
1584 * so the RCU grace period before sk freeing
1585 * is guaranteed by sk_destruct()
1586 */
d1db275d 1587 net->ipv6.devconf_all->mc_forwarding--;
927265bc 1588 write_unlock_bh(&mrt_lock);
85b3daad 1589 inet6_netconf_notify_devconf(net, RTM_NEWNETCONF,
d67b8c61
ND		 1590 NETCONFA_MC_FORWARDING,
1591 NETCONFA_IFINDEX_ALL,
1592 net->ipv6.devconf_all);
7bc570c8 1593
ca8d4794 1594 mroute_clean_tables(mrt, MRT6_FLUSH_MIFS | MRT6_FLUSH_MFC);
d1db275d
PM		 1595 err = 0;
1596 break;
1597 }
1598 }
7bc570c8
YH		 1599 rtnl_unlock();
1600
1601 return err;
1602}
1603
8571ab47 1604bool mroute6_is_socket(struct net *net, struct sk_buff *skb)
6bd52143 1605{
b70432f7 1606 struct mr_table *mrt;
4c9483b2 1607 struct flowi6 fl6 = {
e374c618 1608 .flowi6_iif = skb->skb_iif ? : LOOPBACK_IFINDEX,
4c9483b2
DM		 1609 .flowi6_oif = skb->dev->ifindex,
1610 .flowi6_mark = skb->mark,
d1db275d
PM		 1611 };
1612
4c9483b2 1613 if (ip6mr_fib_lookup(net, &fl6, &mrt) < 0)
d1db275d 1614 return NULL;
6bd52143 1615
b70432f7 1616 return rcu_access_pointer(mrt->mroute_sk);
6bd52143 1617}
8571ab47 1618EXPORT_SYMBOL(mroute6_is_socket);
6bd52143 1619
7bc570c8
YH		 1620/*
1621 * Socket options and virtual interface manipulation. The whole
1622 * virtual interface system is a complete heap, but unfortunately
1623 * that's how BSD mrouted happens to think. Maybe one day with a proper
1624 * MOSPF/PIM router set up we can clean this up.
1625 */
1626
b7058842 1627int ip6_mroute_setsockopt(struct sock *sk, int optname, char __user *optval, unsigned int optlen)
7bc570c8 1628{
660b26dc 1629 int ret, parent = 0;
7bc570c8
YH		 1630 struct mif6ctl vif;
1631 struct mf6cctl mfc;
1632 mifi_t mifi;
8229efda 1633 struct net *net = sock_net(sk);
b70432f7 1634 struct mr_table *mrt;
d1db275d 1635
99253eb7
XL		 1636 if (sk->sk_type != SOCK_RAW ||
1637 inet_sk(sk)->inet_num != IPPROTO_ICMPV6)
1638 return -EOPNOTSUPP;
1639
d1db275d 1640 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
63159f29 1641 if (!mrt)
d1db275d 1642 return -ENOENT;
7bc570c8
YH		 1643
1644 if (optname != MRT6_INIT) {
b70432f7 1645 if (sk != rcu_access_pointer(mrt->mroute_sk) &&
8571ab47 1646 !ns_capable(net->user_ns, CAP_NET_ADMIN))
7bc570c8
YH		 1647 return -EACCES;
1648 }
1649
1650 switch (optname) {
1651 case MRT6_INIT:
7bc570c8
YH		 1652 if (optlen < sizeof(int))
1653 return -EINVAL;
1654
6bd52143 1655 return ip6mr_sk_init(mrt, sk);
7bc570c8
YH		 1656
1657 case MRT6_DONE:
1658 return ip6mr_sk_done(sk);
1659
1660 case MRT6_ADD_MIF:
1661 if (optlen < sizeof(vif))
1662 return -EINVAL;
1663 if (copy_from_user(&vif, optval, sizeof(vif)))
1664 return -EFAULT;
6ac7eb08 1665 if (vif.mif6c_mifi >= MAXMIFS)
7bc570c8
YH		 1666 return -ENFILE;
1667 rtnl_lock();
8571ab47 1668 ret = mif6_add(net, mrt, &vif,
b70432f7 1669 sk == rtnl_dereference(mrt->mroute_sk));
7bc570c8
YH		 1670 rtnl_unlock();
1671 return ret;
1672
1673 case MRT6_DEL_MIF:
1674 if (optlen < sizeof(mifi_t))
1675 return -EINVAL;
1676 if (copy_from_user(&mifi, optval, sizeof(mifi_t)))
1677 return -EFAULT;
1678 rtnl_lock();
723b929c 1679 ret = mif6_delete(mrt, mifi, 0, NULL);
7bc570c8
YH		 1680 rtnl_unlock();
1681 return ret;
1682
1683 /*
1684 * Manipulate the forwarding caches. These live
1685 * in a sort of kernel/user symbiosis.
