Commit | Line | Data |
---|---|---|
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 |
53 | struct ip6mr_rule { |
54 | struct fib_rule common; | |
55 | }; | |
56 | ||
57 | struct 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 | ||
a96f7a6a | 65 | static DEFINE_SPINLOCK(mrt_lock); |
7bc570c8 | 66 | |
ebc31979 ED |
67 | static struct net_device *vif_dev_read(const struct vif_device *vif) |
68 | { | |
a96f7a6a | 69 | return rcu_dereference(vif->dev); |
ebc31979 ED |
70 | } |
71 | ||
b70432f7 | 72 | /* Multicast router control variables */ |
7bc570c8 | 73 | |
7bc570c8 YH |
74 | /* Special spinlock for queue of unresolved entries */ |
75 | static DEFINE_SPINLOCK(mfc_unres_lock); | |
76 | ||
77 | /* We return to original Alan's scheme. Hash table of resolved | |
78 | entries is changed only in process context and protected | |
79 | with weak lock mrt_lock. Queue of unresolved entries is protected | |
80 | with strong spinlock mfc_unres_lock. | |
81 | ||
82 | In this case data path is free of exclusive locks at all. | |
83 | */ | |
84 | ||
85 | static struct kmem_cache *mrt_cachep __read_mostly; | |
86 | ||
b70432f7 YM |
87 | static struct mr_table *ip6mr_new_table(struct net *net, u32 id); |
88 | static void ip6mr_free_table(struct mr_table *mrt); | |
d1db275d | 89 | |
b70432f7 | 90 | static void ip6_mr_forward(struct net *net, struct mr_table *mrt, |
e4a38c0c PR |
91 | struct net_device *dev, struct sk_buff *skb, |
92 | struct mfc6_cache *cache); | |
3493a5b7 | 93 | static int ip6mr_cache_report(const struct mr_table *mrt, struct sk_buff *pkt, |
8229efda | 94 | mifi_t mifi, int assert); |
b70432f7 | 95 | static void mr6_netlink_event(struct mr_table *mrt, struct mfc6_cache *mfc, |
812e44dd | 96 | int cmd); |
3493a5b7 | 97 | static void mrt6msg_netlink_event(const struct mr_table *mrt, struct sk_buff *pkt); |
d7c31cbd DL |
98 | static int ip6mr_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh, |
99 | struct netlink_ext_ack *extack); | |
5b285cac PM |
100 | static int ip6mr_rtm_dumproute(struct sk_buff *skb, |
101 | struct netlink_callback *cb); | |
ca8d4794 | 102 | static void mroute_clean_tables(struct mr_table *mrt, int flags); |
e99e88a9 | 103 | static void ipmr_expire_process(struct timer_list *t); |
d1db275d PM |
104 | |
105 | #ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES | |
8ffb335e | 106 | #define ip6mr_for_each_table(mrt, net) \ |
28b380e2 | 107 | list_for_each_entry_rcu(mrt, &net->ipv6.mr6_tables, list, \ |
b6dd5acd MB |
108 | lockdep_rtnl_is_held() || \ |
109 | list_empty(&net->ipv6.mr6_tables)) | |
d1db275d | 110 | |
7b0db857 YM |
111 | static struct mr_table *ip6mr_mr_table_iter(struct net *net, |
112 | struct mr_table *mrt) | |
113 | { | |
114 | struct mr_table *ret; | |
115 | ||
116 | if (!mrt) | |
117 | ret = list_entry_rcu(net->ipv6.mr6_tables.next, | |
118 | struct mr_table, list); | |
119 | else | |
120 | ret = list_entry_rcu(mrt->list.next, | |
121 | struct mr_table, list); | |
122 | ||
123 | if (&ret->list == &net->ipv6.mr6_tables) | |
124 | return NULL; | |
125 | return ret; | |
126 | } | |
127 | ||
b70432f7 | 128 | static struct mr_table *ip6mr_get_table(struct net *net, u32 id) |
d1db275d | 129 | { |
b70432f7 | 130 | struct mr_table *mrt; |
d1db275d PM |
131 | |
132 | ip6mr_for_each_table(mrt, net) { | |
133 | if (mrt->id == id) | |
134 | return mrt; | |
135 | } | |
136 | return NULL; | |
137 | } | |
138 | ||
4c9483b2 | 139 | static int ip6mr_fib_lookup(struct net *net, struct flowi6 *flp6, |
b70432f7 | 140 | struct mr_table **mrt) |
d1db275d | 141 | { |
d1db275d | 142 | int err; |
95f4a45d HFS |
143 | struct ip6mr_result res; |
144 | struct fib_lookup_arg arg = { | |
145 | .result = &res, | |
146 | .flags = FIB_LOOKUP_NOREF, | |
147 | }; | |
d1db275d | 148 | |
e4a38c0c PR |
149 | /* update flow if oif or iif point to device enslaved to l3mdev */ |
150 | l3mdev_update_flow(net, flowi6_to_flowi(flp6)); | |
151 | ||
4c9483b2 DM |
152 | err = fib_rules_lookup(net->ipv6.mr6_rules_ops, |
153 | flowi6_to_flowi(flp6), 0, &arg); | |
d1db275d PM |
154 | if (err < 0) |
155 | return err; | |
156 | *mrt = res.mrt; | |
157 | return 0; | |
158 | } | |
159 | ||
160 | static int ip6mr_rule_action(struct fib_rule *rule, struct flowi *flp, | |
161 | int flags, struct fib_lookup_arg *arg) | |
162 | { | |
163 | struct ip6mr_result *res = arg->result; | |
b70432f7 | 164 | struct mr_table *mrt; |
d1db275d PM |
165 | |
166 | switch (rule->action) { | |
167 | case FR_ACT_TO_TBL: | |
168 | break; | |
169 | case FR_ACT_UNREACHABLE: | |
170 | return -ENETUNREACH; | |
171 | case FR_ACT_PROHIBIT: | |
172 | return -EACCES; | |
173 | case FR_ACT_BLACKHOLE: | |
174 | default: | |
175 | return -EINVAL; | |
176 | } | |
177 | ||
e4a38c0c PR |
178 | arg->table = fib_rule_get_table(rule, arg); |
179 | ||
180 | mrt = ip6mr_get_table(rule->fr_net, arg->table); | |
63159f29 | 181 | if (!mrt) |
d1db275d PM |
182 | return -EAGAIN; |
183 | res->mrt = mrt; | |
184 | return 0; | |
185 | } | |
186 | ||
187 | static int ip6mr_rule_match(struct fib_rule *rule, struct flowi *flp, int flags) | |
188 | { | |
189 | return 1; | |
190 | } | |
191 | ||
d1db275d | 192 | static int ip6mr_rule_configure(struct fib_rule *rule, struct sk_buff *skb, |
b16fb418 RP |
193 | struct fib_rule_hdr *frh, struct nlattr **tb, |
194 | struct netlink_ext_ack *extack) | |
d1db275d PM |
195 | { |
196 | return 0; | |
197 | } | |
198 | ||
199 | static int ip6mr_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh, | |
200 | struct nlattr **tb) | |
201 | { | |
202 | return 1; | |
203 | } | |
204 | ||
205 | static int ip6mr_rule_fill(struct fib_rule *rule, struct sk_buff *skb, | |
206 | struct fib_rule_hdr *frh) | |
207 | { | |
208 | frh->dst_len = 0; | |
209 | frh->src_len = 0; | |
210 | frh->tos = 0; | |
211 | return 0; | |
212 | } | |
213 | ||
04a6f82c | 214 | static const struct fib_rules_ops __net_initconst ip6mr_rules_ops_template = { |
d1db275d PM |
215 | .family = RTNL_FAMILY_IP6MR, |
216 | .rule_size = sizeof(struct ip6mr_rule), | |
217 | .addr_size = sizeof(struct in6_addr), | |
218 | .action = ip6mr_rule_action, | |
219 | .match = ip6mr_rule_match, | |
220 | .configure = ip6mr_rule_configure, | |
221 | .compare = ip6mr_rule_compare, | |
d1db275d PM |
222 | .fill = ip6mr_rule_fill, |
223 | .nlgroup = RTNLGRP_IPV6_RULE, | |
d1db275d PM |
224 | .owner = THIS_MODULE, |
225 | }; | |
226 | ||
227 | static int __net_init ip6mr_rules_init(struct net *net) | |
228 | { | |
229 | struct fib_rules_ops *ops; | |
b70432f7 | 230 | struct mr_table *mrt; |
d1db275d PM |
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); | |
e783bb00 SD |
240 | if (IS_ERR(mrt)) { |
241 | err = PTR_ERR(mrt); | |
d1db275d PM |
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 | ||
252 | err2: | |
5611a006 | 253 | rtnl_lock(); |
f243e5a7 | 254 | ip6mr_free_table(mrt); |
5611a006 | 255 | rtnl_unlock(); |
d1db275d PM |
256 | err1: |
257 | fib_rules_unregister(ops); | |
258 | return err; | |
259 | } | |
260 | ||
261 | static void __net_exit ip6mr_rules_exit(struct net *net) | |
262 | { | |
b70432f7 | 263 | struct mr_table *mrt, *next; |
d1db275d | 264 | |
e2f736b7 | 265 | ASSERT_RTNL(); |
035320d5 ED |
266 | list_for_each_entry_safe(mrt, next, &net->ipv6.mr6_tables, list) { |
267 | list_del(&mrt->list); | |
d1db275d | 268 | ip6mr_free_table(mrt); |
035320d5 | 269 | } |
d1db275d PM |
270 | fib_rules_unregister(net->ipv6.mr6_rules_ops); |
271 | } | |
088aa3ee | 272 | |
b7a59557 JP |
273 | static int ip6mr_rules_dump(struct net *net, struct notifier_block *nb, |
274 | struct netlink_ext_ack *extack) | |
088aa3ee | 275 | { |
b7a59557 | 276 | return fib_rules_dump(net, nb, RTNL_FAMILY_IP6MR, extack); |
088aa3ee YM |
277 | } |
278 | ||
279 | static unsigned int ip6mr_rules_seq_read(struct net *net) | |
280 | { | |
281 | return fib_rules_seq_read(net, RTNL_FAMILY_IP6MR); | |
282 | } | |
d3c07e5b YM |
283 | |
284 | bool ip6mr_rule_default(const struct fib_rule *rule) | |
285 | { | |
286 | return fib_rule_matchall(rule) && rule->action == FR_ACT_TO_TBL && | |
287 | rule->table == RT6_TABLE_DFLT && !rule->l3mdev; | |
288 | } | |
289 | EXPORT_SYMBOL(ip6mr_rule_default); | |
d1db275d PM |
290 | #else |
291 | #define ip6mr_for_each_table(mrt, net) \ | |
292 | for (mrt = net->ipv6.mrt6; mrt; mrt = NULL) | |
293 | ||
7b0db857 YM |
294 | static struct mr_table *ip6mr_mr_table_iter(struct net *net, |
295 | struct mr_table *mrt) | |
296 | { | |
297 | if (!mrt) | |
298 | return net->ipv6.mrt6; | |
299 | return NULL; | |
300 | } | |
301 | ||
b70432f7 | 302 | static struct mr_table *ip6mr_get_table(struct net *net, u32 id) |
d1db275d PM |
303 | { |
304 | return net->ipv6.mrt6; | |
305 | } | |
306 | ||
4c9483b2 | 307 | static int ip6mr_fib_lookup(struct net *net, struct flowi6 *flp6, |
b70432f7 | 308 | struct mr_table **mrt) |
d1db275d PM |
309 | { |
310 | *mrt = net->ipv6.mrt6; | |
311 | return 0; | |
312 | } | |
313 | ||
314 | static int __net_init ip6mr_rules_init(struct net *net) | |
315 | { | |
e783bb00 SD |
316 | struct mr_table *mrt; |
317 | ||
318 | mrt = ip6mr_new_table(net, RT6_TABLE_DFLT); | |
319 | if (IS_ERR(mrt)) | |
320 | return PTR_ERR(mrt); | |
321 | net->ipv6.mrt6 = mrt; | |
322 | return 0; | |
d1db275d PM |
323 | } |
324 | ||
325 | static void __net_exit ip6mr_rules_exit(struct net *net) | |
326 | { | |
e2f736b7 | 327 | ASSERT_RTNL(); |
d1db275d | 328 | ip6mr_free_table(net->ipv6.mrt6); |
905a6f96 | 329 | net->ipv6.mrt6 = NULL; |
d1db275d | 330 | } |
088aa3ee | 331 | |
b7a59557 JP |
332 | static int ip6mr_rules_dump(struct net *net, struct notifier_block *nb, |
333 | struct netlink_ext_ack *extack) | |
088aa3ee YM |
334 | { |
335 | return 0; | |
336 | } | |
337 | ||
338 | static unsigned int ip6mr_rules_seq_read(struct net *net) | |
339 | { | |
340 | return 0; | |
341 | } | |
d1db275d PM |
342 | #endif |
343 | ||
87c418bf YM |
344 | static int ip6mr_hash_cmp(struct rhashtable_compare_arg *arg, |
345 | const void *ptr) | |
346 | { | |
347 | const struct mfc6_cache_cmp_arg *cmparg = arg->key; | |
348 | struct mfc6_cache *c = (struct mfc6_cache *)ptr; | |
349 | ||
350 | return !ipv6_addr_equal(&c->mf6c_mcastgrp, &cmparg->mf6c_mcastgrp) || | |
351 | !ipv6_addr_equal(&c->mf6c_origin, &cmparg->mf6c_origin); | |
352 | } | |
353 | ||
354 | static const struct rhashtable_params ip6mr_rht_params = { | |
494fff56 | 355 | .head_offset = offsetof(struct mr_mfc, mnode), |
87c418bf YM |
356 | .key_offset = offsetof(struct mfc6_cache, cmparg), |
357 | .key_len = sizeof(struct mfc6_cache_cmp_arg), | |
358 | .nelem_hint = 3, | |
87c418bf YM |
359 | .obj_cmpfn = ip6mr_hash_cmp, |
360 | .automatic_shrinking = true, | |
361 | }; | |
362 | ||
0bbbf0e7 YM |
363 | static void ip6mr_new_table_set(struct mr_table *mrt, |
364 | struct net *net) | |
365 | { | |
366 | #ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES | |
367 | list_add_tail_rcu(&mrt->list, &net->ipv6.mr6_tables); | |
368 | #endif | |
369 | } | |
370 | ||
845c9a7a YM |
371 | static struct mfc6_cache_cmp_arg ip6mr_mr_table_ops_cmparg_any = { |
372 | .mf6c_origin = IN6ADDR_ANY_INIT, | |
373 | .mf6c_mcastgrp = IN6ADDR_ANY_INIT, | |
374 | }; | |
375 | ||
376 | static struct mr_table_ops ip6mr_mr_table_ops = { | |
377 | .rht_params = &ip6mr_rht_params, | |
378 | .cmparg_any = &ip6mr_mr_table_ops_cmparg_any, | |
379 | }; | |
380 | ||
b70432f7 | 381 | static struct mr_table *ip6mr_new_table(struct net *net, u32 id) |
d1db275d | 382 | { |
b70432f7 | 383 | struct mr_table *mrt; |
d1db275d PM |
384 | |
385 | mrt = ip6mr_get_table(net, id); | |
53b24b8f | 386 | if (mrt) |
d1db275d PM |
387 | return mrt; |
388 | ||
845c9a7a | 389 | return mr_table_alloc(net, id, &ip6mr_mr_table_ops, |
0bbbf0e7 | 390 | ipmr_expire_process, ip6mr_new_table_set); |
d1db275d | 391 | } |
7bc570c8 | 392 | |
b70432f7 | 393 | static void ip6mr_free_table(struct mr_table *mrt) |
d1db275d | 394 | { |
292a089d | 395 | timer_shutdown_sync(&mrt->ipmr_expire_timer); |
ca8d4794 CS |
396 | mroute_clean_tables(mrt, MRT6_FLUSH_MIFS | MRT6_FLUSH_MIFS_STATIC | |
