]>
Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * IP multicast routing support for mrouted 3.6/3.8 | |
3 | * | |
113aa838 | 4 | * (c) 1995 Alan Cox, <alan@lxorguk.ukuu.org.uk> |
1da177e4 LT |
5 | * Linux Consultancy and Custom Driver Development |
6 | * | |
7 | * This program is free software; you can redistribute it and/or | |
8 | * modify it under the terms of the GNU General Public License | |
9 | * as published by the Free Software Foundation; either version | |
10 | * 2 of the License, or (at your option) any later version. | |
11 | * | |
1da177e4 LT |
12 | * Fixes: |
13 | * Michael Chastain : Incorrect size of copying. | |
14 | * Alan Cox : Added the cache manager code | |
15 | * Alan Cox : Fixed the clone/copy bug and device race. | |
16 | * Mike McLagan : Routing by source | |
17 | * Malcolm Beattie : Buffer handling fixes. | |
18 | * Alexey Kuznetsov : Double buffer free and other fixes. | |
19 | * SVR Anand : Fixed several multicast bugs and problems. | |
20 | * Alexey Kuznetsov : Status, optimisations and more. | |
21 | * Brad Parker : Better behaviour on mrouted upcall | |
22 | * overflow. | |
23 | * Carlos Picoto : PIMv1 Support | |
24 | * Pavlin Ivanov Radoslavov: PIMv2 Registers must checksum only PIM header | |
f77f13e2 | 25 | * Relax this requirement to work with older peers. |
1da177e4 LT |
26 | * |
27 | */ | |
28 | ||
7c0f6ba6 | 29 | #include <linux/uaccess.h> |
1da177e4 | 30 | #include <linux/types.h> |
4fc268d2 | 31 | #include <linux/capability.h> |
1da177e4 LT |
32 | #include <linux/errno.h> |
33 | #include <linux/timer.h> | |
34 | #include <linux/mm.h> | |
35 | #include <linux/kernel.h> | |
36 | #include <linux/fcntl.h> | |
37 | #include <linux/stat.h> | |
38 | #include <linux/socket.h> | |
39 | #include <linux/in.h> | |
40 | #include <linux/inet.h> | |
41 | #include <linux/netdevice.h> | |
42 | #include <linux/inetdevice.h> | |
43 | #include <linux/igmp.h> | |
44 | #include <linux/proc_fs.h> | |
45 | #include <linux/seq_file.h> | |
46 | #include <linux/mroute.h> | |
47 | #include <linux/init.h> | |
46f25dff | 48 | #include <linux/if_ether.h> |
5a0e3ad6 | 49 | #include <linux/slab.h> |
457c4cbc | 50 | #include <net/net_namespace.h> |
1da177e4 LT |
51 | #include <net/ip.h> |
52 | #include <net/protocol.h> | |
53 | #include <linux/skbuff.h> | |
14c85021 | 54 | #include <net/route.h> |
1da177e4 LT |
55 | #include <net/sock.h> |
56 | #include <net/icmp.h> | |
57 | #include <net/udp.h> | |
58 | #include <net/raw.h> | |
59 | #include <linux/notifier.h> | |
60 | #include <linux/if_arp.h> | |
61 | #include <linux/netfilter_ipv4.h> | |
709b46e8 | 62 | #include <linux/compat.h> |
bc3b2d7f | 63 | #include <linux/export.h> |
c5441932 | 64 | #include <net/ip_tunnels.h> |
1da177e4 | 65 | #include <net/checksum.h> |
dc5fc579 | 66 | #include <net/netlink.h> |
f0ad0860 | 67 | #include <net/fib_rules.h> |
d67b8c61 | 68 | #include <linux/netconf.h> |
ccbb0aa6 | 69 | #include <net/nexthop.h> |
1da177e4 | 70 | |
f0ad0860 PM |
71 | struct ipmr_rule { |
72 | struct fib_rule common; | |
73 | }; | |
74 | ||
75 | struct ipmr_result { | |
76 | struct mr_table *mrt; | |
77 | }; | |
78 | ||
1da177e4 | 79 | /* Big lock, protecting vif table, mrt cache and mroute socket state. |
a8cb16dd | 80 | * Note that the changes are semaphored via rtnl_lock. |
1da177e4 LT |
81 | */ |
82 | ||
83 | static DEFINE_RWLOCK(mrt_lock); | |
84 | ||
7ef8f65d | 85 | /* Multicast router control variables */ |
1da177e4 | 86 | |
1da177e4 LT |
87 | /* Special spinlock for queue of unresolved entries */ |
88 | static DEFINE_SPINLOCK(mfc_unres_lock); | |
89 | ||
90 | /* We return to original Alan's scheme. Hash table of resolved | |
a8cb16dd ED |
91 | * entries is changed only in process context and protected |
92 | * with weak lock mrt_lock. Queue of unresolved entries is protected | |
93 | * with strong spinlock mfc_unres_lock. | |
94 | * | |
95 | * In this case data path is free of exclusive locks at all. | |
1da177e4 LT |
96 | */ |
97 | ||
e18b890b | 98 | static struct kmem_cache *mrt_cachep __read_mostly; |
1da177e4 | 99 | |
f0ad0860 | 100 | static struct mr_table *ipmr_new_table(struct net *net, u32 id); |
acbb219d FR |
101 | static void ipmr_free_table(struct mr_table *mrt); |
102 | ||
c4854ec8 RR |
103 | static void ip_mr_forward(struct net *net, struct mr_table *mrt, |
104 | struct sk_buff *skb, struct mfc_cache *cache, | |
105 | int local); | |
0c12295a | 106 | static int ipmr_cache_report(struct mr_table *mrt, |
4feb88e5 | 107 | struct sk_buff *pkt, vifi_t vifi, int assert); |
cb6a4e46 PM |
108 | static int __ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
109 | struct mfc_cache *c, struct rtmsg *rtm); | |
8cd3ac9f ND |
110 | static void mroute_netlink_event(struct mr_table *mrt, struct mfc_cache *mfc, |
111 | int cmd); | |
0e615e96 | 112 | static void mroute_clean_tables(struct mr_table *mrt, bool all); |
f0ad0860 PM |
113 | static void ipmr_expire_process(unsigned long arg); |
114 | ||
115 | #ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES | |
116 | #define ipmr_for_each_table(mrt, net) \ | |
117 | list_for_each_entry_rcu(mrt, &net->ipv4.mr_tables, list) | |
118 | ||
119 | static struct mr_table *ipmr_get_table(struct net *net, u32 id) | |
120 | { | |
121 | struct mr_table *mrt; | |
122 | ||
123 | ipmr_for_each_table(mrt, net) { | |
124 | if (mrt->id == id) | |
125 | return mrt; | |
126 | } | |
127 | return NULL; | |
128 | } | |
129 | ||
da91981b | 130 | static int ipmr_fib_lookup(struct net *net, struct flowi4 *flp4, |
f0ad0860 PM |
131 | struct mr_table **mrt) |
132 | { | |
f0ad0860 | 133 | int err; |
95f4a45d HFS |
134 | struct ipmr_result res; |
135 | struct fib_lookup_arg arg = { | |
136 | .result = &res, | |
137 | .flags = FIB_LOOKUP_NOREF, | |
138 | }; | |
f0ad0860 | 139 | |
e58e4159 DA |
140 | /* update flow if oif or iif point to device enslaved to l3mdev */ |
141 | l3mdev_update_flow(net, flowi4_to_flowi(flp4)); | |
142 | ||
da91981b DM |
143 | err = fib_rules_lookup(net->ipv4.mr_rules_ops, |
144 | flowi4_to_flowi(flp4), 0, &arg); | |
f0ad0860 PM |
145 | if (err < 0) |
146 | return err; | |
147 | *mrt = res.mrt; | |
148 | return 0; | |
149 | } | |
150 | ||
151 | static int ipmr_rule_action(struct fib_rule *rule, struct flowi *flp, | |
152 | int flags, struct fib_lookup_arg *arg) | |
153 | { | |
154 | struct ipmr_result *res = arg->result; | |
155 | struct mr_table *mrt; | |
1da177e4 | 156 | |
f0ad0860 PM |
157 | switch (rule->action) { |
158 | case FR_ACT_TO_TBL: | |
159 | break; | |
160 | case FR_ACT_UNREACHABLE: | |
161 | return -ENETUNREACH; | |
162 | case FR_ACT_PROHIBIT: | |
163 | return -EACCES; | |
164 | case FR_ACT_BLACKHOLE: | |
165 | default: | |
166 | return -EINVAL; | |
167 | } | |
168 | ||
e58e4159 DA |
169 | arg->table = fib_rule_get_table(rule, arg); |
170 | ||
171 | mrt = ipmr_get_table(rule->fr_net, arg->table); | |
51456b29 | 172 | if (!mrt) |
f0ad0860 PM |
173 | return -EAGAIN; |
174 | res->mrt = mrt; | |
175 | return 0; | |
176 | } | |
177 | ||
178 | static int ipmr_rule_match(struct fib_rule *rule, struct flowi *fl, int flags) | |
179 | { | |
180 | return 1; | |
181 | } | |
182 | ||
183 | static const struct nla_policy ipmr_rule_policy[FRA_MAX + 1] = { | |
184 | FRA_GENERIC_POLICY, | |
185 | }; | |
186 | ||
187 | static int ipmr_rule_configure(struct fib_rule *rule, struct sk_buff *skb, | |
188 | struct fib_rule_hdr *frh, struct nlattr **tb) | |
189 | { | |
190 | return 0; | |
191 | } | |
192 | ||
193 | static int ipmr_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh, | |
194 | struct nlattr **tb) | |
195 | { | |
196 | return 1; | |
197 | } | |
198 | ||
199 | static int ipmr_rule_fill(struct fib_rule *rule, struct sk_buff *skb, | |
200 | struct fib_rule_hdr *frh) | |
201 | { | |
202 | frh->dst_len = 0; | |
203 | frh->src_len = 0; | |
204 | frh->tos = 0; | |
205 | return 0; | |
206 | } | |
207 | ||
04a6f82c | 208 | static const struct fib_rules_ops __net_initconst ipmr_rules_ops_template = { |
25239cee | 209 | .family = RTNL_FAMILY_IPMR, |
f0ad0860 PM |
210 | .rule_size = sizeof(struct ipmr_rule), |
211 | .addr_size = sizeof(u32), | |
212 | .action = ipmr_rule_action, | |
213 | .match = ipmr_rule_match, | |
214 | .configure = ipmr_rule_configure, | |
215 | .compare = ipmr_rule_compare, | |
f0ad0860 PM |
216 | .fill = ipmr_rule_fill, |
217 | .nlgroup = RTNLGRP_IPV4_RULE, | |
218 | .policy = ipmr_rule_policy, | |
219 | .owner = THIS_MODULE, | |
220 | }; | |
221 | ||
222 | static int __net_init ipmr_rules_init(struct net *net) | |
223 | { | |
224 | struct fib_rules_ops *ops; | |
225 | struct mr_table *mrt; | |
226 | int err; | |
227 | ||
228 | ops = fib_rules_register(&ipmr_rules_ops_template, net); | |
229 | if (IS_ERR(ops)) | |
230 | return PTR_ERR(ops); | |
231 | ||
232 | INIT_LIST_HEAD(&net->ipv4.mr_tables); | |
233 | ||
234 | mrt = ipmr_new_table(net, RT_TABLE_DEFAULT); | |
1113ebbc NA |
235 | if (IS_ERR(mrt)) { |
236 | err = PTR_ERR(mrt); | |
f0ad0860 PM |
237 | goto err1; |
238 | } | |
239 | ||
240 | err = fib_default_rule_add(ops, 0x7fff, RT_TABLE_DEFAULT, 0); | |
241 | if (err < 0) | |
242 | goto err2; | |
243 | ||
244 | net->ipv4.mr_rules_ops = ops; | |
245 | return 0; | |
246 | ||
247 | err2: | |
f243e5a7 | 248 | ipmr_free_table(mrt); |
f0ad0860 PM |
249 | err1: |
250 | fib_rules_unregister(ops); | |
251 | return err; | |
252 | } | |
253 | ||
254 | static void __net_exit ipmr_rules_exit(struct net *net) | |
255 | { | |
256 | struct mr_table *mrt, *next; | |
257 | ||
ed785309 | 258 | rtnl_lock(); |
035320d5 ED |
259 | list_for_each_entry_safe(mrt, next, &net->ipv4.mr_tables, list) { |
260 | list_del(&mrt->list); | |
acbb219d | 261 | ipmr_free_table(mrt); |
035320d5 | 262 | } |
f0ad0860 | 263 | fib_rules_unregister(net->ipv4.mr_rules_ops); |
419df12f | 264 | rtnl_unlock(); |
f0ad0860 PM |
265 | } |
266 | #else | |
267 | #define ipmr_for_each_table(mrt, net) \ | |
268 | for (mrt = net->ipv4.mrt; mrt; mrt = NULL) | |
269 | ||
270 | static struct mr_table *ipmr_get_table(struct net *net, u32 id) | |
271 | { | |
272 | return net->ipv4.mrt; | |
273 | } | |
274 | ||
da91981b | 275 | static int ipmr_fib_lookup(struct net *net, struct flowi4 *flp4, |
f0ad0860 PM |
276 | struct mr_table **mrt) |
277 | { | |
278 | *mrt = net->ipv4.mrt; | |
279 | return 0; | |
280 | } | |
281 | ||
282 | static int __net_init ipmr_rules_init(struct net *net) | |
283 | { | |
1113ebbc NA |
284 | struct mr_table *mrt; |
285 | ||
286 | mrt = ipmr_new_table(net, RT_TABLE_DEFAULT); | |
287 | if (IS_ERR(mrt)) | |
288 | return PTR_ERR(mrt); | |
289 | net->ipv4.mrt = mrt; | |
290 | return 0; | |
f0ad0860 PM |
291 | } |
292 | ||
293 | static void __net_exit ipmr_rules_exit(struct net *net) | |
294 | { | |
ed785309 | 295 | rtnl_lock(); |
acbb219d | 296 | ipmr_free_table(net->ipv4.mrt); |
ed785309 WC |
297 | net->ipv4.mrt = NULL; |
298 | rtnl_unlock(); | |
f0ad0860 PM |
299 | } |
300 | #endif | |
301 | ||
8fb472c0 NA |
302 | static inline int ipmr_hash_cmp(struct rhashtable_compare_arg *arg, |
303 | const void *ptr) | |
304 | { | |
305 | const struct mfc_cache_cmp_arg *cmparg = arg->key; | |
306 | struct mfc_cache *c = (struct mfc_cache *)ptr; | |
307 | ||
308 | return cmparg->mfc_mcastgrp != c->mfc_mcastgrp || | |
309 | cmparg->mfc_origin != c->mfc_origin; | |
310 | } | |
311 | ||
312 | static const struct rhashtable_params ipmr_rht_params = { | |
313 | .head_offset = offsetof(struct mfc_cache, mnode), | |
314 | .key_offset = offsetof(struct mfc_cache, cmparg), | |
315 | .key_len = sizeof(struct mfc_cache_cmp_arg), | |
316 | .nelem_hint = 3, | |
317 | .locks_mul = 1, | |
318 | .obj_cmpfn = ipmr_hash_cmp, | |
319 | .automatic_shrinking = true, | |
320 | }; | |
321 | ||
f0ad0860 PM |
322 | static struct mr_table *ipmr_new_table(struct net *net, u32 id) |
323 | { | |
324 | struct mr_table *mrt; | |
1da177e4 | 325 | |
1113ebbc NA |
326 | /* "pimreg%u" should not exceed 16 bytes (IFNAMSIZ) */ |
327 | if (id != RT_TABLE_DEFAULT && id >= 1000000000) | |
328 | return ERR_PTR(-EINVAL); | |
329 | ||
f0ad0860 | 330 | mrt = ipmr_get_table(net, id); |
00db4124 | 331 | if (mrt) |
f0ad0860 PM |
332 | return mrt; |
333 | ||
334 | mrt = kzalloc(sizeof(*mrt), GFP_KERNEL); | |
51456b29 | 335 | if (!mrt) |
1113ebbc | 336 | return ERR_PTR(-ENOMEM); |
8de53dfb | 337 | write_pnet(&mrt->net, net); |
f0ad0860 PM |
338 | mrt->id = id; |
339 | ||
8fb472c0 NA |
340 | rhltable_init(&mrt->mfc_hash, &ipmr_rht_params); |
341 | INIT_LIST_HEAD(&mrt->mfc_cache_list); | |
f0ad0860 PM |
342 | INIT_LIST_HEAD(&mrt->mfc_unres_queue); |
343 | ||
344 | setup_timer(&mrt->ipmr_expire_timer, ipmr_expire_process, | |
345 | (unsigned long)mrt); | |
346 | ||
f0ad0860 | 347 | mrt->mroute_reg_vif_num = -1; |
f0ad0860 PM |
348 | #ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES |
349 | list_add_tail_rcu(&mrt->list, &net->ipv4.mr_tables); | |
350 | #endif | |
351 | return mrt; | |
352 | } | |
1da177e4 | 353 | |
acbb219d FR |
354 | static void ipmr_free_table(struct mr_table *mrt) |
355 | { | |
356 | del_timer_sync(&mrt->ipmr_expire_timer); | |
0e615e96 | 357 | mroute_clean_tables(mrt, true); |
8fb472c0 | 358 | rhltable_destroy(&mrt->mfc_hash); |
acbb219d FR |
359 | kfree(mrt); |
360 | } | |
361 | ||
1da177e4 LT |
362 | /* Service routines creating virtual interfaces: DVMRP tunnels and PIMREG */ |
363 | ||
d607032d WC |
364 | static void ipmr_del_tunnel(struct net_device *dev, struct vifctl *v) |
365 | { | |
4feb88e5 BT |
