]>
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 | |
25 | * Relax this requrement to work with older peers. | |
26 | * | |
27 | */ | |
28 | ||
1da177e4 LT |
29 | #include <asm/system.h> |
30 | #include <asm/uaccess.h> | |
31 | #include <linux/types.h> | |
4fc268d2 | 32 | #include <linux/capability.h> |
1da177e4 LT |
33 | #include <linux/errno.h> |
34 | #include <linux/timer.h> | |
35 | #include <linux/mm.h> | |
36 | #include <linux/kernel.h> | |
37 | #include <linux/fcntl.h> | |
38 | #include <linux/stat.h> | |
39 | #include <linux/socket.h> | |
40 | #include <linux/in.h> | |
41 | #include <linux/inet.h> | |
42 | #include <linux/netdevice.h> | |
43 | #include <linux/inetdevice.h> | |
44 | #include <linux/igmp.h> | |
45 | #include <linux/proc_fs.h> | |
46 | #include <linux/seq_file.h> | |
47 | #include <linux/mroute.h> | |
48 | #include <linux/init.h> | |
46f25dff | 49 | #include <linux/if_ether.h> |
5a0e3ad6 | 50 | #include <linux/slab.h> |
457c4cbc | 51 | #include <net/net_namespace.h> |
1da177e4 LT |
52 | #include <net/ip.h> |
53 | #include <net/protocol.h> | |
54 | #include <linux/skbuff.h> | |
14c85021 | 55 | #include <net/route.h> |
1da177e4 LT |
56 | #include <net/sock.h> |
57 | #include <net/icmp.h> | |
58 | #include <net/udp.h> | |
59 | #include <net/raw.h> | |
60 | #include <linux/notifier.h> | |
61 | #include <linux/if_arp.h> | |
62 | #include <linux/netfilter_ipv4.h> | |
63 | #include <net/ipip.h> | |
64 | #include <net/checksum.h> | |
dc5fc579 | 65 | #include <net/netlink.h> |
1da177e4 LT |
66 | |
67 | #if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2) | |
68 | #define CONFIG_IP_PIMSM 1 | |
69 | #endif | |
70 | ||
1da177e4 LT |
71 | /* Big lock, protecting vif table, mrt cache and mroute socket state. |
72 | Note that the changes are semaphored via rtnl_lock. | |
73 | */ | |
74 | ||
75 | static DEFINE_RWLOCK(mrt_lock); | |
76 | ||
77 | /* | |
78 | * Multicast router control variables | |
79 | */ | |
80 | ||
cf958ae3 | 81 | #define VIF_EXISTS(_net, _idx) ((_net)->ipv4.vif_table[_idx].dev != NULL) |
1da177e4 | 82 | |
1da177e4 LT |
83 | /* Special spinlock for queue of unresolved entries */ |
84 | static DEFINE_SPINLOCK(mfc_unres_lock); | |
85 | ||
86 | /* We return to original Alan's scheme. Hash table of resolved | |
87 | entries is changed only in process context and protected | |
88 | with weak lock mrt_lock. Queue of unresolved entries is protected | |
89 | with strong spinlock mfc_unres_lock. | |
90 | ||
91 | In this case data path is free of exclusive locks at all. | |
92 | */ | |
93 | ||
e18b890b | 94 | static struct kmem_cache *mrt_cachep __read_mostly; |
1da177e4 | 95 | |
d658f8a0 PM |
96 | static int ip_mr_forward(struct net *net, struct sk_buff *skb, |
97 | struct mfc_cache *cache, int local); | |
4feb88e5 BT |
98 | static int ipmr_cache_report(struct net *net, |
99 | struct sk_buff *pkt, vifi_t vifi, int assert); | |
d658f8a0 PM |
100 | static int ipmr_fill_mroute(struct net *net, struct sk_buff *skb, |
101 | struct mfc_cache *c, struct rtmsg *rtm); | |
1da177e4 | 102 | |
1da177e4 LT |
103 | /* Service routines creating virtual interfaces: DVMRP tunnels and PIMREG */ |
104 | ||
d607032d WC |
105 | static void ipmr_del_tunnel(struct net_device *dev, struct vifctl *v) |
106 | { | |
4feb88e5 BT |
107 | struct net *net = dev_net(dev); |
108 | ||
d607032d WC |
109 | dev_close(dev); |
110 | ||
4feb88e5 | 111 | dev = __dev_get_by_name(net, "tunl0"); |
d607032d | 112 | if (dev) { |
5bc3eb7e | 113 | const struct net_device_ops *ops = dev->netdev_ops; |
d607032d | 114 | struct ifreq ifr; |
d607032d WC |
115 | struct ip_tunnel_parm p; |
116 | ||
117 | memset(&p, 0, sizeof(p)); | |
118 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
119 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
120 | p.iph.version = 4; | |
121 | p.iph.ihl = 5; | |
122 | p.iph.protocol = IPPROTO_IPIP; | |
123 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
124 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; | |
125 | ||
5bc3eb7e SH |
126 | if (ops->ndo_do_ioctl) { |
127 | mm_segment_t oldfs = get_fs(); | |
128 | ||
129 | set_fs(KERNEL_DS); | |
130 | ops->ndo_do_ioctl(dev, &ifr, SIOCDELTUNNEL); | |
131 | set_fs(oldfs); | |
132 | } | |
d607032d WC |
133 | } |
134 | } | |
135 | ||
1da177e4 | 136 | static |
4feb88e5 | 137 | struct net_device *ipmr_new_tunnel(struct net *net, struct vifctl *v) |
1da177e4 LT |
138 | { |
139 | struct net_device *dev; | |
140 | ||
4feb88e5 | 141 | dev = __dev_get_by_name(net, "tunl0"); |
1da177e4 LT |
142 | |
143 | if (dev) { | |
5bc3eb7e | 144 | const struct net_device_ops *ops = dev->netdev_ops; |
1da177e4 LT |
145 | int err; |
146 | struct ifreq ifr; | |
1da177e4 LT |
147 | struct ip_tunnel_parm p; |
148 | struct in_device *in_dev; | |
149 | ||
150 | memset(&p, 0, sizeof(p)); | |
151 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
152 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
153 | p.iph.version = 4; | |
154 | p.iph.ihl = 5; | |
155 | p.iph.protocol = IPPROTO_IPIP; | |
156 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
ba93ef74 | 157 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; |
1da177e4 | 158 | |
5bc3eb7e SH |
159 | if (ops->ndo_do_ioctl) { |
160 | mm_segment_t oldfs = get_fs(); | |
161 | ||
162 | set_fs(KERNEL_DS); | |
163 | err = ops->ndo_do_ioctl(dev, &ifr, SIOCADDTUNNEL); | |
164 | set_fs(oldfs); | |
165 | } else | |
166 | err = -EOPNOTSUPP; | |
1da177e4 LT |
167 | |
168 | dev = NULL; | |
169 | ||
4feb88e5 BT |
170 | if (err == 0 && |
171 | (dev = __dev_get_by_name(net, p.name)) != NULL) { | |
1da177e4 LT |
172 | dev->flags |= IFF_MULTICAST; |
173 | ||
e5ed6399 | 174 | in_dev = __in_dev_get_rtnl(dev); |
71e27da9 | 175 | if (in_dev == NULL) |
1da177e4 | 176 | goto failure; |
71e27da9 HX |
177 | |
178 | ipv4_devconf_setall(in_dev); | |
179 | IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0; | |
1da177e4 LT |
180 | |
181 | if (dev_open(dev)) | |
182 | goto failure; | |
7dc00c82 | 183 | dev_hold(dev); |
1da177e4 LT |
184 | } |
185 | } | |
186 | return dev; | |
187 | ||
188 | failure: | |
189 | /* allow the register to be completed before unregistering. */ | |
190 | rtnl_unlock(); | |
191 | rtnl_lock(); | |
192 | ||
193 | unregister_netdevice(dev); | |
194 | return NULL; | |
195 | } | |
196 | ||
197 | #ifdef CONFIG_IP_PIMSM | |
198 | ||
6fef4c0c | 199 | static netdev_tx_t reg_vif_xmit(struct sk_buff *skb, struct net_device *dev) |
1da177e4 | 200 | { |
4feb88e5 BT |
201 | struct net *net = dev_net(dev); |
202 | ||
1da177e4 | 203 | read_lock(&mrt_lock); |
cf3677ae PE |
204 | dev->stats.tx_bytes += skb->len; |
205 | dev->stats.tx_packets++; | |
4feb88e5 BT |
206 | ipmr_cache_report(net, skb, net->ipv4.mroute_reg_vif_num, |
207 | IGMPMSG_WHOLEPKT); | |
1da177e4 LT |
208 | read_unlock(&mrt_lock); |
209 | kfree_skb(skb); | |
6ed10654 | 210 | return NETDEV_TX_OK; |
1da177e4 LT |
211 | } |
212 | ||
007c3838 SH |
213 | static const struct net_device_ops reg_vif_netdev_ops = { |
214 | .ndo_start_xmit = reg_vif_xmit, | |
215 | }; | |
216 | ||
1da177e4 LT |
217 | static void reg_vif_setup(struct net_device *dev) |
218 | { | |
219 | dev->type = ARPHRD_PIMREG; | |
46f25dff | 220 | dev->mtu = ETH_DATA_LEN - sizeof(struct iphdr) - 8; |
1da177e4 | 221 | dev->flags = IFF_NOARP; |
007c3838 | 222 | dev->netdev_ops = ®_vif_netdev_ops, |
1da177e4 | 223 | dev->destructor = free_netdev; |
403dbb97 | 224 | dev->features |= NETIF_F_NETNS_LOCAL; |
1da177e4 LT |
225 | } |
226 | ||
403dbb97 | 227 | static struct net_device *ipmr_reg_vif(struct net *net) |
1da177e4 LT |
228 | { |
229 | struct net_device *dev; | |
230 | struct in_device *in_dev; | |
231 | ||
cf3677ae | 232 | dev = alloc_netdev(0, "pimreg", reg_vif_setup); |
1da177e4 LT |
233 | |
234 | if (dev == NULL) | |
235 | return NULL; | |
236 | ||
403dbb97 TG |
237 | dev_net_set(dev, net); |
238 | ||
1da177e4 LT |
239 | if (register_netdevice(dev)) { |
240 | free_netdev(dev); | |
241 | return NULL; | |
242 | } | |
243 | dev->iflink = 0; | |
244 | ||
71e27da9 HX |
245 | rcu_read_lock(); |
246 | if ((in_dev = __in_dev_get_rcu(dev)) == NULL) { | |
247 | rcu_read_unlock(); | |
1da177e4 | 248 | goto failure; |
71e27da9 | 249 | } |
1da177e4 | 250 | |
71e27da9 HX |
251 | ipv4_devconf_setall(in_dev); |
252 | IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0; | |
253 | rcu_read_unlock(); | |
1da177e4 LT |
254 | |
255 | if (dev_open(dev)) | |
256 | goto failure; | |
257 | ||
7dc00c82 WC |
258 | dev_hold(dev); |
259 | ||
1da177e4 LT |
260 | return dev; |
261 | ||
262 | failure: | |
263 | /* allow the register to be completed before unregistering. */ | |
264 | rtnl_unlock(); | |
265 | rtnl_lock(); | |
266 | ||
267 | unregister_netdevice(dev); | |
268 | return NULL; | |
269 | } | |
270 | #endif | |
271 | ||
272 | /* | |
273 | * Delete a VIF entry | |
7dc00c82 | 274 | * @notify: Set to 1, if the caller is a notifier_call |
1da177e4 | 275 | */ |
e905a9ed | 276 | |
d17fa6fa ED |
277 | static int vif_delete(struct net *net, int vifi, int notify, |
278 | struct list_head *head) | |
1da177e4 LT |
279 | { |
280 | struct vif_device *v; | |
