]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blame - net/ipv4/ipmr.c
projects / mirror_ubuntu-jammy-kernel.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
mroute*: Make mr_table a common struct
[mirror_ubuntu-jammy-kernel.git] / net / ipv4 / ipmr.c
CommitLineData
1da177e4
LT		 1/*
2 * IP multicast routing support for mrouted 3.6/3.8
3 *
113aa838 4 * (c) 1995 Alan Cox, <alan@lxorguk.ukuu.org.uk>
1da177e4
LT		 5 * Linux Consultancy and Custom Driver Development
6 *
7 * This program is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation; either version
10 * 2 of the License, or (at your option) any later version.
11 *
1da177e4
LT		 12 * Fixes:
13 * Michael Chastain : Incorrect size of copying.
14 * Alan Cox : Added the cache manager code
15 * Alan Cox : Fixed the clone/copy bug and device race.
16 * Mike McLagan : Routing by source
17 * Malcolm Beattie : Buffer handling fixes.
18 * Alexey Kuznetsov : Double buffer free and other fixes.
19 * SVR Anand : Fixed several multicast bugs and problems.
20 * Alexey Kuznetsov : Status, optimisations and more.
21 * Brad Parker : Better behaviour on mrouted upcall
22 * overflow.
23 * Carlos Picoto : PIMv1 Support
24 * Pavlin Ivanov Radoslavov: PIMv2 Registers must checksum only PIM header
f77f13e2 25 * Relax this requirement to work with older peers.
1da177e4
LT		 26 *
27 */
28
7c0f6ba6 29#include <linux/uaccess.h>
1da177e4 30#include <linux/types.h>
08009a76 31#include <linux/cache.h>
4fc268d2 32#include <linux/capability.h>
1da177e4 33#include <linux/errno.h>
1da177e4
LT		 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/icmp.h>
56#include <net/udp.h>
57#include <net/raw.h>
58#include <linux/notifier.h>
59#include <linux/if_arp.h>
60#include <linux/netfilter_ipv4.h>
709b46e8 61#include <linux/compat.h>
bc3b2d7f 62#include <linux/export.h>
c5441932 63#include <net/ip_tunnels.h>
1da177e4 64#include <net/checksum.h>
dc5fc579 65#include <net/netlink.h>
f0ad0860 66#include <net/fib_rules.h>
d67b8c61 67#include <linux/netconf.h>
ccbb0aa6 68#include <net/nexthop.h>
5d8b3e69 69#include <net/switchdev.h>
1da177e4 70
f0ad0860
PM		 71struct ipmr_rule {
72 struct fib_rule common;
73};
74
75struct ipmr_result {
76 struct mr_table *mrt;
77};
78
1da177e4 79/* Big lock, protecting vif table, mrt cache and mroute socket state.
a8cb16dd 80 * Note that the changes are semaphored via rtnl_lock.
1da177e4
LT		 81 */
82
83static DEFINE_RWLOCK(mrt_lock);
84
7ef8f65d 85/* Multicast router control variables */
1da177e4 86
1da177e4
LT		 87/* Special spinlock for queue of unresolved entries */
88static DEFINE_SPINLOCK(mfc_unres_lock);
89
90/* We return to original Alan's scheme. Hash table of resolved
a8cb16dd
ED		 91 * entries is changed only in process context and protected
92 * with weak lock mrt_lock. Queue of unresolved entries is protected
93 * with strong spinlock mfc_unres_lock.
94 *
95 * In this case data path is free of exclusive locks at all.