397 | MRT6_FLUSH_MFC | MRT6_FLUSH_MFC_STATIC); | |
b70432f7 | 398 | rhltable_destroy(&mrt->mfc_hash); |
d1db275d PM |
399 | kfree(mrt); |
400 | } | |
7bc570c8 YH |
401 | |
402 | #ifdef CONFIG_PROC_FS | |
c8d61968 YM |
403 | /* The /proc interfaces to multicast routing |
404 | * /proc/ip6_mr_cache /proc/ip6_mr_vif | |
7bc570c8 YH |
405 | */ |
406 | ||
7bc570c8 | 407 | static void *ip6mr_vif_seq_start(struct seq_file *seq, loff_t *pos) |
b96ef16d | 408 | __acquires(RCU) |
7bc570c8 | 409 | { |
3feda6b4 | 410 | struct mr_vif_iter *iter = seq->private; |
8b90fc7e | 411 | struct net *net = seq_file_net(seq); |
b70432f7 | 412 | struct mr_table *mrt; |
d1db275d PM |
413 | |
414 | mrt = ip6mr_get_table(net, RT6_TABLE_DFLT); | |
63159f29 | 415 | if (!mrt) |
d1db275d PM |
416 | return ERR_PTR(-ENOENT); |
417 | ||
418 | iter->mrt = mrt; | |
8b90fc7e | 419 | |
b96ef16d | 420 | rcu_read_lock(); |
3feda6b4 | 421 | return mr_vif_seq_start(seq, pos); |
7bc570c8 YH |
422 | } |
423 | ||
424 | static void ip6mr_vif_seq_stop(struct seq_file *seq, void *v) | |
b96ef16d | 425 | __releases(RCU) |
7bc570c8 | 426 | { |
b96ef16d | 427 | rcu_read_unlock(); |
7bc570c8 YH |
428 | } |
429 | ||
430 | static int ip6mr_vif_seq_show(struct seq_file *seq, void *v) | |
431 | { | |
3feda6b4 | 432 | struct mr_vif_iter *iter = seq->private; |
b70432f7 | 433 | struct mr_table *mrt = iter->mrt; |
8b90fc7e | 434 | |
7bc570c8 YH |
435 | if (v == SEQ_START_TOKEN) { |
436 | seq_puts(seq, | |
437 | "Interface BytesIn PktsIn BytesOut PktsOut Flags\n"); | |
438 | } else { | |
6853f21f | 439 | const struct vif_device *vif = v; |
ebc31979 ED |
440 | const struct net_device *vif_dev; |
441 | const char *name; | |
442 | ||
443 | vif_dev = vif_dev_read(vif); | |
444 | name = vif_dev ? vif_dev->name : "none"; | |
7bc570c8 YH |
445 | |
446 | seq_printf(seq, | |
d430a227 | 447 | "%2td %-10s %8ld %7ld %8ld %7ld %05X\n", |
b70432f7 | 448 | vif - mrt->vif_table, |
7bc570c8 YH |
449 | name, vif->bytes_in, vif->pkt_in, |
450 | vif->bytes_out, vif->pkt_out, | |
451 | vif->flags); | |
452 | } | |
453 | return 0; | |
454 | } | |
455 | ||
98147d52 | 456 | static const struct seq_operations ip6mr_vif_seq_ops = { |
7bc570c8 | 457 | .start = ip6mr_vif_seq_start, |
3feda6b4 | 458 | .next = mr_vif_seq_next, |
7bc570c8 YH |
459 | .stop = ip6mr_vif_seq_stop, |
460 | .show = ip6mr_vif_seq_show, | |
461 | }; | |
462 | ||
7bc570c8 YH |
463 | static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos) |
464 | { | |
8b90fc7e | 465 | struct net *net = seq_file_net(seq); |
b70432f7 | 466 | struct mr_table *mrt; |
8b90fc7e | 467 | |
d1db275d | 468 | mrt = ip6mr_get_table(net, RT6_TABLE_DFLT); |
63159f29 | 469 | if (!mrt) |
d1db275d PM |
470 | return ERR_PTR(-ENOENT); |
471 | ||
c8d61968 | 472 | return mr_mfc_seq_start(seq, pos, mrt, &mfc_unres_lock); |
7bc570c8 YH |
473 | } |
474 | ||
475 | static int ipmr_mfc_seq_show(struct seq_file *seq, void *v) | |
476 | { | |
477 | int n; | |
478 | ||
479 | if (v == SEQ_START_TOKEN) { | |
480 | seq_puts(seq, | |
481 | "Group " | |
482 | "Origin " | |
483 | "Iif Pkts Bytes Wrong Oifs\n"); | |
484 | } else { | |
485 | const struct mfc6_cache *mfc = v; | |
c8d61968 | 486 | const struct mr_mfc_iter *it = seq->private; |
b70432f7 | 487 | struct mr_table *mrt = it->mrt; |
7bc570c8 | 488 | |
999890b2 | 489 | seq_printf(seq, "%pI6 %pI6 %-3hd", |
0c6ce78a | 490 | &mfc->mf6c_mcastgrp, &mfc->mf6c_origin, |
494fff56 | 491 | mfc->_c.mfc_parent); |
7bc570c8 | 492 | |
b70432f7 | 493 | if (it->cache != &mrt->mfc_unres_queue) { |
1ea472e2 | 494 | seq_printf(seq, " %8lu %8lu %8lu", |
494fff56 YM |
495 | mfc->_c.mfc_un.res.pkt, |
496 | mfc->_c.mfc_un.res.bytes, | |
497 | mfc->_c.mfc_un.res.wrong_if); | |
498 | for (n = mfc->_c.mfc_un.res.minvif; | |
499 | n < mfc->_c.mfc_un.res.maxvif; n++) { | |
b70432f7 | 500 | if (VIF_EXISTS(mrt, n) && |
494fff56 | 501 | mfc->_c.mfc_un.res.ttls[n] < 255) |
7bc570c8 | 502 | seq_printf(seq, |
494fff56 YM |
503 | " %2d:%-3d", n, |
504 | mfc->_c.mfc_un.res.ttls[n]); | |
7bc570c8 | 505 | } |
1ea472e2 BT |
506 | } else { |
507 | /* unresolved mfc_caches don't contain | |
508 | * pkt, bytes and wrong_if values | |
509 | */ | |
510 | seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul); | |
7bc570c8 YH |
511 | } |
512 | seq_putc(seq, '\n'); | |
513 | } | |
514 | return 0; | |
515 | } | |
516 | ||
88e9d34c | 517 | static const struct seq_operations ipmr_mfc_seq_ops = { |
7bc570c8 | 518 | .start = ipmr_mfc_seq_start, |
c8d61968 YM |
519 | .next = mr_mfc_seq_next, |
520 | .stop = mr_mfc_seq_stop, | |
7bc570c8 YH |
521 | .show = ipmr_mfc_seq_show, |
522 | }; | |
7bc570c8 YH |
523 | #endif |
524 | ||
14fb64e1 | 525 | #ifdef CONFIG_IPV6_PIMSM_V2 |
14fb64e1 YH |
526 | |
527 | static int pim6_rcv(struct sk_buff *skb) | |
528 | { | |
529 | struct pimreghdr *pim; | |
530 | struct ipv6hdr *encap; | |
531 | struct net_device *reg_dev = NULL; | |
8229efda | 532 | struct net *net = dev_net(skb->dev); |
b70432f7 | 533 | struct mr_table *mrt; |
4c9483b2 DM |
534 | struct flowi6 fl6 = { |
535 | .flowi6_iif = skb->dev->ifindex, | |
536 | .flowi6_mark = skb->mark, | |
d1db275d PM |
537 | }; |
538 | int reg_vif_num; | |
14fb64e1 YH |
539 | |
540 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(*encap))) | |
541 | goto drop; | |
542 | ||
543 | pim = (struct pimreghdr *)skb_transport_header(skb); | |
56245cae | 544 | if (pim->type != ((PIM_VERSION << 4) | PIM_TYPE_REGISTER) || |
14fb64e1 | 545 | (pim->flags & PIM_NULL_REGISTER) || |
1d6e55f1 TG |
546 | (csum_ipv6_magic(&ipv6_hdr(skb)->saddr, &ipv6_hdr(skb)->daddr, |
547 | sizeof(*pim), IPPROTO_PIM, | |
548 | csum_partial((void *)pim, sizeof(*pim), 0)) && | |
ec6b486f | 549 | csum_fold(skb_checksum(skb, 0, skb->len, 0)))) |
14fb64e1 YH |
550 | goto drop; |
551 | ||
552 | /* check if the inner packet is destined to mcast group */ | |
553 | encap = (struct ipv6hdr *)(skb_transport_header(skb) + | |
554 | sizeof(*pim)); | |
555 | ||
556 | if (!ipv6_addr_is_multicast(&encap->daddr) || | |
557 | encap->payload_len == 0 || | |
558 | ntohs(encap->payload_len) + sizeof(*pim) > skb->len) | |
559 | goto drop; | |
560 | ||
4c9483b2 | 561 | if (ip6mr_fib_lookup(net, &fl6, &mrt) < 0) |
d1db275d | 562 | goto drop; |
d1db275d | 563 | |
6d086587 ED |
564 | /* Pairs with WRITE_ONCE() in mif6_add()/mif6_delete() */ |
565 | reg_vif_num = READ_ONCE(mrt->mroute_reg_vif_num); | |
14fb64e1 | 566 | if (reg_vif_num >= 0) |
ebc31979 | 567 | reg_dev = vif_dev_read(&mrt->vif_table[reg_vif_num]); |
14fb64e1 | 568 | |
63159f29 | 569 | if (!reg_dev) |
14fb64e1 YH |
570 | goto drop; |
571 | ||
572 | skb->mac_header = skb->network_header; | |
573 | skb_pull(skb, (u8 *)encap - skb->data); | |
574 | skb_reset_network_header(skb); | |
1d6e55f1 | 575 | skb->protocol = htons(ETH_P_IPV6); |
3e49e6d5 | 576 | skb->ip_summed = CHECKSUM_NONE; |
d19d56dd | 577 | |
ea23192e | 578 | skb_tunnel_rx(skb, reg_dev, dev_net(reg_dev)); |
d19d56dd | 579 | |
caf586e5 | 580 | netif_rx(skb); |
8990f468 | 581 | |
14fb64e1 YH |
582 | return 0; |
583 | drop: | |
584 | kfree_skb(skb); | |
585 | return 0; | |
586 | } | |
587 | ||
41135cc8 | 588 | static const struct inet6_protocol pim6_protocol = { |
14fb64e1 YH |
589 | .handler = pim6_rcv, |
590 | }; | |
591 | ||
592 | /* Service routines creating virtual interfaces: PIMREG */ | |
593 | ||
6fef4c0c SH |
594 | static netdev_tx_t reg_vif_xmit(struct sk_buff *skb, |
595 | struct net_device *dev) | |
14fb64e1 | 596 | { |
8229efda | 597 | struct net *net = dev_net(dev); |
b70432f7 | 598 | struct mr_table *mrt; |
4c9483b2 DM |
599 | struct flowi6 fl6 = { |
600 | .flowi6_oif = dev->ifindex, | |
6a662719 | 601 | .flowi6_iif = skb->skb_iif ? : LOOPBACK_IFINDEX, |
4c9483b2 | 602 | .flowi6_mark = skb->mark, |
d1db275d | 603 | }; |
d1db275d | 604 | |
cb9f1b78 WB |
605 | if (!pskb_inet_may_pull(skb)) |
606 | goto tx_err; | |
607 | ||
608 | if (ip6mr_fib_lookup(net, &fl6, &mrt) < 0) | |
609 | goto tx_err; | |
8229efda | 610 | |
2fad1ba3 ED |
611 | DEV_STATS_ADD(dev, tx_bytes, skb->len); |
612 | DEV_STATS_INC(dev, tx_packets); | |
3493a5b7 ED |
613 | rcu_read_lock(); |
614 | ip6mr_cache_report(mrt, skb, READ_ONCE(mrt->mroute_reg_vif_num), | |
615 | MRT6MSG_WHOLEPKT); | |
616 | rcu_read_unlock(); | |
14fb64e1 | 617 | kfree_skb(skb); |
6ed10654 | 618 | return NETDEV_TX_OK; |
cb9f1b78 WB |
619 | |
620 | tx_err: | |
2fad1ba3 | 621 | DEV_STATS_INC(dev, tx_errors); |
cb9f1b78 WB |
622 | kfree_skb(skb); |
623 | return NETDEV_TX_OK; | |
14fb64e1 YH |
624 | } |
625 | ||
ee9b9596 ND |
626 | static int reg_vif_get_iflink(const struct net_device *dev) |
627 | { | |
628 | return 0; | |
629 | } | |
630 | ||
007c3838 SH |
631 | static const struct net_device_ops reg_vif_netdev_ops = { |
632 | .ndo_start_xmit = reg_vif_xmit, | |
ee9b9596 | 633 | .ndo_get_iflink = reg_vif_get_iflink, |
007c3838 SH |
634 | }; |
635 | ||
14fb64e1 YH |
636 | static void reg_vif_setup(struct net_device *dev) |
637 | { | |
638 | dev->type = ARPHRD_PIMREG; | |
639 | dev->mtu = 1500 - sizeof(struct ipv6hdr) - 8; | |
640 | dev->flags = IFF_NOARP; | |
007c3838 | 641 | dev->netdev_ops = ®_vif_netdev_ops; |
cf124db5 | 642 | dev->needs_free_netdev = true; |
403dbb97 | 643 | dev->features |= NETIF_F_NETNS_LOCAL; |
14fb64e1 YH |
644 | } |
645 | ||
b70432f7 | 646 | static struct net_device *ip6mr_reg_vif(struct net *net, struct mr_table *mrt) |
14fb64e1 YH |
647 | { |
648 | struct net_device *dev; | |
d1db275d PM |
649 | char name[IFNAMSIZ]; |
650 | ||
651 | if (mrt->id == RT6_TABLE_DFLT) | |
652 | sprintf(name, "pim6reg"); | |
653 | else | |
654 | sprintf(name, "pim6reg%u", mrt->id); | |
14fb64e1 | 655 | |
c835a677 | 656 | dev = alloc_netdev(0, name, NET_NAME_UNKNOWN, reg_vif_setup); |
63159f29 | 657 | if (!dev) |
14fb64e1 YH |
658 | return NULL; |
659 | ||
8229efda BT |
660 | dev_net_set(dev, net); |
661 | ||
14fb64e1 YH |
662 | if (register_netdevice(dev)) { |
663 | free_netdev(dev); | |
664 | return NULL; | |
665 | } | |
14fb64e1 | 666 | |
00f54e68 | 667 | if (dev_open(dev, NULL)) |
14fb64e1 YH |
668 | goto failure; |
669 | ||
7af3db78 | 670 | dev_hold(dev); |
14fb64e1 YH |
671 | return dev; |
672 | ||
673 | failure: | |
14fb64e1 YH |
674 | unregister_netdevice(dev); |
675 | return NULL; | |
676 | } | |
677 | #endif | |
678 | ||
088aa3ee YM |
679 | static int call_ip6mr_vif_entry_notifiers(struct net *net, |
680 | enum fib_event_type event_type, | |
681 | struct vif_device *vif, | |
ebc31979 | 682 | struct net_device *vif_dev, |
088aa3ee YM |
683 | mifi_t vif_index, u32 tb_id) |
684 | { | |
685 | return mr_call_vif_notifiers(net, RTNL_FAMILY_IP6MR, event_type, | |
ebc31979 | 686 | vif, vif_dev, vif_index, tb_id, |
088aa3ee YM |
687 | &net->ipv6.ipmr_seq); |
688 | } | |
7bc570c8 | 689 | |
088aa3ee YM |
690 | static int call_ip6mr_mfc_entry_notifiers(struct net *net, |
691 | enum fib_event_type event_type, | |
692 | struct mfc6_cache *mfc, u32 tb_id) | |
693 | { | |
694 | return mr_call_mfc_notifiers(net, RTNL_FAMILY_IP6MR, event_type, | |
695 | &mfc->_c, tb_id, &net->ipv6.ipmr_seq); | |
696 | } | |
697 | ||
698 | /* Delete a VIF entry */ | |
b70432f7 | 699 | static int mif6_delete(struct mr_table *mrt, int vifi, int notify, |
723b929c | 700 | struct list_head *head) |
7bc570c8 | 701 | { |
6853f21f | 702 | struct vif_device *v; |
7bc570c8 | 703 | struct net_device *dev; |
1d6e55f1 | 704 | struct inet6_dev *in6_dev; |
6bd52143 PM |
705 | |
706 | if (vifi < 0 || vifi >= mrt->maxvif) | |
7bc570c8 YH |
707 | return -EADDRNOTAVAIL; |
708 | ||
b70432f7 | 709 | v = &mrt->vif_table[vifi]; |
7bc570c8 | 710 | |
ebc31979 ED |
711 | dev = rtnl_dereference(v->dev); |
712 | if (!dev) | |
713 | return -EADDRNOTAVAIL; | |
088aa3ee | 714 | |
ebc31979 ED |
715 | call_ip6mr_vif_entry_notifiers(read_pnet(&mrt->net), |
716 | FIB_EVENT_VIF_DEL, v, dev, | |
717 | vifi, mrt->id); | |
a96f7a6a | 718 | spin_lock(&mrt_lock); |
ebc31979 | 719 | RCU_INIT_POINTER(v->dev, NULL); |
7bc570c8 | 720 | |
14fb64e1 | 721 | #ifdef CONFIG_IPV6_PIMSM_V2 |
3493a5b7 ED |
722 | if (vifi == mrt->mroute_reg_vif_num) { |
