]>
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 | ||
302 | static struct mr_table *ipmr_new_table(struct net *net, u32 id) | |
303 | { | |
304 | struct mr_table *mrt; | |
305 | unsigned int i; | |
1da177e4 | 306 | |
1113ebbc NA |
307 | /* "pimreg%u" should not exceed 16 bytes (IFNAMSIZ) */ |
308 | if (id != RT_TABLE_DEFAULT && id >= 1000000000) | |
309 | return ERR_PTR(-EINVAL); | |
310 | ||
f0ad0860 | 311 | mrt = ipmr_get_table(net, id); |
00db4124 | 312 | if (mrt) |
f0ad0860 PM |
313 | return mrt; |
314 | ||
315 | mrt = kzalloc(sizeof(*mrt), GFP_KERNEL); | |
51456b29 | 316 | if (!mrt) |
1113ebbc | 317 | return ERR_PTR(-ENOMEM); |
8de53dfb | 318 | write_pnet(&mrt->net, net); |
f0ad0860 PM |
319 | mrt->id = id; |
320 | ||
321 | /* Forwarding cache */ | |
322 | for (i = 0; i < MFC_LINES; i++) | |
323 | INIT_LIST_HEAD(&mrt->mfc_cache_array[i]); | |
324 | ||
325 | INIT_LIST_HEAD(&mrt->mfc_unres_queue); | |
326 | ||
327 | setup_timer(&mrt->ipmr_expire_timer, ipmr_expire_process, | |
328 | (unsigned long)mrt); | |
329 | ||
f0ad0860 | 330 | mrt->mroute_reg_vif_num = -1; |
f0ad0860 PM |
331 | #ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES |
332 | list_add_tail_rcu(&mrt->list, &net->ipv4.mr_tables); | |
333 | #endif | |
334 | return mrt; | |
335 | } | |
1da177e4 | 336 | |
acbb219d FR |
337 | static void ipmr_free_table(struct mr_table *mrt) |
338 | { | |
339 | del_timer_sync(&mrt->ipmr_expire_timer); | |
0e615e96 | 340 | mroute_clean_tables(mrt, true); |
acbb219d FR |
341 | kfree(mrt); |
342 | } | |
343 | ||
1da177e4 LT |
344 | /* Service routines creating virtual interfaces: DVMRP tunnels and PIMREG */ |
345 | ||
d607032d WC |
346 | static void ipmr_del_tunnel(struct net_device *dev, struct vifctl *v) |
347 | { | |
4feb88e5 BT |
348 | struct net *net = dev_net(dev); |
349 | ||
d607032d WC |
350 | dev_close(dev); |
351 | ||
4feb88e5 | 352 | dev = __dev_get_by_name(net, "tunl0"); |
d607032d | 353 | if (dev) { |
5bc3eb7e | 354 | const struct net_device_ops *ops = dev->netdev_ops; |
d607032d | 355 | struct ifreq ifr; |
d607032d WC |
356 | struct ip_tunnel_parm p; |
357 | ||
358 | memset(&p, 0, sizeof(p)); | |
359 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
360 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
361 | p.iph.version = 4; | |
362 | p.iph.ihl = 5; | |
363 | p.iph.protocol = IPPROTO_IPIP; | |
364 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
365 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; | |
366 | ||
5bc3eb7e SH |
367 | if (ops->ndo_do_ioctl) { |
368 | mm_segment_t oldfs = get_fs(); | |
369 | ||
370 | set_fs(KERNEL_DS); | |
371 | ops->ndo_do_ioctl(dev, &ifr, SIOCDELTUNNEL); | |
372 | set_fs(oldfs); | |
373 | } | |
d607032d WC |
374 | } |
375 | } | |
376 | ||
a0b47736 NA |
377 | /* Initialize ipmr pimreg/tunnel in_device */ |
378 | static bool ipmr_init_vif_indev(const struct net_device *dev) | |
379 | { | |
380 | struct in_device *in_dev; | |
381 | ||
382 | ASSERT_RTNL(); | |
383 | ||
384 | in_dev = __in_dev_get_rtnl(dev); | |
385 | if (!in_dev) | |
386 | return false; | |
387 | ipv4_devconf_setall(in_dev); | |
388 | neigh_parms_data_state_setall(in_dev->arp_parms); | |
389 | IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0; | |
390 | ||
391 | return true; | |
392 | } | |
393 | ||
7ef8f65d | 394 | static struct net_device *ipmr_new_tunnel(struct net *net, struct vifctl *v) |
1da177e4 LT |
395 | { |
396 | struct net_device *dev; | |
397 | ||
4feb88e5 | 398 | dev = __dev_get_by_name(net, "tunl0"); |
1da177e4 LT |
399 | |
400 | if (dev) { | |
5bc3eb7e | 401 | const struct net_device_ops *ops = dev->netdev_ops; |
1da177e4 LT |
402 | int err; |
403 | struct ifreq ifr; | |
1da177e4 | 404 | struct ip_tunnel_parm p; |
1da177e4 LT |
405 | |
406 | memset(&p, 0, sizeof(p)); | |
407 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
408 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
409 | p.iph.version = 4; | |
410 | p.iph.ihl = 5; | |
411 | p.iph.protocol = IPPROTO_IPIP; | |
412 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
ba93ef74 | 413 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; |
1da177e4 | 414 | |
5bc3eb7e SH |
415 | if (ops->ndo_do_ioctl) { |
416 | mm_segment_t oldfs = get_fs(); | |
417 | ||
418 | set_fs(KERNEL_DS); | |
419 | err = ops->ndo_do_ioctl(dev, &ifr, SIOCADDTUNNEL); | |
420 | set_fs(oldfs); | |
a8cb16dd | 421 | } else { |
5bc3eb7e | 422 | err = -EOPNOTSUPP; |
a8cb16dd | 423 | } |
1da177e4 LT |
424 | dev = NULL; |
425 | ||
4feb88e5 BT |
426 | if (err == 0 && |
427 | (dev = __dev_get_by_name(net, p.name)) != NULL) { | |
1da177e4 | 428 | dev->flags |= IFF_MULTICAST; |
a0b47736 | 429 | if (!ipmr_init_vif_indev(dev)) |
1da177e4 | 430 | goto failure; |
1da177e4 LT |
431 | if (dev_open(dev)) |
432 | goto failure; | |
7dc00c82 | 433 | dev_hold(dev); |
1da177e4 LT |
434 | } |
435 | } | |
436 | return dev; | |
437 | ||
438 | failure: | |
1da177e4 LT |
439 | unregister_netdevice(dev); |
440 | return NULL; | |
441 | } | |
442 | ||
c316c629 | 443 | #if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2) |
6fef4c0c | 444 | static netdev_tx_t reg_vif_xmit(struct sk_buff *skb, struct net_device *dev) |
1da177e4 | 445 | { |
4feb88e5 | 446 | struct net *net = dev_net(dev); |
f0ad0860 | 447 | struct mr_table *mrt; |
da91981b DM |
448 | struct flowi4 fl4 = { |
449 | .flowi4_oif = dev->ifindex, | |
6a662719 | 450 | .flowi4_iif = skb->skb_iif ? : LOOPBACK_IFINDEX, |
da91981b | 451 | .flowi4_mark = skb->mark, |
f0ad0860 PM |
452 | }; |
453 | int err; | |
454 | ||
da91981b | 455 | err = ipmr_fib_lookup(net, &fl4, &mrt); |
e40dbc51 BG |
456 | if (err < 0) { |
457 | kfree_skb(skb); | |
f0ad0860 | 458 | return err; |
e40dbc51 | 459 | } |
4feb88e5 | 460 | |
1da177e4 | 461 | read_lock(&mrt_lock); |
cf3677ae PE |
462 | dev->stats.tx_bytes += skb->len; |
463 | dev->stats.tx_packets++; | |
0c12295a | 464 | ipmr_cache_report(mrt, skb, mrt->mroute_reg_vif_num, IGMPMSG_WHOLEPKT); |
1da177e4 LT |
465 | read_unlock(&mrt_lock); |
466 | kfree_skb(skb); | |
6ed10654 | 467 | return NETDEV_TX_OK; |
1da177e4 LT |
468 | } |
469 | ||
ee9b9596 ND |
470 | static int reg_vif_get_iflink(const struct net_device *dev) |
471 | { | |
472 | return 0; | |
473 | } | |
474 | ||
007c3838 SH |
475 | static const struct net_device_ops reg_vif_netdev_ops = { |
476 | .ndo_start_xmit = reg_vif_xmit, | |
ee9b9596 | 477 | .ndo_get_iflink = reg_vif_get_iflink, |
007c3838 SH |
478 | }; |
479 | ||
1da177e4 LT |
480 | static void reg_vif_setup(struct net_device *dev) |
481 | { | |
482 | dev->type = ARPHRD_PIMREG; | |
46f25dff | 483 | dev->mtu = ETH_DATA_LEN - sizeof(struct iphdr) - 8; |
1da177e4 | 484 | dev->flags = IFF_NOARP; |
70cb4a45 | 485 | dev->netdev_ops = ®_vif_netdev_ops; |
1da177e4 | 486 | dev->destructor = free_netdev; |
403dbb97 | 487 | dev->features |= NETIF_F_NETNS_LOCAL; |
1da177e4 LT |
488 | } |
489 | ||
f0ad0860 | 490 | static struct net_device *ipmr_reg_vif(struct net *net, struct mr_table *mrt) |
1da177e4 LT |
491 | { |
492 | struct net_device *dev; | |
f0ad0860 | 493 | char name[IFNAMSIZ]; |
1da177e4 | 494 | |
f0ad0860 PM |
495 | if (mrt->id == RT_TABLE_DEFAULT) |
496 | sprintf(name, "pimreg"); | |
497 | else | |
498 | sprintf(name, "pimreg%u", mrt->id); | |
1da177e4 | 499 | |
c835a677 | 500 | dev = alloc_netdev(0, name, NET_NAME_UNKNOWN, reg_vif_setup); |
1da177e4 | 501 | |
51456b29 | 502 | if (!dev) |
1da177e4 LT |
503 | return NULL; |
504 | ||
403dbb97 TG |
505 | dev_net_set(dev, net); |
506 | ||
1da177e4 LT |
507 | if (register_netdevice(dev)) { |
508 | free_netdev(dev); | |
509 | return NULL; | |
510 | } | |
1da177e4 | 511 | |
a0b47736 | 512 | if (!ipmr_init_vif_indev(dev)) |
1da177e4 | 513 | goto failure; |
1da177e4 LT |
514 | if (dev_open(dev)) |
515 | goto failure; | |
516 | ||
7dc00c82 WC |
517 | dev_hold(dev); |
518 | ||
1da177e4 LT |
519 | return dev; |
520 | ||
521 | failure: | |
1da177e4 LT |
522 | unregister_netdevice(dev); |
523 | return NULL; | |
524 | } | |
c316c629 NA |
525 | |
526 | /* called with rcu_read_lock() */ | |
527 | static int __pim_rcv(struct mr_table *mrt, struct sk_buff *skb, | |
528 | unsigned int pimlen) | |
529 | { | |
530 | struct net_device *reg_dev = NULL; | |
531 | struct iphdr *encap; | |
532 | ||
533 | encap = (struct iphdr *)(skb_transport_header(skb) + pimlen); | |
7ef8f65d | 534 | /* Check that: |
c316c629 NA |
535 | * a. packet is really sent to a multicast group |
536 | * b. packet is not a NULL-REGISTER | |
537 | * c. packet is not truncated | |
538 | */ | |
539 | if (!ipv4_is_multicast(encap->daddr) || | |
540 | encap->tot_len == 0 || | |
541 | ntohs(encap->tot_len) + pimlen > skb->len) | |
542 | return 1; | |
543 | ||
544 | read_lock(&mrt_lock); | |
545 | if (mrt->mroute_reg_vif_num >= 0) | |
546 | reg_dev = mrt->vif_table[mrt->mroute_reg_vif_num].dev; | |
547 | read_unlock(&mrt_lock); | |
548 | ||
549 | if (!reg_dev) | |
550 | return 1; | |
551 | ||
552 | skb->mac_header = skb->network_header; | |
553 | skb_pull(skb, (u8 *)encap - skb->data); | |
554 | skb_reset_network_header(skb); | |
555 | skb->protocol = htons(ETH_P_IP); | |
556 | skb->ip_summed = CHECKSUM_NONE; | |
557 | ||
558 | skb_tunnel_rx(skb, reg_dev, dev_net(reg_dev)); | |
559 | ||
560 | netif_rx(skb); | |
561 | ||
562 | return NET_RX_SUCCESS; | |
563 | } | |
564 | #else | |
565 | static struct net_device *ipmr_reg_vif(struct net *net, struct mr_table *mrt) | |
566 | { | |
567 | return NULL; | |
568 | } | |
1da177e4 LT |
569 | #endif |
570 | ||
2c53040f BH |
571 | /** |
572 | * vif_delete - Delete a VIF entry | |
7dc00c82 | 573 | * @notify: Set to 1, if the caller is a notifier_call |
1da177e4 | 574 | */ |
0c12295a | 575 | static int vif_delete(struct mr_table *mrt, int vifi, int notify, |
d17fa6fa | 576 | struct list_head *head) |
1da177e4 LT |
577 | { |
578 | struct vif_device *v; | |
579 | struct net_device *dev; | |
580 | struct in_device *in_dev; | |
581 | ||
0c12295a | 582 | if (vifi < 0 || vifi >= mrt->maxvif) |
1da177e4 LT |
583 | return -EADDRNOTAVAIL; |
584 | ||
0c12295a | 585 | v = &mrt->vif_table[vifi]; |
1da177e4 LT |
586 | |
587 | write_lock_bh(&mrt_lock); | |
588 | dev = v->dev; | |
589 | v->dev = NULL; | |
590 | ||
591 | if (!dev) { | |
592 | write_unlock_bh(&mrt_lock); | |
593 | return -EADDRNOTAVAIL; | |
594 | } | |
595 | ||
0c12295a PM |
596 | if (vifi == mrt->mroute_reg_vif_num) |
597 | mrt->mroute_reg_vif_num = -1; | |
1da177e4 | 598 | |
a8cb16dd | 599 | if (vifi + 1 == mrt->maxvif) { |
1da177e4 | 600 | int tmp; |
a8cb16dd ED |
601 | |
602 | for (tmp = vifi - 1; tmp >= 0; tmp--) { | |
0c12295a | 603 | if (VIF_EXISTS(mrt, tmp)) |
1da177e4 LT |
604 | break; |
605 | } | |
0c12295a | 606 | mrt->maxvif = tmp+1; |
1da177e4 LT |
607 | } |
608 | ||
609 | write_unlock_bh(&mrt_lock); | |
610 | ||
611 | dev_set_allmulti(dev, -1); | |
612 | ||
a8cb16dd ED |
613 | in_dev = __in_dev_get_rtnl(dev); |
614 | if (in_dev) { | |
42f811b8 | 615 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)--; |
d67b8c61 ND |
616 | inet_netconf_notify_devconf(dev_net(dev), |
617 | NETCONFA_MC_FORWARDING, | |
618 | dev->ifindex, &in_dev->cnf); | |
1da177e4 LT |
619 | ip_rt_multicast_event(in_dev); |
620 | } | |
621 | ||
a8cb16dd | 622 | if (v->flags & (VIFF_TUNNEL | VIFF_REGISTER) && !notify) |
d17fa6fa | 623 | unregister_netdevice_queue(dev, head); |
1da177e4 LT |
624 | |
625 | dev_put(dev); | |
626 | return 0; | |
627 | } | |
628 | ||
a8c9486b | 629 | static void ipmr_cache_free_rcu(struct rcu_head *head) |
5c0a66f5 | 630 | { |
a8c9486b ED |
631 | struct mfc_cache *c = container_of(head, struct mfc_cache, rcu); |
632 | ||
5c0a66f5 BT |
633 | kmem_cache_free(mrt_cachep, c); |
634 | } | |
635 | ||
a8c9486b ED |
636 | static inline void ipmr_cache_free(struct mfc_cache *c) |
637 | { | |
638 | call_rcu(&c->rcu, ipmr_cache_free_rcu); | |
639 | } | |
640 | ||
1da177e4 | 641 | /* Destroy an unresolved cache entry, killing queued skbs |
a8cb16dd | 642 | * and reporting error to netlink readers. |