1da177e4
LT		 96 */
97
08009a76 98static struct kmem_cache *mrt_cachep __ro_after_init;
1da177e4 99
f0ad0860 100static struct mr_table *ipmr_new_table(struct net *net, u32 id);
acbb219d
FR		 101static void ipmr_free_table(struct mr_table *mrt);
102
c4854ec8 103static void ip_mr_forward(struct net *net, struct mr_table *mrt,
4b1f0d33
DS		 104 struct net_device *dev, struct sk_buff *skb,
105 struct mfc_cache *cache, int local);
0c12295a 106static int ipmr_cache_report(struct mr_table *mrt,
4feb88e5 107 struct sk_buff *pkt, vifi_t vifi, int assert);
cb6a4e46
PM		 108static int __ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb,
109 struct mfc_cache *c, struct rtmsg *rtm);
8cd3ac9f
ND		 110static void mroute_netlink_event(struct mr_table *mrt, struct mfc_cache *mfc,
111 int cmd);
5a645dd8 112static void igmpmsg_netlink_event(struct mr_table *mrt, struct sk_buff *pkt);
0e615e96 113static void mroute_clean_tables(struct mr_table *mrt, bool all);
e99e88a9 114static void ipmr_expire_process(struct timer_list *t);
f0ad0860
PM		 115
116#ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES
117#define ipmr_for_each_table(mrt, net) \
118 list_for_each_entry_rcu(mrt, &net->ipv4.mr_tables, list)
119
120static struct mr_table *ipmr_get_table(struct net *net, u32 id)
121{
122 struct mr_table *mrt;
123
124 ipmr_for_each_table(mrt, net) {
125 if (mrt->id == id)
126 return mrt;
127 }
128 return NULL;
129}
130
da91981b 131static int ipmr_fib_lookup(struct net *net, struct flowi4 *flp4,
f0ad0860
PM		 132 struct mr_table **mrt)
133{
f0ad0860 134 int err;
95f4a45d
HFS		 135 struct ipmr_result res;
136 struct fib_lookup_arg arg = {
137 .result = &res,
138 .flags = FIB_LOOKUP_NOREF,
139 };
f0ad0860 140
e58e4159
DA		 141 /* update flow if oif or iif point to device enslaved to l3mdev */
142 l3mdev_update_flow(net, flowi4_to_flowi(flp4));
143
da91981b
DM		 144 err = fib_rules_lookup(net->ipv4.mr_rules_ops,
145 flowi4_to_flowi(flp4), 0, &arg);
f0ad0860
PM		 146 if (err < 0)
147 return err;
148 *mrt = res.mrt;
149 return 0;
150}
151
152static int ipmr_rule_action(struct fib_rule *rule, struct flowi *flp,
153 int flags, struct fib_lookup_arg *arg)
154{
155 struct ipmr_result *res = arg->result;
156 struct mr_table *mrt;
1da177e4 157
f0ad0860
PM		 158 switch (rule->action) {
159 case FR_ACT_TO_TBL:
160 break;
161 case FR_ACT_UNREACHABLE:
162 return -ENETUNREACH;
163 case FR_ACT_PROHIBIT:
164 return -EACCES;
165 case FR_ACT_BLACKHOLE:
166 default:
167 return -EINVAL;
168 }
169
e58e4159
DA		 170 arg->table = fib_rule_get_table(rule, arg);
171
172 mrt = ipmr_get_table(rule->fr_net, arg->table);
51456b29 173 if (!mrt)
f0ad0860
PM		 174 return -EAGAIN;
175 res->mrt = mrt;
176 return 0;
177}
178
179static int ipmr_rule_match(struct fib_rule *rule, struct flowi *fl, int flags)
180{
181 return 1;
182}
183
184static const struct nla_policy ipmr_rule_policy[FRA_MAX + 1] = {
185 FRA_GENERIC_POLICY,
186};
187
188static int ipmr_rule_configure(struct fib_rule *rule, struct sk_buff *skb,
189 struct fib_rule_hdr *frh, struct nlattr **tb)
190{
191 return 0;
192}
193
194static int ipmr_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh,
195 struct nlattr **tb)
196{
197 return 1;
198}
199
200static int ipmr_rule_fill(struct fib_rule *rule, struct sk_buff *skb,