366 | struct net *net = dev_net(dev); |
367 | ||
d607032d WC |
368 | dev_close(dev); |
369 | ||
4feb88e5 | 370 | dev = __dev_get_by_name(net, "tunl0"); |
d607032d | 371 | if (dev) { |
5bc3eb7e | 372 | const struct net_device_ops *ops = dev->netdev_ops; |
d607032d | 373 | struct ifreq ifr; |
d607032d WC |
374 | struct ip_tunnel_parm p; |
375 | ||
376 | memset(&p, 0, sizeof(p)); | |
377 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
378 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
379 | p.iph.version = 4; | |
380 | p.iph.ihl = 5; | |
381 | p.iph.protocol = IPPROTO_IPIP; | |
382 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
383 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; | |
384 | ||
5bc3eb7e SH |
385 | if (ops->ndo_do_ioctl) { |
386 | mm_segment_t oldfs = get_fs(); | |
387 | ||
388 | set_fs(KERNEL_DS); | |
389 | ops->ndo_do_ioctl(dev, &ifr, SIOCDELTUNNEL); | |
390 | set_fs(oldfs); | |
391 | } | |
d607032d WC |
392 | } |
393 | } | |
394 | ||
a0b47736 NA |
395 | /* Initialize ipmr pimreg/tunnel in_device */ |
396 | static bool ipmr_init_vif_indev(const struct net_device *dev) | |
397 | { | |
398 | struct in_device *in_dev; | |
399 | ||
400 | ASSERT_RTNL(); | |
401 | ||
402 | in_dev = __in_dev_get_rtnl(dev); | |
403 | if (!in_dev) | |
404 | return false; | |
405 | ipv4_devconf_setall(in_dev); | |
406 | neigh_parms_data_state_setall(in_dev->arp_parms); | |
407 | IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0; | |
408 | ||
409 | return true; | |
410 | } | |
411 | ||
7ef8f65d | 412 | static struct net_device *ipmr_new_tunnel(struct net *net, struct vifctl *v) |
1da177e4 LT |
413 | { |
414 | struct net_device *dev; | |
415 | ||
4feb88e5 | 416 | dev = __dev_get_by_name(net, "tunl0"); |
1da177e4 LT |
417 | |
418 | if (dev) { | |
5bc3eb7e | 419 | const struct net_device_ops *ops = dev->netdev_ops; |
1da177e4 LT |
420 | int err; |
421 | struct ifreq ifr; | |
1da177e4 | 422 | struct ip_tunnel_parm p; |
1da177e4 LT |
423 | |
424 | memset(&p, 0, sizeof(p)); | |
425 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
426 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
427 | p.iph.version = 4; | |
428 | p.iph.ihl = 5; | |
429 | p.iph.protocol = IPPROTO_IPIP; | |
430 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
ba93ef74 | 431 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; |
1da177e4 | 432 | |
5bc3eb7e SH |
433 | if (ops->ndo_do_ioctl) { |
434 | mm_segment_t oldfs = get_fs(); | |
435 | ||
436 | set_fs(KERNEL_DS); | |
437 | err = ops->ndo_do_ioctl(dev, &ifr, SIOCADDTUNNEL); | |
438 | set_fs(oldfs); | |
a8cb16dd | 439 | } else { |
5bc3eb7e | 440 | err = -EOPNOTSUPP; |
a8cb16dd | 441 | } |
1da177e4 LT |
442 | dev = NULL; |
443 | ||
4feb88e5 BT |
444 | if (err == 0 && |
445 | (dev = __dev_get_by_name(net, p.name)) != NULL) { | |
1da177e4 | 446 | dev->flags |= IFF_MULTICAST; |
a0b47736 | 447 | if (!ipmr_init_vif_indev(dev)) |
1da177e4 | 448 | goto failure; |
1da177e4 LT |
449 | if (dev_open(dev)) |
450 | goto failure; | |
7dc00c82 | 451 | dev_hold(dev); |
1da177e4 LT |
452 | } |
453 | } | |
454 | return dev; | |
455 | ||
456 | failure: | |
1da177e4 LT |
457 | unregister_netdevice(dev); |
458 | return NULL; | |
459 | } | |
460 | ||
c316c629 | 461 | #if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2) |
6fef4c0c | 462 | static netdev_tx_t reg_vif_xmit(struct sk_buff *skb, struct net_device *dev) |
1da177e4 | 463 | { |
4feb88e5 | 464 | struct net *net = dev_net(dev); |
f0ad0860 | 465 | struct mr_table *mrt; |
da91981b DM |
466 | struct flowi4 fl4 = { |
467 | .flowi4_oif = dev->ifindex, | |
6a662719 | 468 | .flowi4_iif = skb->skb_iif ? : LOOPBACK_IFINDEX, |
da91981b | 469 | .flowi4_mark = skb->mark, |
f0ad0860 PM |
470 | }; |
471 | int err; | |
472 | ||
da91981b | 473 | err = ipmr_fib_lookup(net, &fl4, &mrt); |
e40dbc51 BG |
474 | if (err < 0) { |
475 | kfree_skb(skb); | |
f0ad0860 | 476 | return err; |
e40dbc51 | 477 | } |
4feb88e5 | 478 | |
1da177e4 | 479 | read_lock(&mrt_lock); |
cf3677ae PE |
480 | dev->stats.tx_bytes += skb->len; |
481 | dev->stats.tx_packets++; | |
0c12295a | 482 | ipmr_cache_report(mrt, skb, mrt->mroute_reg_vif_num, IGMPMSG_WHOLEPKT); |
1da177e4 LT |
483 | read_unlock(&mrt_lock); |
484 | kfree_skb(skb); | |
6ed10654 | 485 | return NETDEV_TX_OK; |
1da177e4 LT |
486 | } |
487 | ||
ee9b9596 ND |
488 | static int reg_vif_get_iflink(const struct net_device *dev) |
489 | { | |
490 | return 0; | |
491 | } | |
492 | ||
007c3838 SH |
493 | static const struct net_device_ops reg_vif_netdev_ops = { |
494 | .ndo_start_xmit = reg_vif_xmit, | |
ee9b9596 | 495 | .ndo_get_iflink = reg_vif_get_iflink, |
007c3838 SH |
496 | }; |
497 | ||
1da177e4 LT |
498 | static void reg_vif_setup(struct net_device *dev) |
499 | { | |
500 | dev->type = ARPHRD_PIMREG; | |
46f25dff | 501 | dev->mtu = ETH_DATA_LEN - sizeof(struct iphdr) - 8; |
1da177e4 | 502 | dev->flags = IFF_NOARP; |
70cb4a45 | 503 | dev->netdev_ops = ®_vif_netdev_ops; |
1da177e4 | 504 | dev->destructor = free_netdev; |
403dbb97 | 505 | dev->features |= NETIF_F_NETNS_LOCAL; |
1da177e4 LT |
506 | } |
507 | ||
f0ad0860 | 508 | static struct net_device *ipmr_reg_vif(struct net *net, struct mr_table *mrt) |
1da177e4 LT |
509 | { |
510 | struct net_device *dev; | |
f0ad0860 | 511 | char name[IFNAMSIZ]; |
1da177e4 | 512 | |
f0ad0860 PM |
513 | if (mrt->id == RT_TABLE_DEFAULT) |
514 | sprintf(name, "pimreg"); | |
515 | else | |
516 | sprintf(name, "pimreg%u", mrt->id); | |
1da177e4 | 517 | |
c835a677 | 518 | dev = alloc_netdev(0, name, NET_NAME_UNKNOWN, reg_vif_setup); |
1da177e4 | 519 | |
51456b29 | 520 | if (!dev) |
1da177e4 LT |
521 | return NULL; |
522 | ||
403dbb97 TG |
523 | dev_net_set(dev, net); |
524 | ||
1da177e4 LT |
525 | if (register_netdevice(dev)) { |
526 | free_netdev(dev); | |
527 | return NULL; | |
528 | } | |
1da177e4 | 529 | |
a0b47736 | 530 | if (!ipmr_init_vif_indev(dev)) |
1da177e4 | 531 | goto failure; |
1da177e4 LT |
532 | if (dev_open(dev)) |
533 | goto failure; | |
534 | ||
7dc00c82 WC |
535 | dev_hold(dev); |
536 | ||
1da177e4 LT |
537 | return dev; |
538 | ||
539 | failure: | |
1da177e4 LT |
540 | unregister_netdevice(dev); |
541 | return NULL; | |
542 | } | |
c316c629 NA |
543 | |
544 | /* called with rcu_read_lock() */ | |
545 | static int __pim_rcv(struct mr_table *mrt, struct sk_buff *skb, | |
546 | unsigned int pimlen) | |
547 | { | |
548 | struct net_device *reg_dev = NULL; | |
549 | struct iphdr *encap; | |
550 | ||
551 | encap = (struct iphdr *)(skb_transport_header(skb) + pimlen); | |
7ef8f65d | 552 | /* Check that: |
c316c629 NA |
553 | * a. packet is really sent to a multicast group |
554 | * b. packet is not a NULL-REGISTER | |
555 | * c. packet is not truncated | |
556 | */ | |
557 | if (!ipv4_is_multicast(encap->daddr) || | |
558 | encap->tot_len == 0 || | |
559 | ntohs(encap->tot_len) + pimlen > skb->len) | |
560 | return 1; | |
561 | ||
562 | read_lock(&mrt_lock); | |
563 | if (mrt->mroute_reg_vif_num >= 0) | |
564 | reg_dev = mrt->vif_table[mrt->mroute_reg_vif_num].dev; | |
565 | read_unlock(&mrt_lock); | |
566 | ||
567 | if (!reg_dev) | |
568 | return 1; | |
569 | ||
570 | skb->mac_header = skb->network_header; | |
571 | skb_pull(skb, (u8 *)encap - skb->data); | |
572 | skb_reset_network_header(skb); | |
573 | skb->protocol = htons(ETH_P_IP); | |
574 | skb->ip_summed = CHECKSUM_NONE; | |
575 | ||
576 | skb_tunnel_rx(skb, reg_dev, dev_net(reg_dev)); | |
577 | ||
578 | netif_rx(skb); | |
579 | ||
580 | return NET_RX_SUCCESS; | |
581 | } | |
582 | #else | |
583 | static struct net_device *ipmr_reg_vif(struct net *net, struct mr_table *mrt) | |
584 | { | |
585 | return NULL; | |
586 | } | |
1da177e4 LT |
587 | #endif |
588 | ||
2c53040f BH |
589 | /** |
590 | * vif_delete - Delete a VIF entry | |
7dc00c82 | 591 | * @notify: Set to 1, if the caller is a notifier_call |
1da177e4 | 592 | */ |
0c12295a | 593 | static int vif_delete(struct mr_table *mrt, int vifi, int notify, |
d17fa6fa | 594 | struct list_head *head) |
1da177e4 LT |
595 | { |
596 | struct vif_device *v; | |
597 | struct net_device *dev; | |
598 | struct in_device *in_dev; | |
599 | ||
0c12295a | 600 | if (vifi < 0 || vifi >= mrt->maxvif) |
1da177e4 LT |
601 | return -EADDRNOTAVAIL; |
602 | ||
0c12295a | 603 | v = &mrt->vif_table[vifi]; |
1da177e4 LT |
604 | |
605 | write_lock_bh(&mrt_lock); | |
606 | dev = v->dev; | |
607 | v->dev = NULL; | |
608 | ||
609 | if (!dev) { | |
610 | write_unlock_bh(&mrt_lock); | |
611 | return -EADDRNOTAVAIL; | |
612 | } | |
613 | ||
0c12295a PM |
614 | if (vifi == mrt->mroute_reg_vif_num) |
615 | mrt->mroute_reg_vif_num = -1; | |
1da177e4 | 616 | |
a8cb16dd | 617 | if (vifi + 1 == mrt->maxvif) { |
1da177e4 | 618 | int tmp; |
a8cb16dd ED |
619 | |
620 | for (tmp = vifi - 1; tmp >= 0; tmp--) { | |
0c12295a | 621 | if (VIF_EXISTS(mrt, tmp)) |
1da177e4 LT |
622 | break; |
623 | } | |
0c12295a | 624 | mrt->maxvif = tmp+1; |
1da177e4 LT |
625 | } |
626 | ||
627 | write_unlock_bh(&mrt_lock); | |
628 | ||
629 | dev_set_allmulti(dev, -1); | |
630 | ||
a8cb16dd ED |
631 | in_dev = __in_dev_get_rtnl(dev); |
632 | if (in_dev) { | |
42f811b8 | 633 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)--; |
d67b8c61 ND |
634 | inet_netconf_notify_devconf(dev_net(dev), |
635 | NETCONFA_MC_FORWARDING, | |
636 | dev->ifindex, &in_dev->cnf); | |
1da177e4 LT |
637 | ip_rt_multicast_event(in_dev); |
638 | } | |
639 | ||
a8cb16dd | 640 | if (v->flags & (VIFF_TUNNEL | VIFF_REGISTER) && !notify) |
d17fa6fa | 641 | unregister_netdevice_queue(dev, head); |
1da177e4 LT |
642 | |
643 | dev_put(dev); | |
644 | return 0; | |
645 | } | |
646 | ||
a8c9486b | 647 | static void ipmr_cache_free_rcu(struct rcu_head *head) |
5c0a66f5 | 648 | { |
a8c9486b ED |
649 | struct mfc_cache *c = container_of(head, struct mfc_cache, rcu); |
650 | ||
5c0a66f5 BT |
651 | kmem_cache_free(mrt_cachep, c); |
652 | } | |
653 | ||
a8c9486b ED |
654 | static inline void ipmr_cache_free(struct mfc_cache *c) |
655 | { | |
656 | call_rcu(&c->rcu, ipmr_cache_free_rcu); | |
657 | } | |
658 | ||
1da177e4 | 659 | /* Destroy an unresolved cache entry, killing queued skbs |
a8cb16dd | 660 | * and reporting error to netlink readers. |
1da177e4 | 661 | */ |
0c12295a | 662 | static void ipmr_destroy_unres(struct mr_table *mrt, struct mfc_cache *c) |
1da177e4 | 663 | { |
8de53dfb | 664 | struct net *net = read_pnet(&mrt->net); |
1da177e4 | 665 | struct sk_buff *skb; |
9ef1d4c7 | 666 | struct nlmsgerr *e; |
1da177e4 | 667 | |
0c12295a | 668 | atomic_dec(&mrt->cache_resolve_queue_len); |
1da177e4 | 669 | |
c354e124 | 670 | while ((skb = skb_dequeue(&c->mfc_un.unres.unresolved))) { |
eddc9ec5 | 671 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
672 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
673 | nlh->nlmsg_type = NLMSG_ERROR; | |
573ce260 | 674 | nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr)); |
1da177e4 | 675 | skb_trim(skb, nlh->nlmsg_len); |
573ce260 | 676 | e = nlmsg_data(nlh); |
9ef1d4c7 PM |
677 | e->error = -ETIMEDOUT; |
678 | memset(&e->msg, 0, sizeof(e->msg)); | |
2942e900 | 679 | |
15e47304 | 680 | rtnl_unicast(skb, net, NETLINK_CB(skb).portid); |
a8cb16dd | 681 | } else { |
1da177e4 | 682 | kfree_skb(skb); |
a8cb16dd | 683 | } |
1da177e4 LT |
684 | } |
685 | ||
5c0a66f5 | 686 | ipmr_cache_free(c); |
1da177e4 LT |
687 | } |
688 | ||
e258beb2 | 689 | /* Timer process for the unresolved queue. */ |
e258beb2 | 690 | static void ipmr_expire_process(unsigned long arg) |
1da177e4 | 691 | { |
0c12295a | 692 | struct mr_table *mrt = (struct mr_table *)arg; |
1da177e4 LT |
693 | unsigned long now; |
694 | unsigned long expires; | |
862465f2 | 695 | struct mfc_cache *c, *next; |
1da177e4 LT |
696 | |
697 | if (!spin_trylock(&mfc_unres_lock)) { | |
0c12295a | 698 | mod_timer(&mrt->ipmr_expire_timer, jiffies+HZ/10); |
1da177e4 LT |
699 | return; |
700 | } | |
701 | ||
0c12295a | 702 | if (list_empty(&mrt->mfc_unres_queue)) |
1da177e4 LT |
703 | goto out; |
704 | ||
705 | now = jiffies; | |
706 | expires = 10*HZ; | |
1da177e4 | 707 | |
0c12295a | 708 | list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) { |
1da177e4 LT |
709 | if (time_after(c->mfc_un.unres.expires, now)) { |
710 | unsigned long interval = c->mfc_un.unres.expires - now; | |
711 | if (interval < expires) | |
712 | expires = interval; | |
1da177e4 LT |
713 | continue; |
714 | } | |
715 | ||
862465f2 | 716 | list_del(&c->list); |
8cd3ac9f | 717 | mroute_netlink_event(mrt, c, RTM_DELROUTE); |
0c12295a | 718 | ipmr_destroy_unres(mrt, c); |
1da177e4 LT |
719 | } |
720 | ||
0c12295a PM |
721 | if (!list_empty(&mrt->mfc_unres_queue)) |
722 | mod_timer(&mrt->ipmr_expire_timer, jiffies + expires); | |
1da177e4 LT |
723 | |
724 | out: | |
725 | spin_unlock(&mfc_unres_lock); | |
726 | } | |
727 | ||
728 | /* Fill oifs list. It is called under write locked mrt_lock. */ | |
0c12295a | 729 | static void ipmr_update_thresholds(struct mr_table *mrt, struct mfc_cache *cache, |
d658f8a0 | 730 | unsigned char *ttls) |
1da177e4 LT |
731 | { |
732 | int vifi; | |
733 | ||
734 | cache->mfc_un.res.minvif = MAXVIFS; | |
735 | cache->mfc_un.res.maxvif = 0; | |
736 | memset(cache->mfc_un.res.ttls, 255, MAXVIFS); | |
737 | ||
0c12295a PM |
738 | for (vifi = 0; vifi < mrt->maxvif; vifi++) { |
739 | if (VIF_EXISTS(mrt, vifi) && | |
cf958ae3 | 740 | ttls[vifi] && ttls[vifi] < 255) { |
1da177e4 LT |
741 | cache->mfc_un.res.ttls[vifi] = ttls[vifi]; |