281 | struct net_device *dev; | |
282 | struct in_device *in_dev; | |
283 | ||
4feb88e5 | 284 | if (vifi < 0 || vifi >= net->ipv4.maxvif) |
1da177e4 LT |
285 | return -EADDRNOTAVAIL; |
286 | ||
4feb88e5 | 287 | v = &net->ipv4.vif_table[vifi]; |
1da177e4 LT |
288 | |
289 | write_lock_bh(&mrt_lock); | |
290 | dev = v->dev; | |
291 | v->dev = NULL; | |
292 | ||
293 | if (!dev) { | |
294 | write_unlock_bh(&mrt_lock); | |
295 | return -EADDRNOTAVAIL; | |
296 | } | |
297 | ||
298 | #ifdef CONFIG_IP_PIMSM | |
4feb88e5 BT |
299 | if (vifi == net->ipv4.mroute_reg_vif_num) |
300 | net->ipv4.mroute_reg_vif_num = -1; | |
1da177e4 LT |
301 | #endif |
302 | ||
4feb88e5 | 303 | if (vifi+1 == net->ipv4.maxvif) { |
1da177e4 LT |
304 | int tmp; |
305 | for (tmp=vifi-1; tmp>=0; tmp--) { | |
4feb88e5 | 306 | if (VIF_EXISTS(net, tmp)) |
1da177e4 LT |
307 | break; |
308 | } | |
4feb88e5 | 309 | net->ipv4.maxvif = tmp+1; |
1da177e4 LT |
310 | } |
311 | ||
312 | write_unlock_bh(&mrt_lock); | |
313 | ||
314 | dev_set_allmulti(dev, -1); | |
315 | ||
e5ed6399 | 316 | if ((in_dev = __in_dev_get_rtnl(dev)) != NULL) { |
42f811b8 | 317 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)--; |
1da177e4 LT |
318 | ip_rt_multicast_event(in_dev); |
319 | } | |
320 | ||
7dc00c82 | 321 | if (v->flags&(VIFF_TUNNEL|VIFF_REGISTER) && !notify) |
d17fa6fa | 322 | unregister_netdevice_queue(dev, head); |
1da177e4 LT |
323 | |
324 | dev_put(dev); | |
325 | return 0; | |
326 | } | |
327 | ||
5c0a66f5 BT |
328 | static inline void ipmr_cache_free(struct mfc_cache *c) |
329 | { | |
5c0a66f5 BT |
330 | kmem_cache_free(mrt_cachep, c); |
331 | } | |
332 | ||
1da177e4 LT |
333 | /* Destroy an unresolved cache entry, killing queued skbs |
334 | and reporting error to netlink readers. | |
335 | */ | |
336 | ||
d658f8a0 | 337 | static void ipmr_destroy_unres(struct net *net, struct mfc_cache *c) |
1da177e4 LT |
338 | { |
339 | struct sk_buff *skb; | |
9ef1d4c7 | 340 | struct nlmsgerr *e; |
1da177e4 | 341 | |
4feb88e5 | 342 | atomic_dec(&net->ipv4.cache_resolve_queue_len); |
1da177e4 | 343 | |
c354e124 | 344 | while ((skb = skb_dequeue(&c->mfc_un.unres.unresolved))) { |
eddc9ec5 | 345 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
346 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
347 | nlh->nlmsg_type = NLMSG_ERROR; | |
348 | nlh->nlmsg_len = NLMSG_LENGTH(sizeof(struct nlmsgerr)); | |
349 | skb_trim(skb, nlh->nlmsg_len); | |
9ef1d4c7 PM |
350 | e = NLMSG_DATA(nlh); |
351 | e->error = -ETIMEDOUT; | |
352 | memset(&e->msg, 0, sizeof(e->msg)); | |
2942e900 | 353 | |
4feb88e5 | 354 | rtnl_unicast(skb, net, NETLINK_CB(skb).pid); |
1da177e4 LT |
355 | } else |
356 | kfree_skb(skb); | |
357 | } | |
358 | ||
5c0a66f5 | 359 | ipmr_cache_free(c); |
1da177e4 LT |
360 | } |
361 | ||
362 | ||
e258beb2 | 363 | /* Timer process for the unresolved queue. */ |
1da177e4 | 364 | |
e258beb2 | 365 | static void ipmr_expire_process(unsigned long arg) |
1da177e4 | 366 | { |
e258beb2 | 367 | struct net *net = (struct net *)arg; |
1da177e4 LT |
368 | unsigned long now; |
369 | unsigned long expires; | |
370 | struct mfc_cache *c, **cp; | |
371 | ||
372 | if (!spin_trylock(&mfc_unres_lock)) { | |
e258beb2 | 373 | mod_timer(&net->ipv4.ipmr_expire_timer, jiffies+HZ/10); |
1da177e4 LT |
374 | return; |
375 | } | |
376 | ||
e258beb2 | 377 | if (net->ipv4.mfc_unres_queue == NULL) |
1da177e4 LT |
378 | goto out; |
379 | ||
380 | now = jiffies; | |
381 | expires = 10*HZ; | |
e258beb2 | 382 | cp = &net->ipv4.mfc_unres_queue; |
1da177e4 LT |
383 | |
384 | while ((c=*cp) != NULL) { | |
385 | if (time_after(c->mfc_un.unres.expires, now)) { | |
386 | unsigned long interval = c->mfc_un.unres.expires - now; | |
387 | if (interval < expires) | |
388 | expires = interval; | |
389 | cp = &c->next; | |
390 | continue; | |
391 | } | |
392 | ||
393 | *cp = c->next; | |
394 | ||
d658f8a0 | 395 | ipmr_destroy_unres(net, c); |
1da177e4 LT |
396 | } |
397 | ||
e258beb2 PM |
398 | if (net->ipv4.mfc_unres_queue != NULL) |
399 | mod_timer(&net->ipv4.ipmr_expire_timer, jiffies + expires); | |
1da177e4 LT |
400 | |
401 | out: | |
402 | spin_unlock(&mfc_unres_lock); | |
403 | } | |
404 | ||
405 | /* Fill oifs list. It is called under write locked mrt_lock. */ | |
406 | ||
d658f8a0 PM |
407 | static void ipmr_update_thresholds(struct net *net, struct mfc_cache *cache, |
408 | unsigned char *ttls) | |
1da177e4 LT |
409 | { |
410 | int vifi; | |
411 | ||
412 | cache->mfc_un.res.minvif = MAXVIFS; | |
413 | cache->mfc_un.res.maxvif = 0; | |
414 | memset(cache->mfc_un.res.ttls, 255, MAXVIFS); | |
415 | ||
4feb88e5 BT |
416 | for (vifi = 0; vifi < net->ipv4.maxvif; vifi++) { |
417 | if (VIF_EXISTS(net, vifi) && | |
cf958ae3 | 418 | ttls[vifi] && ttls[vifi] < 255) { |
1da177e4 LT |
419 | cache->mfc_un.res.ttls[vifi] = ttls[vifi]; |
420 | if (cache->mfc_un.res.minvif > vifi) | |
421 | cache->mfc_un.res.minvif = vifi; | |
422 | if (cache->mfc_un.res.maxvif <= vifi) | |
423 | cache->mfc_un.res.maxvif = vifi + 1; | |
424 | } | |
425 | } | |
426 | } | |
427 | ||
4feb88e5 | 428 | static int vif_add(struct net *net, struct vifctl *vifc, int mrtsock) |
1da177e4 LT |
429 | { |
430 | int vifi = vifc->vifc_vifi; | |
4feb88e5 | 431 | struct vif_device *v = &net->ipv4.vif_table[vifi]; |
1da177e4 LT |
432 | struct net_device *dev; |
433 | struct in_device *in_dev; | |
d607032d | 434 | int err; |
1da177e4 LT |
435 | |
436 | /* Is vif busy ? */ | |
4feb88e5 | 437 | if (VIF_EXISTS(net, vifi)) |
1da177e4 LT |
438 | return -EADDRINUSE; |
439 | ||
440 | switch (vifc->vifc_flags) { | |
441 | #ifdef CONFIG_IP_PIMSM | |
442 | case VIFF_REGISTER: | |
443 | /* | |
444 | * Special Purpose VIF in PIM | |
445 | * All the packets will be sent to the daemon | |
446 | */ | |
4feb88e5 | 447 | if (net->ipv4.mroute_reg_vif_num >= 0) |
1da177e4 | 448 | return -EADDRINUSE; |
403dbb97 | 449 | dev = ipmr_reg_vif(net); |
1da177e4 LT |
450 | if (!dev) |
451 | return -ENOBUFS; | |
d607032d WC |
452 | err = dev_set_allmulti(dev, 1); |
453 | if (err) { | |
454 | unregister_netdevice(dev); | |
7dc00c82 | 455 | dev_put(dev); |
d607032d WC |
456 | return err; |
457 | } | |
1da177e4 LT |
458 | break; |
459 | #endif | |
e905a9ed | 460 | case VIFF_TUNNEL: |
4feb88e5 | 461 | dev = ipmr_new_tunnel(net, vifc); |
1da177e4 LT |
462 | if (!dev) |
463 | return -ENOBUFS; | |
d607032d WC |
464 | err = dev_set_allmulti(dev, 1); |
465 | if (err) { | |
466 | ipmr_del_tunnel(dev, vifc); | |
7dc00c82 | 467 | dev_put(dev); |
d607032d WC |
468 | return err; |
469 | } | |
1da177e4 | 470 | break; |
ee5e81f0 I |
471 | |
472 | case VIFF_USE_IFINDEX: | |
1da177e4 | 473 | case 0: |
ee5e81f0 I |
474 | if (vifc->vifc_flags == VIFF_USE_IFINDEX) { |
475 | dev = dev_get_by_index(net, vifc->vifc_lcl_ifindex); | |
476 | if (dev && dev->ip_ptr == NULL) { | |
477 | dev_put(dev); | |
478 | return -EADDRNOTAVAIL; | |
479 | } | |
480 | } else | |
481 | dev = ip_dev_find(net, vifc->vifc_lcl_addr.s_addr); | |
482 | ||
1da177e4 LT |
483 | if (!dev) |
484 | return -EADDRNOTAVAIL; | |
d607032d | 485 | err = dev_set_allmulti(dev, 1); |
7dc00c82 WC |
486 | if (err) { |
487 | dev_put(dev); | |
d607032d | 488 | return err; |
7dc00c82 | 489 | } |
1da177e4 LT |
490 | break; |
491 | default: | |
492 | return -EINVAL; | |
493 | } | |
494 | ||
d0490cfd DC |
495 | if ((in_dev = __in_dev_get_rtnl(dev)) == NULL) { |
496 | dev_put(dev); | |
1da177e4 | 497 | return -EADDRNOTAVAIL; |
d0490cfd | 498 | } |
42f811b8 | 499 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)++; |
1da177e4 LT |
500 | ip_rt_multicast_event(in_dev); |
501 | ||
502 | /* | |
503 | * Fill in the VIF structures | |
504 | */ | |
c354e124 JK |
505 | v->rate_limit = vifc->vifc_rate_limit; |
506 | v->local = vifc->vifc_lcl_addr.s_addr; | |
507 | v->remote = vifc->vifc_rmt_addr.s_addr; | |
508 | v->flags = vifc->vifc_flags; | |
1da177e4 LT |
509 | if (!mrtsock) |
510 | v->flags |= VIFF_STATIC; | |
c354e124 | 511 | v->threshold = vifc->vifc_threshold; |
1da177e4 LT |
512 | v->bytes_in = 0; |
513 | v->bytes_out = 0; | |
514 | v->pkt_in = 0; | |
515 | v->pkt_out = 0; | |
516 | v->link = dev->ifindex; | |
517 | if (v->flags&(VIFF_TUNNEL|VIFF_REGISTER)) | |
518 | v->link = dev->iflink; | |
519 | ||
520 | /* And finish update writing critical data */ | |
521 | write_lock_bh(&mrt_lock); | |
c354e124 | 522 | v->dev = dev; |
1da177e4 LT |
523 | #ifdef CONFIG_IP_PIMSM |
524 | if (v->flags&VIFF_REGISTER) | |
4feb88e5 | 525 | net->ipv4.mroute_reg_vif_num = vifi; |
1da177e4 | 526 | #endif |
4feb88e5 BT |
527 | if (vifi+1 > net->ipv4.maxvif) |
528 | net->ipv4.maxvif = vifi+1; | |
1da177e4 LT |
529 | write_unlock_bh(&mrt_lock); |
530 | return 0; | |
531 | } | |
532 | ||
4feb88e5 BT |
533 | static struct mfc_cache *ipmr_cache_find(struct net *net, |
534 | __be32 origin, | |
535 | __be32 mcastgrp) | |
1da177e4 | 536 | { |
c354e124 | 537 | int line = MFC_HASH(mcastgrp, origin); |
1da177e4 LT |
538 | struct mfc_cache *c; |
539 | ||
4feb88e5 | 540 | for (c = net->ipv4.mfc_cache_array[line]; c; c = c->next) { |
1da177e4 LT |
541 | if (c->mfc_origin==origin && c->mfc_mcastgrp==mcastgrp) |