1686 */
1687 case MRT6_ADD_MFC:
1688 case MRT6_DEL_MFC:
660b26dc 1689 parent = -1;
275757e6 1690 /* fall through */
660b26dc
ND		 1691 case MRT6_ADD_MFC_PROXY:
1692 case MRT6_DEL_MFC_PROXY:
7bc570c8
YH		 1693 if (optlen < sizeof(mfc))
1694 return -EINVAL;
1695 if (copy_from_user(&mfc, optval, sizeof(mfc)))
1696 return -EFAULT;
660b26dc
ND		 1697 if (parent == 0)
1698 parent = mfc.mf6cc_parent;
7bc570c8 1699 rtnl_lock();
660b26dc
ND		 1700 if (optname == MRT6_DEL_MFC || optname == MRT6_DEL_MFC_PROXY)
1701 ret = ip6mr_mfc_delete(mrt, &mfc, parent);
7bc570c8 1702 else
660b26dc 1703 ret = ip6mr_mfc_add(net, mrt, &mfc,
8571ab47 1704 sk ==
b70432f7 1705 rtnl_dereference(mrt->mroute_sk),
8571ab47 1706 parent);
7bc570c8
YH		 1707 rtnl_unlock();
1708 return ret;
1709
ca8d4794
CS		 1710 case MRT6_FLUSH:
1711 {
1712 int flags;
1713
1714 if (optlen != sizeof(flags))
1715 return -EINVAL;
1716 if (get_user(flags, (int __user *)optval))
1717 return -EFAULT;
1718 rtnl_lock();
1719 mroute_clean_tables(mrt, flags);
1720 rtnl_unlock();
1721 return 0;
1722 }
1723
14fb64e1
YH		 1724 /*
1725 * Control PIM assert (to activate pim will activate assert)
1726 */
1727 case MRT6_ASSERT:
1728 {
1729 int v;
03f52a0a
JP		 1730
1731 if (optlen != sizeof(v))
1732 return -EINVAL;
14fb64e1
YH		 1733 if (get_user(v, (int __user *)optval))
1734 return -EFAULT;
53d6841d 1735 mrt->mroute_do_assert = v;
14fb64e1
YH		 1736 return 0;
1737 }
1738
1739#ifdef CONFIG_IPV6_PIMSM_V2
1740 case MRT6_PIM:
1741 {
a9f83bf3 1742 int v;
03f52a0a
JP		 1743
1744 if (optlen != sizeof(v))
1745 return -EINVAL;
14fb64e1
YH		 1746 if (get_user(v, (int __user *)optval))
1747 return -EFAULT;
1748 v = !!v;
1749 rtnl_lock();
1750 ret = 0;
6bd52143
PM		 1751 if (v != mrt->mroute_do_pim) {
1752 mrt->mroute_do_pim = v;
1753 mrt->mroute_do_assert = v;
14fb64e1
YH		 1754 }
1755 rtnl_unlock();
1756 return ret;
1757 }
1758
d1db275d
PM		 1759#endif
1760#ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES
1761 case MRT6_TABLE:
1762 {
1763 u32 v;
1764
1765 if (optlen != sizeof(u32))
1766 return -EINVAL;
1767 if (get_user(v, (u32 __user *)optval))
1768 return -EFAULT;
75356a81
DC		 1769 /* "pim6reg%u" should not exceed 16 bytes (IFNAMSIZ) */
1770 if (v != RT_TABLE_DEFAULT && v >= 100000000)
1771 return -EINVAL;
b70432f7 1772 if (sk == rcu_access_pointer(mrt->mroute_sk))
d1db275d
PM		 1773 return -EBUSY;
1774
1775 rtnl_lock();
1776 ret = 0;
e783bb00
SD		 1777 mrt = ip6mr_new_table(net, v);
1778 if (IS_ERR(mrt))
1779 ret = PTR_ERR(mrt);
848235ed
SD		 1780 else
1781 raw6_sk(sk)->ip6mr_table = v;
d1db275d
PM		 1782 rtnl_unlock();
1783 return ret;
1784 }
14fb64e1 1785#endif
7bc570c8 1786 /*
7d120c55 1787 * Spurious command, or MRT6_VERSION which you cannot
7bc570c8
YH		 1788 * set.
1789 */
1790 default:
1791 return -ENOPROTOOPT;
1792 }
1793}
1794
1795/*
1796 * Getsock opt support for the multicast routing system.
1797 */
1798
1799int ip6_mroute_getsockopt(struct sock *sk, int optname, char __user *optval,
1800 int __user *optlen)
1801{
1802 int olr;
1803 int val;
8229efda 1804 struct net *net = sock_net(sk);
b70432f7 1805 struct mr_table *mrt;
d1db275d 1806
99253eb7
XL		 1807 if (sk->sk_type != SOCK_RAW ||
1808 inet_sk(sk)->inet_num != IPPROTO_ICMPV6)
1809 return -EOPNOTSUPP;
1810
d1db275d 1811 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
63159f29 1812 if (!mrt)
d1db275d 1813 return -ENOENT;
7bc570c8
YH		 1814
1815 switch (optname) {
1816 case MRT6_VERSION:
1817 val = 0x0305;
1818 break;
14fb64e1
YH		 1819#ifdef CONFIG_IPV6_PIMSM_V2
1820 case MRT6_PIM:
6bd52143 1821 val = mrt->mroute_do_pim;
14fb64e1
YH		 1822 break;
1823#endif
1824 case MRT6_ASSERT:
6bd52143 1825 val = mrt->mroute_do_assert;
14fb64e1 1826 break;
7bc570c8
YH		 1827 default:
1828 return -ENOPROTOOPT;
1829 }
1830
1831 if (get_user(olr, optlen))
1832 return -EFAULT;
1833
1834 olr = min_t(int, olr, sizeof(int));
1835 if (olr < 0)
1836 return -EINVAL;
1837
1838 if (put_user(olr, optlen))
1839 return -EFAULT;
1840 if (copy_to_user(optval, &val, olr))
1841 return -EFAULT;
1842 return 0;
1843}
1844
1845/*
1846 * The IP multicast ioctl support routines.