1da177e4 | 643 | */ |
0c12295a | 644 | static void ipmr_destroy_unres(struct mr_table *mrt, struct mfc_cache *c) |
1da177e4 | 645 | { |
8de53dfb | 646 | struct net *net = read_pnet(&mrt->net); |
1da177e4 | 647 | struct sk_buff *skb; |
9ef1d4c7 | 648 | struct nlmsgerr *e; |
1da177e4 | 649 | |
0c12295a | 650 | atomic_dec(&mrt->cache_resolve_queue_len); |
1da177e4 | 651 | |
c354e124 | 652 | while ((skb = skb_dequeue(&c->mfc_un.unres.unresolved))) { |
eddc9ec5 | 653 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
654 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
655 | nlh->nlmsg_type = NLMSG_ERROR; | |
573ce260 | 656 | nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr)); |
1da177e4 | 657 | skb_trim(skb, nlh->nlmsg_len); |
573ce260 | 658 | e = nlmsg_data(nlh); |
9ef1d4c7 PM |
659 | e->error = -ETIMEDOUT; |
660 | memset(&e->msg, 0, sizeof(e->msg)); | |
2942e900 | 661 | |
15e47304 | 662 | rtnl_unicast(skb, net, NETLINK_CB(skb).portid); |
a8cb16dd | 663 | } else { |
1da177e4 | 664 | kfree_skb(skb); |
a8cb16dd | 665 | } |
1da177e4 LT |
666 | } |
667 | ||
5c0a66f5 | 668 | ipmr_cache_free(c); |
1da177e4 LT |
669 | } |
670 | ||
e258beb2 | 671 | /* Timer process for the unresolved queue. */ |
e258beb2 | 672 | static void ipmr_expire_process(unsigned long arg) |
1da177e4 | 673 | { |
0c12295a | 674 | struct mr_table *mrt = (struct mr_table *)arg; |
1da177e4 LT |
675 | unsigned long now; |
676 | unsigned long expires; | |
862465f2 | 677 | struct mfc_cache *c, *next; |
1da177e4 LT |
678 | |
679 | if (!spin_trylock(&mfc_unres_lock)) { | |
0c12295a | 680 | mod_timer(&mrt->ipmr_expire_timer, jiffies+HZ/10); |
1da177e4 LT |
681 | return; |
682 | } | |
683 | ||
0c12295a | 684 | if (list_empty(&mrt->mfc_unres_queue)) |
1da177e4 LT |
685 | goto out; |
686 | ||
687 | now = jiffies; | |
688 | expires = 10*HZ; | |
1da177e4 | 689 | |
0c12295a | 690 | list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) { |
1da177e4 LT |
691 | if (time_after(c->mfc_un.unres.expires, now)) { |
692 | unsigned long interval = c->mfc_un.unres.expires - now; | |
693 | if (interval < expires) | |
694 | expires = interval; | |
1da177e4 LT |
695 | continue; |
696 | } | |
697 | ||
862465f2 | 698 | list_del(&c->list); |
8cd3ac9f | 699 | mroute_netlink_event(mrt, c, RTM_DELROUTE); |
0c12295a | 700 | ipmr_destroy_unres(mrt, c); |
1da177e4 LT |
701 | } |
702 | ||
0c12295a PM |
703 | if (!list_empty(&mrt->mfc_unres_queue)) |
704 | mod_timer(&mrt->ipmr_expire_timer, jiffies + expires); | |
1da177e4 LT |
705 | |
706 | out: | |
707 | spin_unlock(&mfc_unres_lock); | |
708 | } | |
709 | ||
710 | /* Fill oifs list. It is called under write locked mrt_lock. */ | |
0c12295a | 711 | static void ipmr_update_thresholds(struct mr_table *mrt, struct mfc_cache *cache, |
d658f8a0 | 712 | unsigned char *ttls) |
1da177e4 LT |
713 | { |
714 | int vifi; | |
715 | ||
716 | cache->mfc_un.res.minvif = MAXVIFS; | |
717 | cache->mfc_un.res.maxvif = 0; | |
718 | memset(cache->mfc_un.res.ttls, 255, MAXVIFS); | |
719 | ||
0c12295a PM |
720 | for (vifi = 0; vifi < mrt->maxvif; vifi++) { |
721 | if (VIF_EXISTS(mrt, vifi) && | |
cf958ae3 | 722 | ttls[vifi] && ttls[vifi] < 255) { |
1da177e4 LT |
723 | cache->mfc_un.res.ttls[vifi] = ttls[vifi]; |
724 | if (cache->mfc_un.res.minvif > vifi) | |
725 | cache->mfc_un.res.minvif = vifi; | |
726 | if (cache->mfc_un.res.maxvif <= vifi) | |
727 | cache->mfc_un.res.maxvif = vifi + 1; | |
728 | } | |
729 | } | |
90b5ca17 | 730 | cache->mfc_un.res.lastuse = jiffies; |
1da177e4 LT |
731 | } |
732 | ||
0c12295a PM |
733 | static int vif_add(struct net *net, struct mr_table *mrt, |
734 | struct vifctl *vifc, int mrtsock) | |
1da177e4 LT |
735 | { |
736 | int vifi = vifc->vifc_vifi; | |
0c12295a | 737 | struct vif_device *v = &mrt->vif_table[vifi]; |
1da177e4 LT |
738 | struct net_device *dev; |
739 | struct in_device *in_dev; | |
d607032d | 740 | int err; |
1da177e4 LT |
741 | |
742 | /* Is vif busy ? */ | |
0c12295a | 743 | if (VIF_EXISTS(mrt, vifi)) |
1da177e4 LT |
744 | return -EADDRINUSE; |
745 | ||
746 | switch (vifc->vifc_flags) { | |
1da177e4 | 747 | case VIFF_REGISTER: |
1973a4ea | 748 | if (!ipmr_pimsm_enabled()) |
c316c629 NA |
749 | return -EINVAL; |
750 | /* Special Purpose VIF in PIM | |
1da177e4 LT |
751 | * All the packets will be sent to the daemon |
752 | */ | |
0c12295a | 753 | if (mrt->mroute_reg_vif_num >= 0) |
1da177e4 | 754 | return -EADDRINUSE; |
f0ad0860 | 755 | dev = ipmr_reg_vif(net, mrt); |
1da177e4 LT |
756 | if (!dev) |
757 | return -ENOBUFS; | |
d607032d WC |
758 | err = dev_set_allmulti(dev, 1); |
759 | if (err) { | |
760 | unregister_netdevice(dev); | |
7dc00c82 | 761 | dev_put(dev); |
d607032d WC |
762 | return err; |
763 | } | |
1da177e4 | 764 | break; |
e905a9ed | 765 | case VIFF_TUNNEL: |
4feb88e5 | 766 | dev = ipmr_new_tunnel(net, vifc); |
1da177e4 LT |
767 | if (!dev) |
768 | return -ENOBUFS; | |
d607032d WC |
769 | err = dev_set_allmulti(dev, 1); |
770 | if (err) { | |
771 | ipmr_del_tunnel(dev, vifc); | |
7dc00c82 | 772 | dev_put(dev); |
d607032d WC |
773 | return err; |
774 | } | |
1da177e4 | 775 | break; |
ee5e81f0 | 776 | case VIFF_USE_IFINDEX: |
1da177e4 | 777 | case 0: |
ee5e81f0 I |
778 | if (vifc->vifc_flags == VIFF_USE_IFINDEX) { |
779 | dev = dev_get_by_index(net, vifc->vifc_lcl_ifindex); | |
51456b29 | 780 | if (dev && !__in_dev_get_rtnl(dev)) { |
ee5e81f0 I |
781 | dev_put(dev); |
782 | return -EADDRNOTAVAIL; | |
783 | } | |
a8cb16dd | 784 | } else { |
ee5e81f0 | 785 | dev = ip_dev_find(net, vifc->vifc_lcl_addr.s_addr); |
a8cb16dd | 786 | } |
1da177e4 LT |
787 | if (!dev) |
788 | return -EADDRNOTAVAIL; | |
d607032d | 789 | err = dev_set_allmulti(dev, 1); |
7dc00c82 WC |
790 | if (err) { |
791 | dev_put(dev); | |
d607032d | 792 | return err; |
7dc00c82 | 793 | } |
1da177e4 LT |
794 | break; |
795 | default: | |
796 | return -EINVAL; | |
797 | } | |
798 | ||
a8cb16dd ED |
799 | in_dev = __in_dev_get_rtnl(dev); |
800 | if (!in_dev) { | |
d0490cfd | 801 | dev_put(dev); |
1da177e4 | 802 | return -EADDRNOTAVAIL; |
d0490cfd | 803 | } |
42f811b8 | 804 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)++; |
d67b8c61 ND |
805 | inet_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING, dev->ifindex, |
806 | &in_dev->cnf); | |
1da177e4 LT |
807 | ip_rt_multicast_event(in_dev); |
808 | ||
a8cb16dd ED |
809 | /* Fill in the VIF structures */ |
810 | ||
c354e124 JK |
811 | v->rate_limit = vifc->vifc_rate_limit; |
812 | v->local = vifc->vifc_lcl_addr.s_addr; | |
813 | v->remote = vifc->vifc_rmt_addr.s_addr; | |
814 | v->flags = vifc->vifc_flags; | |
1da177e4 LT |
815 | if (!mrtsock) |
816 | v->flags |= VIFF_STATIC; | |
c354e124 | 817 | v->threshold = vifc->vifc_threshold; |
1da177e4 LT |
818 | v->bytes_in = 0; |
819 | v->bytes_out = 0; | |
820 | v->pkt_in = 0; | |
821 | v->pkt_out = 0; | |
822 | v->link = dev->ifindex; | |
a8cb16dd | 823 | if (v->flags & (VIFF_TUNNEL | VIFF_REGISTER)) |
a54acb3a | 824 | v->link = dev_get_iflink(dev); |
1da177e4 LT |
825 | |
826 | /* And finish update writing critical data */ | |
827 | write_lock_bh(&mrt_lock); | |
c354e124 | 828 | v->dev = dev; |
a8cb16dd | 829 | if (v->flags & VIFF_REGISTER) |
0c12295a | 830 | mrt->mroute_reg_vif_num = vifi; |
0c12295a PM |
831 | if (vifi+1 > mrt->maxvif) |
832 | mrt->maxvif = vifi+1; | |
1da177e4 LT |
833 | write_unlock_bh(&mrt_lock); |
834 | return 0; | |
835 | } | |
836 | ||
a8c9486b | 837 | /* called with rcu_read_lock() */ |
0c12295a | 838 | static struct mfc_cache *ipmr_cache_find(struct mr_table *mrt, |
4feb88e5 BT |
839 | __be32 origin, |
840 | __be32 mcastgrp) | |
1da177e4 | 841 | { |
c354e124 | 842 | int line = MFC_HASH(mcastgrp, origin); |
1da177e4 LT |
843 | struct mfc_cache *c; |
844 | ||
a8c9486b | 845 | list_for_each_entry_rcu(c, &mrt->mfc_cache_array[line], list) { |
862465f2 PM |
846 | if (c->mfc_origin == origin && c->mfc_mcastgrp == mcastgrp) |
847 | return c; | |
1da177e4 | 848 | } |
862465f2 | 849 | return NULL; |
1da177e4 LT |
850 | } |
851 | ||
660b26dc ND |
852 | /* Look for a (*,*,oif) entry */ |
853 | static struct mfc_cache *ipmr_cache_find_any_parent(struct mr_table *mrt, | |
854 | int vifi) | |
855 | { | |
360eb5da | 856 | int line = MFC_HASH(htonl(INADDR_ANY), htonl(INADDR_ANY)); |
660b26dc ND |
857 | struct mfc_cache *c; |
858 | ||
859 | list_for_each_entry_rcu(c, &mrt->mfc_cache_array[line], list) | |
360eb5da ND |
860 | if (c->mfc_origin == htonl(INADDR_ANY) && |
861 | c->mfc_mcastgrp == htonl(INADDR_ANY) && | |
660b26dc ND |
862 | c->mfc_un.res.ttls[vifi] < 255) |
863 | return c; | |
864 | ||
865 | return NULL; | |
866 | } | |
867 | ||
868 | /* Look for a (*,G) entry */ | |
869 | static struct mfc_cache *ipmr_cache_find_any(struct mr_table *mrt, | |
870 | __be32 mcastgrp, int vifi) | |
871 | { | |
360eb5da | 872 | int line = MFC_HASH(mcastgrp, htonl(INADDR_ANY)); |
660b26dc ND |
873 | struct mfc_cache *c, *proxy; |
874 | ||
360eb5da | 875 | if (mcastgrp == htonl(INADDR_ANY)) |
660b26dc ND |
876 | goto skip; |
877 | ||
878 | list_for_each_entry_rcu(c, &mrt->mfc_cache_array[line], list) | |
360eb5da | 879 | if (c->mfc_origin == htonl(INADDR_ANY) && |
660b26dc ND |
880 | c->mfc_mcastgrp == mcastgrp) { |
881 | if (c->mfc_un.res.ttls[vifi] < 255) | |
882 | return c; | |
883 | ||
884 | /* It's ok if the vifi is part of the static tree */ | |
885 | proxy = ipmr_cache_find_any_parent(mrt, | |
886 | c->mfc_parent); | |
887 | if (proxy && proxy->mfc_un.res.ttls[vifi] < 255) | |
888 | return c; | |
889 | } | |
890 | ||
891 | skip: | |
892 | return ipmr_cache_find_any_parent(mrt, vifi); | |
893 | } | |
894 | ||
7ef8f65d | 895 | /* Allocate a multicast cache entry */ |
d658f8a0 | 896 | static struct mfc_cache *ipmr_cache_alloc(void) |
1da177e4 | 897 | { |
c354e124 | 898 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL); |
a8c9486b | 899 | |
70a0dec4 TG |
900 | if (c) { |
901 | c->mfc_un.res.last_assert = jiffies - MFC_ASSERT_THRESH - 1; | |
a8c9486b | 902 | c->mfc_un.res.minvif = MAXVIFS; |
70a0dec4 | 903 | } |
1da177e4 LT |
904 | return c; |
905 | } | |
906 | ||
d658f8a0 | 907 | static struct mfc_cache *ipmr_cache_alloc_unres(void) |
1da177e4 | 908 | { |
c354e124 | 909 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC); |
a8c9486b ED |
910 | |
911 | if (c) { | |
912 | skb_queue_head_init(&c->mfc_un.unres.unresolved); | |
913 | c->mfc_un.unres.expires = jiffies + 10*HZ; | |
914 | } | |
1da177e4 LT |
915 | return c; |
916 | } | |
917 | ||
7ef8f65d | 918 | /* A cache entry has gone into a resolved state from queued */ |
0c12295a PM |
919 | static void ipmr_cache_resolve(struct net *net, struct mr_table *mrt, |
920 | struct mfc_cache *uc, struct mfc_cache *c) | |
1da177e4 LT |
921 | { |
922 | struct sk_buff *skb; | |
9ef1d4c7 | 923 | struct nlmsgerr *e; |
1da177e4 | 924 | |
a8cb16dd | 925 | /* Play the pending entries through our router */ |
c354e124 | 926 | while ((skb = __skb_dequeue(&uc->mfc_un.unres.unresolved))) { |
eddc9ec5 | 927 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
928 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
929 | ||
573ce260 | 930 | if (__ipmr_fill_mroute(mrt, skb, c, nlmsg_data(nlh)) > 0) { |
a8cb16dd ED |
931 | nlh->nlmsg_len = skb_tail_pointer(skb) - |
932 | (u8 *)nlh; | |
1da177e4 LT |
933 | } else { |
934 | nlh->nlmsg_type = NLMSG_ERROR; | |
573ce260 | 935 | nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr)); |
1da177e4 | 936 | skb_trim(skb, nlh->nlmsg_len); |
573ce260 | 937 | e = nlmsg_data(nlh); |
9ef1d4c7 PM |
938 | e->error = -EMSGSIZE; |
939 | memset(&e->msg, 0, sizeof(e->msg)); | |
1da177e4 | 940 | } |
2942e900 | 941 | |
15e47304 | 942 | rtnl_unicast(skb, net, NETLINK_CB(skb).portid); |
a8cb16dd | 943 | } else { |
0c12295a | 944 | ip_mr_forward(net, mrt, skb, c, 0); |
a8cb16dd | 945 | } |
1da177e4 LT |
946 | } |
947 | } | |
948 | ||
c316c629 NA |
949 | /* Bounce a cache query up to mrouted. We could use netlink for this but mrouted |
950 | * expects the following bizarre scheme. | |