542 | break; | |
543 | } | |
544 | return c; | |
545 | } | |
546 | ||
547 | /* | |
548 | * Allocate a multicast cache entry | |
549 | */ | |
d658f8a0 | 550 | static struct mfc_cache *ipmr_cache_alloc(void) |
1da177e4 | 551 | { |
c354e124 JK |
552 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL); |
553 | if (c == NULL) | |
1da177e4 | 554 | return NULL; |
1da177e4 LT |
555 | c->mfc_un.res.minvif = MAXVIFS; |
556 | return c; | |
557 | } | |
558 | ||
d658f8a0 | 559 | static struct mfc_cache *ipmr_cache_alloc_unres(void) |
1da177e4 | 560 | { |
c354e124 JK |
561 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC); |
562 | if (c == NULL) | |
1da177e4 | 563 | return NULL; |
1da177e4 LT |
564 | skb_queue_head_init(&c->mfc_un.unres.unresolved); |
565 | c->mfc_un.unres.expires = jiffies + 10*HZ; | |
566 | return c; | |
567 | } | |
568 | ||
569 | /* | |
570 | * A cache entry has gone into a resolved state from queued | |
571 | */ | |
e905a9ed | 572 | |
d658f8a0 PM |
573 | static void ipmr_cache_resolve(struct net *net, struct mfc_cache *uc, |
574 | struct mfc_cache *c) | |
1da177e4 LT |
575 | { |
576 | struct sk_buff *skb; | |
9ef1d4c7 | 577 | struct nlmsgerr *e; |
1da177e4 LT |
578 | |
579 | /* | |
580 | * Play the pending entries through our router | |
581 | */ | |
582 | ||
c354e124 | 583 | while ((skb = __skb_dequeue(&uc->mfc_un.unres.unresolved))) { |
eddc9ec5 | 584 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
585 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
586 | ||
d658f8a0 | 587 | if (ipmr_fill_mroute(net, skb, c, NLMSG_DATA(nlh)) > 0) { |
27a884dc ACM |
588 | nlh->nlmsg_len = (skb_tail_pointer(skb) - |
589 | (u8 *)nlh); | |
1da177e4 LT |
590 | } else { |
591 | nlh->nlmsg_type = NLMSG_ERROR; | |
592 | nlh->nlmsg_len = NLMSG_LENGTH(sizeof(struct nlmsgerr)); | |
593 | skb_trim(skb, nlh->nlmsg_len); | |
9ef1d4c7 PM |
594 | e = NLMSG_DATA(nlh); |
595 | e->error = -EMSGSIZE; | |
596 | memset(&e->msg, 0, sizeof(e->msg)); | |
1da177e4 | 597 | } |
2942e900 | 598 | |
d658f8a0 | 599 | rtnl_unicast(skb, net, NETLINK_CB(skb).pid); |
1da177e4 | 600 | } else |
d658f8a0 | 601 | ip_mr_forward(net, skb, c, 0); |
1da177e4 LT |
602 | } |
603 | } | |
604 | ||
605 | /* | |
606 | * Bounce a cache query up to mrouted. We could use netlink for this but mrouted | |
607 | * expects the following bizarre scheme. | |
608 | * | |
609 | * Called under mrt_lock. | |
610 | */ | |
e905a9ed | 611 | |
4feb88e5 BT |
612 | static int ipmr_cache_report(struct net *net, |
613 | struct sk_buff *pkt, vifi_t vifi, int assert) | |
1da177e4 LT |
614 | { |
615 | struct sk_buff *skb; | |
c9bdd4b5 | 616 | const int ihl = ip_hdrlen(pkt); |
1da177e4 LT |
617 | struct igmphdr *igmp; |
618 | struct igmpmsg *msg; | |
619 | int ret; | |
620 | ||
621 | #ifdef CONFIG_IP_PIMSM | |
622 | if (assert == IGMPMSG_WHOLEPKT) | |
623 | skb = skb_realloc_headroom(pkt, sizeof(struct iphdr)); | |
624 | else | |
625 | #endif | |
626 | skb = alloc_skb(128, GFP_ATOMIC); | |
627 | ||
132adf54 | 628 | if (!skb) |
1da177e4 LT |
629 | return -ENOBUFS; |
630 | ||
631 | #ifdef CONFIG_IP_PIMSM | |
632 | if (assert == IGMPMSG_WHOLEPKT) { | |
633 | /* Ugly, but we have no choice with this interface. | |
634 | Duplicate old header, fix ihl, length etc. | |
635 | And all this only to mangle msg->im_msgtype and | |
636 | to set msg->im_mbz to "mbz" :-) | |
637 | */ | |
878c8145 ACM |
638 | skb_push(skb, sizeof(struct iphdr)); |
639 | skb_reset_network_header(skb); | |
badff6d0 | 640 | skb_reset_transport_header(skb); |
0272ffc4 | 641 | msg = (struct igmpmsg *)skb_network_header(skb); |
d56f90a7 | 642 | memcpy(msg, skb_network_header(pkt), sizeof(struct iphdr)); |
1da177e4 LT |
643 | msg->im_msgtype = IGMPMSG_WHOLEPKT; |
644 | msg->im_mbz = 0; | |
4feb88e5 | 645 | msg->im_vif = net->ipv4.mroute_reg_vif_num; |
eddc9ec5 ACM |
646 | ip_hdr(skb)->ihl = sizeof(struct iphdr) >> 2; |
647 | ip_hdr(skb)->tot_len = htons(ntohs(ip_hdr(pkt)->tot_len) + | |
648 | sizeof(struct iphdr)); | |
e905a9ed | 649 | } else |
1da177e4 | 650 | #endif |
e905a9ed YH |
651 | { |
652 | ||
1da177e4 LT |
653 | /* |
654 | * Copy the IP header | |
655 | */ | |
656 | ||
27a884dc | 657 | skb->network_header = skb->tail; |
ddc7b8e3 | 658 | skb_put(skb, ihl); |
27d7ff46 | 659 | skb_copy_to_linear_data(skb, pkt->data, ihl); |
eddc9ec5 ACM |
660 | ip_hdr(skb)->protocol = 0; /* Flag to the kernel this is a route add */ |
661 | msg = (struct igmpmsg *)skb_network_header(skb); | |
1da177e4 | 662 | msg->im_vif = vifi; |
adf30907 | 663 | skb_dst_set(skb, dst_clone(skb_dst(pkt))); |
1da177e4 LT |
664 | |
665 | /* | |
666 | * Add our header | |
667 | */ | |
668 | ||
c354e124 | 669 | igmp=(struct igmphdr *)skb_put(skb, sizeof(struct igmphdr)); |
1da177e4 LT |
670 | igmp->type = |
671 | msg->im_msgtype = assert; | |
672 | igmp->code = 0; | |
eddc9ec5 | 673 | ip_hdr(skb)->tot_len = htons(skb->len); /* Fix the length */ |
b0e380b1 | 674 | skb->transport_header = skb->network_header; |
e905a9ed | 675 | } |
1da177e4 | 676 | |
4feb88e5 | 677 | if (net->ipv4.mroute_sk == NULL) { |
1da177e4 LT |
678 | kfree_skb(skb); |
679 | return -EINVAL; | |
680 | } | |
681 | ||
682 | /* | |
683 | * Deliver to mrouted | |
684 | */ | |
4feb88e5 | 685 | ret = sock_queue_rcv_skb(net->ipv4.mroute_sk, skb); |
70a269e6 | 686 | if (ret < 0) { |
1da177e4 LT |
687 | if (net_ratelimit()) |
688 | printk(KERN_WARNING "mroute: pending queue full, dropping entries.\n"); | |
689 | kfree_skb(skb); | |
690 | } | |
691 | ||
692 | return ret; | |
693 | } | |
694 | ||
695 | /* | |
696 | * Queue a packet for resolution. It gets locked cache entry! | |
697 | */ | |
e905a9ed | 698 | |
1da177e4 | 699 | static int |
4feb88e5 | 700 | ipmr_cache_unresolved(struct net *net, vifi_t vifi, struct sk_buff *skb) |
1da177e4 LT |
701 | { |
702 | int err; | |
703 | struct mfc_cache *c; | |
eddc9ec5 | 704 | const struct iphdr *iph = ip_hdr(skb); |
1da177e4 LT |
705 | |
706 | spin_lock_bh(&mfc_unres_lock); | |
e258beb2 PM |
707 | for (c=net->ipv4.mfc_unres_queue; c; c=c->next) { |
708 | if (c->mfc_mcastgrp == iph->daddr && | |
eddc9ec5 | 709 | c->mfc_origin == iph->saddr) |
1da177e4 LT |
710 | break; |
711 | } | |
712 | ||
713 | if (c == NULL) { | |
714 | /* | |
715 | * Create a new entry if allowable | |
716 | */ | |
717 | ||
4feb88e5 | 718 | if (atomic_read(&net->ipv4.cache_resolve_queue_len) >= 10 || |
d658f8a0 | 719 | (c = ipmr_cache_alloc_unres()) == NULL) { |
1da177e4 LT |
720 | spin_unlock_bh(&mfc_unres_lock); |
721 | ||
722 | kfree_skb(skb); | |
723 | return -ENOBUFS; | |
724 | } | |
725 | ||
726 | /* | |
727 | * Fill in the new cache entry | |
728 | */ | |
eddc9ec5 ACM |
729 | c->mfc_parent = -1; |
730 | c->mfc_origin = iph->saddr; | |
731 | c->mfc_mcastgrp = iph->daddr; | |
1da177e4 LT |
732 | |
733 | /* | |
734 | * Reflect first query at mrouted. | |
735 | */ | |
4feb88e5 BT |
736 | err = ipmr_cache_report(net, skb, vifi, IGMPMSG_NOCACHE); |
737 | if (err < 0) { | |
e905a9ed | 738 | /* If the report failed throw the cache entry |
1da177e4 LT |
739 | out - Brad Parker |
740 | */ | |
741 | spin_unlock_bh(&mfc_unres_lock); | |
742 | ||
5c0a66f5 | 743 | ipmr_cache_free(c); |
1da177e4 LT |
744 | kfree_skb(skb); |
745 | return err; | |
746 | } | |
747 | ||
4feb88e5 | 748 | atomic_inc(&net->ipv4.cache_resolve_queue_len); |
e258beb2 PM |
749 | c->next = net->ipv4.mfc_unres_queue; |
750 | net->ipv4.mfc_unres_queue = c; | |
1da177e4 | 751 | |
e258beb2 | 752 | mod_timer(&net->ipv4.ipmr_expire_timer, c->mfc_un.unres.expires); |
1da177e4 LT |
753 | } |
754 | ||
755 | /* | |
756 | * See if we can append the packet | |
757 | */ | |
758 | if (c->mfc_un.unres.unresolved.qlen>3) { | |
759 | kfree_skb(skb); | |
760 | err = -ENOBUFS; | |
761 | } else { | |
c354e124 | 762 | skb_queue_tail(&c->mfc_un.unres.unresolved, skb); |
1da177e4 LT |
763 | err = 0; |
764 | } | |
765 | ||
766 | spin_unlock_bh(&mfc_unres_lock); | |
767 | return err; | |
768 | } | |
769 | ||
770 | /* | |
771 | * MFC cache manipulation by user space mroute daemon | |
772 | */ | |
773 | ||
4feb88e5 | 774 | static int ipmr_mfc_delete(struct net *net, struct mfcctl *mfc) |
1da177e4 LT |
775 | { |
776 | int line; | |
777 | struct mfc_cache *c, **cp; | |
778 | ||
c354e124 | 779 | line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr); |
1da177e4 | 780 | |
4feb88e5 | 781 | for (cp = &net->ipv4.mfc_cache_array[line]; |
2bb8b26c | 782 | (c = *cp) != NULL; cp = &c->next) { |
1da177e4 LT |
783 | if (c->mfc_origin == mfc->mfcc_origin.s_addr && |
784 | c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr) { | |
785 | write_lock_bh(&mrt_lock); | |
786 | *cp = c->next; | |
787 | write_unlock_bh(&mrt_lock); | |
788 | ||
5c0a66f5 | 789 | ipmr_cache_free(c); |
1da177e4 LT |
790 | return 0; |
791 | } | |
792 | } | |
793 | return -ENOENT; | |
794 | } | |
795 | ||
4feb88e5 | 796 | static int ipmr_mfc_add(struct net *net, struct mfcctl *mfc, int mrtsock) |
1da177e4 LT |
797 | { |
798 | int line; | |
799 | struct mfc_cache *uc, *c, **cp; | |
800 | ||
a50436f2 PM |
801 | if (mfc->mfcc_parent >= MAXVIFS) |
802 | return -ENFILE; | |