1847 */
1848
1849int ip6mr_ioctl(struct sock *sk, int cmd, void __user *arg)
1850{
1851 struct sioc_sg_req6 sr;
1852 struct sioc_mif_req6 vr;
6853f21f 1853 struct vif_device *vif;
7bc570c8 1854 struct mfc6_cache *c;
8229efda 1855 struct net *net = sock_net(sk);
b70432f7 1856 struct mr_table *mrt;
d1db275d
PM		 1857
1858 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
63159f29 1859 if (!mrt)
d1db275d 1860 return -ENOENT;
7bc570c8
YH		 1861
1862 switch (cmd) {
1863 case SIOCGETMIFCNT_IN6:
1864 if (copy_from_user(&vr, arg, sizeof(vr)))
1865 return -EFAULT;
6bd52143 1866 if (vr.mifi >= mrt->maxvif)
7bc570c8 1867 return -EINVAL;
69d2c867 1868 vr.mifi = array_index_nospec(vr.mifi, mrt->maxvif);
7bc570c8 1869 read_lock(&mrt_lock);
b70432f7
YM		 1870 vif = &mrt->vif_table[vr.mifi];
1871 if (VIF_EXISTS(mrt, vr.mifi)) {
7bc570c8
YH		 1872 vr.icount = vif->pkt_in;
1873 vr.ocount = vif->pkt_out;
1874 vr.ibytes = vif->bytes_in;
1875 vr.obytes = vif->bytes_out;
1876 read_unlock(&mrt_lock);
1877
1878 if (copy_to_user(arg, &vr, sizeof(vr)))
1879 return -EFAULT;
1880 return 0;
1881 }
1882 read_unlock(&mrt_lock);
1883 return -EADDRNOTAVAIL;
1884 case SIOCGETSGCNT_IN6:
1885 if (copy_from_user(&sr, arg, sizeof(sr)))
1886 return -EFAULT;
1887
87c418bf 1888 rcu_read_lock();
6bd52143 1889 c = ip6mr_cache_find(mrt, &sr.src.sin6_addr, &sr.grp.sin6_addr);
7bc570c8 1890 if (c) {
494fff56
YM		 1891 sr.pktcnt = c->_c.mfc_un.res.pkt;
1892 sr.bytecnt = c->_c.mfc_un.res.bytes;
1893 sr.wrong_if = c->_c.mfc_un.res.wrong_if;
87c418bf 1894 rcu_read_unlock();
7bc570c8
YH		 1895
1896 if (copy_to_user(arg, &sr, sizeof(sr)))
1897 return -EFAULT;
1898 return 0;
1899 }
87c418bf 1900 rcu_read_unlock();
7bc570c8
YH		 1901 return -EADDRNOTAVAIL;
1902 default:
1903 return -ENOIOCTLCMD;
1904 }
1905}
1906
e2d57766
DM		 1907#ifdef CONFIG_COMPAT
1908struct compat_sioc_sg_req6 {
1909 struct sockaddr_in6 src;
1910 struct sockaddr_in6 grp;
1911 compat_ulong_t pktcnt;
1912 compat_ulong_t bytecnt;
1913 compat_ulong_t wrong_if;
1914};
1915
1916struct compat_sioc_mif_req6 {
1917 mifi_t mifi;
1918 compat_ulong_t icount;
1919 compat_ulong_t ocount;
1920 compat_ulong_t ibytes;
1921 compat_ulong_t obytes;
1922};
1923
1924int ip6mr_compat_ioctl(struct sock *sk, unsigned int cmd, void __user *arg)
1925{
1926 struct compat_sioc_sg_req6 sr;
1927 struct compat_sioc_mif_req6 vr;
6853f21f 1928 struct vif_device *vif;
e2d57766
DM		 1929 struct mfc6_cache *c;
1930 struct net *net = sock_net(sk);
b70432f7 1931 struct mr_table *mrt;
e2d57766
DM		 1932
1933 mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT);
63159f29 1934 if (!mrt)
e2d57766
DM		 1935 return -ENOENT;
1936
1937 switch (cmd) {
1938 case SIOCGETMIFCNT_IN6:
1939 if (copy_from_user(&vr, arg, sizeof(vr)))
1940 return -EFAULT;
1941 if (vr.mifi >= mrt->maxvif)
1942 return -EINVAL;
69d2c867 1943 vr.mifi = array_index_nospec(vr.mifi, mrt->maxvif);
e2d57766 1944 read_lock(&mrt_lock);
b70432f7
YM		 1945 vif = &mrt->vif_table[vr.mifi];
1946 if (VIF_EXISTS(mrt, vr.mifi)) {
e2d57766
DM		 1947 vr.icount = vif->pkt_in;
1948 vr.ocount = vif->pkt_out;
1949 vr.ibytes = vif->bytes_in;