1da177e4 | 951 | * |
c316c629 | 952 | * Called under mrt_lock. |
1da177e4 | 953 | */ |
0c12295a | 954 | static int ipmr_cache_report(struct mr_table *mrt, |
4feb88e5 | 955 | struct sk_buff *pkt, vifi_t vifi, int assert) |
1da177e4 | 956 | { |
c9bdd4b5 | 957 | const int ihl = ip_hdrlen(pkt); |
c316c629 | 958 | struct sock *mroute_sk; |
1da177e4 LT |
959 | struct igmphdr *igmp; |
960 | struct igmpmsg *msg; | |
c316c629 | 961 | struct sk_buff *skb; |
1da177e4 LT |
962 | int ret; |
963 | ||
1da177e4 LT |
964 | if (assert == IGMPMSG_WHOLEPKT) |
965 | skb = skb_realloc_headroom(pkt, sizeof(struct iphdr)); | |
966 | else | |
1da177e4 LT |
967 | skb = alloc_skb(128, GFP_ATOMIC); |
968 | ||
132adf54 | 969 | if (!skb) |
1da177e4 LT |
970 | return -ENOBUFS; |
971 | ||
1da177e4 LT |
972 | if (assert == IGMPMSG_WHOLEPKT) { |
973 | /* Ugly, but we have no choice with this interface. | |
a8cb16dd ED |
974 | * Duplicate old header, fix ihl, length etc. |
975 | * And all this only to mangle msg->im_msgtype and | |
976 | * to set msg->im_mbz to "mbz" :-) | |
1da177e4 | 977 | */ |
878c8145 ACM |
978 | skb_push(skb, sizeof(struct iphdr)); |
979 | skb_reset_network_header(skb); | |
badff6d0 | 980 | skb_reset_transport_header(skb); |
0272ffc4 | 981 | msg = (struct igmpmsg *)skb_network_header(skb); |
d56f90a7 | 982 | memcpy(msg, skb_network_header(pkt), sizeof(struct iphdr)); |
1da177e4 LT |
983 | msg->im_msgtype = IGMPMSG_WHOLEPKT; |
984 | msg->im_mbz = 0; | |
0c12295a | 985 | msg->im_vif = mrt->mroute_reg_vif_num; |
eddc9ec5 ACM |
986 | ip_hdr(skb)->ihl = sizeof(struct iphdr) >> 2; |
987 | ip_hdr(skb)->tot_len = htons(ntohs(ip_hdr(pkt)->tot_len) + | |
988 | sizeof(struct iphdr)); | |
c316c629 NA |
989 | } else { |
990 | /* Copy the IP header */ | |
991 | skb_set_network_header(skb, skb->len); | |
992 | skb_put(skb, ihl); | |
993 | skb_copy_to_linear_data(skb, pkt->data, ihl); | |
994 | /* Flag to the kernel this is a route add */ | |
995 | ip_hdr(skb)->protocol = 0; | |
996 | msg = (struct igmpmsg *)skb_network_header(skb); | |
997 | msg->im_vif = vifi; | |
998 | skb_dst_set(skb, dst_clone(skb_dst(pkt))); | |
999 | /* Add our header */ | |
1000 | igmp = (struct igmphdr *)skb_put(skb, sizeof(struct igmphdr)); | |
1001 | igmp->type = assert; | |
1002 | msg->im_msgtype = assert; | |
1003 | igmp->code = 0; | |
1004 | ip_hdr(skb)->tot_len = htons(skb->len); /* Fix the length */ | |
1005 | skb->transport_header = skb->network_header; | |
e905a9ed | 1006 | } |
1da177e4 | 1007 | |
4c968709 ED |
1008 | rcu_read_lock(); |
1009 | mroute_sk = rcu_dereference(mrt->mroute_sk); | |
51456b29 | 1010 | if (!mroute_sk) { |
4c968709 | 1011 | rcu_read_unlock(); |
1da177e4 LT |
1012 | kfree_skb(skb); |
1013 | return -EINVAL; | |
1014 | } | |
1015 | ||
a8cb16dd | 1016 | /* Deliver to mrouted */ |
4c968709 ED |
1017 | ret = sock_queue_rcv_skb(mroute_sk, skb); |
1018 | rcu_read_unlock(); | |
70a269e6 | 1019 | if (ret < 0) { |
e87cc472 | 1020 | net_warn_ratelimited("mroute: pending queue full, dropping entries\n"); |
1da177e4 LT |
1021 | kfree_skb(skb); |
1022 | } | |
1023 | ||
1024 | return ret; | |
1025 | } | |
1026 | ||
7ef8f65d NA |
1027 | /* Queue a packet for resolution. It gets locked cache entry! */ |
1028 | static int ipmr_cache_unresolved(struct mr_table *mrt, vifi_t vifi, | |
1029 | struct sk_buff *skb) | |
1da177e4 | 1030 | { |
862465f2 | 1031 | bool found = false; |
1da177e4 LT |
1032 | int err; |
1033 | struct mfc_cache *c; | |
eddc9ec5 | 1034 | const struct iphdr *iph = ip_hdr(skb); |
1da177e4 LT |
1035 | |
1036 | spin_lock_bh(&mfc_unres_lock); | |
0c12295a | 1037 | list_for_each_entry(c, &mrt->mfc_unres_queue, list) { |
e258beb2 | 1038 | if (c->mfc_mcastgrp == iph->daddr && |
862465f2 PM |
1039 | c->mfc_origin == iph->saddr) { |
1040 | found = true; | |
1da177e4 | 1041 | break; |
862465f2 | 1042 | } |
1da177e4 LT |
1043 | } |
1044 | ||
862465f2 | 1045 | if (!found) { |
a8cb16dd | 1046 | /* Create a new entry if allowable */ |
0c12295a | 1047 | if (atomic_read(&mrt->cache_resolve_queue_len) >= 10 || |
d658f8a0 | 1048 | (c = ipmr_cache_alloc_unres()) == NULL) { |
1da177e4 LT |
1049 | spin_unlock_bh(&mfc_unres_lock); |
1050 | ||
1051 | kfree_skb(skb); | |
1052 | return -ENOBUFS; | |
1053 | } | |
1054 | ||
a8cb16dd | 1055 | /* Fill in the new cache entry */ |
eddc9ec5 ACM |
1056 | c->mfc_parent = -1; |
1057 | c->mfc_origin = iph->saddr; | |
1058 | c->mfc_mcastgrp = iph->daddr; | |
1da177e4 | 1059 | |
a8cb16dd | 1060 | /* Reflect first query at mrouted. */ |
0c12295a | 1061 | err = ipmr_cache_report(mrt, skb, vifi, IGMPMSG_NOCACHE); |
4feb88e5 | 1062 | if (err < 0) { |
e905a9ed | 1063 | /* If the report failed throw the cache entry |
1da177e4 LT |
1064 | out - Brad Parker |
1065 | */ | |
1066 | spin_unlock_bh(&mfc_unres_lock); | |
1067 | ||
5c0a66f5 | 1068 | ipmr_cache_free(c); |
1da177e4 LT |
1069 | kfree_skb(skb); |
1070 | return err; | |
1071 | } | |
1072 | ||
0c12295a PM |
1073 | atomic_inc(&mrt->cache_resolve_queue_len); |
1074 | list_add(&c->list, &mrt->mfc_unres_queue); | |
8cd3ac9f | 1075 | mroute_netlink_event(mrt, c, RTM_NEWROUTE); |
1da177e4 | 1076 | |
278554bd DM |
1077 | if (atomic_read(&mrt->cache_resolve_queue_len) == 1) |
1078 | mod_timer(&mrt->ipmr_expire_timer, c->mfc_un.unres.expires); | |
1da177e4 LT |
1079 | } |
1080 | ||
a8cb16dd | 1081 | /* See if we can append the packet */ |
a8cb16dd | 1082 | if (c->mfc_un.unres.unresolved.qlen > 3) { |
1da177e4 LT |
1083 | kfree_skb(skb); |
1084 | err = -ENOBUFS; | |
1085 | } else { | |
c354e124 | 1086 | skb_queue_tail(&c->mfc_un.unres.unresolved, skb); |
1da177e4 LT |
1087 | err = 0; |
1088 | } | |
1089 | ||
1090 | spin_unlock_bh(&mfc_unres_lock); | |
1091 | return err; | |
1092 | } | |
1093 | ||
7ef8f65d | 1094 | /* MFC cache manipulation by user space mroute daemon */ |
1da177e4 | 1095 | |
660b26dc | 1096 | static int ipmr_mfc_delete(struct mr_table *mrt, struct mfcctl *mfc, int parent) |
1da177e4 LT |
1097 | { |
1098 | int line; | |
862465f2 | 1099 | struct mfc_cache *c, *next; |
1da177e4 | 1100 | |
c354e124 | 1101 | line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr); |
1da177e4 | 1102 | |
0c12295a | 1103 | list_for_each_entry_safe(c, next, &mrt->mfc_cache_array[line], list) { |
1da177e4 | 1104 | if (c->mfc_origin == mfc->mfcc_origin.s_addr && |
660b26dc ND |
1105 | c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr && |
1106 | (parent == -1 || parent == c->mfc_parent)) { | |
a8c9486b | 1107 | list_del_rcu(&c->list); |
8cd3ac9f | 1108 | mroute_netlink_event(mrt, c, RTM_DELROUTE); |
5c0a66f5 | 1109 | ipmr_cache_free(c); |
1da177e4 LT |
1110 | return 0; |
1111 | } | |
1112 | } | |
1113 | return -ENOENT; | |
1114 | } | |
1115 | ||
0c12295a | 1116 | static int ipmr_mfc_add(struct net *net, struct mr_table *mrt, |
660b26dc | 1117 | struct mfcctl *mfc, int mrtsock, int parent) |
1da177e4 | 1118 | { |
862465f2 | 1119 | bool found = false; |
1da177e4 | 1120 | int line; |
862465f2 | 1121 | struct mfc_cache *uc, *c; |
1da177e4 | 1122 | |
a50436f2 PM |
1123 | if (mfc->mfcc_parent >= MAXVIFS) |
1124 | return -ENFILE; | |
1125 | ||
c354e124 | 1126 | line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr); |
1da177e4 | 1127 | |
0c12295a | 1128 | list_for_each_entry(c, &mrt->mfc_cache_array[line], list) { |
1da177e4 | 1129 | if (c->mfc_origin == mfc->mfcc_origin.s_addr && |
660b26dc ND |
1130 | c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr && |
1131 | (parent == -1 || parent == c->mfc_parent)) { | |
862465f2 | 1132 | found = true; |
1da177e4 | 1133 | break; |
862465f2 | 1134 | } |
1da177e4 LT |
1135 | } |
1136 | ||
862465f2 | 1137 | if (found) { |
1da177e4 LT |
1138 | write_lock_bh(&mrt_lock); |
1139 | c->mfc_parent = mfc->mfcc_parent; | |
0c12295a | 1140 | ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls); |
1da177e4 LT |
1141 | if (!mrtsock) |
1142 | c->mfc_flags |= MFC_STATIC; | |
1143 | write_unlock_bh(&mrt_lock); | |
8cd3ac9f | 1144 | mroute_netlink_event(mrt, c, RTM_NEWROUTE); |
1da177e4 LT |
1145 | return 0; |
1146 | } | |
1147 | ||
360eb5da | 1148 | if (mfc->mfcc_mcastgrp.s_addr != htonl(INADDR_ANY) && |
660b26dc | 1149 | !ipv4_is_multicast(mfc->mfcc_mcastgrp.s_addr)) |
1da177e4 LT |
1150 | return -EINVAL; |
1151 | ||
d658f8a0 | 1152 | c = ipmr_cache_alloc(); |
51456b29 | 1153 | if (!c) |
1da177e4 LT |
1154 | return -ENOMEM; |
1155 | ||
c354e124 JK |
1156 | c->mfc_origin = mfc->mfcc_origin.s_addr; |
1157 | c->mfc_mcastgrp = mfc->mfcc_mcastgrp.s_addr; | |
1158 | c->mfc_parent = mfc->mfcc_parent; | |
0c12295a | 1159 | ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls); |
1da177e4 LT |
1160 | if (!mrtsock) |
1161 | c->mfc_flags |= MFC_STATIC; | |
1162 | ||
a8c9486b | 1163 | list_add_rcu(&c->list, &mrt->mfc_cache_array[line]); |
1da177e4 | 1164 | |
7ef8f65d NA |
1165 | /* Check to see if we resolved a queued list. If so we |
1166 | * need to send on the frames and tidy up. | |
1da177e4 | 1167 | */ |
b0ebb739 | 1168 | found = false; |
1da177e4 | 1169 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 1170 | list_for_each_entry(uc, &mrt->mfc_unres_queue, list) { |
e258beb2 | 1171 | if (uc->mfc_origin == c->mfc_origin && |
1da177e4 | 1172 | uc->mfc_mcastgrp == c->mfc_mcastgrp) { |
862465f2 | 1173 | list_del(&uc->list); |
0c12295a | 1174 | atomic_dec(&mrt->cache_resolve_queue_len); |
b0ebb739 | 1175 | found = true; |
1da177e4 LT |
1176 | break; |
1177 | } | |
1178 | } | |
0c12295a PM |
1179 | if (list_empty(&mrt->mfc_unres_queue)) |
1180 | del_timer(&mrt->ipmr_expire_timer); | |
1da177e4 LT |
1181 | spin_unlock_bh(&mfc_unres_lock); |
1182 | ||
b0ebb739 | 1183 | if (found) { |
0c12295a | 1184 | ipmr_cache_resolve(net, mrt, uc, c); |
5c0a66f5 | 1185 | ipmr_cache_free(uc); |
1da177e4 | 1186 | } |
8cd3ac9f | 1187 | mroute_netlink_event(mrt, c, RTM_NEWROUTE); |
1da177e4 LT |
1188 | return 0; |
1189 | } | |
1190 | ||
7ef8f65d | 1191 | /* Close the multicast socket, and clear the vif tables etc */ |
0e615e96 | 1192 | static void mroute_clean_tables(struct mr_table *mrt, bool all) |
1da177e4 LT |
1193 | { |
1194 | int i; | |
d17fa6fa | 1195 | LIST_HEAD(list); |
862465f2 | 1196 | struct mfc_cache *c, *next; |
e905a9ed | 1197 | |
a8cb16dd | 1198 | /* Shut down all active vif entries */ |
0c12295a | 1199 | for (i = 0; i < mrt->maxvif; i++) { |
0e615e96 NA |
1200 | if (!all && (mrt->vif_table[i].flags & VIFF_STATIC)) |
1201 | continue; | |
1202 | vif_delete(mrt, i, 0, &list); | |
1da177e4 | 1203 | } |
d17fa6fa | 1204 | unregister_netdevice_many(&list); |
1da177e4 | 1205 | |
a8cb16dd | 1206 | /* Wipe the cache */ |
862465f2 | 1207 | for (i = 0; i < MFC_LINES; i++) { |
0c12295a | 1208 | list_for_each_entry_safe(c, next, &mrt->mfc_cache_array[i], list) { |
0e615e96 | 1209 | if (!all && (c->mfc_flags & MFC_STATIC)) |
1da177e4 | 1210 | continue; |
a8c9486b | 1211 | list_del_rcu(&c->list); |
8cd3ac9f | 1212 | mroute_netlink_event(mrt, c, RTM_DELROUTE); |
5c0a66f5 | 1213 | ipmr_cache_free(c); |
1da177e4 LT |
1214 | } |
1215 | } | |
1216 | ||
0c12295a | 1217 | if (atomic_read(&mrt->cache_resolve_queue_len) != 0) { |
1da177e4 | 1218 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 1219 | list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) { |
862465f2 | 1220 | list_del(&c->list); |
8cd3ac9f | 1221 | mroute_netlink_event(mrt, c, RTM_DELROUTE); |
0c12295a | 1222 | ipmr_destroy_unres(mrt, c); |
1da177e4 LT |
1223 | } |
1224 | spin_unlock_bh(&mfc_unres_lock); | |
1225 | } | |
1226 | } | |
1227 | ||
4c968709 ED |
1228 | /* called from ip_ra_control(), before an RCU grace period, |
1229 | * we dont need to call synchronize_rcu() here | |
1230 | */ | |
1da177e4 LT |
1231 | static void mrtsock_destruct(struct sock *sk) |
1232 | { | |
4feb88e5 | 1233 | struct net *net = sock_net(sk); |
f0ad0860 | 1234 | struct mr_table *mrt; |
4feb88e5 | 1235 | |
1da177e4 | 1236 | rtnl_lock(); |
f0ad0860 | 1237 | ipmr_for_each_table(mrt, net) { |
4c968709 | 1238 | if (sk == rtnl_dereference(mrt->mroute_sk)) { |