803 | ||
c354e124 | 804 | line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr); |
1da177e4 | 805 | |
4feb88e5 | 806 | for (cp = &net->ipv4.mfc_cache_array[line]; |
2bb8b26c | 807 | (c = *cp) != NULL; cp = &c->next) { |
1da177e4 LT |
808 | if (c->mfc_origin == mfc->mfcc_origin.s_addr && |
809 | c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr) | |
810 | break; | |
811 | } | |
812 | ||
813 | if (c != NULL) { | |
814 | write_lock_bh(&mrt_lock); | |
815 | c->mfc_parent = mfc->mfcc_parent; | |
d658f8a0 | 816 | ipmr_update_thresholds(net, c, mfc->mfcc_ttls); |
1da177e4 LT |
817 | if (!mrtsock) |
818 | c->mfc_flags |= MFC_STATIC; | |
819 | write_unlock_bh(&mrt_lock); | |
820 | return 0; | |
821 | } | |
822 | ||
f97c1e0c | 823 | if (!ipv4_is_multicast(mfc->mfcc_mcastgrp.s_addr)) |
1da177e4 LT |
824 | return -EINVAL; |
825 | ||
d658f8a0 | 826 | c = ipmr_cache_alloc(); |
c354e124 | 827 | if (c == NULL) |
1da177e4 LT |
828 | return -ENOMEM; |
829 | ||
c354e124 JK |
830 | c->mfc_origin = mfc->mfcc_origin.s_addr; |
831 | c->mfc_mcastgrp = mfc->mfcc_mcastgrp.s_addr; | |
832 | c->mfc_parent = mfc->mfcc_parent; | |
d658f8a0 | 833 | ipmr_update_thresholds(net, c, mfc->mfcc_ttls); |
1da177e4 LT |
834 | if (!mrtsock) |
835 | c->mfc_flags |= MFC_STATIC; | |
836 | ||
837 | write_lock_bh(&mrt_lock); | |
4feb88e5 BT |
838 | c->next = net->ipv4.mfc_cache_array[line]; |
839 | net->ipv4.mfc_cache_array[line] = c; | |
1da177e4 LT |
840 | write_unlock_bh(&mrt_lock); |
841 | ||
842 | /* | |
843 | * Check to see if we resolved a queued list. If so we | |
844 | * need to send on the frames and tidy up. | |
845 | */ | |
846 | spin_lock_bh(&mfc_unres_lock); | |
e258beb2 | 847 | for (cp = &net->ipv4.mfc_unres_queue; (uc=*cp) != NULL; |
1da177e4 | 848 | cp = &uc->next) { |
e258beb2 | 849 | if (uc->mfc_origin == c->mfc_origin && |
1da177e4 LT |
850 | uc->mfc_mcastgrp == c->mfc_mcastgrp) { |
851 | *cp = uc->next; | |
4feb88e5 | 852 | atomic_dec(&net->ipv4.cache_resolve_queue_len); |
1da177e4 LT |
853 | break; |
854 | } | |
855 | } | |
e258beb2 PM |
856 | if (net->ipv4.mfc_unres_queue == NULL) |
857 | del_timer(&net->ipv4.ipmr_expire_timer); | |
1da177e4 LT |
858 | spin_unlock_bh(&mfc_unres_lock); |
859 | ||
860 | if (uc) { | |
d658f8a0 | 861 | ipmr_cache_resolve(net, uc, c); |
5c0a66f5 | 862 | ipmr_cache_free(uc); |
1da177e4 LT |
863 | } |
864 | return 0; | |
865 | } | |
866 | ||
867 | /* | |
868 | * Close the multicast socket, and clear the vif tables etc | |
869 | */ | |
e905a9ed | 870 | |
4feb88e5 | 871 | static void mroute_clean_tables(struct net *net) |
1da177e4 LT |
872 | { |
873 | int i; | |
d17fa6fa | 874 | LIST_HEAD(list); |
e905a9ed | 875 | |
1da177e4 LT |
876 | /* |
877 | * Shut down all active vif entries | |
878 | */ | |
4feb88e5 BT |
879 | for (i = 0; i < net->ipv4.maxvif; i++) { |
880 | if (!(net->ipv4.vif_table[i].flags&VIFF_STATIC)) | |
d17fa6fa | 881 | vif_delete(net, i, 0, &list); |
1da177e4 | 882 | } |
d17fa6fa | 883 | unregister_netdevice_many(&list); |
1da177e4 LT |
884 | |
885 | /* | |
886 | * Wipe the cache | |
887 | */ | |
c354e124 | 888 | for (i=0; i<MFC_LINES; i++) { |
1da177e4 LT |
889 | struct mfc_cache *c, **cp; |
890 | ||
4feb88e5 | 891 | cp = &net->ipv4.mfc_cache_array[i]; |
1da177e4 LT |
892 | while ((c = *cp) != NULL) { |
893 | if (c->mfc_flags&MFC_STATIC) { | |
894 | cp = &c->next; | |
895 | continue; | |
896 | } | |
897 | write_lock_bh(&mrt_lock); | |
898 | *cp = c->next; | |
899 | write_unlock_bh(&mrt_lock); | |
900 | ||
5c0a66f5 | 901 | ipmr_cache_free(c); |
1da177e4 LT |
902 | } |
903 | } | |
904 | ||
4feb88e5 | 905 | if (atomic_read(&net->ipv4.cache_resolve_queue_len) != 0) { |
1e8fb3b6 | 906 | struct mfc_cache *c, **cp; |
1da177e4 LT |
907 | |
908 | spin_lock_bh(&mfc_unres_lock); | |
e258beb2 | 909 | cp = &net->ipv4.mfc_unres_queue; |
1e8fb3b6 | 910 | while ((c = *cp) != NULL) { |
1e8fb3b6 | 911 | *cp = c->next; |
d658f8a0 | 912 | ipmr_destroy_unres(net, c); |
1da177e4 LT |
913 | } |
914 | spin_unlock_bh(&mfc_unres_lock); | |
915 | } | |
916 | } | |
917 | ||
918 | static void mrtsock_destruct(struct sock *sk) | |
919 | { | |
4feb88e5 BT |
920 | struct net *net = sock_net(sk); |
921 | ||
1da177e4 | 922 | rtnl_lock(); |
4feb88e5 BT |
923 | if (sk == net->ipv4.mroute_sk) { |
924 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)--; | |
1da177e4 LT |
925 | |
926 | write_lock_bh(&mrt_lock); | |
4feb88e5 | 927 | net->ipv4.mroute_sk = NULL; |
1da177e4 LT |
928 | write_unlock_bh(&mrt_lock); |
929 | ||
4feb88e5 | 930 | mroute_clean_tables(net); |
1da177e4 LT |
931 | } |
932 | rtnl_unlock(); | |
933 | } | |
934 | ||
935 | /* | |
936 | * Socket options and virtual interface manipulation. The whole | |
937 | * virtual interface system is a complete heap, but unfortunately | |
938 | * that's how BSD mrouted happens to think. Maybe one day with a proper | |
939 | * MOSPF/PIM router set up we can clean this up. | |
940 | */ | |
e905a9ed | 941 | |
b7058842 | 942 | int ip_mroute_setsockopt(struct sock *sk, int optname, char __user *optval, unsigned int optlen) |
1da177e4 LT |
943 | { |
944 | int ret; | |
945 | struct vifctl vif; | |
946 | struct mfcctl mfc; | |
4feb88e5 | 947 | struct net *net = sock_net(sk); |
e905a9ed | 948 | |
132adf54 | 949 | if (optname != MRT_INIT) { |
4feb88e5 | 950 | if (sk != net->ipv4.mroute_sk && !capable(CAP_NET_ADMIN)) |
1da177e4 LT |
951 | return -EACCES; |
952 | } | |
953 | ||
132adf54 SH |
954 | switch (optname) { |
955 | case MRT_INIT: | |
956 | if (sk->sk_type != SOCK_RAW || | |
c720c7e8 | 957 | inet_sk(sk)->inet_num != IPPROTO_IGMP) |
132adf54 | 958 | return -EOPNOTSUPP; |
c354e124 | 959 | if (optlen != sizeof(int)) |
132adf54 | 960 | return -ENOPROTOOPT; |
1da177e4 | 961 | |
132adf54 | 962 | rtnl_lock(); |
4feb88e5 | 963 | if (net->ipv4.mroute_sk) { |
1da177e4 | 964 | rtnl_unlock(); |
132adf54 SH |
965 | return -EADDRINUSE; |
966 | } | |
967 | ||
968 | ret = ip_ra_control(sk, 1, mrtsock_destruct); | |
969 | if (ret == 0) { | |
970 | write_lock_bh(&mrt_lock); | |
4feb88e5 | 971 | net->ipv4.mroute_sk = sk; |
132adf54 SH |
972 | write_unlock_bh(&mrt_lock); |
973 | ||
4feb88e5 | 974 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)++; |
132adf54 SH |
975 | } |
976 | rtnl_unlock(); | |
977 | return ret; | |
978 | case MRT_DONE: | |
4feb88e5 | 979 | if (sk != net->ipv4.mroute_sk) |
132adf54 SH |
980 | return -EACCES; |
981 | return ip_ra_control(sk, 0, NULL); | |
982 | case MRT_ADD_VIF: | |
983 | case MRT_DEL_VIF: | |
c354e124 | 984 | if (optlen != sizeof(vif)) |
132adf54 | 985 | return -EINVAL; |
c354e124 | 986 | if (copy_from_user(&vif, optval, sizeof(vif))) |
132adf54 SH |
987 | return -EFAULT; |
988 | if (vif.vifc_vifi >= MAXVIFS) | |
989 | return -ENFILE; | |
990 | rtnl_lock(); | |
c354e124 | 991 | if (optname == MRT_ADD_VIF) { |
4feb88e5 | 992 | ret = vif_add(net, &vif, sk == net->ipv4.mroute_sk); |
132adf54 | 993 | } else { |
d17fa6fa | 994 | ret = vif_delete(net, vif.vifc_vifi, 0, NULL); |
132adf54 SH |
995 | } |
996 | rtnl_unlock(); | |
997 | return ret; | |
1da177e4 LT |
998 | |
999 | /* | |
1000 | * Manipulate the forwarding caches. These live | |
1001 | * in a sort of kernel/user symbiosis. | |
1002 | */ | |
132adf54 SH |
1003 | case MRT_ADD_MFC: |
1004 | case MRT_DEL_MFC: | |
c354e124 | 1005 | if (optlen != sizeof(mfc)) |
132adf54 | 1006 | return -EINVAL; |
c354e124 | 1007 | if (copy_from_user(&mfc, optval, sizeof(mfc))) |
132adf54 SH |
1008 | return -EFAULT; |
1009 | rtnl_lock(); | |
c354e124 | 1010 | if (optname == MRT_DEL_MFC) |
4feb88e5 | 1011 | ret = ipmr_mfc_delete(net, &mfc); |
132adf54 | 1012 | else |
4feb88e5 | 1013 | ret = ipmr_mfc_add(net, &mfc, sk == net->ipv4.mroute_sk); |
132adf54 SH |
1014 | rtnl_unlock(); |
1015 | return ret; | |
1da177e4 LT |
1016 | /* |
1017 | * Control PIM assert. | |
1018 | */ | |
132adf54 SH |
1019 | case MRT_ASSERT: |
1020 | { | |
1021 | int v; | |
1022 | if (get_user(v,(int __user *)optval)) | |
1023 | return -EFAULT; | |
4feb88e5 | 1024 | net->ipv4.mroute_do_assert = (v) ? 1 : 0; |
132adf54 SH |
1025 | return 0; |
1026 | } | |
1da177e4 | 1027 | #ifdef CONFIG_IP_PIMSM |
132adf54 SH |
1028 | case MRT_PIM: |
1029 | { | |
ba93ef74 SH |
1030 | int v; |
1031 | ||
132adf54 SH |
1032 | if (get_user(v,(int __user *)optval)) |
1033 | return -EFAULT; | |
ba93ef74 SH |
1034 | v = (v) ? 1 : 0; |
1035 | ||
132adf54 SH |
1036 | rtnl_lock(); |
1037 | ret = 0; | |
4feb88e5 BT |
1038 | if (v != net->ipv4.mroute_do_pim) { |
1039 | net->ipv4.mroute_do_pim = v; | |
1040 | net->ipv4.mroute_do_assert = v; | |
1da177e4 | 1041 | } |
132adf54 SH |
1042 | rtnl_unlock(); |
1043 | return ret; | |
1044 | } | |
1da177e4 | 1045 | #endif |
132adf54 SH |
1046 | /* |
1047 | * Spurious command, or MRT_VERSION which you cannot | |
1048 | * set. | |
1049 | */ | |
1050 | default: | |
1051 | return -ENOPROTOOPT; | |
1da177e4 LT |
1052 | } |
1053 | } | |
1054 | ||
1055 | /* | |
1056 | * Getsock opt support for the multicast routing system. | |
1057 | */ | |
e905a9ed | 1058 | |
c354e124 | 1059 | int ip_mroute_getsockopt(struct sock *sk, int optname, char __user *optval, int __user *optlen) |
