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 | ||
1da177e4 LT |
29 | #include <asm/uaccess.h> |
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> |
1da177e4 LT |
64 | #include <net/ipip.h> |
65 | #include <net/checksum.h> | |
dc5fc579 | 66 | #include <net/netlink.h> |
f0ad0860 | 67 | #include <net/fib_rules.h> |
d67b8c61 | 68 | #include <linux/netconf.h> |
1da177e4 LT |
69 | |
70 | #if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2) | |
71 | #define CONFIG_IP_PIMSM 1 | |
72 | #endif | |
73 | ||
0c12295a | 74 | struct mr_table { |
f0ad0860 | 75 | struct list_head list; |
8de53dfb PM |
76 | #ifdef CONFIG_NET_NS |
77 | struct net *net; | |
78 | #endif | |
f0ad0860 | 79 | u32 id; |
4c968709 | 80 | struct sock __rcu *mroute_sk; |
0c12295a PM |
81 | struct timer_list ipmr_expire_timer; |
82 | struct list_head mfc_unres_queue; | |
83 | struct list_head mfc_cache_array[MFC_LINES]; | |
84 | struct vif_device vif_table[MAXVIFS]; | |
85 | int maxvif; | |
86 | atomic_t cache_resolve_queue_len; | |
53d6841d JP |
87 | bool mroute_do_assert; |
88 | bool mroute_do_pim; | |
0c12295a PM |
89 | #if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2) |
90 | int mroute_reg_vif_num; | |
91 | #endif | |
92 | }; | |
93 | ||
f0ad0860 PM |
94 | struct ipmr_rule { |
95 | struct fib_rule common; | |
96 | }; | |
97 | ||
98 | struct ipmr_result { | |
99 | struct mr_table *mrt; | |
100 | }; | |
101 | ||
1da177e4 | 102 | /* Big lock, protecting vif table, mrt cache and mroute socket state. |
a8cb16dd | 103 | * Note that the changes are semaphored via rtnl_lock. |
1da177e4 LT |
104 | */ |
105 | ||
106 | static DEFINE_RWLOCK(mrt_lock); | |
107 | ||
108 | /* | |
109 | * Multicast router control variables | |
110 | */ | |
111 | ||
0c12295a | 112 | #define VIF_EXISTS(_mrt, _idx) ((_mrt)->vif_table[_idx].dev != NULL) |
1da177e4 LT |
113 | |
114 | /* Special spinlock for queue of unresolved entries */ | |
115 | static DEFINE_SPINLOCK(mfc_unres_lock); | |
116 | ||
117 | /* We return to original Alan's scheme. Hash table of resolved | |
a8cb16dd ED |
118 | * entries is changed only in process context and protected |
119 | * with weak lock mrt_lock. Queue of unresolved entries is protected | |
120 | * with strong spinlock mfc_unres_lock. | |
121 | * | |
122 | * In this case data path is free of exclusive locks at all. | |
1da177e4 LT |
123 | */ |
124 | ||
e18b890b | 125 | static struct kmem_cache *mrt_cachep __read_mostly; |
1da177e4 | 126 | |
f0ad0860 | 127 | static struct mr_table *ipmr_new_table(struct net *net, u32 id); |
acbb219d FR |
128 | static void ipmr_free_table(struct mr_table *mrt); |
129 | ||
0c12295a PM |
130 | static int ip_mr_forward(struct net *net, struct mr_table *mrt, |
131 | struct sk_buff *skb, struct mfc_cache *cache, | |
132 | int local); | |
133 | static int ipmr_cache_report(struct mr_table *mrt, | |
4feb88e5 | 134 | struct sk_buff *pkt, vifi_t vifi, int assert); |
cb6a4e46 PM |
135 | static int __ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
136 | struct mfc_cache *c, struct rtmsg *rtm); | |
acbb219d | 137 | static void mroute_clean_tables(struct mr_table *mrt); |
f0ad0860 PM |
138 | static void ipmr_expire_process(unsigned long arg); |
139 | ||
140 | #ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES | |
141 | #define ipmr_for_each_table(mrt, net) \ | |
142 | list_for_each_entry_rcu(mrt, &net->ipv4.mr_tables, list) | |
143 | ||
144 | static struct mr_table *ipmr_get_table(struct net *net, u32 id) | |
145 | { | |
146 | struct mr_table *mrt; | |
147 | ||
148 | ipmr_for_each_table(mrt, net) { | |
149 | if (mrt->id == id) | |
150 | return mrt; | |
151 | } | |
152 | return NULL; | |
153 | } | |
154 | ||
da91981b | 155 | static int ipmr_fib_lookup(struct net *net, struct flowi4 *flp4, |
f0ad0860 PM |
156 | struct mr_table **mrt) |
157 | { | |
158 | struct ipmr_result res; | |
159 | struct fib_lookup_arg arg = { .result = &res, }; | |
160 | int err; | |
161 | ||
da91981b DM |
162 | err = fib_rules_lookup(net->ipv4.mr_rules_ops, |
163 | flowi4_to_flowi(flp4), 0, &arg); | |
f0ad0860 PM |
164 | if (err < 0) |
165 | return err; | |
166 | *mrt = res.mrt; | |
167 | return 0; | |
168 | } | |
169 | ||
170 | static int ipmr_rule_action(struct fib_rule *rule, struct flowi *flp, | |
171 | int flags, struct fib_lookup_arg *arg) | |
172 | { | |
173 | struct ipmr_result *res = arg->result; | |
174 | struct mr_table *mrt; | |
1da177e4 | 175 | |
f0ad0860 PM |
176 | switch (rule->action) { |
177 | case FR_ACT_TO_TBL: | |
178 | break; | |
179 | case FR_ACT_UNREACHABLE: | |
180 | return -ENETUNREACH; | |
181 | case FR_ACT_PROHIBIT: | |
182 | return -EACCES; | |
183 | case FR_ACT_BLACKHOLE: | |
184 | default: | |
185 | return -EINVAL; | |
186 | } | |
187 | ||
188 | mrt = ipmr_get_table(rule->fr_net, rule->table); | |
189 | if (mrt == NULL) | |
190 | return -EAGAIN; | |
191 | res->mrt = mrt; | |
192 | return 0; | |
193 | } | |
194 | ||
195 | static int ipmr_rule_match(struct fib_rule *rule, struct flowi *fl, int flags) | |
196 | { | |
197 | return 1; | |
198 | } | |
199 | ||
200 | static const struct nla_policy ipmr_rule_policy[FRA_MAX + 1] = { | |
201 | FRA_GENERIC_POLICY, | |
202 | }; | |
203 | ||
204 | static int ipmr_rule_configure(struct fib_rule *rule, struct sk_buff *skb, | |
205 | struct fib_rule_hdr *frh, struct nlattr **tb) | |
206 | { | |
207 | return 0; | |
208 | } | |
209 | ||
210 | static int ipmr_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh, | |
211 | struct nlattr **tb) | |
212 | { | |
213 | return 1; | |
214 | } | |
215 | ||
216 | static int ipmr_rule_fill(struct fib_rule *rule, struct sk_buff *skb, | |
217 | struct fib_rule_hdr *frh) | |
218 | { | |
219 | frh->dst_len = 0; | |
220 | frh->src_len = 0; | |
221 | frh->tos = 0; | |
222 | return 0; | |
223 | } | |
224 | ||
04a6f82c | 225 | static const struct fib_rules_ops __net_initconst ipmr_rules_ops_template = { |
25239cee | 226 | .family = RTNL_FAMILY_IPMR, |
f0ad0860 PM |
227 | .rule_size = sizeof(struct ipmr_rule), |
228 | .addr_size = sizeof(u32), | |
229 | .action = ipmr_rule_action, | |
230 | .match = ipmr_rule_match, | |
231 | .configure = ipmr_rule_configure, | |
232 | .compare = ipmr_rule_compare, | |
233 | .default_pref = fib_default_rule_pref, | |
234 | .fill = ipmr_rule_fill, | |
235 | .nlgroup = RTNLGRP_IPV4_RULE, | |
236 | .policy = ipmr_rule_policy, | |
237 | .owner = THIS_MODULE, | |
238 | }; | |
239 | ||
240 | static int __net_init ipmr_rules_init(struct net *net) | |
241 | { | |
242 | struct fib_rules_ops *ops; | |
243 | struct mr_table *mrt; | |
244 | int err; | |
245 | ||
246 | ops = fib_rules_register(&ipmr_rules_ops_template, net); | |
247 | if (IS_ERR(ops)) | |
248 | return PTR_ERR(ops); | |
249 | ||
250 | INIT_LIST_HEAD(&net->ipv4.mr_tables); | |
251 | ||
252 | mrt = ipmr_new_table(net, RT_TABLE_DEFAULT); | |
253 | if (mrt == NULL) { | |
254 | err = -ENOMEM; | |
255 | goto err1; | |
256 | } | |
257 | ||
258 | err = fib_default_rule_add(ops, 0x7fff, RT_TABLE_DEFAULT, 0); | |
259 | if (err < 0) | |
260 | goto err2; | |
261 | ||
262 | net->ipv4.mr_rules_ops = ops; | |
263 | return 0; | |
264 | ||
265 | err2: | |
266 | kfree(mrt); | |
267 | err1: | |
268 | fib_rules_unregister(ops); | |
269 | return err; | |
270 | } | |
271 | ||
272 | static void __net_exit ipmr_rules_exit(struct net *net) | |
273 | { | |
274 | struct mr_table *mrt, *next; | |
275 | ||
035320d5 ED |
276 | list_for_each_entry_safe(mrt, next, &net->ipv4.mr_tables, list) { |
277 | list_del(&mrt->list); | |
acbb219d | 278 | ipmr_free_table(mrt); |
035320d5 | 279 | } |
f0ad0860 PM |
280 | fib_rules_unregister(net->ipv4.mr_rules_ops); |
281 | } | |
282 | #else | |
283 | #define ipmr_for_each_table(mrt, net) \ | |
284 | for (mrt = net->ipv4.mrt; mrt; mrt = NULL) | |
285 | ||
286 | static struct mr_table *ipmr_get_table(struct net *net, u32 id) | |
287 | { | |
288 | return net->ipv4.mrt; | |
289 | } | |
290 | ||
da91981b | 291 | static int ipmr_fib_lookup(struct net *net, struct flowi4 *flp4, |
f0ad0860 PM |
292 | struct mr_table **mrt) |
293 | { | |
294 | *mrt = net->ipv4.mrt; | |
295 | return 0; | |
296 | } | |
297 | ||
298 | static int __net_init ipmr_rules_init(struct net *net) | |
299 | { | |
300 | net->ipv4.mrt = ipmr_new_table(net, RT_TABLE_DEFAULT); | |
301 | return net->ipv4.mrt ? 0 : -ENOMEM; | |
302 | } | |
303 | ||
304 | static void __net_exit ipmr_rules_exit(struct net *net) | |
305 | { | |
acbb219d | 306 | ipmr_free_table(net->ipv4.mrt); |
f0ad0860 PM |
307 | } |
308 | #endif | |
309 | ||
310 | static struct mr_table *ipmr_new_table(struct net *net, u32 id) | |
311 | { | |
312 | struct mr_table *mrt; | |
313 | unsigned int i; | |
1da177e4 | 314 | |
f0ad0860 PM |
315 | mrt = ipmr_get_table(net, id); |
316 | if (mrt != NULL) | |
317 | return mrt; | |
318 | ||
319 | mrt = kzalloc(sizeof(*mrt), GFP_KERNEL); | |
320 | if (mrt == NULL) | |
321 | return NULL; | |
8de53dfb | 322 | write_pnet(&mrt->net, net); |
f0ad0860 PM |
323 | mrt->id = id; |
324 | ||
325 | /* Forwarding cache */ | |
326 | for (i = 0; i < MFC_LINES; i++) | |
327 | INIT_LIST_HEAD(&mrt->mfc_cache_array[i]); | |
328 | ||
329 | INIT_LIST_HEAD(&mrt->mfc_unres_queue); | |
330 | ||
331 | setup_timer(&mrt->ipmr_expire_timer, ipmr_expire_process, | |
332 | (unsigned long)mrt); | |
333 | ||
334 | #ifdef CONFIG_IP_PIMSM | |
335 | mrt->mroute_reg_vif_num = -1; | |
336 | #endif | |
337 | #ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES | |
338 | list_add_tail_rcu(&mrt->list, &net->ipv4.mr_tables); | |
339 | #endif | |
340 | return mrt; | |
341 | } | |
1da177e4 | 342 | |
acbb219d FR |
343 | static void ipmr_free_table(struct mr_table *mrt) |
344 | { | |
345 | del_timer_sync(&mrt->ipmr_expire_timer); | |
346 | mroute_clean_tables(mrt); | |
347 | kfree(mrt); | |
348 | } | |
349 | ||
1da177e4 LT |
350 | /* Service routines creating virtual interfaces: DVMRP tunnels and PIMREG */ |
351 | ||
d607032d WC |
352 | static void ipmr_del_tunnel(struct net_device *dev, struct vifctl *v) |
353 | { | |
4feb88e5 BT |
354 | struct net *net = dev_net(dev); |
355 | ||
d607032d WC |
356 | dev_close(dev); |
357 | ||
4feb88e5 | 358 | dev = __dev_get_by_name(net, "tunl0"); |
d607032d | 359 | if (dev) { |
5bc3eb7e | 360 | const struct net_device_ops *ops = dev->netdev_ops; |
d607032d | 361 | struct ifreq ifr; |
d607032d WC |
362 | struct ip_tunnel_parm p; |
363 | ||
364 | memset(&p, 0, sizeof(p)); | |
365 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
366 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
367 | p.iph.version = 4; | |
368 | p.iph.ihl = 5; | |
369 | p.iph.protocol = IPPROTO_IPIP; | |
370 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
371 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; | |
372 | ||
5bc3eb7e SH |
373 | if (ops->ndo_do_ioctl) { |
374 | mm_segment_t oldfs = get_fs(); | |
375 | ||
376 | set_fs(KERNEL_DS); | |
377 | ops->ndo_do_ioctl(dev, &ifr, SIOCDELTUNNEL); | |
378 | set_fs(oldfs); | |
379 | } | |
d607032d WC |
380 | } |
381 | } | |
382 | ||
1da177e4 | 383 | static |
4feb88e5 | 384 | struct net_device *ipmr_new_tunnel(struct net *net, struct vifctl *v) |
1da177e4 LT |
385 | { |
386 | struct net_device *dev; | |
387 | ||
4feb88e5 | 388 | dev = __dev_get_by_name(net, "tunl0"); |
1da177e4 LT |
389 | |
390 | if (dev) { | |
5bc3eb7e | 391 | const struct net_device_ops *ops = dev->netdev_ops; |
1da177e4 LT |
392 | int err; |
393 | struct ifreq ifr; | |
1da177e4 LT |
394 | struct ip_tunnel_parm p; |
395 | struct in_device *in_dev; | |
396 | ||