723 | /* Pairs with READ_ONCE() in ip6mr_cache_report() and reg_vif_xmit() */ | |
724 | WRITE_ONCE(mrt->mroute_reg_vif_num, -1); | |
725 | } | |
14fb64e1 YH |
726 | #endif |
727 | ||
6bd52143 | 728 | if (vifi + 1 == mrt->maxvif) { |
7bc570c8 YH |
729 | int tmp; |
730 | for (tmp = vifi - 1; tmp >= 0; tmp--) { | |
b70432f7 | 731 | if (VIF_EXISTS(mrt, tmp)) |
7bc570c8 YH |
732 | break; |
733 | } | |
db9eb7c8 | 734 | WRITE_ONCE(mrt->maxvif, tmp + 1); |
7bc570c8 YH |
735 | } |
736 | ||
a96f7a6a | 737 | spin_unlock(&mrt_lock); |
7bc570c8 YH |
738 | |
739 | dev_set_allmulti(dev, -1); | |
740 | ||
1d6e55f1 | 741 | in6_dev = __in6_dev_get(dev); |
d67b8c61 | 742 | if (in6_dev) { |
145c7a79 | 743 | atomic_dec(&in6_dev->cnf.mc_forwarding); |
85b3daad | 744 | inet6_netconf_notify_devconf(dev_net(dev), RTM_NEWNETCONF, |
d67b8c61 ND |
745 | NETCONFA_MC_FORWARDING, |
746 | dev->ifindex, &in6_dev->cnf); | |
747 | } | |
1d6e55f1 | 748 | |
723b929c | 749 | if ((v->flags & MIFF_REGISTER) && !notify) |
c871e664 | 750 | unregister_netdevice_queue(dev, head); |
7bc570c8 | 751 | |
d62607c3 | 752 | netdev_put(dev, &v->dev_tracker); |
7bc570c8 YH |
753 | return 0; |
754 | } | |
755 | ||
87c418bf | 756 | static inline void ip6mr_cache_free_rcu(struct rcu_head *head) |
58701ad4 | 757 | { |
494fff56 | 758 | struct mr_mfc *c = container_of(head, struct mr_mfc, rcu); |
87c418bf | 759 | |
494fff56 | 760 | kmem_cache_free(mrt_cachep, (struct mfc6_cache *)c); |
58701ad4 BT |
761 | } |
762 | ||
87c418bf YM |
763 | static inline void ip6mr_cache_free(struct mfc6_cache *c) |
764 | { | |
494fff56 | 765 | call_rcu(&c->_c.rcu, ip6mr_cache_free_rcu); |
87c418bf YM |
766 | } |
767 | ||
7bc570c8 YH |
768 | /* Destroy an unresolved cache entry, killing queued skbs |
769 | and reporting error to netlink readers. | |
770 | */ | |
771 | ||
b70432f7 | 772 | static void ip6mr_destroy_unres(struct mr_table *mrt, struct mfc6_cache *c) |
7bc570c8 | 773 | { |
6bd52143 | 774 | struct net *net = read_pnet(&mrt->net); |
7bc570c8 YH |
775 | struct sk_buff *skb; |
776 | ||
6bd52143 | 777 | atomic_dec(&mrt->cache_resolve_queue_len); |
7bc570c8 | 778 | |
494fff56 | 779 | while ((skb = skb_dequeue(&c->_c.mfc_un.unres.unresolved)) != NULL) { |
7bc570c8 | 780 | if (ipv6_hdr(skb)->version == 0) { |
af72868b JB |
781 | struct nlmsghdr *nlh = skb_pull(skb, |
782 | sizeof(struct ipv6hdr)); | |
7bc570c8 | 783 | nlh->nlmsg_type = NLMSG_ERROR; |
573ce260 | 784 | nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr)); |
7bc570c8 | 785 | skb_trim(skb, nlh->nlmsg_len); |
573ce260 | 786 | ((struct nlmsgerr *)nlmsg_data(nlh))->error = -ETIMEDOUT; |
15e47304 | 787 | rtnl_unicast(skb, net, NETLINK_CB(skb).portid); |
7bc570c8 YH |
788 | } else |
789 | kfree_skb(skb); | |
790 | } | |
791 | ||
58701ad4 | 792 | ip6mr_cache_free(c); |
7bc570c8 YH |
793 | } |
794 | ||
795 | ||
c476efbc | 796 | /* Timer process for all the unresolved queue. */ |
7bc570c8 | 797 | |
b70432f7 | 798 | static void ipmr_do_expire_process(struct mr_table *mrt) |
7bc570c8 YH |
799 | { |
800 | unsigned long now = jiffies; | |
801 | unsigned long expires = 10 * HZ; | |
494fff56 | 802 | struct mr_mfc *c, *next; |
7bc570c8 | 803 | |
b70432f7 | 804 | list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) { |
7bc570c8 YH |
805 | if (time_after(c->mfc_un.unres.expires, now)) { |
806 | /* not yet... */ | |
807 | unsigned long interval = c->mfc_un.unres.expires - now; | |
808 | if (interval < expires) | |
809 | expires = interval; | |
7bc570c8 YH |
810 | continue; |
811 | } | |
812 | ||
f30a7784 | 813 | list_del(&c->list); |
494fff56 YM |
814 | mr6_netlink_event(mrt, (struct mfc6_cache *)c, RTM_DELROUTE); |
815 | ip6mr_destroy_unres(mrt, (struct mfc6_cache *)c); | |
7bc570c8 YH |
816 | } |
817 | ||
b70432f7 | 818 | if (!list_empty(&mrt->mfc_unres_queue)) |
6bd52143 | 819 | mod_timer(&mrt->ipmr_expire_timer, jiffies + expires); |
7bc570c8 YH |
820 | } |
821 | ||
e99e88a9 | 822 | static void ipmr_expire_process(struct timer_list *t) |
7bc570c8 | 823 | { |
b70432f7 | 824 | struct mr_table *mrt = from_timer(mrt, t, ipmr_expire_timer); |
c476efbc | 825 | |
7bc570c8 | 826 | if (!spin_trylock(&mfc_unres_lock)) { |
6bd52143 | 827 | mod_timer(&mrt->ipmr_expire_timer, jiffies + 1); |
7bc570c8 YH |
828 | return; |
829 | } | |
830 | ||
b70432f7 | 831 | if (!list_empty(&mrt->mfc_unres_queue)) |
6bd52143 | 832 | ipmr_do_expire_process(mrt); |
7bc570c8 YH |
833 | |
834 | spin_unlock(&mfc_unres_lock); | |
835 | } | |
836 | ||
a96f7a6a | 837 | /* Fill oifs list. It is called under locked mrt_lock. */ |
7bc570c8 | 838 | |
b70432f7 | 839 | static void ip6mr_update_thresholds(struct mr_table *mrt, |
494fff56 | 840 | struct mr_mfc *cache, |
b5aa30b1 | 841 | unsigned char *ttls) |
7bc570c8 YH |
842 | { |
843 | int vifi; | |
844 | ||
6ac7eb08 | 845 | cache->mfc_un.res.minvif = MAXMIFS; |
7bc570c8 | 846 | cache->mfc_un.res.maxvif = 0; |
6ac7eb08 | 847 | memset(cache->mfc_un.res.ttls, 255, MAXMIFS); |
7bc570c8 | 848 | |
6bd52143 | 849 | for (vifi = 0; vifi < mrt->maxvif; vifi++) { |
b70432f7 | 850 | if (VIF_EXISTS(mrt, vifi) && |
4e16880c | 851 | ttls[vifi] && ttls[vifi] < 255) { |
7bc570c8 YH |
852 | cache->mfc_un.res.ttls[vifi] = ttls[vifi]; |
853 | if (cache->mfc_un.res.minvif > vifi) | |
854 | cache->mfc_un.res.minvif = vifi; | |
855 | if (cache->mfc_un.res.maxvif <= vifi) | |
856 | cache->mfc_un.res.maxvif = vifi + 1; | |
857 | } | |
858 | } | |
90b5ca17 | 859 | cache->mfc_un.res.lastuse = jiffies; |
7bc570c8 YH |
860 | } |
861 | ||
b70432f7 | 862 | static int mif6_add(struct net *net, struct mr_table *mrt, |
6bd52143 | 863 | struct mif6ctl *vifc, int mrtsock) |
7bc570c8 YH |
864 | { |
865 | int vifi = vifc->mif6c_mifi; | |
b70432f7 | 866 | struct vif_device *v = &mrt->vif_table[vifi]; |
7bc570c8 | 867 | struct net_device *dev; |
1d6e55f1 | 868 | struct inet6_dev *in6_dev; |
5ae7b444 | 869 | int err; |
7bc570c8 YH |
870 | |
871 | /* Is vif busy ? */ | |
b70432f7 | 872 | if (VIF_EXISTS(mrt, vifi)) |
7bc570c8 YH |
873 | return -EADDRINUSE; |
874 | ||
875 | switch (vifc->mif6c_flags) { | |
14fb64e1 YH |
876 | #ifdef CONFIG_IPV6_PIMSM_V2 |
877 | case MIFF_REGISTER: | |
878 | /* | |
879 | * Special Purpose VIF in PIM | |
880 | * All the packets will be sent to the daemon | |
881 | */ | |
6bd52143 | 882 | if (mrt->mroute_reg_vif_num >= 0) |
14fb64e1 | 883 | return -EADDRINUSE; |
d1db275d | 884 | dev = ip6mr_reg_vif(net, mrt); |
14fb64e1 YH |
885 | if (!dev) |
886 | return -ENOBUFS; | |
5ae7b444 WC |
887 | err = dev_set_allmulti(dev, 1); |
888 | if (err) { | |
889 | unregister_netdevice(dev); | |
7af3db78 | 890 | dev_put(dev); |
5ae7b444 WC |
891 | return err; |
892 | } | |
14fb64e1 YH |
893 | break; |
894 | #endif | |
7bc570c8 | 895 | case 0: |
8229efda | 896 | dev = dev_get_by_index(net, vifc->mif6c_pifi); |
7bc570c8 YH |
897 | if (!dev) |
898 | return -EADDRNOTAVAIL; | |
5ae7b444 | 899 | err = dev_set_allmulti(dev, 1); |
7af3db78 WC |
900 | if (err) { |
901 | dev_put(dev); | |
5ae7b444 | 902 | return err; |
7af3db78 | 903 | } |
7bc570c8 YH |
904 | break; |
905 | default: | |
906 | return -EINVAL; | |
907 | } | |
908 | ||
1d6e55f1 | 909 | in6_dev = __in6_dev_get(dev); |
d67b8c61 | 910 | if (in6_dev) { |
145c7a79 | 911 | atomic_inc(&in6_dev->cnf.mc_forwarding); |
85b3daad | 912 | inet6_netconf_notify_devconf(dev_net(dev), RTM_NEWNETCONF, |
d67b8c61 ND |
913 | NETCONFA_MC_FORWARDING, |
914 | dev->ifindex, &in6_dev->cnf); | |
915 | } | |
1d6e55f1 | 916 | |
6853f21f YM |
917 | /* Fill in the VIF structures */ |
918 | vif_device_init(v, dev, vifc->vifc_rate_limit, vifc->vifc_threshold, | |
919 | vifc->mif6c_flags | (!mrtsock ? VIFF_STATIC : 0), | |
920 | MIFF_REGISTER); | |
7bc570c8 YH |
921 | |
922 | /* And finish update writing critical data */ | |
a96f7a6a | 923 | spin_lock(&mrt_lock); |
ebc31979 | 924 | rcu_assign_pointer(v->dev, dev); |
42120a86 | 925 | netdev_tracker_alloc(dev, &v->dev_tracker, GFP_ATOMIC); |
14fb64e1 YH |
926 | #ifdef CONFIG_IPV6_PIMSM_V2 |
927 | if (v->flags & MIFF_REGISTER) | |
3493a5b7 | 928 | WRITE_ONCE(mrt->mroute_reg_vif_num, vifi); |
14fb64e1 | 929 | #endif |
6bd52143 | 930 | if (vifi + 1 > mrt->maxvif) |
db9eb7c8 | 931 | WRITE_ONCE(mrt->maxvif, vifi + 1); |
a96f7a6a | 932 | spin_unlock(&mrt_lock); |
088aa3ee | 933 | call_ip6mr_vif_entry_notifiers(net, FIB_EVENT_VIF_ADD, |
ebc31979 | 934 | v, dev, vifi, mrt->id); |
7bc570c8 YH |
935 | return 0; |
936 | } | |
937 | ||
b70432f7 | 938 | static struct mfc6_cache *ip6mr_cache_find(struct mr_table *mrt, |
b71d1d42 ED |
939 | const struct in6_addr *origin, |
940 | const struct in6_addr *mcastgrp) | |
7bc570c8 | 941 | { |
87c418bf YM |
942 | struct mfc6_cache_cmp_arg arg = { |
943 | .mf6c_origin = *origin, | |
944 | .mf6c_mcastgrp = *mcastgrp, | |
945 | }; | |
87c418bf | 946 | |
845c9a7a | 947 | return mr_mfc_find(mrt, &arg); |
660b26dc ND |
948 | } |
949 | ||
950 | /* Look for a (*,G) entry */ | |
b70432f7 | 951 | static struct mfc6_cache *ip6mr_cache_find_any(struct mr_table *mrt, |
660b26dc ND |
952 | struct in6_addr *mcastgrp, |
953 | mifi_t mifi) | |
954 | { | |
87c418bf YM |
955 | struct mfc6_cache_cmp_arg arg = { |
956 | .mf6c_origin = in6addr_any, | |
957 | .mf6c_mcastgrp = *mcastgrp, | |
958 | }; | |
660b26dc ND |
959 | |
960 | if (ipv6_addr_any(mcastgrp)) | |
845c9a7a YM |
961 | return mr_mfc_find_any_parent(mrt, mifi); |
962 | return mr_mfc_find_any(mrt, mifi, &arg); | |
660b26dc ND |
963 | } |
964 | ||
87c418bf YM |
965 | /* Look for a (S,G,iif) entry if parent != -1 */ |
966 | static struct mfc6_cache * | |
b70432f7 | 967 | ip6mr_cache_find_parent(struct mr_table *mrt, |
87c418bf YM |
968 | const struct in6_addr *origin, |
969 | const struct in6_addr *mcastgrp, | |
970 | int parent) | |
971 | { | |
972 | struct mfc6_cache_cmp_arg arg = { | |
973 | .mf6c_origin = *origin, | |
974 | .mf6c_mcastgrp = *mcastgrp, | |
975 | }; | |
87c418bf | 976 | |
845c9a7a | 977 | return mr_mfc_find_parent(mrt, &arg, parent); |
87c418bf YM |
978 | } |
979 | ||
845c9a7a | 980 | /* Allocate a multicast cache entry */ |
b5aa30b1 | 981 | static struct mfc6_cache *ip6mr_cache_alloc(void) |
7bc570c8 | 982 | { |
36cbac59 | 983 | struct mfc6_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL); |
63159f29 | 984 | if (!c) |
7bc570c8 | 985 | return NULL; |
494fff56 YM |
986 | c->_c.mfc_un.res.last_assert = jiffies - MFC_ASSERT_THRESH - 1; |
987 | c->_c.mfc_un.res.minvif = MAXMIFS; | |
8c13af2a YM |
988 | c->_c.free = ip6mr_cache_free_rcu; |
989 | refcount_set(&c->_c.mfc_un.res.refcount, 1); | |
7bc570c8 YH |
990 | return c; |
991 | } | |
992 | ||
b5aa30b1 | 993 | static struct mfc6_cache *ip6mr_cache_alloc_unres(void) |
7bc570c8 | 994 | { |
36cbac59 | 995 | struct mfc6_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC); |
63159f29 | 996 | if (!c) |
7bc570c8 | 997 | return NULL; |
494fff56 YM |
998 | skb_queue_head_init(&c->_c.mfc_un.unres.unresolved); |
999 | c->_c.mfc_un.unres.expires = jiffies + 10 * HZ; | |
7bc570c8 YH |
1000 | return c; |
1001 | } | |
1002 | ||
1003 | /* | |
1004 | * A cache entry has gone into a resolved state from queued | |
1005 | */ | |
1006 | ||
b70432f7 | 1007 | static void ip6mr_cache_resolve(struct net *net, struct mr_table *mrt, |
6bd52143 | 1008 | struct mfc6_cache *uc, struct mfc6_cache *c) |
7bc570c8 YH |
1009 | { |
1010 | struct sk_buff *skb; | |
1011 | ||
1012 | /* | |
1013 | * Play the pending entries through our router | |
1014 | */ | |
1015 | ||
494fff56 | 1016 | while ((skb = __skb_dequeue(&uc->_c.mfc_un.unres.unresolved))) { |
7bc570c8 | 1017 | if (ipv6_hdr(skb)->version == 0) { |
af72868b JB |
1018 | struct nlmsghdr *nlh = skb_pull(skb, |
1019 | sizeof(struct ipv6hdr)); | |
7bc570c8 | 1020 | |
7b0db857 YM |
1021 | if (mr_fill_mroute(mrt, skb, &c->_c, |
1022 | nlmsg_data(nlh)) > 0) { | |
549e028d | 1023 | nlh->nlmsg_len = skb_tail_pointer(skb) - (u8 *)nlh; |
7bc570c8 YH |
1024 | } else { |
1025 | nlh->nlmsg_type = NLMSG_ERROR; | |
573ce260 | 1026 | nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr)); |
7bc570c8 | 1027 | skb_trim(skb, nlh->nlmsg_len); |