201 struct fib_rule_hdr *frh)
202{
203 frh->dst_len = 0;
204 frh->src_len = 0;
205 frh->tos = 0;
206 return 0;
207}
208
04a6f82c 209static const struct fib_rules_ops __net_initconst ipmr_rules_ops_template = {
25239cee 210 .family = RTNL_FAMILY_IPMR,
f0ad0860
PM		 211 .rule_size = sizeof(struct ipmr_rule),
212 .addr_size = sizeof(u32),
213 .action = ipmr_rule_action,
214 .match = ipmr_rule_match,
215 .configure = ipmr_rule_configure,
216 .compare = ipmr_rule_compare,
f0ad0860
PM		 217 .fill = ipmr_rule_fill,
218 .nlgroup = RTNLGRP_IPV4_RULE,
219 .policy = ipmr_rule_policy,
220 .owner = THIS_MODULE,
221};
222
223static int __net_init ipmr_rules_init(struct net *net)
224{
225 struct fib_rules_ops *ops;
226 struct mr_table *mrt;
227 int err;
228
229 ops = fib_rules_register(&ipmr_rules_ops_template, net);
230 if (IS_ERR(ops))
231 return PTR_ERR(ops);
232
233 INIT_LIST_HEAD(&net->ipv4.mr_tables);
234
235 mrt = ipmr_new_table(net, RT_TABLE_DEFAULT);
1113ebbc
NA		 236 if (IS_ERR(mrt)) {
237 err = PTR_ERR(mrt);
f0ad0860
PM		 238 goto err1;
239 }
240
241 err = fib_default_rule_add(ops, 0x7fff, RT_TABLE_DEFAULT, 0);
242 if (err < 0)
243 goto err2;
244
245 net->ipv4.mr_rules_ops = ops;
246 return 0;
247
248err2:
f243e5a7 249 ipmr_free_table(mrt);
f0ad0860
PM		 250err1:
251 fib_rules_unregister(ops);
252 return err;
253}
254
255static void __net_exit ipmr_rules_exit(struct net *net)
256{
257 struct mr_table *mrt, *next;
258
ed785309 259 rtnl_lock();
035320d5
ED		 260 list_for_each_entry_safe(mrt, next, &net->ipv4.mr_tables, list) {
261 list_del(&mrt->list);
acbb219d 262 ipmr_free_table(mrt);
035320d5 263 }
f0ad0860 264 fib_rules_unregister(net->ipv4.mr_rules_ops);
419df12f 265 rtnl_unlock();
f0ad0860 266}
4d65b948
YG		 267
268static int ipmr_rules_dump(struct net *net, struct notifier_block *nb)
269{
270 return fib_rules_dump(net, nb, RTNL_FAMILY_IPMR);
271}
272
273static unsigned int ipmr_rules_seq_read(struct net *net)
274{
275 return fib_rules_seq_read(net, RTNL_FAMILY_IPMR);
276}
478e4c2f
YG		 277
278bool ipmr_rule_default(const struct fib_rule *rule)
279{
280 return fib_rule_matchall(rule) && rule->table == RT_TABLE_DEFAULT;
281}
282EXPORT_SYMBOL(ipmr_rule_default);
f0ad0860
PM		 283#else
284#define ipmr_for_each_table(mrt, net) \
285 for (mrt = net->ipv4.mrt; mrt; mrt = NULL)
286
287static struct mr_table *ipmr_get_table(struct net *net, u32 id)
288{
289 return net->ipv4.mrt;
290}
291
da91981b 292static int ipmr_fib_lookup(struct net *net, struct flowi4 *flp4,
f0ad0860
PM		 293 struct mr_table **mrt)
294{
295 *mrt = net->ipv4.mrt;
296 return 0;
297}
298
299static int __net_init ipmr_rules_init(struct net *net)
300{
1113ebbc
NA		 301 struct mr_table *mrt;
302
303 mrt = ipmr_new_table(net, RT_TABLE_DEFAULT);
304 if (IS_ERR(mrt))
305 return PTR_ERR(mrt);
306 net->ipv4.mrt = mrt;
307 return 0;
f0ad0860
PM		 308}
309
310static void __net_exit ipmr_rules_exit(struct net *net)
311{
ed785309 312 rtnl_lock();
acbb219d 313 ipmr_free_table(net->ipv4.mrt);
ed785309
WC		 314 net->ipv4.mrt = NULL;
315 rtnl_unlock();
f0ad0860 316}
4d65b948
YG		 317
318static int ipmr_rules_dump(struct net *net, struct notifier_block *nb)
319{
320 return 0;
321}
322