f0ad0860 | 1239 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)--; |
d67b8c61 ND |
1240 | inet_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING, |
1241 | NETCONFA_IFINDEX_ALL, | |
1242 | net->ipv4.devconf_all); | |
a9b3cd7f | 1243 | RCU_INIT_POINTER(mrt->mroute_sk, NULL); |
0e615e96 | 1244 | mroute_clean_tables(mrt, false); |
f0ad0860 | 1245 | } |
1da177e4 LT |
1246 | } |
1247 | rtnl_unlock(); | |
1248 | } | |
1249 | ||
7ef8f65d NA |
1250 | /* Socket options and virtual interface manipulation. The whole |
1251 | * virtual interface system is a complete heap, but unfortunately | |
1252 | * that's how BSD mrouted happens to think. Maybe one day with a proper | |
1253 | * MOSPF/PIM router set up we can clean this up. | |
1da177e4 | 1254 | */ |
e905a9ed | 1255 | |
29e97d21 NA |
1256 | int ip_mroute_setsockopt(struct sock *sk, int optname, char __user *optval, |
1257 | unsigned int optlen) | |
1da177e4 | 1258 | { |
4feb88e5 | 1259 | struct net *net = sock_net(sk); |
29e97d21 | 1260 | int val, ret = 0, parent = 0; |
f0ad0860 | 1261 | struct mr_table *mrt; |
29e97d21 NA |
1262 | struct vifctl vif; |
1263 | struct mfcctl mfc; | |
1264 | u32 uval; | |
f0ad0860 | 1265 | |
29e97d21 NA |
1266 | /* There's one exception to the lock - MRT_DONE which needs to unlock */ |
1267 | rtnl_lock(); | |
5e1859fb | 1268 | if (sk->sk_type != SOCK_RAW || |
29e97d21 NA |
1269 | inet_sk(sk)->inet_num != IPPROTO_IGMP) { |
1270 | ret = -EOPNOTSUPP; | |
1271 | goto out_unlock; | |
1272 | } | |
5e1859fb | 1273 | |
f0ad0860 | 1274 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); |
29e97d21 NA |
1275 | if (!mrt) { |
1276 | ret = -ENOENT; | |
1277 | goto out_unlock; | |
1278 | } | |
132adf54 | 1279 | if (optname != MRT_INIT) { |
33d480ce | 1280 | if (sk != rcu_access_pointer(mrt->mroute_sk) && |
29e97d21 NA |
1281 | !ns_capable(net->user_ns, CAP_NET_ADMIN)) { |
1282 | ret = -EACCES; | |
1283 | goto out_unlock; | |
1284 | } | |
1da177e4 LT |
1285 | } |
1286 | ||
132adf54 SH |
1287 | switch (optname) { |
1288 | case MRT_INIT: | |
42e6b89c | 1289 | if (optlen != sizeof(int)) { |
29e97d21 | 1290 | ret = -EINVAL; |
42e6b89c NA |
1291 | break; |
1292 | } | |
1293 | if (rtnl_dereference(mrt->mroute_sk)) { | |
29e97d21 | 1294 | ret = -EADDRINUSE; |
29e97d21 | 1295 | break; |
42e6b89c | 1296 | } |
132adf54 SH |
1297 | |
1298 | ret = ip_ra_control(sk, 1, mrtsock_destruct); | |
1299 | if (ret == 0) { | |
cf778b00 | 1300 | rcu_assign_pointer(mrt->mroute_sk, sk); |
4feb88e5 | 1301 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)++; |
d67b8c61 ND |
1302 | inet_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING, |
1303 | NETCONFA_IFINDEX_ALL, | |
1304 | net->ipv4.devconf_all); | |
132adf54 | 1305 | } |
29e97d21 | 1306 | break; |
132adf54 | 1307 | case MRT_DONE: |
29e97d21 NA |
1308 | if (sk != rcu_access_pointer(mrt->mroute_sk)) { |
1309 | ret = -EACCES; | |
1310 | } else { | |
1311 | /* We need to unlock here because mrtsock_destruct takes | |
1312 | * care of rtnl itself and we can't change that due to | |
1313 | * the IP_ROUTER_ALERT setsockopt which runs without it. | |
1314 | */ | |
1315 | rtnl_unlock(); | |
1316 | ret = ip_ra_control(sk, 0, NULL); | |
1317 | goto out; | |
1318 | } | |
1319 | break; | |
132adf54 SH |
1320 | case MRT_ADD_VIF: |
1321 | case MRT_DEL_VIF: | |
29e97d21 NA |
1322 | if (optlen != sizeof(vif)) { |
1323 | ret = -EINVAL; | |
1324 | break; | |
1325 | } | |
1326 | if (copy_from_user(&vif, optval, sizeof(vif))) { | |
1327 | ret = -EFAULT; | |
1328 | break; | |
1329 | } | |
1330 | if (vif.vifc_vifi >= MAXVIFS) { | |
1331 | ret = -ENFILE; | |
1332 | break; | |
1333 | } | |
c354e124 | 1334 | if (optname == MRT_ADD_VIF) { |
4c968709 ED |
1335 | ret = vif_add(net, mrt, &vif, |
1336 | sk == rtnl_dereference(mrt->mroute_sk)); | |
132adf54 | 1337 | } else { |
0c12295a | 1338 | ret = vif_delete(mrt, vif.vifc_vifi, 0, NULL); |
132adf54 | 1339 | } |
29e97d21 | 1340 | break; |
7ef8f65d NA |
1341 | /* Manipulate the forwarding caches. These live |
1342 | * in a sort of kernel/user symbiosis. | |
1343 | */ | |
132adf54 SH |
1344 | case MRT_ADD_MFC: |
1345 | case MRT_DEL_MFC: | |
660b26dc ND |
1346 | parent = -1; |
1347 | case MRT_ADD_MFC_PROXY: | |
1348 | case MRT_DEL_MFC_PROXY: | |
29e97d21 NA |
1349 | if (optlen != sizeof(mfc)) { |
1350 | ret = -EINVAL; | |
1351 | break; | |
1352 | } | |
1353 | if (copy_from_user(&mfc, optval, sizeof(mfc))) { | |
1354 | ret = -EFAULT; | |
1355 | break; | |
1356 | } | |
660b26dc ND |
1357 | if (parent == 0) |
1358 | parent = mfc.mfcc_parent; | |
660b26dc ND |
1359 | if (optname == MRT_DEL_MFC || optname == MRT_DEL_MFC_PROXY) |
1360 | ret = ipmr_mfc_delete(mrt, &mfc, parent); | |
132adf54 | 1361 | else |
4c968709 | 1362 | ret = ipmr_mfc_add(net, mrt, &mfc, |
660b26dc ND |
1363 | sk == rtnl_dereference(mrt->mroute_sk), |
1364 | parent); | |
29e97d21 | 1365 | break; |
7ef8f65d | 1366 | /* Control PIM assert. */ |
132adf54 | 1367 | case MRT_ASSERT: |
29e97d21 NA |
1368 | if (optlen != sizeof(val)) { |
1369 | ret = -EINVAL; | |
1370 | break; | |
1371 | } | |
1372 | if (get_user(val, (int __user *)optval)) { | |
1373 | ret = -EFAULT; | |
1374 | break; | |
1375 | } | |
1376 | mrt->mroute_do_assert = val; | |
1377 | break; | |
132adf54 | 1378 | case MRT_PIM: |
1973a4ea | 1379 | if (!ipmr_pimsm_enabled()) { |
29e97d21 NA |
1380 | ret = -ENOPROTOOPT; |
1381 | break; | |
1382 | } | |
1383 | if (optlen != sizeof(val)) { | |
1384 | ret = -EINVAL; | |
1385 | break; | |
1386 | } | |
1387 | if (get_user(val, (int __user *)optval)) { | |
1388 | ret = -EFAULT; | |
1389 | break; | |
1390 | } | |
ba93ef74 | 1391 | |
29e97d21 NA |
1392 | val = !!val; |
1393 | if (val != mrt->mroute_do_pim) { | |
1394 | mrt->mroute_do_pim = val; | |
1395 | mrt->mroute_do_assert = val; | |
1da177e4 | 1396 | } |
29e97d21 | 1397 | break; |
f0ad0860 | 1398 | case MRT_TABLE: |
29e97d21 NA |
1399 | if (!IS_BUILTIN(CONFIG_IP_MROUTE_MULTIPLE_TABLES)) { |
1400 | ret = -ENOPROTOOPT; | |
1401 | break; | |
1402 | } | |
1403 | if (optlen != sizeof(uval)) { | |
1404 | ret = -EINVAL; | |
1405 | break; | |
1406 | } | |
1407 | if (get_user(uval, (u32 __user *)optval)) { | |
1408 | ret = -EFAULT; | |
1409 | break; | |
1410 | } | |
f0ad0860 | 1411 | |
4c968709 ED |
1412 | if (sk == rtnl_dereference(mrt->mroute_sk)) { |
1413 | ret = -EBUSY; | |
1414 | } else { | |
29e97d21 | 1415 | mrt = ipmr_new_table(net, uval); |
1113ebbc NA |
1416 | if (IS_ERR(mrt)) |
1417 | ret = PTR_ERR(mrt); | |
5e1859fb | 1418 | else |
29e97d21 | 1419 | raw_sk(sk)->ipmr_table = uval; |
4c968709 | 1420 | } |
29e97d21 | 1421 | break; |
7ef8f65d | 1422 | /* Spurious command, or MRT_VERSION which you cannot set. */ |
132adf54 | 1423 | default: |
29e97d21 | 1424 | ret = -ENOPROTOOPT; |
1da177e4 | 1425 | } |
29e97d21 NA |
1426 | out_unlock: |
1427 | rtnl_unlock(); | |
1428 | out: | |
1429 | return ret; | |
1da177e4 LT |
1430 | } |
1431 | ||
7ef8f65d | 1432 | /* Getsock opt support for the multicast routing system. */ |
c354e124 | 1433 | int ip_mroute_getsockopt(struct sock *sk, int optname, char __user *optval, int __user *optlen) |
1da177e4 LT |
1434 | { |
1435 | int olr; | |
1436 | int val; | |
4feb88e5 | 1437 | struct net *net = sock_net(sk); |
f0ad0860 PM |
1438 | struct mr_table *mrt; |
1439 | ||
5e1859fb ED |
1440 | if (sk->sk_type != SOCK_RAW || |
1441 | inet_sk(sk)->inet_num != IPPROTO_IGMP) | |
1442 | return -EOPNOTSUPP; | |
1443 | ||
f0ad0860 | 1444 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); |
51456b29 | 1445 | if (!mrt) |
f0ad0860 | 1446 | return -ENOENT; |
1da177e4 | 1447 | |
fe9ef3ce NA |
1448 | switch (optname) { |
1449 | case MRT_VERSION: | |
1450 | val = 0x0305; | |
1451 | break; | |
1452 | case MRT_PIM: | |
1973a4ea | 1453 | if (!ipmr_pimsm_enabled()) |
fe9ef3ce NA |
1454 | return -ENOPROTOOPT; |
1455 | val = mrt->mroute_do_pim; | |
1456 | break; | |
1457 | case MRT_ASSERT: | |
1458 | val = mrt->mroute_do_assert; | |
1459 | break; | |
1460 | default: | |
1da177e4 | 1461 | return -ENOPROTOOPT; |
fe9ef3ce | 1462 | } |
1da177e4 LT |
1463 | |
1464 | if (get_user(olr, optlen)) | |
1465 | return -EFAULT; | |
1da177e4 LT |
1466 | olr = min_t(unsigned int, olr, sizeof(int)); |
1467 | if (olr < 0) | |
1468 | return -EINVAL; | |
c354e124 | 1469 | if (put_user(olr, optlen)) |
1da177e4 | 1470 | return -EFAULT; |
c354e124 | 1471 | if (copy_to_user(optval, &val, olr)) |
1da177e4 LT |
1472 | return -EFAULT; |
1473 | return 0; | |
1474 | } | |
1475 | ||
7ef8f65d | 1476 | /* The IP multicast ioctl support routines. */ |
1da177e4 LT |
1477 | int ipmr_ioctl(struct sock *sk, int cmd, void __user *arg) |
1478 | { | |
1479 | struct sioc_sg_req sr; | |
1480 | struct sioc_vif_req vr; | |
1481 | struct vif_device *vif; | |
1482 | struct mfc_cache *c; | |
4feb88e5 | 1483 | struct net *net = sock_net(sk); |
f0ad0860 PM |
1484 | struct mr_table *mrt; |
1485 | ||
1486 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); | |
51456b29 | 1487 | if (!mrt) |
f0ad0860 | 1488 | return -ENOENT; |
e905a9ed | 1489 | |
132adf54 SH |
1490 | switch (cmd) { |
1491 | case SIOCGETVIFCNT: | |
c354e124 | 1492 | if (copy_from_user(&vr, arg, sizeof(vr))) |
132adf54 | 1493 | return -EFAULT; |
0c12295a | 1494 | if (vr.vifi >= mrt->maxvif) |
132adf54 SH |
1495 | return -EINVAL; |
1496 | read_lock(&mrt_lock); | |
0c12295a PM |
1497 | vif = &mrt->vif_table[vr.vifi]; |
1498 | if (VIF_EXISTS(mrt, vr.vifi)) { | |
c354e124 JK |
1499 | vr.icount = vif->pkt_in; |
1500 | vr.ocount = vif->pkt_out; | |
1501 | vr.ibytes = vif->bytes_in; | |
1502 | vr.obytes = vif->bytes_out; | |
1da177e4 | 1503 | read_unlock(&mrt_lock); |
1da177e4 | 1504 | |
c354e124 | 1505 | if (copy_to_user(arg, &vr, sizeof(vr))) |
132adf54 SH |
1506 | return -EFAULT; |
1507 | return 0; | |
1508 | } | |
1509 | read_unlock(&mrt_lock); | |
1510 | return -EADDRNOTAVAIL; | |
1511 | case SIOCGETSGCNT: | |
c354e124 | 1512 | if (copy_from_user(&sr, arg, sizeof(sr))) |
132adf54 SH |
1513 | return -EFAULT; |
1514 | ||
a8c9486b | 1515 | rcu_read_lock(); |
0c12295a | 1516 | c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr); |
132adf54 SH |
1517 | if (c) { |
1518 | sr.pktcnt = c->mfc_un.res.pkt; | |
1519 | sr.bytecnt = c->mfc_un.res.bytes; | |
1520 | sr.wrong_if = c->mfc_un.res.wrong_if; | |
a8c9486b | 1521 | rcu_read_unlock(); |
132adf54 | 1522 | |
c354e124 | 1523 | if (copy_to_user(arg, &sr, sizeof(sr))) |
132adf54 SH |
1524 | return -EFAULT; |
1525 | return 0; | |
1526 | } | |
a8c9486b | 1527 | rcu_read_unlock(); |
132adf54 SH |
1528 | return -EADDRNOTAVAIL; |
1529 | default: | |
1530 | return -ENOIOCTLCMD; | |
1da177e4 LT |
1531 | } |
1532 | } | |
1533 | ||
709b46e8 EB |
1534 | #ifdef CONFIG_COMPAT |
1535 | struct compat_sioc_sg_req { | |
1536 | struct in_addr src; | |
1537 | struct in_addr grp; | |
1538 | compat_ulong_t pktcnt; | |
1539 | compat_ulong_t bytecnt; | |
1540 | compat_ulong_t wrong_if; | |
1541 | }; | |
1542 | ||
ca6b8bb0 DM |
1543 | struct compat_sioc_vif_req { |
1544 | vifi_t vifi; /* Which iface */ | |
1545 | compat_ulong_t icount; | |
1546 | compat_ulong_t ocount; | |
1547 | compat_ulong_t ibytes; | |
1548 | compat_ulong_t obytes; | |
1549 | }; | |
1550 | ||
709b46e8 EB |
1551 | int ipmr_compat_ioctl(struct sock *sk, unsigned int cmd, void __user *arg) |
1552 | { | |
0033d5ad | 1553 | struct compat_sioc_sg_req sr; |
ca6b8bb0 DM |
1554 | struct compat_sioc_vif_req vr; |
1555 | struct vif_device *vif; | |
709b46e8 EB |
1556 | struct mfc_cache *c; |
1557 | struct net *net = sock_net(sk); | |
1558 | struct mr_table *mrt; | |
1559 | ||
1560 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); | |
51456b29 | 1561 | if (!mrt) |
709b46e8 EB |
1562 | return -ENOENT; |
1563 | ||
1564 | switch (cmd) { | |
ca6b8bb0 DM |
1565 | case SIOCGETVIFCNT: |
1566 | if (copy_from_user(&vr, arg, sizeof(vr))) | |
1567 | return -EFAULT; | |
1568 | if (vr.vifi >= mrt->maxvif) | |
1569 | return -EINVAL; | |
1570 | read_lock(&mrt_lock); | |
1571 | vif = &mrt->vif_table[vr.vifi]; | |