1950 vr.obytes = vif->bytes_out;
1951 read_unlock(&mrt_lock);
1952
1953 if (copy_to_user(arg, &vr, sizeof(vr)))
1954 return -EFAULT;
1955 return 0;
1956 }
1957 read_unlock(&mrt_lock);
1958 return -EADDRNOTAVAIL;
1959 case SIOCGETSGCNT_IN6:
1960 if (copy_from_user(&sr, arg, sizeof(sr)))
1961 return -EFAULT;
1962
87c418bf 1963 rcu_read_lock();
e2d57766
DM		 1964 c = ip6mr_cache_find(mrt, &sr.src.sin6_addr, &sr.grp.sin6_addr);
1965 if (c) {
494fff56
YM		 1966 sr.pktcnt = c->_c.mfc_un.res.pkt;
1967 sr.bytecnt = c->_c.mfc_un.res.bytes;
1968 sr.wrong_if = c->_c.mfc_un.res.wrong_if;
87c418bf 1969 rcu_read_unlock();
e2d57766
DM		 1970
1971 if (copy_to_user(arg, &sr, sizeof(sr)))
1972 return -EFAULT;
1973 return 0;
1974 }
87c418bf 1975 rcu_read_unlock();
e2d57766
DM		 1976 return -EADDRNOTAVAIL;
1977 default:
1978 return -ENOIOCTLCMD;
1979 }
1980}
1981#endif
7bc570c8 1982
0c4b51f0 1983static inline int ip6mr_forward2_finish(struct net *net, struct sock *sk, struct sk_buff *skb)
7bc570c8 1984{
87c11f1d
IS		 1985 IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)),
1986 IPSTATS_MIB_OUTFORWDATAGRAMS);
1987 IP6_ADD_STATS(net, ip6_dst_idev(skb_dst(skb)),
1988 IPSTATS_MIB_OUTOCTETS, skb->len);
13206b6b 1989 return dst_output(net, sk, skb);
7bc570c8
YH		 1990}
1991
1992/*
1993 * Processing handlers for ip6mr_forward
1994 */
1995
b70432f7 1996static int ip6mr_forward2(struct net *net, struct mr_table *mrt,
f5c6dfde 1997 struct sk_buff *skb, int vifi)
7bc570c8
YH		 1998{
1999 struct ipv6hdr *ipv6h;
b70432f7 2000 struct vif_device *vif = &mrt->vif_table[vifi];
7bc570c8
YH		 2001 struct net_device *dev;
2002 struct dst_entry *dst;
4c9483b2 2003 struct flowi6 fl6;
7bc570c8 2004
63159f29 2005 if (!vif->dev)
7bc570c8
YH		 2006 goto out_free;
2007
14fb64e1
YH		 2008#ifdef CONFIG_IPV6_PIMSM_V2
2009 if (vif->flags & MIFF_REGISTER) {
2010 vif->pkt_out++;
2011 vif->bytes_out += skb->len;
dc58c78c
PE		 2012 vif->dev->stats.tx_bytes += skb->len;
2013 vif->dev->stats.tx_packets++;
6bd52143 2014 ip6mr_cache_report(mrt, skb, vifi, MRT6MSG_WHOLEPKT);
8da73b73 2015 goto out_free;
14fb64e1
YH		 2016 }
2017#endif
2018
7bc570c8
YH		 2019 ipv6h = ipv6_hdr(skb);
2020
4c9483b2
DM		 2021 fl6 = (struct flowi6) {
2022 .flowi6_oif = vif->link,
2023 .daddr = ipv6h->daddr,
7bc570c8
YH		 2024 };
2025
4c9483b2 2026 dst = ip6_route_output(net, NULL, &fl6);
5095d64d
RL		 2027 if (dst->error) {
2028 dst_release(dst);
7bc570c8 2029 goto out_free;
5095d64d 2030 }
7bc570c8 2031
adf30907
ED		 2032 skb_dst_drop(skb);
2033 skb_dst_set(skb, dst);
7bc570c8
YH		 2034
2035 /*
2036 * RFC1584 teaches, that DVMRP/PIM router must deliver packets locally
2037 * not only before forwarding, but after forwarding on all output
2038 * interfaces. It is clear, if mrouter runs a multicasting
2039 * program, it should receive packets not depending to what interface
2040 * program is joined.
2041 * If we will not make it, the program will have to join on all
2042 * interfaces. On the other hand, multihoming host (or router, but
2043 * not mrouter) cannot join to more than one interface - it will
2044 * result in receiving multiple packets.