323static unsigned int ipmr_rules_seq_read(struct net *net)
324{
325 return 0;
326}
478e4c2f
YG		 327
328bool ipmr_rule_default(const struct fib_rule *rule)
329{
330 return true;
331}
332EXPORT_SYMBOL(ipmr_rule_default);
f0ad0860
PM		 333#endif
334
8fb472c0
NA		 335static inline int ipmr_hash_cmp(struct rhashtable_compare_arg *arg,
336 const void *ptr)
337{
338 const struct mfc_cache_cmp_arg *cmparg = arg->key;
339 struct mfc_cache *c = (struct mfc_cache *)ptr;
340
341 return cmparg->mfc_mcastgrp != c->mfc_mcastgrp ||
342 cmparg->mfc_origin != c->mfc_origin;
343}
344
345static const struct rhashtable_params ipmr_rht_params = {
346 .head_offset = offsetof(struct mfc_cache, mnode),
347 .key_offset = offsetof(struct mfc_cache, cmparg),
348 .key_len = sizeof(struct mfc_cache_cmp_arg),
349 .nelem_hint = 3,
350 .locks_mul = 1,
351 .obj_cmpfn = ipmr_hash_cmp,
352 .automatic_shrinking = true,
353};
354
f0ad0860
PM		 355static struct mr_table *ipmr_new_table(struct net *net, u32 id)
356{
357 struct mr_table *mrt;
1da177e4 358
1113ebbc
NA		 359 /* "pimreg%u" should not exceed 16 bytes (IFNAMSIZ) */
360 if (id != RT_TABLE_DEFAULT && id >= 1000000000)
361 return ERR_PTR(-EINVAL);
362
f0ad0860 363 mrt = ipmr_get_table(net, id);
00db4124 364 if (mrt)
f0ad0860
PM		 365 return mrt;
366
367 mrt = kzalloc(sizeof(*mrt), GFP_KERNEL);
51456b29 368 if (!mrt)
1113ebbc 369 return ERR_PTR(-ENOMEM);
8de53dfb 370 write_pnet(&mrt->net, net);
f0ad0860
PM		 371 mrt->id = id;
372
8fb472c0
NA		 373 rhltable_init(&mrt->mfc_hash, &ipmr_rht_params);
374 INIT_LIST_HEAD(&mrt->mfc_cache_list);
f0ad0860
PM		 375 INIT_LIST_HEAD(&mrt->mfc_unres_queue);
376
e99e88a9 377 timer_setup(&mrt->ipmr_expire_timer, ipmr_expire_process, 0);
f0ad0860 378
f0ad0860 379 mrt->mroute_reg_vif_num = -1;
f0ad0860
PM		 380#ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES
381 list_add_tail_rcu(&mrt->list, &net->ipv4.mr_tables);
382#endif
383 return mrt;
384}
1da177e4 385
acbb219d
FR		 386static void ipmr_free_table(struct mr_table *mrt)
387{
388 del_timer_sync(&mrt->ipmr_expire_timer);
0e615e96 389 mroute_clean_tables(mrt, true);
8fb472c0 390 rhltable_destroy(&mrt->mfc_hash);
acbb219d
FR		 391 kfree(mrt);
392}
393
1da177e4
LT		 394/* Service routines creating virtual interfaces: DVMRP tunnels and PIMREG */
395
d607032d
WC		 396static void ipmr_del_tunnel(struct net_device *dev, struct vifctl *v)
397{
4feb88e5
BT		 398 struct net *net = dev_net(dev);
399
d607032d
WC		 400 dev_close(dev);
401
4feb88e5 402 dev = __dev_get_by_name(net, "tunl0");
d607032d 403 if (dev) {
5bc3eb7e 404 const struct net_device_ops *ops = dev->netdev_ops;
d607032d 405 struct ifreq ifr;
d607032d
WC		 406 struct ip_tunnel_parm p;
407
408 memset(&p, 0, sizeof(p));
409 p.iph.daddr = v->vifc_rmt_addr.s_addr;
410 p.iph.saddr = v->vifc_lcl_addr.s_addr;
411 p.iph.version = 4;
412 p.iph.ihl = 5;
413 p.iph.protocol = IPPROTO_IPIP;
414 sprintf(p.name, "dvmrp%d", v->vifc_vifi);
415 ifr.ifr_ifru.ifru_data = (__force void __user *)&p;
416
5bc3eb7e
SH		 417 if (ops->ndo_do_ioctl) {
418 mm_segment_t oldfs = get_fs();
419
420 set_fs(KERNEL_DS);
421 ops->ndo_do_ioctl(dev, &ifr, SIOCDELTUNNEL);
422 set_fs(oldfs);
423 }
d607032d
WC		 424 }
425}
426
a0b47736
NA		 427/* Initialize ipmr pimreg/tunnel in_device */