1572 | if (VIF_EXISTS(mrt, vr.vifi)) { | |
1573 | vr.icount = vif->pkt_in; | |
1574 | vr.ocount = vif->pkt_out; | |
1575 | vr.ibytes = vif->bytes_in; | |
1576 | vr.obytes = vif->bytes_out; | |
1577 | read_unlock(&mrt_lock); | |
1578 | ||
1579 | if (copy_to_user(arg, &vr, sizeof(vr))) | |
1580 | return -EFAULT; | |
1581 | return 0; | |
1582 | } | |
1583 | read_unlock(&mrt_lock); | |
1584 | return -EADDRNOTAVAIL; | |
709b46e8 EB |
1585 | case SIOCGETSGCNT: |
1586 | if (copy_from_user(&sr, arg, sizeof(sr))) | |
1587 | return -EFAULT; | |
1588 | ||
1589 | rcu_read_lock(); | |
1590 | c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr); | |
1591 | if (c) { | |
1592 | sr.pktcnt = c->mfc_un.res.pkt; | |
1593 | sr.bytecnt = c->mfc_un.res.bytes; | |
1594 | sr.wrong_if = c->mfc_un.res.wrong_if; | |
1595 | rcu_read_unlock(); | |
1596 | ||
1597 | if (copy_to_user(arg, &sr, sizeof(sr))) | |
1598 | return -EFAULT; | |
1599 | return 0; | |
1600 | } | |
1601 | rcu_read_unlock(); | |
1602 | return -EADDRNOTAVAIL; | |
1603 | default: | |
1604 | return -ENOIOCTLCMD; | |
1605 | } | |
1606 | } | |
1607 | #endif | |
1608 | ||
1da177e4 LT |
1609 | static int ipmr_device_event(struct notifier_block *this, unsigned long event, void *ptr) |
1610 | { | |
351638e7 | 1611 | struct net_device *dev = netdev_notifier_info_to_dev(ptr); |
4feb88e5 | 1612 | struct net *net = dev_net(dev); |
f0ad0860 | 1613 | struct mr_table *mrt; |
1da177e4 LT |
1614 | struct vif_device *v; |
1615 | int ct; | |
e9dc8653 | 1616 | |
1da177e4 LT |
1617 | if (event != NETDEV_UNREGISTER) |
1618 | return NOTIFY_DONE; | |
f0ad0860 PM |
1619 | |
1620 | ipmr_for_each_table(mrt, net) { | |
1621 | v = &mrt->vif_table[0]; | |
1622 | for (ct = 0; ct < mrt->maxvif; ct++, v++) { | |
1623 | if (v->dev == dev) | |
e92036a6 | 1624 | vif_delete(mrt, ct, 1, NULL); |
f0ad0860 | 1625 | } |
1da177e4 LT |
1626 | } |
1627 | return NOTIFY_DONE; | |
1628 | } | |
1629 | ||
c354e124 | 1630 | static struct notifier_block ip_mr_notifier = { |
1da177e4 LT |
1631 | .notifier_call = ipmr_device_event, |
1632 | }; | |
1633 | ||
7ef8f65d NA |
1634 | /* Encapsulate a packet by attaching a valid IPIP header to it. |
1635 | * This avoids tunnel drivers and other mess and gives us the speed so | |
1636 | * important for multicast video. | |
1da177e4 | 1637 | */ |
b6a7719a HFS |
1638 | static void ip_encap(struct net *net, struct sk_buff *skb, |
1639 | __be32 saddr, __be32 daddr) | |
1da177e4 | 1640 | { |
8856dfa3 | 1641 | struct iphdr *iph; |
b71d1d42 | 1642 | const struct iphdr *old_iph = ip_hdr(skb); |
8856dfa3 ACM |
1643 | |
1644 | skb_push(skb, sizeof(struct iphdr)); | |
b0e380b1 | 1645 | skb->transport_header = skb->network_header; |
8856dfa3 | 1646 | skb_reset_network_header(skb); |
eddc9ec5 | 1647 | iph = ip_hdr(skb); |
1da177e4 | 1648 | |
a8cb16dd | 1649 | iph->version = 4; |
e023dd64 ACM |
1650 | iph->tos = old_iph->tos; |
1651 | iph->ttl = old_iph->ttl; | |
1da177e4 LT |
1652 | iph->frag_off = 0; |
1653 | iph->daddr = daddr; | |
1654 | iph->saddr = saddr; | |
1655 | iph->protocol = IPPROTO_IPIP; | |
1656 | iph->ihl = 5; | |
1657 | iph->tot_len = htons(skb->len); | |
b6a7719a | 1658 | ip_select_ident(net, skb, NULL); |
1da177e4 LT |
1659 | ip_send_check(iph); |
1660 | ||
1da177e4 LT |
1661 | memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt)); |
1662 | nf_reset(skb); | |
1663 | } | |
1664 | ||
0c4b51f0 EB |
1665 | static inline int ipmr_forward_finish(struct net *net, struct sock *sk, |
1666 | struct sk_buff *skb) | |
1da177e4 | 1667 | { |
a8cb16dd | 1668 | struct ip_options *opt = &(IPCB(skb)->opt); |
1da177e4 | 1669 | |
73186df8 DM |
1670 | IP_INC_STATS(net, IPSTATS_MIB_OUTFORWDATAGRAMS); |
1671 | IP_ADD_STATS(net, IPSTATS_MIB_OUTOCTETS, skb->len); | |
1da177e4 LT |
1672 | |
1673 | if (unlikely(opt->optlen)) | |
1674 | ip_forward_options(skb); | |
1675 | ||
13206b6b | 1676 | return dst_output(net, sk, skb); |
1da177e4 LT |
1677 | } |
1678 | ||
7ef8f65d | 1679 | /* Processing handlers for ipmr_forward */ |
1da177e4 | 1680 | |
0c12295a PM |
1681 | static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt, |
1682 | struct sk_buff *skb, struct mfc_cache *c, int vifi) | |
1da177e4 | 1683 | { |
eddc9ec5 | 1684 | const struct iphdr *iph = ip_hdr(skb); |
0c12295a | 1685 | struct vif_device *vif = &mrt->vif_table[vifi]; |
1da177e4 LT |
1686 | struct net_device *dev; |
1687 | struct rtable *rt; | |
31e4543d | 1688 | struct flowi4 fl4; |
1da177e4 LT |
1689 | int encap = 0; |
1690 | ||
51456b29 | 1691 | if (!vif->dev) |
1da177e4 LT |
1692 | goto out_free; |
1693 | ||
1da177e4 LT |
1694 | if (vif->flags & VIFF_REGISTER) { |
1695 | vif->pkt_out++; | |
c354e124 | 1696 | vif->bytes_out += skb->len; |
cf3677ae PE |
1697 | vif->dev->stats.tx_bytes += skb->len; |
1698 | vif->dev->stats.tx_packets++; | |
0c12295a | 1699 | ipmr_cache_report(mrt, skb, vifi, IGMPMSG_WHOLEPKT); |
69ebbf58 | 1700 | goto out_free; |
1da177e4 | 1701 | } |
1da177e4 | 1702 | |
a8cb16dd | 1703 | if (vif->flags & VIFF_TUNNEL) { |
31e4543d | 1704 | rt = ip_route_output_ports(net, &fl4, NULL, |
78fbfd8a DM |
1705 | vif->remote, vif->local, |
1706 | 0, 0, | |
1707 | IPPROTO_IPIP, | |
1708 | RT_TOS(iph->tos), vif->link); | |
b23dd4fe | 1709 | if (IS_ERR(rt)) |
1da177e4 LT |
1710 | goto out_free; |
1711 | encap = sizeof(struct iphdr); | |
1712 | } else { | |
31e4543d | 1713 | rt = ip_route_output_ports(net, &fl4, NULL, iph->daddr, 0, |
78fbfd8a DM |
1714 | 0, 0, |
1715 | IPPROTO_IPIP, | |
1716 | RT_TOS(iph->tos), vif->link); | |
b23dd4fe | 1717 | if (IS_ERR(rt)) |
1da177e4 LT |
1718 | goto out_free; |
1719 | } | |
1720 | ||
d8d1f30b | 1721 | dev = rt->dst.dev; |
1da177e4 | 1722 | |
d8d1f30b | 1723 | if (skb->len+encap > dst_mtu(&rt->dst) && (ntohs(iph->frag_off) & IP_DF)) { |
1da177e4 | 1724 | /* Do not fragment multicasts. Alas, IPv4 does not |
a8cb16dd ED |
1725 | * allow to send ICMP, so that packets will disappear |
1726 | * to blackhole. | |
1da177e4 | 1727 | */ |
73186df8 | 1728 | IP_INC_STATS(net, IPSTATS_MIB_FRAGFAILS); |
1da177e4 LT |
1729 | ip_rt_put(rt); |
1730 | goto out_free; | |
1731 | } | |
1732 | ||
d8d1f30b | 1733 | encap += LL_RESERVED_SPACE(dev) + rt->dst.header_len; |
1da177e4 LT |
1734 | |
1735 | if (skb_cow(skb, encap)) { | |
e905a9ed | 1736 | ip_rt_put(rt); |
1da177e4 LT |
1737 | goto out_free; |
1738 | } | |
1739 | ||
1740 | vif->pkt_out++; | |
c354e124 | 1741 | vif->bytes_out += skb->len; |
1da177e4 | 1742 | |
adf30907 | 1743 | skb_dst_drop(skb); |
d8d1f30b | 1744 | skb_dst_set(skb, &rt->dst); |
eddc9ec5 | 1745 | ip_decrease_ttl(ip_hdr(skb)); |
1da177e4 LT |
1746 | |
1747 | /* FIXME: forward and output firewalls used to be called here. | |
a8cb16dd ED |
1748 | * What do we do with netfilter? -- RR |
1749 | */ | |
1da177e4 | 1750 | if (vif->flags & VIFF_TUNNEL) { |
b6a7719a | 1751 | ip_encap(net, skb, vif->local, vif->remote); |
1da177e4 | 1752 | /* FIXME: extra output firewall step used to be here. --RR */ |
2f4c02d4 PE |
1753 | vif->dev->stats.tx_packets++; |
1754 | vif->dev->stats.tx_bytes += skb->len; | |
1da177e4 LT |
1755 | } |
1756 | ||
9ee6c5dc | 1757 | IPCB(skb)->flags |= IPSKB_FORWARDED; |
1da177e4 | 1758 | |
7ef8f65d | 1759 | /* RFC1584 teaches, that DVMRP/PIM router must deliver packets locally |
1da177e4 LT |
1760 | * not only before forwarding, but after forwarding on all output |
1761 | * interfaces. It is clear, if mrouter runs a multicasting | |
1762 | * program, it should receive packets not depending to what interface | |
1763 | * program is joined. | |
1764 | * If we will not make it, the program will have to join on all | |
1765 | * interfaces. On the other hand, multihoming host (or router, but | |
1766 | * not mrouter) cannot join to more than one interface - it will | |
1767 | * result in receiving multiple packets. | |
1768 | */ | |
29a26a56 EB |
1769 | NF_HOOK(NFPROTO_IPV4, NF_INET_FORWARD, |
1770 | net, NULL, skb, skb->dev, dev, | |
1da177e4 LT |
1771 | ipmr_forward_finish); |
1772 | return; | |
1773 | ||
1774 | out_free: | |
1775 | kfree_skb(skb); | |
1da177e4 LT |
1776 | } |
1777 | ||
0c12295a | 1778 | static int ipmr_find_vif(struct mr_table *mrt, struct net_device *dev) |
1da177e4 LT |
1779 | { |
1780 | int ct; | |
0c12295a PM |
1781 | |
1782 | for (ct = mrt->maxvif-1; ct >= 0; ct--) { | |
1783 | if (mrt->vif_table[ct].dev == dev) | |
1da177e4 LT |
1784 | break; |
1785 | } | |
1786 | return ct; | |
1787 | } | |
1788 | ||
1789 | /* "local" means that we should preserve one skb (for local delivery) */ | |
c4854ec8 RR |
1790 | static void ip_mr_forward(struct net *net, struct mr_table *mrt, |
1791 | struct sk_buff *skb, struct mfc_cache *cache, | |
1792 | int local) | |
1da177e4 LT |
1793 | { |
1794 | int psend = -1; | |
1795 | int vif, ct; | |
660b26dc | 1796 | int true_vifi = ipmr_find_vif(mrt, skb->dev); |
1da177e4 LT |
1797 | |
1798 | vif = cache->mfc_parent; | |
1799 | cache->mfc_un.res.pkt++; | |
1800 | cache->mfc_un.res.bytes += skb->len; | |
43b9e127 | 1801 | cache->mfc_un.res.lastuse = jiffies; |
1da177e4 | 1802 | |
360eb5da | 1803 | if (cache->mfc_origin == htonl(INADDR_ANY) && true_vifi >= 0) { |
660b26dc ND |
1804 | struct mfc_cache *cache_proxy; |
1805 | ||
1806 | /* For an (*,G) entry, we only check that the incomming | |
1807 | * interface is part of the static tree. | |
1808 | */ | |
1809 | cache_proxy = ipmr_cache_find_any_parent(mrt, vif); | |
1810 | if (cache_proxy && | |
1811 | cache_proxy->mfc_un.res.ttls[true_vifi] < 255) | |
1812 | goto forward; | |
1813 | } | |
1814 | ||
7ef8f65d | 1815 | /* Wrong interface: drop packet and (maybe) send PIM assert. */ |
0c12295a | 1816 | if (mrt->vif_table[vif].dev != skb->dev) { |
e58e4159 DA |
1817 | struct net_device *mdev; |
1818 | ||
1819 | mdev = l3mdev_master_dev_rcu(mrt->vif_table[vif].dev); | |
1820 | if (mdev == skb->dev) | |
1821 | goto forward; | |
1822 | ||
c7537967 | 1823 | if (rt_is_output_route(skb_rtable(skb))) { |
1da177e4 | 1824 | /* It is our own packet, looped back. |
a8cb16dd ED |
1825 | * Very complicated situation... |
1826 | * | |
1827 | * The best workaround until routing daemons will be | |
1828 | * fixed is not to redistribute packet, if it was | |
1829 | * send through wrong interface. It means, that | |
1830 | * multicast applications WILL NOT work for | |
1831 | * (S,G), which have default multicast route pointing | |
1832 | * to wrong oif. In any case, it is not a good | |
1833 | * idea to use multicasting applications on router. | |
1da177e4 LT |
1834 | */ |
1835 | goto dont_forward; | |
1836 | } | |
1837 | ||
1838 | cache->mfc_un.res.wrong_if++; | |
1da177e4 | 1839 | |
0c12295a | 1840 | if (true_vifi >= 0 && mrt->mroute_do_assert && |
1da177e4 | 1841 | /* pimsm uses asserts, when switching from RPT to SPT, |
a8cb16dd ED |
1842 | * so that we cannot check that packet arrived on an oif. |
1843 | * It is bad, but otherwise we would need to move pretty | |
1844 | * large chunk of pimd to kernel. Ough... --ANK | |
1da177e4 | 1845 | */ |
0c12295a | 1846 | (mrt->mroute_do_pim || |
6f9374a9 | 1847 | cache->mfc_un.res.ttls[true_vifi] < 255) && |
e905a9ed | 1848 | time_after(jiffies, |
1da177e4 LT |
1849 | cache->mfc_un.res.last_assert + MFC_ASSERT_THRESH)) { |
1850 | cache->mfc_un.res.last_assert = jiffies; | |
0c12295a | 1851 | ipmr_cache_report(mrt, skb, true_vifi, IGMPMSG_WRONGVIF); |
1da177e4 LT |
1852 | } |
1853 | goto dont_forward; | |
1854 | } | |
1855 | ||
660b26dc | 1856 | forward: |
0c12295a PM |
1857 | mrt->vif_table[vif].pkt_in++; |
1858 | mrt->vif_table[vif].bytes_in += skb->len; | |
1da177e4 | 1859 | |
7ef8f65d | 1860 | /* Forward the frame */ |
360eb5da ND |
1861 | if (cache->mfc_origin == htonl(INADDR_ANY) && |
1862 | cache->mfc_mcastgrp == htonl(INADDR_ANY)) { | |
660b26dc ND |
1863 | if (true_vifi >= 0 && |
1864 | true_vifi != cache->mfc_parent && | |
1865 | ip_hdr(skb)->ttl > | |
1866 | cache->mfc_un.res.ttls[cache->mfc_parent]) { | |
1867 | /* It's an (*,*) entry and the packet is not coming from | |