397 | memset(&p, 0, sizeof(p)); | |
398 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
399 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
400 | p.iph.version = 4; | |
401 | p.iph.ihl = 5; | |
402 | p.iph.protocol = IPPROTO_IPIP; | |
403 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
ba93ef74 | 404 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; |
1da177e4 | 405 | |
5bc3eb7e SH |
406 | if (ops->ndo_do_ioctl) { |
407 | mm_segment_t oldfs = get_fs(); | |
408 | ||
409 | set_fs(KERNEL_DS); | |
410 | err = ops->ndo_do_ioctl(dev, &ifr, SIOCADDTUNNEL); | |
411 | set_fs(oldfs); | |
a8cb16dd | 412 | } else { |
5bc3eb7e | 413 | err = -EOPNOTSUPP; |
a8cb16dd | 414 | } |
1da177e4 LT |
415 | dev = NULL; |
416 | ||
4feb88e5 BT |
417 | if (err == 0 && |
418 | (dev = __dev_get_by_name(net, p.name)) != NULL) { | |
1da177e4 LT |
419 | dev->flags |= IFF_MULTICAST; |
420 | ||
e5ed6399 | 421 | in_dev = __in_dev_get_rtnl(dev); |
71e27da9 | 422 | if (in_dev == NULL) |
1da177e4 | 423 | goto failure; |
71e27da9 HX |
424 | |
425 | ipv4_devconf_setall(in_dev); | |
426 | IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0; | |
1da177e4 LT |
427 | |
428 | if (dev_open(dev)) | |
429 | goto failure; | |
7dc00c82 | 430 | dev_hold(dev); |
1da177e4 LT |
431 | } |
432 | } | |
433 | return dev; | |
434 | ||
435 | failure: | |
436 | /* allow the register to be completed before unregistering. */ | |
437 | rtnl_unlock(); | |
438 | rtnl_lock(); | |
439 | ||
440 | unregister_netdevice(dev); | |
441 | return NULL; | |
442 | } | |
443 | ||
444 | #ifdef CONFIG_IP_PIMSM | |
445 | ||
6fef4c0c | 446 | static netdev_tx_t reg_vif_xmit(struct sk_buff *skb, struct net_device *dev) |
1da177e4 | 447 | { |
4feb88e5 | 448 | struct net *net = dev_net(dev); |
f0ad0860 | 449 | struct mr_table *mrt; |
da91981b DM |
450 | struct flowi4 fl4 = { |
451 | .flowi4_oif = dev->ifindex, | |
452 | .flowi4_iif = skb->skb_iif, | |
453 | .flowi4_mark = skb->mark, | |
f0ad0860 PM |
454 | }; |
455 | int err; | |
456 | ||
da91981b | 457 | err = ipmr_fib_lookup(net, &fl4, &mrt); |
e40dbc51 BG |
458 | if (err < 0) { |
459 | kfree_skb(skb); | |
f0ad0860 | 460 | return err; |
e40dbc51 | 461 | } |
4feb88e5 | 462 | |
1da177e4 | 463 | read_lock(&mrt_lock); |
cf3677ae PE |
464 | dev->stats.tx_bytes += skb->len; |
465 | dev->stats.tx_packets++; | |
0c12295a | 466 | ipmr_cache_report(mrt, skb, mrt->mroute_reg_vif_num, IGMPMSG_WHOLEPKT); |
1da177e4 LT |
467 | read_unlock(&mrt_lock); |
468 | kfree_skb(skb); | |
6ed10654 | 469 | return NETDEV_TX_OK; |
1da177e4 LT |
470 | } |
471 | ||
007c3838 SH |
472 | static const struct net_device_ops reg_vif_netdev_ops = { |
473 | .ndo_start_xmit = reg_vif_xmit, | |
474 | }; | |
475 | ||
1da177e4 LT |
476 | static void reg_vif_setup(struct net_device *dev) |
477 | { | |
478 | dev->type = ARPHRD_PIMREG; | |
46f25dff | 479 | dev->mtu = ETH_DATA_LEN - sizeof(struct iphdr) - 8; |
1da177e4 | 480 | dev->flags = IFF_NOARP; |
007c3838 | 481 | dev->netdev_ops = ®_vif_netdev_ops, |
1da177e4 | 482 | dev->destructor = free_netdev; |
403dbb97 | 483 | dev->features |= NETIF_F_NETNS_LOCAL; |
1da177e4 LT |
484 | } |
485 | ||
f0ad0860 | 486 | static struct net_device *ipmr_reg_vif(struct net *net, struct mr_table *mrt) |
1da177e4 LT |
487 | { |
488 | struct net_device *dev; | |
489 | struct in_device *in_dev; | |
f0ad0860 | 490 | char name[IFNAMSIZ]; |
1da177e4 | 491 | |
f0ad0860 PM |
492 | if (mrt->id == RT_TABLE_DEFAULT) |
493 | sprintf(name, "pimreg"); | |
494 | else | |
495 | sprintf(name, "pimreg%u", mrt->id); | |
1da177e4 | 496 | |
f0ad0860 | 497 | dev = alloc_netdev(0, name, reg_vif_setup); |
1da177e4 LT |
498 | |
499 | if (dev == NULL) | |
500 | return NULL; | |
501 | ||
403dbb97 TG |
502 | dev_net_set(dev, net); |
503 | ||
1da177e4 LT |
504 | if (register_netdevice(dev)) { |
505 | free_netdev(dev); | |
506 | return NULL; | |
507 | } | |
508 | dev->iflink = 0; | |
509 | ||
71e27da9 | 510 | rcu_read_lock(); |
a8cb16dd ED |
511 | in_dev = __in_dev_get_rcu(dev); |
512 | if (!in_dev) { | |
71e27da9 | 513 | rcu_read_unlock(); |
1da177e4 | 514 | goto failure; |
71e27da9 | 515 | } |
1da177e4 | 516 | |
71e27da9 HX |
517 | ipv4_devconf_setall(in_dev); |
518 | IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0; | |
519 | rcu_read_unlock(); | |
1da177e4 LT |
520 | |
521 | if (dev_open(dev)) | |
522 | goto failure; | |
523 | ||
7dc00c82 WC |
524 | dev_hold(dev); |
525 | ||
1da177e4 LT |
526 | return dev; |
527 | ||
528 | failure: | |
529 | /* allow the register to be completed before unregistering. */ | |
530 | rtnl_unlock(); | |
531 | rtnl_lock(); | |
532 | ||
533 | unregister_netdevice(dev); | |
534 | return NULL; | |
535 | } | |
536 | #endif | |
537 | ||
2c53040f BH |
538 | /** |
539 | * vif_delete - Delete a VIF entry | |
7dc00c82 | 540 | * @notify: Set to 1, if the caller is a notifier_call |
1da177e4 | 541 | */ |
e905a9ed | 542 | |
0c12295a | 543 | static int vif_delete(struct mr_table *mrt, int vifi, int notify, |
d17fa6fa | 544 | struct list_head *head) |
1da177e4 LT |
545 | { |
546 | struct vif_device *v; | |
547 | struct net_device *dev; | |
548 | struct in_device *in_dev; | |
549 | ||
0c12295a | 550 | if (vifi < 0 || vifi >= mrt->maxvif) |
1da177e4 LT |
551 | return -EADDRNOTAVAIL; |
552 | ||
0c12295a | 553 | v = &mrt->vif_table[vifi]; |
1da177e4 LT |
554 | |
555 | write_lock_bh(&mrt_lock); | |
556 | dev = v->dev; | |
557 | v->dev = NULL; | |
558 | ||
559 | if (!dev) { | |
560 | write_unlock_bh(&mrt_lock); | |
561 | return -EADDRNOTAVAIL; | |
562 | } | |
563 | ||
564 | #ifdef CONFIG_IP_PIMSM | |
0c12295a PM |
565 | if (vifi == mrt->mroute_reg_vif_num) |
566 | mrt->mroute_reg_vif_num = -1; | |
1da177e4 LT |
567 | #endif |
568 | ||
a8cb16dd | 569 | if (vifi + 1 == mrt->maxvif) { |
1da177e4 | 570 | int tmp; |
a8cb16dd ED |
571 | |
572 | for (tmp = vifi - 1; tmp >= 0; tmp--) { | |
0c12295a | 573 | if (VIF_EXISTS(mrt, tmp)) |
1da177e4 LT |
574 | break; |
575 | } | |
0c12295a | 576 | mrt->maxvif = tmp+1; |
1da177e4 LT |
577 | } |
578 | ||
579 | write_unlock_bh(&mrt_lock); | |
580 | ||
581 | dev_set_allmulti(dev, -1); | |
582 | ||
a8cb16dd ED |
583 | in_dev = __in_dev_get_rtnl(dev); |
584 | if (in_dev) { | |
42f811b8 | 585 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)--; |
d67b8c61 ND |
586 | inet_netconf_notify_devconf(dev_net(dev), |
587 | NETCONFA_MC_FORWARDING, | |
588 | dev->ifindex, &in_dev->cnf); | |
1da177e4 LT |
589 | ip_rt_multicast_event(in_dev); |
590 | } | |
591 | ||
a8cb16dd | 592 | if (v->flags & (VIFF_TUNNEL | VIFF_REGISTER) && !notify) |
d17fa6fa | 593 | unregister_netdevice_queue(dev, head); |
1da177e4 LT |
594 | |
595 | dev_put(dev); | |
596 | return 0; | |
597 | } | |
598 | ||
a8c9486b | 599 | static void ipmr_cache_free_rcu(struct rcu_head *head) |
5c0a66f5 | 600 | { |
a8c9486b ED |
601 | struct mfc_cache *c = container_of(head, struct mfc_cache, rcu); |
602 | ||
5c0a66f5 BT |
603 | kmem_cache_free(mrt_cachep, c); |
604 | } | |
605 | ||
a8c9486b ED |
606 | static inline void ipmr_cache_free(struct mfc_cache *c) |
607 | { | |
608 | call_rcu(&c->rcu, ipmr_cache_free_rcu); | |
609 | } | |
610 | ||
1da177e4 | 611 | /* Destroy an unresolved cache entry, killing queued skbs |
a8cb16dd | 612 | * and reporting error to netlink readers. |
1da177e4 LT |
613 | */ |
614 | ||
0c12295a | 615 | static void ipmr_destroy_unres(struct mr_table *mrt, struct mfc_cache *c) |
1da177e4 | 616 | { |
8de53dfb | 617 | struct net *net = read_pnet(&mrt->net); |
1da177e4 | 618 | struct sk_buff *skb; |
9ef1d4c7 | 619 | struct nlmsgerr *e; |
1da177e4 | 620 | |
0c12295a | 621 | atomic_dec(&mrt->cache_resolve_queue_len); |
1da177e4 | 622 | |
c354e124 | 623 | while ((skb = skb_dequeue(&c->mfc_un.unres.unresolved))) { |
eddc9ec5 | 624 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
625 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
626 | nlh->nlmsg_type = NLMSG_ERROR; | |
627 | nlh->nlmsg_len = NLMSG_LENGTH(sizeof(struct nlmsgerr)); | |
628 | skb_trim(skb, nlh->nlmsg_len); | |
9ef1d4c7 PM |
629 | e = NLMSG_DATA(nlh); |
630 | e->error = -ETIMEDOUT; | |
631 | memset(&e->msg, 0, sizeof(e->msg)); | |
2942e900 | 632 | |
15e47304 | 633 | rtnl_unicast(skb, net, NETLINK_CB(skb).portid); |
a8cb16dd | 634 | } else { |
1da177e4 | 635 | kfree_skb(skb); |
a8cb16dd | 636 | } |
1da177e4 LT |
637 | } |
638 | ||
5c0a66f5 | 639 | ipmr_cache_free(c); |
1da177e4 LT |
640 | } |
641 | ||
642 | ||
e258beb2 | 643 | /* Timer process for the unresolved queue. */ |
1da177e4 | 644 | |
e258beb2 | 645 | static void ipmr_expire_process(unsigned long arg) |
1da177e4 | 646 | { |
0c12295a | 647 | struct mr_table *mrt = (struct mr_table *)arg; |
1da177e4 LT |
648 | unsigned long now; |
649 | unsigned long expires; | |
862465f2 | 650 | struct mfc_cache *c, *next; |
1da177e4 LT |
651 | |
652 | if (!spin_trylock(&mfc_unres_lock)) { | |
0c12295a | 653 | mod_timer(&mrt->ipmr_expire_timer, jiffies+HZ/10); |
1da177e4 LT |
654 | return; |
655 | } | |
656 | ||
0c12295a | 657 | if (list_empty(&mrt->mfc_unres_queue)) |
1da177e4 LT |
658 | goto out; |
659 | ||
660 | now = jiffies; | |
661 | expires = 10*HZ; | |
1da177e4 | 662 | |
0c12295a | 663 | list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) { |
1da177e4 LT |
664 | if (time_after(c->mfc_un.unres.expires, now)) { |
665 | unsigned long interval = c->mfc_un.unres.expires - now; | |
666 | if (interval < expires) | |
667 | expires = interval; | |
1da177e4 LT |
668 | continue; |
669 | } | |
670 | ||
862465f2 | 671 | list_del(&c->list); |
0c12295a | 672 | ipmr_destroy_unres(mrt, c); |
1da177e4 LT |
673 | } |
674 | ||
0c12295a PM |
675 | if (!list_empty(&mrt->mfc_unres_queue)) |
676 | mod_timer(&mrt->ipmr_expire_timer, jiffies + expires); | |
1da177e4 LT |
677 | |
678 | out: | |
679 | spin_unlock(&mfc_unres_lock); | |
680 | } | |
681 | ||
682 | /* Fill oifs list. It is called under write locked mrt_lock. */ | |
683 | ||
0c12295a | 684 | static void ipmr_update_thresholds(struct mr_table *mrt, struct mfc_cache *cache, |
d658f8a0 | 685 | unsigned char *ttls) |
1da177e4 LT |
686 | { |
687 | int vifi; | |
688 | ||
689 | cache->mfc_un.res.minvif = MAXVIFS; | |
690 | cache->mfc_un.res.maxvif = 0; | |
691 | memset(cache->mfc_un.res.ttls, 255, MAXVIFS); | |
692 | ||
0c12295a PM |
693 | for (vifi = 0; vifi < mrt->maxvif; vifi++) { |
694 | if (VIF_EXISTS(mrt, vifi) && | |
cf958ae3 | 695 | ttls[vifi] && ttls[vifi] < 255) { |
1da177e4 LT |
696 | cache->mfc_un.res.ttls[vifi] = ttls[vifi]; |
697 | if (cache->mfc_un.res.minvif > vifi) | |
698 | cache->mfc_un.res.minvif = vifi; | |
699 | if (cache->mfc_un.res.maxvif <= vifi) | |
700 | cache->mfc_un.res.maxvif = vifi + 1; | |
701 | } | |
702 | } | |
703 | } | |
704 | ||
0c12295a PM |
705 | static int vif_add(struct net *net, struct mr_table *mrt, |
706 | struct vifctl *vifc, int mrtsock) | |
1da177e4 LT |
707 | { |
708 | int vifi = vifc->vifc_vifi; | |
0c12295a | 709 | struct vif_device *v = &mrt->vif_table[vifi]; |
1da177e4 LT |
710 | struct net_device *dev; |
711 | struct in_device *in_dev; | |
d607032d | 712 | int err; |
1da177e4 LT |
713 | |
714 | /* Is vif busy ? */ | |
0c12295a | 715 | if (VIF_EXISTS(mrt, vifi)) |
1da177e4 LT |
716 | return -EADDRINUSE; |
717 | ||
718 | switch (vifc->vifc_flags) { | |
719 | #ifdef CONFIG_IP_PIMSM | |
720 | case VIFF_REGISTER: | |
721 | /* | |
722 | * Special Purpose VIF in PIM | |
723 | * All the packets will be sent to the daemon | |
724 | */ | |
0c12295a | 725 | if (mrt->mroute_reg_vif_num >= 0) |
1da177e4 | 726 | return -EADDRINUSE; |
f0ad0860 | 727 | dev = ipmr_reg_vif(net, mrt); |