573ce260 | 1028 | ((struct nlmsgerr *)nlmsg_data(nlh))->error = -EMSGSIZE; |
7bc570c8 | 1029 | } |
15e47304 | 1030 | rtnl_unicast(skb, net, NETLINK_CB(skb).portid); |
b07a9b26 IS |
1031 | } else { |
1032 | rcu_read_lock(); | |
e4a38c0c | 1033 | ip6_mr_forward(net, mrt, skb->dev, skb, c); |
b07a9b26 IS |
1034 | rcu_read_unlock(); |
1035 | } | |
7bc570c8 YH |
1036 | } |
1037 | } | |
1038 | ||
1039 | /* | |
dd12d15c | 1040 | * Bounce a cache query up to pim6sd and netlink. |
7bc570c8 | 1041 | * |
3493a5b7 | 1042 | * Called under rcu_read_lock() |
7bc570c8 YH |
1043 | */ |
1044 | ||
3493a5b7 | 1045 | static int ip6mr_cache_report(const struct mr_table *mrt, struct sk_buff *pkt, |
6bd52143 | 1046 | mifi_t mifi, int assert) |
7bc570c8 | 1047 | { |
8571ab47 | 1048 | struct sock *mroute6_sk; |
7bc570c8 YH |
1049 | struct sk_buff *skb; |
1050 | struct mrt6msg *msg; | |
1051 | int ret; | |
1052 | ||
14fb64e1 | 1053 | #ifdef CONFIG_IPV6_PIMSM_V2 |
4b340a5a | 1054 | if (assert == MRT6MSG_WHOLEPKT || assert == MRT6MSG_WRMIFWHOLE) |
14fb64e1 YH |
1055 | skb = skb_realloc_headroom(pkt, -skb_network_offset(pkt) |
1056 | +sizeof(*msg)); | |
1057 | else | |
1058 | #endif | |
1059 | skb = alloc_skb(sizeof(struct ipv6hdr) + sizeof(*msg), GFP_ATOMIC); | |
7bc570c8 YH |
1060 | |
1061 | if (!skb) | |
1062 | return -ENOBUFS; | |
1063 | ||
1064 | /* I suppose that internal messages | |
1065 | * do not require checksums */ | |
1066 | ||
1067 | skb->ip_summed = CHECKSUM_UNNECESSARY; | |
1068 | ||
14fb64e1 | 1069 | #ifdef CONFIG_IPV6_PIMSM_V2 |
4b340a5a | 1070 | if (assert == MRT6MSG_WHOLEPKT || assert == MRT6MSG_WRMIFWHOLE) { |
14fb64e1 YH |
1071 | /* Ugly, but we have no choice with this interface. |
1072 | Duplicate old header, fix length etc. | |
1073 | And all this only to mangle msg->im6_msgtype and | |
1074 | to set msg->im6_mbz to "mbz" :-) | |
1075 | */ | |
1076 | skb_push(skb, -skb_network_offset(pkt)); | |
1077 | ||
1078 | skb_push(skb, sizeof(*msg)); | |
1079 | skb_reset_transport_header(skb); | |
1080 | msg = (struct mrt6msg *)skb_transport_header(skb); | |
1081 | msg->im6_mbz = 0; | |
4b340a5a MR |
1082 | msg->im6_msgtype = assert; |
1083 | if (assert == MRT6MSG_WRMIFWHOLE) | |
1084 | msg->im6_mif = mifi; | |
1085 | else | |
3493a5b7 | 1086 | msg->im6_mif = READ_ONCE(mrt->mroute_reg_vif_num); |
14fb64e1 | 1087 | msg->im6_pad = 0; |
4e3fd7a0 AD |
1088 | msg->im6_src = ipv6_hdr(pkt)->saddr; |
1089 | msg->im6_dst = ipv6_hdr(pkt)->daddr; | |
14fb64e1 YH |
1090 | |
1091 | skb->ip_summed = CHECKSUM_UNNECESSARY; | |
1092 | } else | |
1093 | #endif | |
1094 | { | |
7bc570c8 YH |
1095 | /* |
1096 | * Copy the IP header | |
1097 | */ | |
1098 | ||
1099 | skb_put(skb, sizeof(struct ipv6hdr)); | |
1100 | skb_reset_network_header(skb); | |
1101 | skb_copy_to_linear_data(skb, ipv6_hdr(pkt), sizeof(struct ipv6hdr)); | |
1102 | ||
1103 | /* | |
1104 | * Add our header | |
1105 | */ | |
1106 | skb_put(skb, sizeof(*msg)); | |
1107 | skb_reset_transport_header(skb); | |
1108 | msg = (struct mrt6msg *)skb_transport_header(skb); | |
1109 | ||
1110 | msg->im6_mbz = 0; | |
1111 | msg->im6_msgtype = assert; | |
6ac7eb08 | 1112 | msg->im6_mif = mifi; |
7bc570c8 | 1113 | msg->im6_pad = 0; |
4e3fd7a0 AD |
1114 | msg->im6_src = ipv6_hdr(pkt)->saddr; |
1115 | msg->im6_dst = ipv6_hdr(pkt)->daddr; | |
7bc570c8 | 1116 | |
adf30907 | 1117 | skb_dst_set(skb, dst_clone(skb_dst(pkt))); |
7bc570c8 | 1118 | skb->ip_summed = CHECKSUM_UNNECESSARY; |
14fb64e1 | 1119 | } |
7bc570c8 | 1120 | |
b70432f7 | 1121 | mroute6_sk = rcu_dereference(mrt->mroute_sk); |
8571ab47 | 1122 | if (!mroute6_sk) { |
7bc570c8 YH |
1123 | kfree_skb(skb); |
1124 | return -EINVAL; | |
1125 | } | |
1126 | ||
dd12d15c JG |
1127 | mrt6msg_netlink_event(mrt, skb); |
1128 | ||
8571ab47 YM |
1129 | /* Deliver to user space multicast routing algorithms */ |
1130 | ret = sock_queue_rcv_skb(mroute6_sk, skb); | |
3493a5b7 | 1131 | |
bd91b8bf | 1132 | if (ret < 0) { |
e87cc472 | 1133 | net_warn_ratelimited("mroute6: pending queue full, dropping entries\n"); |
7bc570c8 YH |
1134 | kfree_skb(skb); |
1135 | } | |
1136 | ||
1137 | return ret; | |
1138 | } | |
1139 | ||
494fff56 YM |
1140 | /* Queue a packet for resolution. It gets locked cache entry! */ |
1141 | static int ip6mr_cache_unresolved(struct mr_table *mrt, mifi_t mifi, | |
e4a38c0c | 1142 | struct sk_buff *skb, struct net_device *dev) |
7bc570c8 | 1143 | { |
494fff56 | 1144 | struct mfc6_cache *c; |
f30a7784 | 1145 | bool found = false; |
7bc570c8 | 1146 | int err; |
7bc570c8 YH |
1147 | |
1148 | spin_lock_bh(&mfc_unres_lock); | |
494fff56 | 1149 | list_for_each_entry(c, &mrt->mfc_unres_queue, _c.list) { |
c476efbc | 1150 | if (ipv6_addr_equal(&c->mf6c_mcastgrp, &ipv6_hdr(skb)->daddr) && |
f30a7784 PM |
1151 | ipv6_addr_equal(&c->mf6c_origin, &ipv6_hdr(skb)->saddr)) { |
1152 | found = true; | |
7bc570c8 | 1153 | break; |
f30a7784 | 1154 | } |
7bc570c8 YH |
1155 | } |
1156 | ||
f30a7784 | 1157 | if (!found) { |
7bc570c8 YH |
1158 | /* |
1159 | * Create a new entry if allowable | |
1160 | */ | |
1161 | ||
0079ad8e HL |
1162 | c = ip6mr_cache_alloc_unres(); |
1163 | if (!c) { | |
7bc570c8 YH |
1164 | spin_unlock_bh(&mfc_unres_lock); |
1165 | ||
1166 | kfree_skb(skb); | |
1167 | return -ENOBUFS; | |
1168 | } | |
1169 | ||
494fff56 YM |
1170 | /* Fill in the new cache entry */ |
1171 | c->_c.mfc_parent = -1; | |
7bc570c8 YH |
1172 | c->mf6c_origin = ipv6_hdr(skb)->saddr; |
1173 | c->mf6c_mcastgrp = ipv6_hdr(skb)->daddr; | |
1174 | ||
1175 | /* | |
1176 | * Reflect first query at pim6sd | |
1177 | */ | |
6bd52143 | 1178 | err = ip6mr_cache_report(mrt, skb, mifi, MRT6MSG_NOCACHE); |
8229efda | 1179 | if (err < 0) { |
7bc570c8 YH |
1180 | /* If the report failed throw the cache entry |
1181 | out - Brad Parker | |
1182 | */ | |
1183 | spin_unlock_bh(&mfc_unres_lock); | |
1184 | ||
58701ad4 | 1185 | ip6mr_cache_free(c); |
7bc570c8 YH |
1186 | kfree_skb(skb); |
1187 | return err; | |
1188 | } | |
1189 | ||
6bd52143 | 1190 | atomic_inc(&mrt->cache_resolve_queue_len); |
494fff56 | 1191 | list_add(&c->_c.list, &mrt->mfc_unres_queue); |
812e44dd | 1192 | mr6_netlink_event(mrt, c, RTM_NEWROUTE); |
7bc570c8 | 1193 | |
6bd52143 | 1194 | ipmr_do_expire_process(mrt); |
7bc570c8 YH |
1195 | } |
1196 | ||
494fff56 YM |
1197 | /* See if we can append the packet */ |
1198 | if (c->_c.mfc_un.unres.unresolved.qlen > 3) { | |
7bc570c8 YH |
1199 | kfree_skb(skb); |
1200 | err = -ENOBUFS; | |
1201 | } else { | |
e4a38c0c PR |
1202 | if (dev) { |
1203 | skb->dev = dev; | |
1204 | skb->skb_iif = dev->ifindex; | |
1205 | } | |
494fff56 | 1206 | skb_queue_tail(&c->_c.mfc_un.unres.unresolved, skb); |
7bc570c8 YH |
1207 | err = 0; |
1208 | } | |
1209 | ||
1210 | spin_unlock_bh(&mfc_unres_lock); | |
1211 | return err; | |
1212 | } | |
1213 | ||
1214 | /* | |
1215 | * MFC6 cache manipulation by user space | |
1216 | */ | |
1217 | ||
b70432f7 | 1218 | static int ip6mr_mfc_delete(struct mr_table *mrt, struct mf6cctl *mfc, |
660b26dc | 1219 | int parent) |
7bc570c8 | 1220 | { |
87c418bf | 1221 | struct mfc6_cache *c; |
7bc570c8 | 1222 | |
87c418bf YM |
1223 | /* The entries are added/deleted only under RTNL */ |
1224 | rcu_read_lock(); | |
1225 | c = ip6mr_cache_find_parent(mrt, &mfc->mf6cc_origin.sin6_addr, | |
1226 | &mfc->mf6cc_mcastgrp.sin6_addr, parent); | |
1227 | rcu_read_unlock(); | |
1228 | if (!c) | |
1229 | return -ENOENT; | |
494fff56 YM |
1230 | rhltable_remove(&mrt->mfc_hash, &c->_c.mnode, ip6mr_rht_params); |
1231 | list_del_rcu(&c->_c.list); | |
7bc570c8 | 1232 | |
088aa3ee YM |
1233 | call_ip6mr_mfc_entry_notifiers(read_pnet(&mrt->net), |
1234 | FIB_EVENT_ENTRY_DEL, c, mrt->id); | |
87c418bf | 1235 | mr6_netlink_event(mrt, c, RTM_DELROUTE); |
8c13af2a | 1236 | mr_cache_put(&c->_c); |
87c418bf | 1237 | return 0; |
7bc570c8 YH |
1238 | } |
1239 | ||
1240 | static int ip6mr_device_event(struct notifier_block *this, | |
1241 | unsigned long event, void *ptr) | |
1242 | { | |
351638e7 | 1243 | struct net_device *dev = netdev_notifier_info_to_dev(ptr); |
8229efda | 1244 | struct net *net = dev_net(dev); |
b70432f7 | 1245 | struct mr_table *mrt; |
6853f21f | 1246 | struct vif_device *v; |
7bc570c8 YH |
1247 | int ct; |
1248 | ||
7bc570c8 YH |
1249 | if (event != NETDEV_UNREGISTER) |
1250 | return NOTIFY_DONE; | |
1251 | ||
d1db275d | 1252 | ip6mr_for_each_table(mrt, net) { |
b70432f7 | 1253 | v = &mrt->vif_table[0]; |
d1db275d | 1254 | for (ct = 0; ct < mrt->maxvif; ct++, v++) { |
ebc31979 | 1255 | if (rcu_access_pointer(v->dev) == dev) |
723b929c | 1256 | mif6_delete(mrt, ct, 1, NULL); |
d1db275d | 1257 | } |
7bc570c8 | 1258 | } |
c871e664 | 1259 | |
7bc570c8 YH |
1260 | return NOTIFY_DONE; |
1261 | } | |
1262 | ||
088aa3ee YM |
1263 | static unsigned int ip6mr_seq_read(struct net *net) |
1264 | { | |
1265 | ASSERT_RTNL(); | |
1266 | ||
1267 | return net->ipv6.ipmr_seq + ip6mr_rules_seq_read(net); | |
1268 | } | |
1269 | ||
b7a59557 JP |
1270 | static int ip6mr_dump(struct net *net, struct notifier_block *nb, |
1271 | struct netlink_ext_ack *extack) | |
088aa3ee YM |
1272 | { |
1273 | return mr_dump(net, nb, RTNL_FAMILY_IP6MR, ip6mr_rules_dump, | |
194366b2 | 1274 | ip6mr_mr_table_iter, extack); |
088aa3ee YM |
1275 | } |
1276 | ||
7bc570c8 YH |
1277 | static struct notifier_block ip6_mr_notifier = { |
1278 | .notifier_call = ip6mr_device_event | |
1279 | }; | |
1280 | ||
088aa3ee YM |
1281 | static const struct fib_notifier_ops ip6mr_notifier_ops_template = { |
1282 | .family = RTNL_FAMILY_IP6MR, | |
1283 | .fib_seq_read = ip6mr_seq_read, | |
1284 | .fib_dump = ip6mr_dump, | |
1285 | .owner = THIS_MODULE, | |
1286 | }; | |
1287 | ||
1288 | static int __net_init ip6mr_notifier_init(struct net *net) | |
1289 | { | |
1290 | struct fib_notifier_ops *ops; | |
1291 | ||
1292 | net->ipv6.ipmr_seq = 0; | |
7bc570c8 | 1293 | |
088aa3ee YM |
1294 | ops = fib_notifier_ops_register(&ip6mr_notifier_ops_template, net); |
1295 | if (IS_ERR(ops)) | |
1296 | return PTR_ERR(ops); | |
1297 | ||
1298 | net->ipv6.ip6mr_notifier_ops = ops; | |
1299 | ||
1300 | return 0; | |
1301 | } | |
1302 | ||
1303 | static void __net_exit ip6mr_notifier_exit(struct net *net) | |
1304 | { | |
1305 | fib_notifier_ops_unregister(net->ipv6.ip6mr_notifier_ops); | |
1306 | net->ipv6.ip6mr_notifier_ops = NULL; | |
1307 | } | |
1308 | ||
1309 | /* Setup for IP multicast routing */ | |
4e16880c BT |
1310 | static int __net_init ip6mr_net_init(struct net *net) |
1311 | { | |
d1db275d | 1312 | int err; |
f30a7784 | 1313 | |
088aa3ee YM |
1314 | err = ip6mr_notifier_init(net); |
1315 | if (err) | |
1316 | return err; | |
1317 | ||
d1db275d PM |
1318 | err = ip6mr_rules_init(net); |
1319 | if (err < 0) | |
088aa3ee | 1320 | goto ip6mr_rules_fail; |
8b90fc7e BT |
1321 | |
1322 | #ifdef CONFIG_PROC_FS | |
1323 | err = -ENOMEM; | |
c3506372 CH |
1324 | if (!proc_create_net("ip6_mr_vif", 0, net->proc_net, &ip6mr_vif_seq_ops, |
1325 | sizeof(struct mr_vif_iter))) | |
8b90fc7e | 1326 | goto proc_vif_fail; |
c3506372 CH |
1327 | if (!proc_create_net("ip6_mr_cache", 0, net->proc_net, &ipmr_mfc_seq_ops, |
1328 | sizeof(struct mr_mfc_iter))) | |
8b90fc7e BT |
1329 | goto proc_cache_fail; |
1330 | #endif | |
6bd52143 | 1331 | |
4a6258a0 BT |
1332 | return 0; |
1333 | ||
8b90fc7e BT |
1334 | #ifdef CONFIG_PROC_FS |
1335 | proc_cache_fail: | |
ece31ffd | 1336 | remove_proc_entry("ip6_mr_vif", net->proc_net); |
8b90fc7e | 1337 | proc_vif_fail: |
e2f736b7 | 1338 | rtnl_lock(); |
d1db275d | 1339 | ip6mr_rules_exit(net); |
e2f736b7 | 1340 | rtnl_unlock(); |
8b90fc7e | 1341 | #endif |
088aa3ee YM |
1342 | ip6mr_rules_fail: |
1343 | ip6mr_notifier_exit(net); | |
4e16880c BT |
1344 | return err; |
1345 | } | |
1346 | ||
1347 | static void __net_exit ip6mr_net_exit(struct net *net) | |
1348 | { | |
8b90fc7e | 1349 | #ifdef CONFIG_PROC_FS |
ece31ffd G |
1350 | remove_proc_entry("ip6_mr_cache", net->proc_net); |
1351 | remove_proc_entry("ip6_mr_vif", net->proc_net); | |
8b90fc7e | 1352 | #endif |
088aa3ee | 1353 | ip6mr_notifier_exit(net); |
4e16880c BT |