428static bool ipmr_init_vif_indev(const struct net_device *dev)
429{
430 struct in_device *in_dev;
431
432 ASSERT_RTNL();
433
434 in_dev = __in_dev_get_rtnl(dev);
435 if (!in_dev)
436 return false;
437 ipv4_devconf_setall(in_dev);
438 neigh_parms_data_state_setall(in_dev->arp_parms);
439 IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0;
440
441 return true;
442}
443
7ef8f65d 444static struct net_device *ipmr_new_tunnel(struct net *net, struct vifctl *v)
1da177e4
LT		 445{
446 struct net_device *dev;
447
4feb88e5 448 dev = __dev_get_by_name(net, "tunl0");
1da177e4
LT		 449
450 if (dev) {
5bc3eb7e 451 const struct net_device_ops *ops = dev->netdev_ops;
1da177e4
LT		 452 int err;
453 struct ifreq ifr;
1da177e4 454 struct ip_tunnel_parm p;
1da177e4
LT		 455
456 memset(&p, 0, sizeof(p));
457 p.iph.daddr = v->vifc_rmt_addr.s_addr;
458 p.iph.saddr = v->vifc_lcl_addr.s_addr;
459 p.iph.version = 4;
460 p.iph.ihl = 5;
461 p.iph.protocol = IPPROTO_IPIP;
462 sprintf(p.name, "dvmrp%d", v->vifc_vifi);
ba93ef74 463 ifr.ifr_ifru.ifru_data = (__force void __user *)&p;
1da177e4 464
5bc3eb7e
SH		 465 if (ops->ndo_do_ioctl) {
466 mm_segment_t oldfs = get_fs();
467
468 set_fs(KERNEL_DS);
469 err = ops->ndo_do_ioctl(dev, &ifr, SIOCADDTUNNEL);
470 set_fs(oldfs);
a8cb16dd 471 } else {
5bc3eb7e 472 err = -EOPNOTSUPP;
a8cb16dd 473 }
1da177e4
LT		 474 dev = NULL;
475
4feb88e5
BT		 476 if (err == 0 &&
477 (dev = __dev_get_by_name(net, p.name)) != NULL) {
1da177e4 478 dev->flags |= IFF_MULTICAST;
a0b47736 479 if (!ipmr_init_vif_indev(dev))
1da177e4 480 goto failure;
1da177e4
LT		 481 if (dev_open(dev))
482 goto failure;
7dc00c82 483 dev_hold(dev);
1da177e4
LT		 484 }
485 }
486 return dev;
487
488failure:
1da177e4
LT		 489 unregister_netdevice(dev);
490 return NULL;
491}
492
c316c629 493#if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2)
6fef4c0c 494static netdev_tx_t reg_vif_xmit(struct sk_buff *skb, struct net_device *dev)
1da177e4 495{
4feb88e5 496 struct net *net = dev_net(dev);
f0ad0860 497 struct mr_table *mrt;
da91981b
DM		 498 struct flowi4 fl4 = {
499 .flowi4_oif = dev->ifindex,
6a662719 500 .flowi4_iif = skb->skb_iif ? : LOOPBACK_IFINDEX,
da91981b 501 .flowi4_mark = skb->mark,
f0ad0860
PM		 502 };
503 int err;
504
da91981b 505 err = ipmr_fib_lookup(net, &fl4, &mrt);
e40dbc51
BG		 506 if (err < 0) {
507 kfree_skb(skb);
f0ad0860 508 return err;
e40dbc51 509 }
4feb88e5 510
1da177e4 511 read_lock(&mrt_lock);
cf3677ae
PE		 512 dev->stats.tx_bytes += skb->len;
513 dev->stats.tx_packets++;
0c12295a 514 ipmr_cache_report(mrt, skb, mrt->mroute_reg_vif_num, IGMPMSG_WHOLEPKT);
1da177e4
LT		 515 read_unlock(&mrt_lock);
516 kfree_skb(skb);
6ed10654 517 return NETDEV_TX_OK;
1da177e4
LT		 518}
519
ee9b9596
ND		 520static int reg_vif_get_iflink(const struct net_device *dev)
521{
522 return 0;
523}
524
007c3838
SH		 525static const struct net_device_ops reg_vif_netdev_ops = {
526 .ndo_start_xmit = reg_vif_xmit,
ee9b9596 527 .ndo_get_iflink = reg_vif_get_iflink,
007c3838
SH		 528};
529
1da177e4
LT		 530static void reg_vif_setup(struct net_device *dev)
531{
532 dev->type = ARPHRD_PIMREG;
46f25dff 533 dev->mtu = ETH_DATA_LEN - sizeof(struct iphdr) - 8;
1da177e4 534 dev->flags = IFF_NOARP;
70cb4a45 535 dev->netdev_ops = &reg_vif_netdev_ops;