1da177e4 LT |
728 | if (!dev) |
729 | return -ENOBUFS; | |
d607032d WC |
730 | err = dev_set_allmulti(dev, 1); |
731 | if (err) { | |
732 | unregister_netdevice(dev); | |
7dc00c82 | 733 | dev_put(dev); |
d607032d WC |
734 | return err; |
735 | } | |
1da177e4 LT |
736 | break; |
737 | #endif | |
e905a9ed | 738 | case VIFF_TUNNEL: |
4feb88e5 | 739 | dev = ipmr_new_tunnel(net, vifc); |
1da177e4 LT |
740 | if (!dev) |
741 | return -ENOBUFS; | |
d607032d WC |
742 | err = dev_set_allmulti(dev, 1); |
743 | if (err) { | |
744 | ipmr_del_tunnel(dev, vifc); | |
7dc00c82 | 745 | dev_put(dev); |
d607032d WC |
746 | return err; |
747 | } | |
1da177e4 | 748 | break; |
ee5e81f0 I |
749 | |
750 | case VIFF_USE_IFINDEX: | |
1da177e4 | 751 | case 0: |
ee5e81f0 I |
752 | if (vifc->vifc_flags == VIFF_USE_IFINDEX) { |
753 | dev = dev_get_by_index(net, vifc->vifc_lcl_ifindex); | |
95ae6b22 | 754 | if (dev && __in_dev_get_rtnl(dev) == NULL) { |
ee5e81f0 I |
755 | dev_put(dev); |
756 | return -EADDRNOTAVAIL; | |
757 | } | |
a8cb16dd | 758 | } else { |
ee5e81f0 | 759 | dev = ip_dev_find(net, vifc->vifc_lcl_addr.s_addr); |
a8cb16dd | 760 | } |
1da177e4 LT |
761 | if (!dev) |
762 | return -EADDRNOTAVAIL; | |
d607032d | 763 | err = dev_set_allmulti(dev, 1); |
7dc00c82 WC |
764 | if (err) { |
765 | dev_put(dev); | |
d607032d | 766 | return err; |
7dc00c82 | 767 | } |
1da177e4 LT |
768 | break; |
769 | default: | |
770 | return -EINVAL; | |
771 | } | |
772 | ||
a8cb16dd ED |
773 | in_dev = __in_dev_get_rtnl(dev); |
774 | if (!in_dev) { | |
d0490cfd | 775 | dev_put(dev); |
1da177e4 | 776 | return -EADDRNOTAVAIL; |
d0490cfd | 777 | } |
42f811b8 | 778 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)++; |
d67b8c61 ND |
779 | inet_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING, dev->ifindex, |
780 | &in_dev->cnf); | |
1da177e4 LT |
781 | ip_rt_multicast_event(in_dev); |
782 | ||
a8cb16dd ED |
783 | /* Fill in the VIF structures */ |
784 | ||
c354e124 JK |
785 | v->rate_limit = vifc->vifc_rate_limit; |
786 | v->local = vifc->vifc_lcl_addr.s_addr; | |
787 | v->remote = vifc->vifc_rmt_addr.s_addr; | |
788 | v->flags = vifc->vifc_flags; | |
1da177e4 LT |
789 | if (!mrtsock) |
790 | v->flags |= VIFF_STATIC; | |
c354e124 | 791 | v->threshold = vifc->vifc_threshold; |
1da177e4 LT |
792 | v->bytes_in = 0; |
793 | v->bytes_out = 0; | |
794 | v->pkt_in = 0; | |
795 | v->pkt_out = 0; | |
796 | v->link = dev->ifindex; | |
a8cb16dd | 797 | if (v->flags & (VIFF_TUNNEL | VIFF_REGISTER)) |
1da177e4 LT |
798 | v->link = dev->iflink; |
799 | ||
800 | /* And finish update writing critical data */ | |
801 | write_lock_bh(&mrt_lock); | |
c354e124 | 802 | v->dev = dev; |
1da177e4 | 803 | #ifdef CONFIG_IP_PIMSM |
a8cb16dd | 804 | if (v->flags & VIFF_REGISTER) |
0c12295a | 805 | mrt->mroute_reg_vif_num = vifi; |
1da177e4 | 806 | #endif |
0c12295a PM |
807 | if (vifi+1 > mrt->maxvif) |
808 | mrt->maxvif = vifi+1; | |
1da177e4 LT |
809 | write_unlock_bh(&mrt_lock); |
810 | return 0; | |
811 | } | |
812 | ||
a8c9486b | 813 | /* called with rcu_read_lock() */ |
0c12295a | 814 | static struct mfc_cache *ipmr_cache_find(struct mr_table *mrt, |
4feb88e5 BT |
815 | __be32 origin, |
816 | __be32 mcastgrp) | |
1da177e4 | 817 | { |
c354e124 | 818 | int line = MFC_HASH(mcastgrp, origin); |
1da177e4 LT |
819 | struct mfc_cache *c; |
820 | ||
a8c9486b | 821 | list_for_each_entry_rcu(c, &mrt->mfc_cache_array[line], list) { |
862465f2 PM |
822 | if (c->mfc_origin == origin && c->mfc_mcastgrp == mcastgrp) |
823 | return c; | |
1da177e4 | 824 | } |
862465f2 | 825 | return NULL; |
1da177e4 LT |
826 | } |
827 | ||
828 | /* | |
829 | * Allocate a multicast cache entry | |
830 | */ | |
d658f8a0 | 831 | static struct mfc_cache *ipmr_cache_alloc(void) |
1da177e4 | 832 | { |
c354e124 | 833 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL); |
a8c9486b ED |
834 | |
835 | if (c) | |
836 | c->mfc_un.res.minvif = MAXVIFS; | |
1da177e4 LT |
837 | return c; |
838 | } | |
839 | ||
d658f8a0 | 840 | static struct mfc_cache *ipmr_cache_alloc_unres(void) |
1da177e4 | 841 | { |
c354e124 | 842 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC); |
a8c9486b ED |
843 | |
844 | if (c) { | |
845 | skb_queue_head_init(&c->mfc_un.unres.unresolved); | |
846 | c->mfc_un.unres.expires = jiffies + 10*HZ; | |
847 | } | |
1da177e4 LT |
848 | return c; |
849 | } | |
850 | ||
851 | /* | |
852 | * A cache entry has gone into a resolved state from queued | |
853 | */ | |
e905a9ed | 854 | |
0c12295a PM |
855 | static void ipmr_cache_resolve(struct net *net, struct mr_table *mrt, |
856 | struct mfc_cache *uc, struct mfc_cache *c) | |
1da177e4 LT |
857 | { |
858 | struct sk_buff *skb; | |
9ef1d4c7 | 859 | struct nlmsgerr *e; |
1da177e4 | 860 | |
a8cb16dd | 861 | /* Play the pending entries through our router */ |
1da177e4 | 862 | |
c354e124 | 863 | while ((skb = __skb_dequeue(&uc->mfc_un.unres.unresolved))) { |
eddc9ec5 | 864 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
865 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
866 | ||
cb6a4e46 | 867 | if (__ipmr_fill_mroute(mrt, skb, c, NLMSG_DATA(nlh)) > 0) { |
a8cb16dd ED |
868 | nlh->nlmsg_len = skb_tail_pointer(skb) - |
869 | (u8 *)nlh; | |
1da177e4 LT |
870 | } else { |
871 | nlh->nlmsg_type = NLMSG_ERROR; | |
872 | nlh->nlmsg_len = NLMSG_LENGTH(sizeof(struct nlmsgerr)); | |
873 | skb_trim(skb, nlh->nlmsg_len); | |
9ef1d4c7 PM |
874 | e = NLMSG_DATA(nlh); |
875 | e->error = -EMSGSIZE; | |
876 | memset(&e->msg, 0, sizeof(e->msg)); | |
1da177e4 | 877 | } |
2942e900 | 878 | |
15e47304 | 879 | rtnl_unicast(skb, net, NETLINK_CB(skb).portid); |
a8cb16dd | 880 | } else { |
0c12295a | 881 | ip_mr_forward(net, mrt, skb, c, 0); |
a8cb16dd | 882 | } |
1da177e4 LT |
883 | } |
884 | } | |
885 | ||
886 | /* | |
887 | * Bounce a cache query up to mrouted. We could use netlink for this but mrouted | |
888 | * expects the following bizarre scheme. | |
889 | * | |
890 | * Called under mrt_lock. | |
891 | */ | |
e905a9ed | 892 | |
0c12295a | 893 | static int ipmr_cache_report(struct mr_table *mrt, |
4feb88e5 | 894 | struct sk_buff *pkt, vifi_t vifi, int assert) |
1da177e4 LT |
895 | { |
896 | struct sk_buff *skb; | |
c9bdd4b5 | 897 | const int ihl = ip_hdrlen(pkt); |
1da177e4 LT |
898 | struct igmphdr *igmp; |
899 | struct igmpmsg *msg; | |
4c968709 | 900 | struct sock *mroute_sk; |
1da177e4 LT |
901 | int ret; |
902 | ||
903 | #ifdef CONFIG_IP_PIMSM | |
904 | if (assert == IGMPMSG_WHOLEPKT) | |
905 | skb = skb_realloc_headroom(pkt, sizeof(struct iphdr)); | |
906 | else | |
907 | #endif | |
908 | skb = alloc_skb(128, GFP_ATOMIC); | |
909 | ||
132adf54 | 910 | if (!skb) |
1da177e4 LT |
911 | return -ENOBUFS; |
912 | ||
913 | #ifdef CONFIG_IP_PIMSM | |
914 | if (assert == IGMPMSG_WHOLEPKT) { | |
915 | /* Ugly, but we have no choice with this interface. | |
a8cb16dd ED |
916 | * Duplicate old header, fix ihl, length etc. |
917 | * And all this only to mangle msg->im_msgtype and | |
918 | * to set msg->im_mbz to "mbz" :-) | |
1da177e4 | 919 | */ |
878c8145 ACM |
920 | skb_push(skb, sizeof(struct iphdr)); |
921 | skb_reset_network_header(skb); | |
badff6d0 | 922 | skb_reset_transport_header(skb); |
0272ffc4 | 923 | msg = (struct igmpmsg *)skb_network_header(skb); |
d56f90a7 | 924 | memcpy(msg, skb_network_header(pkt), sizeof(struct iphdr)); |
1da177e4 LT |
925 | msg->im_msgtype = IGMPMSG_WHOLEPKT; |
926 | msg->im_mbz = 0; | |
0c12295a | 927 | msg->im_vif = mrt->mroute_reg_vif_num; |
eddc9ec5 ACM |
928 | ip_hdr(skb)->ihl = sizeof(struct iphdr) >> 2; |
929 | ip_hdr(skb)->tot_len = htons(ntohs(ip_hdr(pkt)->tot_len) + | |
930 | sizeof(struct iphdr)); | |
e905a9ed | 931 | } else |
1da177e4 | 932 | #endif |
e905a9ed YH |
933 | { |
934 | ||
a8cb16dd | 935 | /* Copy the IP header */ |
1da177e4 | 936 | |
27a884dc | 937 | skb->network_header = skb->tail; |
ddc7b8e3 | 938 | skb_put(skb, ihl); |
27d7ff46 | 939 | skb_copy_to_linear_data(skb, pkt->data, ihl); |
a8cb16dd | 940 | ip_hdr(skb)->protocol = 0; /* Flag to the kernel this is a route add */ |
eddc9ec5 | 941 | msg = (struct igmpmsg *)skb_network_header(skb); |
1da177e4 | 942 | msg->im_vif = vifi; |
adf30907 | 943 | skb_dst_set(skb, dst_clone(skb_dst(pkt))); |
1da177e4 | 944 | |
a8cb16dd | 945 | /* Add our header */ |
1da177e4 | 946 | |
a8cb16dd | 947 | igmp = (struct igmphdr *)skb_put(skb, sizeof(struct igmphdr)); |
1da177e4 LT |
948 | igmp->type = |
949 | msg->im_msgtype = assert; | |
a8cb16dd ED |
950 | igmp->code = 0; |
951 | ip_hdr(skb)->tot_len = htons(skb->len); /* Fix the length */ | |
b0e380b1 | 952 | skb->transport_header = skb->network_header; |
e905a9ed | 953 | } |
1da177e4 | 954 | |
4c968709 ED |
955 | rcu_read_lock(); |
956 | mroute_sk = rcu_dereference(mrt->mroute_sk); | |
957 | if (mroute_sk == NULL) { | |
958 | rcu_read_unlock(); | |
1da177e4 LT |
959 | kfree_skb(skb); |
960 | return -EINVAL; | |
961 | } | |
962 | ||
a8cb16dd ED |
963 | /* Deliver to mrouted */ |
964 | ||
4c968709 ED |
965 | ret = sock_queue_rcv_skb(mroute_sk, skb); |
966 | rcu_read_unlock(); | |
70a269e6 | 967 | if (ret < 0) { |
e87cc472 | 968 | net_warn_ratelimited("mroute: pending queue full, dropping entries\n"); |
1da177e4 LT |
969 | kfree_skb(skb); |
970 | } | |
971 | ||
972 | return ret; | |
973 | } | |
974 | ||
975 | /* | |
976 | * Queue a packet for resolution. It gets locked cache entry! | |
977 | */ | |
e905a9ed | 978 | |
1da177e4 | 979 | static int |
0c12295a | 980 | ipmr_cache_unresolved(struct mr_table *mrt, vifi_t vifi, struct sk_buff *skb) |
1da177e4 | 981 | { |
862465f2 | 982 | bool found = false; |
1da177e4 LT |
983 | int err; |
984 | struct mfc_cache *c; | |
eddc9ec5 | 985 | const struct iphdr *iph = ip_hdr(skb); |
1da177e4 LT |
986 | |
987 | spin_lock_bh(&mfc_unres_lock); | |
0c12295a | 988 | list_for_each_entry(c, &mrt->mfc_unres_queue, list) { |
e258beb2 | 989 | if (c->mfc_mcastgrp == iph->daddr && |
862465f2 PM |
990 | c->mfc_origin == iph->saddr) { |
991 | found = true; | |
1da177e4 | 992 | break; |
862465f2 | 993 | } |
1da177e4 LT |
994 | } |
995 | ||
862465f2 | 996 | if (!found) { |
a8cb16dd | 997 | /* Create a new entry if allowable */ |
1da177e4 | 998 | |
0c12295a | 999 | if (atomic_read(&mrt->cache_resolve_queue_len) >= 10 || |
d658f8a0 | 1000 | (c = ipmr_cache_alloc_unres()) == NULL) { |
1da177e4 LT |
1001 | spin_unlock_bh(&mfc_unres_lock); |
1002 | ||
1003 | kfree_skb(skb); | |
1004 | return -ENOBUFS; | |
1005 | } | |
1006 | ||
a8cb16dd ED |
1007 | /* Fill in the new cache entry */ |
1008 | ||
eddc9ec5 ACM |
1009 | c->mfc_parent = -1; |
1010 | c->mfc_origin = iph->saddr; | |
1011 | c->mfc_mcastgrp = iph->daddr; | |
1da177e4 | 1012 | |
a8cb16dd ED |
1013 | /* Reflect first query at mrouted. */ |
1014 | ||
0c12295a | 1015 | err = ipmr_cache_report(mrt, skb, vifi, IGMPMSG_NOCACHE); |
4feb88e5 | 1016 | if (err < 0) { |
e905a9ed | 1017 | /* If the report failed throw the cache entry |
1da177e4 LT |
1018 | out - Brad Parker |
1019 | */ | |
1020 | spin_unlock_bh(&mfc_unres_lock); | |
1021 | ||
5c0a66f5 | 1022 | ipmr_cache_free(c); |
1da177e4 LT |
1023 | kfree_skb(skb); |
1024 | return err; | |
1025 | } | |
1026 | ||
0c12295a PM |
1027 | atomic_inc(&mrt->cache_resolve_queue_len); |
1028 | list_add(&c->list, &mrt->mfc_unres_queue); | |
1da177e4 | 1029 | |
278554bd DM |
1030 | if (atomic_read(&mrt->cache_resolve_queue_len) == 1) |
1031 | mod_timer(&mrt->ipmr_expire_timer, c->mfc_un.unres.expires); | |
1da177e4 LT |
1032 | } |
1033 | ||
a8cb16dd ED |
1034 | /* See if we can append the packet */ |