1868 | * the upstream: forward the packet to the upstream | |
1869 | * only. | |
1870 | */ | |
1871 | psend = cache->mfc_parent; | |
1872 | goto last_forward; | |
1873 | } | |
1874 | goto dont_forward; | |
1875 | } | |
a8cb16dd ED |
1876 | for (ct = cache->mfc_un.res.maxvif - 1; |
1877 | ct >= cache->mfc_un.res.minvif; ct--) { | |
660b26dc | 1878 | /* For (*,G) entry, don't forward to the incoming interface */ |
360eb5da ND |
1879 | if ((cache->mfc_origin != htonl(INADDR_ANY) || |
1880 | ct != true_vifi) && | |
660b26dc | 1881 | ip_hdr(skb)->ttl > cache->mfc_un.res.ttls[ct]) { |
1da177e4 LT |
1882 | if (psend != -1) { |
1883 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
a8cb16dd | 1884 | |
1da177e4 | 1885 | if (skb2) |
0c12295a PM |
1886 | ipmr_queue_xmit(net, mrt, skb2, cache, |
1887 | psend); | |
1da177e4 | 1888 | } |
c354e124 | 1889 | psend = ct; |
1da177e4 LT |
1890 | } |
1891 | } | |
660b26dc | 1892 | last_forward: |
1da177e4 LT |
1893 | if (psend != -1) { |
1894 | if (local) { | |
1895 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
a8cb16dd | 1896 | |
1da177e4 | 1897 | if (skb2) |
0c12295a | 1898 | ipmr_queue_xmit(net, mrt, skb2, cache, psend); |
1da177e4 | 1899 | } else { |
0c12295a | 1900 | ipmr_queue_xmit(net, mrt, skb, cache, psend); |
c4854ec8 | 1901 | return; |
1da177e4 LT |
1902 | } |
1903 | } | |
1904 | ||
1905 | dont_forward: | |
1906 | if (!local) | |
1907 | kfree_skb(skb); | |
1da177e4 LT |
1908 | } |
1909 | ||
417da66f | 1910 | static struct mr_table *ipmr_rt_fib_lookup(struct net *net, struct sk_buff *skb) |
ee3f1aaf | 1911 | { |
417da66f DM |
1912 | struct rtable *rt = skb_rtable(skb); |
1913 | struct iphdr *iph = ip_hdr(skb); | |
da91981b | 1914 | struct flowi4 fl4 = { |
417da66f DM |
1915 | .daddr = iph->daddr, |
1916 | .saddr = iph->saddr, | |
b0fe4a31 | 1917 | .flowi4_tos = RT_TOS(iph->tos), |
4fd551d7 DM |
1918 | .flowi4_oif = (rt_is_output_route(rt) ? |
1919 | skb->dev->ifindex : 0), | |
1920 | .flowi4_iif = (rt_is_output_route(rt) ? | |
1fb9489b | 1921 | LOOPBACK_IFINDEX : |
4fd551d7 | 1922 | skb->dev->ifindex), |
b4869889 | 1923 | .flowi4_mark = skb->mark, |
ee3f1aaf DM |
1924 | }; |
1925 | struct mr_table *mrt; | |
1926 | int err; | |
1927 | ||
da91981b | 1928 | err = ipmr_fib_lookup(net, &fl4, &mrt); |
ee3f1aaf DM |
1929 | if (err) |
1930 | return ERR_PTR(err); | |
1931 | return mrt; | |
1932 | } | |
1da177e4 | 1933 | |
7ef8f65d NA |
1934 | /* Multicast packets for forwarding arrive here |
1935 | * Called with rcu_read_lock(); | |
1da177e4 | 1936 | */ |
1da177e4 LT |
1937 | int ip_mr_input(struct sk_buff *skb) |
1938 | { | |
1939 | struct mfc_cache *cache; | |
4feb88e5 | 1940 | struct net *net = dev_net(skb->dev); |
511c3f92 | 1941 | int local = skb_rtable(skb)->rt_flags & RTCF_LOCAL; |
f0ad0860 | 1942 | struct mr_table *mrt; |
1da177e4 LT |
1943 | |
1944 | /* Packet is looped back after forward, it should not be | |
a8cb16dd | 1945 | * forwarded second time, but still can be delivered locally. |
1da177e4 | 1946 | */ |
4c968709 | 1947 | if (IPCB(skb)->flags & IPSKB_FORWARDED) |
1da177e4 LT |
1948 | goto dont_forward; |
1949 | ||
417da66f | 1950 | mrt = ipmr_rt_fib_lookup(net, skb); |
ee3f1aaf DM |
1951 | if (IS_ERR(mrt)) { |
1952 | kfree_skb(skb); | |
1953 | return PTR_ERR(mrt); | |
e40dbc51 | 1954 | } |
1da177e4 | 1955 | if (!local) { |
4c968709 ED |
1956 | if (IPCB(skb)->opt.router_alert) { |
1957 | if (ip_call_ra_chain(skb)) | |
1958 | return 0; | |
1959 | } else if (ip_hdr(skb)->protocol == IPPROTO_IGMP) { | |
1960 | /* IGMPv1 (and broken IGMPv2 implementations sort of | |
1961 | * Cisco IOS <= 11.2(8)) do not put router alert | |
1962 | * option to IGMP packets destined to routable | |
1963 | * groups. It is very bad, because it means | |
1964 | * that we can forward NO IGMP messages. | |
1965 | */ | |
1966 | struct sock *mroute_sk; | |
1967 | ||
1968 | mroute_sk = rcu_dereference(mrt->mroute_sk); | |
1969 | if (mroute_sk) { | |
1970 | nf_reset(skb); | |
1971 | raw_rcv(mroute_sk, skb); | |
1972 | return 0; | |
1973 | } | |
1da177e4 LT |
1974 | } |
1975 | } | |
1976 | ||
a8c9486b | 1977 | /* already under rcu_read_lock() */ |
0c12295a | 1978 | cache = ipmr_cache_find(mrt, ip_hdr(skb)->saddr, ip_hdr(skb)->daddr); |
51456b29 | 1979 | if (!cache) { |
660b26dc ND |
1980 | int vif = ipmr_find_vif(mrt, skb->dev); |
1981 | ||
1982 | if (vif >= 0) | |
1983 | cache = ipmr_cache_find_any(mrt, ip_hdr(skb)->daddr, | |
1984 | vif); | |
1985 | } | |
1da177e4 | 1986 | |
7ef8f65d | 1987 | /* No usable cache entry */ |
51456b29 | 1988 | if (!cache) { |
1da177e4 LT |
1989 | int vif; |
1990 | ||
1991 | if (local) { | |
1992 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
1993 | ip_local_deliver(skb); | |
51456b29 | 1994 | if (!skb2) |
1da177e4 | 1995 | return -ENOBUFS; |
1da177e4 LT |
1996 | skb = skb2; |
1997 | } | |
1998 | ||
a8c9486b | 1999 | read_lock(&mrt_lock); |
0c12295a | 2000 | vif = ipmr_find_vif(mrt, skb->dev); |
1da177e4 | 2001 | if (vif >= 0) { |
0eae88f3 | 2002 | int err2 = ipmr_cache_unresolved(mrt, vif, skb); |
1da177e4 LT |
2003 | read_unlock(&mrt_lock); |
2004 | ||
0eae88f3 | 2005 | return err2; |
1da177e4 LT |
2006 | } |
2007 | read_unlock(&mrt_lock); | |
2008 | kfree_skb(skb); | |
2009 | return -ENODEV; | |
2010 | } | |
2011 | ||
a8c9486b | 2012 | read_lock(&mrt_lock); |
0c12295a | 2013 | ip_mr_forward(net, mrt, skb, cache, local); |
1da177e4 LT |
2014 | read_unlock(&mrt_lock); |
2015 | ||
2016 | if (local) | |
2017 | return ip_local_deliver(skb); | |
2018 | ||
2019 | return 0; | |
2020 | ||
2021 | dont_forward: | |
2022 | if (local) | |
2023 | return ip_local_deliver(skb); | |
2024 | kfree_skb(skb); | |
2025 | return 0; | |
2026 | } | |
2027 | ||
b1879204 | 2028 | #ifdef CONFIG_IP_PIMSM_V1 |
7ef8f65d | 2029 | /* Handle IGMP messages of PIMv1 */ |
a8cb16dd | 2030 | int pim_rcv_v1(struct sk_buff *skb) |
b1879204 IJ |
2031 | { |
2032 | struct igmphdr *pim; | |
4feb88e5 | 2033 | struct net *net = dev_net(skb->dev); |
f0ad0860 | 2034 | struct mr_table *mrt; |
b1879204 IJ |
2035 | |
2036 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) | |
2037 | goto drop; | |
2038 | ||
2039 | pim = igmp_hdr(skb); | |
2040 | ||
417da66f | 2041 | mrt = ipmr_rt_fib_lookup(net, skb); |
ee3f1aaf DM |
2042 | if (IS_ERR(mrt)) |
2043 | goto drop; | |
0c12295a | 2044 | if (!mrt->mroute_do_pim || |
b1879204 IJ |
2045 | pim->group != PIM_V1_VERSION || pim->code != PIM_V1_REGISTER) |
2046 | goto drop; | |
2047 | ||
f0ad0860 | 2048 | if (__pim_rcv(mrt, skb, sizeof(*pim))) { |
b1879204 IJ |
2049 | drop: |
2050 | kfree_skb(skb); | |
2051 | } | |
1da177e4 LT |
2052 | return 0; |
2053 | } | |
2054 | #endif | |
2055 | ||
2056 | #ifdef CONFIG_IP_PIMSM_V2 | |
a8cb16dd | 2057 | static int pim_rcv(struct sk_buff *skb) |
1da177e4 LT |
2058 | { |
2059 | struct pimreghdr *pim; | |
f0ad0860 PM |
2060 | struct net *net = dev_net(skb->dev); |
2061 | struct mr_table *mrt; | |
1da177e4 | 2062 | |
b1879204 | 2063 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) |
1da177e4 LT |
2064 | goto drop; |
2065 | ||
9c70220b | 2066 | pim = (struct pimreghdr *)skb_transport_header(skb); |
56245cae | 2067 | if (pim->type != ((PIM_VERSION << 4) | (PIM_TYPE_REGISTER)) || |
a8cb16dd | 2068 | (pim->flags & PIM_NULL_REGISTER) || |
e905a9ed | 2069 | (ip_compute_csum((void *)pim, sizeof(*pim)) != 0 && |
d3bc23e7 | 2070 | csum_fold(skb_checksum(skb, 0, skb->len, 0)))) |
1da177e4 LT |
2071 | goto drop; |
2072 | ||
417da66f | 2073 | mrt = ipmr_rt_fib_lookup(net, skb); |
ee3f1aaf DM |
2074 | if (IS_ERR(mrt)) |
2075 | goto drop; | |
f0ad0860 | 2076 | if (__pim_rcv(mrt, skb, sizeof(*pim))) { |
b1879204 IJ |
2077 | drop: |
2078 | kfree_skb(skb); | |
2079 | } | |
1da177e4 LT |
2080 | return 0; |
2081 | } | |
2082 | #endif | |
2083 | ||
cb6a4e46 PM |
2084 | static int __ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
2085 | struct mfc_cache *c, struct rtmsg *rtm) | |
1da177e4 | 2086 | { |
adfa85e4 | 2087 | struct rta_mfc_stats mfcs; |
43b9e127 NA |
2088 | struct nlattr *mp_attr; |
2089 | struct rtnexthop *nhp; | |
b5036cd4 | 2090 | unsigned long lastuse; |
43b9e127 | 2091 | int ct; |
1da177e4 | 2092 | |
7438189b | 2093 | /* If cache is unresolved, don't try to parse IIF and OIF */ |
ed0f160a | 2094 | if (c->mfc_parent >= MAXVIFS) |
7438189b ND |
2095 | return -ENOENT; |
2096 | ||
92a395e5 TG |
2097 | if (VIF_EXISTS(mrt, c->mfc_parent) && |
2098 | nla_put_u32(skb, RTA_IIF, mrt->vif_table[c->mfc_parent].dev->ifindex) < 0) | |
2099 | return -EMSGSIZE; | |
1da177e4 | 2100 | |
92a395e5 TG |
2101 | if (!(mp_attr = nla_nest_start(skb, RTA_MULTIPATH))) |
2102 | return -EMSGSIZE; | |
1da177e4 LT |
2103 | |
2104 | for (ct = c->mfc_un.res.minvif; ct < c->mfc_un.res.maxvif; ct++) { | |
0c12295a | 2105 | if (VIF_EXISTS(mrt, ct) && c->mfc_un.res.ttls[ct] < 255) { |
92a395e5 TG |
2106 | if (!(nhp = nla_reserve_nohdr(skb, sizeof(*nhp)))) { |
2107 | nla_nest_cancel(skb, mp_attr); | |
2108 | return -EMSGSIZE; | |
2109 | } | |
2110 | ||
1da177e4 LT |
2111 | nhp->rtnh_flags = 0; |
2112 | nhp->rtnh_hops = c->mfc_un.res.ttls[ct]; | |
0c12295a | 2113 | nhp->rtnh_ifindex = mrt->vif_table[ct].dev->ifindex; |
1da177e4 LT |
2114 | nhp->rtnh_len = sizeof(*nhp); |
2115 | } | |
2116 | } | |
92a395e5 TG |
2117 | |
2118 | nla_nest_end(skb, mp_attr); | |
2119 | ||
b5036cd4 NA |
2120 | lastuse = READ_ONCE(c->mfc_un.res.lastuse); |
2121 | lastuse = time_after_eq(jiffies, lastuse) ? jiffies - lastuse : 0; | |
2122 | ||
adfa85e4 ND |
2123 | mfcs.mfcs_packets = c->mfc_un.res.pkt; |
2124 | mfcs.mfcs_bytes = c->mfc_un.res.bytes; | |
2125 | mfcs.mfcs_wrong_if = c->mfc_un.res.wrong_if; | |
43b9e127 | 2126 | if (nla_put_64bit(skb, RTA_MFC_STATS, sizeof(mfcs), &mfcs, RTA_PAD) || |
b5036cd4 | 2127 | nla_put_u64_64bit(skb, RTA_EXPIRES, jiffies_to_clock_t(lastuse), |
43b9e127 | 2128 | RTA_PAD)) |
adfa85e4 ND |
2129 | return -EMSGSIZE; |
2130 | ||
1da177e4 LT |
2131 | rtm->rtm_type = RTN_MULTICAST; |
2132 | return 1; | |
1da177e4 LT |
2133 | } |
2134 | ||
9a1b9496 DM |
2135 | int ipmr_get_route(struct net *net, struct sk_buff *skb, |
2136 | __be32 saddr, __be32 daddr, | |
2cf75070 | 2137 | struct rtmsg *rtm, int nowait, u32 portid) |
1da177e4 | 2138 | { |
1da177e4 | 2139 | struct mfc_cache *cache; |
9a1b9496 DM |
2140 | struct mr_table *mrt; |
2141 | int err; | |
1da177e4 | 2142 | |
f0ad0860 | 2143 | mrt = ipmr_get_table(net, RT_TABLE_DEFAULT); |
51456b29 | 2144 | if (!mrt) |
f0ad0860 PM |
2145 | return -ENOENT; |
2146 | ||
a8c9486b | 2147 | rcu_read_lock(); |
9a1b9496 | 2148 | cache = ipmr_cache_find(mrt, saddr, daddr); |
51456b29 | 2149 | if (!cache && skb->dev) { |
660b26dc | 2150 | int vif = ipmr_find_vif(mrt, skb->dev); |
1da177e4 | 2151 | |
660b26dc ND |
2152 | if (vif >= 0) |
2153 | cache = ipmr_cache_find_any(mrt, daddr, vif); | |
2154 | } | |
51456b29 | 2155 | if (!cache) { |
72287490 | 2156 | struct sk_buff *skb2; |
eddc9ec5 | 2157 | struct iphdr *iph; |
1da177e4 | 2158 | struct net_device *dev; |
a8cb16dd | 2159 | int vif = -1; |
1da177e4 LT |
2160 | |
2161 | if (nowait) { | |
a8c9486b | 2162 | rcu_read_unlock(); |
1da177e4 LT |
2163 | return -EAGAIN; |
2164 | } | |
2165 | ||
2166 | dev = skb->dev; | |
a8c9486b | 2167 | read_lock(&mrt_lock); |
a8cb16dd ED |
2168 | if (dev) |
2169 | vif = ipmr_find_vif(mrt, dev); | |
2170 | if (vif < 0) { | |
1da177e4 | 2171 | read_unlock(&mrt_lock); |
a8c9486b | 2172 | rcu_read_unlock(); |
1da177e4 LT |
2173 | return -ENODEV; |
2174 | } | |
72287490 AK |
2175 | skb2 = skb_clone(skb, GFP_ATOMIC); |
2176 | if (!skb2) { | |
2177 | read_unlock(&mrt_lock); | |
a8c9486b | 2178 | rcu_read_unlock(); |
72287490 AK |
2179 | return -ENOMEM; |
2180 | } | |
2181 | ||
2cf75070 | 2182 | NETLINK_CB(skb2).portid = portid; |
e2d1bca7 ACM |
2183 | skb_push(skb2, sizeof(struct iphdr)); |
2184 | skb_reset_network_header(skb2); | |
eddc9ec5 ACM |
2185 | iph = ip_hdr(skb2); |
2186 | iph->ihl = sizeof(struct iphdr) >> 2; | |
9a1b9496 DM |
2187 | iph->saddr = saddr; |