1354 | } |
1355 | ||
e2f736b7 ED |
1356 | static void __net_exit ip6mr_net_exit_batch(struct list_head *net_list) |
1357 | { | |
1358 | struct net *net; | |
1359 | ||
1360 | rtnl_lock(); | |
1361 | list_for_each_entry(net, net_list, exit_list) | |
1362 | ip6mr_rules_exit(net); | |
1363 | rtnl_unlock(); | |
1364 | } | |
1365 | ||
4e16880c BT |
1366 | static struct pernet_operations ip6mr_net_ops = { |
1367 | .init = ip6mr_net_init, | |
1368 | .exit = ip6mr_net_exit, | |
e2f736b7 | 1369 | .exit_batch = ip6mr_net_exit_batch, |
4e16880c BT |
1370 | }; |
1371 | ||
623d1a1a | 1372 | int __init ip6_mr_init(void) |
7bc570c8 | 1373 | { |
623d1a1a WC |
1374 | int err; |
1375 | ||
7bc570c8 YH |
1376 | mrt_cachep = kmem_cache_create("ip6_mrt_cache", |
1377 | sizeof(struct mfc6_cache), | |
1378 | 0, SLAB_HWCACHE_ALIGN, | |
1379 | NULL); | |
1380 | if (!mrt_cachep) | |
623d1a1a | 1381 | return -ENOMEM; |
7bc570c8 | 1382 | |
4e16880c BT |
1383 | err = register_pernet_subsys(&ip6mr_net_ops); |
1384 | if (err) | |
1385 | goto reg_pernet_fail; | |
1386 | ||
623d1a1a WC |
1387 | err = register_netdevice_notifier(&ip6_mr_notifier); |
1388 | if (err) | |
1389 | goto reg_notif_fail; | |
403dbb97 TG |
1390 | #ifdef CONFIG_IPV6_PIMSM_V2 |
1391 | if (inet6_add_protocol(&pim6_protocol, IPPROTO_PIM) < 0) { | |
f3213831 | 1392 | pr_err("%s: can't add PIM protocol\n", __func__); |
403dbb97 TG |
1393 | err = -EAGAIN; |
1394 | goto add_proto_fail; | |
1395 | } | |
1396 | #endif | |
a3fde2ad | 1397 | err = rtnl_register_module(THIS_MODULE, RTNL_FAMILY_IP6MR, RTM_GETROUTE, |
d7c31cbd | 1398 | ip6mr_rtm_getroute, ip6mr_rtm_dumproute, 0); |
a3fde2ad FW |
1399 | if (err == 0) |
1400 | return 0; | |
1401 | ||
403dbb97 | 1402 | #ifdef CONFIG_IPV6_PIMSM_V2 |
a3fde2ad | 1403 | inet6_del_protocol(&pim6_protocol, IPPROTO_PIM); |
403dbb97 TG |
1404 | add_proto_fail: |
1405 | unregister_netdevice_notifier(&ip6_mr_notifier); | |
1406 | #endif | |
87b30a65 | 1407 | reg_notif_fail: |
4e16880c BT |
1408 | unregister_pernet_subsys(&ip6mr_net_ops); |
1409 | reg_pernet_fail: | |
87b30a65 | 1410 | kmem_cache_destroy(mrt_cachep); |
623d1a1a | 1411 | return err; |
7bc570c8 YH |
1412 | } |
1413 | ||
623d1a1a WC |
1414 | void ip6_mr_cleanup(void) |
1415 | { | |
ffb1388a DJ |
1416 | rtnl_unregister(RTNL_FAMILY_IP6MR, RTM_GETROUTE); |
1417 | #ifdef CONFIG_IPV6_PIMSM_V2 | |
1418 | inet6_del_protocol(&pim6_protocol, IPPROTO_PIM); | |
1419 | #endif | |
623d1a1a | 1420 | unregister_netdevice_notifier(&ip6_mr_notifier); |
4e16880c | 1421 | unregister_pernet_subsys(&ip6mr_net_ops); |
623d1a1a WC |
1422 | kmem_cache_destroy(mrt_cachep); |
1423 | } | |
7bc570c8 | 1424 | |
b70432f7 | 1425 | static int ip6mr_mfc_add(struct net *net, struct mr_table *mrt, |
660b26dc | 1426 | struct mf6cctl *mfc, int mrtsock, int parent) |
7bc570c8 | 1427 | { |
6ac7eb08 | 1428 | unsigned char ttls[MAXMIFS]; |
87c418bf | 1429 | struct mfc6_cache *uc, *c; |
494fff56 | 1430 | struct mr_mfc *_uc; |
87c418bf YM |
1431 | bool found; |
1432 | int i, err; | |
7bc570c8 | 1433 | |
a50436f2 PM |
1434 | if (mfc->mf6cc_parent >= MAXMIFS) |
1435 | return -ENFILE; | |
1436 | ||
6ac7eb08 RR |
1437 | memset(ttls, 255, MAXMIFS); |
1438 | for (i = 0; i < MAXMIFS; i++) { | |
7bc570c8 YH |
1439 | if (IF_ISSET(i, &mfc->mf6cc_ifset)) |
1440 | ttls[i] = 1; | |
7bc570c8 YH |
1441 | } |
1442 | ||
87c418bf YM |
1443 | /* The entries are added/deleted only under RTNL */ |
1444 | rcu_read_lock(); | |
1445 | c = ip6mr_cache_find_parent(mrt, &mfc->mf6cc_origin.sin6_addr, | |
1446 | &mfc->mf6cc_mcastgrp.sin6_addr, parent); | |
1447 | rcu_read_unlock(); | |
1448 | if (c) { | |
a96f7a6a | 1449 | spin_lock(&mrt_lock); |
494fff56 YM |
1450 | c->_c.mfc_parent = mfc->mf6cc_parent; |
1451 | ip6mr_update_thresholds(mrt, &c->_c, ttls); | |
7bc570c8 | 1452 | if (!mrtsock) |
494fff56 | 1453 | c->_c.mfc_flags |= MFC_STATIC; |
a96f7a6a | 1454 | spin_unlock(&mrt_lock); |
088aa3ee YM |
1455 | call_ip6mr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_REPLACE, |
1456 | c, mrt->id); | |
812e44dd | 1457 | mr6_netlink_event(mrt, c, RTM_NEWROUTE); |
7bc570c8 YH |
1458 | return 0; |
1459 | } | |
1460 | ||
660b26dc ND |
1461 | if (!ipv6_addr_any(&mfc->mf6cc_mcastgrp.sin6_addr) && |
1462 | !ipv6_addr_is_multicast(&mfc->mf6cc_mcastgrp.sin6_addr)) | |
7bc570c8 YH |
1463 | return -EINVAL; |
1464 | ||
b5aa30b1 | 1465 | c = ip6mr_cache_alloc(); |
63159f29 | 1466 | if (!c) |
7bc570c8 YH |
1467 | return -ENOMEM; |
1468 | ||
1469 | c->mf6c_origin = mfc->mf6cc_origin.sin6_addr; | |
1470 | c->mf6c_mcastgrp = mfc->mf6cc_mcastgrp.sin6_addr; | |
494fff56 YM |
1471 | c->_c.mfc_parent = mfc->mf6cc_parent; |
1472 | ip6mr_update_thresholds(mrt, &c->_c, ttls); | |
7bc570c8 | 1473 | if (!mrtsock) |
494fff56 | 1474 | c->_c.mfc_flags |= MFC_STATIC; |
7bc570c8 | 1475 | |
494fff56 | 1476 | err = rhltable_insert_key(&mrt->mfc_hash, &c->cmparg, &c->_c.mnode, |
87c418bf YM |
1477 | ip6mr_rht_params); |
1478 | if (err) { | |
1479 | pr_err("ip6mr: rhtable insert error %d\n", err); | |
1480 | ip6mr_cache_free(c); | |
1481 | return err; | |
1482 | } | |
494fff56 | 1483 | list_add_tail_rcu(&c->_c.list, &mrt->mfc_cache_list); |
7bc570c8 | 1484 | |
87c418bf YM |
1485 | /* Check to see if we resolved a queued list. If so we |
1486 | * need to send on the frames and tidy up. | |
7bc570c8 | 1487 | */ |
f30a7784 | 1488 | found = false; |
7bc570c8 | 1489 | spin_lock_bh(&mfc_unres_lock); |
494fff56 YM |
1490 | list_for_each_entry(_uc, &mrt->mfc_unres_queue, list) { |
1491 | uc = (struct mfc6_cache *)_uc; | |
c476efbc | 1492 | if (ipv6_addr_equal(&uc->mf6c_origin, &c->mf6c_origin) && |
7bc570c8 | 1493 | ipv6_addr_equal(&uc->mf6c_mcastgrp, &c->mf6c_mcastgrp)) { |
494fff56 | 1494 | list_del(&_uc->list); |
6bd52143 | 1495 | atomic_dec(&mrt->cache_resolve_queue_len); |
f30a7784 | 1496 | found = true; |
7bc570c8 YH |
1497 | break; |
1498 | } | |
1499 | } | |
b70432f7 | 1500 | if (list_empty(&mrt->mfc_unres_queue)) |
6bd52143 | 1501 | del_timer(&mrt->ipmr_expire_timer); |
7bc570c8 YH |
1502 | spin_unlock_bh(&mfc_unres_lock); |
1503 | ||
f30a7784 | 1504 | if (found) { |
6bd52143 | 1505 | ip6mr_cache_resolve(net, mrt, uc, c); |
58701ad4 | 1506 | ip6mr_cache_free(uc); |
7bc570c8 | 1507 | } |
088aa3ee YM |
1508 | call_ip6mr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_ADD, |
1509 | c, mrt->id); | |
812e44dd | 1510 | mr6_netlink_event(mrt, c, RTM_NEWROUTE); |
7bc570c8 YH |
1511 | return 0; |
1512 | } | |
1513 | ||
1514 | /* | |
1515 | * Close the multicast socket, and clear the vif tables etc | |
1516 | */ | |
1517 | ||
ca8d4794 | 1518 | static void mroute_clean_tables(struct mr_table *mrt, int flags) |
7bc570c8 | 1519 | { |
494fff56 | 1520 | struct mr_mfc *c, *tmp; |
c871e664 | 1521 | LIST_HEAD(list); |
87c418bf | 1522 | int i; |
7bc570c8 | 1523 | |
87c418bf | 1524 | /* Shut down all active vif entries */ |
ca8d4794 CS |
1525 | if (flags & (MRT6_FLUSH_MIFS | MRT6_FLUSH_MIFS_STATIC)) { |
1526 | for (i = 0; i < mrt->maxvif; i++) { | |
1527 | if (((mrt->vif_table[i].flags & VIFF_STATIC) && | |
1528 | !(flags & MRT6_FLUSH_MIFS_STATIC)) || | |
1529 | (!(mrt->vif_table[i].flags & VIFF_STATIC) && !(flags & MRT6_FLUSH_MIFS))) | |
1530 | continue; | |
1531 | mif6_delete(mrt, i, 0, &list); | |
1532 | } | |
1533 | unregister_netdevice_many(&list); | |
7bc570c8 YH |
1534 | } |
1535 | ||
87c418bf | 1536 | /* Wipe the cache */ |
ca8d4794 CS |
1537 | if (flags & (MRT6_FLUSH_MFC | MRT6_FLUSH_MFC_STATIC)) { |
1538 | list_for_each_entry_safe(c, tmp, &mrt->mfc_cache_list, list) { | |
1539 | if (((c->mfc_flags & MFC_STATIC) && !(flags & MRT6_FLUSH_MFC_STATIC)) || | |
1540 | (!(c->mfc_flags & MFC_STATIC) && !(flags & MRT6_FLUSH_MFC))) | |
1541 | continue; | |
1542 | rhltable_remove(&mrt->mfc_hash, &c->mnode, ip6mr_rht_params); | |
1543 | list_del_rcu(&c->list); | |
1544 | call_ip6mr_mfc_entry_notifiers(read_pnet(&mrt->net), | |
1545 | FIB_EVENT_ENTRY_DEL, | |
1546 | (struct mfc6_cache *)c, mrt->id); | |
1547 | mr6_netlink_event(mrt, (struct mfc6_cache *)c, RTM_DELROUTE); | |
1548 | mr_cache_put(c); | |
1549 | } | |
7bc570c8 YH |
1550 | } |
1551 | ||
ca8d4794 CS |
1552 | if (flags & MRT6_FLUSH_MFC) { |
1553 | if (atomic_read(&mrt->cache_resolve_queue_len) != 0) { | |
1554 | spin_lock_bh(&mfc_unres_lock); | |
1555 | list_for_each_entry_safe(c, tmp, &mrt->mfc_unres_queue, list) { | |
1556 | list_del(&c->list); | |
1557 | mr6_netlink_event(mrt, (struct mfc6_cache *)c, | |
1558 | RTM_DELROUTE); | |
1559 | ip6mr_destroy_unres(mrt, (struct mfc6_cache *)c); | |
1560 | } | |
1561 | spin_unlock_bh(&mfc_unres_lock); | |
7bc570c8 | 1562 | } |
7bc570c8 YH |
1563 | } |
1564 | } | |
1565 | ||
b70432f7 | 1566 | static int ip6mr_sk_init(struct mr_table *mrt, struct sock *sk) |
7bc570c8 YH |
1567 | { |
1568 | int err = 0; | |
8229efda | 1569 | struct net *net = sock_net(sk); |
7bc570c8 YH |
1570 | |
1571 | rtnl_lock(); | |
a96f7a6a | 1572 | spin_lock(&mrt_lock); |
b70432f7 | 1573 | if (rtnl_dereference(mrt->mroute_sk)) { |
7bc570c8 | 1574 | err = -EADDRINUSE; |
8571ab47 | 1575 | } else { |
b70432f7 | 1576 | rcu_assign_pointer(mrt->mroute_sk, sk); |
a366e300 | 1577 | sock_set_flag(sk, SOCK_RCU_FREE); |
145c7a79 | 1578 | atomic_inc(&net->ipv6.devconf_all->mc_forwarding); |
927265bc | 1579 | } |
a96f7a6a | 1580 | spin_unlock(&mrt_lock); |
7bc570c8 | 1581 | |
927265bc | 1582 | if (!err) |
85b3daad DA |
1583 | inet6_netconf_notify_devconf(net, RTM_NEWNETCONF, |
1584 | NETCONFA_MC_FORWARDING, | |
927265bc ED |
1585 | NETCONFA_IFINDEX_ALL, |
1586 | net->ipv6.devconf_all); | |
7bc570c8 YH |
1587 | rtnl_unlock(); |
1588 | ||
1589 | return err; | |
1590 | } | |
1591 | ||
1592 | int ip6mr_sk_done(struct sock *sk) | |
1593 | { | |
8229efda | 1594 | struct net *net = sock_net(sk); |
7d9b1b57 | 1595 | struct ipv6_devconf *devconf; |
b70432f7 | 1596 | struct mr_table *mrt; |
7d9b1b57 | 1597 | int err = -EACCES; |
7bc570c8 | 1598 | |
338d182f FR |
1599 | if (sk->sk_type != SOCK_RAW || |
1600 | inet_sk(sk)->inet_num != IPPROTO_ICMPV6) | |
1601 | return err; | |
1602 | ||
7d9b1b57 ED |
1603 | devconf = net->ipv6.devconf_all; |
1604 | if (!devconf || !atomic_read(&devconf->mc_forwarding)) | |
f2f2325e ED |
1605 | return err; |
1606 | ||
7bc570c8 | 1607 | rtnl_lock(); |
d1db275d | 1608 | ip6mr_for_each_table(mrt, net) { |
b70432f7 | 1609 | if (sk == rtnl_dereference(mrt->mroute_sk)) { |
a96f7a6a | 1610 | spin_lock(&mrt_lock); |
b70432f7 | 1611 | RCU_INIT_POINTER(mrt->mroute_sk, NULL); |
a366e300 ED |
1612 | /* Note that mroute_sk had SOCK_RCU_FREE set, |
1613 | * so the RCU grace period before sk freeing | |
1614 | * is guaranteed by sk_destruct() | |
1615 | */ | |
7d9b1b57 | 1616 | atomic_dec(&devconf->mc_forwarding); |
a96f7a6a | 1617 | spin_unlock(&mrt_lock); |
85b3daad | 1618 | inet6_netconf_notify_devconf(net, RTM_NEWNETCONF, |
d67b8c61 ND |
1619 | NETCONFA_MC_FORWARDING, |
1620 | NETCONFA_IFINDEX_ALL, | |
1621 | net->ipv6.devconf_all); | |
7bc570c8 | 1622 | |
ca8d4794 | 1623 | mroute_clean_tables(mrt, MRT6_FLUSH_MIFS | MRT6_FLUSH_MFC); |
d1db275d PM |
1624 | err = 0; |
1625 | break; | |
1626 | } | |
1627 | } | |
7bc570c8 YH |
1628 | rtnl_unlock(); |
1629 | ||
1630 | return err; | |
1631 | } | |
1632 | ||
8571ab47 | 1633 | bool mroute6_is_socket(struct net *net, struct sk_buff *skb) |
6bd52143 | 1634 | { |
b70432f7 | 1635 | struct mr_table *mrt; |
4c9483b2 | 1636 | struct flowi6 fl6 = { |
e374c618 | 1637 | .flowi6_iif = skb->skb_iif ? : LOOPBACK_IFINDEX, |
4c9483b2 DM |
1638 | .flowi6_oif = skb->dev->ifindex, |
1639 | .flowi6_mark = skb->mark, | |
d1db275d PM |
1640 | }; |
1641 | ||
4c9483b2 | 1642 | if (ip6mr_fib_lookup(net, &fl6, &mrt) < 0) |
d1db275d | 1643 | return NULL; |
6bd52143 | 1644 | |
b70432f7 | 1645 | return rcu_access_pointer(mrt->mroute_sk); |
6bd52143 | 1646 | } |
8571ab47 | 1647 | EXPORT_SYMBOL(mroute6_is_socket); |
6bd52143 | 1648 | |
7bc570c8 YH |
1649 | /* |
1650 | * Socket options and virtual interface manipulation. The whole | |
1651 | * virtual interface system is a complete heap, but unfortunately | |