1035 | ||
1036 | if (c->mfc_un.unres.unresolved.qlen > 3) { | |
1da177e4 LT |
1037 | kfree_skb(skb); |
1038 | err = -ENOBUFS; | |
1039 | } else { | |
c354e124 | 1040 | skb_queue_tail(&c->mfc_un.unres.unresolved, skb); |
1da177e4 LT |
1041 | err = 0; |
1042 | } | |
1043 | ||
1044 | spin_unlock_bh(&mfc_unres_lock); | |
1045 | return err; | |
1046 | } | |
1047 | ||
1048 | /* | |
1049 | * MFC cache manipulation by user space mroute daemon | |
1050 | */ | |
1051 | ||
0c12295a | 1052 | static int ipmr_mfc_delete(struct mr_table *mrt, struct mfcctl *mfc) |
1da177e4 LT |
1053 | { |
1054 | int line; | |
862465f2 | 1055 | struct mfc_cache *c, *next; |
1da177e4 | 1056 | |
c354e124 | 1057 | line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr); |
1da177e4 | 1058 | |
0c12295a | 1059 | list_for_each_entry_safe(c, next, &mrt->mfc_cache_array[line], list) { |
1da177e4 LT |
1060 | if (c->mfc_origin == mfc->mfcc_origin.s_addr && |
1061 | c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr) { | |
a8c9486b | 1062 | list_del_rcu(&c->list); |
1da177e4 | 1063 | |
5c0a66f5 | 1064 | ipmr_cache_free(c); |
1da177e4 LT |
1065 | return 0; |
1066 | } | |
1067 | } | |
1068 | return -ENOENT; | |
1069 | } | |
1070 | ||
0c12295a PM |
1071 | static int ipmr_mfc_add(struct net *net, struct mr_table *mrt, |
1072 | struct mfcctl *mfc, int mrtsock) | |
1da177e4 | 1073 | { |
862465f2 | 1074 | bool found = false; |
1da177e4 | 1075 | int line; |
862465f2 | 1076 | struct mfc_cache *uc, *c; |
1da177e4 | 1077 | |
a50436f2 PM |
1078 | if (mfc->mfcc_parent >= MAXVIFS) |
1079 | return -ENFILE; | |
1080 | ||
c354e124 | 1081 | line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr); |
1da177e4 | 1082 | |
0c12295a | 1083 | list_for_each_entry(c, &mrt->mfc_cache_array[line], list) { |
1da177e4 | 1084 | if (c->mfc_origin == mfc->mfcc_origin.s_addr && |
862465f2 PM |
1085 | c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr) { |
1086 | found = true; | |
1da177e4 | 1087 | break; |
862465f2 | 1088 | } |
1da177e4 LT |
1089 | } |
1090 | ||
862465f2 | 1091 | if (found) { |
1da177e4 LT |
1092 | write_lock_bh(&mrt_lock); |
1093 | c->mfc_parent = mfc->mfcc_parent; | |
0c12295a | 1094 | ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls); |
1da177e4 LT |
1095 | if (!mrtsock) |
1096 | c->mfc_flags |= MFC_STATIC; | |
1097 | write_unlock_bh(&mrt_lock); | |
1098 | return 0; | |
1099 | } | |
1100 | ||
f97c1e0c | 1101 | if (!ipv4_is_multicast(mfc->mfcc_mcastgrp.s_addr)) |
1da177e4 LT |
1102 | return -EINVAL; |
1103 | ||
d658f8a0 | 1104 | c = ipmr_cache_alloc(); |
c354e124 | 1105 | if (c == NULL) |
1da177e4 LT |
1106 | return -ENOMEM; |
1107 | ||
c354e124 JK |
1108 | c->mfc_origin = mfc->mfcc_origin.s_addr; |
1109 | c->mfc_mcastgrp = mfc->mfcc_mcastgrp.s_addr; | |
1110 | c->mfc_parent = mfc->mfcc_parent; | |
0c12295a | 1111 | ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls); |
1da177e4 LT |
1112 | if (!mrtsock) |
1113 | c->mfc_flags |= MFC_STATIC; | |
1114 | ||
a8c9486b | 1115 | list_add_rcu(&c->list, &mrt->mfc_cache_array[line]); |
1da177e4 LT |
1116 | |
1117 | /* | |
1118 | * Check to see if we resolved a queued list. If so we | |
1119 | * need to send on the frames and tidy up. | |
1120 | */ | |
b0ebb739 | 1121 | found = false; |
1da177e4 | 1122 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 1123 | list_for_each_entry(uc, &mrt->mfc_unres_queue, list) { |
e258beb2 | 1124 | if (uc->mfc_origin == c->mfc_origin && |
1da177e4 | 1125 | uc->mfc_mcastgrp == c->mfc_mcastgrp) { |
862465f2 | 1126 | list_del(&uc->list); |
0c12295a | 1127 | atomic_dec(&mrt->cache_resolve_queue_len); |
b0ebb739 | 1128 | found = true; |
1da177e4 LT |
1129 | break; |
1130 | } | |
1131 | } | |
0c12295a PM |
1132 | if (list_empty(&mrt->mfc_unres_queue)) |
1133 | del_timer(&mrt->ipmr_expire_timer); | |
1da177e4 LT |
1134 | spin_unlock_bh(&mfc_unres_lock); |
1135 | ||
b0ebb739 | 1136 | if (found) { |
0c12295a | 1137 | ipmr_cache_resolve(net, mrt, uc, c); |
5c0a66f5 | 1138 | ipmr_cache_free(uc); |
1da177e4 LT |
1139 | } |
1140 | return 0; | |
1141 | } | |
1142 | ||
1143 | /* | |
1144 | * Close the multicast socket, and clear the vif tables etc | |
1145 | */ | |
e905a9ed | 1146 | |
0c12295a | 1147 | static void mroute_clean_tables(struct mr_table *mrt) |
1da177e4 LT |
1148 | { |
1149 | int i; | |
d17fa6fa | 1150 | LIST_HEAD(list); |
862465f2 | 1151 | struct mfc_cache *c, *next; |
e905a9ed | 1152 | |
a8cb16dd ED |
1153 | /* Shut down all active vif entries */ |
1154 | ||
0c12295a | 1155 | for (i = 0; i < mrt->maxvif; i++) { |
a8cb16dd | 1156 | if (!(mrt->vif_table[i].flags & VIFF_STATIC)) |
0c12295a | 1157 | vif_delete(mrt, i, 0, &list); |
1da177e4 | 1158 | } |
d17fa6fa | 1159 | unregister_netdevice_many(&list); |
1da177e4 | 1160 | |
a8cb16dd ED |
1161 | /* Wipe the cache */ |
1162 | ||
862465f2 | 1163 | for (i = 0; i < MFC_LINES; i++) { |
0c12295a | 1164 | list_for_each_entry_safe(c, next, &mrt->mfc_cache_array[i], list) { |
a8c9486b | 1165 | if (c->mfc_flags & MFC_STATIC) |
1da177e4 | 1166 | continue; |
a8c9486b | 1167 | list_del_rcu(&c->list); |
5c0a66f5 | 1168 | ipmr_cache_free(c); |
1da177e4 LT |
1169 | } |
1170 | } | |
1171 | ||
0c12295a | 1172 | if (atomic_read(&mrt->cache_resolve_queue_len) != 0) { |
1da177e4 | 1173 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 1174 | list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) { |
862465f2 | 1175 | list_del(&c->list); |
0c12295a | 1176 | ipmr_destroy_unres(mrt, c); |
1da177e4 LT |
1177 | } |
1178 | spin_unlock_bh(&mfc_unres_lock); | |
1179 | } | |
1180 | } | |
1181 | ||
4c968709 ED |
1182 | /* called from ip_ra_control(), before an RCU grace period, |
1183 | * we dont need to call synchronize_rcu() here | |
1184 | */ | |
1da177e4 LT |
1185 | static void mrtsock_destruct(struct sock *sk) |
1186 | { | |
4feb88e5 | 1187 | struct net *net = sock_net(sk); |
f0ad0860 | 1188 | struct mr_table *mrt; |
4feb88e5 | 1189 | |
1da177e4 | 1190 | rtnl_lock(); |
f0ad0860 | 1191 | ipmr_for_each_table(mrt, net) { |
4c968709 | 1192 | if (sk == rtnl_dereference(mrt->mroute_sk)) { |
f0ad0860 | 1193 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)--; |
d67b8c61 ND |
1194 | inet_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING, |
1195 | NETCONFA_IFINDEX_ALL, | |
1196 | net->ipv4.devconf_all); | |
a9b3cd7f | 1197 | RCU_INIT_POINTER(mrt->mroute_sk, NULL); |
f0ad0860 PM |
1198 | mroute_clean_tables(mrt); |
1199 | } | |
1da177e4 LT |
1200 | } |
1201 | rtnl_unlock(); | |
1202 | } | |
1203 | ||
1204 | /* | |
1205 | * Socket options and virtual interface manipulation. The whole | |
1206 | * virtual interface system is a complete heap, but unfortunately | |
1207 | * that's how BSD mrouted happens to think. Maybe one day with a proper | |
1208 | * MOSPF/PIM router set up we can clean this up. | |
1209 | */ | |
e905a9ed | 1210 | |
b7058842 | 1211 | int ip_mroute_setsockopt(struct sock *sk, int optname, char __user *optval, unsigned int optlen) |
1da177e4 LT |
1212 | { |
1213 | int ret; | |
1214 | struct vifctl vif; | |
1215 | struct mfcctl mfc; | |
4feb88e5 | 1216 | struct net *net = sock_net(sk); |
f0ad0860 PM |
1217 | struct mr_table *mrt; |
1218 | ||
5e1859fb ED |
1219 | if (sk->sk_type != SOCK_RAW || |
1220 | inet_sk(sk)->inet_num != IPPROTO_IGMP) | |
1221 | return -EOPNOTSUPP; | |
1222 | ||
f0ad0860 PM |
1223 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); |
1224 | if (mrt == NULL) | |
1225 | return -ENOENT; | |
e905a9ed | 1226 | |
132adf54 | 1227 | if (optname != MRT_INIT) { |
33d480ce | 1228 | if (sk != rcu_access_pointer(mrt->mroute_sk) && |
52e804c6 | 1229 | !ns_capable(net->user_ns, CAP_NET_ADMIN)) |
1da177e4 LT |
1230 | return -EACCES; |
1231 | } | |
1232 | ||
132adf54 SH |
1233 | switch (optname) { |
1234 | case MRT_INIT: | |
c354e124 | 1235 | if (optlen != sizeof(int)) |
5e1859fb | 1236 | return -EINVAL; |
1da177e4 | 1237 | |
132adf54 | 1238 | rtnl_lock(); |
4c968709 | 1239 | if (rtnl_dereference(mrt->mroute_sk)) { |
1da177e4 | 1240 | rtnl_unlock(); |
132adf54 SH |
1241 | return -EADDRINUSE; |
1242 | } | |
1243 | ||
1244 | ret = ip_ra_control(sk, 1, mrtsock_destruct); | |
1245 | if (ret == 0) { | |
cf778b00 | 1246 | rcu_assign_pointer(mrt->mroute_sk, sk); |
4feb88e5 | 1247 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)++; |
d67b8c61 ND |
1248 | inet_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING, |
1249 | NETCONFA_IFINDEX_ALL, | |
1250 | net->ipv4.devconf_all); | |
132adf54 SH |
1251 | } |
1252 | rtnl_unlock(); | |
1253 | return ret; | |
1254 | case MRT_DONE: | |
33d480ce | 1255 | if (sk != rcu_access_pointer(mrt->mroute_sk)) |
132adf54 SH |
1256 | return -EACCES; |
1257 | return ip_ra_control(sk, 0, NULL); | |
1258 | case MRT_ADD_VIF: | |
1259 | case MRT_DEL_VIF: | |
c354e124 | 1260 | if (optlen != sizeof(vif)) |
132adf54 | 1261 | return -EINVAL; |
c354e124 | 1262 | if (copy_from_user(&vif, optval, sizeof(vif))) |
132adf54 SH |
1263 | return -EFAULT; |
1264 | if (vif.vifc_vifi >= MAXVIFS) | |
1265 | return -ENFILE; | |
1266 | rtnl_lock(); | |
c354e124 | 1267 | if (optname == MRT_ADD_VIF) { |
4c968709 ED |
1268 | ret = vif_add(net, mrt, &vif, |
1269 | sk == rtnl_dereference(mrt->mroute_sk)); | |
132adf54 | 1270 | } else { |
0c12295a | 1271 | ret = vif_delete(mrt, vif.vifc_vifi, 0, NULL); |
132adf54 SH |
1272 | } |
1273 | rtnl_unlock(); | |
1274 | return ret; | |
1da177e4 LT |
1275 | |
1276 | /* | |
1277 | * Manipulate the forwarding caches. These live | |
1278 | * in a sort of kernel/user symbiosis. | |
1279 | */ | |
132adf54 SH |
1280 | case MRT_ADD_MFC: |
1281 | case MRT_DEL_MFC: | |
c354e124 | 1282 | if (optlen != sizeof(mfc)) |
132adf54 | 1283 | return -EINVAL; |
c354e124 | 1284 | if (copy_from_user(&mfc, optval, sizeof(mfc))) |
132adf54 SH |
1285 | return -EFAULT; |
1286 | rtnl_lock(); | |
c354e124 | 1287 | if (optname == MRT_DEL_MFC) |
0c12295a | 1288 | ret = ipmr_mfc_delete(mrt, &mfc); |
132adf54 | 1289 | else |
4c968709 ED |
1290 | ret = ipmr_mfc_add(net, mrt, &mfc, |
1291 | sk == rtnl_dereference(mrt->mroute_sk)); | |
132adf54 SH |
1292 | rtnl_unlock(); |
1293 | return ret; | |
1da177e4 LT |
1294 | /* |
1295 | * Control PIM assert. | |
1296 | */ | |
132adf54 SH |
1297 | case MRT_ASSERT: |
1298 | { | |
1299 | int v; | |
5e1859fb ED |
1300 | if (optlen != sizeof(v)) |
1301 | return -EINVAL; | |
a8cb16dd | 1302 | if (get_user(v, (int __user *)optval)) |
132adf54 | 1303 | return -EFAULT; |
53d6841d | 1304 | mrt->mroute_do_assert = v; |
132adf54 SH |
1305 | return 0; |
1306 | } | |
1da177e4 | 1307 | #ifdef CONFIG_IP_PIMSM |
132adf54 SH |
1308 | case MRT_PIM: |
1309 | { | |
ba93ef74 SH |
1310 | int v; |
1311 | ||
5e1859fb ED |
1312 | if (optlen != sizeof(v)) |
1313 | return -EINVAL; | |
a8cb16dd | 1314 | if (get_user(v, (int __user *)optval)) |
132adf54 | 1315 | return -EFAULT; |
5e1859fb | 1316 | v = !!v; |
ba93ef74 | 1317 | |
132adf54 SH |
1318 | rtnl_lock(); |
1319 | ret = 0; | |
0c12295a PM |
1320 | if (v != mrt->mroute_do_pim) { |
1321 | mrt->mroute_do_pim = v; | |
1322 | mrt->mroute_do_assert = v; | |
1da177e4 | 1323 | } |
132adf54 SH |
1324 | rtnl_unlock(); |
1325 | return ret; | |
1326 | } | |
f0ad0860 PM |
1327 | #endif |
1328 | #ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES | |
1329 | case MRT_TABLE: | |
1330 | { | |
1331 | u32 v; | |
1332 | ||
1333 | if (optlen != sizeof(u32)) | |
1334 | return -EINVAL; | |
1335 | if (get_user(v, (u32 __user *)optval)) | |
1336 | return -EFAULT; | |
f0ad0860 | 1337 | |
b49d3c1e ED |
1338 | /* "pimreg%u" should not exceed 16 bytes (IFNAMSIZ) */ |