2045 */
2046 dev = vif->dev;
2047 skb->dev = dev;
2048 vif->pkt_out++;
2049 vif->bytes_out += skb->len;
2050
2051 /* We are about to write */
2052 /* XXX: extension headers? */
2053 if (skb_cow(skb, sizeof(*ipv6h) + LL_RESERVED_SPACE(dev)))
2054 goto out_free;
2055
2056 ipv6h = ipv6_hdr(skb);
2057 ipv6h->hop_limit--;
2058
2059 IP6CB(skb)->flags |= IP6SKB_FORWARDED;
2060
29a26a56
EB		 2061 return NF_HOOK(NFPROTO_IPV6, NF_INET_FORWARD,
2062 net, NULL, skb, skb->dev, dev,
7bc570c8
YH		 2063 ip6mr_forward2_finish);
2064
2065out_free:
2066 kfree_skb(skb);
2067 return 0;
2068}
2069
b70432f7 2070static int ip6mr_find_vif(struct mr_table *mrt, struct net_device *dev)
7bc570c8
YH		 2071{
2072 int ct;
6bd52143
PM		 2073
2074 for (ct = mrt->maxvif - 1; ct >= 0; ct--) {
b70432f7 2075 if (mrt->vif_table[ct].dev == dev)
7bc570c8
YH		 2076 break;
2077 }
2078 return ct;
2079}
2080
b70432f7 2081static void ip6_mr_forward(struct net *net, struct mr_table *mrt,
e4a38c0c
PR		 2082 struct net_device *dev, struct sk_buff *skb,
2083 struct mfc6_cache *c)
7bc570c8
YH		 2084{
2085 int psend = -1;
2086 int vif, ct;
e4a38c0c 2087 int true_vifi = ip6mr_find_vif(mrt, dev);
7bc570c8 2088
494fff56
YM		 2089 vif = c->_c.mfc_parent;
2090 c->_c.mfc_un.res.pkt++;
2091 c->_c.mfc_un.res.bytes += skb->len;
2092 c->_c.mfc_un.res.lastuse = jiffies;
7bc570c8 2093
494fff56 2094 if (ipv6_addr_any(&c->mf6c_origin) && true_vifi >= 0) {
660b26dc
ND		 2095 struct mfc6_cache *cache_proxy;
2096
40dc2ca3 2097 /* For an (*,G) entry, we only check that the incoming
660b26dc
ND		 2098 * interface is part of the static tree.
2099 */
87c418bf 2100 rcu_read_lock();
845c9a7a 2101 cache_proxy = mr_mfc_find_any_parent(mrt, vif);
660b26dc 2102 if (cache_proxy &&
494fff56 2103 cache_proxy->_c.mfc_un.res.ttls[true_vifi] < 255) {
87c418bf 2104 rcu_read_unlock();
660b26dc 2105 goto forward;
87c418bf
YM		 2106 }
2107 rcu_read_unlock();
660b26dc
ND		 2108 }
2109
14fb64e1
YH		 2110 /*
2111 * Wrong interface: drop packet and (maybe) send PIM assert.
2112 */
e4a38c0c 2113 if (mrt->vif_table[vif].dev != dev) {
494fff56 2114 c->_c.mfc_un.res.wrong_if++;
14fb64e1 2115
6bd52143 2116 if (true_vifi >= 0 && mrt->mroute_do_assert &&
14fb64e1
YH		 2117 /* pimsm uses asserts, when switching from RPT to SPT,
2118 so that we cannot check that packet arrived on an oif.
2119 It is bad, but otherwise we would need to move pretty
2120 large chunk of pimd to kernel. Ough... --ANK
2121 */
6bd52143 2122 (mrt->mroute_do_pim ||
494fff56 2123 c->_c.mfc_un.res.ttls[true_vifi] < 255) &&
14fb64e1 2124 time_after(jiffies,
494fff56
YM		 2125 c->_c.mfc_un.res.last_assert +
2126 MFC_ASSERT_THRESH)) {
2127 c->_c.mfc_un.res.last_assert = jiffies;
6bd52143 2128 ip6mr_cache_report(mrt, skb, true_vifi, MRT6MSG_WRONGMIF);
14fb64e1
YH		 2129 }
2130 goto dont_forward;
2131 }
2132
660b26dc 2133forward:
b70432f7
YM		 2134 mrt->vif_table[vif].pkt_in++;
2135 mrt->vif_table[vif].bytes_in += skb->len;
7bc570c8
YH		 2136
2137 /*
2138 * Forward the frame
2139 */
494fff56
YM		 2140 if (ipv6_addr_any(&c->mf6c_origin) &&
2141 ipv6_addr_any(&c->mf6c_mcastgrp)) {
660b26dc 2142 if (true_vifi >= 0 &&
494fff56 2143 true_vifi != c->_c.mfc_parent &&
660b26dc 2144 ipv6_hdr(skb)->hop_limit >
494fff56 2145 c->_c.mfc_un.res.ttls[c->_c.mfc_parent]) {
660b26dc
ND		 2146 /* It's an (*,*) entry and the packet is not coming from
2147 * the upstream: forward the packet to the upstream
2148 * only.
2149 */
494fff56 2150 psend = c->_c.mfc_parent;
660b26dc
ND		 2151 goto last_forward;
2152 }
2153 goto dont_forward;
2154 }
494fff56
YM		 2155 for (ct = c->_c.mfc_un.res.maxvif - 1;
2156 ct >= c->_c.mfc_un.res.minvif; ct--) {
660b26dc 2157 /* For (*,G) entry, don't forward to the incoming interface */
494fff56
YM		 2158 if ((!ipv6_addr_any(&c->mf6c_origin) || ct != true_vifi) &&
2159 ipv6_hdr(skb)->hop_limit > c->_c.mfc_un.res.ttls[ct]) {
7bc570c8
YH		 2160 if (psend != -1) {
2161 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
2162 if (skb2)
f5c6dfde 2163 ip6mr_forward2(net, mrt, skb2, psend);
7bc570c8
YH		 2164 }
2165 psend = ct;
2166 }
2167 }
660b26dc 2168last_forward:
7bc570c8 2169 if (psend != -1) {
f5c6dfde 2170 ip6mr_forward2(net, mrt, skb, psend);
2b52c3ad 2171 return;
7bc570c8
YH		 2172 }
2173
14fb64e1 2174dont_forward:
7bc570c8 2175 kfree_skb(skb);
7bc570c8
YH		 2176}
2177
2178
2179/*
2180 * Multicast packets for forwarding arrive here
2181 */
2182
2183int ip6_mr_input(struct sk_buff *skb)
2184{
2185 struct mfc6_cache *cache;
8229efda 2186 struct net *net = dev_net(skb->dev);
b70432f7 2187 struct mr_table *mrt;
4c9483b2
DM		 2188 struct flowi6 fl6 = {
2189 .flowi6_iif = skb->dev->ifindex,
2190 .flowi6_mark = skb->mark,
d1db275d
PM		 2191 };
2192 int err;
e4a38c0c
PR		 2193 struct net_device *dev;
2194
2195 /* skb->dev passed in is the master dev for vrfs.