1652 | * that's how BSD mrouted happens to think. Maybe one day with a proper | |
1653 | * MOSPF/PIM router set up we can clean this up. | |
1654 | */ | |
1655 | ||
b43c6153 CH |
1656 | int ip6_mroute_setsockopt(struct sock *sk, int optname, sockptr_t optval, |
1657 | unsigned int optlen) | |
7bc570c8 | 1658 | { |
660b26dc | 1659 | int ret, parent = 0; |
7bc570c8 YH |
1660 | struct mif6ctl vif; |
1661 | struct mf6cctl mfc; | |
1662 | mifi_t mifi; | |
8229efda | 1663 | struct net *net = sock_net(sk); |
b70432f7 | 1664 | struct mr_table *mrt; |
d1db275d | 1665 | |
99253eb7 XL |
1666 | if (sk->sk_type != SOCK_RAW || |
1667 | inet_sk(sk)->inet_num != IPPROTO_ICMPV6) | |
1668 | return -EOPNOTSUPP; | |
1669 | ||
d1db275d | 1670 | mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT); |
63159f29 | 1671 | if (!mrt) |
d1db275d | 1672 | return -ENOENT; |
7bc570c8 YH |
1673 | |
1674 | if (optname != MRT6_INIT) { | |
b70432f7 | 1675 | if (sk != rcu_access_pointer(mrt->mroute_sk) && |
8571ab47 | 1676 | !ns_capable(net->user_ns, CAP_NET_ADMIN)) |
7bc570c8 YH |
1677 | return -EACCES; |
1678 | } | |
1679 | ||
1680 | switch (optname) { | |
1681 | case MRT6_INIT: | |
7bc570c8 YH |
1682 | if (optlen < sizeof(int)) |
1683 | return -EINVAL; | |
1684 | ||
6bd52143 | 1685 | return ip6mr_sk_init(mrt, sk); |
7bc570c8 YH |
1686 | |
1687 | case MRT6_DONE: | |
1688 | return ip6mr_sk_done(sk); | |
1689 | ||
1690 | case MRT6_ADD_MIF: | |
1691 | if (optlen < sizeof(vif)) | |
1692 | return -EINVAL; | |
b43c6153 | 1693 | if (copy_from_sockptr(&vif, optval, sizeof(vif))) |
7bc570c8 | 1694 | return -EFAULT; |
6ac7eb08 | 1695 | if (vif.mif6c_mifi >= MAXMIFS) |
7bc570c8 YH |
1696 | return -ENFILE; |
1697 | rtnl_lock(); | |
8571ab47 | 1698 | ret = mif6_add(net, mrt, &vif, |
b70432f7 | 1699 | sk == rtnl_dereference(mrt->mroute_sk)); |
7bc570c8 YH |
1700 | rtnl_unlock(); |
1701 | return ret; | |
1702 | ||
1703 | case MRT6_DEL_MIF: | |
1704 | if (optlen < sizeof(mifi_t)) | |
1705 | return -EINVAL; | |
b43c6153 | 1706 | if (copy_from_sockptr(&mifi, optval, sizeof(mifi_t))) |
7bc570c8 YH |
1707 | return -EFAULT; |
1708 | rtnl_lock(); | |
723b929c | 1709 | ret = mif6_delete(mrt, mifi, 0, NULL); |
7bc570c8 YH |
1710 | rtnl_unlock(); |
1711 | return ret; | |
1712 | ||
1713 | /* | |
1714 | * Manipulate the forwarding caches. These live | |
1715 | * in a sort of kernel/user symbiosis. | |
1716 | */ | |
1717 | case MRT6_ADD_MFC: | |
1718 | case MRT6_DEL_MFC: | |
660b26dc | 1719 | parent = -1; |
a8eceea8 | 1720 | fallthrough; |
660b26dc ND |
1721 | case MRT6_ADD_MFC_PROXY: |
1722 | case MRT6_DEL_MFC_PROXY: | |
7bc570c8 YH |
1723 | if (optlen < sizeof(mfc)) |
1724 | return -EINVAL; | |
b43c6153 | 1725 | if (copy_from_sockptr(&mfc, optval, sizeof(mfc))) |
7bc570c8 | 1726 | return -EFAULT; |
660b26dc ND |
1727 | if (parent == 0) |
1728 | parent = mfc.mf6cc_parent; | |
7bc570c8 | 1729 | rtnl_lock(); |
660b26dc ND |
1730 | if (optname == MRT6_DEL_MFC || optname == MRT6_DEL_MFC_PROXY) |
1731 | ret = ip6mr_mfc_delete(mrt, &mfc, parent); | |
7bc570c8 | 1732 | else |
660b26dc | 1733 | ret = ip6mr_mfc_add(net, mrt, &mfc, |
8571ab47 | 1734 | sk == |
b70432f7 | 1735 | rtnl_dereference(mrt->mroute_sk), |
8571ab47 | 1736 | parent); |
7bc570c8 YH |
1737 | rtnl_unlock(); |
1738 | return ret; | |
1739 | ||
ca8d4794 CS |
1740 | case MRT6_FLUSH: |
1741 | { | |
1742 | int flags; | |
1743 | ||
1744 | if (optlen != sizeof(flags)) | |
1745 | return -EINVAL; | |
b43c6153 | 1746 | if (copy_from_sockptr(&flags, optval, sizeof(flags))) |
ca8d4794 CS |
1747 | return -EFAULT; |
1748 | rtnl_lock(); | |
1749 | mroute_clean_tables(mrt, flags); | |
1750 | rtnl_unlock(); | |
1751 | return 0; | |
1752 | } | |
1753 | ||
14fb64e1 YH |
1754 | /* |
1755 | * Control PIM assert (to activate pim will activate assert) | |
1756 | */ | |
1757 | case MRT6_ASSERT: | |
1758 | { | |
1759 | int v; | |
03f52a0a JP |
1760 | |
1761 | if (optlen != sizeof(v)) | |
1762 | return -EINVAL; | |
b43c6153 | 1763 | if (copy_from_sockptr(&v, optval, sizeof(v))) |
14fb64e1 | 1764 | return -EFAULT; |
53d6841d | 1765 | mrt->mroute_do_assert = v; |
14fb64e1 YH |
1766 | return 0; |
1767 | } | |
1768 | ||
1769 | #ifdef CONFIG_IPV6_PIMSM_V2 | |
1770 | case MRT6_PIM: | |
1771 | { | |
a3ebe92a | 1772 | bool do_wrmifwhole; |
a9f83bf3 | 1773 | int v; |
03f52a0a JP |
1774 | |
1775 | if (optlen != sizeof(v)) | |
1776 | return -EINVAL; | |
b43c6153 | 1777 | if (copy_from_sockptr(&v, optval, sizeof(v))) |
14fb64e1 | 1778 | return -EFAULT; |
4b340a5a MR |
1779 | |
1780 | do_wrmifwhole = (v == MRT6MSG_WRMIFWHOLE); | |
14fb64e1 YH |
1781 | v = !!v; |
1782 | rtnl_lock(); | |
1783 | ret = 0; | |
6bd52143 PM |
1784 | if (v != mrt->mroute_do_pim) { |
1785 | mrt->mroute_do_pim = v; | |
1786 | mrt->mroute_do_assert = v; | |
4b340a5a | 1787 | mrt->mroute_do_wrvifwhole = do_wrmifwhole; |
14fb64e1 YH |
1788 | } |
1789 | rtnl_unlock(); | |
1790 | return ret; | |
1791 | } | |
1792 | ||
d1db275d PM |
1793 | #endif |
1794 | #ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES | |
1795 | case MRT6_TABLE: | |
1796 | { | |
1797 | u32 v; | |
1798 | ||
1799 | if (optlen != sizeof(u32)) | |
1800 | return -EINVAL; | |
b43c6153 | 1801 | if (copy_from_sockptr(&v, optval, sizeof(v))) |
d1db275d | 1802 | return -EFAULT; |
75356a81 DC |
1803 | /* "pim6reg%u" should not exceed 16 bytes (IFNAMSIZ) */ |
1804 | if (v != RT_TABLE_DEFAULT && v >= 100000000) | |
1805 | return -EINVAL; | |
b70432f7 | 1806 | if (sk == rcu_access_pointer(mrt->mroute_sk)) |
d1db275d PM |
1807 | return -EBUSY; |
1808 | ||
1809 | rtnl_lock(); | |
1810 | ret = 0; | |
e783bb00 SD |
1811 | mrt = ip6mr_new_table(net, v); |
1812 | if (IS_ERR(mrt)) | |
1813 | ret = PTR_ERR(mrt); | |
848235ed SD |
1814 | else |
1815 | raw6_sk(sk)->ip6mr_table = v; | |
d1db275d PM |
1816 | rtnl_unlock(); |
1817 | return ret; | |
1818 | } | |
14fb64e1 | 1819 | #endif |
7bc570c8 | 1820 | /* |
7d120c55 | 1821 | * Spurious command, or MRT6_VERSION which you cannot |
7bc570c8 YH |
1822 | * set. |
1823 | */ | |
1824 | default: | |
1825 | return -ENOPROTOOPT; | |
1826 | } | |
1827 | } | |
1828 | ||
1829 | /* | |
1830 | * Getsock opt support for the multicast routing system. | |
1831 | */ | |
1832 | ||
6dadbe4b MKL |
1833 | int ip6_mroute_getsockopt(struct sock *sk, int optname, sockptr_t optval, |
1834 | sockptr_t optlen) | |
7bc570c8 YH |
1835 | { |
1836 | int olr; | |
1837 | int val; | |
8229efda | 1838 | struct net *net = sock_net(sk); |
b70432f7 | 1839 | struct mr_table *mrt; |
d1db275d | 1840 | |
99253eb7 XL |
1841 | if (sk->sk_type != SOCK_RAW || |
1842 | inet_sk(sk)->inet_num != IPPROTO_ICMPV6) | |
1843 | return -EOPNOTSUPP; | |
1844 | ||
d1db275d | 1845 | mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT); |
63159f29 | 1846 | if (!mrt) |
d1db275d | 1847 | return -ENOENT; |
7bc570c8 YH |
1848 | |
1849 | switch (optname) { | |
1850 | case MRT6_VERSION: | |
1851 | val = 0x0305; | |
1852 | break; | |
14fb64e1 YH |
1853 | #ifdef CONFIG_IPV6_PIMSM_V2 |
1854 | case MRT6_PIM: | |
6bd52143 | 1855 | val = mrt->mroute_do_pim; |
14fb64e1 YH |
1856 | break; |
1857 | #endif | |
1858 | case MRT6_ASSERT: | |
6bd52143 | 1859 | val = mrt->mroute_do_assert; |
14fb64e1 | 1860 | break; |
7bc570c8 YH |
1861 | default: |
1862 | return -ENOPROTOOPT; | |
1863 | } | |
1864 | ||
6dadbe4b | 1865 | if (copy_from_sockptr(&olr, optlen, sizeof(int))) |
7bc570c8 YH |
1866 | return -EFAULT; |
1867 | ||
1868 | olr = min_t(int, olr, sizeof(int)); | |
1869 | if (olr < 0) | |
1870 | return -EINVAL; | |
1871 | ||
6dadbe4b | 1872 | if (copy_to_sockptr(optlen, &olr, sizeof(int))) |
7bc570c8 | 1873 | return -EFAULT; |
6dadbe4b | 1874 | if (copy_to_sockptr(optval, &val, olr)) |
7bc570c8 YH |
1875 | return -EFAULT; |
1876 | return 0; | |
1877 | } | |
1878 | ||
1879 | /* | |
1880 | * The IP multicast ioctl support routines. | |
1881 | */ | |
1882 | ||
1883 | int ip6mr_ioctl(struct sock *sk, int cmd, void __user *arg) | |
1884 | { | |
1885 | struct sioc_sg_req6 sr; | |
1886 | struct sioc_mif_req6 vr; | |
6853f21f | 1887 | struct vif_device *vif; |
7bc570c8 | 1888 | struct mfc6_cache *c; |
8229efda | 1889 | struct net *net = sock_net(sk); |
b70432f7 | 1890 | struct mr_table *mrt; |
d1db275d PM |
1891 | |
1892 | mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT); | |
63159f29 | 1893 | if (!mrt) |
d1db275d | 1894 | return -ENOENT; |
7bc570c8 YH |
1895 | |
1896 | switch (cmd) { | |
1897 | case SIOCGETMIFCNT_IN6: | |
1898 | if (copy_from_user(&vr, arg, sizeof(vr))) | |
1899 | return -EFAULT; | |
6bd52143 | 1900 | if (vr.mifi >= mrt->maxvif) |
7bc570c8 | 1901 | return -EINVAL; |
69d2c867 | 1902 | vr.mifi = array_index_nospec(vr.mifi, mrt->maxvif); |
638cf4a2 | 1903 | rcu_read_lock(); |
b70432f7 YM |
1904 | vif = &mrt->vif_table[vr.mifi]; |
1905 | if (VIF_EXISTS(mrt, vr.mifi)) { | |
638cf4a2 ED |
1906 | vr.icount = READ_ONCE(vif->pkt_in); |
1907 | vr.ocount = READ_ONCE(vif->pkt_out); | |
1908 | vr.ibytes = READ_ONCE(vif->bytes_in); | |
1909 | vr.obytes = READ_ONCE(vif->bytes_out); | |
1910 | rcu_read_unlock(); | |
7bc570c8 YH |
1911 | |
1912 | if (copy_to_user(arg, &vr, sizeof(vr))) | |
1913 | return -EFAULT; | |
1914 | return 0; | |
1915 | } | |
638cf4a2 | 1916 | rcu_read_unlock(); |
7bc570c8 YH |
1917 | return -EADDRNOTAVAIL; |
1918 | case SIOCGETSGCNT_IN6: | |
1919 | if (copy_from_user(&sr, arg, sizeof(sr))) | |
1920 | return -EFAULT; | |
1921 | ||
87c418bf | 1922 | rcu_read_lock(); |
6bd52143 | 1923 | c = ip6mr_cache_find(mrt, &sr.src.sin6_addr, &sr.grp.sin6_addr); |
7bc570c8 | 1924 | if (c) { |
494fff56 YM |
1925 | sr.pktcnt = c->_c.mfc_un.res.pkt; |
1926 | sr.bytecnt = c->_c.mfc_un.res.bytes; | |
1927 | sr.wrong_if = c->_c.mfc_un.res.wrong_if; | |
87c418bf | 1928 | rcu_read_unlock(); |
7bc570c8 YH |
1929 | |
1930 | if (copy_to_user(arg, &sr, sizeof(sr))) | |
1931 | return -EFAULT; | |
1932 | return 0; | |
1933 | } | |
87c418bf | 1934 | rcu_read_unlock(); |
7bc570c8 YH |
1935 | return -EADDRNOTAVAIL; |
1936 | default: | |
1937 | return -ENOIOCTLCMD; | |
1938 | } | |
1939 | } | |
1940 | ||
e2d57766 DM |
1941 | #ifdef CONFIG_COMPAT |
1942 | struct compat_sioc_sg_req6 { | |
1943 | struct sockaddr_in6 src; | |
1944 | struct sockaddr_in6 grp; | |
1945 | compat_ulong_t pktcnt; | |
1946 | compat_ulong_t bytecnt; | |
1947 | compat_ulong_t wrong_if; | |
1948 | }; | |
1949 | ||
1950 | struct compat_sioc_mif_req6 { | |
1951 | mifi_t mifi; | |
1952 | compat_ulong_t icount; | |
1953 | compat_ulong_t ocount; | |
1954 | compat_ulong_t ibytes; | |
1955 | compat_ulong_t obytes; | |
1956 | }; | |
1957 | ||
1958 | int ip6mr_compat_ioctl(struct sock *sk, unsigned int cmd, void __user *arg) | |
1959 | { | |
1960 | struct compat_sioc_sg_req6 sr; | |
1961 | struct compat_sioc_mif_req6 vr; | |
6853f21f | 1962 | struct vif_device *vif; |
e2d57766 DM |
1963 | struct mfc6_cache *c; |
1964 | struct net *net = sock_net(sk); | |
b70432f7 | 1965 | struct mr_table *mrt; |
e2d57766 DM |
1966 | |
1967 | mrt = ip6mr_get_table(net, raw6_sk(sk)->ip6mr_table ? : RT6_TABLE_DFLT); | |
63159f29 | 1968 | if (!mrt) |
e2d57766 DM |
1969 | return -ENOENT; |
1970 | ||
1971 | switch (cmd) { | |
1972 | case SIOCGETMIFCNT_IN6: | |
1973 | if (copy_from_user(&vr, arg, sizeof(vr))) | |
1974 | return -EFAULT; | |
1975 | if (vr.mifi >= mrt->maxvif) | |
1976 | return -EINVAL; | |
69d2c867 | 1977 | vr.mifi = array_index_nospec(vr.mifi, mrt->maxvif); |
638cf4a2 | 1978 | rcu_read_lock(); |
b70432f7 YM |
1979 | vif = &mrt->vif_table[vr.mifi]; |
1980 | if (VIF_EXISTS(mrt, vr.mifi)) { | |
638cf4a2 ED |
1981 | vr.icount = READ_ONCE(vif->pkt_in); |
1982 | vr.ocount = READ_ONCE(vif->pkt_out); | |
1983 | vr.ibytes = READ_ONCE(vif->bytes_in); | |
1984 | vr.obytes = READ_ONCE(vif->bytes_out); | |
1985 | rcu_read_unlock(); | |
e2d57766 DM |
1986 | |
1987 | if (copy_to_user(arg, &vr, sizeof(vr))) | |
1988 | return -EFAULT; | |
1989 | return 0; | |
1990 | } | |
638cf4a2 | 1991 | rcu_read_unlock(); |
e2d57766 DM |
1992 | return -EADDRNOTAVAIL; |
1993 | case SIOCGETSGCNT_IN6: | |
1994 | if (copy_from_user(&sr, arg, sizeof(sr))) | |