1339 | if (v != RT_TABLE_DEFAULT && v >= 1000000000) | |
1340 | return -EINVAL; | |
1341 | ||
f0ad0860 PM |
1342 | rtnl_lock(); |
1343 | ret = 0; | |
4c968709 ED |
1344 | if (sk == rtnl_dereference(mrt->mroute_sk)) { |
1345 | ret = -EBUSY; | |
1346 | } else { | |
1347 | if (!ipmr_new_table(net, v)) | |
1348 | ret = -ENOMEM; | |
5e1859fb ED |
1349 | else |
1350 | raw_sk(sk)->ipmr_table = v; | |
4c968709 | 1351 | } |
f0ad0860 PM |
1352 | rtnl_unlock(); |
1353 | return ret; | |
1354 | } | |
1da177e4 | 1355 | #endif |
132adf54 SH |
1356 | /* |
1357 | * Spurious command, or MRT_VERSION which you cannot | |
1358 | * set. | |
1359 | */ | |
1360 | default: | |
1361 | return -ENOPROTOOPT; | |
1da177e4 LT |
1362 | } |
1363 | } | |
1364 | ||
1365 | /* | |
1366 | * Getsock opt support for the multicast routing system. | |
1367 | */ | |
e905a9ed | 1368 | |
c354e124 | 1369 | int ip_mroute_getsockopt(struct sock *sk, int optname, char __user *optval, int __user *optlen) |
1da177e4 LT |
1370 | { |
1371 | int olr; | |
1372 | int val; | |
4feb88e5 | 1373 | struct net *net = sock_net(sk); |
f0ad0860 PM |
1374 | struct mr_table *mrt; |
1375 | ||
5e1859fb ED |
1376 | if (sk->sk_type != SOCK_RAW || |
1377 | inet_sk(sk)->inet_num != IPPROTO_IGMP) | |
1378 | return -EOPNOTSUPP; | |
1379 | ||
f0ad0860 PM |
1380 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); |
1381 | if (mrt == NULL) | |
1382 | return -ENOENT; | |
1da177e4 | 1383 | |
c354e124 | 1384 | if (optname != MRT_VERSION && |
1da177e4 | 1385 | #ifdef CONFIG_IP_PIMSM |
a8cb16dd | 1386 | optname != MRT_PIM && |
1da177e4 | 1387 | #endif |
a8cb16dd | 1388 | optname != MRT_ASSERT) |
1da177e4 LT |
1389 | return -ENOPROTOOPT; |
1390 | ||
1391 | if (get_user(olr, optlen)) | |
1392 | return -EFAULT; | |
1393 | ||
1394 | olr = min_t(unsigned int, olr, sizeof(int)); | |
1395 | if (olr < 0) | |
1396 | return -EINVAL; | |
e905a9ed | 1397 | |
c354e124 | 1398 | if (put_user(olr, optlen)) |
1da177e4 | 1399 | return -EFAULT; |
c354e124 JK |
1400 | if (optname == MRT_VERSION) |
1401 | val = 0x0305; | |
1da177e4 | 1402 | #ifdef CONFIG_IP_PIMSM |
c354e124 | 1403 | else if (optname == MRT_PIM) |
0c12295a | 1404 | val = mrt->mroute_do_pim; |
1da177e4 LT |
1405 | #endif |
1406 | else | |
0c12295a | 1407 | val = mrt->mroute_do_assert; |
c354e124 | 1408 | if (copy_to_user(optval, &val, olr)) |
1da177e4 LT |
1409 | return -EFAULT; |
1410 | return 0; | |
1411 | } | |
1412 | ||
1413 | /* | |
1414 | * The IP multicast ioctl support routines. | |
1415 | */ | |
e905a9ed | 1416 | |
1da177e4 LT |
1417 | int ipmr_ioctl(struct sock *sk, int cmd, void __user *arg) |
1418 | { | |
1419 | struct sioc_sg_req sr; | |
1420 | struct sioc_vif_req vr; | |
1421 | struct vif_device *vif; | |
1422 | struct mfc_cache *c; | |
4feb88e5 | 1423 | struct net *net = sock_net(sk); |
f0ad0860 PM |
1424 | struct mr_table *mrt; |
1425 | ||
1426 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); | |
1427 | if (mrt == NULL) | |
1428 | return -ENOENT; | |
e905a9ed | 1429 | |
132adf54 SH |
1430 | switch (cmd) { |
1431 | case SIOCGETVIFCNT: | |
c354e124 | 1432 | if (copy_from_user(&vr, arg, sizeof(vr))) |
132adf54 | 1433 | return -EFAULT; |
0c12295a | 1434 | if (vr.vifi >= mrt->maxvif) |
132adf54 SH |
1435 | return -EINVAL; |
1436 | read_lock(&mrt_lock); | |
0c12295a PM |
1437 | vif = &mrt->vif_table[vr.vifi]; |
1438 | if (VIF_EXISTS(mrt, vr.vifi)) { | |
c354e124 JK |
1439 | vr.icount = vif->pkt_in; |
1440 | vr.ocount = vif->pkt_out; | |
1441 | vr.ibytes = vif->bytes_in; | |
1442 | vr.obytes = vif->bytes_out; | |
1da177e4 | 1443 | read_unlock(&mrt_lock); |
1da177e4 | 1444 | |
c354e124 | 1445 | if (copy_to_user(arg, &vr, sizeof(vr))) |
132adf54 SH |
1446 | return -EFAULT; |
1447 | return 0; | |
1448 | } | |
1449 | read_unlock(&mrt_lock); | |
1450 | return -EADDRNOTAVAIL; | |
1451 | case SIOCGETSGCNT: | |
c354e124 | 1452 | if (copy_from_user(&sr, arg, sizeof(sr))) |
132adf54 SH |
1453 | return -EFAULT; |
1454 | ||
a8c9486b | 1455 | rcu_read_lock(); |
0c12295a | 1456 | c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr); |
132adf54 SH |
1457 | if (c) { |
1458 | sr.pktcnt = c->mfc_un.res.pkt; | |
1459 | sr.bytecnt = c->mfc_un.res.bytes; | |
1460 | sr.wrong_if = c->mfc_un.res.wrong_if; | |
a8c9486b | 1461 | rcu_read_unlock(); |
132adf54 | 1462 | |
c354e124 | 1463 | if (copy_to_user(arg, &sr, sizeof(sr))) |
132adf54 SH |
1464 | return -EFAULT; |
1465 | return 0; | |
1466 | } | |
a8c9486b | 1467 | rcu_read_unlock(); |
132adf54 SH |
1468 | return -EADDRNOTAVAIL; |
1469 | default: | |
1470 | return -ENOIOCTLCMD; | |
1da177e4 LT |
1471 | } |
1472 | } | |
1473 | ||
709b46e8 EB |
1474 | #ifdef CONFIG_COMPAT |
1475 | struct compat_sioc_sg_req { | |
1476 | struct in_addr src; | |
1477 | struct in_addr grp; | |
1478 | compat_ulong_t pktcnt; | |
1479 | compat_ulong_t bytecnt; | |
1480 | compat_ulong_t wrong_if; | |
1481 | }; | |
1482 | ||
ca6b8bb0 DM |
1483 | struct compat_sioc_vif_req { |
1484 | vifi_t vifi; /* Which iface */ | |
1485 | compat_ulong_t icount; | |
1486 | compat_ulong_t ocount; | |
1487 | compat_ulong_t ibytes; | |
1488 | compat_ulong_t obytes; | |
1489 | }; | |
1490 | ||
709b46e8 EB |
1491 | int ipmr_compat_ioctl(struct sock *sk, unsigned int cmd, void __user *arg) |
1492 | { | |
0033d5ad | 1493 | struct compat_sioc_sg_req sr; |
ca6b8bb0 DM |
1494 | struct compat_sioc_vif_req vr; |
1495 | struct vif_device *vif; | |
709b46e8 EB |
1496 | struct mfc_cache *c; |
1497 | struct net *net = sock_net(sk); | |
1498 | struct mr_table *mrt; | |
1499 | ||
1500 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); | |
1501 | if (mrt == NULL) | |
1502 | return -ENOENT; | |
1503 | ||
1504 | switch (cmd) { | |
ca6b8bb0 DM |
1505 | case SIOCGETVIFCNT: |
1506 | if (copy_from_user(&vr, arg, sizeof(vr))) | |
1507 | return -EFAULT; | |
1508 | if (vr.vifi >= mrt->maxvif) | |
1509 | return -EINVAL; | |
1510 | read_lock(&mrt_lock); | |
1511 | vif = &mrt->vif_table[vr.vifi]; | |
1512 | if (VIF_EXISTS(mrt, vr.vifi)) { | |
1513 | vr.icount = vif->pkt_in; | |
1514 | vr.ocount = vif->pkt_out; | |
1515 | vr.ibytes = vif->bytes_in; | |
1516 | vr.obytes = vif->bytes_out; | |
1517 | read_unlock(&mrt_lock); | |
1518 | ||
1519 | if (copy_to_user(arg, &vr, sizeof(vr))) | |
1520 | return -EFAULT; | |
1521 | return 0; | |
1522 | } | |
1523 | read_unlock(&mrt_lock); | |
1524 | return -EADDRNOTAVAIL; | |
709b46e8 EB |
1525 | case SIOCGETSGCNT: |
1526 | if (copy_from_user(&sr, arg, sizeof(sr))) | |
1527 | return -EFAULT; | |
1528 | ||
1529 | rcu_read_lock(); | |
1530 | c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr); | |
1531 | if (c) { | |
1532 | sr.pktcnt = c->mfc_un.res.pkt; | |
1533 | sr.bytecnt = c->mfc_un.res.bytes; | |
1534 | sr.wrong_if = c->mfc_un.res.wrong_if; | |
1535 | rcu_read_unlock(); | |
1536 | ||
1537 | if (copy_to_user(arg, &sr, sizeof(sr))) | |
1538 | return -EFAULT; | |
1539 | return 0; | |
1540 | } | |
1541 | rcu_read_unlock(); | |
1542 | return -EADDRNOTAVAIL; | |
1543 | default: | |
1544 | return -ENOIOCTLCMD; | |
1545 | } | |
1546 | } | |
1547 | #endif | |
1548 | ||
1da177e4 LT |
1549 | |
1550 | static int ipmr_device_event(struct notifier_block *this, unsigned long event, void *ptr) | |
1551 | { | |
e9dc8653 | 1552 | struct net_device *dev = ptr; |
4feb88e5 | 1553 | struct net *net = dev_net(dev); |
f0ad0860 | 1554 | struct mr_table *mrt; |
1da177e4 LT |
1555 | struct vif_device *v; |
1556 | int ct; | |
e9dc8653 | 1557 | |
1da177e4 LT |
1558 | if (event != NETDEV_UNREGISTER) |
1559 | return NOTIFY_DONE; | |
f0ad0860 PM |
1560 | |
1561 | ipmr_for_each_table(mrt, net) { | |
1562 | v = &mrt->vif_table[0]; | |
1563 | for (ct = 0; ct < mrt->maxvif; ct++, v++) { | |
1564 | if (v->dev == dev) | |
e92036a6 | 1565 | vif_delete(mrt, ct, 1, NULL); |
f0ad0860 | 1566 | } |
1da177e4 LT |
1567 | } |
1568 | return NOTIFY_DONE; | |
1569 | } | |
1570 | ||
1571 | ||
c354e124 | 1572 | static struct notifier_block ip_mr_notifier = { |
1da177e4 LT |
1573 | .notifier_call = ipmr_device_event, |
1574 | }; | |
1575 | ||
1576 | /* | |
a8cb16dd | 1577 | * Encapsulate a packet by attaching a valid IPIP header to it. |
1da177e4 LT |
1578 | * This avoids tunnel drivers and other mess and gives us the speed so |
1579 | * important for multicast video. | |
1580 | */ | |
e905a9ed | 1581 | |
114c7844 | 1582 | static void ip_encap(struct sk_buff *skb, __be32 saddr, __be32 daddr) |
1da177e4 | 1583 | { |
8856dfa3 | 1584 | struct iphdr *iph; |
b71d1d42 | 1585 | const struct iphdr *old_iph = ip_hdr(skb); |
8856dfa3 ACM |
1586 | |
1587 | skb_push(skb, sizeof(struct iphdr)); | |
b0e380b1 | 1588 | skb->transport_header = skb->network_header; |
8856dfa3 | 1589 | skb_reset_network_header(skb); |
eddc9ec5 | 1590 | iph = ip_hdr(skb); |
1da177e4 | 1591 | |
a8cb16dd | 1592 | iph->version = 4; |
e023dd64 ACM |
1593 | iph->tos = old_iph->tos; |
1594 | iph->ttl = old_iph->ttl; | |
1da177e4 LT |
1595 | iph->frag_off = 0; |
1596 | iph->daddr = daddr; | |
1597 | iph->saddr = saddr; | |
1598 | iph->protocol = IPPROTO_IPIP; | |
1599 | iph->ihl = 5; | |
1600 | iph->tot_len = htons(skb->len); | |
adf30907 | 1601 | ip_select_ident(iph, skb_dst(skb), NULL); |
1da177e4 LT |
1602 | ip_send_check(iph); |
1603 | ||
1da177e4 LT |
1604 | memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt)); |
1605 | nf_reset(skb); | |
1606 | } | |
1607 | ||
1608 | static inline int ipmr_forward_finish(struct sk_buff *skb) | |
1609 | { | |
a8cb16dd | 1610 | struct ip_options *opt = &(IPCB(skb)->opt); |
1da177e4 | 1611 | |
adf30907 | 1612 | IP_INC_STATS_BH(dev_net(skb_dst(skb)->dev), IPSTATS_MIB_OUTFORWDATAGRAMS); |
2d8dbb04 | 1613 | IP_ADD_STATS_BH(dev_net(skb_dst(skb)->dev), IPSTATS_MIB_OUTOCTETS, skb->len); |
1da177e4 LT |
1614 | |
1615 | if (unlikely(opt->optlen)) | |
1616 | ip_forward_options(skb); | |
1617 | ||
1618 | return dst_output(skb); | |
1619 | } | |
1620 | ||
1621 | /* | |
1622 | * Processing handlers for ipmr_forward | |
1623 | */ | |
1624 | ||
0c12295a PM |
1625 | static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt, |
1626 | struct sk_buff *skb, struct mfc_cache *c, int vifi) | |
1da177e4 | 1627 | { |
eddc9ec5 | 1628 | const struct iphdr *iph = ip_hdr(skb); |
0c12295a | 1629 | struct vif_device *vif = &mrt->vif_table[vifi]; |
1da177e4 LT |
1630 | struct net_device *dev; |
1631 | struct rtable *rt; | |
31e4543d | 1632 | struct flowi4 fl4; |
1da177e4 LT |
1633 | int encap = 0; |
1634 | ||
1635 | if (vif->dev == NULL) | |
1636 | goto out_free; | |
1637 | ||
1638 | #ifdef CONFIG_IP_PIMSM | |
1639 | if (vif->flags & VIFF_REGISTER) { | |
1640 | vif->pkt_out++; | |
c354e124 | 1641 | vif->bytes_out += skb->len; |
cf3677ae PE |
1642 | vif->dev->stats.tx_bytes += skb->len; |
1643 | vif->dev->stats.tx_packets++; | |
0c12295a | 1644 | ipmr_cache_report(mrt, skb, vifi, IGMPMSG_WHOLEPKT); |
69ebbf58 | 1645 | goto out_free; |
1da177e4 LT |
1646 | } |
1647 | #endif | |
1648 | ||
a8cb16dd | 1649 | if (vif->flags & VIFF_TUNNEL) { |
31e4543d | 1650 | rt = ip_route_output_ports(net, &fl4, NULL, |
78fbfd8a DM |
1651 | vif->remote, vif->local, |
1652 | 0, 0, | |
1653 | IPPROTO_IPIP, | |
1654 | RT_TOS(iph->tos), vif->link); | |
b23dd4fe | 1655 | if (IS_ERR(rt)) |
1da177e4 LT |
1656 | goto out_free; |
1657 | encap = sizeof(struct iphdr); | |
1658 | } else { | |
31e4543d | 1659 | rt = ip_route_output_ports(net, &fl4, NULL, iph->daddr, 0, |
78fbfd8a DM |
1660 | 0, 0, |
1661 | IPPROTO_IPIP, | |
1662 | RT_TOS(iph->tos), vif->link); | |
b23dd4fe | 1663 | if (IS_ERR(rt)) |