2196 * Get the proper interface that does have a vif associated with it.
2197 */
2198 dev = skb->dev;
2199 if (netif_is_l3_master(skb->dev)) {
2200 dev = dev_get_by_index_rcu(net, IPCB(skb)->iif);
2201 if (!dev) {
2202 kfree_skb(skb);
2203 return -ENODEV;
2204 }
2205 }
d1db275d 2206
4c9483b2 2207 err = ip6mr_fib_lookup(net, &fl6, &mrt);
2015de5f
BG		 2208 if (err < 0) {
2209 kfree_skb(skb);
d1db275d 2210 return err;
2015de5f 2211 }
7bc570c8
YH		 2212
2213 read_lock(&mrt_lock);
6bd52143 2214 cache = ip6mr_cache_find(mrt,
8229efda 2215 &ipv6_hdr(skb)->saddr, &ipv6_hdr(skb)->daddr);
63159f29 2216 if (!cache) {
e4a38c0c 2217 int vif = ip6mr_find_vif(mrt, dev);
660b26dc
ND		 2218
2219 if (vif >= 0)
2220 cache = ip6mr_cache_find_any(mrt,
2221 &ipv6_hdr(skb)->daddr,
2222 vif);
2223 }
7bc570c8
YH		 2224
2225 /*
2226 * No usable cache entry
2227 */
63159f29 2228 if (!cache) {
7bc570c8
YH		 2229 int vif;
2230
e4a38c0c 2231 vif = ip6mr_find_vif(mrt, dev);
7bc570c8 2232 if (vif >= 0) {
e4a38c0c 2233 int err = ip6mr_cache_unresolved(mrt, vif, skb, dev);
7bc570c8
YH		 2234 read_unlock(&mrt_lock);
2235
2236 return err;
2237 }
2238 read_unlock(&mrt_lock);
2239 kfree_skb(skb);
2240 return -ENODEV;
2241 }
2242
e4a38c0c 2243 ip6_mr_forward(net, mrt, dev, skb, cache);
7bc570c8
YH		 2244
2245 read_unlock(&mrt_lock);
2246
2247 return 0;
2248}
2249
2cf75070 2250int ip6mr_get_route(struct net *net, struct sk_buff *skb, struct rtmsg *rtm,
fd61c6ba 2251 u32 portid)
7bc570c8
YH		 2252{
2253 int err;
b70432f7 2254 struct mr_table *mrt;
7bc570c8 2255 struct mfc6_cache *cache;
adf30907 2256 struct rt6_info *rt = (struct rt6_info *)skb_dst(skb);
7bc570c8 2257
d1db275d 2258 mrt = ip6mr_get_table(net, RT6_TABLE_DFLT);
63159f29 2259 if (!mrt)
d1db275d
PM		 2260 return -ENOENT;
2261
7bc570c8 2262 read_lock(&mrt_lock);
6bd52143 2263 cache = ip6mr_cache_find(mrt, &rt->rt6i_src.addr, &rt->rt6i_dst.addr);
660b26dc
ND		 2264 if (!cache && skb->dev) {
2265 int vif = ip6mr_find_vif(mrt, skb->dev);
2266
2267 if (vif >= 0)
2268 cache = ip6mr_cache_find_any(mrt, &rt->rt6i_dst.addr,
2269 vif);
2270 }
7bc570c8
YH		 2271
2272 if (!cache) {
2273 struct sk_buff *skb2;
2274 struct ipv6hdr *iph;
2275 struct net_device *dev;
2276 int vif;
2277
7bc570c8 2278 dev = skb->dev;
63159f29 2279 if (!dev || (vif = ip6mr_find_vif(mrt, dev)) < 0) {
7bc570c8
YH		 2280 read_unlock(&mrt_lock);
2281 return -ENODEV;
2282 }
2283
2284 /* really correct? */
2285 skb2 = alloc_skb(sizeof(struct ipv6hdr), GFP_ATOMIC);
2286 if (!skb2) {
2287 read_unlock(&mrt_lock);
2288 return -ENOMEM;
2289 }
2290
2cf75070 2291 NETLINK_CB(skb2).portid = portid;
7bc570c8
YH		 2292 skb_reset_transport_header(skb2);
2293
2294 skb_put(skb2, sizeof(struct ipv6hdr));
2295 skb_reset_network_header(skb2);
2296
2297 iph = ipv6_hdr(skb2);
2298 iph->version = 0;
2299 iph->priority = 0;
2300 iph->flow_lbl[0] = 0;
2301 iph->flow_lbl[1] = 0;
2302 iph->flow_lbl[2] = 0;
2303 iph->payload_len = 0;
2304 iph->nexthdr = IPPROTO_NONE;
2305 iph->hop_limit = 0;
4e3fd7a0
AD		 2306 iph->saddr = rt->rt6i_src.addr;
2307 iph->daddr = rt->rt6i_dst.addr;
7bc570c8 2308
e4a38c0c 2309 err = ip6mr_cache_unresolved(mrt, vif, skb2, dev);
7bc570c8
YH		 2310 read_unlock(&mrt_lock);
2311