1da177e4 LT |
1060 | { |
1061 | int olr; | |
1062 | int val; | |
4feb88e5 | 1063 | struct net *net = sock_net(sk); |
1da177e4 | 1064 | |
c354e124 | 1065 | if (optname != MRT_VERSION && |
1da177e4 LT |
1066 | #ifdef CONFIG_IP_PIMSM |
1067 | optname!=MRT_PIM && | |
1068 | #endif | |
1069 | optname!=MRT_ASSERT) | |
1070 | return -ENOPROTOOPT; | |
1071 | ||
1072 | if (get_user(olr, optlen)) | |
1073 | return -EFAULT; | |
1074 | ||
1075 | olr = min_t(unsigned int, olr, sizeof(int)); | |
1076 | if (olr < 0) | |
1077 | return -EINVAL; | |
e905a9ed | 1078 | |
c354e124 | 1079 | if (put_user(olr, optlen)) |
1da177e4 | 1080 | return -EFAULT; |
c354e124 JK |
1081 | if (optname == MRT_VERSION) |
1082 | val = 0x0305; | |
1da177e4 | 1083 | #ifdef CONFIG_IP_PIMSM |
c354e124 | 1084 | else if (optname == MRT_PIM) |
4feb88e5 | 1085 | val = net->ipv4.mroute_do_pim; |
1da177e4 LT |
1086 | #endif |
1087 | else | |
4feb88e5 | 1088 | val = net->ipv4.mroute_do_assert; |
c354e124 | 1089 | if (copy_to_user(optval, &val, olr)) |
1da177e4 LT |
1090 | return -EFAULT; |
1091 | return 0; | |
1092 | } | |
1093 | ||
1094 | /* | |
1095 | * The IP multicast ioctl support routines. | |
1096 | */ | |
e905a9ed | 1097 | |
1da177e4 LT |
1098 | int ipmr_ioctl(struct sock *sk, int cmd, void __user *arg) |
1099 | { | |
1100 | struct sioc_sg_req sr; | |
1101 | struct sioc_vif_req vr; | |
1102 | struct vif_device *vif; | |
1103 | struct mfc_cache *c; | |
4feb88e5 | 1104 | struct net *net = sock_net(sk); |
e905a9ed | 1105 | |
132adf54 SH |
1106 | switch (cmd) { |
1107 | case SIOCGETVIFCNT: | |
c354e124 | 1108 | if (copy_from_user(&vr, arg, sizeof(vr))) |
132adf54 | 1109 | return -EFAULT; |
4feb88e5 | 1110 | if (vr.vifi >= net->ipv4.maxvif) |
132adf54 SH |
1111 | return -EINVAL; |
1112 | read_lock(&mrt_lock); | |
4feb88e5 BT |
1113 | vif = &net->ipv4.vif_table[vr.vifi]; |
1114 | if (VIF_EXISTS(net, vr.vifi)) { | |
c354e124 JK |
1115 | vr.icount = vif->pkt_in; |
1116 | vr.ocount = vif->pkt_out; | |
1117 | vr.ibytes = vif->bytes_in; | |
1118 | vr.obytes = vif->bytes_out; | |
1da177e4 | 1119 | read_unlock(&mrt_lock); |
1da177e4 | 1120 | |
c354e124 | 1121 | if (copy_to_user(arg, &vr, sizeof(vr))) |
132adf54 SH |
1122 | return -EFAULT; |
1123 | return 0; | |
1124 | } | |
1125 | read_unlock(&mrt_lock); | |
1126 | return -EADDRNOTAVAIL; | |
1127 | case SIOCGETSGCNT: | |
c354e124 | 1128 | if (copy_from_user(&sr, arg, sizeof(sr))) |
132adf54 SH |
1129 | return -EFAULT; |
1130 | ||
1131 | read_lock(&mrt_lock); | |
4feb88e5 | 1132 | c = ipmr_cache_find(net, sr.src.s_addr, sr.grp.s_addr); |
132adf54 SH |
1133 | if (c) { |
1134 | sr.pktcnt = c->mfc_un.res.pkt; | |
1135 | sr.bytecnt = c->mfc_un.res.bytes; | |
1136 | sr.wrong_if = c->mfc_un.res.wrong_if; | |
1da177e4 | 1137 | read_unlock(&mrt_lock); |
132adf54 | 1138 | |
c354e124 | 1139 | if (copy_to_user(arg, &sr, sizeof(sr))) |
132adf54 SH |
1140 | return -EFAULT; |
1141 | return 0; | |
1142 | } | |
1143 | read_unlock(&mrt_lock); | |
1144 | return -EADDRNOTAVAIL; | |
1145 | default: | |
1146 | return -ENOIOCTLCMD; | |
1da177e4 LT |
1147 | } |
1148 | } | |
1149 | ||
1150 | ||
1151 | static int ipmr_device_event(struct notifier_block *this, unsigned long event, void *ptr) | |
1152 | { | |
e9dc8653 | 1153 | struct net_device *dev = ptr; |
4feb88e5 | 1154 | struct net *net = dev_net(dev); |
1da177e4 LT |
1155 | struct vif_device *v; |
1156 | int ct; | |
d17fa6fa | 1157 | LIST_HEAD(list); |
e9dc8653 | 1158 | |
1da177e4 LT |
1159 | if (event != NETDEV_UNREGISTER) |
1160 | return NOTIFY_DONE; | |
4feb88e5 BT |
1161 | v = &net->ipv4.vif_table[0]; |
1162 | for (ct = 0; ct < net->ipv4.maxvif; ct++, v++) { | |
c354e124 | 1163 | if (v->dev == dev) |
d17fa6fa | 1164 | vif_delete(net, ct, 1, &list); |
1da177e4 | 1165 | } |
d17fa6fa | 1166 | unregister_netdevice_many(&list); |
1da177e4 LT |
1167 | return NOTIFY_DONE; |
1168 | } | |
1169 | ||
1170 | ||
c354e124 | 1171 | static struct notifier_block ip_mr_notifier = { |
1da177e4 LT |
1172 | .notifier_call = ipmr_device_event, |
1173 | }; | |
1174 | ||
1175 | /* | |
1176 | * Encapsulate a packet by attaching a valid IPIP header to it. | |
1177 | * This avoids tunnel drivers and other mess and gives us the speed so | |
1178 | * important for multicast video. | |
1179 | */ | |
e905a9ed | 1180 | |
114c7844 | 1181 | static void ip_encap(struct sk_buff *skb, __be32 saddr, __be32 daddr) |
1da177e4 | 1182 | { |
8856dfa3 | 1183 | struct iphdr *iph; |
eddc9ec5 | 1184 | struct iphdr *old_iph = ip_hdr(skb); |
8856dfa3 ACM |
1185 | |
1186 | skb_push(skb, sizeof(struct iphdr)); | |
b0e380b1 | 1187 | skb->transport_header = skb->network_header; |
8856dfa3 | 1188 | skb_reset_network_header(skb); |
eddc9ec5 | 1189 | iph = ip_hdr(skb); |
1da177e4 LT |
1190 | |
1191 | iph->version = 4; | |
e023dd64 ACM |
1192 | iph->tos = old_iph->tos; |
1193 | iph->ttl = old_iph->ttl; | |
1da177e4 LT |
1194 | iph->frag_off = 0; |
1195 | iph->daddr = daddr; | |
1196 | iph->saddr = saddr; | |
1197 | iph->protocol = IPPROTO_IPIP; | |
1198 | iph->ihl = 5; | |
1199 | iph->tot_len = htons(skb->len); | |
adf30907 | 1200 | ip_select_ident(iph, skb_dst(skb), NULL); |
1da177e4 LT |
1201 | ip_send_check(iph); |
1202 | ||
1da177e4 LT |
1203 | memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt)); |
1204 | nf_reset(skb); | |
1205 | } | |
1206 | ||
1207 | static inline int ipmr_forward_finish(struct sk_buff *skb) | |
1208 | { | |
1209 | struct ip_options * opt = &(IPCB(skb)->opt); | |
1210 | ||
adf30907 | 1211 | IP_INC_STATS_BH(dev_net(skb_dst(skb)->dev), IPSTATS_MIB_OUTFORWDATAGRAMS); |
1da177e4 LT |
1212 | |
1213 | if (unlikely(opt->optlen)) | |
1214 | ip_forward_options(skb); | |
1215 | ||
1216 | return dst_output(skb); | |
1217 | } | |
1218 | ||
1219 | /* | |
1220 | * Processing handlers for ipmr_forward | |
1221 | */ | |
1222 | ||
d658f8a0 PM |
1223 | static void ipmr_queue_xmit(struct net *net, struct sk_buff *skb, |
1224 | struct mfc_cache *c, int vifi) | |
1da177e4 | 1225 | { |
eddc9ec5 | 1226 | const struct iphdr *iph = ip_hdr(skb); |
4feb88e5 | 1227 | struct vif_device *vif = &net->ipv4.vif_table[vifi]; |
1da177e4 LT |
1228 | struct net_device *dev; |
1229 | struct rtable *rt; | |
1230 | int encap = 0; | |
1231 | ||
1232 | if (vif->dev == NULL) | |
1233 | goto out_free; | |
1234 | ||
1235 | #ifdef CONFIG_IP_PIMSM | |
1236 | if (vif->flags & VIFF_REGISTER) { | |
1237 | vif->pkt_out++; | |
c354e124 | 1238 | vif->bytes_out += skb->len; |
cf3677ae PE |
1239 | vif->dev->stats.tx_bytes += skb->len; |
1240 | vif->dev->stats.tx_packets++; | |
4feb88e5 | 1241 | ipmr_cache_report(net, skb, vifi, IGMPMSG_WHOLEPKT); |
69ebbf58 | 1242 | goto out_free; |
1da177e4 LT |
1243 | } |
1244 | #endif | |
1245 | ||
1246 | if (vif->flags&VIFF_TUNNEL) { | |
1247 | struct flowi fl = { .oif = vif->link, | |
1248 | .nl_u = { .ip4_u = | |
1249 | { .daddr = vif->remote, | |
1250 | .saddr = vif->local, | |
1251 | .tos = RT_TOS(iph->tos) } }, | |
1252 | .proto = IPPROTO_IPIP }; | |
4feb88e5 | 1253 | if (ip_route_output_key(net, &rt, &fl)) |
1da177e4 LT |
1254 | goto out_free; |
1255 | encap = sizeof(struct iphdr); | |
1256 | } else { | |
1257 | struct flowi fl = { .oif = vif->link, | |
1258 | .nl_u = { .ip4_u = | |
1259 | { .daddr = iph->daddr, | |
1260 | .tos = RT_TOS(iph->tos) } }, | |
1261 | .proto = IPPROTO_IPIP }; | |
4feb88e5 | 1262 | if (ip_route_output_key(net, &rt, &fl)) |
1da177e4 LT |
1263 | goto out_free; |
1264 | } | |
1265 | ||
1266 | dev = rt->u.dst.dev; | |
1267 | ||
1268 | if (skb->len+encap > dst_mtu(&rt->u.dst) && (ntohs(iph->frag_off) & IP_DF)) { | |
1269 | /* Do not fragment multicasts. Alas, IPv4 does not | |
1270 | allow to send ICMP, so that packets will disappear | |
1271 | to blackhole. | |
1272 | */ | |
1273 | ||
7c73a6fa | 1274 | IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_FRAGFAILS); |
1da177e4 LT |
1275 | ip_rt_put(rt); |
1276 | goto out_free; | |
1277 | } | |
1278 | ||
1279 | encap += LL_RESERVED_SPACE(dev) + rt->u.dst.header_len; | |
1280 | ||
1281 | if (skb_cow(skb, encap)) { | |
e905a9ed | 1282 | ip_rt_put(rt); |
1da177e4 LT |
1283 | goto out_free; |
1284 | } | |
1285 | ||
1286 | vif->pkt_out++; | |
c354e124 | 1287 | vif->bytes_out += skb->len; |
1da177e4 | 1288 | |
adf30907 ED |
1289 | skb_dst_drop(skb); |
1290 | skb_dst_set(skb, &rt->u.dst); | |
eddc9ec5 | 1291 | ip_decrease_ttl(ip_hdr(skb)); |
1da177e4 LT |
1292 | |
1293 | /* FIXME: forward and output firewalls used to be called here. | |
1294 | * What do we do with netfilter? -- RR */ | |
1295 | if (vif->flags & VIFF_TUNNEL) { | |
1296 | ip_encap(skb, vif->local, vif->remote); | |
1297 | /* FIXME: extra output firewall step used to be here. --RR */ | |
2f4c02d4 PE |
1298 | vif->dev->stats.tx_packets++; |
1299 | vif->dev->stats.tx_bytes += skb->len; | |
1da177e4 LT |
1300 | } |
1301 | ||
1302 | IPCB(skb)->flags |= IPSKB_FORWARDED; | |
1303 | ||
1304 | /* | |
1305 | * RFC1584 teaches, that DVMRP/PIM router must deliver packets locally | |
1306 | * not only before forwarding, but after forwarding on all output | |
1307 | * interfaces. It is clear, if mrouter runs a multicasting | |