1da177e4 LT |
1664 | goto out_free; |
1665 | } | |
1666 | ||
d8d1f30b | 1667 | dev = rt->dst.dev; |
1da177e4 | 1668 | |
d8d1f30b | 1669 | if (skb->len+encap > dst_mtu(&rt->dst) && (ntohs(iph->frag_off) & IP_DF)) { |
1da177e4 | 1670 | /* Do not fragment multicasts. Alas, IPv4 does not |
a8cb16dd ED |
1671 | * allow to send ICMP, so that packets will disappear |
1672 | * to blackhole. | |
1da177e4 LT |
1673 | */ |
1674 | ||
7c73a6fa | 1675 | IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_FRAGFAILS); |
1da177e4 LT |
1676 | ip_rt_put(rt); |
1677 | goto out_free; | |
1678 | } | |
1679 | ||
d8d1f30b | 1680 | encap += LL_RESERVED_SPACE(dev) + rt->dst.header_len; |
1da177e4 LT |
1681 | |
1682 | if (skb_cow(skb, encap)) { | |
e905a9ed | 1683 | ip_rt_put(rt); |
1da177e4 LT |
1684 | goto out_free; |
1685 | } | |
1686 | ||
1687 | vif->pkt_out++; | |
c354e124 | 1688 | vif->bytes_out += skb->len; |
1da177e4 | 1689 | |
adf30907 | 1690 | skb_dst_drop(skb); |
d8d1f30b | 1691 | skb_dst_set(skb, &rt->dst); |
eddc9ec5 | 1692 | ip_decrease_ttl(ip_hdr(skb)); |
1da177e4 LT |
1693 | |
1694 | /* FIXME: forward and output firewalls used to be called here. | |
a8cb16dd ED |
1695 | * What do we do with netfilter? -- RR |
1696 | */ | |
1da177e4 LT |
1697 | if (vif->flags & VIFF_TUNNEL) { |
1698 | ip_encap(skb, vif->local, vif->remote); | |
1699 | /* FIXME: extra output firewall step used to be here. --RR */ | |
2f4c02d4 PE |
1700 | vif->dev->stats.tx_packets++; |
1701 | vif->dev->stats.tx_bytes += skb->len; | |
1da177e4 LT |
1702 | } |
1703 | ||
1704 | IPCB(skb)->flags |= IPSKB_FORWARDED; | |
1705 | ||
1706 | /* | |
1707 | * RFC1584 teaches, that DVMRP/PIM router must deliver packets locally | |
1708 | * not only before forwarding, but after forwarding on all output | |
1709 | * interfaces. It is clear, if mrouter runs a multicasting | |
1710 | * program, it should receive packets not depending to what interface | |
1711 | * program is joined. | |
1712 | * If we will not make it, the program will have to join on all | |
1713 | * interfaces. On the other hand, multihoming host (or router, but | |
1714 | * not mrouter) cannot join to more than one interface - it will | |
1715 | * result in receiving multiple packets. | |
1716 | */ | |
9bbc768a | 1717 | NF_HOOK(NFPROTO_IPV4, NF_INET_FORWARD, skb, skb->dev, dev, |
1da177e4 LT |
1718 | ipmr_forward_finish); |
1719 | return; | |
1720 | ||
1721 | out_free: | |
1722 | kfree_skb(skb); | |
1da177e4 LT |
1723 | } |
1724 | ||
0c12295a | 1725 | static int ipmr_find_vif(struct mr_table *mrt, struct net_device *dev) |
1da177e4 LT |
1726 | { |
1727 | int ct; | |
0c12295a PM |
1728 | |
1729 | for (ct = mrt->maxvif-1; ct >= 0; ct--) { | |
1730 | if (mrt->vif_table[ct].dev == dev) | |
1da177e4 LT |
1731 | break; |
1732 | } | |
1733 | return ct; | |
1734 | } | |
1735 | ||
1736 | /* "local" means that we should preserve one skb (for local delivery) */ | |
1737 | ||
0c12295a PM |
1738 | static int ip_mr_forward(struct net *net, struct mr_table *mrt, |
1739 | struct sk_buff *skb, struct mfc_cache *cache, | |
1740 | int local) | |
1da177e4 LT |
1741 | { |
1742 | int psend = -1; | |
1743 | int vif, ct; | |
1744 | ||
1745 | vif = cache->mfc_parent; | |
1746 | cache->mfc_un.res.pkt++; | |
1747 | cache->mfc_un.res.bytes += skb->len; | |
1748 | ||
1749 | /* | |
1750 | * Wrong interface: drop packet and (maybe) send PIM assert. | |
1751 | */ | |
0c12295a | 1752 | if (mrt->vif_table[vif].dev != skb->dev) { |
1da177e4 LT |
1753 | int true_vifi; |
1754 | ||
c7537967 | 1755 | if (rt_is_output_route(skb_rtable(skb))) { |
1da177e4 | 1756 | /* It is our own packet, looped back. |
a8cb16dd ED |
1757 | * Very complicated situation... |
1758 | * | |
1759 | * The best workaround until routing daemons will be | |
1760 | * fixed is not to redistribute packet, if it was | |
1761 | * send through wrong interface. It means, that | |
1762 | * multicast applications WILL NOT work for | |
1763 | * (S,G), which have default multicast route pointing | |
1764 | * to wrong oif. In any case, it is not a good | |
1765 | * idea to use multicasting applications on router. | |
1da177e4 LT |
1766 | */ |
1767 | goto dont_forward; | |
1768 | } | |
1769 | ||
1770 | cache->mfc_un.res.wrong_if++; | |
0c12295a | 1771 | true_vifi = ipmr_find_vif(mrt, skb->dev); |
1da177e4 | 1772 | |
0c12295a | 1773 | if (true_vifi >= 0 && mrt->mroute_do_assert && |
1da177e4 | 1774 | /* pimsm uses asserts, when switching from RPT to SPT, |
a8cb16dd ED |
1775 | * so that we cannot check that packet arrived on an oif. |
1776 | * It is bad, but otherwise we would need to move pretty | |
1777 | * large chunk of pimd to kernel. Ough... --ANK | |
1da177e4 | 1778 | */ |
0c12295a | 1779 | (mrt->mroute_do_pim || |
6f9374a9 | 1780 | cache->mfc_un.res.ttls[true_vifi] < 255) && |
e905a9ed | 1781 | time_after(jiffies, |
1da177e4 LT |
1782 | cache->mfc_un.res.last_assert + MFC_ASSERT_THRESH)) { |
1783 | cache->mfc_un.res.last_assert = jiffies; | |
0c12295a | 1784 | ipmr_cache_report(mrt, skb, true_vifi, IGMPMSG_WRONGVIF); |
1da177e4 LT |
1785 | } |
1786 | goto dont_forward; | |
1787 | } | |
1788 | ||
0c12295a PM |
1789 | mrt->vif_table[vif].pkt_in++; |
1790 | mrt->vif_table[vif].bytes_in += skb->len; | |
1da177e4 LT |
1791 | |
1792 | /* | |
1793 | * Forward the frame | |
1794 | */ | |
a8cb16dd ED |
1795 | for (ct = cache->mfc_un.res.maxvif - 1; |
1796 | ct >= cache->mfc_un.res.minvif; ct--) { | |
eddc9ec5 | 1797 | if (ip_hdr(skb)->ttl > cache->mfc_un.res.ttls[ct]) { |
1da177e4 LT |
1798 | if (psend != -1) { |
1799 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
a8cb16dd | 1800 | |
1da177e4 | 1801 | if (skb2) |
0c12295a PM |
1802 | ipmr_queue_xmit(net, mrt, skb2, cache, |
1803 | psend); | |
1da177e4 | 1804 | } |
c354e124 | 1805 | psend = ct; |
1da177e4 LT |
1806 | } |
1807 | } | |
1808 | if (psend != -1) { | |
1809 | if (local) { | |
1810 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
a8cb16dd | 1811 | |
1da177e4 | 1812 | if (skb2) |
0c12295a | 1813 | ipmr_queue_xmit(net, mrt, skb2, cache, psend); |
1da177e4 | 1814 | } else { |
0c12295a | 1815 | ipmr_queue_xmit(net, mrt, skb, cache, psend); |
1da177e4 LT |
1816 | return 0; |
1817 | } | |
1818 | } | |
1819 | ||
1820 | dont_forward: | |
1821 | if (!local) | |
1822 | kfree_skb(skb); | |
1823 | return 0; | |
1824 | } | |
1825 | ||
417da66f | 1826 | static struct mr_table *ipmr_rt_fib_lookup(struct net *net, struct sk_buff *skb) |
ee3f1aaf | 1827 | { |
417da66f DM |
1828 | struct rtable *rt = skb_rtable(skb); |
1829 | struct iphdr *iph = ip_hdr(skb); | |
da91981b | 1830 | struct flowi4 fl4 = { |
417da66f DM |
1831 | .daddr = iph->daddr, |
1832 | .saddr = iph->saddr, | |
b0fe4a31 | 1833 | .flowi4_tos = RT_TOS(iph->tos), |
4fd551d7 DM |
1834 | .flowi4_oif = (rt_is_output_route(rt) ? |
1835 | skb->dev->ifindex : 0), | |
1836 | .flowi4_iif = (rt_is_output_route(rt) ? | |
1fb9489b | 1837 | LOOPBACK_IFINDEX : |
4fd551d7 | 1838 | skb->dev->ifindex), |
b4869889 | 1839 | .flowi4_mark = skb->mark, |
ee3f1aaf DM |
1840 | }; |
1841 | struct mr_table *mrt; | |
1842 | int err; | |
1843 | ||
da91981b | 1844 | err = ipmr_fib_lookup(net, &fl4, &mrt); |
ee3f1aaf DM |
1845 | if (err) |
1846 | return ERR_PTR(err); | |
1847 | return mrt; | |
1848 | } | |
1da177e4 LT |
1849 | |
1850 | /* | |
1851 | * Multicast packets for forwarding arrive here | |
4c968709 | 1852 | * Called with rcu_read_lock(); |
1da177e4 LT |
1853 | */ |
1854 | ||
1855 | int ip_mr_input(struct sk_buff *skb) | |
1856 | { | |
1857 | struct mfc_cache *cache; | |
4feb88e5 | 1858 | struct net *net = dev_net(skb->dev); |
511c3f92 | 1859 | int local = skb_rtable(skb)->rt_flags & RTCF_LOCAL; |
f0ad0860 | 1860 | struct mr_table *mrt; |
1da177e4 LT |
1861 | |
1862 | /* Packet is looped back after forward, it should not be | |
a8cb16dd | 1863 | * forwarded second time, but still can be delivered locally. |
1da177e4 | 1864 | */ |
4c968709 | 1865 | if (IPCB(skb)->flags & IPSKB_FORWARDED) |
1da177e4 LT |
1866 | goto dont_forward; |
1867 | ||
417da66f | 1868 | mrt = ipmr_rt_fib_lookup(net, skb); |
ee3f1aaf DM |
1869 | if (IS_ERR(mrt)) { |
1870 | kfree_skb(skb); | |
1871 | return PTR_ERR(mrt); | |
e40dbc51 | 1872 | } |
1da177e4 | 1873 | if (!local) { |
4c968709 ED |
1874 | if (IPCB(skb)->opt.router_alert) { |
1875 | if (ip_call_ra_chain(skb)) | |
1876 | return 0; | |
1877 | } else if (ip_hdr(skb)->protocol == IPPROTO_IGMP) { | |
1878 | /* IGMPv1 (and broken IGMPv2 implementations sort of | |
1879 | * Cisco IOS <= 11.2(8)) do not put router alert | |
1880 | * option to IGMP packets destined to routable | |
1881 | * groups. It is very bad, because it means | |
1882 | * that we can forward NO IGMP messages. | |
1883 | */ | |
1884 | struct sock *mroute_sk; | |
1885 | ||
1886 | mroute_sk = rcu_dereference(mrt->mroute_sk); | |
1887 | if (mroute_sk) { | |
1888 | nf_reset(skb); | |
1889 | raw_rcv(mroute_sk, skb); | |
1890 | return 0; | |
1891 | } | |
1da177e4 LT |
1892 | } |
1893 | } | |
1894 | ||
a8c9486b | 1895 | /* already under rcu_read_lock() */ |
0c12295a | 1896 | cache = ipmr_cache_find(mrt, ip_hdr(skb)->saddr, ip_hdr(skb)->daddr); |
1da177e4 LT |
1897 | |
1898 | /* | |
1899 | * No usable cache entry | |
1900 | */ | |
c354e124 | 1901 | if (cache == NULL) { |
1da177e4 LT |
1902 | int vif; |
1903 | ||
1904 | if (local) { | |
1905 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
1906 | ip_local_deliver(skb); | |
a8c9486b | 1907 | if (skb2 == NULL) |
1da177e4 | 1908 | return -ENOBUFS; |
1da177e4 LT |
1909 | skb = skb2; |
1910 | } | |
1911 | ||
a8c9486b | 1912 | read_lock(&mrt_lock); |
0c12295a | 1913 | vif = ipmr_find_vif(mrt, skb->dev); |
1da177e4 | 1914 | if (vif >= 0) { |
0eae88f3 | 1915 | int err2 = ipmr_cache_unresolved(mrt, vif, skb); |
1da177e4 LT |
1916 | read_unlock(&mrt_lock); |
1917 | ||
0eae88f3 | 1918 | return err2; |
1da177e4 LT |
1919 | } |
1920 | read_unlock(&mrt_lock); | |
1921 | kfree_skb(skb); | |
1922 | return -ENODEV; | |
1923 | } | |
1924 | ||
a8c9486b | 1925 | read_lock(&mrt_lock); |
0c12295a | 1926 | ip_mr_forward(net, mrt, skb, cache, local); |
1da177e4 LT |
1927 | read_unlock(&mrt_lock); |
1928 | ||
1929 | if (local) | |
1930 | return ip_local_deliver(skb); | |
1931 | ||
1932 | return 0; | |
1933 | ||
1934 | dont_forward: | |
1935 | if (local) | |
1936 | return ip_local_deliver(skb); | |
1937 | kfree_skb(skb); | |
1938 | return 0; | |
1939 | } | |
1940 | ||
b1879204 | 1941 | #ifdef CONFIG_IP_PIMSM |
55747a0a | 1942 | /* called with rcu_read_lock() */ |
f0ad0860 PM |
1943 | static int __pim_rcv(struct mr_table *mrt, struct sk_buff *skb, |
1944 | unsigned int pimlen) | |
1da177e4 | 1945 | { |
b1879204 IJ |
1946 | struct net_device *reg_dev = NULL; |
1947 | struct iphdr *encap; | |
1da177e4 | 1948 | |
b1879204 | 1949 | encap = (struct iphdr *)(skb_transport_header(skb) + pimlen); |
1da177e4 | 1950 | /* |
a8cb16dd ED |
1951 | * Check that: |
1952 | * a. packet is really sent to a multicast group | |
1953 | * b. packet is not a NULL-REGISTER | |
1954 | * c. packet is not truncated | |
1da177e4 | 1955 | */ |
f97c1e0c | 1956 | if (!ipv4_is_multicast(encap->daddr) || |
1da177e4 | 1957 | encap->tot_len == 0 || |
b1879204 IJ |
1958 | ntohs(encap->tot_len) + pimlen > skb->len) |
1959 | return 1; | |
1da177e4 LT |
1960 | |
1961 | read_lock(&mrt_lock); | |
0c12295a PM |
1962 | if (mrt->mroute_reg_vif_num >= 0) |
1963 | reg_dev = mrt->vif_table[mrt->mroute_reg_vif_num].dev; | |
1da177e4 LT |
1964 | read_unlock(&mrt_lock); |
1965 | ||
e905a9ed | 1966 | if (reg_dev == NULL) |