2188 | iph->daddr = daddr; | |
eddc9ec5 | 2189 | iph->version = 0; |
0c12295a | 2190 | err = ipmr_cache_unresolved(mrt, vif, skb2); |
1da177e4 | 2191 | read_unlock(&mrt_lock); |
a8c9486b | 2192 | rcu_read_unlock(); |
1da177e4 LT |
2193 | return err; |
2194 | } | |
2195 | ||
a8c9486b | 2196 | read_lock(&mrt_lock); |
cb6a4e46 | 2197 | err = __ipmr_fill_mroute(mrt, skb, cache, rtm); |
1da177e4 | 2198 | read_unlock(&mrt_lock); |
a8c9486b | 2199 | rcu_read_unlock(); |
1da177e4 LT |
2200 | return err; |
2201 | } | |
2202 | ||
cb6a4e46 | 2203 | static int ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
65886f43 ND |
2204 | u32 portid, u32 seq, struct mfc_cache *c, int cmd, |
2205 | int flags) | |
cb6a4e46 PM |
2206 | { |
2207 | struct nlmsghdr *nlh; | |
2208 | struct rtmsg *rtm; | |
1eb99af5 | 2209 | int err; |
cb6a4e46 | 2210 | |
65886f43 | 2211 | nlh = nlmsg_put(skb, portid, seq, cmd, sizeof(*rtm), flags); |
51456b29 | 2212 | if (!nlh) |
cb6a4e46 PM |
2213 | return -EMSGSIZE; |
2214 | ||
2215 | rtm = nlmsg_data(nlh); | |
2216 | rtm->rtm_family = RTNL_FAMILY_IPMR; | |
2217 | rtm->rtm_dst_len = 32; | |
2218 | rtm->rtm_src_len = 32; | |
2219 | rtm->rtm_tos = 0; | |
2220 | rtm->rtm_table = mrt->id; | |
f3756b79 DM |
2221 | if (nla_put_u32(skb, RTA_TABLE, mrt->id)) |
2222 | goto nla_put_failure; | |
cb6a4e46 PM |
2223 | rtm->rtm_type = RTN_MULTICAST; |
2224 | rtm->rtm_scope = RT_SCOPE_UNIVERSE; | |
9a68ac72 ND |
2225 | if (c->mfc_flags & MFC_STATIC) |
2226 | rtm->rtm_protocol = RTPROT_STATIC; | |
2227 | else | |
2228 | rtm->rtm_protocol = RTPROT_MROUTED; | |
cb6a4e46 PM |
2229 | rtm->rtm_flags = 0; |
2230 | ||
930345ea JB |
2231 | if (nla_put_in_addr(skb, RTA_SRC, c->mfc_origin) || |
2232 | nla_put_in_addr(skb, RTA_DST, c->mfc_mcastgrp)) | |
f3756b79 | 2233 | goto nla_put_failure; |
1eb99af5 ND |
2234 | err = __ipmr_fill_mroute(mrt, skb, c, rtm); |
2235 | /* do not break the dump if cache is unresolved */ | |
2236 | if (err < 0 && err != -ENOENT) | |
cb6a4e46 PM |
2237 | goto nla_put_failure; |
2238 | ||
053c095a JB |
2239 | nlmsg_end(skb, nlh); |
2240 | return 0; | |
cb6a4e46 PM |
2241 | |
2242 | nla_put_failure: | |
2243 | nlmsg_cancel(skb, nlh); | |
2244 | return -EMSGSIZE; | |
2245 | } | |
2246 | ||
8cd3ac9f ND |
2247 | static size_t mroute_msgsize(bool unresolved, int maxvif) |
2248 | { | |
2249 | size_t len = | |
2250 | NLMSG_ALIGN(sizeof(struct rtmsg)) | |
2251 | + nla_total_size(4) /* RTA_TABLE */ | |
2252 | + nla_total_size(4) /* RTA_SRC */ | |
2253 | + nla_total_size(4) /* RTA_DST */ | |
2254 | ; | |
2255 | ||
2256 | if (!unresolved) | |
2257 | len = len | |
2258 | + nla_total_size(4) /* RTA_IIF */ | |
2259 | + nla_total_size(0) /* RTA_MULTIPATH */ | |
2260 | + maxvif * NLA_ALIGN(sizeof(struct rtnexthop)) | |
2261 | /* RTA_MFC_STATS */ | |
a9a08042 | 2262 | + nla_total_size_64bit(sizeof(struct rta_mfc_stats)) |
8cd3ac9f ND |
2263 | ; |
2264 | ||
2265 | return len; | |
2266 | } | |
2267 | ||
2268 | static void mroute_netlink_event(struct mr_table *mrt, struct mfc_cache *mfc, | |
2269 | int cmd) | |
2270 | { | |
2271 | struct net *net = read_pnet(&mrt->net); | |
2272 | struct sk_buff *skb; | |
2273 | int err = -ENOBUFS; | |
2274 | ||
2275 | skb = nlmsg_new(mroute_msgsize(mfc->mfc_parent >= MAXVIFS, mrt->maxvif), | |
2276 | GFP_ATOMIC); | |
51456b29 | 2277 | if (!skb) |
8cd3ac9f ND |
2278 | goto errout; |
2279 | ||
65886f43 | 2280 | err = ipmr_fill_mroute(mrt, skb, 0, 0, mfc, cmd, 0); |
8cd3ac9f ND |
2281 | if (err < 0) |
2282 | goto errout; | |
2283 | ||
2284 | rtnl_notify(skb, net, 0, RTNLGRP_IPV4_MROUTE, NULL, GFP_ATOMIC); | |
2285 | return; | |
2286 | ||
2287 | errout: | |
2288 | kfree_skb(skb); | |
2289 | if (err < 0) | |
2290 | rtnl_set_sk_err(net, RTNLGRP_IPV4_MROUTE, err); | |
2291 | } | |
2292 | ||
cb6a4e46 PM |
2293 | static int ipmr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb) |
2294 | { | |
2295 | struct net *net = sock_net(skb->sk); | |
2296 | struct mr_table *mrt; | |
2297 | struct mfc_cache *mfc; | |
2298 | unsigned int t = 0, s_t; | |
2299 | unsigned int h = 0, s_h; | |
2300 | unsigned int e = 0, s_e; | |
2301 | ||
2302 | s_t = cb->args[0]; | |
2303 | s_h = cb->args[1]; | |
2304 | s_e = cb->args[2]; | |
2305 | ||
a8c9486b | 2306 | rcu_read_lock(); |
cb6a4e46 PM |
2307 | ipmr_for_each_table(mrt, net) { |
2308 | if (t < s_t) | |
2309 | goto next_table; | |
2310 | if (t > s_t) | |
2311 | s_h = 0; | |
2312 | for (h = s_h; h < MFC_LINES; h++) { | |
a8c9486b | 2313 | list_for_each_entry_rcu(mfc, &mrt->mfc_cache_array[h], list) { |
cb6a4e46 PM |
2314 | if (e < s_e) |
2315 | goto next_entry; | |
2316 | if (ipmr_fill_mroute(mrt, skb, | |
15e47304 | 2317 | NETLINK_CB(cb->skb).portid, |
cb6a4e46 | 2318 | cb->nlh->nlmsg_seq, |
65886f43 ND |
2319 | mfc, RTM_NEWROUTE, |
2320 | NLM_F_MULTI) < 0) | |
cb6a4e46 PM |
2321 | goto done; |
2322 | next_entry: | |
2323 | e++; | |
2324 | } | |
2325 | e = s_e = 0; | |
2326 | } | |
1eb99af5 ND |
2327 | spin_lock_bh(&mfc_unres_lock); |
2328 | list_for_each_entry(mfc, &mrt->mfc_unres_queue, list) { | |
2329 | if (e < s_e) | |
2330 | goto next_entry2; | |
2331 | if (ipmr_fill_mroute(mrt, skb, | |
2332 | NETLINK_CB(cb->skb).portid, | |
2333 | cb->nlh->nlmsg_seq, | |
65886f43 ND |
2334 | mfc, RTM_NEWROUTE, |
2335 | NLM_F_MULTI) < 0) { | |
1eb99af5 ND |
2336 | spin_unlock_bh(&mfc_unres_lock); |
2337 | goto done; | |
2338 | } | |
2339 | next_entry2: | |
2340 | e++; | |
2341 | } | |
2342 | spin_unlock_bh(&mfc_unres_lock); | |
2343 | e = s_e = 0; | |
cb6a4e46 PM |
2344 | s_h = 0; |
2345 | next_table: | |
2346 | t++; | |
2347 | } | |
2348 | done: | |
a8c9486b | 2349 | rcu_read_unlock(); |
cb6a4e46 PM |
2350 | |
2351 | cb->args[2] = e; | |
2352 | cb->args[1] = h; | |
2353 | cb->args[0] = t; | |
2354 | ||
2355 | return skb->len; | |
2356 | } | |
2357 | ||
ccbb0aa6 NA |
2358 | static const struct nla_policy rtm_ipmr_policy[RTA_MAX + 1] = { |
2359 | [RTA_SRC] = { .type = NLA_U32 }, | |
2360 | [RTA_DST] = { .type = NLA_U32 }, | |
2361 | [RTA_IIF] = { .type = NLA_U32 }, | |
2362 | [RTA_TABLE] = { .type = NLA_U32 }, | |
2363 | [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) }, | |
2364 | }; | |
2365 | ||
2366 | static bool ipmr_rtm_validate_proto(unsigned char rtm_protocol) | |
2367 | { | |
2368 | switch (rtm_protocol) { | |
2369 | case RTPROT_STATIC: | |
2370 | case RTPROT_MROUTED: | |
2371 | return true; | |
2372 | } | |
2373 | return false; | |
2374 | } | |
2375 | ||
2376 | static int ipmr_nla_get_ttls(const struct nlattr *nla, struct mfcctl *mfcc) | |
2377 | { | |
2378 | struct rtnexthop *rtnh = nla_data(nla); | |
2379 | int remaining = nla_len(nla), vifi = 0; | |
2380 | ||
2381 | while (rtnh_ok(rtnh, remaining)) { | |
2382 | mfcc->mfcc_ttls[vifi] = rtnh->rtnh_hops; | |
2383 | if (++vifi == MAXVIFS) | |
2384 | break; | |
2385 | rtnh = rtnh_next(rtnh, &remaining); | |
2386 | } | |
2387 | ||
2388 | return remaining > 0 ? -EINVAL : vifi; | |
2389 | } | |
2390 | ||
2391 | /* returns < 0 on error, 0 for ADD_MFC and 1 for ADD_MFC_PROXY */ | |
2392 | static int rtm_to_ipmr_mfcc(struct net *net, struct nlmsghdr *nlh, | |
2393 | struct mfcctl *mfcc, int *mrtsock, | |
2394 | struct mr_table **mrtret) | |
2395 | { | |
2396 | struct net_device *dev = NULL; | |
2397 | u32 tblid = RT_TABLE_DEFAULT; | |
2398 | struct mr_table *mrt; | |
2399 | struct nlattr *attr; | |
2400 | struct rtmsg *rtm; | |
2401 | int ret, rem; | |
2402 | ||
2403 | ret = nlmsg_validate(nlh, sizeof(*rtm), RTA_MAX, rtm_ipmr_policy); | |
2404 | if (ret < 0) | |
2405 | goto out; | |
2406 | rtm = nlmsg_data(nlh); | |
2407 | ||
2408 | ret = -EINVAL; | |
2409 | if (rtm->rtm_family != RTNL_FAMILY_IPMR || rtm->rtm_dst_len != 32 || | |
2410 | rtm->rtm_type != RTN_MULTICAST || | |
2411 | rtm->rtm_scope != RT_SCOPE_UNIVERSE || | |
2412 | !ipmr_rtm_validate_proto(rtm->rtm_protocol)) | |
2413 | goto out; | |
2414 | ||
2415 | memset(mfcc, 0, sizeof(*mfcc)); | |
2416 | mfcc->mfcc_parent = -1; | |
2417 | ret = 0; | |
2418 | nlmsg_for_each_attr(attr, nlh, sizeof(struct rtmsg), rem) { | |
2419 | switch (nla_type(attr)) { | |
2420 | case RTA_SRC: | |
2421 | mfcc->mfcc_origin.s_addr = nla_get_be32(attr); | |
2422 | break; | |
2423 | case RTA_DST: | |
2424 | mfcc->mfcc_mcastgrp.s_addr = nla_get_be32(attr); | |
2425 | break; | |
2426 | case RTA_IIF: | |
2427 | dev = __dev_get_by_index(net, nla_get_u32(attr)); | |
2428 | if (!dev) { | |
2429 | ret = -ENODEV; | |
2430 | goto out; | |
2431 | } | |
2432 | break; | |
2433 | case RTA_MULTIPATH: | |
2434 | if (ipmr_nla_get_ttls(attr, mfcc) < 0) { | |
2435 | ret = -EINVAL; | |
2436 | goto out; | |
2437 | } | |
2438 | break; | |
2439 | case RTA_PREFSRC: | |
2440 | ret = 1; | |
2441 | break; | |
2442 | case RTA_TABLE: | |
2443 | tblid = nla_get_u32(attr); | |
2444 | break; | |
2445 | } | |
2446 | } | |
2447 | mrt = ipmr_get_table(net, tblid); | |
2448 | if (!mrt) { | |
2449 | ret = -ENOENT; | |
2450 | goto out; | |
2451 | } | |
2452 | *mrtret = mrt; | |
2453 | *mrtsock = rtm->rtm_protocol == RTPROT_MROUTED ? 1 : 0; | |
2454 | if (dev) | |
2455 | mfcc->mfcc_parent = ipmr_find_vif(mrt, dev); | |
2456 | ||
2457 | out: | |
2458 | return ret; | |
2459 | } | |
2460 | ||
2461 | /* takes care of both newroute and delroute */ | |
2462 | static int ipmr_rtm_route(struct sk_buff *skb, struct nlmsghdr *nlh) | |
2463 | { | |
2464 | struct net *net = sock_net(skb->sk); | |
2465 | int ret, mrtsock, parent; | |
2466 | struct mr_table *tbl; | |
2467 | struct mfcctl mfcc; | |
2468 | ||
2469 | mrtsock = 0; | |
2470 | tbl = NULL; | |
2471 | ret = rtm_to_ipmr_mfcc(net, nlh, &mfcc, &mrtsock, &tbl); | |
2472 | if (ret < 0) | |
2473 | return ret; | |
2474 | ||
2475 | parent = ret ? mfcc.mfcc_parent : -1; | |
2476 | if (nlh->nlmsg_type == RTM_NEWROUTE) | |
2477 | return ipmr_mfc_add(net, tbl, &mfcc, mrtsock, parent); | |
2478 | else | |
2479 | return ipmr_mfc_delete(tbl, &mfcc, parent); | |
2480 | } | |
2481 | ||
e905a9ed | 2482 | #ifdef CONFIG_PROC_FS |
7ef8f65d NA |
2483 | /* The /proc interfaces to multicast routing : |
2484 | * /proc/net/ip_mr_cache & /proc/net/ip_mr_vif | |
1da177e4 LT |
2485 | */ |
2486 | struct ipmr_vif_iter { | |
f6bb4514 | 2487 | struct seq_net_private p; |
f0ad0860 | 2488 | struct mr_table *mrt; |
1da177e4 LT |
2489 | int ct; |
2490 | }; | |
2491 | ||
f6bb4514 BT |
2492 | static struct vif_device *ipmr_vif_seq_idx(struct net *net, |
2493 | struct ipmr_vif_iter *iter, | |
1da177e4 LT |
2494 | loff_t pos) |
2495 | { | |
f0ad0860 | 2496 | struct mr_table *mrt = iter->mrt; |
0c12295a PM |
2497 | |
2498 | for (iter->ct = 0; iter->ct < mrt->maxvif; ++iter->ct) { | |
2499 | if (!VIF_EXISTS(mrt, iter->ct)) | |
1da177e4 | 2500 | continue; |
e905a9ed | 2501 | if (pos-- == 0) |
0c12295a | 2502 | return &mrt->vif_table[iter->ct]; |
1da177e4 LT |
2503 | } |
2504 | return NULL; | |
2505 | } | |
2506 | ||
2507 | static void *ipmr_vif_seq_start(struct seq_file *seq, loff_t *pos) | |
ba93ef74 | 2508 | __acquires(mrt_lock) |
1da177e4 | 2509 | { |
f0ad0860 | 2510 | struct ipmr_vif_iter *iter = seq->private; |
f6bb4514 | 2511 | struct net *net = seq_file_net(seq); |
f0ad0860 PM |
2512 | struct mr_table *mrt; |
2513 | ||
2514 | mrt = ipmr_get_table(net, RT_TABLE_DEFAULT); | |
51456b29 | 2515 | if (!mrt) |
f0ad0860 PM |
2516 | return ERR_PTR(-ENOENT); |
2517 | ||
2518 | iter->mrt = mrt; | |
f6bb4514 | 2519 | |
1da177e4 | 2520 | read_lock(&mrt_lock); |
f6bb4514 | 2521 | return *pos ? ipmr_vif_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
2522 | : SEQ_START_TOKEN; |
2523 | } | |
2524 | ||
2525 | static void *ipmr_vif_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
2526 | { | |
2527 | struct ipmr_vif_iter *iter = seq->private; | |
f6bb4514 | 2528 | struct net *net = seq_file_net(seq); |
f0ad0860 | 2529 | struct mr_table *mrt = iter->mrt; |
1da177e4 LT |
2530 | |
2531 | ++*pos; | |
2532 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 2533 | return ipmr_vif_seq_idx(net, iter, 0); |
e905a9ed | 2534 | |
0c12295a PM |