2312 return err;
2313 }
2314
7b0db857 2315 err = mr_fill_mroute(mrt, skb, &cache->_c, rtm);
7bc570c8
YH		 2316 read_unlock(&mrt_lock);
2317 return err;
2318}
2319
b70432f7 2320static int ip6mr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb,
f518338b
ND		 2321 u32 portid, u32 seq, struct mfc6_cache *c, int cmd,
2322 int flags)
5b285cac
PM		 2323{
2324 struct nlmsghdr *nlh;
2325 struct rtmsg *rtm;
1eb99af5 2326 int err;
5b285cac 2327
f518338b 2328 nlh = nlmsg_put(skb, portid, seq, cmd, sizeof(*rtm), flags);
63159f29 2329 if (!nlh)
5b285cac
PM		 2330 return -EMSGSIZE;
2331
2332 rtm = nlmsg_data(nlh);
193c1e47 2333 rtm->rtm_family = RTNL_FAMILY_IP6MR;
5b285cac
PM		 2334 rtm->rtm_dst_len = 128;
2335 rtm->rtm_src_len = 128;
2336 rtm->rtm_tos = 0;
2337 rtm->rtm_table = mrt->id;
c78679e8
DM		 2338 if (nla_put_u32(skb, RTA_TABLE, mrt->id))
2339 goto nla_put_failure;
1eb99af5 2340 rtm->rtm_type = RTN_MULTICAST;
5b285cac 2341 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
494fff56 2342 if (c->_c.mfc_flags & MFC_STATIC)
9a68ac72
ND		 2343 rtm->rtm_protocol = RTPROT_STATIC;
2344 else
2345 rtm->rtm_protocol = RTPROT_MROUTED;
5b285cac
PM		 2346 rtm->rtm_flags = 0;
2347
930345ea
JB		 2348 if (nla_put_in6_addr(skb, RTA_SRC, &c->mf6c_origin) ||
2349 nla_put_in6_addr(skb, RTA_DST, &c->mf6c_mcastgrp))
c78679e8 2350 goto nla_put_failure;
7b0db857 2351 err = mr_fill_mroute(mrt, skb, &c->_c, rtm);
1eb99af5
ND		 2352 /* do not break the dump if cache is unresolved */
2353 if (err < 0 && err != -ENOENT)
5b285cac
PM		 2354 goto nla_put_failure;
2355
053c095a
JB		 2356 nlmsg_end(skb, nlh);
2357 return 0;
5b285cac
PM		 2358
2359nla_put_failure:
2360 nlmsg_cancel(skb, nlh);
2361 return -EMSGSIZE;
2362}
2363
7b0db857
YM		 2364static int _ip6mr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb,
2365 u32 portid, u32 seq, struct mr_mfc *c,
2366 int cmd, int flags)
2367{
2368 return ip6mr_fill_mroute(mrt, skb, portid, seq, (struct mfc6_cache *)c,
2369 cmd, flags);
2370}
2371
812e44dd
ND		 2372static int mr6_msgsize(bool unresolved, int maxvif)
2373{
2374 size_t len =
2375 NLMSG_ALIGN(sizeof(struct rtmsg))
2376 + nla_total_size(4) /* RTA_TABLE */
2377 + nla_total_size(sizeof(struct in6_addr)) /* RTA_SRC */
2378 + nla_total_size(sizeof(struct in6_addr)) /* RTA_DST */
2379 ;
2380
2381 if (!unresolved)
2382 len = len
2383 + nla_total_size(4) /* RTA_IIF */
2384 + nla_total_size(0) /* RTA_MULTIPATH */
2385 + maxvif * NLA_ALIGN(sizeof(struct rtnexthop))
2386 /* RTA_MFC_STATS */
3d6b66c1 2387 + nla_total_size_64bit(sizeof(struct rta_mfc_stats))
812e44dd
ND		 2388 ;
2389
2390 return len;
2391}
2392
b70432f7 2393static void mr6_netlink_event(struct mr_table *mrt, struct mfc6_cache *mfc,
812e44dd
ND		 2394 int cmd)
2395{
2396 struct net *net = read_pnet(&mrt->net);
2397 struct sk_buff *skb;
2398 int err = -ENOBUFS;
2399
494fff56 2400 skb = nlmsg_new(mr6_msgsize(mfc->_c.mfc_parent >= MAXMIFS, mrt->maxvif),
812e44dd 2401 GFP_ATOMIC);
63159f29 2402 if (!skb)
812e44dd
ND		 2403 goto errout;
2404
f518338b 2405 err = ip6mr_fill_mroute(mrt, skb, 0, 0, mfc, cmd, 0);
812e44dd
ND		 2406 if (err < 0)
2407 goto errout;
2408
2409 rtnl_notify(skb, net, 0, RTNLGRP_IPV6_MROUTE, NULL, GFP_ATOMIC);
2410 return;