b1879204 | 1967 | return 1; |
1da177e4 | 1968 | |
b0e380b1 | 1969 | skb->mac_header = skb->network_header; |
55747a0a | 1970 | skb_pull(skb, (u8 *)encap - skb->data); |
31c7711b | 1971 | skb_reset_network_header(skb); |
1da177e4 | 1972 | skb->protocol = htons(ETH_P_IP); |
55747a0a | 1973 | skb->ip_summed = CHECKSUM_NONE; |
1da177e4 | 1974 | skb->pkt_type = PACKET_HOST; |
d19d56dd ED |
1975 | |
1976 | skb_tunnel_rx(skb, reg_dev); | |
1977 | ||
1da177e4 | 1978 | netif_rx(skb); |
b1879204 | 1979 | |
55747a0a | 1980 | return NET_RX_SUCCESS; |
b1879204 IJ |
1981 | } |
1982 | #endif | |
1983 | ||
1984 | #ifdef CONFIG_IP_PIMSM_V1 | |
1985 | /* | |
1986 | * Handle IGMP messages of PIMv1 | |
1987 | */ | |
1988 | ||
a8cb16dd | 1989 | int pim_rcv_v1(struct sk_buff *skb) |
b1879204 IJ |
1990 | { |
1991 | struct igmphdr *pim; | |
4feb88e5 | 1992 | struct net *net = dev_net(skb->dev); |
f0ad0860 | 1993 | struct mr_table *mrt; |
b1879204 IJ |
1994 | |
1995 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) | |
1996 | goto drop; | |
1997 | ||
1998 | pim = igmp_hdr(skb); | |
1999 | ||
417da66f | 2000 | mrt = ipmr_rt_fib_lookup(net, skb); |
ee3f1aaf DM |
2001 | if (IS_ERR(mrt)) |
2002 | goto drop; | |
0c12295a | 2003 | if (!mrt->mroute_do_pim || |
b1879204 IJ |
2004 | pim->group != PIM_V1_VERSION || pim->code != PIM_V1_REGISTER) |
2005 | goto drop; | |
2006 | ||
f0ad0860 | 2007 | if (__pim_rcv(mrt, skb, sizeof(*pim))) { |
b1879204 IJ |
2008 | drop: |
2009 | kfree_skb(skb); | |
2010 | } | |
1da177e4 LT |
2011 | return 0; |
2012 | } | |
2013 | #endif | |
2014 | ||
2015 | #ifdef CONFIG_IP_PIMSM_V2 | |
a8cb16dd | 2016 | static int pim_rcv(struct sk_buff *skb) |
1da177e4 LT |
2017 | { |
2018 | struct pimreghdr *pim; | |
f0ad0860 PM |
2019 | struct net *net = dev_net(skb->dev); |
2020 | struct mr_table *mrt; | |
1da177e4 | 2021 | |
b1879204 | 2022 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) |
1da177e4 LT |
2023 | goto drop; |
2024 | ||
9c70220b | 2025 | pim = (struct pimreghdr *)skb_transport_header(skb); |
a8cb16dd ED |
2026 | if (pim->type != ((PIM_VERSION << 4) | (PIM_REGISTER)) || |
2027 | (pim->flags & PIM_NULL_REGISTER) || | |
e905a9ed | 2028 | (ip_compute_csum((void *)pim, sizeof(*pim)) != 0 && |
d3bc23e7 | 2029 | csum_fold(skb_checksum(skb, 0, skb->len, 0)))) |
1da177e4 LT |
2030 | goto drop; |
2031 | ||
417da66f | 2032 | mrt = ipmr_rt_fib_lookup(net, skb); |
ee3f1aaf DM |
2033 | if (IS_ERR(mrt)) |
2034 | goto drop; | |
f0ad0860 | 2035 | if (__pim_rcv(mrt, skb, sizeof(*pim))) { |
b1879204 IJ |
2036 | drop: |
2037 | kfree_skb(skb); | |
2038 | } | |
1da177e4 LT |
2039 | return 0; |
2040 | } | |
2041 | #endif | |
2042 | ||
cb6a4e46 PM |
2043 | static int __ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
2044 | struct mfc_cache *c, struct rtmsg *rtm) | |
1da177e4 LT |
2045 | { |
2046 | int ct; | |
2047 | struct rtnexthop *nhp; | |
92a395e5 | 2048 | struct nlattr *mp_attr; |
1da177e4 | 2049 | |
7438189b | 2050 | /* If cache is unresolved, don't try to parse IIF and OIF */ |
ed0f160a | 2051 | if (c->mfc_parent >= MAXVIFS) |
7438189b ND |
2052 | return -ENOENT; |
2053 | ||
92a395e5 TG |
2054 | if (VIF_EXISTS(mrt, c->mfc_parent) && |
2055 | nla_put_u32(skb, RTA_IIF, mrt->vif_table[c->mfc_parent].dev->ifindex) < 0) | |
2056 | return -EMSGSIZE; | |
1da177e4 | 2057 | |
92a395e5 TG |
2058 | if (!(mp_attr = nla_nest_start(skb, RTA_MULTIPATH))) |
2059 | return -EMSGSIZE; | |
1da177e4 LT |
2060 | |
2061 | for (ct = c->mfc_un.res.minvif; ct < c->mfc_un.res.maxvif; ct++) { | |
0c12295a | 2062 | if (VIF_EXISTS(mrt, ct) && c->mfc_un.res.ttls[ct] < 255) { |
92a395e5 TG |
2063 | if (!(nhp = nla_reserve_nohdr(skb, sizeof(*nhp)))) { |
2064 | nla_nest_cancel(skb, mp_attr); | |
2065 | return -EMSGSIZE; | |
2066 | } | |
2067 | ||
1da177e4 LT |
2068 | nhp->rtnh_flags = 0; |
2069 | nhp->rtnh_hops = c->mfc_un.res.ttls[ct]; | |
0c12295a | 2070 | nhp->rtnh_ifindex = mrt->vif_table[ct].dev->ifindex; |
1da177e4 LT |
2071 | nhp->rtnh_len = sizeof(*nhp); |
2072 | } | |
2073 | } | |
92a395e5 TG |
2074 | |
2075 | nla_nest_end(skb, mp_attr); | |
2076 | ||
1da177e4 LT |
2077 | rtm->rtm_type = RTN_MULTICAST; |
2078 | return 1; | |
1da177e4 LT |
2079 | } |
2080 | ||
9a1b9496 DM |
2081 | int ipmr_get_route(struct net *net, struct sk_buff *skb, |
2082 | __be32 saddr, __be32 daddr, | |
2083 | struct rtmsg *rtm, int nowait) | |
1da177e4 | 2084 | { |
1da177e4 | 2085 | struct mfc_cache *cache; |
9a1b9496 DM |
2086 | struct mr_table *mrt; |
2087 | int err; | |
1da177e4 | 2088 | |
f0ad0860 PM |
2089 | mrt = ipmr_get_table(net, RT_TABLE_DEFAULT); |
2090 | if (mrt == NULL) | |
2091 | return -ENOENT; | |
2092 | ||
a8c9486b | 2093 | rcu_read_lock(); |
9a1b9496 | 2094 | cache = ipmr_cache_find(mrt, saddr, daddr); |
1da177e4 | 2095 | |
c354e124 | 2096 | if (cache == NULL) { |
72287490 | 2097 | struct sk_buff *skb2; |
eddc9ec5 | 2098 | struct iphdr *iph; |
1da177e4 | 2099 | struct net_device *dev; |
a8cb16dd | 2100 | int vif = -1; |
1da177e4 LT |
2101 | |
2102 | if (nowait) { | |
a8c9486b | 2103 | rcu_read_unlock(); |
1da177e4 LT |
2104 | return -EAGAIN; |
2105 | } | |
2106 | ||
2107 | dev = skb->dev; | |
a8c9486b | 2108 | read_lock(&mrt_lock); |
a8cb16dd ED |
2109 | if (dev) |
2110 | vif = ipmr_find_vif(mrt, dev); | |
2111 | if (vif < 0) { | |
1da177e4 | 2112 | read_unlock(&mrt_lock); |
a8c9486b | 2113 | rcu_read_unlock(); |
1da177e4 LT |
2114 | return -ENODEV; |
2115 | } | |
72287490 AK |
2116 | skb2 = skb_clone(skb, GFP_ATOMIC); |
2117 | if (!skb2) { | |
2118 | read_unlock(&mrt_lock); | |
a8c9486b | 2119 | rcu_read_unlock(); |
72287490 AK |
2120 | return -ENOMEM; |
2121 | } | |
2122 | ||
e2d1bca7 ACM |
2123 | skb_push(skb2, sizeof(struct iphdr)); |
2124 | skb_reset_network_header(skb2); | |
eddc9ec5 ACM |
2125 | iph = ip_hdr(skb2); |
2126 | iph->ihl = sizeof(struct iphdr) >> 2; | |
9a1b9496 DM |
2127 | iph->saddr = saddr; |
2128 | iph->daddr = daddr; | |
eddc9ec5 | 2129 | iph->version = 0; |
0c12295a | 2130 | err = ipmr_cache_unresolved(mrt, vif, skb2); |
1da177e4 | 2131 | read_unlock(&mrt_lock); |
a8c9486b | 2132 | rcu_read_unlock(); |
1da177e4 LT |
2133 | return err; |
2134 | } | |
2135 | ||
a8c9486b ED |
2136 | read_lock(&mrt_lock); |
2137 | if (!nowait && (rtm->rtm_flags & RTM_F_NOTIFY)) | |
1da177e4 | 2138 | cache->mfc_flags |= MFC_NOTIFY; |
cb6a4e46 | 2139 | err = __ipmr_fill_mroute(mrt, skb, cache, rtm); |
1da177e4 | 2140 | read_unlock(&mrt_lock); |
a8c9486b | 2141 | rcu_read_unlock(); |
1da177e4 LT |
2142 | return err; |
2143 | } | |
2144 | ||
cb6a4e46 | 2145 | static int ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
15e47304 | 2146 | u32 portid, u32 seq, struct mfc_cache *c) |
cb6a4e46 PM |
2147 | { |
2148 | struct nlmsghdr *nlh; | |
2149 | struct rtmsg *rtm; | |
2150 | ||
15e47304 | 2151 | nlh = nlmsg_put(skb, portid, seq, RTM_NEWROUTE, sizeof(*rtm), NLM_F_MULTI); |
cb6a4e46 PM |
2152 | if (nlh == NULL) |
2153 | return -EMSGSIZE; | |
2154 | ||
2155 | rtm = nlmsg_data(nlh); | |
2156 | rtm->rtm_family = RTNL_FAMILY_IPMR; | |
2157 | rtm->rtm_dst_len = 32; | |
2158 | rtm->rtm_src_len = 32; | |
2159 | rtm->rtm_tos = 0; | |
2160 | rtm->rtm_table = mrt->id; | |
f3756b79 DM |
2161 | if (nla_put_u32(skb, RTA_TABLE, mrt->id)) |
2162 | goto nla_put_failure; | |
cb6a4e46 PM |
2163 | rtm->rtm_type = RTN_MULTICAST; |
2164 | rtm->rtm_scope = RT_SCOPE_UNIVERSE; | |
2165 | rtm->rtm_protocol = RTPROT_UNSPEC; | |
2166 | rtm->rtm_flags = 0; | |
2167 | ||
f3756b79 DM |
2168 | if (nla_put_be32(skb, RTA_SRC, c->mfc_origin) || |
2169 | nla_put_be32(skb, RTA_DST, c->mfc_mcastgrp)) | |
2170 | goto nla_put_failure; | |
cb6a4e46 PM |
2171 | if (__ipmr_fill_mroute(mrt, skb, c, rtm) < 0) |
2172 | goto nla_put_failure; | |
2173 | ||
2174 | return nlmsg_end(skb, nlh); | |
2175 | ||
2176 | nla_put_failure: | |
2177 | nlmsg_cancel(skb, nlh); | |
2178 | return -EMSGSIZE; | |
2179 | } | |
2180 | ||
2181 | static int ipmr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb) | |
2182 | { | |
2183 | struct net *net = sock_net(skb->sk); | |
2184 | struct mr_table *mrt; | |
2185 | struct mfc_cache *mfc; | |
2186 | unsigned int t = 0, s_t; | |
2187 | unsigned int h = 0, s_h; | |
2188 | unsigned int e = 0, s_e; | |
2189 | ||
2190 | s_t = cb->args[0]; | |
2191 | s_h = cb->args[1]; | |
2192 | s_e = cb->args[2]; | |
2193 | ||
a8c9486b | 2194 | rcu_read_lock(); |
cb6a4e46 PM |
2195 | ipmr_for_each_table(mrt, net) { |
2196 | if (t < s_t) | |
2197 | goto next_table; | |
2198 | if (t > s_t) | |
2199 | s_h = 0; | |
2200 | for (h = s_h; h < MFC_LINES; h++) { | |
a8c9486b | 2201 | list_for_each_entry_rcu(mfc, &mrt->mfc_cache_array[h], list) { |
cb6a4e46 PM |
2202 | if (e < s_e) |
2203 | goto next_entry; | |
2204 | if (ipmr_fill_mroute(mrt, skb, | |
15e47304 | 2205 | NETLINK_CB(cb->skb).portid, |
cb6a4e46 PM |
2206 | cb->nlh->nlmsg_seq, |
2207 | mfc) < 0) | |
2208 | goto done; | |
2209 | next_entry: | |
2210 | e++; | |
2211 | } | |
2212 | e = s_e = 0; | |
2213 | } | |
2214 | s_h = 0; | |
2215 | next_table: | |
2216 | t++; | |
2217 | } | |
2218 | done: | |
a8c9486b | 2219 | rcu_read_unlock(); |
cb6a4e46 PM |
2220 | |
2221 | cb->args[2] = e; | |
2222 | cb->args[1] = h; | |
2223 | cb->args[0] = t; | |
2224 | ||
2225 | return skb->len; | |
2226 | } | |
2227 | ||
e905a9ed | 2228 | #ifdef CONFIG_PROC_FS |
1da177e4 | 2229 | /* |
a8cb16dd ED |
2230 | * The /proc interfaces to multicast routing : |
2231 | * /proc/net/ip_mr_cache & /proc/net/ip_mr_vif | |
1da177e4 LT |
2232 | */ |
2233 | struct ipmr_vif_iter { | |
f6bb4514 | 2234 | struct seq_net_private p; |
f0ad0860 | 2235 | struct mr_table *mrt; |
1da177e4 LT |
2236 | int ct; |
2237 | }; | |
2238 | ||
f6bb4514 BT |
2239 | static struct vif_device *ipmr_vif_seq_idx(struct net *net, |
2240 | struct ipmr_vif_iter *iter, | |
1da177e4 LT |
2241 | loff_t pos) |
2242 | { | |
f0ad0860 | 2243 | struct mr_table *mrt = iter->mrt; |
0c12295a PM |
2244 | |
2245 | for (iter->ct = 0; iter->ct < mrt->maxvif; ++iter->ct) { | |
2246 | if (!VIF_EXISTS(mrt, iter->ct)) | |
1da177e4 | 2247 | continue; |
e905a9ed | 2248 | if (pos-- == 0) |
0c12295a | 2249 | return &mrt->vif_table[iter->ct]; |
1da177e4 LT |
2250 | } |
2251 | return NULL; | |
2252 | } | |
2253 | ||
2254 | static void *ipmr_vif_seq_start(struct seq_file *seq, loff_t *pos) | |
ba93ef74 | 2255 | __acquires(mrt_lock) |
1da177e4 | 2256 | { |
f0ad0860 | 2257 | struct ipmr_vif_iter *iter = seq->private; |
f6bb4514 | 2258 | struct net *net = seq_file_net(seq); |
f0ad0860 PM |
2259 | struct mr_table *mrt; |
2260 | ||
2261 | mrt = ipmr_get_table(net, RT_TABLE_DEFAULT); | |
2262 | if (mrt == NULL) | |
2263 | return ERR_PTR(-ENOENT); | |
2264 | ||
2265 | iter->mrt = mrt; | |
f6bb4514 | 2266 | |
1da177e4 | 2267 | read_lock(&mrt_lock); |
f6bb4514 | 2268 | return *pos ? ipmr_vif_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
2269 | : SEQ_START_TOKEN; |
2270 | } | |
2271 | ||
2272 | static void *ipmr_vif_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
2273 | { | |
2274 | struct ipmr_vif_iter *iter = seq->private; | |
f6bb4514 | 2275 | struct net *net = seq_file_net(seq); |
f0ad0860 | 2276 | struct mr_table *mrt = iter->mrt; |
1da177e4 LT |
2277 | |
2278 | ++*pos; | |
2279 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 2280 | return ipmr_vif_seq_idx(net, iter, 0); |
e905a9ed | 2281 | |
0c12295a PM |
2282 | while (++iter->ct < mrt->maxvif) { |
2283 | if (!VIF_EXISTS(mrt, iter->ct)) | |
1da177e4 | 2284 | continue; |
0c12295a | 2285 | return &mrt->vif_table[iter->ct]; |
1da177e4 LT |
2286 | } |