742 | if (cache->mfc_un.res.minvif > vifi) | |
743 | cache->mfc_un.res.minvif = vifi; | |
744 | if (cache->mfc_un.res.maxvif <= vifi) | |
745 | cache->mfc_un.res.maxvif = vifi + 1; | |
746 | } | |
747 | } | |
90b5ca17 | 748 | cache->mfc_un.res.lastuse = jiffies; |
1da177e4 LT |
749 | } |
750 | ||
0c12295a PM |
751 | static int vif_add(struct net *net, struct mr_table *mrt, |
752 | struct vifctl *vifc, int mrtsock) | |
1da177e4 LT |
753 | { |
754 | int vifi = vifc->vifc_vifi; | |
0c12295a | 755 | struct vif_device *v = &mrt->vif_table[vifi]; |
1da177e4 LT |
756 | struct net_device *dev; |
757 | struct in_device *in_dev; | |
d607032d | 758 | int err; |
1da177e4 LT |
759 | |
760 | /* Is vif busy ? */ | |
0c12295a | 761 | if (VIF_EXISTS(mrt, vifi)) |
1da177e4 LT |
762 | return -EADDRINUSE; |
763 | ||
764 | switch (vifc->vifc_flags) { | |
1da177e4 | 765 | case VIFF_REGISTER: |
1973a4ea | 766 | if (!ipmr_pimsm_enabled()) |
c316c629 NA |
767 | return -EINVAL; |
768 | /* Special Purpose VIF in PIM | |
1da177e4 LT |
769 | * All the packets will be sent to the daemon |
770 | */ | |
0c12295a | 771 | if (mrt->mroute_reg_vif_num >= 0) |
1da177e4 | 772 | return -EADDRINUSE; |
f0ad0860 | 773 | dev = ipmr_reg_vif(net, mrt); |
1da177e4 LT |
774 | if (!dev) |
775 | return -ENOBUFS; | |
d607032d WC |
776 | err = dev_set_allmulti(dev, 1); |
777 | if (err) { | |
778 | unregister_netdevice(dev); | |
7dc00c82 | 779 | dev_put(dev); |
d607032d WC |
780 | return err; |
781 | } | |
1da177e4 | 782 | break; |
e905a9ed | 783 | case VIFF_TUNNEL: |
4feb88e5 | 784 | dev = ipmr_new_tunnel(net, vifc); |
1da177e4 LT |
785 | if (!dev) |
786 | return -ENOBUFS; | |
d607032d WC |
787 | err = dev_set_allmulti(dev, 1); |
788 | if (err) { | |
789 | ipmr_del_tunnel(dev, vifc); | |
7dc00c82 | 790 | dev_put(dev); |
d607032d WC |
791 | return err; |
792 | } | |
1da177e4 | 793 | break; |
ee5e81f0 | 794 | case VIFF_USE_IFINDEX: |
1da177e4 | 795 | case 0: |
ee5e81f0 I |
796 | if (vifc->vifc_flags == VIFF_USE_IFINDEX) { |
797 | dev = dev_get_by_index(net, vifc->vifc_lcl_ifindex); | |
51456b29 | 798 | if (dev && !__in_dev_get_rtnl(dev)) { |
ee5e81f0 I |
799 | dev_put(dev); |
800 | return -EADDRNOTAVAIL; | |
801 | } | |
a8cb16dd | 802 | } else { |
ee5e81f0 | 803 | dev = ip_dev_find(net, vifc->vifc_lcl_addr.s_addr); |
a8cb16dd | 804 | } |
1da177e4 LT |
805 | if (!dev) |
806 | return -EADDRNOTAVAIL; | |
d607032d | 807 | err = dev_set_allmulti(dev, 1); |
7dc00c82 WC |
808 | if (err) { |
809 | dev_put(dev); | |
d607032d | 810 | return err; |
7dc00c82 | 811 | } |
1da177e4 LT |
812 | break; |
813 | default: | |
814 | return -EINVAL; | |
815 | } | |
816 | ||
a8cb16dd ED |
817 | in_dev = __in_dev_get_rtnl(dev); |
818 | if (!in_dev) { | |
d0490cfd | 819 | dev_put(dev); |
1da177e4 | 820 | return -EADDRNOTAVAIL; |
d0490cfd | 821 | } |
42f811b8 | 822 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)++; |
d67b8c61 ND |
823 | inet_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING, dev->ifindex, |
824 | &in_dev->cnf); | |
1da177e4 LT |
825 | ip_rt_multicast_event(in_dev); |
826 | ||
a8cb16dd ED |
827 | /* Fill in the VIF structures */ |
828 | ||
c354e124 JK |
829 | v->rate_limit = vifc->vifc_rate_limit; |
830 | v->local = vifc->vifc_lcl_addr.s_addr; | |
831 | v->remote = vifc->vifc_rmt_addr.s_addr; | |
832 | v->flags = vifc->vifc_flags; | |
1da177e4 LT |
833 | if (!mrtsock) |
834 | v->flags |= VIFF_STATIC; | |
c354e124 | 835 | v->threshold = vifc->vifc_threshold; |
1da177e4 LT |
836 | v->bytes_in = 0; |
837 | v->bytes_out = 0; | |
838 | v->pkt_in = 0; | |
839 | v->pkt_out = 0; | |
840 | v->link = dev->ifindex; | |
a8cb16dd | 841 | if (v->flags & (VIFF_TUNNEL | VIFF_REGISTER)) |
a54acb3a | 842 | v->link = dev_get_iflink(dev); |
1da177e4 LT |
843 | |
844 | /* And finish update writing critical data */ | |
845 | write_lock_bh(&mrt_lock); | |
c354e124 | 846 | v->dev = dev; |
a8cb16dd | 847 | if (v->flags & VIFF_REGISTER) |
0c12295a | 848 | mrt->mroute_reg_vif_num = vifi; |
0c12295a PM |
849 | if (vifi+1 > mrt->maxvif) |
850 | mrt->maxvif = vifi+1; | |
1da177e4 LT |
851 | write_unlock_bh(&mrt_lock); |
852 | return 0; | |
853 | } | |
854 | ||
a8c9486b | 855 | /* called with rcu_read_lock() */ |
0c12295a | 856 | static struct mfc_cache *ipmr_cache_find(struct mr_table *mrt, |
4feb88e5 BT |
857 | __be32 origin, |
858 | __be32 mcastgrp) | |
1da177e4 | 859 | { |
8fb472c0 NA |
860 | struct mfc_cache_cmp_arg arg = { |
861 | .mfc_mcastgrp = mcastgrp, | |
862 | .mfc_origin = origin | |
863 | }; | |
864 | struct rhlist_head *tmp, *list; | |
1da177e4 LT |
865 | struct mfc_cache *c; |
866 | ||
8fb472c0 NA |
867 | list = rhltable_lookup(&mrt->mfc_hash, &arg, ipmr_rht_params); |
868 | rhl_for_each_entry_rcu(c, tmp, list, mnode) | |
869 | return c; | |
870 | ||
862465f2 | 871 | return NULL; |
1da177e4 LT |
872 | } |
873 | ||
660b26dc ND |
874 | /* Look for a (*,*,oif) entry */ |
875 | static struct mfc_cache *ipmr_cache_find_any_parent(struct mr_table *mrt, | |
876 | int vifi) | |
877 | { | |
8fb472c0 NA |
878 | struct mfc_cache_cmp_arg arg = { |
879 | .mfc_mcastgrp = htonl(INADDR_ANY), | |
880 | .mfc_origin = htonl(INADDR_ANY) | |
881 | }; | |
882 | struct rhlist_head *tmp, *list; | |
660b26dc ND |
883 | struct mfc_cache *c; |
884 | ||
8fb472c0 NA |
885 | list = rhltable_lookup(&mrt->mfc_hash, &arg, ipmr_rht_params); |
886 | rhl_for_each_entry_rcu(c, tmp, list, mnode) | |
887 | if (c->mfc_un.res.ttls[vifi] < 255) | |
660b26dc ND |
888 | return c; |
889 | ||
890 | return NULL; | |
891 | } | |
892 | ||
893 | /* Look for a (*,G) entry */ | |
894 | static struct mfc_cache *ipmr_cache_find_any(struct mr_table *mrt, | |
895 | __be32 mcastgrp, int vifi) | |
896 | { | |
8fb472c0 NA |
897 | struct mfc_cache_cmp_arg arg = { |
898 | .mfc_mcastgrp = mcastgrp, | |
899 | .mfc_origin = htonl(INADDR_ANY) | |
900 | }; | |
901 | struct rhlist_head *tmp, *list; | |
660b26dc ND |
902 | struct mfc_cache *c, *proxy; |
903 | ||
360eb5da | 904 | if (mcastgrp == htonl(INADDR_ANY)) |
660b26dc ND |
905 | goto skip; |
906 | ||
8fb472c0 NA |
907 | list = rhltable_lookup(&mrt->mfc_hash, &arg, ipmr_rht_params); |
908 | rhl_for_each_entry_rcu(c, tmp, list, mnode) { | |
909 | if (c->mfc_un.res.ttls[vifi] < 255) | |
910 | return c; | |
911 | ||
912 | /* It's ok if the vifi is part of the static tree */ | |
913 | proxy = ipmr_cache_find_any_parent(mrt, c->mfc_parent); | |
914 | if (proxy && proxy->mfc_un.res.ttls[vifi] < 255) | |
915 | return c; | |
916 | } | |
660b26dc ND |
917 | |
918 | skip: | |
919 | return ipmr_cache_find_any_parent(mrt, vifi); | |
920 | } | |
921 | ||
8fb472c0 NA |
922 | /* Look for a (S,G,iif) entry if parent != -1 */ |
923 | static struct mfc_cache *ipmr_cache_find_parent(struct mr_table *mrt, | |
924 | __be32 origin, __be32 mcastgrp, | |
925 | int parent) | |
926 | { | |
927 | struct mfc_cache_cmp_arg arg = { | |
928 | .mfc_mcastgrp = mcastgrp, | |
929 | .mfc_origin = origin, | |
930 | }; | |
931 | struct rhlist_head *tmp, *list; | |
932 | struct mfc_cache *c; | |
933 | ||
934 | list = rhltable_lookup(&mrt->mfc_hash, &arg, ipmr_rht_params); | |
935 | rhl_for_each_entry_rcu(c, tmp, list, mnode) | |
936 | if (parent == -1 || parent == c->mfc_parent) | |
937 | return c; | |
938 | ||
939 | return NULL; | |
940 | } | |
941 | ||
7ef8f65d | 942 | /* Allocate a multicast cache entry */ |
d658f8a0 | 943 | static struct mfc_cache *ipmr_cache_alloc(void) |
1da177e4 | 944 | { |
c354e124 | 945 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL); |
a8c9486b | 946 | |
70a0dec4 TG |
947 | if (c) { |
948 | c->mfc_un.res.last_assert = jiffies - MFC_ASSERT_THRESH - 1; | |
a8c9486b | 949 | c->mfc_un.res.minvif = MAXVIFS; |
70a0dec4 | 950 | } |
1da177e4 LT |
951 | return c; |
952 | } | |
953 | ||
d658f8a0 | 954 | static struct mfc_cache *ipmr_cache_alloc_unres(void) |
1da177e4 | 955 | { |
c354e124 | 956 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC); |
a8c9486b ED |
957 | |
958 | if (c) { | |
959 | skb_queue_head_init(&c->mfc_un.unres.unresolved); | |
960 | c->mfc_un.unres.expires = jiffies + 10*HZ; | |
961 | } | |
1da177e4 LT |
962 | return c; |
963 | } | |
964 | ||
7ef8f65d | 965 | /* A cache entry has gone into a resolved state from queued */ |
0c12295a PM |
966 | static void ipmr_cache_resolve(struct net *net, struct mr_table *mrt, |
967 | struct mfc_cache *uc, struct mfc_cache *c) | |
1da177e4 LT |
968 | { |
969 | struct sk_buff *skb; | |
9ef1d4c7 | 970 | struct nlmsgerr *e; |
1da177e4 | 971 | |
a8cb16dd | 972 | /* Play the pending entries through our router */ |
c354e124 | 973 | while ((skb = __skb_dequeue(&uc->mfc_un.unres.unresolved))) { |
eddc9ec5 | 974 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
975 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
976 | ||
573ce260 | 977 | if (__ipmr_fill_mroute(mrt, skb, c, nlmsg_data(nlh)) > 0) { |
a8cb16dd ED |
978 | nlh->nlmsg_len = skb_tail_pointer(skb) - |
979 | (u8 *)nlh; | |
1da177e4 LT |
980 | } else { |
981 | nlh->nlmsg_type = NLMSG_ERROR; | |
573ce260 | 982 | nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr)); |
1da177e4 | 983 | skb_trim(skb, nlh->nlmsg_len); |
573ce260 | 984 | e = nlmsg_data(nlh); |
9ef1d4c7 PM |
985 | e->error = -EMSGSIZE; |
986 | memset(&e->msg, 0, sizeof(e->msg)); | |
1da177e4 | 987 | } |
2942e900 | 988 | |
15e47304 | 989 | rtnl_unicast(skb, net, NETLINK_CB(skb).portid); |
a8cb16dd | 990 | } else { |
0c12295a | 991 | ip_mr_forward(net, mrt, skb, c, 0); |
a8cb16dd | 992 | } |
1da177e4 LT |
993 | } |
994 | } | |
995 | ||
c316c629 NA |
996 | /* Bounce a cache query up to mrouted. We could use netlink for this but mrouted |
997 | * expects the following bizarre scheme. | |
1da177e4 | 998 | * |
c316c629 | 999 | * Called under mrt_lock. |
1da177e4 | 1000 | */ |
0c12295a | 1001 | static int ipmr_cache_report(struct mr_table *mrt, |
4feb88e5 | 1002 | struct sk_buff *pkt, vifi_t vifi, int assert) |
1da177e4 | 1003 | { |
c9bdd4b5 | 1004 | const int ihl = ip_hdrlen(pkt); |
c316c629 | 1005 | struct sock *mroute_sk; |
1da177e4 LT |
1006 | struct igmphdr *igmp; |
1007 | struct igmpmsg *msg; | |
c316c629 | 1008 | struct sk_buff *skb; |
1da177e4 LT |
1009 | int ret; |
1010 | ||
1da177e4 LT |
1011 | if (assert == IGMPMSG_WHOLEPKT) |
1012 | skb = skb_realloc_headroom(pkt, sizeof(struct iphdr)); | |
1013 | else | |
1da177e4 LT |
1014 | skb = alloc_skb(128, GFP_ATOMIC); |
1015 | ||
132adf54 | 1016 | if (!skb) |
1da177e4 LT |
1017 | return -ENOBUFS; |
1018 | ||
1da177e4 LT |
1019 | if (assert == IGMPMSG_WHOLEPKT) { |
1020 | /* Ugly, but we have no choice with this interface. | |
a8cb16dd ED |
1021 | * Duplicate old header, fix ihl, length etc. |
1022 | * And all this only to mangle msg->im_msgtype and | |
1023 | * to set msg->im_mbz to "mbz" :-) | |
1da177e4 | 1024 | */ |
878c8145 ACM |
1025 | skb_push(skb, sizeof(struct iphdr)); |
1026 | skb_reset_network_header(skb); | |
badff6d0 | 1027 | skb_reset_transport_header(skb); |
0272ffc4 | 1028 | msg = (struct igmpmsg *)skb_network_header(skb); |
d56f90a7 | 1029 | memcpy(msg, skb_network_header(pkt), sizeof(struct iphdr)); |
1da177e4 LT |
1030 | msg->im_msgtype = IGMPMSG_WHOLEPKT; |
1031 | msg->im_mbz = 0; | |
0c12295a | 1032 | msg->im_vif = mrt->mroute_reg_vif_num; |
eddc9ec5 ACM |
1033 | ip_hdr(skb)->ihl = sizeof(struct iphdr) >> 2; |
1034 | ip_hdr(skb)->tot_len = htons(ntohs(ip_hdr(pkt)->tot_len) + | |
1035 | sizeof(struct iphdr)); | |
c316c629 NA |
1036 | } else { |
1037 | /* Copy the IP header */ | |
1038 | skb_set_network_header(skb, skb->len); | |
1039 | skb_put(skb, ihl); | |
1040 | skb_copy_to_linear_data(skb, pkt->data, ihl); | |
1041 | /* Flag to the kernel this is a route add */ | |
1042 | ip_hdr(skb)->protocol = 0; | |
1043 | msg = (struct igmpmsg *)skb_network_header(skb); | |
1044 | msg->im_vif = vifi; | |
1045 | skb_dst_set(skb, dst_clone(skb_dst(pkt))); | |
1046 | /* Add our header */ | |
1047 | igmp = (struct igmphdr *)skb_put(skb, sizeof(struct igmphdr)); | |
1048 | igmp->type = assert; | |
1049 | msg->im_msgtype = assert; | |
1050 | igmp->code = 0; | |
1051 | ip_hdr(skb)->tot_len = htons(skb->len); /* Fix the length */ | |
1052 | skb->transport_header = skb->network_header; | |
e905a9ed | 1053 | } |
1da177e4 | 1054 | |
4c968709 ED |
1055 | rcu_read_lock(); |
1056 | mroute_sk = rcu_dereference(mrt->mroute_sk); | |
51456b29 | 1057 | if (!mroute_sk) { |
4c968709 | 1058 | rcu_read_unlock(); |
1da177e4 LT |
1059 | kfree_skb(skb); |
1060 | return -EINVAL; | |
1061 | } | |
1062 | ||
a8cb16dd | 1063 | /* Deliver to mrouted */ |
4c968709 ED |
1064 | ret = sock_queue_rcv_skb(mroute_sk, skb); |
1065 | rcu_read_unlock(); | |
70a269e6 | 1066 | if (ret < 0) { |
e87cc472 | 1067 | net_warn_ratelimited("mroute: pending queue full, dropping entries\n"); |
1da177e4 LT |
1068 | kfree_skb(skb); |
1069 | } | |
1070 | ||
1071 | return ret; | |
1072 | } | |
1073 | ||
7ef8f65d NA |
1074 | /* Queue a packet for resolution. It gets locked cache entry! */ |
1075 | static int ipmr_cache_unresolved(struct mr_table *mrt, vifi_t vifi, | |
1076 | struct sk_buff *skb) | |
1da177e4 | 1077 | { |
8fb472c0 NA |
1078 | const struct iphdr *iph = ip_hdr(skb); |
1079 | struct mfc_cache *c; | |
862465f2 | 1080 | bool found = false; |
1da177e4 | 1081 | int err; |
1da177e4 LT |
1082 | |
1083 | spin_lock_bh(&mfc_unres_lock); | |
0c12295a | 1084 | list_for_each_entry(c, &mrt->mfc_unres_queue, list) { |
e258beb2 | 1085 | if (c->mfc_mcastgrp == iph->daddr && |