cf124db5 536 dev->needs_free_netdev = true;
403dbb97 537 dev->features |= NETIF_F_NETNS_LOCAL;
1da177e4
LT		 538}
539
f0ad0860 540static struct net_device *ipmr_reg_vif(struct net *net, struct mr_table *mrt)
1da177e4
LT		 541{
542 struct net_device *dev;
f0ad0860 543 char name[IFNAMSIZ];
1da177e4 544
f0ad0860
PM		 545 if (mrt->id == RT_TABLE_DEFAULT)
546 sprintf(name, "pimreg");
547 else
548 sprintf(name, "pimreg%u", mrt->id);
1da177e4 549
c835a677 550 dev = alloc_netdev(0, name, NET_NAME_UNKNOWN, reg_vif_setup);
1da177e4 551
51456b29 552 if (!dev)
1da177e4
LT		 553 return NULL;
554
403dbb97
TG		 555 dev_net_set(dev, net);
556
1da177e4
LT		 557 if (register_netdevice(dev)) {
558 free_netdev(dev);
559 return NULL;
560 }
1da177e4 561
a0b47736 562 if (!ipmr_init_vif_indev(dev))
1da177e4 563 goto failure;
1da177e4
LT		 564 if (dev_open(dev))
565 goto failure;
566
7dc00c82
WC		 567 dev_hold(dev);
568
1da177e4
LT		 569 return dev;
570
571failure:
1da177e4
LT		 572 unregister_netdevice(dev);
573 return NULL;
574}
c316c629
NA		 575
576/* called with rcu_read_lock() */
577static int __pim_rcv(struct mr_table *mrt, struct sk_buff *skb,
578 unsigned int pimlen)
579{
580 struct net_device *reg_dev = NULL;
581 struct iphdr *encap;
582
583 encap = (struct iphdr *)(skb_transport_header(skb) + pimlen);
7ef8f65d 584 /* Check that:
c316c629
NA		 585 * a. packet is really sent to a multicast group
586 * b. packet is not a NULL-REGISTER
587 * c. packet is not truncated
588 */
589 if (!ipv4_is_multicast(encap->daddr) ||
590 encap->tot_len == 0 ||
591 ntohs(encap->tot_len) + pimlen > skb->len)
592 return 1;
593
594 read_lock(&mrt_lock);
595 if (mrt->mroute_reg_vif_num >= 0)
596 reg_dev = mrt->vif_table[mrt->mroute_reg_vif_num].dev;
597 read_unlock(&mrt_lock);
598
599 if (!reg_dev)
600 return 1;
601
602 skb->mac_header = skb->network_header;
603 skb_pull(skb, (u8 *)encap - skb->data);
604 skb_reset_network_header(skb);
605 skb->protocol = htons(ETH_P_IP);
606 skb->ip_summed = CHECKSUM_NONE;
607
608 skb_tunnel_rx(skb, reg_dev, dev_net(reg_dev));
609
610 netif_rx(skb);
611
612 return NET_RX_SUCCESS;
613}
614#else
615static struct net_device *ipmr_reg_vif(struct net *net, struct mr_table *mrt)
616{
617 return NULL;
618}
1da177e4
LT		 619#endif
620
4d65b948
YG		 621static int call_ipmr_vif_entry_notifier(struct notifier_block *nb,
622 struct net *net,
623 enum fib_event_type event_type,
624 struct vif_device *vif,
625 vifi_t vif_index, u32 tb_id)
626{
627 struct vif_entry_notifier_info info = {
628 .info = {
629 .family = RTNL_FAMILY_IPMR,
630 .net = net,
631 },
632 .dev = vif->dev,
633 .vif_index = vif_index,
634 .vif_flags = vif->flags,
635 .tb_id = tb_id,
636 };
637
638 return call_fib_notifier(nb, net, event_type, &info.info);
639}
640
b362053a
YG		 641static int call_ipmr_vif_entry_notifiers(struct net *net,
642 enum fib_event_type event_type,
643 struct vif_device *vif,
644 vifi_t vif_index, u32 tb_id)
645{
646 struct vif_entry_notifier_info info = {
647 .info = {
648 .family = RTNL_FAMILY_IPMR,
649 .net = net,
650 },
651 .dev = vif->dev,
652 .vif_index = vif_index,
653 .vif_flags = vif->flags,
654 .tb_id = tb_id,
655 };
656
657 ASSERT_RTNL();
658 net->ipv4.ipmr_seq++;