2535 | while (++iter->ct < mrt->maxvif) { |
2536 | if (!VIF_EXISTS(mrt, iter->ct)) | |
1da177e4 | 2537 | continue; |
0c12295a | 2538 | return &mrt->vif_table[iter->ct]; |
1da177e4 LT |
2539 | } |
2540 | return NULL; | |
2541 | } | |
2542 | ||
2543 | static void ipmr_vif_seq_stop(struct seq_file *seq, void *v) | |
ba93ef74 | 2544 | __releases(mrt_lock) |
1da177e4 LT |
2545 | { |
2546 | read_unlock(&mrt_lock); | |
2547 | } | |
2548 | ||
2549 | static int ipmr_vif_seq_show(struct seq_file *seq, void *v) | |
2550 | { | |
f0ad0860 PM |
2551 | struct ipmr_vif_iter *iter = seq->private; |
2552 | struct mr_table *mrt = iter->mrt; | |
f6bb4514 | 2553 | |
1da177e4 | 2554 | if (v == SEQ_START_TOKEN) { |
e905a9ed | 2555 | seq_puts(seq, |
1da177e4 LT |
2556 | "Interface BytesIn PktsIn BytesOut PktsOut Flags Local Remote\n"); |
2557 | } else { | |
2558 | const struct vif_device *vif = v; | |
2559 | const char *name = vif->dev ? vif->dev->name : "none"; | |
2560 | ||
2561 | seq_printf(seq, | |
2562 | "%2Zd %-10s %8ld %7ld %8ld %7ld %05X %08X %08X\n", | |
0c12295a | 2563 | vif - mrt->vif_table, |
e905a9ed | 2564 | name, vif->bytes_in, vif->pkt_in, |
1da177e4 LT |
2565 | vif->bytes_out, vif->pkt_out, |
2566 | vif->flags, vif->local, vif->remote); | |
2567 | } | |
2568 | return 0; | |
2569 | } | |
2570 | ||
f690808e | 2571 | static const struct seq_operations ipmr_vif_seq_ops = { |
1da177e4 LT |
2572 | .start = ipmr_vif_seq_start, |
2573 | .next = ipmr_vif_seq_next, | |
2574 | .stop = ipmr_vif_seq_stop, | |
2575 | .show = ipmr_vif_seq_show, | |
2576 | }; | |
2577 | ||
2578 | static int ipmr_vif_open(struct inode *inode, struct file *file) | |
2579 | { | |
f6bb4514 BT |
2580 | return seq_open_net(inode, file, &ipmr_vif_seq_ops, |
2581 | sizeof(struct ipmr_vif_iter)); | |
1da177e4 LT |
2582 | } |
2583 | ||
9a32144e | 2584 | static const struct file_operations ipmr_vif_fops = { |
1da177e4 LT |
2585 | .owner = THIS_MODULE, |
2586 | .open = ipmr_vif_open, | |
2587 | .read = seq_read, | |
2588 | .llseek = seq_lseek, | |
f6bb4514 | 2589 | .release = seq_release_net, |
1da177e4 LT |
2590 | }; |
2591 | ||
2592 | struct ipmr_mfc_iter { | |
f6bb4514 | 2593 | struct seq_net_private p; |
f0ad0860 | 2594 | struct mr_table *mrt; |
862465f2 | 2595 | struct list_head *cache; |
1da177e4 LT |
2596 | int ct; |
2597 | }; | |
2598 | ||
2599 | ||
f6bb4514 BT |
2600 | static struct mfc_cache *ipmr_mfc_seq_idx(struct net *net, |
2601 | struct ipmr_mfc_iter *it, loff_t pos) | |
1da177e4 | 2602 | { |
f0ad0860 | 2603 | struct mr_table *mrt = it->mrt; |
1da177e4 LT |
2604 | struct mfc_cache *mfc; |
2605 | ||
a8c9486b | 2606 | rcu_read_lock(); |
862465f2 | 2607 | for (it->ct = 0; it->ct < MFC_LINES; it->ct++) { |
0c12295a | 2608 | it->cache = &mrt->mfc_cache_array[it->ct]; |
a8c9486b | 2609 | list_for_each_entry_rcu(mfc, it->cache, list) |
e905a9ed | 2610 | if (pos-- == 0) |
1da177e4 | 2611 | return mfc; |
862465f2 | 2612 | } |
a8c9486b | 2613 | rcu_read_unlock(); |
1da177e4 | 2614 | |
1da177e4 | 2615 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 2616 | it->cache = &mrt->mfc_unres_queue; |
862465f2 | 2617 | list_for_each_entry(mfc, it->cache, list) |
e258beb2 | 2618 | if (pos-- == 0) |
1da177e4 LT |
2619 | return mfc; |
2620 | spin_unlock_bh(&mfc_unres_lock); | |
2621 | ||
2622 | it->cache = NULL; | |
2623 | return NULL; | |
2624 | } | |
2625 | ||
2626 | ||
2627 | static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos) | |
2628 | { | |
2629 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 | 2630 | struct net *net = seq_file_net(seq); |
f0ad0860 | 2631 | struct mr_table *mrt; |
f6bb4514 | 2632 | |
f0ad0860 | 2633 | mrt = ipmr_get_table(net, RT_TABLE_DEFAULT); |
51456b29 | 2634 | if (!mrt) |
f0ad0860 | 2635 | return ERR_PTR(-ENOENT); |
f6bb4514 | 2636 | |
f0ad0860 | 2637 | it->mrt = mrt; |
1da177e4 LT |
2638 | it->cache = NULL; |
2639 | it->ct = 0; | |
f6bb4514 | 2640 | return *pos ? ipmr_mfc_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
2641 | : SEQ_START_TOKEN; |
2642 | } | |
2643 | ||
2644 | static void *ipmr_mfc_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
2645 | { | |
2646 | struct mfc_cache *mfc = v; | |
2647 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 | 2648 | struct net *net = seq_file_net(seq); |
f0ad0860 | 2649 | struct mr_table *mrt = it->mrt; |
1da177e4 LT |
2650 | |
2651 | ++*pos; | |
2652 | ||
2653 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 2654 | return ipmr_mfc_seq_idx(net, seq->private, 0); |
1da177e4 | 2655 | |
862465f2 PM |
2656 | if (mfc->list.next != it->cache) |
2657 | return list_entry(mfc->list.next, struct mfc_cache, list); | |
e905a9ed | 2658 | |
0c12295a | 2659 | if (it->cache == &mrt->mfc_unres_queue) |
1da177e4 LT |
2660 | goto end_of_list; |
2661 | ||
0c12295a | 2662 | BUG_ON(it->cache != &mrt->mfc_cache_array[it->ct]); |
1da177e4 LT |
2663 | |
2664 | while (++it->ct < MFC_LINES) { | |
0c12295a | 2665 | it->cache = &mrt->mfc_cache_array[it->ct]; |
862465f2 PM |
2666 | if (list_empty(it->cache)) |
2667 | continue; | |
2668 | return list_first_entry(it->cache, struct mfc_cache, list); | |
1da177e4 LT |
2669 | } |
2670 | ||
2671 | /* exhausted cache_array, show unresolved */ | |
a8c9486b | 2672 | rcu_read_unlock(); |
0c12295a | 2673 | it->cache = &mrt->mfc_unres_queue; |
1da177e4 | 2674 | it->ct = 0; |
e905a9ed | 2675 | |
1da177e4 | 2676 | spin_lock_bh(&mfc_unres_lock); |
862465f2 PM |
2677 | if (!list_empty(it->cache)) |
2678 | return list_first_entry(it->cache, struct mfc_cache, list); | |
1da177e4 | 2679 | |
a8cb16dd | 2680 | end_of_list: |
1da177e4 LT |
2681 | spin_unlock_bh(&mfc_unres_lock); |
2682 | it->cache = NULL; | |
2683 | ||
2684 | return NULL; | |
2685 | } | |
2686 | ||
2687 | static void ipmr_mfc_seq_stop(struct seq_file *seq, void *v) | |
2688 | { | |
2689 | struct ipmr_mfc_iter *it = seq->private; | |
f0ad0860 | 2690 | struct mr_table *mrt = it->mrt; |
1da177e4 | 2691 | |
0c12295a | 2692 | if (it->cache == &mrt->mfc_unres_queue) |
1da177e4 | 2693 | spin_unlock_bh(&mfc_unres_lock); |
0c12295a | 2694 | else if (it->cache == &mrt->mfc_cache_array[it->ct]) |
a8c9486b | 2695 | rcu_read_unlock(); |
1da177e4 LT |
2696 | } |
2697 | ||
2698 | static int ipmr_mfc_seq_show(struct seq_file *seq, void *v) | |
2699 | { | |
2700 | int n; | |
2701 | ||
2702 | if (v == SEQ_START_TOKEN) { | |
e905a9ed | 2703 | seq_puts(seq, |
1da177e4 LT |
2704 | "Group Origin Iif Pkts Bytes Wrong Oifs\n"); |
2705 | } else { | |
2706 | const struct mfc_cache *mfc = v; | |
2707 | const struct ipmr_mfc_iter *it = seq->private; | |
f0ad0860 | 2708 | const struct mr_table *mrt = it->mrt; |
e905a9ed | 2709 | |
0eae88f3 ED |
2710 | seq_printf(seq, "%08X %08X %-3hd", |
2711 | (__force u32) mfc->mfc_mcastgrp, | |
2712 | (__force u32) mfc->mfc_origin, | |
1ea472e2 | 2713 | mfc->mfc_parent); |
1da177e4 | 2714 | |
0c12295a | 2715 | if (it->cache != &mrt->mfc_unres_queue) { |
1ea472e2 BT |
2716 | seq_printf(seq, " %8lu %8lu %8lu", |
2717 | mfc->mfc_un.res.pkt, | |
2718 | mfc->mfc_un.res.bytes, | |
2719 | mfc->mfc_un.res.wrong_if); | |
132adf54 | 2720 | for (n = mfc->mfc_un.res.minvif; |
a8cb16dd | 2721 | n < mfc->mfc_un.res.maxvif; n++) { |
0c12295a | 2722 | if (VIF_EXISTS(mrt, n) && |
cf958ae3 BT |
2723 | mfc->mfc_un.res.ttls[n] < 255) |
2724 | seq_printf(seq, | |
e905a9ed | 2725 | " %2d:%-3d", |
1da177e4 LT |
2726 | n, mfc->mfc_un.res.ttls[n]); |
2727 | } | |
1ea472e2 BT |
2728 | } else { |
2729 | /* unresolved mfc_caches don't contain | |
2730 | * pkt, bytes and wrong_if values | |
2731 | */ | |
2732 | seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul); | |
1da177e4 LT |
2733 | } |
2734 | seq_putc(seq, '\n'); | |
2735 | } | |
2736 | return 0; | |
2737 | } | |
2738 | ||
f690808e | 2739 | static const struct seq_operations ipmr_mfc_seq_ops = { |
1da177e4 LT |
2740 | .start = ipmr_mfc_seq_start, |
2741 | .next = ipmr_mfc_seq_next, | |
2742 | .stop = ipmr_mfc_seq_stop, | |
2743 | .show = ipmr_mfc_seq_show, | |
2744 | }; | |
2745 | ||
2746 | static int ipmr_mfc_open(struct inode *inode, struct file *file) | |
2747 | { | |
f6bb4514 BT |
2748 | return seq_open_net(inode, file, &ipmr_mfc_seq_ops, |
2749 | sizeof(struct ipmr_mfc_iter)); | |
1da177e4 LT |
2750 | } |
2751 | ||
9a32144e | 2752 | static const struct file_operations ipmr_mfc_fops = { |
1da177e4 LT |
2753 | .owner = THIS_MODULE, |
2754 | .open = ipmr_mfc_open, | |
2755 | .read = seq_read, | |
2756 | .llseek = seq_lseek, | |
f6bb4514 | 2757 | .release = seq_release_net, |
1da177e4 | 2758 | }; |
e905a9ed | 2759 | #endif |
1da177e4 LT |
2760 | |
2761 | #ifdef CONFIG_IP_PIMSM_V2 | |
32613090 | 2762 | static const struct net_protocol pim_protocol = { |
1da177e4 | 2763 | .handler = pim_rcv, |
403dbb97 | 2764 | .netns_ok = 1, |
1da177e4 LT |
2765 | }; |
2766 | #endif | |
2767 | ||
7ef8f65d | 2768 | /* Setup for IP multicast routing */ |
cf958ae3 BT |
2769 | static int __net_init ipmr_net_init(struct net *net) |
2770 | { | |
f0ad0860 | 2771 | int err; |
cf958ae3 | 2772 | |
f0ad0860 PM |
2773 | err = ipmr_rules_init(net); |
2774 | if (err < 0) | |
cf958ae3 | 2775 | goto fail; |
f6bb4514 BT |
2776 | |
2777 | #ifdef CONFIG_PROC_FS | |
2778 | err = -ENOMEM; | |
d4beaa66 | 2779 | if (!proc_create("ip_mr_vif", 0, net->proc_net, &ipmr_vif_fops)) |
f6bb4514 | 2780 | goto proc_vif_fail; |
d4beaa66 | 2781 | if (!proc_create("ip_mr_cache", 0, net->proc_net, &ipmr_mfc_fops)) |
f6bb4514 BT |
2782 | goto proc_cache_fail; |
2783 | #endif | |
2bb8b26c BT |
2784 | return 0; |
2785 | ||
f6bb4514 BT |
2786 | #ifdef CONFIG_PROC_FS |
2787 | proc_cache_fail: | |
ece31ffd | 2788 | remove_proc_entry("ip_mr_vif", net->proc_net); |
f6bb4514 | 2789 | proc_vif_fail: |
f0ad0860 | 2790 | ipmr_rules_exit(net); |
f6bb4514 | 2791 | #endif |
cf958ae3 BT |
2792 | fail: |
2793 | return err; | |
2794 | } | |
2795 | ||
2796 | static void __net_exit ipmr_net_exit(struct net *net) | |
2797 | { | |
f6bb4514 | 2798 | #ifdef CONFIG_PROC_FS |
ece31ffd G |
2799 | remove_proc_entry("ip_mr_cache", net->proc_net); |
2800 | remove_proc_entry("ip_mr_vif", net->proc_net); | |
f6bb4514 | 2801 | #endif |
f0ad0860 | 2802 | ipmr_rules_exit(net); |
cf958ae3 BT |
2803 | } |
2804 | ||
2805 | static struct pernet_operations ipmr_net_ops = { | |
2806 | .init = ipmr_net_init, | |
2807 | .exit = ipmr_net_exit, | |
2808 | }; | |
e905a9ed | 2809 | |
03d2f897 | 2810 | int __init ip_mr_init(void) |
1da177e4 | 2811 | { |
03d2f897 WC |
2812 | int err; |
2813 | ||
1da177e4 LT |
2814 | mrt_cachep = kmem_cache_create("ip_mrt_cache", |
2815 | sizeof(struct mfc_cache), | |
a8c9486b | 2816 | 0, SLAB_HWCACHE_ALIGN | SLAB_PANIC, |
20c2df83 | 2817 | NULL); |
03d2f897 | 2818 | |
cf958ae3 BT |
2819 | err = register_pernet_subsys(&ipmr_net_ops); |
2820 | if (err) | |
2821 | goto reg_pernet_fail; | |
2822 | ||
03d2f897 WC |
2823 | err = register_netdevice_notifier(&ip_mr_notifier); |
2824 | if (err) | |
2825 | goto reg_notif_fail; | |
403dbb97 TG |
2826 | #ifdef CONFIG_IP_PIMSM_V2 |
2827 | if (inet_add_protocol(&pim_protocol, IPPROTO_PIM) < 0) { | |
058bd4d2 | 2828 | pr_err("%s: can't add PIM protocol\n", __func__); |
403dbb97 TG |
2829 | err = -EAGAIN; |
2830 | goto add_proto_fail; | |
2831 | } | |
2832 | #endif | |
c7ac8679 GR |
2833 | rtnl_register(RTNL_FAMILY_IPMR, RTM_GETROUTE, |
2834 | NULL, ipmr_rtm_dumproute, NULL); | |
ccbb0aa6 NA |
2835 | rtnl_register(RTNL_FAMILY_IPMR, RTM_NEWROUTE, |
2836 | ipmr_rtm_route, NULL, NULL); | |
2837 | rtnl_register(RTNL_FAMILY_IPMR, RTM_DELROUTE, | |
2838 | ipmr_rtm_route, NULL, NULL); | |
03d2f897 | 2839 | return 0; |
f6bb4514 | 2840 | |
403dbb97 TG |
2841 | #ifdef CONFIG_IP_PIMSM_V2 |
2842 | add_proto_fail: | |
2843 | unregister_netdevice_notifier(&ip_mr_notifier); | |
2844 | #endif | |
c3e38896 | 2845 | reg_notif_fail: |
cf958ae3 BT |
2846 | unregister_pernet_subsys(&ipmr_net_ops); |
2847 | reg_pernet_fail: | |
c3e38896 | 2848 | kmem_cache_destroy(mrt_cachep); |
03d2f897 | 2849 | return err; |
1da177e4 | 2850 | } |