2287 | return NULL; | |
2288 | } | |
2289 | ||
2290 | static void ipmr_vif_seq_stop(struct seq_file *seq, void *v) | |
ba93ef74 | 2291 | __releases(mrt_lock) |
1da177e4 LT |
2292 | { |
2293 | read_unlock(&mrt_lock); | |
2294 | } | |
2295 | ||
2296 | static int ipmr_vif_seq_show(struct seq_file *seq, void *v) | |
2297 | { | |
f0ad0860 PM |
2298 | struct ipmr_vif_iter *iter = seq->private; |
2299 | struct mr_table *mrt = iter->mrt; | |
f6bb4514 | 2300 | |
1da177e4 | 2301 | if (v == SEQ_START_TOKEN) { |
e905a9ed | 2302 | seq_puts(seq, |
1da177e4 LT |
2303 | "Interface BytesIn PktsIn BytesOut PktsOut Flags Local Remote\n"); |
2304 | } else { | |
2305 | const struct vif_device *vif = v; | |
2306 | const char *name = vif->dev ? vif->dev->name : "none"; | |
2307 | ||
2308 | seq_printf(seq, | |
2309 | "%2Zd %-10s %8ld %7ld %8ld %7ld %05X %08X %08X\n", | |
0c12295a | 2310 | vif - mrt->vif_table, |
e905a9ed | 2311 | name, vif->bytes_in, vif->pkt_in, |
1da177e4 LT |
2312 | vif->bytes_out, vif->pkt_out, |
2313 | vif->flags, vif->local, vif->remote); | |
2314 | } | |
2315 | return 0; | |
2316 | } | |
2317 | ||
f690808e | 2318 | static const struct seq_operations ipmr_vif_seq_ops = { |
1da177e4 LT |
2319 | .start = ipmr_vif_seq_start, |
2320 | .next = ipmr_vif_seq_next, | |
2321 | .stop = ipmr_vif_seq_stop, | |
2322 | .show = ipmr_vif_seq_show, | |
2323 | }; | |
2324 | ||
2325 | static int ipmr_vif_open(struct inode *inode, struct file *file) | |
2326 | { | |
f6bb4514 BT |
2327 | return seq_open_net(inode, file, &ipmr_vif_seq_ops, |
2328 | sizeof(struct ipmr_vif_iter)); | |
1da177e4 LT |
2329 | } |
2330 | ||
9a32144e | 2331 | static const struct file_operations ipmr_vif_fops = { |
1da177e4 LT |
2332 | .owner = THIS_MODULE, |
2333 | .open = ipmr_vif_open, | |
2334 | .read = seq_read, | |
2335 | .llseek = seq_lseek, | |
f6bb4514 | 2336 | .release = seq_release_net, |
1da177e4 LT |
2337 | }; |
2338 | ||
2339 | struct ipmr_mfc_iter { | |
f6bb4514 | 2340 | struct seq_net_private p; |
f0ad0860 | 2341 | struct mr_table *mrt; |
862465f2 | 2342 | struct list_head *cache; |
1da177e4 LT |
2343 | int ct; |
2344 | }; | |
2345 | ||
2346 | ||
f6bb4514 BT |
2347 | static struct mfc_cache *ipmr_mfc_seq_idx(struct net *net, |
2348 | struct ipmr_mfc_iter *it, loff_t pos) | |
1da177e4 | 2349 | { |
f0ad0860 | 2350 | struct mr_table *mrt = it->mrt; |
1da177e4 LT |
2351 | struct mfc_cache *mfc; |
2352 | ||
a8c9486b | 2353 | rcu_read_lock(); |
862465f2 | 2354 | for (it->ct = 0; it->ct < MFC_LINES; it->ct++) { |
0c12295a | 2355 | it->cache = &mrt->mfc_cache_array[it->ct]; |
a8c9486b | 2356 | list_for_each_entry_rcu(mfc, it->cache, list) |
e905a9ed | 2357 | if (pos-- == 0) |
1da177e4 | 2358 | return mfc; |
862465f2 | 2359 | } |
a8c9486b | 2360 | rcu_read_unlock(); |
1da177e4 | 2361 | |
1da177e4 | 2362 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 2363 | it->cache = &mrt->mfc_unres_queue; |
862465f2 | 2364 | list_for_each_entry(mfc, it->cache, list) |
e258beb2 | 2365 | if (pos-- == 0) |
1da177e4 LT |
2366 | return mfc; |
2367 | spin_unlock_bh(&mfc_unres_lock); | |
2368 | ||
2369 | it->cache = NULL; | |
2370 | return NULL; | |
2371 | } | |
2372 | ||
2373 | ||
2374 | static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos) | |
2375 | { | |
2376 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 | 2377 | struct net *net = seq_file_net(seq); |
f0ad0860 | 2378 | struct mr_table *mrt; |
f6bb4514 | 2379 | |
f0ad0860 PM |
2380 | mrt = ipmr_get_table(net, RT_TABLE_DEFAULT); |
2381 | if (mrt == NULL) | |
2382 | return ERR_PTR(-ENOENT); | |
f6bb4514 | 2383 | |
f0ad0860 | 2384 | it->mrt = mrt; |
1da177e4 LT |
2385 | it->cache = NULL; |
2386 | it->ct = 0; | |
f6bb4514 | 2387 | return *pos ? ipmr_mfc_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
2388 | : SEQ_START_TOKEN; |
2389 | } | |
2390 | ||
2391 | static void *ipmr_mfc_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
2392 | { | |
2393 | struct mfc_cache *mfc = v; | |
2394 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 | 2395 | struct net *net = seq_file_net(seq); |
f0ad0860 | 2396 | struct mr_table *mrt = it->mrt; |
1da177e4 LT |
2397 | |
2398 | ++*pos; | |
2399 | ||
2400 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 2401 | return ipmr_mfc_seq_idx(net, seq->private, 0); |
1da177e4 | 2402 | |
862465f2 PM |
2403 | if (mfc->list.next != it->cache) |
2404 | return list_entry(mfc->list.next, struct mfc_cache, list); | |
e905a9ed | 2405 | |
0c12295a | 2406 | if (it->cache == &mrt->mfc_unres_queue) |
1da177e4 LT |
2407 | goto end_of_list; |
2408 | ||
0c12295a | 2409 | BUG_ON(it->cache != &mrt->mfc_cache_array[it->ct]); |
1da177e4 LT |
2410 | |
2411 | while (++it->ct < MFC_LINES) { | |
0c12295a | 2412 | it->cache = &mrt->mfc_cache_array[it->ct]; |
862465f2 PM |
2413 | if (list_empty(it->cache)) |
2414 | continue; | |
2415 | return list_first_entry(it->cache, struct mfc_cache, list); | |
1da177e4 LT |
2416 | } |
2417 | ||
2418 | /* exhausted cache_array, show unresolved */ | |
a8c9486b | 2419 | rcu_read_unlock(); |
0c12295a | 2420 | it->cache = &mrt->mfc_unres_queue; |
1da177e4 | 2421 | it->ct = 0; |
e905a9ed | 2422 | |
1da177e4 | 2423 | spin_lock_bh(&mfc_unres_lock); |
862465f2 PM |
2424 | if (!list_empty(it->cache)) |
2425 | return list_first_entry(it->cache, struct mfc_cache, list); | |
1da177e4 | 2426 | |
a8cb16dd | 2427 | end_of_list: |
1da177e4 LT |
2428 | spin_unlock_bh(&mfc_unres_lock); |
2429 | it->cache = NULL; | |
2430 | ||
2431 | return NULL; | |
2432 | } | |
2433 | ||
2434 | static void ipmr_mfc_seq_stop(struct seq_file *seq, void *v) | |
2435 | { | |
2436 | struct ipmr_mfc_iter *it = seq->private; | |
f0ad0860 | 2437 | struct mr_table *mrt = it->mrt; |
1da177e4 | 2438 | |
0c12295a | 2439 | if (it->cache == &mrt->mfc_unres_queue) |
1da177e4 | 2440 | spin_unlock_bh(&mfc_unres_lock); |
0c12295a | 2441 | else if (it->cache == &mrt->mfc_cache_array[it->ct]) |
a8c9486b | 2442 | rcu_read_unlock(); |
1da177e4 LT |
2443 | } |
2444 | ||
2445 | static int ipmr_mfc_seq_show(struct seq_file *seq, void *v) | |
2446 | { | |
2447 | int n; | |
2448 | ||
2449 | if (v == SEQ_START_TOKEN) { | |
e905a9ed | 2450 | seq_puts(seq, |
1da177e4 LT |
2451 | "Group Origin Iif Pkts Bytes Wrong Oifs\n"); |
2452 | } else { | |
2453 | const struct mfc_cache *mfc = v; | |
2454 | const struct ipmr_mfc_iter *it = seq->private; | |
f0ad0860 | 2455 | const struct mr_table *mrt = it->mrt; |
e905a9ed | 2456 | |
0eae88f3 ED |
2457 | seq_printf(seq, "%08X %08X %-3hd", |
2458 | (__force u32) mfc->mfc_mcastgrp, | |
2459 | (__force u32) mfc->mfc_origin, | |
1ea472e2 | 2460 | mfc->mfc_parent); |
1da177e4 | 2461 | |
0c12295a | 2462 | if (it->cache != &mrt->mfc_unres_queue) { |
1ea472e2 BT |
2463 | seq_printf(seq, " %8lu %8lu %8lu", |
2464 | mfc->mfc_un.res.pkt, | |
2465 | mfc->mfc_un.res.bytes, | |
2466 | mfc->mfc_un.res.wrong_if); | |
132adf54 | 2467 | for (n = mfc->mfc_un.res.minvif; |
a8cb16dd | 2468 | n < mfc->mfc_un.res.maxvif; n++) { |
0c12295a | 2469 | if (VIF_EXISTS(mrt, n) && |
cf958ae3 BT |
2470 | mfc->mfc_un.res.ttls[n] < 255) |
2471 | seq_printf(seq, | |
e905a9ed | 2472 | " %2d:%-3d", |
1da177e4 LT |
2473 | n, mfc->mfc_un.res.ttls[n]); |
2474 | } | |
1ea472e2 BT |
2475 | } else { |
2476 | /* unresolved mfc_caches don't contain | |
2477 | * pkt, bytes and wrong_if values | |
2478 | */ | |
2479 | seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul); | |
1da177e4 LT |
2480 | } |
2481 | seq_putc(seq, '\n'); | |
2482 | } | |
2483 | return 0; | |
2484 | } | |
2485 | ||
f690808e | 2486 | static const struct seq_operations ipmr_mfc_seq_ops = { |
1da177e4 LT |
2487 | .start = ipmr_mfc_seq_start, |
2488 | .next = ipmr_mfc_seq_next, | |
2489 | .stop = ipmr_mfc_seq_stop, | |
2490 | .show = ipmr_mfc_seq_show, | |
2491 | }; | |
2492 | ||
2493 | static int ipmr_mfc_open(struct inode *inode, struct file *file) | |
2494 | { | |
f6bb4514 BT |
2495 | return seq_open_net(inode, file, &ipmr_mfc_seq_ops, |
2496 | sizeof(struct ipmr_mfc_iter)); | |
1da177e4 LT |
2497 | } |
2498 | ||
9a32144e | 2499 | static const struct file_operations ipmr_mfc_fops = { |
1da177e4 LT |
2500 | .owner = THIS_MODULE, |
2501 | .open = ipmr_mfc_open, | |
2502 | .read = seq_read, | |
2503 | .llseek = seq_lseek, | |
f6bb4514 | 2504 | .release = seq_release_net, |
1da177e4 | 2505 | }; |
e905a9ed | 2506 | #endif |
1da177e4 LT |
2507 | |
2508 | #ifdef CONFIG_IP_PIMSM_V2 | |
32613090 | 2509 | static const struct net_protocol pim_protocol = { |
1da177e4 | 2510 | .handler = pim_rcv, |
403dbb97 | 2511 | .netns_ok = 1, |
1da177e4 LT |
2512 | }; |
2513 | #endif | |
2514 | ||
2515 | ||
2516 | /* | |
2517 | * Setup for IP multicast routing | |
2518 | */ | |
cf958ae3 BT |
2519 | static int __net_init ipmr_net_init(struct net *net) |
2520 | { | |
f0ad0860 | 2521 | int err; |
cf958ae3 | 2522 | |
f0ad0860 PM |
2523 | err = ipmr_rules_init(net); |
2524 | if (err < 0) | |
cf958ae3 | 2525 | goto fail; |
f6bb4514 BT |
2526 | |
2527 | #ifdef CONFIG_PROC_FS | |
2528 | err = -ENOMEM; | |
2529 | if (!proc_net_fops_create(net, "ip_mr_vif", 0, &ipmr_vif_fops)) | |
2530 | goto proc_vif_fail; | |
2531 | if (!proc_net_fops_create(net, "ip_mr_cache", 0, &ipmr_mfc_fops)) | |
2532 | goto proc_cache_fail; | |
2533 | #endif | |
2bb8b26c BT |
2534 | return 0; |
2535 | ||
f6bb4514 BT |
2536 | #ifdef CONFIG_PROC_FS |
2537 | proc_cache_fail: | |
2538 | proc_net_remove(net, "ip_mr_vif"); | |
2539 | proc_vif_fail: | |
f0ad0860 | 2540 | ipmr_rules_exit(net); |
f6bb4514 | 2541 | #endif |
cf958ae3 BT |
2542 | fail: |
2543 | return err; | |
2544 | } | |
2545 | ||
2546 | static void __net_exit ipmr_net_exit(struct net *net) | |
2547 | { | |
f6bb4514 BT |
2548 | #ifdef CONFIG_PROC_FS |
2549 | proc_net_remove(net, "ip_mr_cache"); | |
2550 | proc_net_remove(net, "ip_mr_vif"); | |
2551 | #endif | |
f0ad0860 | 2552 | ipmr_rules_exit(net); |
cf958ae3 BT |
2553 | } |
2554 | ||
2555 | static struct pernet_operations ipmr_net_ops = { | |
2556 | .init = ipmr_net_init, | |
2557 | .exit = ipmr_net_exit, | |
2558 | }; | |
e905a9ed | 2559 | |
03d2f897 | 2560 | int __init ip_mr_init(void) |
1da177e4 | 2561 | { |
03d2f897 WC |
2562 | int err; |
2563 | ||
1da177e4 LT |
2564 | mrt_cachep = kmem_cache_create("ip_mrt_cache", |
2565 | sizeof(struct mfc_cache), | |
a8c9486b | 2566 | 0, SLAB_HWCACHE_ALIGN | SLAB_PANIC, |
20c2df83 | 2567 | NULL); |
03d2f897 WC |
2568 | if (!mrt_cachep) |
2569 | return -ENOMEM; | |
2570 | ||
cf958ae3 BT |
2571 | err = register_pernet_subsys(&ipmr_net_ops); |
2572 | if (err) | |
2573 | goto reg_pernet_fail; | |
2574 | ||
03d2f897 WC |
2575 | err = register_netdevice_notifier(&ip_mr_notifier); |
2576 | if (err) | |
2577 | goto reg_notif_fail; | |
403dbb97 TG |
2578 | #ifdef CONFIG_IP_PIMSM_V2 |
2579 | if (inet_add_protocol(&pim_protocol, IPPROTO_PIM) < 0) { | |
058bd4d2 | 2580 | pr_err("%s: can't add PIM protocol\n", __func__); |
403dbb97 TG |
2581 | err = -EAGAIN; |
2582 | goto add_proto_fail; | |
2583 | } | |
2584 | #endif | |
c7ac8679 GR |
2585 | rtnl_register(RTNL_FAMILY_IPMR, RTM_GETROUTE, |
2586 | NULL, ipmr_rtm_dumproute, NULL); | |
03d2f897 | 2587 | return 0; |
f6bb4514 | 2588 | |
403dbb97 TG |
2589 | #ifdef CONFIG_IP_PIMSM_V2 |
2590 | add_proto_fail: | |
2591 | unregister_netdevice_notifier(&ip_mr_notifier); | |
2592 | #endif | |
c3e38896 | 2593 | reg_notif_fail: |
cf958ae3 BT |
2594 | unregister_pernet_subsys(&ipmr_net_ops); |
2595 | reg_pernet_fail: | |
c3e38896 | 2596 | kmem_cache_destroy(mrt_cachep); |
03d2f897 | 2597 | return err; |
1da177e4 | 2598 | } |