659 return call_fib_notifiers(net, event_type, &info.info);
660}
661
4d65b948
YG		 662static int call_ipmr_mfc_entry_notifier(struct notifier_block *nb,
663 struct net *net,
664 enum fib_event_type event_type,
665 struct mfc_cache *mfc, u32 tb_id)
666{
667 struct mfc_entry_notifier_info info = {
668 .info = {
669 .family = RTNL_FAMILY_IPMR,
670 .net = net,
671 },
672 .mfc = mfc,
673 .tb_id = tb_id
674 };
675
676 return call_fib_notifier(nb, net, event_type, &info.info);
677}
678
b362053a
YG		 679static int call_ipmr_mfc_entry_notifiers(struct net *net,
680 enum fib_event_type event_type,
681 struct mfc_cache *mfc, u32 tb_id)
682{
683 struct mfc_entry_notifier_info info = {
684 .info = {
685 .family = RTNL_FAMILY_IPMR,
686 .net = net,
687 },
688 .mfc = mfc,
689 .tb_id = tb_id
690 };
691
692 ASSERT_RTNL();
693 net->ipv4.ipmr_seq++;
694 return call_fib_notifiers(net, event_type, &info.info);
695}
696
2c53040f
BH		 697/**
698 * vif_delete - Delete a VIF entry
7dc00c82 699 * @notify: Set to 1, if the caller is a notifier_call
1da177e4 700 */
0c12295a 701static int vif_delete(struct mr_table *mrt, int vifi, int notify,
d17fa6fa 702 struct list_head *head)
1da177e4 703{
b362053a 704 struct net *net = read_pnet(&mrt->net);
1da177e4
LT		 705 struct vif_device *v;
706 struct net_device *dev;
707 struct in_device *in_dev;
708
0c12295a 709 if (vifi < 0 || vifi >= mrt->maxvif)
1da177e4
LT		 710 return -EADDRNOTAVAIL;
711
0c12295a 712 v = &mrt->vif_table[vifi];
1da177e4 713
b362053a
YG		 714 if (VIF_EXISTS(mrt, vifi))
715 call_ipmr_vif_entry_notifiers(net, FIB_EVENT_VIF_DEL, v, vifi,
716 mrt->id);
717
1da177e4
LT		 718 write_lock_bh(&mrt_lock);
719 dev = v->dev;
720 v->dev = NULL;
721
722 if (!dev) {
723 write_unlock_bh(&mrt_lock);
724 return -EADDRNOTAVAIL;
725 }
726
0c12295a
PM		 727 if (vifi == mrt->mroute_reg_vif_num)
728 mrt->mroute_reg_vif_num = -1;
1da177e4 729
a8cb16dd 730 if (vifi + 1 == mrt->maxvif) {
1da177e4 731 int tmp;
a8cb16dd
ED		 732
733 for (tmp = vifi - 1; tmp >= 0; tmp--) {
0c12295a 734 if (VIF_EXISTS(mrt, tmp))
1da177e4
LT		 735 break;
736 }
0c12295a 737 mrt->maxvif = tmp+1;
1da177e4
LT		 738 }
739
740 write_unlock_bh(&mrt_lock);
741
742 dev_set_allmulti(dev, -1);
743
a8cb16dd
ED		 744 in_dev = __in_dev_get_rtnl(dev);
745 if (in_dev) {
42f811b8 746 IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)--;
3b022865 747 inet_netconf_notify_devconf(dev_net(dev), RTM_NEWNETCONF,
d67b8c61
ND		 748 NETCONFA_MC_FORWARDING,
749 dev->ifindex, &in_dev->cnf);
1da177e4
LT		 750 ip_rt_multicast_event(in_dev);
751 }
752
a8cb16dd 753 if (v->flags & (VIFF_TUNNEL | VIFF_REGISTER) && !notify)
d17fa6fa 754 unregister_netdevice_queue(dev, head);
1da177e4
LT		 755
756 dev_put(dev);
757 return 0;
758}
759
a8c9486b 760static void ipmr_cache_free_rcu(struct rcu_head *head)
5c0a66f5 761{
a8c9486b
ED		 762 struct mfc_cache *c = container_of(head, struct mfc_cache, rcu);
763
5c0a66f5
BT		 764 kmem_cache_free(mrt_cachep, c);
765}
766
310ebbba 767void ipmr_cache_free(struct mfc_cache *c)
a8c9486b
ED		 768{
769 call_rcu(&c->rcu, ipmr_cache_free_rcu);
770}
310ebbba 771EXPORT_SYMBOL(ipmr_cache_free);
a8c9486b 772
1da177e4 773/* Destroy an unresolved cache entry, killing queued skbs
a8cb16dd 774 * and reporting error to netlink readers.