862465f2 PM |
1086 | c->mfc_origin == iph->saddr) { |
1087 | found = true; | |
1da177e4 | 1088 | break; |
862465f2 | 1089 | } |
1da177e4 LT |
1090 | } |
1091 | ||
862465f2 | 1092 | if (!found) { |
a8cb16dd | 1093 | /* Create a new entry if allowable */ |
0c12295a | 1094 | if (atomic_read(&mrt->cache_resolve_queue_len) >= 10 || |
d658f8a0 | 1095 | (c = ipmr_cache_alloc_unres()) == NULL) { |
1da177e4 LT |
1096 | spin_unlock_bh(&mfc_unres_lock); |
1097 | ||
1098 | kfree_skb(skb); | |
1099 | return -ENOBUFS; | |
1100 | } | |
1101 | ||
a8cb16dd | 1102 | /* Fill in the new cache entry */ |
eddc9ec5 ACM |
1103 | c->mfc_parent = -1; |
1104 | c->mfc_origin = iph->saddr; | |
1105 | c->mfc_mcastgrp = iph->daddr; | |
1da177e4 | 1106 | |
a8cb16dd | 1107 | /* Reflect first query at mrouted. */ |
0c12295a | 1108 | err = ipmr_cache_report(mrt, skb, vifi, IGMPMSG_NOCACHE); |
4feb88e5 | 1109 | if (err < 0) { |
e905a9ed | 1110 | /* If the report failed throw the cache entry |
1da177e4 LT |
1111 | out - Brad Parker |
1112 | */ | |
1113 | spin_unlock_bh(&mfc_unres_lock); | |
1114 | ||
5c0a66f5 | 1115 | ipmr_cache_free(c); |
1da177e4 LT |
1116 | kfree_skb(skb); |
1117 | return err; | |
1118 | } | |
1119 | ||
0c12295a PM |
1120 | atomic_inc(&mrt->cache_resolve_queue_len); |
1121 | list_add(&c->list, &mrt->mfc_unres_queue); | |
8cd3ac9f | 1122 | mroute_netlink_event(mrt, c, RTM_NEWROUTE); |
1da177e4 | 1123 | |
278554bd DM |
1124 | if (atomic_read(&mrt->cache_resolve_queue_len) == 1) |
1125 | mod_timer(&mrt->ipmr_expire_timer, c->mfc_un.unres.expires); | |
1da177e4 LT |
1126 | } |
1127 | ||
a8cb16dd | 1128 | /* See if we can append the packet */ |
a8cb16dd | 1129 | if (c->mfc_un.unres.unresolved.qlen > 3) { |
1da177e4 LT |
1130 | kfree_skb(skb); |
1131 | err = -ENOBUFS; | |
1132 | } else { | |
c354e124 | 1133 | skb_queue_tail(&c->mfc_un.unres.unresolved, skb); |
1da177e4 LT |
1134 | err = 0; |
1135 | } | |
1136 | ||
1137 | spin_unlock_bh(&mfc_unres_lock); | |
1138 | return err; | |
1139 | } | |
1140 | ||
7ef8f65d | 1141 | /* MFC cache manipulation by user space mroute daemon */ |
1da177e4 | 1142 | |
660b26dc | 1143 | static int ipmr_mfc_delete(struct mr_table *mrt, struct mfcctl *mfc, int parent) |
1da177e4 | 1144 | { |
8fb472c0 | 1145 | struct mfc_cache *c; |
1da177e4 | 1146 | |
8fb472c0 NA |
1147 | /* The entries are added/deleted only under RTNL */ |
1148 | rcu_read_lock(); | |
1149 | c = ipmr_cache_find_parent(mrt, mfc->mfcc_origin.s_addr, | |
1150 | mfc->mfcc_mcastgrp.s_addr, parent); | |
1151 | rcu_read_unlock(); | |
1152 | if (!c) | |
1153 | return -ENOENT; | |
1154 | rhltable_remove(&mrt->mfc_hash, &c->mnode, ipmr_rht_params); | |
1155 | list_del_rcu(&c->list); | |
1156 | mroute_netlink_event(mrt, c, RTM_DELROUTE); | |
1157 | ipmr_cache_free(c); | |
1da177e4 | 1158 | |
8fb472c0 | 1159 | return 0; |
1da177e4 LT |
1160 | } |
1161 | ||
0c12295a | 1162 | static int ipmr_mfc_add(struct net *net, struct mr_table *mrt, |
660b26dc | 1163 | struct mfcctl *mfc, int mrtsock, int parent) |
1da177e4 | 1164 | { |
862465f2 | 1165 | struct mfc_cache *uc, *c; |
8fb472c0 NA |
1166 | bool found; |
1167 | int ret; | |
1da177e4 | 1168 | |
a50436f2 PM |
1169 | if (mfc->mfcc_parent >= MAXVIFS) |
1170 | return -ENFILE; | |
1171 | ||
8fb472c0 NA |
1172 | /* The entries are added/deleted only under RTNL */ |
1173 | rcu_read_lock(); | |
1174 | c = ipmr_cache_find_parent(mrt, mfc->mfcc_origin.s_addr, | |
1175 | mfc->mfcc_mcastgrp.s_addr, parent); | |
1176 | rcu_read_unlock(); | |
1177 | if (c) { | |
1da177e4 LT |
1178 | write_lock_bh(&mrt_lock); |
1179 | c->mfc_parent = mfc->mfcc_parent; | |
0c12295a | 1180 | ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls); |
1da177e4 LT |
1181 | if (!mrtsock) |
1182 | c->mfc_flags |= MFC_STATIC; | |
1183 | write_unlock_bh(&mrt_lock); | |
8cd3ac9f | 1184 | mroute_netlink_event(mrt, c, RTM_NEWROUTE); |
1da177e4 LT |
1185 | return 0; |
1186 | } | |
1187 | ||
360eb5da | 1188 | if (mfc->mfcc_mcastgrp.s_addr != htonl(INADDR_ANY) && |
660b26dc | 1189 | !ipv4_is_multicast(mfc->mfcc_mcastgrp.s_addr)) |
1da177e4 LT |
1190 | return -EINVAL; |
1191 | ||
d658f8a0 | 1192 | c = ipmr_cache_alloc(); |
51456b29 | 1193 | if (!c) |
1da177e4 LT |
1194 | return -ENOMEM; |
1195 | ||
c354e124 JK |
1196 | c->mfc_origin = mfc->mfcc_origin.s_addr; |
1197 | c->mfc_mcastgrp = mfc->mfcc_mcastgrp.s_addr; | |
1198 | c->mfc_parent = mfc->mfcc_parent; | |
0c12295a | 1199 | ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls); |
1da177e4 LT |
1200 | if (!mrtsock) |
1201 | c->mfc_flags |= MFC_STATIC; | |
1202 | ||
8fb472c0 NA |
1203 | ret = rhltable_insert_key(&mrt->mfc_hash, &c->cmparg, &c->mnode, |
1204 | ipmr_rht_params); | |
1205 | if (ret) { | |
1206 | pr_err("ipmr: rhtable insert error %d\n", ret); | |
1207 | ipmr_cache_free(c); | |
1208 | return ret; | |
1209 | } | |
1210 | list_add_tail_rcu(&c->list, &mrt->mfc_cache_list); | |
7ef8f65d NA |
1211 | /* Check to see if we resolved a queued list. If so we |
1212 | * need to send on the frames and tidy up. | |
1da177e4 | 1213 | */ |
b0ebb739 | 1214 | found = false; |
1da177e4 | 1215 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 1216 | list_for_each_entry(uc, &mrt->mfc_unres_queue, list) { |
e258beb2 | 1217 | if (uc->mfc_origin == c->mfc_origin && |
1da177e4 | 1218 | uc->mfc_mcastgrp == c->mfc_mcastgrp) { |
862465f2 | 1219 | list_del(&uc->list); |
0c12295a | 1220 | atomic_dec(&mrt->cache_resolve_queue_len); |
b0ebb739 | 1221 | found = true; |
1da177e4 LT |
1222 | break; |
1223 | } | |
1224 | } | |
0c12295a PM |
1225 | if (list_empty(&mrt->mfc_unres_queue)) |
1226 | del_timer(&mrt->ipmr_expire_timer); | |
1da177e4 LT |
1227 | spin_unlock_bh(&mfc_unres_lock); |
1228 | ||
b0ebb739 | 1229 | if (found) { |
0c12295a | 1230 | ipmr_cache_resolve(net, mrt, uc, c); |
5c0a66f5 | 1231 | ipmr_cache_free(uc); |
1da177e4 | 1232 | } |
8cd3ac9f | 1233 | mroute_netlink_event(mrt, c, RTM_NEWROUTE); |
1da177e4 LT |
1234 | return 0; |
1235 | } | |
1236 | ||
7ef8f65d | 1237 | /* Close the multicast socket, and clear the vif tables etc */ |
0e615e96 | 1238 | static void mroute_clean_tables(struct mr_table *mrt, bool all) |
1da177e4 | 1239 | { |
8fb472c0 | 1240 | struct mfc_cache *c, *tmp; |
d17fa6fa | 1241 | LIST_HEAD(list); |
8fb472c0 | 1242 | int i; |
e905a9ed | 1243 | |
a8cb16dd | 1244 | /* Shut down all active vif entries */ |
0c12295a | 1245 | for (i = 0; i < mrt->maxvif; i++) { |
0e615e96 NA |
1246 | if (!all && (mrt->vif_table[i].flags & VIFF_STATIC)) |
1247 | continue; | |
1248 | vif_delete(mrt, i, 0, &list); | |
1da177e4 | 1249 | } |
d17fa6fa | 1250 | unregister_netdevice_many(&list); |
1da177e4 | 1251 | |
a8cb16dd | 1252 | /* Wipe the cache */ |
8fb472c0 NA |
1253 | list_for_each_entry_safe(c, tmp, &mrt->mfc_cache_list, list) { |
1254 | if (!all && (c->mfc_flags & MFC_STATIC)) | |
1255 | continue; | |
1256 | rhltable_remove(&mrt->mfc_hash, &c->mnode, ipmr_rht_params); | |
1257 | list_del_rcu(&c->list); | |
1258 | mroute_netlink_event(mrt, c, RTM_DELROUTE); | |
1259 | ipmr_cache_free(c); | |
1da177e4 LT |
1260 | } |
1261 | ||
0c12295a | 1262 | if (atomic_read(&mrt->cache_resolve_queue_len) != 0) { |
1da177e4 | 1263 | spin_lock_bh(&mfc_unres_lock); |
8fb472c0 | 1264 | list_for_each_entry_safe(c, tmp, &mrt->mfc_unres_queue, list) { |
862465f2 | 1265 | list_del(&c->list); |
8cd3ac9f | 1266 | mroute_netlink_event(mrt, c, RTM_DELROUTE); |
0c12295a | 1267 | ipmr_destroy_unres(mrt, c); |
1da177e4 LT |
1268 | } |
1269 | spin_unlock_bh(&mfc_unres_lock); | |
1270 | } | |
1271 | } | |
1272 | ||
4c968709 ED |
1273 | /* called from ip_ra_control(), before an RCU grace period, |
1274 | * we dont need to call synchronize_rcu() here | |
1275 | */ | |
1da177e4 LT |
1276 | static void mrtsock_destruct(struct sock *sk) |
1277 | { | |
4feb88e5 | 1278 | struct net *net = sock_net(sk); |
f0ad0860 | 1279 | struct mr_table *mrt; |
4feb88e5 | 1280 | |
1da177e4 | 1281 | rtnl_lock(); |
f0ad0860 | 1282 | ipmr_for_each_table(mrt, net) { |
4c968709 | 1283 | if (sk == rtnl_dereference(mrt->mroute_sk)) { |
f0ad0860 | 1284 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)--; |
d67b8c61 ND |
1285 | inet_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING, |
1286 | NETCONFA_IFINDEX_ALL, | |
1287 | net->ipv4.devconf_all); | |
a9b3cd7f | 1288 | RCU_INIT_POINTER(mrt->mroute_sk, NULL); |
0e615e96 | 1289 | mroute_clean_tables(mrt, false); |
f0ad0860 | 1290 | } |
1da177e4 LT |
1291 | } |
1292 | rtnl_unlock(); | |
1293 | } | |
1294 | ||
7ef8f65d NA |
1295 | /* Socket options and virtual interface manipulation. The whole |
1296 | * virtual interface system is a complete heap, but unfortunately | |
1297 | * that's how BSD mrouted happens to think. Maybe one day with a proper | |
1298 | * MOSPF/PIM router set up we can clean this up. | |
1da177e4 | 1299 | */ |
e905a9ed | 1300 | |
29e97d21 NA |
1301 | int ip_mroute_setsockopt(struct sock *sk, int optname, char __user *optval, |
1302 | unsigned int optlen) | |
1da177e4 | 1303 | { |
4feb88e5 | 1304 | struct net *net = sock_net(sk); |
29e97d21 | 1305 | int val, ret = 0, parent = 0; |
f0ad0860 | 1306 | struct mr_table *mrt; |
29e97d21 NA |
1307 | struct vifctl vif; |
1308 | struct mfcctl mfc; | |
1309 | u32 uval; | |
f0ad0860 | 1310 | |
29e97d21 NA |
1311 | /* There's one exception to the lock - MRT_DONE which needs to unlock */ |
1312 | rtnl_lock(); | |
5e1859fb | 1313 | if (sk->sk_type != SOCK_RAW || |
29e97d21 NA |
1314 | inet_sk(sk)->inet_num != IPPROTO_IGMP) { |
1315 | ret = -EOPNOTSUPP; | |
1316 | goto out_unlock; | |
1317 | } | |
5e1859fb | 1318 | |
f0ad0860 | 1319 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); |
29e97d21 NA |
1320 | if (!mrt) { |
1321 | ret = -ENOENT; | |
1322 | goto out_unlock; | |
1323 | } | |
132adf54 | 1324 | if (optname != MRT_INIT) { |
33d480ce | 1325 | if (sk != rcu_access_pointer(mrt->mroute_sk) && |
29e97d21 NA |
1326 | !ns_capable(net->user_ns, CAP_NET_ADMIN)) { |
1327 | ret = -EACCES; | |
1328 | goto out_unlock; | |
1329 | } | |
1da177e4 LT |
1330 | } |
1331 | ||
132adf54 SH |
1332 | switch (optname) { |
1333 | case MRT_INIT: | |
42e6b89c | 1334 | if (optlen != sizeof(int)) { |
29e97d21 | 1335 | ret = -EINVAL; |
42e6b89c NA |
1336 | break; |
1337 | } | |
1338 | if (rtnl_dereference(mrt->mroute_sk)) { | |
29e97d21 | 1339 | ret = -EADDRINUSE; |
29e97d21 | 1340 | break; |
42e6b89c | 1341 | } |
132adf54 SH |
1342 | |
1343 | ret = ip_ra_control(sk, 1, mrtsock_destruct); | |
1344 | if (ret == 0) { | |
cf778b00 | 1345 | rcu_assign_pointer(mrt->mroute_sk, sk); |
4feb88e5 | 1346 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)++; |
d67b8c61 ND |
1347 | inet_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING, |
1348 | NETCONFA_IFINDEX_ALL, | |
1349 | net->ipv4.devconf_all); | |
132adf54 | 1350 | } |
29e97d21 | 1351 | break; |
132adf54 | 1352 | case MRT_DONE: |
29e97d21 NA |
1353 | if (sk != rcu_access_pointer(mrt->mroute_sk)) { |
1354 | ret = -EACCES; | |
1355 | } else { | |
1356 | /* We need to unlock here because mrtsock_destruct takes | |
1357 | * care of rtnl itself and we can't change that due to | |
1358 | * the IP_ROUTER_ALERT setsockopt which runs without it. | |
1359 | */ | |
1360 | rtnl_unlock(); | |
1361 | ret = ip_ra_control(sk, 0, NULL); | |
1362 | goto out; | |
1363 | } | |
1364 | break; | |
132adf54 SH |
1365 | case MRT_ADD_VIF: |
1366 | case MRT_DEL_VIF: | |
29e97d21 NA |
1367 | if (optlen != sizeof(vif)) { |
1368 | ret = -EINVAL; | |
1369 | break; | |
1370 | } | |
1371 | if (copy_from_user(&vif, optval, sizeof(vif))) { | |
1372 | ret = -EFAULT; | |
1373 | break; | |
1374 | } | |
1375 | if (vif.vifc_vifi >= MAXVIFS) { | |
1376 | ret = -ENFILE; | |
1377 | break; | |
1378 | } | |
c354e124 | 1379 | if (optname == MRT_ADD_VIF) { |
4c968709 ED |
1380 | ret = vif_add(net, mrt, &vif, |
1381 | sk == rtnl_dereference(mrt->mroute_sk)); | |
132adf54 | 1382 | } else { |
0c12295a | 1383 | ret = vif_delete(mrt, vif.vifc_vifi, 0, NULL); |
132adf54 | 1384 | } |
29e97d21 | 1385 | break; |
7ef8f65d NA |
1386 | /* Manipulate the forwarding caches. These live |
1387 | * in a sort of kernel/user symbiosis. | |
1388 | */ | |
132adf54 SH |
1389 | case MRT_ADD_MFC: |
1390 | case MRT_DEL_MFC: | |
660b26dc ND |
1391 | parent = -1; |
1392 | case MRT_ADD_MFC_PROXY: | |
1393 | case MRT_DEL_MFC_PROXY: | |
29e97d21 NA |
1394 | if (optlen != sizeof(mfc)) { |
1395 | ret = -EINVAL; | |
1396 | break; | |
1397 | } | |
1398 | if (copy_from_user(&mfc, optval, sizeof(mfc))) { | |
1399 | ret = -EFAULT; | |
1400 | break; | |
1401 | } | |
660b26dc ND |
1402 | if (parent == 0) |
1403 | parent = mfc.mfcc_parent; | |
660b26dc ND |
1404 | if (optname == MRT_DEL_MFC || optname == MRT_DEL_MFC_PROXY) |
1405 | ret = ipmr_mfc_delete(mrt, &mfc, parent); | |
132adf54 | 1406 | else |
4c968709 | 1407 | ret = ipmr_mfc_add(net, mrt, &mfc, |
660b26dc ND |
1408 | sk == rtnl_dereference(mrt->mroute_sk), |
1409 | parent); | |
29e97d21 | 1410 | break; |
7ef8f65d | 1411 | /* Control PIM assert. */ |
132adf54 | 1412 | case MRT_ASSERT: |
29e97d21 NA |
1413 | if (optlen != sizeof(val)) { |
1414 | ret = -EINVAL; | |
1415 | break; | |
1416 | } | |
1417 | if (get_user(val, (int __user *)optval)) { | |
1418 | ret = -EFAULT; | |
1419 | break; | |
1420 | } | |
1421 | mrt->mroute_do_assert = val; | |
1422 | break; | |
132adf54 | 1423 | case MRT_PIM: |
1973a4ea | 1424 | if (!ipmr_pimsm_enabled()) { |
29e97d21 NA |
1425 | ret = -ENOPROTOOPT; |
1426 | break; | |