1308 | * program, it should receive packets not depending to what interface | |
1309 | * program is joined. | |
1310 | * If we will not make it, the program will have to join on all | |
1311 | * interfaces. On the other hand, multihoming host (or router, but | |
1312 | * not mrouter) cannot join to more than one interface - it will | |
1313 | * result in receiving multiple packets. | |
1314 | */ | |
6e23ae2a | 1315 | NF_HOOK(PF_INET, NF_INET_FORWARD, skb, skb->dev, dev, |
1da177e4 LT |
1316 | ipmr_forward_finish); |
1317 | return; | |
1318 | ||
1319 | out_free: | |
1320 | kfree_skb(skb); | |
1321 | return; | |
1322 | } | |
1323 | ||
1324 | static int ipmr_find_vif(struct net_device *dev) | |
1325 | { | |
4feb88e5 | 1326 | struct net *net = dev_net(dev); |
1da177e4 | 1327 | int ct; |
4feb88e5 BT |
1328 | for (ct = net->ipv4.maxvif-1; ct >= 0; ct--) { |
1329 | if (net->ipv4.vif_table[ct].dev == dev) | |
1da177e4 LT |
1330 | break; |
1331 | } | |
1332 | return ct; | |
1333 | } | |
1334 | ||
1335 | /* "local" means that we should preserve one skb (for local delivery) */ | |
1336 | ||
d658f8a0 PM |
1337 | static int ip_mr_forward(struct net *net, struct sk_buff *skb, |
1338 | struct mfc_cache *cache, int local) | |
1da177e4 LT |
1339 | { |
1340 | int psend = -1; | |
1341 | int vif, ct; | |
1342 | ||
1343 | vif = cache->mfc_parent; | |
1344 | cache->mfc_un.res.pkt++; | |
1345 | cache->mfc_un.res.bytes += skb->len; | |
1346 | ||
1347 | /* | |
1348 | * Wrong interface: drop packet and (maybe) send PIM assert. | |
1349 | */ | |
4feb88e5 | 1350 | if (net->ipv4.vif_table[vif].dev != skb->dev) { |
1da177e4 LT |
1351 | int true_vifi; |
1352 | ||
511c3f92 | 1353 | if (skb_rtable(skb)->fl.iif == 0) { |
1da177e4 LT |
1354 | /* It is our own packet, looped back. |
1355 | Very complicated situation... | |
1356 | ||
1357 | The best workaround until routing daemons will be | |
1358 | fixed is not to redistribute packet, if it was | |
1359 | send through wrong interface. It means, that | |
1360 | multicast applications WILL NOT work for | |
1361 | (S,G), which have default multicast route pointing | |
1362 | to wrong oif. In any case, it is not a good | |
1363 | idea to use multicasting applications on router. | |
1364 | */ | |
1365 | goto dont_forward; | |
1366 | } | |
1367 | ||
1368 | cache->mfc_un.res.wrong_if++; | |
1369 | true_vifi = ipmr_find_vif(skb->dev); | |
1370 | ||
4feb88e5 | 1371 | if (true_vifi >= 0 && net->ipv4.mroute_do_assert && |
1da177e4 LT |
1372 | /* pimsm uses asserts, when switching from RPT to SPT, |
1373 | so that we cannot check that packet arrived on an oif. | |
1374 | It is bad, but otherwise we would need to move pretty | |
1375 | large chunk of pimd to kernel. Ough... --ANK | |
1376 | */ | |
4feb88e5 | 1377 | (net->ipv4.mroute_do_pim || |
6f9374a9 | 1378 | cache->mfc_un.res.ttls[true_vifi] < 255) && |
e905a9ed | 1379 | time_after(jiffies, |
1da177e4 LT |
1380 | cache->mfc_un.res.last_assert + MFC_ASSERT_THRESH)) { |
1381 | cache->mfc_un.res.last_assert = jiffies; | |
4feb88e5 | 1382 | ipmr_cache_report(net, skb, true_vifi, IGMPMSG_WRONGVIF); |
1da177e4 LT |
1383 | } |
1384 | goto dont_forward; | |
1385 | } | |
1386 | ||
4feb88e5 BT |
1387 | net->ipv4.vif_table[vif].pkt_in++; |
1388 | net->ipv4.vif_table[vif].bytes_in += skb->len; | |
1da177e4 LT |
1389 | |
1390 | /* | |
1391 | * Forward the frame | |
1392 | */ | |
1393 | for (ct = cache->mfc_un.res.maxvif-1; ct >= cache->mfc_un.res.minvif; ct--) { | |
eddc9ec5 | 1394 | if (ip_hdr(skb)->ttl > cache->mfc_un.res.ttls[ct]) { |
1da177e4 LT |
1395 | if (psend != -1) { |
1396 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
1397 | if (skb2) | |
d658f8a0 | 1398 | ipmr_queue_xmit(net, skb2, cache, psend); |
1da177e4 | 1399 | } |
c354e124 | 1400 | psend = ct; |
1da177e4 LT |
1401 | } |
1402 | } | |
1403 | if (psend != -1) { | |
1404 | if (local) { | |
1405 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
1406 | if (skb2) | |
d658f8a0 | 1407 | ipmr_queue_xmit(net, skb2, cache, psend); |
1da177e4 | 1408 | } else { |
d658f8a0 | 1409 | ipmr_queue_xmit(net, skb, cache, psend); |
1da177e4 LT |
1410 | return 0; |
1411 | } | |
1412 | } | |
1413 | ||
1414 | dont_forward: | |
1415 | if (!local) | |
1416 | kfree_skb(skb); | |
1417 | return 0; | |
1418 | } | |
1419 | ||
1420 | ||
1421 | /* | |
1422 | * Multicast packets for forwarding arrive here | |
1423 | */ | |
1424 | ||
1425 | int ip_mr_input(struct sk_buff *skb) | |
1426 | { | |
1427 | struct mfc_cache *cache; | |
4feb88e5 | 1428 | struct net *net = dev_net(skb->dev); |
511c3f92 | 1429 | int local = skb_rtable(skb)->rt_flags & RTCF_LOCAL; |
1da177e4 LT |
1430 | |
1431 | /* Packet is looped back after forward, it should not be | |
1432 | forwarded second time, but still can be delivered locally. | |
1433 | */ | |
1434 | if (IPCB(skb)->flags&IPSKB_FORWARDED) | |
1435 | goto dont_forward; | |
1436 | ||
1437 | if (!local) { | |
1438 | if (IPCB(skb)->opt.router_alert) { | |
1439 | if (ip_call_ra_chain(skb)) | |
1440 | return 0; | |
eddc9ec5 | 1441 | } else if (ip_hdr(skb)->protocol == IPPROTO_IGMP){ |
1da177e4 LT |
1442 | /* IGMPv1 (and broken IGMPv2 implementations sort of |
1443 | Cisco IOS <= 11.2(8)) do not put router alert | |
1444 | option to IGMP packets destined to routable | |
1445 | groups. It is very bad, because it means | |
1446 | that we can forward NO IGMP messages. | |
1447 | */ | |
1448 | read_lock(&mrt_lock); | |
4feb88e5 | 1449 | if (net->ipv4.mroute_sk) { |
2715bcf9 | 1450 | nf_reset(skb); |
4feb88e5 | 1451 | raw_rcv(net->ipv4.mroute_sk, skb); |
1da177e4 LT |
1452 | read_unlock(&mrt_lock); |
1453 | return 0; | |
1454 | } | |
1455 | read_unlock(&mrt_lock); | |
1456 | } | |
1457 | } | |
1458 | ||
1459 | read_lock(&mrt_lock); | |
4feb88e5 | 1460 | cache = ipmr_cache_find(net, ip_hdr(skb)->saddr, ip_hdr(skb)->daddr); |
1da177e4 LT |
1461 | |
1462 | /* | |
1463 | * No usable cache entry | |
1464 | */ | |
c354e124 | 1465 | if (cache == NULL) { |
1da177e4 LT |
1466 | int vif; |
1467 | ||
1468 | if (local) { | |
1469 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
1470 | ip_local_deliver(skb); | |
1471 | if (skb2 == NULL) { | |
1472 | read_unlock(&mrt_lock); | |
1473 | return -ENOBUFS; | |
1474 | } | |
1475 | skb = skb2; | |
1476 | } | |
1477 | ||
1478 | vif = ipmr_find_vif(skb->dev); | |
1479 | if (vif >= 0) { | |
4feb88e5 | 1480 | int err = ipmr_cache_unresolved(net, vif, skb); |
1da177e4 LT |
1481 | read_unlock(&mrt_lock); |
1482 | ||
1483 | return err; | |
1484 | } | |
1485 | read_unlock(&mrt_lock); | |
1486 | kfree_skb(skb); | |
1487 | return -ENODEV; | |
1488 | } | |
1489 | ||
d658f8a0 | 1490 | ip_mr_forward(net, skb, cache, local); |
1da177e4 LT |
1491 | |
1492 | read_unlock(&mrt_lock); | |
1493 | ||
1494 | if (local) | |
1495 | return ip_local_deliver(skb); | |
1496 | ||
1497 | return 0; | |
1498 | ||
1499 | dont_forward: | |
1500 | if (local) | |
1501 | return ip_local_deliver(skb); | |
1502 | kfree_skb(skb); | |
1503 | return 0; | |
1504 | } | |
1505 | ||
b1879204 IJ |
1506 | #ifdef CONFIG_IP_PIMSM |
1507 | static int __pim_rcv(struct sk_buff *skb, unsigned int pimlen) | |
1da177e4 | 1508 | { |
b1879204 IJ |
1509 | struct net_device *reg_dev = NULL; |
1510 | struct iphdr *encap; | |
4feb88e5 | 1511 | struct net *net = dev_net(skb->dev); |
1da177e4 | 1512 | |
b1879204 | 1513 | encap = (struct iphdr *)(skb_transport_header(skb) + pimlen); |
1da177e4 LT |
1514 | /* |
1515 | Check that: | |
1516 | a. packet is really destinted to a multicast group | |
1517 | b. packet is not a NULL-REGISTER | |
1518 | c. packet is not truncated | |
1519 | */ | |
f97c1e0c | 1520 | if (!ipv4_is_multicast(encap->daddr) || |
1da177e4 | 1521 | encap->tot_len == 0 || |
b1879204 IJ |
1522 | ntohs(encap->tot_len) + pimlen > skb->len) |
1523 | return 1; | |
1da177e4 LT |
1524 | |
1525 | read_lock(&mrt_lock); | |
4feb88e5 BT |
1526 | if (net->ipv4.mroute_reg_vif_num >= 0) |
1527 | reg_dev = net->ipv4.vif_table[net->ipv4.mroute_reg_vif_num].dev; | |
1da177e4 LT |
1528 | if (reg_dev) |
1529 | dev_hold(reg_dev); | |
1530 | read_unlock(&mrt_lock); | |
1531 | ||
e905a9ed | 1532 | if (reg_dev == NULL) |
b1879204 | 1533 | return 1; |
1da177e4 | 1534 | |
b0e380b1 | 1535 | skb->mac_header = skb->network_header; |
1da177e4 | 1536 | skb_pull(skb, (u8*)encap - skb->data); |
31c7711b | 1537 | skb_reset_network_header(skb); |
1da177e4 | 1538 | skb->dev = reg_dev; |
1da177e4 LT |
1539 | skb->protocol = htons(ETH_P_IP); |
1540 | skb->ip_summed = 0; | |
1541 | skb->pkt_type = PACKET_HOST; | |
adf30907 | 1542 | skb_dst_drop(skb); |
cf3677ae PE |
1543 | reg_dev->stats.rx_bytes += skb->len; |
1544 | reg_dev->stats.rx_packets++; | |
1da177e4 LT |
1545 | nf_reset(skb); |
1546 | netif_rx(skb); | |
1547 | dev_put(reg_dev); | |
b1879204 | 1548 | |
1da177e4 | 1549 | return 0; |
b1879204 IJ |
1550 | } |
1551 | #endif | |
1552 | ||
1553 | #ifdef CONFIG_IP_PIMSM_V1 | |
1554 | /* | |
1555 | * Handle IGMP messages of PIMv1 | |
1556 | */ | |
1557 | ||
1558 | int pim_rcv_v1(struct sk_buff * skb) | |
1559 | { | |
1560 | struct igmphdr *pim; | |