1995 | return -EFAULT; | |
1996 | ||
87c418bf | 1997 | rcu_read_lock(); |
e2d57766 DM |
1998 | c = ip6mr_cache_find(mrt, &sr.src.sin6_addr, &sr.grp.sin6_addr); |
1999 | if (c) { | |
494fff56 YM |
2000 | sr.pktcnt = c->_c.mfc_un.res.pkt; |
2001 | sr.bytecnt = c->_c.mfc_un.res.bytes; | |
2002 | sr.wrong_if = c->_c.mfc_un.res.wrong_if; | |
87c418bf | 2003 | rcu_read_unlock(); |
e2d57766 DM |
2004 | |
2005 | if (copy_to_user(arg, &sr, sizeof(sr))) | |
2006 | return -EFAULT; | |
2007 | return 0; | |
2008 | } | |
87c418bf | 2009 | rcu_read_unlock(); |
e2d57766 DM |
2010 | return -EADDRNOTAVAIL; |
2011 | default: | |
2012 | return -ENOIOCTLCMD; | |
2013 | } | |
2014 | } | |
2015 | #endif | |
7bc570c8 | 2016 | |
0c4b51f0 | 2017 | static inline int ip6mr_forward2_finish(struct net *net, struct sock *sk, struct sk_buff *skb) |
7bc570c8 | 2018 | { |
87c11f1d IS |
2019 | IP6_INC_STATS(net, ip6_dst_idev(skb_dst(skb)), |
2020 | IPSTATS_MIB_OUTFORWDATAGRAMS); | |
2021 | IP6_ADD_STATS(net, ip6_dst_idev(skb_dst(skb)), | |
2022 | IPSTATS_MIB_OUTOCTETS, skb->len); | |
13206b6b | 2023 | return dst_output(net, sk, skb); |
7bc570c8 YH |
2024 | } |
2025 | ||
2026 | /* | |
2027 | * Processing handlers for ip6mr_forward | |
2028 | */ | |
2029 | ||
b70432f7 | 2030 | static int ip6mr_forward2(struct net *net, struct mr_table *mrt, |
f5c6dfde | 2031 | struct sk_buff *skb, int vifi) |
7bc570c8 | 2032 | { |
b70432f7 | 2033 | struct vif_device *vif = &mrt->vif_table[vifi]; |
ebc31979 ED |
2034 | struct net_device *vif_dev; |
2035 | struct ipv6hdr *ipv6h; | |
7bc570c8 | 2036 | struct dst_entry *dst; |
4c9483b2 | 2037 | struct flowi6 fl6; |
7bc570c8 | 2038 | |
ebc31979 ED |
2039 | vif_dev = vif_dev_read(vif); |
2040 | if (!vif_dev) | |
7bc570c8 YH |
2041 | goto out_free; |
2042 | ||
14fb64e1 YH |
2043 | #ifdef CONFIG_IPV6_PIMSM_V2 |
2044 | if (vif->flags & MIFF_REGISTER) { | |
638cf4a2 ED |
2045 | WRITE_ONCE(vif->pkt_out, vif->pkt_out + 1); |
2046 | WRITE_ONCE(vif->bytes_out, vif->bytes_out + skb->len); | |
2fad1ba3 ED |
2047 | DEV_STATS_ADD(vif_dev, tx_bytes, skb->len); |
2048 | DEV_STATS_INC(vif_dev, tx_packets); | |
6bd52143 | 2049 | ip6mr_cache_report(mrt, skb, vifi, MRT6MSG_WHOLEPKT); |
8da73b73 | 2050 | goto out_free; |
14fb64e1 YH |
2051 | } |
2052 | #endif | |
2053 | ||
7bc570c8 YH |
2054 | ipv6h = ipv6_hdr(skb); |
2055 | ||
4c9483b2 DM |
2056 | fl6 = (struct flowi6) { |
2057 | .flowi6_oif = vif->link, | |
2058 | .daddr = ipv6h->daddr, | |
7bc570c8 YH |
2059 | }; |
2060 | ||
4c9483b2 | 2061 | dst = ip6_route_output(net, NULL, &fl6); |
5095d64d RL |
2062 | if (dst->error) { |
2063 | dst_release(dst); | |
7bc570c8 | 2064 | goto out_free; |
5095d64d | 2065 | } |
7bc570c8 | 2066 | |
adf30907 ED |
2067 | skb_dst_drop(skb); |
2068 | skb_dst_set(skb, dst); | |
7bc570c8 YH |
2069 | |
2070 | /* | |
2071 | * RFC1584 teaches, that DVMRP/PIM router must deliver packets locally | |
2072 | * not only before forwarding, but after forwarding on all output | |
2073 | * interfaces. It is clear, if mrouter runs a multicasting | |
2074 | * program, it should receive packets not depending to what interface | |
2075 | * program is joined. | |
2076 | * If we will not make it, the program will have to join on all | |
2077 | * interfaces. On the other hand, multihoming host (or router, but | |
2078 | * not mrouter) cannot join to more than one interface - it will | |
2079 | * result in receiving multiple packets. | |
2080 | */ | |
ebc31979 | 2081 | skb->dev = vif_dev; |
638cf4a2 ED |
2082 | WRITE_ONCE(vif->pkt_out, vif->pkt_out + 1); |
2083 | WRITE_ONCE(vif->bytes_out, vif->bytes_out + skb->len); | |
7bc570c8 YH |
2084 | |
2085 | /* We are about to write */ | |
2086 | /* XXX: extension headers? */ | |
ebc31979 | 2087 | if (skb_cow(skb, sizeof(*ipv6h) + LL_RESERVED_SPACE(vif_dev))) |
7bc570c8 YH |
2088 | goto out_free; |
2089 | ||
2090 | ipv6h = ipv6_hdr(skb); | |
2091 | ipv6h->hop_limit--; | |
2092 | ||
2093 | IP6CB(skb)->flags |= IP6SKB_FORWARDED; | |
2094 | ||
29a26a56 | 2095 | return NF_HOOK(NFPROTO_IPV6, NF_INET_FORWARD, |
ebc31979 | 2096 | net, NULL, skb, skb->dev, vif_dev, |
7bc570c8 YH |
2097 | ip6mr_forward2_finish); |
2098 | ||
2099 | out_free: | |
2100 | kfree_skb(skb); | |
2101 | return 0; | |
2102 | } | |
2103 | ||
a96f7a6a | 2104 | /* Called with rcu_read_lock() */ |
b70432f7 | 2105 | static int ip6mr_find_vif(struct mr_table *mrt, struct net_device *dev) |
7bc570c8 YH |
2106 | { |
2107 | int ct; | |
6bd52143 | 2108 | |
db9eb7c8 ED |
2109 | /* Pairs with WRITE_ONCE() in mif6_delete()/mif6_add() */ |
2110 | for (ct = READ_ONCE(mrt->maxvif) - 1; ct >= 0; ct--) { | |
ebc31979 | 2111 | if (rcu_access_pointer(mrt->vif_table[ct].dev) == dev) |
7bc570c8 YH |
2112 | break; |
2113 | } | |
2114 | return ct; | |
2115 | } | |
2116 | ||
9b1c21d8 | 2117 | /* Called under rcu_read_lock() */ |
b70432f7 | 2118 | static void ip6_mr_forward(struct net *net, struct mr_table *mrt, |
e4a38c0c PR |
2119 | struct net_device *dev, struct sk_buff *skb, |
2120 | struct mfc6_cache *c) | |
7bc570c8 YH |
2121 | { |
2122 | int psend = -1; | |
2123 | int vif, ct; | |
e4a38c0c | 2124 | int true_vifi = ip6mr_find_vif(mrt, dev); |
7bc570c8 | 2125 | |
494fff56 YM |
2126 | vif = c->_c.mfc_parent; |
2127 | c->_c.mfc_un.res.pkt++; | |
2128 | c->_c.mfc_un.res.bytes += skb->len; | |
2129 | c->_c.mfc_un.res.lastuse = jiffies; | |
7bc570c8 | 2130 | |
494fff56 | 2131 | if (ipv6_addr_any(&c->mf6c_origin) && true_vifi >= 0) { |
660b26dc ND |
2132 | struct mfc6_cache *cache_proxy; |
2133 | ||
40dc2ca3 | 2134 | /* For an (*,G) entry, we only check that the incoming |
660b26dc ND |
2135 | * interface is part of the static tree. |
2136 | */ | |
845c9a7a | 2137 | cache_proxy = mr_mfc_find_any_parent(mrt, vif); |
660b26dc | 2138 | if (cache_proxy && |
a7e555d4 | 2139 | cache_proxy->_c.mfc_un.res.ttls[true_vifi] < 255) |
660b26dc ND |
2140 | goto forward; |
2141 | } | |
2142 | ||
14fb64e1 YH |
2143 | /* |
2144 | * Wrong interface: drop packet and (maybe) send PIM assert. | |
2145 | */ | |
ebc31979 | 2146 | if (rcu_access_pointer(mrt->vif_table[vif].dev) != dev) { |
494fff56 | 2147 | c->_c.mfc_un.res.wrong_if++; |
14fb64e1 | 2148 | |
6bd52143 | 2149 | if (true_vifi >= 0 && mrt->mroute_do_assert && |
14fb64e1 YH |
2150 | /* pimsm uses asserts, when switching from RPT to SPT, |
2151 | so that we cannot check that packet arrived on an oif. | |
2152 | It is bad, but otherwise we would need to move pretty | |
2153 | large chunk of pimd to kernel. Ough... --ANK | |
2154 | */ | |
6bd52143 | 2155 | (mrt->mroute_do_pim || |
494fff56 | 2156 | c->_c.mfc_un.res.ttls[true_vifi] < 255) && |
14fb64e1 | 2157 | time_after(jiffies, |
494fff56 YM |
2158 | c->_c.mfc_un.res.last_assert + |
2159 | MFC_ASSERT_THRESH)) { | |
2160 | c->_c.mfc_un.res.last_assert = jiffies; | |
6bd52143 | 2161 | ip6mr_cache_report(mrt, skb, true_vifi, MRT6MSG_WRONGMIF); |
4b340a5a MR |
2162 | if (mrt->mroute_do_wrvifwhole) |
2163 | ip6mr_cache_report(mrt, skb, true_vifi, | |
2164 | MRT6MSG_WRMIFWHOLE); | |
14fb64e1 YH |
2165 | } |
2166 | goto dont_forward; | |
2167 | } | |
2168 | ||
660b26dc | 2169 | forward: |
638cf4a2 ED |
2170 | WRITE_ONCE(mrt->vif_table[vif].pkt_in, |
2171 | mrt->vif_table[vif].pkt_in + 1); | |
2172 | WRITE_ONCE(mrt->vif_table[vif].bytes_in, | |
2173 | mrt->vif_table[vif].bytes_in + skb->len); | |
7bc570c8 YH |
2174 | |
2175 | /* | |
2176 | * Forward the frame | |
2177 | */ | |
494fff56 YM |
2178 | if (ipv6_addr_any(&c->mf6c_origin) && |
2179 | ipv6_addr_any(&c->mf6c_mcastgrp)) { | |
660b26dc | 2180 | if (true_vifi >= 0 && |
494fff56 | 2181 | true_vifi != c->_c.mfc_parent && |
660b26dc | 2182 | ipv6_hdr(skb)->hop_limit > |
494fff56 | 2183 | c->_c.mfc_un.res.ttls[c->_c.mfc_parent]) { |
660b26dc ND |
2184 | /* It's an (*,*) entry and the packet is not coming from |
2185 | * the upstream: forward the packet to the upstream | |
2186 | * only. | |
2187 | */ | |
494fff56 | 2188 | psend = c->_c.mfc_parent; |
660b26dc ND |
2189 | goto last_forward; |
2190 | } | |
2191 | goto dont_forward; | |
2192 | } | |
494fff56 YM |
2193 | for (ct = c->_c.mfc_un.res.maxvif - 1; |
2194 | ct >= c->_c.mfc_un.res.minvif; ct--) { | |
660b26dc | 2195 | /* For (*,G) entry, don't forward to the incoming interface */ |
494fff56 YM |
2196 | if ((!ipv6_addr_any(&c->mf6c_origin) || ct != true_vifi) && |
2197 | ipv6_hdr(skb)->hop_limit > c->_c.mfc_un.res.ttls[ct]) { | |
7bc570c8 YH |
2198 | if (psend != -1) { |
2199 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
2200 | if (skb2) | |
f5c6dfde | 2201 | ip6mr_forward2(net, mrt, skb2, psend); |
7bc570c8 YH |
2202 | } |
2203 | psend = ct; | |
2204 | } | |
2205 | } | |
660b26dc | 2206 | last_forward: |
7bc570c8 | 2207 | if (psend != -1) { |
f5c6dfde | 2208 | ip6mr_forward2(net, mrt, skb, psend); |
2b52c3ad | 2209 | return; |
7bc570c8 YH |
2210 | } |
2211 | ||
14fb64e1 | 2212 | dont_forward: |
7bc570c8 | 2213 | kfree_skb(skb); |
7bc570c8 YH |
2214 | } |
2215 | ||
2216 | ||
2217 | /* | |
2218 | * Multicast packets for forwarding arrive here | |
2219 | */ | |
2220 | ||
2221 | int ip6_mr_input(struct sk_buff *skb) | |
2222 | { | |
2223 | struct mfc6_cache *cache; | |
8229efda | 2224 | struct net *net = dev_net(skb->dev); |
b70432f7 | 2225 | struct mr_table *mrt; |
4c9483b2 DM |
2226 | struct flowi6 fl6 = { |
2227 | .flowi6_iif = skb->dev->ifindex, | |
2228 | .flowi6_mark = skb->mark, | |
d1db275d PM |
2229 | }; |
2230 | int err; | |
e4a38c0c PR |
2231 | struct net_device *dev; |
2232 | ||
2233 | /* skb->dev passed in is the master dev for vrfs. | |
2234 | * Get the proper interface that does have a vif associated with it. | |
2235 | */ | |
2236 | dev = skb->dev; | |
2237 | if (netif_is_l3_master(skb->dev)) { | |
2238 | dev = dev_get_by_index_rcu(net, IPCB(skb)->iif); | |
2239 | if (!dev) { | |
2240 | kfree_skb(skb); | |
2241 | return -ENODEV; | |
2242 | } | |
2243 | } | |
d1db275d | 2244 | |
4c9483b2 | 2245 | err = ip6mr_fib_lookup(net, &fl6, &mrt); |
2015de5f BG |
2246 | if (err < 0) { |
2247 | kfree_skb(skb); | |
d1db275d | 2248 | return err; |
2015de5f | 2249 | } |
7bc570c8 | 2250 | |
6bd52143 | 2251 | cache = ip6mr_cache_find(mrt, |
8229efda | 2252 | &ipv6_hdr(skb)->saddr, &ipv6_hdr(skb)->daddr); |
63159f29 | 2253 | if (!cache) { |
e4a38c0c | 2254 | int vif = ip6mr_find_vif(mrt, dev); |
660b26dc ND |
2255 | |
2256 | if (vif >= 0) | |
2257 | cache = ip6mr_cache_find_any(mrt, | |
2258 | &ipv6_hdr(skb)->daddr, | |
2259 | vif); | |
2260 | } | |
7bc570c8 YH |
2261 | |
2262 | /* | |
2263 | * No usable cache entry | |
2264 | */ | |
63159f29 | 2265 | if (!cache) { |
7bc570c8 YH |
2266 | int vif; |
2267 | ||
e4a38c0c | 2268 | vif = ip6mr_find_vif(mrt, dev); |
7bc570c8 | 2269 | if (vif >= 0) { |
e4a38c0c | 2270 | int err = ip6mr_cache_unresolved(mrt, vif, skb, dev); |
7bc570c8 YH |
2271 | |
2272 | return err; | |
2273 | } | |
7bc570c8 YH |
2274 | kfree_skb(skb); |
2275 | return -ENODEV; | |
2276 | } | |
2277 | ||
e4a38c0c | 2278 | ip6_mr_forward(net, mrt, dev, skb, cache); |
7bc570c8 | 2279 | |
7bc570c8 YH |
2280 | return 0; |
2281 | } | |
2282 | ||
2cf75070 | 2283 | int ip6mr_get_route(struct net *net, struct sk_buff *skb, struct rtmsg *rtm, |
fd61c6ba | 2284 | u32 portid) |
7bc570c8 YH |
2285 | { |
2286 | int err; | |
b70432f7 | 2287 | struct mr_table *mrt; |
7bc570c8 | 2288 | struct mfc6_cache *cache; |
adf30907 | 2289 | struct rt6_info *rt = (struct rt6_info *)skb_dst(skb); |
7bc570c8 | 2290 | |
d1db275d | 2291 | mrt = ip6mr_get_table(net, RT6_TABLE_DFLT); |
63159f29 | 2292 | if (!mrt) |
d1db275d PM |
2293 | return -ENOENT; |
2294 | ||
6fa40a29 | 2295 | rcu_read_lock(); |
6bd52143 | 2296 | cache = ip6mr_cache_find(mrt, &rt->rt6i_src.addr, &rt->rt6i_dst.addr); |
660b26dc ND |
2297 | if (!cache && skb->dev) { |
2298 | int vif = ip6mr_find_vif(mrt, skb->dev); | |
2299 | ||
2300 | if (vif >= 0) | |
2301 | cache = ip6mr_cache_find_any(mrt, &rt->rt6i_dst.addr, | |
2302 | vif); | |
2303 | } | |
7bc570c8 YH |
2304 | |
2305 | if (!cache) { | |
2306 | struct sk_buff *skb2; | |
2307 | struct ipv6hdr *iph; | |
2308 | struct net_device *dev; | |