1427 | } | |
1428 | if (optlen != sizeof(val)) { | |
1429 | ret = -EINVAL; | |
1430 | break; | |
1431 | } | |
1432 | if (get_user(val, (int __user *)optval)) { | |
1433 | ret = -EFAULT; | |
1434 | break; | |
1435 | } | |
ba93ef74 | 1436 | |
29e97d21 NA |
1437 | val = !!val; |
1438 | if (val != mrt->mroute_do_pim) { | |
1439 | mrt->mroute_do_pim = val; | |
1440 | mrt->mroute_do_assert = val; | |
1da177e4 | 1441 | } |
29e97d21 | 1442 | break; |
f0ad0860 | 1443 | case MRT_TABLE: |
29e97d21 NA |
1444 | if (!IS_BUILTIN(CONFIG_IP_MROUTE_MULTIPLE_TABLES)) { |
1445 | ret = -ENOPROTOOPT; | |
1446 | break; | |
1447 | } | |
1448 | if (optlen != sizeof(uval)) { | |
1449 | ret = -EINVAL; | |
1450 | break; | |
1451 | } | |
1452 | if (get_user(uval, (u32 __user *)optval)) { | |
1453 | ret = -EFAULT; | |
1454 | break; | |
1455 | } | |
f0ad0860 | 1456 | |
4c968709 ED |
1457 | if (sk == rtnl_dereference(mrt->mroute_sk)) { |
1458 | ret = -EBUSY; | |
1459 | } else { | |
29e97d21 | 1460 | mrt = ipmr_new_table(net, uval); |
1113ebbc NA |
1461 | if (IS_ERR(mrt)) |
1462 | ret = PTR_ERR(mrt); | |
5e1859fb | 1463 | else |
29e97d21 | 1464 | raw_sk(sk)->ipmr_table = uval; |
4c968709 | 1465 | } |
29e97d21 | 1466 | break; |
7ef8f65d | 1467 | /* Spurious command, or MRT_VERSION which you cannot set. */ |
132adf54 | 1468 | default: |
29e97d21 | 1469 | ret = -ENOPROTOOPT; |
1da177e4 | 1470 | } |
29e97d21 NA |
1471 | out_unlock: |
1472 | rtnl_unlock(); | |
1473 | out: | |
1474 | return ret; | |
1da177e4 LT |
1475 | } |
1476 | ||
7ef8f65d | 1477 | /* Getsock opt support for the multicast routing system. */ |
c354e124 | 1478 | int ip_mroute_getsockopt(struct sock *sk, int optname, char __user *optval, int __user *optlen) |
1da177e4 LT |
1479 | { |
1480 | int olr; | |
1481 | int val; | |
4feb88e5 | 1482 | struct net *net = sock_net(sk); |
f0ad0860 PM |
1483 | struct mr_table *mrt; |
1484 | ||
5e1859fb ED |
1485 | if (sk->sk_type != SOCK_RAW || |
1486 | inet_sk(sk)->inet_num != IPPROTO_IGMP) | |
1487 | return -EOPNOTSUPP; | |
1488 | ||
f0ad0860 | 1489 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); |
51456b29 | 1490 | if (!mrt) |
f0ad0860 | 1491 | return -ENOENT; |
1da177e4 | 1492 | |
fe9ef3ce NA |
1493 | switch (optname) { |
1494 | case MRT_VERSION: | |
1495 | val = 0x0305; | |
1496 | break; | |
1497 | case MRT_PIM: | |
1973a4ea | 1498 | if (!ipmr_pimsm_enabled()) |
fe9ef3ce NA |
1499 | return -ENOPROTOOPT; |
1500 | val = mrt->mroute_do_pim; | |
1501 | break; | |
1502 | case MRT_ASSERT: | |
1503 | val = mrt->mroute_do_assert; | |
1504 | break; | |
1505 | default: | |
1da177e4 | 1506 | return -ENOPROTOOPT; |
fe9ef3ce | 1507 | } |
1da177e4 LT |
1508 | |
1509 | if (get_user(olr, optlen)) | |
1510 | return -EFAULT; | |
1da177e4 LT |
1511 | olr = min_t(unsigned int, olr, sizeof(int)); |
1512 | if (olr < 0) | |
1513 | return -EINVAL; | |
c354e124 | 1514 | if (put_user(olr, optlen)) |
1da177e4 | 1515 | return -EFAULT; |
c354e124 | 1516 | if (copy_to_user(optval, &val, olr)) |
1da177e4 LT |
1517 | return -EFAULT; |
1518 | return 0; | |
1519 | } | |
1520 | ||
7ef8f65d | 1521 | /* The IP multicast ioctl support routines. */ |
1da177e4 LT |
1522 | int ipmr_ioctl(struct sock *sk, int cmd, void __user *arg) |
1523 | { | |
1524 | struct sioc_sg_req sr; | |
1525 | struct sioc_vif_req vr; | |
1526 | struct vif_device *vif; | |
1527 | struct mfc_cache *c; | |
4feb88e5 | 1528 | struct net *net = sock_net(sk); |
f0ad0860 PM |
1529 | struct mr_table *mrt; |
1530 | ||
1531 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); | |
51456b29 | 1532 | if (!mrt) |
f0ad0860 | 1533 | return -ENOENT; |
e905a9ed | 1534 | |
132adf54 SH |
1535 | switch (cmd) { |
1536 | case SIOCGETVIFCNT: | |
c354e124 | 1537 | if (copy_from_user(&vr, arg, sizeof(vr))) |
132adf54 | 1538 | return -EFAULT; |
0c12295a | 1539 | if (vr.vifi >= mrt->maxvif) |
132adf54 SH |
1540 | return -EINVAL; |
1541 | read_lock(&mrt_lock); | |
0c12295a PM |
1542 | vif = &mrt->vif_table[vr.vifi]; |
1543 | if (VIF_EXISTS(mrt, vr.vifi)) { | |
c354e124 JK |
1544 | vr.icount = vif->pkt_in; |
1545 | vr.ocount = vif->pkt_out; | |
1546 | vr.ibytes = vif->bytes_in; | |
1547 | vr.obytes = vif->bytes_out; | |
1da177e4 | 1548 | read_unlock(&mrt_lock); |
1da177e4 | 1549 | |
c354e124 | 1550 | if (copy_to_user(arg, &vr, sizeof(vr))) |
132adf54 SH |
1551 | return -EFAULT; |
1552 | return 0; | |
1553 | } | |
1554 | read_unlock(&mrt_lock); | |
1555 | return -EADDRNOTAVAIL; | |
1556 | case SIOCGETSGCNT: | |
c354e124 | 1557 | if (copy_from_user(&sr, arg, sizeof(sr))) |
132adf54 SH |
1558 | return -EFAULT; |
1559 | ||
a8c9486b | 1560 | rcu_read_lock(); |
0c12295a | 1561 | c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr); |
132adf54 SH |
1562 | if (c) { |
1563 | sr.pktcnt = c->mfc_un.res.pkt; | |
1564 | sr.bytecnt = c->mfc_un.res.bytes; | |
1565 | sr.wrong_if = c->mfc_un.res.wrong_if; | |
a8c9486b | 1566 | rcu_read_unlock(); |
132adf54 | 1567 | |
c354e124 | 1568 | if (copy_to_user(arg, &sr, sizeof(sr))) |
132adf54 SH |
1569 | return -EFAULT; |
1570 | return 0; | |
1571 | } | |
a8c9486b | 1572 | rcu_read_unlock(); |
132adf54 SH |
1573 | return -EADDRNOTAVAIL; |
1574 | default: | |
1575 | return -ENOIOCTLCMD; | |
1da177e4 LT |
1576 | } |
1577 | } | |
1578 | ||
709b46e8 EB |
1579 | #ifdef CONFIG_COMPAT |
1580 | struct compat_sioc_sg_req { | |
1581 | struct in_addr src; | |
1582 | struct in_addr grp; | |
1583 | compat_ulong_t pktcnt; | |
1584 | compat_ulong_t bytecnt; | |
1585 | compat_ulong_t wrong_if; | |
1586 | }; | |
1587 | ||
ca6b8bb0 DM |
1588 | struct compat_sioc_vif_req { |
1589 | vifi_t vifi; /* Which iface */ | |
1590 | compat_ulong_t icount; | |
1591 | compat_ulong_t ocount; | |
1592 | compat_ulong_t ibytes; | |
1593 | compat_ulong_t obytes; | |
1594 | }; | |
1595 | ||
709b46e8 EB |
1596 | int ipmr_compat_ioctl(struct sock *sk, unsigned int cmd, void __user *arg) |
1597 | { | |
0033d5ad | 1598 | struct compat_sioc_sg_req sr; |
ca6b8bb0 DM |
1599 | struct compat_sioc_vif_req vr; |
1600 | struct vif_device *vif; | |
709b46e8 EB |
1601 | struct mfc_cache *c; |
1602 | struct net *net = sock_net(sk); | |
1603 | struct mr_table *mrt; | |
1604 | ||
1605 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); | |
51456b29 | 1606 | if (!mrt) |
709b46e8 EB |
1607 | return -ENOENT; |
1608 | ||
1609 | switch (cmd) { | |
ca6b8bb0 DM |
1610 | case SIOCGETVIFCNT: |
1611 | if (copy_from_user(&vr, arg, sizeof(vr))) | |
1612 | return -EFAULT; | |
1613 | if (vr.vifi >= mrt->maxvif) | |
1614 | return -EINVAL; | |
1615 | read_lock(&mrt_lock); | |
1616 | vif = &mrt->vif_table[vr.vifi]; | |
1617 | if (VIF_EXISTS(mrt, vr.vifi)) { | |
1618 | vr.icount = vif->pkt_in; | |
1619 | vr.ocount = vif->pkt_out; | |
1620 | vr.ibytes = vif->bytes_in; | |
1621 | vr.obytes = vif->bytes_out; | |
1622 | read_unlock(&mrt_lock); | |
1623 | ||
1624 | if (copy_to_user(arg, &vr, sizeof(vr))) | |
1625 | return -EFAULT; | |
1626 | return 0; | |
1627 | } | |
1628 | read_unlock(&mrt_lock); | |
1629 | return -EADDRNOTAVAIL; | |
709b46e8 EB |
1630 | case SIOCGETSGCNT: |
1631 | if (copy_from_user(&sr, arg, sizeof(sr))) | |
1632 | return -EFAULT; | |
1633 | ||
1634 | rcu_read_lock(); | |
1635 | c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr); | |
1636 | if (c) { | |
1637 | sr.pktcnt = c->mfc_un.res.pkt; | |
1638 | sr.bytecnt = c->mfc_un.res.bytes; | |
1639 | sr.wrong_if = c->mfc_un.res.wrong_if; | |
1640 | rcu_read_unlock(); | |
1641 | ||
1642 | if (copy_to_user(arg, &sr, sizeof(sr))) | |
1643 | return -EFAULT; | |
1644 | return 0; | |
1645 | } | |
1646 | rcu_read_unlock(); | |
1647 | return -EADDRNOTAVAIL; | |
1648 | default: | |
1649 | return -ENOIOCTLCMD; | |
1650 | } | |
1651 | } | |
1652 | #endif | |
1653 | ||
1da177e4 LT |
1654 | static int ipmr_device_event(struct notifier_block *this, unsigned long event, void *ptr) |
1655 | { | |
351638e7 | 1656 | struct net_device *dev = netdev_notifier_info_to_dev(ptr); |
4feb88e5 | 1657 | struct net *net = dev_net(dev); |
f0ad0860 | 1658 | struct mr_table *mrt; |
1da177e4 LT |
1659 | struct vif_device *v; |
1660 | int ct; | |
e9dc8653 | 1661 | |
1da177e4 LT |
1662 | if (event != NETDEV_UNREGISTER) |
1663 | return NOTIFY_DONE; | |
f0ad0860 PM |
1664 | |
1665 | ipmr_for_each_table(mrt, net) { | |
1666 | v = &mrt->vif_table[0]; | |
1667 | for (ct = 0; ct < mrt->maxvif; ct++, v++) { | |
1668 | if (v->dev == dev) | |
e92036a6 | 1669 | vif_delete(mrt, ct, 1, NULL); |
f0ad0860 | 1670 | } |
1da177e4 LT |
1671 | } |
1672 | return NOTIFY_DONE; | |
1673 | } | |
1674 | ||
c354e124 | 1675 | static struct notifier_block ip_mr_notifier = { |
1da177e4 LT |
1676 | .notifier_call = ipmr_device_event, |
1677 | }; | |
1678 | ||
7ef8f65d NA |
1679 | /* Encapsulate a packet by attaching a valid IPIP header to it. |
1680 | * This avoids tunnel drivers and other mess and gives us the speed so | |
1681 | * important for multicast video. | |
1da177e4 | 1682 | */ |
b6a7719a HFS |
1683 | static void ip_encap(struct net *net, struct sk_buff *skb, |
1684 | __be32 saddr, __be32 daddr) | |
1da177e4 | 1685 | { |
8856dfa3 | 1686 | struct iphdr *iph; |
b71d1d42 | 1687 | const struct iphdr *old_iph = ip_hdr(skb); |
8856dfa3 ACM |
1688 | |
1689 | skb_push(skb, sizeof(struct iphdr)); | |
b0e380b1 | 1690 | skb->transport_header = skb->network_header; |
8856dfa3 | 1691 | skb_reset_network_header(skb); |
eddc9ec5 | 1692 | iph = ip_hdr(skb); |
1da177e4 | 1693 | |
a8cb16dd | 1694 | iph->version = 4; |
e023dd64 ACM |
1695 | iph->tos = old_iph->tos; |
1696 | iph->ttl = old_iph->ttl; | |
1da177e4 LT |
1697 | iph->frag_off = 0; |
1698 | iph->daddr = daddr; | |
1699 | iph->saddr = saddr; | |
1700 | iph->protocol = IPPROTO_IPIP; | |
1701 | iph->ihl = 5; | |
1702 | iph->tot_len = htons(skb->len); | |
b6a7719a | 1703 | ip_select_ident(net, skb, NULL); |
1da177e4 LT |
1704 | ip_send_check(iph); |
1705 | ||
1da177e4 LT |
1706 | memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt)); |
1707 | nf_reset(skb); | |
1708 | } | |
1709 | ||
0c4b51f0 EB |
1710 | static inline int ipmr_forward_finish(struct net *net, struct sock *sk, |
1711 | struct sk_buff *skb) | |
1da177e4 | 1712 | { |
a8cb16dd | 1713 | struct ip_options *opt = &(IPCB(skb)->opt); |
1da177e4 | 1714 | |
73186df8 DM |
1715 | IP_INC_STATS(net, IPSTATS_MIB_OUTFORWDATAGRAMS); |
1716 | IP_ADD_STATS(net, IPSTATS_MIB_OUTOCTETS, skb->len); | |
1da177e4 LT |
1717 | |
1718 | if (unlikely(opt->optlen)) | |
1719 | ip_forward_options(skb); | |
1720 | ||
13206b6b | 1721 | return dst_output(net, sk, skb); |
1da177e4 LT |
1722 | } |
1723 | ||
7ef8f65d | 1724 | /* Processing handlers for ipmr_forward */ |
1da177e4 | 1725 | |
0c12295a PM |
1726 | static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt, |
1727 | struct sk_buff *skb, struct mfc_cache *c, int vifi) | |
1da177e4 | 1728 | { |
eddc9ec5 | 1729 | const struct iphdr *iph = ip_hdr(skb); |
0c12295a | 1730 | struct vif_device *vif = &mrt->vif_table[vifi]; |
1da177e4 LT |
1731 | struct net_device *dev; |
1732 | struct rtable *rt; | |
31e4543d | 1733 | struct flowi4 fl4; |
1da177e4 LT |
1734 | int encap = 0; |
1735 | ||
51456b29 | 1736 | if (!vif->dev) |
1da177e4 LT |
1737 | goto out_free; |
1738 | ||
1da177e4 LT |
1739 | if (vif->flags & VIFF_REGISTER) { |
1740 | vif->pkt_out++; | |
c354e124 | 1741 | vif->bytes_out += skb->len; |
cf3677ae PE |
1742 | vif->dev->stats.tx_bytes += skb->len; |
1743 | vif->dev->stats.tx_packets++; | |
0c12295a | 1744 | ipmr_cache_report(mrt, skb, vifi, IGMPMSG_WHOLEPKT); |
69ebbf58 | 1745 | goto out_free; |
1da177e4 | 1746 | } |
1da177e4 | 1747 | |
a8cb16dd | 1748 | if (vif->flags & VIFF_TUNNEL) { |
31e4543d | 1749 | rt = ip_route_output_ports(net, &fl4, NULL, |
78fbfd8a DM |
1750 | vif->remote, vif->local, |
1751 | 0, 0, | |
1752 | IPPROTO_IPIP, | |
1753 | RT_TOS(iph->tos), vif->link); | |
b23dd4fe | 1754 | if (IS_ERR(rt)) |
1da177e4 LT |
1755 | goto out_free; |
1756 | encap = sizeof(struct iphdr); | |
1757 | } else { | |
31e4543d | 1758 | rt = ip_route_output_ports(net, &fl4, NULL, iph->daddr, 0, |
78fbfd8a DM |
1759 | 0, 0, |
1760 | IPPROTO_IPIP, | |
1761 | RT_TOS(iph->tos), vif->link); | |
b23dd4fe | 1762 | if (IS_ERR(rt)) |
1da177e4 LT |
1763 | goto out_free; |
1764 | } | |
1765 | ||
d8d1f30b | 1766 | dev = rt->dst.dev; |
1da177e4 | 1767 | |
d8d1f30b | 1768 | if (skb->len+encap > dst_mtu(&rt->dst) && (ntohs(iph->frag_off) & IP_DF)) { |
1da177e4 | 1769 | /* Do not fragment multicasts. Alas, IPv4 does not |
a8cb16dd ED |
1770 | * allow to send ICMP, so that packets will disappear |
1771 | * to blackhole. | |
1da177e4 | 1772 | */ |
73186df8 | 1773 | IP_INC_STATS(net, IPSTATS_MIB_FRAGFAILS); |
1da177e4 LT |
1774 | ip_rt_put(rt); |
1775 | goto out_free; | |
1776 | } | |
1777 | ||
d8d1f30b | 1778 | encap += LL_RESERVED_SPACE(dev) + rt->dst.header_len; |
1da177e4 LT |
1779 | |
1780 | if (skb_cow(skb, encap)) { | |
e905a9ed | 1781 | ip_rt_put(rt); |
1da177e4 LT |
1782 | goto out_free; |
1783 | } | |
1784 | ||
1785 | vif->pkt_out++; | |
c354e124 | 1786 | vif->bytes_out += skb->len; |
1da177e4 | 1787 | |
adf30907 | 1788 | skb_dst_drop(skb); |
d8d1f30b | 1789 | skb_dst_set(skb, &rt->dst); |
eddc9ec5 | 1790 | ip_decrease_ttl(ip_hdr(skb)); |
1da177e4 LT |
1791 | |
1792 | /* FIXME: forward and output firewalls used to be called here. | |
a8cb16dd ED |
1793 | * What do we do with netfilter? -- RR |
1794 | */ | |
1da177e4 | 1795 | if (vif->flags & VIFF_TUNNEL) { |