4feb88e5 | 1561 | struct net *net = dev_net(skb->dev); |
b1879204 IJ |
1562 | |
1563 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) | |
1564 | goto drop; | |
1565 | ||
1566 | pim = igmp_hdr(skb); | |
1567 | ||
4feb88e5 | 1568 | if (!net->ipv4.mroute_do_pim || |
b1879204 IJ |
1569 | pim->group != PIM_V1_VERSION || pim->code != PIM_V1_REGISTER) |
1570 | goto drop; | |
1571 | ||
1572 | if (__pim_rcv(skb, sizeof(*pim))) { | |
1573 | drop: | |
1574 | kfree_skb(skb); | |
1575 | } | |
1da177e4 LT |
1576 | return 0; |
1577 | } | |
1578 | #endif | |
1579 | ||
1580 | #ifdef CONFIG_IP_PIMSM_V2 | |
1581 | static int pim_rcv(struct sk_buff * skb) | |
1582 | { | |
1583 | struct pimreghdr *pim; | |
1da177e4 | 1584 | |
b1879204 | 1585 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) |
1da177e4 LT |
1586 | goto drop; |
1587 | ||
9c70220b | 1588 | pim = (struct pimreghdr *)skb_transport_header(skb); |
e905a9ed | 1589 | if (pim->type != ((PIM_VERSION<<4)|(PIM_REGISTER)) || |
1da177e4 | 1590 | (pim->flags&PIM_NULL_REGISTER) || |
e905a9ed | 1591 | (ip_compute_csum((void *)pim, sizeof(*pim)) != 0 && |
d3bc23e7 | 1592 | csum_fold(skb_checksum(skb, 0, skb->len, 0)))) |
1da177e4 LT |
1593 | goto drop; |
1594 | ||
b1879204 IJ |
1595 | if (__pim_rcv(skb, sizeof(*pim))) { |
1596 | drop: | |
1597 | kfree_skb(skb); | |
1598 | } | |
1da177e4 LT |
1599 | return 0; |
1600 | } | |
1601 | #endif | |
1602 | ||
1603 | static int | |
d658f8a0 PM |
1604 | ipmr_fill_mroute(struct net *net, struct sk_buff *skb, struct mfc_cache *c, |
1605 | struct rtmsg *rtm) | |
1da177e4 LT |
1606 | { |
1607 | int ct; | |
1608 | struct rtnexthop *nhp; | |
27a884dc | 1609 | u8 *b = skb_tail_pointer(skb); |
1da177e4 LT |
1610 | struct rtattr *mp_head; |
1611 | ||
7438189b ND |
1612 | /* If cache is unresolved, don't try to parse IIF and OIF */ |
1613 | if (c->mfc_parent > MAXVIFS) | |
1614 | return -ENOENT; | |
1615 | ||
1616 | if (VIF_EXISTS(net, c->mfc_parent)) | |
1617 | RTA_PUT(skb, RTA_IIF, 4, &net->ipv4.vif_table[c->mfc_parent].dev->ifindex); | |
1da177e4 | 1618 | |
c354e124 | 1619 | mp_head = (struct rtattr *)skb_put(skb, RTA_LENGTH(0)); |
1da177e4 LT |
1620 | |
1621 | for (ct = c->mfc_un.res.minvif; ct < c->mfc_un.res.maxvif; ct++) { | |
7438189b | 1622 | if (VIF_EXISTS(net, ct) && c->mfc_un.res.ttls[ct] < 255) { |
1da177e4 LT |
1623 | if (skb_tailroom(skb) < RTA_ALIGN(RTA_ALIGN(sizeof(*nhp)) + 4)) |
1624 | goto rtattr_failure; | |
c354e124 | 1625 | nhp = (struct rtnexthop *)skb_put(skb, RTA_ALIGN(sizeof(*nhp))); |
1da177e4 LT |
1626 | nhp->rtnh_flags = 0; |
1627 | nhp->rtnh_hops = c->mfc_un.res.ttls[ct]; | |
4feb88e5 | 1628 | nhp->rtnh_ifindex = net->ipv4.vif_table[ct].dev->ifindex; |
1da177e4 LT |
1629 | nhp->rtnh_len = sizeof(*nhp); |
1630 | } | |
1631 | } | |
1632 | mp_head->rta_type = RTA_MULTIPATH; | |
27a884dc | 1633 | mp_head->rta_len = skb_tail_pointer(skb) - (u8 *)mp_head; |
1da177e4 LT |
1634 | rtm->rtm_type = RTN_MULTICAST; |
1635 | return 1; | |
1636 | ||
1637 | rtattr_failure: | |
dc5fc579 | 1638 | nlmsg_trim(skb, b); |
1da177e4 LT |
1639 | return -EMSGSIZE; |
1640 | } | |
1641 | ||
4feb88e5 BT |
1642 | int ipmr_get_route(struct net *net, |
1643 | struct sk_buff *skb, struct rtmsg *rtm, int nowait) | |
1da177e4 LT |
1644 | { |
1645 | int err; | |
1646 | struct mfc_cache *cache; | |
511c3f92 | 1647 | struct rtable *rt = skb_rtable(skb); |
1da177e4 LT |
1648 | |
1649 | read_lock(&mrt_lock); | |
4feb88e5 | 1650 | cache = ipmr_cache_find(net, rt->rt_src, rt->rt_dst); |
1da177e4 | 1651 | |
c354e124 | 1652 | if (cache == NULL) { |
72287490 | 1653 | struct sk_buff *skb2; |
eddc9ec5 | 1654 | struct iphdr *iph; |
1da177e4 LT |
1655 | struct net_device *dev; |
1656 | int vif; | |
1657 | ||
1658 | if (nowait) { | |
1659 | read_unlock(&mrt_lock); | |
1660 | return -EAGAIN; | |
1661 | } | |
1662 | ||
1663 | dev = skb->dev; | |
1664 | if (dev == NULL || (vif = ipmr_find_vif(dev)) < 0) { | |
1665 | read_unlock(&mrt_lock); | |
1666 | return -ENODEV; | |
1667 | } | |
72287490 AK |
1668 | skb2 = skb_clone(skb, GFP_ATOMIC); |
1669 | if (!skb2) { | |
1670 | read_unlock(&mrt_lock); | |
1671 | return -ENOMEM; | |
1672 | } | |
1673 | ||
e2d1bca7 ACM |
1674 | skb_push(skb2, sizeof(struct iphdr)); |
1675 | skb_reset_network_header(skb2); | |
eddc9ec5 ACM |
1676 | iph = ip_hdr(skb2); |
1677 | iph->ihl = sizeof(struct iphdr) >> 2; | |
1678 | iph->saddr = rt->rt_src; | |
1679 | iph->daddr = rt->rt_dst; | |
1680 | iph->version = 0; | |
4feb88e5 | 1681 | err = ipmr_cache_unresolved(net, vif, skb2); |
1da177e4 LT |
1682 | read_unlock(&mrt_lock); |
1683 | return err; | |
1684 | } | |
1685 | ||
1686 | if (!nowait && (rtm->rtm_flags&RTM_F_NOTIFY)) | |
1687 | cache->mfc_flags |= MFC_NOTIFY; | |
d658f8a0 | 1688 | err = ipmr_fill_mroute(net, skb, cache, rtm); |
1da177e4 LT |
1689 | read_unlock(&mrt_lock); |
1690 | return err; | |
1691 | } | |
1692 | ||
e905a9ed | 1693 | #ifdef CONFIG_PROC_FS |
1da177e4 LT |
1694 | /* |
1695 | * The /proc interfaces to multicast routing /proc/ip_mr_cache /proc/ip_mr_vif | |
1696 | */ | |
1697 | struct ipmr_vif_iter { | |
f6bb4514 | 1698 | struct seq_net_private p; |
1da177e4 LT |
1699 | int ct; |
1700 | }; | |
1701 | ||
f6bb4514 BT |
1702 | static struct vif_device *ipmr_vif_seq_idx(struct net *net, |
1703 | struct ipmr_vif_iter *iter, | |
1da177e4 LT |
1704 | loff_t pos) |
1705 | { | |
f6bb4514 BT |
1706 | for (iter->ct = 0; iter->ct < net->ipv4.maxvif; ++iter->ct) { |
1707 | if (!VIF_EXISTS(net, iter->ct)) | |
1da177e4 | 1708 | continue; |
e905a9ed | 1709 | if (pos-- == 0) |
f6bb4514 | 1710 | return &net->ipv4.vif_table[iter->ct]; |
1da177e4 LT |
1711 | } |
1712 | return NULL; | |
1713 | } | |
1714 | ||
1715 | static void *ipmr_vif_seq_start(struct seq_file *seq, loff_t *pos) | |
ba93ef74 | 1716 | __acquires(mrt_lock) |
1da177e4 | 1717 | { |
f6bb4514 BT |
1718 | struct net *net = seq_file_net(seq); |
1719 | ||
1da177e4 | 1720 | read_lock(&mrt_lock); |
f6bb4514 | 1721 | return *pos ? ipmr_vif_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
1722 | : SEQ_START_TOKEN; |
1723 | } | |
1724 | ||
1725 | static void *ipmr_vif_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
1726 | { | |
1727 | struct ipmr_vif_iter *iter = seq->private; | |
f6bb4514 | 1728 | struct net *net = seq_file_net(seq); |
1da177e4 LT |
1729 | |
1730 | ++*pos; | |
1731 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 1732 | return ipmr_vif_seq_idx(net, iter, 0); |
e905a9ed | 1733 | |
f6bb4514 BT |
1734 | while (++iter->ct < net->ipv4.maxvif) { |
1735 | if (!VIF_EXISTS(net, iter->ct)) | |
1da177e4 | 1736 | continue; |
f6bb4514 | 1737 | return &net->ipv4.vif_table[iter->ct]; |
1da177e4 LT |
1738 | } |
1739 | return NULL; | |
1740 | } | |
1741 | ||
1742 | static void ipmr_vif_seq_stop(struct seq_file *seq, void *v) | |
ba93ef74 | 1743 | __releases(mrt_lock) |
1da177e4 LT |
1744 | { |
1745 | read_unlock(&mrt_lock); | |
1746 | } | |
1747 | ||
1748 | static int ipmr_vif_seq_show(struct seq_file *seq, void *v) | |
1749 | { | |
f6bb4514 BT |
1750 | struct net *net = seq_file_net(seq); |
1751 | ||
1da177e4 | 1752 | if (v == SEQ_START_TOKEN) { |
e905a9ed | 1753 | seq_puts(seq, |
1da177e4 LT |
1754 | "Interface BytesIn PktsIn BytesOut PktsOut Flags Local Remote\n"); |
1755 | } else { | |
1756 | const struct vif_device *vif = v; | |
1757 | const char *name = vif->dev ? vif->dev->name : "none"; | |
1758 | ||
1759 | seq_printf(seq, | |
1760 | "%2Zd %-10s %8ld %7ld %8ld %7ld %05X %08X %08X\n", | |
f6bb4514 | 1761 | vif - net->ipv4.vif_table, |
e905a9ed | 1762 | name, vif->bytes_in, vif->pkt_in, |
1da177e4 LT |
1763 | vif->bytes_out, vif->pkt_out, |
1764 | vif->flags, vif->local, vif->remote); | |
1765 | } | |
1766 | return 0; | |
1767 | } | |
1768 | ||
f690808e | 1769 | static const struct seq_operations ipmr_vif_seq_ops = { |
1da177e4 LT |
1770 | .start = ipmr_vif_seq_start, |
1771 | .next = ipmr_vif_seq_next, | |
1772 | .stop = ipmr_vif_seq_stop, | |
1773 | .show = ipmr_vif_seq_show, | |
1774 | }; | |
1775 | ||
1776 | static int ipmr_vif_open(struct inode *inode, struct file *file) | |
1777 | { | |
f6bb4514 BT |
1778 | return seq_open_net(inode, file, &ipmr_vif_seq_ops, |
1779 | sizeof(struct ipmr_vif_iter)); | |
1da177e4 LT |
1780 | } |
1781 | ||
9a32144e | 1782 | static const struct file_operations ipmr_vif_fops = { |
1da177e4 LT |
1783 | .owner = THIS_MODULE, |
1784 | .open = ipmr_vif_open, | |
1785 | .read = seq_read, | |
1786 | .llseek = seq_lseek, | |
f6bb4514 | 1787 | .release = seq_release_net, |
1da177e4 LT |
1788 | }; |
1789 | ||
1790 | struct ipmr_mfc_iter { | |
f6bb4514 | 1791 | struct seq_net_private p; |
1da177e4 LT |
1792 | struct mfc_cache **cache; |
1793 | int ct; | |
1794 | }; | |
1795 | ||
1796 | ||
f6bb4514 BT |
1797 | static struct mfc_cache *ipmr_mfc_seq_idx(struct net *net, |
1798 | struct ipmr_mfc_iter *it, loff_t pos) | |
1da177e4 LT |
1799 | { |
1800 | struct mfc_cache *mfc; | |
1801 | ||
f6bb4514 | 1802 | it->cache = net->ipv4.mfc_cache_array; |
1da177e4 | 1803 | read_lock(&mrt_lock); |
e905a9ed | 1804 | for (it->ct = 0; it->ct < MFC_LINES; it->ct++) |