2411
2412errout:
2413 kfree_skb(skb);
2414 if (err < 0)
2415 rtnl_set_sk_err(net, RTNLGRP_IPV6_MROUTE, err);
2416}
2417
dd12d15c
JG		 2418static size_t mrt6msg_netlink_msgsize(size_t payloadlen)
2419{
2420 size_t len =
2421 NLMSG_ALIGN(sizeof(struct rtgenmsg))
2422 + nla_total_size(1) /* IP6MRA_CREPORT_MSGTYPE */
2423 + nla_total_size(4) /* IP6MRA_CREPORT_MIF_ID */
2424 /* IP6MRA_CREPORT_SRC_ADDR */
2425 + nla_total_size(sizeof(struct in6_addr))
2426 /* IP6MRA_CREPORT_DST_ADDR */
2427 + nla_total_size(sizeof(struct in6_addr))
2428 /* IP6MRA_CREPORT_PKT */
2429 + nla_total_size(payloadlen)
2430 ;
2431
2432 return len;
2433}
2434
b70432f7 2435static void mrt6msg_netlink_event(struct mr_table *mrt, struct sk_buff *pkt)
dd12d15c
JG		 2436{
2437 struct net *net = read_pnet(&mrt->net);
2438 struct nlmsghdr *nlh;
2439 struct rtgenmsg *rtgenm;
2440 struct mrt6msg *msg;
2441 struct sk_buff *skb;
2442 struct nlattr *nla;
2443 int payloadlen;
2444
2445 payloadlen = pkt->len - sizeof(struct mrt6msg);
2446 msg = (struct mrt6msg *)skb_transport_header(pkt);
2447
2448 skb = nlmsg_new(mrt6msg_netlink_msgsize(payloadlen), GFP_ATOMIC);
2449 if (!skb)
2450 goto errout;
2451
2452 nlh = nlmsg_put(skb, 0, 0, RTM_NEWCACHEREPORT,
2453 sizeof(struct rtgenmsg), 0);
2454 if (!nlh)
2455 goto errout;
2456 rtgenm = nlmsg_data(nlh);
2457 rtgenm->rtgen_family = RTNL_FAMILY_IP6MR;
2458 if (nla_put_u8(skb, IP6MRA_CREPORT_MSGTYPE, msg->im6_msgtype) ||
2459 nla_put_u32(skb, IP6MRA_CREPORT_MIF_ID, msg->im6_mif) ||
2460 nla_put_in6_addr(skb, IP6MRA_CREPORT_SRC_ADDR,
2461 &msg->im6_src) ||
2462 nla_put_in6_addr(skb, IP6MRA_CREPORT_DST_ADDR,
2463 &msg->im6_dst))
2464 goto nla_put_failure;
2465
2466 nla = nla_reserve(skb, IP6MRA_CREPORT_PKT, payloadlen);
2467 if (!nla || skb_copy_bits(pkt, sizeof(struct mrt6msg),
2468 nla_data(nla), payloadlen))
2469 goto nla_put_failure;
2470
2471 nlmsg_end(skb, nlh);
2472
2473 rtnl_notify(skb, net, 0, RTNLGRP_IPV6_MROUTE_R, NULL, GFP_ATOMIC);
2474 return;
2475
2476nla_put_failure:
2477 nlmsg_cancel(skb, nlh);
2478errout:
2479 kfree_skb(skb);
2480 rtnl_set_sk_err(net, RTNLGRP_IPV6_MROUTE_R, -ENOBUFS);
2481}
2482
5b285cac
PM		 2483static int ip6mr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb)
2484{
e8ba330a 2485 const struct nlmsghdr *nlh = cb->nlh;
4724676d 2486 struct fib_dump_filter filter = {};
cb167893 2487 int err;
e8ba330a
DA		 2488
2489 if (cb->strict_check) {
4724676d 2490 err = ip_valid_fib_dump_req(sock_net(skb->sk), nlh,
effe6792 2491 &filter, cb);
e8ba330a
DA		 2492 if (err < 0)
2493 return err;
2494 }
2495
cb167893
DA		 2496 if (filter.table_id) {
2497 struct mr_table *mrt;
2498
2499 mrt = ip6mr_get_table(sock_net(skb->sk), filter.table_id);
2500 if (!mrt) {
ae677bbb
DA		 2501 if (filter.dump_all_families)
2502 return skb->len;
2503
cb167893
DA		 2504 NL_SET_ERR_MSG_MOD(cb->extack, "MR table does not exist");
2505 return -ENOENT;
2506 }
2507 err = mr_table_dump(mrt, skb, cb, _ip6mr_fill_mroute,
2508 &mfc_unres_lock, &filter);
2509 return skb->len ? : err;
2510 }
2511
7b0db857 2512 return mr_rtm_dumproute(skb, cb, ip6mr_mr_table_iter,
cb167893 2513 _ip6mr_fill_mroute, &mfc_unres_lock, &filter);
5b285cac 2514}
Linux 6.x block layer and io_uring tree(s)