b6a7719a | 1796 | ip_encap(net, skb, vif->local, vif->remote); |
1da177e4 | 1797 | /* FIXME: extra output firewall step used to be here. --RR */ |
2f4c02d4 PE |
1798 | vif->dev->stats.tx_packets++; |
1799 | vif->dev->stats.tx_bytes += skb->len; | |
1da177e4 LT |
1800 | } |
1801 | ||
9ee6c5dc | 1802 | IPCB(skb)->flags |= IPSKB_FORWARDED; |
1da177e4 | 1803 | |
7ef8f65d | 1804 | /* RFC1584 teaches, that DVMRP/PIM router must deliver packets locally |
1da177e4 LT |
1805 | * not only before forwarding, but after forwarding on all output |
1806 | * interfaces. It is clear, if mrouter runs a multicasting | |
1807 | * program, it should receive packets not depending to what interface | |
1808 | * program is joined. | |
1809 | * If we will not make it, the program will have to join on all | |
1810 | * interfaces. On the other hand, multihoming host (or router, but | |
1811 | * not mrouter) cannot join to more than one interface - it will | |
1812 | * result in receiving multiple packets. | |
1813 | */ | |
29a26a56 EB |
1814 | NF_HOOK(NFPROTO_IPV4, NF_INET_FORWARD, |
1815 | net, NULL, skb, skb->dev, dev, | |
1da177e4 LT |
1816 | ipmr_forward_finish); |
1817 | return; | |
1818 | ||
1819 | out_free: | |
1820 | kfree_skb(skb); | |
1da177e4 LT |
1821 | } |
1822 | ||
0c12295a | 1823 | static int ipmr_find_vif(struct mr_table *mrt, struct net_device *dev) |
1da177e4 LT |
1824 | { |
1825 | int ct; | |
0c12295a PM |
1826 | |
1827 | for (ct = mrt->maxvif-1; ct >= 0; ct--) { | |
1828 | if (mrt->vif_table[ct].dev == dev) | |
1da177e4 LT |
1829 | break; |
1830 | } | |
1831 | return ct; | |
1832 | } | |
1833 | ||
1834 | /* "local" means that we should preserve one skb (for local delivery) */ | |
c4854ec8 RR |
1835 | static void ip_mr_forward(struct net *net, struct mr_table *mrt, |
1836 | struct sk_buff *skb, struct mfc_cache *cache, | |
1837 | int local) | |
1da177e4 | 1838 | { |
8fb472c0 | 1839 | int true_vifi = ipmr_find_vif(mrt, skb->dev); |
1da177e4 LT |
1840 | int psend = -1; |
1841 | int vif, ct; | |
1842 | ||
1843 | vif = cache->mfc_parent; | |
1844 | cache->mfc_un.res.pkt++; | |
1845 | cache->mfc_un.res.bytes += skb->len; | |
43b9e127 | 1846 | cache->mfc_un.res.lastuse = jiffies; |
1da177e4 | 1847 | |
360eb5da | 1848 | if (cache->mfc_origin == htonl(INADDR_ANY) && true_vifi >= 0) { |
660b26dc ND |
1849 | struct mfc_cache *cache_proxy; |
1850 | ||
1851 | /* For an (*,G) entry, we only check that the incomming | |
1852 | * interface is part of the static tree. | |
1853 | */ | |
1854 | cache_proxy = ipmr_cache_find_any_parent(mrt, vif); | |
1855 | if (cache_proxy && | |
1856 | cache_proxy->mfc_un.res.ttls[true_vifi] < 255) | |
1857 | goto forward; | |
1858 | } | |
1859 | ||
7ef8f65d | 1860 | /* Wrong interface: drop packet and (maybe) send PIM assert. */ |
0c12295a | 1861 | if (mrt->vif_table[vif].dev != skb->dev) { |
e58e4159 DA |
1862 | struct net_device *mdev; |
1863 | ||
1864 | mdev = l3mdev_master_dev_rcu(mrt->vif_table[vif].dev); | |
1865 | if (mdev == skb->dev) | |
1866 | goto forward; | |
1867 | ||
c7537967 | 1868 | if (rt_is_output_route(skb_rtable(skb))) { |
1da177e4 | 1869 | /* It is our own packet, looped back. |
a8cb16dd ED |
1870 | * Very complicated situation... |
1871 | * | |
1872 | * The best workaround until routing daemons will be | |
1873 | * fixed is not to redistribute packet, if it was | |
1874 | * send through wrong interface. It means, that | |
1875 | * multicast applications WILL NOT work for | |
1876 | * (S,G), which have default multicast route pointing | |
1877 | * to wrong oif. In any case, it is not a good | |
1878 | * idea to use multicasting applications on router. | |
1da177e4 LT |
1879 | */ |
1880 | goto dont_forward; | |
1881 | } | |
1882 | ||
1883 | cache->mfc_un.res.wrong_if++; | |
1da177e4 | 1884 | |
0c12295a | 1885 | if (true_vifi >= 0 && mrt->mroute_do_assert && |
1da177e4 | 1886 | /* pimsm uses asserts, when switching from RPT to SPT, |
a8cb16dd ED |
1887 | * so that we cannot check that packet arrived on an oif. |
1888 | * It is bad, but otherwise we would need to move pretty | |
1889 | * large chunk of pimd to kernel. Ough... --ANK | |
1da177e4 | 1890 | */ |
0c12295a | 1891 | (mrt->mroute_do_pim || |
6f9374a9 | 1892 | cache->mfc_un.res.ttls[true_vifi] < 255) && |
e905a9ed | 1893 | time_after(jiffies, |
1da177e4 LT |
1894 | cache->mfc_un.res.last_assert + MFC_ASSERT_THRESH)) { |
1895 | cache->mfc_un.res.last_assert = jiffies; | |
0c12295a | 1896 | ipmr_cache_report(mrt, skb, true_vifi, IGMPMSG_WRONGVIF); |
1da177e4 LT |
1897 | } |
1898 | goto dont_forward; | |
1899 | } | |
1900 | ||
660b26dc | 1901 | forward: |
0c12295a PM |
1902 | mrt->vif_table[vif].pkt_in++; |
1903 | mrt->vif_table[vif].bytes_in += skb->len; | |
1da177e4 | 1904 | |
7ef8f65d | 1905 | /* Forward the frame */ |
360eb5da ND |
1906 | if (cache->mfc_origin == htonl(INADDR_ANY) && |
1907 | cache->mfc_mcastgrp == htonl(INADDR_ANY)) { | |
660b26dc ND |
1908 | if (true_vifi >= 0 && |
1909 | true_vifi != cache->mfc_parent && | |
1910 | ip_hdr(skb)->ttl > | |
1911 | cache->mfc_un.res.ttls[cache->mfc_parent]) { | |
1912 | /* It's an (*,*) entry and the packet is not coming from | |
1913 | * the upstream: forward the packet to the upstream | |
1914 | * only. | |
1915 | */ | |
1916 | psend = cache->mfc_parent; | |
1917 | goto last_forward; | |
1918 | } | |
1919 | goto dont_forward; | |
1920 | } | |
a8cb16dd ED |
1921 | for (ct = cache->mfc_un.res.maxvif - 1; |
1922 | ct >= cache->mfc_un.res.minvif; ct--) { | |
660b26dc | 1923 | /* For (*,G) entry, don't forward to the incoming interface */ |
360eb5da ND |
1924 | if ((cache->mfc_origin != htonl(INADDR_ANY) || |
1925 | ct != true_vifi) && | |
660b26dc | 1926 | ip_hdr(skb)->ttl > cache->mfc_un.res.ttls[ct]) { |
1da177e4 LT |
1927 | if (psend != -1) { |
1928 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
a8cb16dd | 1929 | |
1da177e4 | 1930 | if (skb2) |
0c12295a PM |
1931 | ipmr_queue_xmit(net, mrt, skb2, cache, |
1932 | psend); | |
1da177e4 | 1933 | } |
c354e124 | 1934 | psend = ct; |
1da177e4 LT |
1935 | } |
1936 | } | |
660b26dc | 1937 | last_forward: |
1da177e4 LT |
1938 | if (psend != -1) { |
1939 | if (local) { | |
1940 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
a8cb16dd | 1941 | |
1da177e4 | 1942 | if (skb2) |
0c12295a | 1943 | ipmr_queue_xmit(net, mrt, skb2, cache, psend); |
1da177e4 | 1944 | } else { |
0c12295a | 1945 | ipmr_queue_xmit(net, mrt, skb, cache, psend); |
c4854ec8 | 1946 | return; |
1da177e4 LT |
1947 | } |
1948 | } | |
1949 | ||
1950 | dont_forward: | |
1951 | if (!local) | |
1952 | kfree_skb(skb); | |
1da177e4 LT |
1953 | } |
1954 | ||
417da66f | 1955 | static struct mr_table *ipmr_rt_fib_lookup(struct net *net, struct sk_buff *skb) |
ee3f1aaf | 1956 | { |
417da66f DM |
1957 | struct rtable *rt = skb_rtable(skb); |
1958 | struct iphdr *iph = ip_hdr(skb); | |
da91981b | 1959 | struct flowi4 fl4 = { |
417da66f DM |
1960 | .daddr = iph->daddr, |
1961 | .saddr = iph->saddr, | |
b0fe4a31 | 1962 | .flowi4_tos = RT_TOS(iph->tos), |
4fd551d7 DM |
1963 | .flowi4_oif = (rt_is_output_route(rt) ? |
1964 | skb->dev->ifindex : 0), | |
1965 | .flowi4_iif = (rt_is_output_route(rt) ? | |
1fb9489b | 1966 | LOOPBACK_IFINDEX : |
4fd551d7 | 1967 | skb->dev->ifindex), |
b4869889 | 1968 | .flowi4_mark = skb->mark, |
ee3f1aaf DM |
1969 | }; |
1970 | struct mr_table *mrt; | |
1971 | int err; | |
1972 | ||
da91981b | 1973 | err = ipmr_fib_lookup(net, &fl4, &mrt); |
ee3f1aaf DM |
1974 | if (err) |
1975 | return ERR_PTR(err); | |
1976 | return mrt; | |
1977 | } | |
1da177e4 | 1978 | |
7ef8f65d NA |
1979 | /* Multicast packets for forwarding arrive here |
1980 | * Called with rcu_read_lock(); | |
1da177e4 | 1981 | */ |
1da177e4 LT |
1982 | int ip_mr_input(struct sk_buff *skb) |
1983 | { | |
1984 | struct mfc_cache *cache; | |
4feb88e5 | 1985 | struct net *net = dev_net(skb->dev); |
511c3f92 | 1986 | int local = skb_rtable(skb)->rt_flags & RTCF_LOCAL; |
f0ad0860 | 1987 | struct mr_table *mrt; |
1da177e4 LT |
1988 | |
1989 | /* Packet is looped back after forward, it should not be | |
a8cb16dd | 1990 | * forwarded second time, but still can be delivered locally. |
1da177e4 | 1991 | */ |
4c968709 | 1992 | if (IPCB(skb)->flags & IPSKB_FORWARDED) |
1da177e4 LT |
1993 | goto dont_forward; |
1994 | ||
417da66f | 1995 | mrt = ipmr_rt_fib_lookup(net, skb); |
ee3f1aaf DM |
1996 | if (IS_ERR(mrt)) { |
1997 | kfree_skb(skb); | |
1998 | return PTR_ERR(mrt); | |
e40dbc51 | 1999 | } |
1da177e4 | 2000 | if (!local) { |
4c968709 ED |
2001 | if (IPCB(skb)->opt.router_alert) { |
2002 | if (ip_call_ra_chain(skb)) | |
2003 | return 0; | |
2004 | } else if (ip_hdr(skb)->protocol == IPPROTO_IGMP) { | |
2005 | /* IGMPv1 (and broken IGMPv2 implementations sort of | |
2006 | * Cisco IOS <= 11.2(8)) do not put router alert | |
2007 | * option to IGMP packets destined to routable | |
2008 | * groups. It is very bad, because it means | |
2009 | * that we can forward NO IGMP messages. | |
2010 | */ | |
2011 | struct sock *mroute_sk; | |
2012 | ||
2013 | mroute_sk = rcu_dereference(mrt->mroute_sk); | |
2014 | if (mroute_sk) { | |
2015 | nf_reset(skb); | |
2016 | raw_rcv(mroute_sk, skb); | |
2017 | return 0; | |
2018 | } | |
1da177e4 LT |
2019 | } |
2020 | } | |
2021 | ||
a8c9486b | 2022 | /* already under rcu_read_lock() */ |
0c12295a | 2023 | cache = ipmr_cache_find(mrt, ip_hdr(skb)->saddr, ip_hdr(skb)->daddr); |
51456b29 | 2024 | if (!cache) { |
660b26dc ND |
2025 | int vif = ipmr_find_vif(mrt, skb->dev); |
2026 | ||
2027 | if (vif >= 0) | |
2028 | cache = ipmr_cache_find_any(mrt, ip_hdr(skb)->daddr, | |
2029 | vif); | |
2030 | } | |
1da177e4 | 2031 | |
7ef8f65d | 2032 | /* No usable cache entry */ |
51456b29 | 2033 | if (!cache) { |
1da177e4 LT |
2034 | int vif; |
2035 | ||
2036 | if (local) { | |
2037 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
2038 | ip_local_deliver(skb); | |
51456b29 | 2039 | if (!skb2) |
1da177e4 | 2040 | return -ENOBUFS; |
1da177e4 LT |
2041 | skb = skb2; |
2042 | } | |
2043 | ||
a8c9486b | 2044 | read_lock(&mrt_lock); |
0c12295a | 2045 | vif = ipmr_find_vif(mrt, skb->dev); |
1da177e4 | 2046 | if (vif >= 0) { |
0eae88f3 | 2047 | int err2 = ipmr_cache_unresolved(mrt, vif, skb); |
1da177e4 LT |
2048 | read_unlock(&mrt_lock); |
2049 | ||
0eae88f3 | 2050 | return err2; |
1da177e4 LT |
2051 | } |
2052 | read_unlock(&mrt_lock); | |
2053 | kfree_skb(skb); | |
2054 | return -ENODEV; | |
2055 | } | |
2056 | ||
a8c9486b | 2057 | read_lock(&mrt_lock); |
0c12295a | 2058 | ip_mr_forward(net, mrt, skb, cache, local); |
1da177e4 LT |
2059 | read_unlock(&mrt_lock); |
2060 | ||
2061 | if (local) | |
2062 | return ip_local_deliver(skb); | |
2063 | ||
2064 | return 0; | |
2065 | ||
2066 | dont_forward: | |
2067 | if (local) | |
2068 | return ip_local_deliver(skb); | |
2069 | kfree_skb(skb); | |
2070 | return 0; | |
2071 | } | |
2072 | ||
b1879204 | 2073 | #ifdef CONFIG_IP_PIMSM_V1 |
7ef8f65d | 2074 | /* Handle IGMP messages of PIMv1 */ |
a8cb16dd | 2075 | int pim_rcv_v1(struct sk_buff *skb) |
b1879204 IJ |
2076 | { |
2077 | struct igmphdr *pim; | |
4feb88e5 | 2078 | struct net *net = dev_net(skb->dev); |
f0ad0860 | 2079 | struct mr_table *mrt; |
b1879204 IJ |
2080 | |
2081 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) | |
2082 | goto drop; | |
2083 | ||
2084 | pim = igmp_hdr(skb); | |
2085 | ||
417da66f | 2086 | mrt = ipmr_rt_fib_lookup(net, skb); |
ee3f1aaf DM |
2087 | if (IS_ERR(mrt)) |
2088 | goto drop; | |
0c12295a | 2089 | if (!mrt->mroute_do_pim || |
b1879204 IJ |
2090 | pim->group != PIM_V1_VERSION || pim->code != PIM_V1_REGISTER) |
2091 | goto drop; | |
2092 | ||
f0ad0860 | 2093 | if (__pim_rcv(mrt, skb, sizeof(*pim))) { |
b1879204 IJ |
2094 | drop: |
2095 | kfree_skb(skb); | |
2096 | } | |
1da177e4 LT |
2097 | return 0; |
2098 | } | |
2099 | #endif | |
2100 | ||
2101 | #ifdef CONFIG_IP_PIMSM_V2 | |
a8cb16dd | 2102 | static int pim_rcv(struct sk_buff *skb) |
1da177e4 LT |
2103 | { |
2104 | struct pimreghdr *pim; | |
f0ad0860 PM |
2105 | struct net *net = dev_net(skb->dev); |
2106 | struct mr_table *mrt; | |
1da177e4 | 2107 | |
b1879204 | 2108 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) |
1da177e4 LT |
2109 | goto drop; |
2110 | ||
9c70220b | 2111 | pim = (struct pimreghdr *)skb_transport_header(skb); |
56245cae | 2112 | if (pim->type != ((PIM_VERSION << 4) | (PIM_TYPE_REGISTER)) || |
a8cb16dd | 2113 | (pim->flags & PIM_NULL_REGISTER) || |
e905a9ed | 2114 | (ip_compute_csum((void *)pim, sizeof(*pim)) != 0 && |
d3bc23e7 | 2115 | csum_fold(skb_checksum(skb, 0, skb->len, 0)))) |
1da177e4 LT |
2116 | goto drop; |
2117 | ||
417da66f | 2118 | mrt = ipmr_rt_fib_lookup(net, skb); |
ee3f1aaf DM |
2119 | if (IS_ERR(mrt)) |
2120 | goto drop; | |
f0ad0860 | 2121 | if (__pim_rcv(mrt, skb, sizeof(*pim))) { |
b1879204 IJ |
2122 | drop: |
2123 | kfree_skb(skb); | |
2124 | } | |
1da177e4 LT |
2125 | return 0; |
2126 | } | |
2127 | #endif | |
2128 | ||
cb6a4e46 PM |
2129 | static int __ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
2130 | struct mfc_cache *c, struct rtmsg *rtm) | |
1da177e4 | 2131 | { |
adfa85e4 | 2132 | struct rta_mfc_stats mfcs; |
43b9e127 NA |
2133 | struct nlattr *mp_attr; |
2134 | struct rtnexthop *nhp; | |
b5036cd4 | 2135 | unsigned long lastuse; |
43b9e127 | 2136 | int ct; |
1da177e4 | 2137 | |
7438189b | 2138 | /* If cache is unresolved, don't try to parse IIF and OIF */ |
1708ebc9 NA |
2139 | if (c->mfc_parent >= MAXVIFS) { |