2309 | int vif; | |
2310 | ||
7bc570c8 | 2311 | dev = skb->dev; |
63159f29 | 2312 | if (!dev || (vif = ip6mr_find_vif(mrt, dev)) < 0) { |
6fa40a29 | 2313 | rcu_read_unlock(); |
7bc570c8 YH |
2314 | return -ENODEV; |
2315 | } | |
2316 | ||
2317 | /* really correct? */ | |
2318 | skb2 = alloc_skb(sizeof(struct ipv6hdr), GFP_ATOMIC); | |
2319 | if (!skb2) { | |
6fa40a29 | 2320 | rcu_read_unlock(); |
7bc570c8 YH |
2321 | return -ENOMEM; |
2322 | } | |
2323 | ||
2cf75070 | 2324 | NETLINK_CB(skb2).portid = portid; |
7bc570c8 YH |
2325 | skb_reset_transport_header(skb2); |
2326 | ||
2327 | skb_put(skb2, sizeof(struct ipv6hdr)); | |
2328 | skb_reset_network_header(skb2); | |
2329 | ||
2330 | iph = ipv6_hdr(skb2); | |
2331 | iph->version = 0; | |
2332 | iph->priority = 0; | |
2333 | iph->flow_lbl[0] = 0; | |
2334 | iph->flow_lbl[1] = 0; | |
2335 | iph->flow_lbl[2] = 0; | |
2336 | iph->payload_len = 0; | |
2337 | iph->nexthdr = IPPROTO_NONE; | |
2338 | iph->hop_limit = 0; | |
4e3fd7a0 AD |
2339 | iph->saddr = rt->rt6i_src.addr; |
2340 | iph->daddr = rt->rt6i_dst.addr; | |
7bc570c8 | 2341 | |
e4a38c0c | 2342 | err = ip6mr_cache_unresolved(mrt, vif, skb2, dev); |
6fa40a29 | 2343 | rcu_read_unlock(); |
7bc570c8 YH |
2344 | |
2345 | return err; | |
2346 | } | |
2347 | ||
7b0db857 | 2348 | err = mr_fill_mroute(mrt, skb, &cache->_c, rtm); |
6fa40a29 | 2349 | rcu_read_unlock(); |
7bc570c8 YH |
2350 | return err; |
2351 | } | |
2352 | ||
b70432f7 | 2353 | static int ip6mr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
f518338b ND |
2354 | u32 portid, u32 seq, struct mfc6_cache *c, int cmd, |
2355 | int flags) | |
5b285cac PM |
2356 | { |
2357 | struct nlmsghdr *nlh; | |
2358 | struct rtmsg *rtm; | |
1eb99af5 | 2359 | int err; |
5b285cac | 2360 | |
f518338b | 2361 | nlh = nlmsg_put(skb, portid, seq, cmd, sizeof(*rtm), flags); |
63159f29 | 2362 | if (!nlh) |
5b285cac PM |
2363 | return -EMSGSIZE; |
2364 | ||
2365 | rtm = nlmsg_data(nlh); | |
193c1e47 | 2366 | rtm->rtm_family = RTNL_FAMILY_IP6MR; |
5b285cac PM |
2367 | rtm->rtm_dst_len = 128; |
2368 | rtm->rtm_src_len = 128; | |
2369 | rtm->rtm_tos = 0; | |
2370 | rtm->rtm_table = mrt->id; | |
c78679e8 DM |
2371 | if (nla_put_u32(skb, RTA_TABLE, mrt->id)) |
2372 | goto nla_put_failure; | |
1eb99af5 | 2373 | rtm->rtm_type = RTN_MULTICAST; |
5b285cac | 2374 | rtm->rtm_scope = RT_SCOPE_UNIVERSE; |
494fff56 | 2375 | if (c->_c.mfc_flags & MFC_STATIC) |
9a68ac72 ND |
2376 | rtm->rtm_protocol = RTPROT_STATIC; |
2377 | else | |
2378 | rtm->rtm_protocol = RTPROT_MROUTED; | |
5b285cac PM |
2379 | rtm->rtm_flags = 0; |
2380 | ||
930345ea JB |
2381 | if (nla_put_in6_addr(skb, RTA_SRC, &c->mf6c_origin) || |
2382 | nla_put_in6_addr(skb, RTA_DST, &c->mf6c_mcastgrp)) | |
c78679e8 | 2383 | goto nla_put_failure; |
7b0db857 | 2384 | err = mr_fill_mroute(mrt, skb, &c->_c, rtm); |
1eb99af5 ND |
2385 | /* do not break the dump if cache is unresolved */ |
2386 | if (err < 0 && err != -ENOENT) | |
5b285cac PM |
2387 | goto nla_put_failure; |
2388 | ||
053c095a JB |
2389 | nlmsg_end(skb, nlh); |
2390 | return 0; | |
5b285cac PM |
2391 | |
2392 | nla_put_failure: | |
2393 | nlmsg_cancel(skb, nlh); | |
2394 | return -EMSGSIZE; | |
2395 | } | |
2396 | ||
7b0db857 YM |
2397 | static int _ip6mr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
2398 | u32 portid, u32 seq, struct mr_mfc *c, | |
2399 | int cmd, int flags) | |
2400 | { | |
2401 | return ip6mr_fill_mroute(mrt, skb, portid, seq, (struct mfc6_cache *)c, | |
2402 | cmd, flags); | |
2403 | } | |
2404 | ||
812e44dd ND |
2405 | static int mr6_msgsize(bool unresolved, int maxvif) |
2406 | { | |
2407 | size_t len = | |
2408 | NLMSG_ALIGN(sizeof(struct rtmsg)) | |
2409 | + nla_total_size(4) /* RTA_TABLE */ | |
2410 | + nla_total_size(sizeof(struct in6_addr)) /* RTA_SRC */ | |
2411 | + nla_total_size(sizeof(struct in6_addr)) /* RTA_DST */ | |
2412 | ; | |
2413 | ||
2414 | if (!unresolved) | |
2415 | len = len | |
2416 | + nla_total_size(4) /* RTA_IIF */ | |
2417 | + nla_total_size(0) /* RTA_MULTIPATH */ | |
2418 | + maxvif * NLA_ALIGN(sizeof(struct rtnexthop)) | |
2419 | /* RTA_MFC_STATS */ | |
3d6b66c1 | 2420 | + nla_total_size_64bit(sizeof(struct rta_mfc_stats)) |
812e44dd ND |
2421 | ; |
2422 | ||
2423 | return len; | |
2424 | } | |
2425 | ||
b70432f7 | 2426 | static void mr6_netlink_event(struct mr_table *mrt, struct mfc6_cache *mfc, |
812e44dd ND |
2427 | int cmd) |
2428 | { | |
2429 | struct net *net = read_pnet(&mrt->net); | |
2430 | struct sk_buff *skb; | |
2431 | int err = -ENOBUFS; | |
2432 | ||
494fff56 | 2433 | skb = nlmsg_new(mr6_msgsize(mfc->_c.mfc_parent >= MAXMIFS, mrt->maxvif), |
812e44dd | 2434 | GFP_ATOMIC); |
63159f29 | 2435 | if (!skb) |
812e44dd ND |
2436 | goto errout; |
2437 | ||
f518338b | 2438 | err = ip6mr_fill_mroute(mrt, skb, 0, 0, mfc, cmd, 0); |
812e44dd ND |
2439 | if (err < 0) |
2440 | goto errout; | |
2441 | ||
2442 | rtnl_notify(skb, net, 0, RTNLGRP_IPV6_MROUTE, NULL, GFP_ATOMIC); | |
2443 | return; | |
2444 | ||
2445 | errout: | |
2446 | kfree_skb(skb); | |
2447 | if (err < 0) | |
2448 | rtnl_set_sk_err(net, RTNLGRP_IPV6_MROUTE, err); | |
2449 | } | |
2450 | ||
dd12d15c JG |
2451 | static size_t mrt6msg_netlink_msgsize(size_t payloadlen) |
2452 | { | |
2453 | size_t len = | |
2454 | NLMSG_ALIGN(sizeof(struct rtgenmsg)) | |
2455 | + nla_total_size(1) /* IP6MRA_CREPORT_MSGTYPE */ | |
2456 | + nla_total_size(4) /* IP6MRA_CREPORT_MIF_ID */ | |
2457 | /* IP6MRA_CREPORT_SRC_ADDR */ | |
2458 | + nla_total_size(sizeof(struct in6_addr)) | |
2459 | /* IP6MRA_CREPORT_DST_ADDR */ | |
2460 | + nla_total_size(sizeof(struct in6_addr)) | |
2461 | /* IP6MRA_CREPORT_PKT */ | |
2462 | + nla_total_size(payloadlen) | |
2463 | ; | |
2464 | ||
2465 | return len; | |
2466 | } | |
2467 | ||
3493a5b7 | 2468 | static void mrt6msg_netlink_event(const struct mr_table *mrt, struct sk_buff *pkt) |
dd12d15c JG |
2469 | { |
2470 | struct net *net = read_pnet(&mrt->net); | |
2471 | struct nlmsghdr *nlh; | |
2472 | struct rtgenmsg *rtgenm; | |
2473 | struct mrt6msg *msg; | |
2474 | struct sk_buff *skb; | |
2475 | struct nlattr *nla; | |
2476 | int payloadlen; | |
2477 | ||
2478 | payloadlen = pkt->len - sizeof(struct mrt6msg); | |
2479 | msg = (struct mrt6msg *)skb_transport_header(pkt); | |
2480 | ||
2481 | skb = nlmsg_new(mrt6msg_netlink_msgsize(payloadlen), GFP_ATOMIC); | |
2482 | if (!skb) | |
2483 | goto errout; | |
2484 | ||
2485 | nlh = nlmsg_put(skb, 0, 0, RTM_NEWCACHEREPORT, | |
2486 | sizeof(struct rtgenmsg), 0); | |
2487 | if (!nlh) | |
2488 | goto errout; | |
2489 | rtgenm = nlmsg_data(nlh); | |
2490 | rtgenm->rtgen_family = RTNL_FAMILY_IP6MR; | |
2491 | if (nla_put_u8(skb, IP6MRA_CREPORT_MSGTYPE, msg->im6_msgtype) || | |
2492 | nla_put_u32(skb, IP6MRA_CREPORT_MIF_ID, msg->im6_mif) || | |
2493 | nla_put_in6_addr(skb, IP6MRA_CREPORT_SRC_ADDR, | |
2494 | &msg->im6_src) || | |
2495 | nla_put_in6_addr(skb, IP6MRA_CREPORT_DST_ADDR, | |
2496 | &msg->im6_dst)) | |
2497 | goto nla_put_failure; | |
2498 | ||
2499 | nla = nla_reserve(skb, IP6MRA_CREPORT_PKT, payloadlen); | |
2500 | if (!nla || skb_copy_bits(pkt, sizeof(struct mrt6msg), | |
2501 | nla_data(nla), payloadlen)) | |
2502 | goto nla_put_failure; | |
2503 | ||
2504 | nlmsg_end(skb, nlh); | |
2505 | ||
2506 | rtnl_notify(skb, net, 0, RTNLGRP_IPV6_MROUTE_R, NULL, GFP_ATOMIC); | |
2507 | return; | |
2508 | ||
2509 | nla_put_failure: | |
2510 | nlmsg_cancel(skb, nlh); | |
2511 | errout: | |
2512 | kfree_skb(skb); | |
2513 | rtnl_set_sk_err(net, RTNLGRP_IPV6_MROUTE_R, -ENOBUFS); | |
2514 | } | |
2515 | ||
d7c31cbd DL |
2516 | static const struct nla_policy ip6mr_getroute_policy[RTA_MAX + 1] = { |
2517 | [RTA_SRC] = NLA_POLICY_EXACT_LEN(sizeof(struct in6_addr)), | |
2518 | [RTA_DST] = NLA_POLICY_EXACT_LEN(sizeof(struct in6_addr)), | |
2519 | [RTA_TABLE] = { .type = NLA_U32 }, | |
2520 | }; | |
2521 | ||
2522 | static int ip6mr_rtm_valid_getroute_req(struct sk_buff *skb, | |
2523 | const struct nlmsghdr *nlh, | |
2524 | struct nlattr **tb, | |
2525 | struct netlink_ext_ack *extack) | |
2526 | { | |
2527 | struct rtmsg *rtm; | |
2528 | int err; | |
2529 | ||
2530 | err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, ip6mr_getroute_policy, | |
2531 | extack); | |
2532 | if (err) | |
2533 | return err; | |
2534 | ||
2535 | rtm = nlmsg_data(nlh); | |
2536 | if ((rtm->rtm_src_len && rtm->rtm_src_len != 128) || | |
2537 | (rtm->rtm_dst_len && rtm->rtm_dst_len != 128) || | |
2538 | rtm->rtm_tos || rtm->rtm_table || rtm->rtm_protocol || | |
2539 | rtm->rtm_scope || rtm->rtm_type || rtm->rtm_flags) { | |
2540 | NL_SET_ERR_MSG_MOD(extack, | |
2541 | "Invalid values in header for multicast route get request"); | |
2542 | return -EINVAL; | |
2543 | } | |
2544 | ||
2545 | if ((tb[RTA_SRC] && !rtm->rtm_src_len) || | |
2546 | (tb[RTA_DST] && !rtm->rtm_dst_len)) { | |
2547 | NL_SET_ERR_MSG_MOD(extack, "rtm_src_len and rtm_dst_len must be 128 for IPv6"); | |
2548 | return -EINVAL; | |
2549 | } | |
2550 | ||
2551 | return 0; | |
2552 | } | |
2553 | ||
2554 | static int ip6mr_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh, | |
2555 | struct netlink_ext_ack *extack) | |
2556 | { | |
2557 | struct net *net = sock_net(in_skb->sk); | |
2558 | struct in6_addr src = {}, grp = {}; | |
2559 | struct nlattr *tb[RTA_MAX + 1]; | |
2560 | struct mfc6_cache *cache; | |
2561 | struct mr_table *mrt; | |
2562 | struct sk_buff *skb; | |
2563 | u32 tableid; | |
2564 | int err; | |
2565 | ||
2566 | err = ip6mr_rtm_valid_getroute_req(in_skb, nlh, tb, extack); | |
2567 | if (err < 0) | |
2568 | return err; | |
2569 | ||
2570 | if (tb[RTA_SRC]) | |
2571 | src = nla_get_in6_addr(tb[RTA_SRC]); | |
2572 | if (tb[RTA_DST]) | |
2573 | grp = nla_get_in6_addr(tb[RTA_DST]); | |
2574 | tableid = tb[RTA_TABLE] ? nla_get_u32(tb[RTA_TABLE]) : 0; | |
2575 | ||
2576 | mrt = ip6mr_get_table(net, tableid ?: RT_TABLE_DEFAULT); | |
2577 | if (!mrt) { | |
2578 | NL_SET_ERR_MSG_MOD(extack, "MR table does not exist"); | |
2579 | return -ENOENT; | |
2580 | } | |
2581 | ||
2582 | /* entries are added/deleted only under RTNL */ | |
2583 | rcu_read_lock(); | |
2584 | cache = ip6mr_cache_find(mrt, &src, &grp); | |
2585 | rcu_read_unlock(); | |
2586 | if (!cache) { | |
2587 | NL_SET_ERR_MSG_MOD(extack, "MR cache entry not found"); | |
2588 | return -ENOENT; | |
2589 | } | |
2590 | ||
2591 | skb = nlmsg_new(mr6_msgsize(false, mrt->maxvif), GFP_KERNEL); | |
2592 | if (!skb) | |
2593 | return -ENOBUFS; | |
2594 | ||
2595 | err = ip6mr_fill_mroute(mrt, skb, NETLINK_CB(in_skb).portid, | |
2596 | nlh->nlmsg_seq, cache, RTM_NEWROUTE, 0); | |
2597 | if (err < 0) { | |
2598 | kfree_skb(skb); | |
2599 | return err; | |
2600 | } | |
2601 | ||
2602 | return rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid); | |
2603 | } | |
2604 | ||
5b285cac PM |
2605 | static int ip6mr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb) |
2606 | { | |
e8ba330a | 2607 | const struct nlmsghdr *nlh = cb->nlh; |
4724676d | 2608 | struct fib_dump_filter filter = {}; |
cb167893 | 2609 | int err; |
e8ba330a DA |
2610 | |
2611 | if (cb->strict_check) { | |
4724676d | 2612 | err = ip_valid_fib_dump_req(sock_net(skb->sk), nlh, |
effe6792 | 2613 | &filter, cb); |
e8ba330a DA |
2614 | if (err < 0) |
2615 | return err; | |
2616 | } | |
2617 | ||
cb167893 DA |
2618 | if (filter.table_id) { |
2619 | struct mr_table *mrt; | |
2620 | ||
2621 | mrt = ip6mr_get_table(sock_net(skb->sk), filter.table_id); | |
2622 | if (!mrt) { | |
41b4bd98 | 2623 | if (rtnl_msg_family(cb->nlh) != RTNL_FAMILY_IP6MR) |
ae677bbb DA |
2624 | return skb->len; |
2625 | ||
cb167893 DA |
2626 | NL_SET_ERR_MSG_MOD(cb->extack, "MR table does not exist"); |
2627 | return -ENOENT; | |
2628 | } | |
2629 | err = mr_table_dump(mrt, skb, cb, _ip6mr_fill_mroute, | |
2630 | &mfc_unres_lock, &filter); | |
2631 | return skb->len ? : err; | |
2632 | } | |
2633 | ||
7b0db857 | 2634 | return mr_rtm_dumproute(skb, cb, ip6mr_mr_table_iter, |
cb167893 | 2635 | _ip6mr_fill_mroute, &mfc_unres_lock, &filter); |
5b285cac | 2636 | } |