f6bb4514 | 1805 | for (mfc = net->ipv4.mfc_cache_array[it->ct]; |
2bb8b26c | 1806 | mfc; mfc = mfc->next) |
e905a9ed | 1807 | if (pos-- == 0) |
1da177e4 LT |
1808 | return mfc; |
1809 | read_unlock(&mrt_lock); | |
1810 | ||
e258beb2 | 1811 | it->cache = &net->ipv4.mfc_unres_queue; |
1da177e4 | 1812 | spin_lock_bh(&mfc_unres_lock); |
e258beb2 PM |
1813 | for (mfc = net->ipv4.mfc_unres_queue; mfc; mfc = mfc->next) |
1814 | if (pos-- == 0) | |
1da177e4 LT |
1815 | return mfc; |
1816 | spin_unlock_bh(&mfc_unres_lock); | |
1817 | ||
1818 | it->cache = NULL; | |
1819 | return NULL; | |
1820 | } | |
1821 | ||
1822 | ||
1823 | static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos) | |
1824 | { | |
1825 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 BT |
1826 | struct net *net = seq_file_net(seq); |
1827 | ||
1da177e4 LT |
1828 | it->cache = NULL; |
1829 | it->ct = 0; | |
f6bb4514 | 1830 | return *pos ? ipmr_mfc_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
1831 | : SEQ_START_TOKEN; |
1832 | } | |
1833 | ||
1834 | static void *ipmr_mfc_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
1835 | { | |
1836 | struct mfc_cache *mfc = v; | |
1837 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 | 1838 | struct net *net = seq_file_net(seq); |
1da177e4 LT |
1839 | |
1840 | ++*pos; | |
1841 | ||
1842 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 1843 | return ipmr_mfc_seq_idx(net, seq->private, 0); |
1da177e4 LT |
1844 | |
1845 | if (mfc->next) | |
1846 | return mfc->next; | |
e905a9ed | 1847 | |
e258beb2 | 1848 | if (it->cache == &net->ipv4.mfc_unres_queue) |
1da177e4 LT |
1849 | goto end_of_list; |
1850 | ||
f6bb4514 | 1851 | BUG_ON(it->cache != net->ipv4.mfc_cache_array); |
1da177e4 LT |
1852 | |
1853 | while (++it->ct < MFC_LINES) { | |
f6bb4514 | 1854 | mfc = net->ipv4.mfc_cache_array[it->ct]; |
1da177e4 LT |
1855 | if (mfc) |
1856 | return mfc; | |
1857 | } | |
1858 | ||
1859 | /* exhausted cache_array, show unresolved */ | |
1860 | read_unlock(&mrt_lock); | |
e258beb2 | 1861 | it->cache = &net->ipv4.mfc_unres_queue; |
1da177e4 | 1862 | it->ct = 0; |
e905a9ed | 1863 | |
1da177e4 | 1864 | spin_lock_bh(&mfc_unres_lock); |
e258beb2 | 1865 | mfc = net->ipv4.mfc_unres_queue; |
e905a9ed | 1866 | if (mfc) |
1da177e4 LT |
1867 | return mfc; |
1868 | ||
1869 | end_of_list: | |
1870 | spin_unlock_bh(&mfc_unres_lock); | |
1871 | it->cache = NULL; | |
1872 | ||
1873 | return NULL; | |
1874 | } | |
1875 | ||
1876 | static void ipmr_mfc_seq_stop(struct seq_file *seq, void *v) | |
1877 | { | |
1878 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 | 1879 | struct net *net = seq_file_net(seq); |
1da177e4 | 1880 | |
e258beb2 | 1881 | if (it->cache == &net->ipv4.mfc_unres_queue) |
1da177e4 | 1882 | spin_unlock_bh(&mfc_unres_lock); |
f6bb4514 | 1883 | else if (it->cache == net->ipv4.mfc_cache_array) |
1da177e4 LT |
1884 | read_unlock(&mrt_lock); |
1885 | } | |
1886 | ||
1887 | static int ipmr_mfc_seq_show(struct seq_file *seq, void *v) | |
1888 | { | |
1889 | int n; | |
f6bb4514 | 1890 | struct net *net = seq_file_net(seq); |
1da177e4 LT |
1891 | |
1892 | if (v == SEQ_START_TOKEN) { | |
e905a9ed | 1893 | seq_puts(seq, |
1da177e4 LT |
1894 | "Group Origin Iif Pkts Bytes Wrong Oifs\n"); |
1895 | } else { | |
1896 | const struct mfc_cache *mfc = v; | |
1897 | const struct ipmr_mfc_iter *it = seq->private; | |
e905a9ed | 1898 | |
999890b2 | 1899 | seq_printf(seq, "%08lX %08lX %-3hd", |
1da177e4 LT |
1900 | (unsigned long) mfc->mfc_mcastgrp, |
1901 | (unsigned long) mfc->mfc_origin, | |
1ea472e2 | 1902 | mfc->mfc_parent); |
1da177e4 | 1903 | |
e258beb2 | 1904 | if (it->cache != &net->ipv4.mfc_unres_queue) { |
1ea472e2 BT |
1905 | seq_printf(seq, " %8lu %8lu %8lu", |
1906 | mfc->mfc_un.res.pkt, | |
1907 | mfc->mfc_un.res.bytes, | |
1908 | mfc->mfc_un.res.wrong_if); | |
132adf54 SH |
1909 | for (n = mfc->mfc_un.res.minvif; |
1910 | n < mfc->mfc_un.res.maxvif; n++ ) { | |
f6bb4514 | 1911 | if (VIF_EXISTS(net, n) && |
cf958ae3 BT |
1912 | mfc->mfc_un.res.ttls[n] < 255) |
1913 | seq_printf(seq, | |
e905a9ed | 1914 | " %2d:%-3d", |
1da177e4 LT |
1915 | n, mfc->mfc_un.res.ttls[n]); |
1916 | } | |
1ea472e2 BT |
1917 | } else { |
1918 | /* unresolved mfc_caches don't contain | |
1919 | * pkt, bytes and wrong_if values | |
1920 | */ | |
1921 | seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul); | |
1da177e4 LT |
1922 | } |
1923 | seq_putc(seq, '\n'); | |
1924 | } | |
1925 | return 0; | |
1926 | } | |
1927 | ||
f690808e | 1928 | static const struct seq_operations ipmr_mfc_seq_ops = { |
1da177e4 LT |
1929 | .start = ipmr_mfc_seq_start, |
1930 | .next = ipmr_mfc_seq_next, | |
1931 | .stop = ipmr_mfc_seq_stop, | |
1932 | .show = ipmr_mfc_seq_show, | |
1933 | }; | |
1934 | ||
1935 | static int ipmr_mfc_open(struct inode *inode, struct file *file) | |
1936 | { | |
f6bb4514 BT |
1937 | return seq_open_net(inode, file, &ipmr_mfc_seq_ops, |
1938 | sizeof(struct ipmr_mfc_iter)); | |
1da177e4 LT |
1939 | } |
1940 | ||
9a32144e | 1941 | static const struct file_operations ipmr_mfc_fops = { |
1da177e4 LT |
1942 | .owner = THIS_MODULE, |
1943 | .open = ipmr_mfc_open, | |
1944 | .read = seq_read, | |
1945 | .llseek = seq_lseek, | |
f6bb4514 | 1946 | .release = seq_release_net, |
1da177e4 | 1947 | }; |
e905a9ed | 1948 | #endif |
1da177e4 LT |
1949 | |
1950 | #ifdef CONFIG_IP_PIMSM_V2 | |
32613090 | 1951 | static const struct net_protocol pim_protocol = { |
1da177e4 | 1952 | .handler = pim_rcv, |
403dbb97 | 1953 | .netns_ok = 1, |
1da177e4 LT |
1954 | }; |
1955 | #endif | |
1956 | ||
1957 | ||
1958 | /* | |
1959 | * Setup for IP multicast routing | |
1960 | */ | |
cf958ae3 BT |
1961 | static int __net_init ipmr_net_init(struct net *net) |
1962 | { | |
1963 | int err = 0; | |
1964 | ||
1965 | net->ipv4.vif_table = kcalloc(MAXVIFS, sizeof(struct vif_device), | |
1966 | GFP_KERNEL); | |
1967 | if (!net->ipv4.vif_table) { | |
1968 | err = -ENOMEM; | |
1969 | goto fail; | |
1970 | } | |
2bb8b26c BT |
1971 | |
1972 | /* Forwarding cache */ | |
1973 | net->ipv4.mfc_cache_array = kcalloc(MFC_LINES, | |
1974 | sizeof(struct mfc_cache *), | |
1975 | GFP_KERNEL); | |
1976 | if (!net->ipv4.mfc_cache_array) { | |
1977 | err = -ENOMEM; | |
1978 | goto fail_mfc_cache; | |
1979 | } | |
6c5143db | 1980 | |
e258beb2 PM |
1981 | setup_timer(&net->ipv4.ipmr_expire_timer, ipmr_expire_process, |
1982 | (unsigned long)net); | |
1983 | ||
6c5143db BT |
1984 | #ifdef CONFIG_IP_PIMSM |
1985 | net->ipv4.mroute_reg_vif_num = -1; | |
1986 | #endif | |
f6bb4514 BT |
1987 | |
1988 | #ifdef CONFIG_PROC_FS | |
1989 | err = -ENOMEM; | |
1990 | if (!proc_net_fops_create(net, "ip_mr_vif", 0, &ipmr_vif_fops)) | |
1991 | goto proc_vif_fail; | |
1992 | if (!proc_net_fops_create(net, "ip_mr_cache", 0, &ipmr_mfc_fops)) | |
1993 | goto proc_cache_fail; | |
1994 | #endif | |
2bb8b26c BT |
1995 | return 0; |
1996 | ||
f6bb4514 BT |
1997 | #ifdef CONFIG_PROC_FS |
1998 | proc_cache_fail: | |
1999 | proc_net_remove(net, "ip_mr_vif"); | |
2000 | proc_vif_fail: | |
2001 | kfree(net->ipv4.mfc_cache_array); | |
2002 | #endif | |
2bb8b26c BT |
2003 | fail_mfc_cache: |
2004 | kfree(net->ipv4.vif_table); | |
cf958ae3 BT |
2005 | fail: |
2006 | return err; | |
2007 | } | |
2008 | ||
2009 | static void __net_exit ipmr_net_exit(struct net *net) | |
2010 | { | |
f6bb4514 BT |
2011 | #ifdef CONFIG_PROC_FS |
2012 | proc_net_remove(net, "ip_mr_cache"); | |
2013 | proc_net_remove(net, "ip_mr_vif"); | |
2014 | #endif | |
2bb8b26c | 2015 | kfree(net->ipv4.mfc_cache_array); |
cf958ae3 BT |
2016 | kfree(net->ipv4.vif_table); |
2017 | } | |
2018 | ||
2019 | static struct pernet_operations ipmr_net_ops = { | |
2020 | .init = ipmr_net_init, | |
2021 | .exit = ipmr_net_exit, | |
2022 | }; | |
e905a9ed | 2023 | |
03d2f897 | 2024 | int __init ip_mr_init(void) |
1da177e4 | 2025 | { |
03d2f897 WC |
2026 | int err; |
2027 | ||
1da177e4 LT |
2028 | mrt_cachep = kmem_cache_create("ip_mrt_cache", |
2029 | sizeof(struct mfc_cache), | |
e5d679f3 | 2030 | 0, SLAB_HWCACHE_ALIGN|SLAB_PANIC, |
20c2df83 | 2031 | NULL); |
03d2f897 WC |
2032 | if (!mrt_cachep) |
2033 | return -ENOMEM; | |
2034 | ||
cf958ae3 BT |
2035 | err = register_pernet_subsys(&ipmr_net_ops); |
2036 | if (err) | |
2037 | goto reg_pernet_fail; | |
2038 | ||
03d2f897 WC |
2039 | err = register_netdevice_notifier(&ip_mr_notifier); |
2040 | if (err) | |
2041 | goto reg_notif_fail; | |
403dbb97 TG |
2042 | #ifdef CONFIG_IP_PIMSM_V2 |
2043 | if (inet_add_protocol(&pim_protocol, IPPROTO_PIM) < 0) { | |
2044 | printk(KERN_ERR "ip_mr_init: can't add PIM protocol\n"); | |
2045 | err = -EAGAIN; | |
2046 | goto add_proto_fail; | |
2047 | } | |
2048 | #endif | |
03d2f897 | 2049 | return 0; |
f6bb4514 | 2050 | |
403dbb97 TG |
2051 | #ifdef CONFIG_IP_PIMSM_V2 |
2052 | add_proto_fail: | |
2053 | unregister_netdevice_notifier(&ip_mr_notifier); | |
2054 | #endif | |
c3e38896 | 2055 | reg_notif_fail: |
cf958ae3 BT |
2056 | unregister_pernet_subsys(&ipmr_net_ops); |
2057 | reg_pernet_fail: | |
c3e38896 | 2058 | kmem_cache_destroy(mrt_cachep); |
03d2f897 | 2059 | return err; |
1da177e4 | 2060 | } |