2140 | rtm->rtm_flags |= RTNH_F_UNRESOLVED; | |
7438189b | 2141 | return -ENOENT; |
1708ebc9 | 2142 | } |
7438189b | 2143 | |
92a395e5 TG |
2144 | if (VIF_EXISTS(mrt, c->mfc_parent) && |
2145 | nla_put_u32(skb, RTA_IIF, mrt->vif_table[c->mfc_parent].dev->ifindex) < 0) | |
2146 | return -EMSGSIZE; | |
1da177e4 | 2147 | |
92a395e5 TG |
2148 | if (!(mp_attr = nla_nest_start(skb, RTA_MULTIPATH))) |
2149 | return -EMSGSIZE; | |
1da177e4 LT |
2150 | |
2151 | for (ct = c->mfc_un.res.minvif; ct < c->mfc_un.res.maxvif; ct++) { | |
0c12295a | 2152 | if (VIF_EXISTS(mrt, ct) && c->mfc_un.res.ttls[ct] < 255) { |
92a395e5 TG |
2153 | if (!(nhp = nla_reserve_nohdr(skb, sizeof(*nhp)))) { |
2154 | nla_nest_cancel(skb, mp_attr); | |
2155 | return -EMSGSIZE; | |
2156 | } | |
2157 | ||
1da177e4 LT |
2158 | nhp->rtnh_flags = 0; |
2159 | nhp->rtnh_hops = c->mfc_un.res.ttls[ct]; | |
0c12295a | 2160 | nhp->rtnh_ifindex = mrt->vif_table[ct].dev->ifindex; |
1da177e4 LT |
2161 | nhp->rtnh_len = sizeof(*nhp); |
2162 | } | |
2163 | } | |
92a395e5 TG |
2164 | |
2165 | nla_nest_end(skb, mp_attr); | |
2166 | ||
b5036cd4 NA |
2167 | lastuse = READ_ONCE(c->mfc_un.res.lastuse); |
2168 | lastuse = time_after_eq(jiffies, lastuse) ? jiffies - lastuse : 0; | |
2169 | ||
adfa85e4 ND |
2170 | mfcs.mfcs_packets = c->mfc_un.res.pkt; |
2171 | mfcs.mfcs_bytes = c->mfc_un.res.bytes; | |
2172 | mfcs.mfcs_wrong_if = c->mfc_un.res.wrong_if; | |
43b9e127 | 2173 | if (nla_put_64bit(skb, RTA_MFC_STATS, sizeof(mfcs), &mfcs, RTA_PAD) || |
b5036cd4 | 2174 | nla_put_u64_64bit(skb, RTA_EXPIRES, jiffies_to_clock_t(lastuse), |
43b9e127 | 2175 | RTA_PAD)) |
adfa85e4 ND |
2176 | return -EMSGSIZE; |
2177 | ||
1da177e4 LT |
2178 | rtm->rtm_type = RTN_MULTICAST; |
2179 | return 1; | |
1da177e4 LT |
2180 | } |
2181 | ||
9a1b9496 DM |
2182 | int ipmr_get_route(struct net *net, struct sk_buff *skb, |
2183 | __be32 saddr, __be32 daddr, | |
9f09eaea | 2184 | struct rtmsg *rtm, u32 portid) |
1da177e4 | 2185 | { |
1da177e4 | 2186 | struct mfc_cache *cache; |
9a1b9496 DM |
2187 | struct mr_table *mrt; |
2188 | int err; | |
1da177e4 | 2189 | |
f0ad0860 | 2190 | mrt = ipmr_get_table(net, RT_TABLE_DEFAULT); |
51456b29 | 2191 | if (!mrt) |
f0ad0860 PM |
2192 | return -ENOENT; |
2193 | ||
a8c9486b | 2194 | rcu_read_lock(); |
9a1b9496 | 2195 | cache = ipmr_cache_find(mrt, saddr, daddr); |
51456b29 | 2196 | if (!cache && skb->dev) { |
660b26dc | 2197 | int vif = ipmr_find_vif(mrt, skb->dev); |
1da177e4 | 2198 | |
660b26dc ND |
2199 | if (vif >= 0) |
2200 | cache = ipmr_cache_find_any(mrt, daddr, vif); | |
2201 | } | |
51456b29 | 2202 | if (!cache) { |
72287490 | 2203 | struct sk_buff *skb2; |
eddc9ec5 | 2204 | struct iphdr *iph; |
1da177e4 | 2205 | struct net_device *dev; |
a8cb16dd | 2206 | int vif = -1; |
1da177e4 | 2207 | |
1da177e4 | 2208 | dev = skb->dev; |
a8c9486b | 2209 | read_lock(&mrt_lock); |
a8cb16dd ED |
2210 | if (dev) |
2211 | vif = ipmr_find_vif(mrt, dev); | |
2212 | if (vif < 0) { | |
1da177e4 | 2213 | read_unlock(&mrt_lock); |
a8c9486b | 2214 | rcu_read_unlock(); |
1da177e4 LT |
2215 | return -ENODEV; |
2216 | } | |
72287490 AK |
2217 | skb2 = skb_clone(skb, GFP_ATOMIC); |
2218 | if (!skb2) { | |
2219 | read_unlock(&mrt_lock); | |
a8c9486b | 2220 | rcu_read_unlock(); |
72287490 AK |
2221 | return -ENOMEM; |
2222 | } | |
2223 | ||
2cf75070 | 2224 | NETLINK_CB(skb2).portid = portid; |
e2d1bca7 ACM |
2225 | skb_push(skb2, sizeof(struct iphdr)); |
2226 | skb_reset_network_header(skb2); | |
eddc9ec5 ACM |
2227 | iph = ip_hdr(skb2); |
2228 | iph->ihl = sizeof(struct iphdr) >> 2; | |
9a1b9496 DM |
2229 | iph->saddr = saddr; |
2230 | iph->daddr = daddr; | |
eddc9ec5 | 2231 | iph->version = 0; |
0c12295a | 2232 | err = ipmr_cache_unresolved(mrt, vif, skb2); |
1da177e4 | 2233 | read_unlock(&mrt_lock); |
a8c9486b | 2234 | rcu_read_unlock(); |
1da177e4 LT |
2235 | return err; |
2236 | } | |
2237 | ||
a8c9486b | 2238 | read_lock(&mrt_lock); |
cb6a4e46 | 2239 | err = __ipmr_fill_mroute(mrt, skb, cache, rtm); |
1da177e4 | 2240 | read_unlock(&mrt_lock); |
a8c9486b | 2241 | rcu_read_unlock(); |
1da177e4 LT |
2242 | return err; |
2243 | } | |
2244 | ||
cb6a4e46 | 2245 | static int ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
65886f43 ND |
2246 | u32 portid, u32 seq, struct mfc_cache *c, int cmd, |
2247 | int flags) | |
cb6a4e46 PM |
2248 | { |
2249 | struct nlmsghdr *nlh; | |
2250 | struct rtmsg *rtm; | |
1eb99af5 | 2251 | int err; |
cb6a4e46 | 2252 | |
65886f43 | 2253 | nlh = nlmsg_put(skb, portid, seq, cmd, sizeof(*rtm), flags); |
51456b29 | 2254 | if (!nlh) |
cb6a4e46 PM |
2255 | return -EMSGSIZE; |
2256 | ||
2257 | rtm = nlmsg_data(nlh); | |
2258 | rtm->rtm_family = RTNL_FAMILY_IPMR; | |
2259 | rtm->rtm_dst_len = 32; | |
2260 | rtm->rtm_src_len = 32; | |
2261 | rtm->rtm_tos = 0; | |
2262 | rtm->rtm_table = mrt->id; | |
f3756b79 DM |
2263 | if (nla_put_u32(skb, RTA_TABLE, mrt->id)) |
2264 | goto nla_put_failure; | |
cb6a4e46 PM |
2265 | rtm->rtm_type = RTN_MULTICAST; |
2266 | rtm->rtm_scope = RT_SCOPE_UNIVERSE; | |
9a68ac72 ND |
2267 | if (c->mfc_flags & MFC_STATIC) |
2268 | rtm->rtm_protocol = RTPROT_STATIC; | |
2269 | else | |
2270 | rtm->rtm_protocol = RTPROT_MROUTED; | |
cb6a4e46 PM |
2271 | rtm->rtm_flags = 0; |
2272 | ||
930345ea JB |
2273 | if (nla_put_in_addr(skb, RTA_SRC, c->mfc_origin) || |
2274 | nla_put_in_addr(skb, RTA_DST, c->mfc_mcastgrp)) | |
f3756b79 | 2275 | goto nla_put_failure; |
1eb99af5 ND |
2276 | err = __ipmr_fill_mroute(mrt, skb, c, rtm); |
2277 | /* do not break the dump if cache is unresolved */ | |
2278 | if (err < 0 && err != -ENOENT) | |
cb6a4e46 PM |
2279 | goto nla_put_failure; |
2280 | ||
053c095a JB |
2281 | nlmsg_end(skb, nlh); |
2282 | return 0; | |
cb6a4e46 PM |
2283 | |
2284 | nla_put_failure: | |
2285 | nlmsg_cancel(skb, nlh); | |
2286 | return -EMSGSIZE; | |
2287 | } | |
2288 | ||
8cd3ac9f ND |
2289 | static size_t mroute_msgsize(bool unresolved, int maxvif) |
2290 | { | |
2291 | size_t len = | |
2292 | NLMSG_ALIGN(sizeof(struct rtmsg)) | |
2293 | + nla_total_size(4) /* RTA_TABLE */ | |
2294 | + nla_total_size(4) /* RTA_SRC */ | |
2295 | + nla_total_size(4) /* RTA_DST */ | |
2296 | ; | |
2297 | ||
2298 | if (!unresolved) | |
2299 | len = len | |
2300 | + nla_total_size(4) /* RTA_IIF */ | |
2301 | + nla_total_size(0) /* RTA_MULTIPATH */ | |
2302 | + maxvif * NLA_ALIGN(sizeof(struct rtnexthop)) | |
2303 | /* RTA_MFC_STATS */ | |
a9a08042 | 2304 | + nla_total_size_64bit(sizeof(struct rta_mfc_stats)) |
8cd3ac9f ND |
2305 | ; |
2306 | ||
2307 | return len; | |
2308 | } | |
2309 | ||
2310 | static void mroute_netlink_event(struct mr_table *mrt, struct mfc_cache *mfc, | |
2311 | int cmd) | |
2312 | { | |
2313 | struct net *net = read_pnet(&mrt->net); | |
2314 | struct sk_buff *skb; | |
2315 | int err = -ENOBUFS; | |
2316 | ||
2317 | skb = nlmsg_new(mroute_msgsize(mfc->mfc_parent >= MAXVIFS, mrt->maxvif), | |
2318 | GFP_ATOMIC); | |
51456b29 | 2319 | if (!skb) |
8cd3ac9f ND |
2320 | goto errout; |
2321 | ||
65886f43 | 2322 | err = ipmr_fill_mroute(mrt, skb, 0, 0, mfc, cmd, 0); |
8cd3ac9f ND |
2323 | if (err < 0) |
2324 | goto errout; | |
2325 | ||
2326 | rtnl_notify(skb, net, 0, RTNLGRP_IPV4_MROUTE, NULL, GFP_ATOMIC); | |
2327 | return; | |
2328 | ||
2329 | errout: | |
2330 | kfree_skb(skb); | |
2331 | if (err < 0) | |
2332 | rtnl_set_sk_err(net, RTNLGRP_IPV4_MROUTE, err); | |
2333 | } | |
2334 | ||
cb6a4e46 PM |
2335 | static int ipmr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb) |
2336 | { | |
2337 | struct net *net = sock_net(skb->sk); | |
2338 | struct mr_table *mrt; | |
2339 | struct mfc_cache *mfc; | |
2340 | unsigned int t = 0, s_t; | |
cb6a4e46 PM |
2341 | unsigned int e = 0, s_e; |
2342 | ||
2343 | s_t = cb->args[0]; | |
8fb472c0 | 2344 | s_e = cb->args[1]; |
cb6a4e46 | 2345 | |
a8c9486b | 2346 | rcu_read_lock(); |
cb6a4e46 PM |
2347 | ipmr_for_each_table(mrt, net) { |
2348 | if (t < s_t) | |
2349 | goto next_table; | |
8fb472c0 NA |
2350 | list_for_each_entry_rcu(mfc, &mrt->mfc_cache_list, list) { |
2351 | if (e < s_e) | |
2352 | goto next_entry; | |
2353 | if (ipmr_fill_mroute(mrt, skb, | |
2354 | NETLINK_CB(cb->skb).portid, | |
2355 | cb->nlh->nlmsg_seq, | |
2356 | mfc, RTM_NEWROUTE, | |
2357 | NLM_F_MULTI) < 0) | |
2358 | goto done; | |
cb6a4e46 | 2359 | next_entry: |
8fb472c0 | 2360 | e++; |
cb6a4e46 | 2361 | } |
8fb472c0 NA |
2362 | e = 0; |
2363 | s_e = 0; | |
2364 | ||
1eb99af5 ND |
2365 | spin_lock_bh(&mfc_unres_lock); |
2366 | list_for_each_entry(mfc, &mrt->mfc_unres_queue, list) { | |
2367 | if (e < s_e) | |
2368 | goto next_entry2; | |
2369 | if (ipmr_fill_mroute(mrt, skb, | |
2370 | NETLINK_CB(cb->skb).portid, | |
2371 | cb->nlh->nlmsg_seq, | |
65886f43 ND |
2372 | mfc, RTM_NEWROUTE, |
2373 | NLM_F_MULTI) < 0) { | |
1eb99af5 ND |
2374 | spin_unlock_bh(&mfc_unres_lock); |
2375 | goto done; | |
2376 | } | |
2377 | next_entry2: | |
2378 | e++; | |
2379 | } | |
2380 | spin_unlock_bh(&mfc_unres_lock); | |
8fb472c0 NA |
2381 | e = 0; |
2382 | s_e = 0; | |
cb6a4e46 PM |
2383 | next_table: |
2384 | t++; | |
2385 | } | |
2386 | done: | |
a8c9486b | 2387 | rcu_read_unlock(); |
cb6a4e46 | 2388 | |
8fb472c0 | 2389 | cb->args[1] = e; |
cb6a4e46 PM |
2390 | cb->args[0] = t; |
2391 | ||
2392 | return skb->len; | |
2393 | } | |
2394 | ||
ccbb0aa6 NA |
2395 | static const struct nla_policy rtm_ipmr_policy[RTA_MAX + 1] = { |
2396 | [RTA_SRC] = { .type = NLA_U32 }, | |
2397 | [RTA_DST] = { .type = NLA_U32 }, | |
2398 | [RTA_IIF] = { .type = NLA_U32 }, | |
2399 | [RTA_TABLE] = { .type = NLA_U32 }, | |
2400 | [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) }, | |
2401 | }; | |
2402 | ||
2403 | static bool ipmr_rtm_validate_proto(unsigned char rtm_protocol) | |
2404 | { | |
2405 | switch (rtm_protocol) { | |
2406 | case RTPROT_STATIC: | |
2407 | case RTPROT_MROUTED: | |
2408 | return true; | |
2409 | } | |
2410 | return false; | |
2411 | } | |
2412 | ||
2413 | static int ipmr_nla_get_ttls(const struct nlattr *nla, struct mfcctl *mfcc) | |
2414 | { | |
2415 | struct rtnexthop *rtnh = nla_data(nla); | |
2416 | int remaining = nla_len(nla), vifi = 0; | |
2417 | ||
2418 | while (rtnh_ok(rtnh, remaining)) { | |
2419 | mfcc->mfcc_ttls[vifi] = rtnh->rtnh_hops; | |
2420 | if (++vifi == MAXVIFS) | |
2421 | break; | |
2422 | rtnh = rtnh_next(rtnh, &remaining); | |
2423 | } | |
2424 | ||
2425 | return remaining > 0 ? -EINVAL : vifi; | |
2426 | } | |
2427 | ||
2428 | /* returns < 0 on error, 0 for ADD_MFC and 1 for ADD_MFC_PROXY */ | |
2429 | static int rtm_to_ipmr_mfcc(struct net *net, struct nlmsghdr *nlh, | |
2430 | struct mfcctl *mfcc, int *mrtsock, | |
2431 | struct mr_table **mrtret) | |
2432 | { | |
2433 | struct net_device *dev = NULL; | |
2434 | u32 tblid = RT_TABLE_DEFAULT; | |
2435 | struct mr_table *mrt; | |
2436 | struct nlattr *attr; | |
2437 | struct rtmsg *rtm; | |
2438 | int ret, rem; | |
2439 | ||
2440 | ret = nlmsg_validate(nlh, sizeof(*rtm), RTA_MAX, rtm_ipmr_policy); | |
2441 | if (ret < 0) | |
2442 | goto out; | |
2443 | rtm = nlmsg_data(nlh); | |
2444 | ||
2445 | ret = -EINVAL; | |
2446 | if (rtm->rtm_family != RTNL_FAMILY_IPMR || rtm->rtm_dst_len != 32 || | |
2447 | rtm->rtm_type != RTN_MULTICAST || | |
2448 | rtm->rtm_scope != RT_SCOPE_UNIVERSE || | |
2449 | !ipmr_rtm_validate_proto(rtm->rtm_protocol)) | |
2450 | goto out; | |
2451 | ||
2452 | memset(mfcc, 0, sizeof(*mfcc)); | |
2453 | mfcc->mfcc_parent = -1; | |
2454 | ret = 0; | |
2455 | nlmsg_for_each_attr(attr, nlh, sizeof(struct rtmsg), rem) { | |
2456 | switch (nla_type(attr)) { | |
2457 | case RTA_SRC: | |
2458 | mfcc->mfcc_origin.s_addr = nla_get_be32(attr); | |
2459 | break; | |
2460 | case RTA_DST: | |
2461 | mfcc->mfcc_mcastgrp.s_addr = nla_get_be32(attr); | |
2462 | break; | |
2463 | case RTA_IIF: | |
2464 | dev = __dev_get_by_index(net, nla_get_u32(attr)); | |
2465 | if (!dev) { | |
2466 | ret = -ENODEV; | |
2467 | goto out; | |
2468 | } | |
2469 | break; | |
2470 | case RTA_MULTIPATH: | |
2471 | if (ipmr_nla_get_ttls(attr, mfcc) < 0) { | |
2472 | ret = -EINVAL; | |
2473 | goto out; | |
2474 | } | |
2475 | break; | |
2476 | case RTA_PREFSRC: | |
2477 | ret = 1; | |
2478 | break; | |
2479 | case RTA_TABLE: | |
2480 | tblid = nla_get_u32(attr); | |
2481 | break; | |
2482 | } | |
2483 | } | |
2484 | mrt = ipmr_get_table(net, tblid); | |
2485 | if (!mrt) { | |
2486 | ret = -ENOENT; | |
2487 | goto out; | |
2488 | } | |
2489 | *mrtret = mrt; | |
2490 | *mrtsock = rtm->rtm_protocol == RTPROT_MROUTED ? 1 : 0; | |
2491 | if (dev) | |
2492 | mfcc->mfcc_parent = ipmr_find_vif(mrt, dev); | |
2493 | ||
2494 | out: | |
2495 | return ret; | |
2496 | } | |
2497 | ||
2498 | /* takes care of both newroute and delroute */ | |
2499 | static int ipmr_rtm_route(struct sk_buff *skb, struct nlmsghdr *nlh) | |
2500 | { | |
2501 | struct net *net = sock_net(skb->sk); | |
2502 | int ret, mrtsock, parent; | |
2503 | struct mr_table *tbl; | |
2504 | struct mfcctl mfcc; | |
2505 | ||
2506 | mrtsock = 0; | |
2507 | tbl = NULL; | |
2508 | ret = rtm_to_ipmr_mfcc(net, nlh, &mfcc, &mrtsock, &tbl); | |
2509 | if (ret < 0) | |
2510 | return ret; | |
2511 | ||
2512 | parent = ret ? mfcc.mfcc_parent : -1; | |
2513 | if (nlh->nlmsg_type == RTM_NEWROUTE) | |
2514 | return ipmr_mfc_add(net, tbl, &mfcc, mrtsock, parent); | |
2515 | else | |
2516 | return ipmr_mfc_delete(tbl, &mfcc, parent); | |
2517 | } | |
2518 | ||
e905a9ed | 2519 | #ifdef CONFIG_PROC_FS |
7ef8f65d NA |
2520 | /* The /proc interfaces to multicast routing : |
2521 | * /proc/net/ip_mr_cache & /proc/net/ip_mr_vif | |
1da177e4 LT |
2522 | */ |
2523 | struct ipmr_vif_iter { | |
f6bb4514 | 2524 | struct seq_net_private p; |
f0ad0860 | 2525 | struct mr_table *mrt; |
1da177e4 LT |
2526 | int ct; |
2527 | }; | |
2528 | ||
f6bb4514 BT |
2529 | static struct vif_device *ipmr_vif_seq_idx(struct net *net, |
2530 | struct ipmr_vif_iter *iter, | |
1da177e4 LT |
2531 | loff_t pos) |
2532 | { | |
f0ad0860 | 2533 | struct mr_table *mrt = iter->mrt; |
0c12295a PM |
2534 | |
2535 | for (iter->ct = 0; iter->ct < mrt->maxvif; ++iter->ct) { | |
2536 | if (!VIF_EXISTS(mrt, iter->ct)) | |
1da177e4 | 2537 | continue; |
e905a9ed | 2538 | if (pos-- == 0) |
0c12295a | 2539 | return &mrt->vif_table[iter->ct]; |
1da177e4 LT |
2540 | } |
2541 | return NULL; | |
2542 | } | |
2543 | ||
2544 | static void *ipmr_vif_seq_start(struct seq_file *seq, loff_t *pos) | |
ba93ef74 | 2545 | __acquires(mrt_lock) |
1da177e4 | 2546 | { |
f0ad0860 | 2547 | struct ipmr_vif_iter *iter = seq->private; |
f6bb4514 | 2548 | struct net *net = seq_file_net(seq); |
f0ad0860 PM |
2549 | struct mr_table *mrt; |
2550 | ||
2551 | mrt = ipmr_get_table(net, RT_TABLE_DEFAULT); | |
51456b29 | 2552 | if (!mrt) |
f0ad0860 PM |
2553 | return ERR_PTR(-ENOENT); |
2554 | ||
2555 | iter->mrt = mrt; | |
f6bb4514 | 2556 | |
1da177e4 | 2557 | read_lock(&mrt_lock); |
f6bb4514 | 2558 | return *pos ? ipmr_vif_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
2559 | : SEQ_START_TOKEN; |
2560 | } | |
2561 | ||
2562 | static void *ipmr_vif_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
2563 | { | |
2564 | struct ipmr_vif_iter *iter = seq->private; | |
f6bb4514 | 2565 | struct net *net = seq_file_net(seq); |
f0ad0860 | 2566 | struct mr_table *mrt = iter->mrt; |
1da177e4 LT |
2567 | |
2568 | ++*pos; | |
2569 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 2570 | return ipmr_vif_seq_idx(net, iter, 0); |
e905a9ed | 2571 | |
0c12295a PM |
2572 | while (++iter->ct < mrt->maxvif) { |
2573 | if (!VIF_EXISTS(mrt, iter->ct)) | |
1da177e4 | 2574 | continue; |
0c12295a | 2575 | return &mrt->vif_table[iter->ct]; |
1da177e4 LT |
2576 | } |
2577 | return NULL; | |
2578 | } | |
2579 | ||
2580 | static void ipmr_vif_seq_stop(struct seq_file *seq, void *v) | |
ba93ef74 | 2581 | __releases(mrt_lock) |
1da177e4 LT |
2582 | { |
2583 | read_unlock(&mrt_lock); | |
2584 | } | |
2585 | ||
2586 | static int ipmr_vif_seq_show(struct seq_file *seq, void *v) | |
2587 | { | |
f0ad0860 PM |
2588 | struct ipmr_vif_iter *iter = seq->private; |
2589 | struct mr_table *mrt = iter->mrt; | |
f6bb4514 | 2590 | |
1da177e4 | 2591 | if (v == SEQ_START_TOKEN) { |
e905a9ed | 2592 | seq_puts(seq, |
1da177e4 LT |
2593 | "Interface BytesIn PktsIn BytesOut PktsOut Flags Local Remote\n"); |
2594 | } else { | |
2595 | const struct vif_device *vif = v; | |
2596 | const char *name = vif->dev ? vif->dev->name : "none"; | |
2597 | ||
2598 | seq_printf(seq, | |
5b5e0928 | 2599 | "%2zd %-10s %8ld %7ld %8ld %7ld %05X %08X %08X\n", |
0c12295a | 2600 | vif - mrt->vif_table, |
e905a9ed | 2601 | name, vif->bytes_in, vif->pkt_in, |
1da177e4 LT |
2602 | vif->bytes_out, vif->pkt_out, |
2603 | vif->flags, vif->local, vif->remote); | |
2604 | } | |
2605 | return 0; | |
2606 | } | |
2607 | ||
f690808e | 2608 | static const struct seq_operations ipmr_vif_seq_ops = { |
1da177e4 LT |
2609 | .start = ipmr_vif_seq_start, |
2610 | .next = ipmr_vif_seq_next, | |
2611 | .stop = ipmr_vif_seq_stop, | |
2612 | .show = ipmr_vif_seq_show, | |
2613 | }; | |
2614 | ||
2615 | static int ipmr_vif_open(struct inode *inode, struct file *file) | |
2616 | { | |
f6bb4514 BT |
2617 | return seq_open_net(inode, file, &ipmr_vif_seq_ops, |
2618 | sizeof(struct ipmr_vif_iter)); | |
1da177e4 LT |
2619 | } |
2620 | ||
9a32144e | 2621 | static const struct file_operations ipmr_vif_fops = { |
1da177e4 LT |
2622 | .owner = THIS_MODULE, |
2623 | .open = ipmr_vif_open, | |
2624 | .read = seq_read, | |
2625 | .llseek = seq_lseek, | |
f6bb4514 | 2626 | .release = seq_release_net, |
1da177e4 LT |
2627 | }; |
2628 | ||
2629 | struct ipmr_mfc_iter { | |
f6bb4514 | 2630 | struct seq_net_private p; |
f0ad0860 | 2631 | struct mr_table *mrt; |
862465f2 | 2632 | struct list_head *cache; |
1da177e4 LT |
2633 | }; |
2634 | ||
f6bb4514 BT |
2635 | static struct mfc_cache *ipmr_mfc_seq_idx(struct net *net, |
2636 | struct ipmr_mfc_iter *it, loff_t pos) | |
1da177e4 | 2637 | { |
f0ad0860 | 2638 | struct mr_table *mrt = it->mrt; |
1da177e4 LT |
2639 | struct mfc_cache *mfc; |
2640 | ||
a8c9486b | 2641 | rcu_read_lock(); |
8fb472c0 NA |
2642 | it->cache = &mrt->mfc_cache_list; |
2643 | list_for_each_entry_rcu(mfc, &mrt->mfc_cache_list, list) | |
2644 | if (pos-- == 0) | |
2645 | return mfc; | |
a8c9486b | 2646 | rcu_read_unlock(); |
1da177e4 | 2647 | |
1da177e4 | 2648 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 2649 | it->cache = &mrt->mfc_unres_queue; |
862465f2 | 2650 | list_for_each_entry(mfc, it->cache, list) |
e258beb2 | 2651 | if (pos-- == 0) |
1da177e4 LT |
2652 | return mfc; |
2653 | spin_unlock_bh(&mfc_unres_lock); | |
2654 | ||
2655 | it->cache = NULL; | |
2656 | return NULL; | |
2657 | } | |
2658 | ||
2659 | ||
2660 | static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos) | |
2661 | { | |
2662 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 | 2663 | struct net *net = seq_file_net(seq); |
f0ad0860 | 2664 | struct mr_table *mrt; |
f6bb4514 | 2665 | |
f0ad0860 | 2666 | mrt = ipmr_get_table(net, RT_TABLE_DEFAULT); |
51456b29 | 2667 | if (!mrt) |
f0ad0860 | 2668 | return ERR_PTR(-ENOENT); |
f6bb4514 | 2669 | |
f0ad0860 | 2670 | it->mrt = mrt; |
1da177e4 | 2671 | it->cache = NULL; |
f6bb4514 | 2672 | return *pos ? ipmr_mfc_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
2673 | : SEQ_START_TOKEN; |
2674 | } | |
2675 | ||
2676 | static void *ipmr_mfc_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
2677 | { | |
1da177e4 | 2678 | struct ipmr_mfc_iter *it = seq->private; |
f6bb4514 | 2679 | struct net *net = seq_file_net(seq); |
f0ad0860 | 2680 | struct mr_table *mrt = it->mrt; |
8fb472c0 | 2681 | struct mfc_cache *mfc = v; |
1da177e4 LT |
2682 | |
2683 | ++*pos; | |
2684 | ||
2685 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 2686 | return ipmr_mfc_seq_idx(net, seq->private, 0); |
1da177e4 | 2687 | |
862465f2 PM |
2688 | if (mfc->list.next != it->cache) |
2689 | return list_entry(mfc->list.next, struct mfc_cache, list); | |
e905a9ed | 2690 | |
0c12295a | 2691 | if (it->cache == &mrt->mfc_unres_queue) |
1da177e4 LT |
2692 | goto end_of_list; |
2693 | ||
1da177e4 | 2694 | /* exhausted cache_array, show unresolved */ |
a8c9486b | 2695 | rcu_read_unlock(); |
0c12295a | 2696 | it->cache = &mrt->mfc_unres_queue; |
e905a9ed | 2697 | |
1da177e4 | 2698 | spin_lock_bh(&mfc_unres_lock); |
862465f2 PM |
2699 | if (!list_empty(it->cache)) |
2700 | return list_first_entry(it->cache, struct mfc_cache, list); | |
1da177e4 | 2701 | |
a8cb16dd | 2702 | end_of_list: |
1da177e4 LT |
2703 | spin_unlock_bh(&mfc_unres_lock); |
2704 | it->cache = NULL; | |
2705 | ||
2706 | return NULL; | |
2707 | } | |
2708 | ||
2709 | static void ipmr_mfc_seq_stop(struct seq_file *seq, void *v) | |
2710 | { | |
2711 | struct ipmr_mfc_iter *it = seq->private; | |
f0ad0860 | 2712 | struct mr_table *mrt = it->mrt; |
1da177e4 | 2713 | |
0c12295a | 2714 | if (it->cache == &mrt->mfc_unres_queue) |
1da177e4 | 2715 | spin_unlock_bh(&mfc_unres_lock); |
8fb472c0 | 2716 | else if (it->cache == &mrt->mfc_cache_list) |
a8c9486b | 2717 | rcu_read_unlock(); |
1da177e4 LT |
2718 | } |
2719 | ||
2720 | static int ipmr_mfc_seq_show(struct seq_file *seq, void *v) | |
2721 | { | |
2722 | int n; | |
2723 | ||
2724 | if (v == SEQ_START_TOKEN) { | |
e905a9ed | 2725 | seq_puts(seq, |
1da177e4 LT |
2726 | "Group Origin Iif Pkts Bytes Wrong Oifs\n"); |
2727 | } else { | |
2728 | const struct mfc_cache *mfc = v; | |
2729 | const struct ipmr_mfc_iter *it = seq->private; | |
f0ad0860 | 2730 | const struct mr_table *mrt = it->mrt; |
e905a9ed | 2731 | |
0eae88f3 ED |
2732 | seq_printf(seq, "%08X %08X %-3hd", |
2733 | (__force u32) mfc->mfc_mcastgrp, | |
2734 | (__force u32) mfc->mfc_origin, | |
1ea472e2 | 2735 | mfc->mfc_parent); |
1da177e4 | 2736 | |
0c12295a | 2737 | if (it->cache != &mrt->mfc_unres_queue) { |
1ea472e2 BT |
2738 | seq_printf(seq, " %8lu %8lu %8lu", |
2739 | mfc->mfc_un.res.pkt, | |
2740 | mfc->mfc_un.res.bytes, | |
2741 | mfc->mfc_un.res.wrong_if); | |
132adf54 | 2742 | for (n = mfc->mfc_un.res.minvif; |
a8cb16dd | 2743 | n < mfc->mfc_un.res.maxvif; n++) { |
0c12295a | 2744 | if (VIF_EXISTS(mrt, n) && |
cf958ae3 BT |
2745 | mfc->mfc_un.res.ttls[n] < 255) |
2746 | seq_printf(seq, | |
e905a9ed | 2747 | " %2d:%-3d", |
1da177e4 LT |
2748 | n, mfc->mfc_un.res.ttls[n]); |
2749 | } | |
1ea472e2 BT |
2750 | } else { |
2751 | /* unresolved mfc_caches don't contain | |
2752 | * pkt, bytes and wrong_if values | |
2753 | */ | |
2754 | seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul); | |
1da177e4 LT |
2755 | } |
2756 | seq_putc(seq, '\n'); | |
2757 | } | |
2758 | return 0; | |
2759 | } | |
2760 | ||
f690808e | 2761 | static const struct seq_operations ipmr_mfc_seq_ops = { |
1da177e4 LT |
2762 | .start = ipmr_mfc_seq_start, |
2763 | .next = ipmr_mfc_seq_next, | |
2764 | .stop = ipmr_mfc_seq_stop, | |
2765 | .show = ipmr_mfc_seq_show, | |
2766 | }; | |
2767 | ||
2768 | static int ipmr_mfc_open(struct inode *inode, struct file *file) | |
2769 | { | |
f6bb4514 BT |
2770 | return seq_open_net(inode, file, &ipmr_mfc_seq_ops, |
2771 | sizeof(struct ipmr_mfc_iter)); | |
1da177e4 LT |
2772 | } |
2773 | ||
9a32144e | 2774 | static const struct file_operations ipmr_mfc_fops = { |
1da177e4 LT |
2775 | .owner = THIS_MODULE, |
2776 | .open = ipmr_mfc_open, | |
2777 | .read = seq_read, | |
2778 | .llseek = seq_lseek, | |
f6bb4514 | 2779 | .release = seq_release_net, |
1da177e4 | 2780 | }; |
e905a9ed | 2781 | #endif |
1da177e4 LT |
2782 | |
2783 | #ifdef CONFIG_IP_PIMSM_V2 | |
32613090 | 2784 | static const struct net_protocol pim_protocol = { |
1da177e4 | 2785 | .handler = pim_rcv, |
403dbb97 | 2786 | .netns_ok = 1, |
1da177e4 LT |
2787 | }; |
2788 | #endif | |
2789 | ||
7ef8f65d | 2790 | /* Setup for IP multicast routing */ |
cf958ae3 BT |
2791 | static int __net_init ipmr_net_init(struct net *net) |
2792 | { | |
f0ad0860 | 2793 | int err; |
cf958ae3 | 2794 | |
f0ad0860 PM |
2795 | err = ipmr_rules_init(net); |
2796 | if (err < 0) | |
cf958ae3 | 2797 | goto fail; |
f6bb4514 BT |
2798 | |
2799 | #ifdef CONFIG_PROC_FS | |
2800 | err = -ENOMEM; | |
d4beaa66 | 2801 | if (!proc_create("ip_mr_vif", 0, net->proc_net, &ipmr_vif_fops)) |
f6bb4514 | 2802 | goto proc_vif_fail; |
d4beaa66 | 2803 | if (!proc_create("ip_mr_cache", 0, net->proc_net, &ipmr_mfc_fops)) |
f6bb4514 BT |
2804 | goto proc_cache_fail; |
2805 | #endif | |
2bb8b26c BT |
2806 | return 0; |
2807 | ||
f6bb4514 BT |
2808 | #ifdef CONFIG_PROC_FS |
2809 | proc_cache_fail: | |
ece31ffd | 2810 | remove_proc_entry("ip_mr_vif", net->proc_net); |
f6bb4514 | 2811 | proc_vif_fail: |
f0ad0860 | 2812 | ipmr_rules_exit(net); |
f6bb4514 | 2813 | #endif |
cf958ae3 BT |
2814 | fail: |
2815 | return err; | |
2816 | } | |
2817 | ||
2818 | static void __net_exit ipmr_net_exit(struct net *net) | |
2819 | { | |
f6bb4514 | 2820 | #ifdef CONFIG_PROC_FS |
ece31ffd G |
2821 | remove_proc_entry("ip_mr_cache", net->proc_net); |
2822 | remove_proc_entry("ip_mr_vif", net->proc_net); | |
f6bb4514 | 2823 | #endif |
f0ad0860 | 2824 | ipmr_rules_exit(net); |
cf958ae3 BT |
2825 | } |
2826 | ||
2827 | static struct pernet_operations ipmr_net_ops = { | |
2828 | .init = ipmr_net_init, | |
2829 | .exit = ipmr_net_exit, | |
2830 | }; | |
e905a9ed | 2831 | |
03d2f897 | 2832 | int __init ip_mr_init(void) |
1da177e4 | 2833 | { |
03d2f897 WC |
2834 | int err; |
2835 | ||
1da177e4 LT |
2836 | mrt_cachep = kmem_cache_create("ip_mrt_cache", |
2837 | sizeof(struct mfc_cache), | |
a8c9486b | 2838 | 0, SLAB_HWCACHE_ALIGN | SLAB_PANIC, |
20c2df83 | 2839 | NULL); |
03d2f897 | 2840 | |
cf958ae3 BT |
2841 | err = register_pernet_subsys(&ipmr_net_ops); |
2842 | if (err) | |
2843 | goto reg_pernet_fail; | |
2844 | ||
03d2f897 WC |
2845 | err = register_netdevice_notifier(&ip_mr_notifier); |
2846 | if (err) | |
2847 | goto reg_notif_fail; | |
403dbb97 TG |
2848 | #ifdef CONFIG_IP_PIMSM_V2 |
2849 | if (inet_add_protocol(&pim_protocol, IPPROTO_PIM) < 0) { | |
058bd4d2 | 2850 | pr_err("%s: can't add PIM protocol\n", __func__); |
403dbb97 TG |
2851 | err = -EAGAIN; |
2852 | goto add_proto_fail; | |
2853 | } | |
2854 | #endif | |
c7ac8679 GR |
2855 | rtnl_register(RTNL_FAMILY_IPMR, RTM_GETROUTE, |
2856 | NULL, ipmr_rtm_dumproute, NULL); | |
ccbb0aa6 NA |
2857 | rtnl_register(RTNL_FAMILY_IPMR, RTM_NEWROUTE, |
2858 | ipmr_rtm_route, NULL, NULL); | |
2859 | rtnl_register(RTNL_FAMILY_IPMR, RTM_DELROUTE, | |
2860 | ipmr_rtm_route, NULL, NULL); | |
03d2f897 | 2861 | return 0; |
f6bb4514 | 2862 | |
403dbb97 TG |
2863 | #ifdef CONFIG_IP_PIMSM_V2 |
2864 | add_proto_fail: | |
2865 | unregister_netdevice_notifier(&ip_mr_notifier); | |
2866 | #endif | |
c3e38896 | 2867 | reg_notif_fail: |
cf958ae3 BT |
2868 | unregister_pernet_subsys(&ipmr_net_ops); |
2869 | reg_pernet_fail: | |
c3e38896 | 2870 | kmem_cache_destroy(mrt_cachep); |
03d2f897 | 2871 | return err; |
1da177e4 | 2872 | } |