1da177e4 775 */
0c12295a 776static void ipmr_destroy_unres(struct mr_table *mrt, struct mfc_cache *c)
1da177e4 777{
8de53dfb 778 struct net *net = read_pnet(&mrt->net);
1da177e4 779 struct sk_buff *skb;
9ef1d4c7 780 struct nlmsgerr *e;
1da177e4 781
0c12295a 782 atomic_dec(&mrt->cache_resolve_queue_len);
1da177e4 783
c354e124 784 while ((skb = skb_dequeue(&c->mfc_un.unres.unresolved))) {
eddc9ec5 785 if (ip_hdr(skb)->version == 0) {
af72868b
JB		 786 struct nlmsghdr *nlh = skb_pull(skb,
787 sizeof(struct iphdr));
1da177e4 788 nlh->nlmsg_type = NLMSG_ERROR;
573ce260 789 nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr));
1da177e4 790 skb_trim(skb, nlh->nlmsg_len);
573ce260 791 e = nlmsg_data(nlh);
9ef1d4c7
PM		 792 e->error = -ETIMEDOUT;
793 memset(&e->msg, 0, sizeof(e->msg));
2942e900 794
15e47304 795 rtnl_unicast(skb, net, NETLINK_CB(skb).portid);
a8cb16dd 796 } else {
1da177e4 797 kfree_skb(skb);
a8cb16dd 798 }
1da177e4
LT		 799 }
800
5c0a66f5 801 ipmr_cache_free(c);
1da177e4
LT		 802}
803
e258beb2 804/* Timer process for the unresolved queue. */
e99e88a9 805static void ipmr_expire_process(struct timer_list *t)
1da177e4 806{
e99e88a9 807 struct mr_table *mrt = from_timer(mrt, t, ipmr_expire_timer);
1da177e4
LT		 808 unsigned long now;
809 unsigned long expires;
862465f2 810 struct mfc_cache *c, *next;
1da177e4
LT		 811
812 if (!spin_trylock(&mfc_unres_lock)) {
0c12295a 813 mod_timer(&mrt->ipmr_expire_timer, jiffies+HZ/10);
1da177e4
LT		 814 return;
815 }
816
0c12295a 817 if (list_empty(&mrt->mfc_unres_queue))
1da177e4
LT		 818 goto out;
819
820 now = jiffies;
821 expires = 10*HZ;
1da177e4 822
0c12295a 823 list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) {
1da177e4
LT		 824 if (time_after(c->mfc_un.unres.expires, now)) {
825 unsigned long interval = c->mfc_un.unres.expires - now;
826 if (interval < expires)
827 expires = interval;
1da177e4
LT		 828 continue;
829 }
830
862465f2 831 list_del(&c->list);
8cd3ac9f 832 mroute_netlink_event(mrt, c, RTM_DELROUTE);
0c12295a 833 ipmr_destroy_unres(mrt, c);
1da177e4
LT		 834 }
835
0c12295a
PM		 836 if (!list_empty(&mrt->mfc_unres_queue))
837 mod_timer(&mrt->ipmr_expire_timer, jiffies + expires);
1da177e4
LT		 838
839out:
840 spin_unlock(&mfc_unres_lock);
841}
842
843/* Fill oifs list. It is called under write locked mrt_lock. */
0c12295a 844static void ipmr_update_thresholds(struct mr_table *mrt, struct mfc_cache *cache,
d658f8a0 845 unsigned char *ttls)
1da177e4
LT		 846{
847 int vifi;
848
849 cache->mfc_un.res.minvif = MAXVIFS;
850 cache->mfc_un.res.maxvif = 0;
851 memset(cache->mfc_un.res.ttls, 255, MAXVIFS);
852
0c12295a
PM		 853 for (vifi = 0; vifi < mrt->maxvif; vifi++) {
854 if (VIF_EXISTS(mrt, vifi) &&
cf958ae3 855 ttls[vifi] && ttls[vifi] < 255) {
1da177e4
LT		 856 cache->mfc_un.res.ttls[vifi] = ttls[vifi];
857 if (cache->mfc_un.res.minvif > vifi)
858 cache->mfc_un.res.minvif = vifi;
859 if (cache->mfc_un.res.maxvif <= vifi)
860 cache->mfc_un.res.maxvif = vifi + 1;
861 }
862 }
90b5ca17 863 cache->mfc_un.res.lastuse = jiffies;
1da177e4
LT		 864}
865
0c12295a
PM		 866static int vif_add(struct net *net, struct mr_table *mrt,
867 struct vifctl *vifc, int mrtsock)
1da177e4
LT		 868{
869 int vifi = vifc->vifc_vifi;
5d8b3e69
YG		 870 struct switchdev_attr attr = {
871 .id = SWITCHDEV_ATTR_ID_PORT_PARENT_ID,
872 };
0c12295a 873 struct vif_device *v = &mrt->vif_table[vifi];
1da177e4
LT		 874 struct net_device *dev;
875 struct in_device *in_dev;
d607032d 876 int err;
1da177e4
LT		 877
878 /* Is vif busy ? */
0c12295a 879 if (VIF_EXISTS(mrt, vifi))
1da177e4
LT		 880 return -EADDRINUSE;
881
882 switch (vifc->vifc_flags) {
1da177e4 883 case VIFF_REGISTER:
1973a4ea 884 if (!ipmr_pimsm_enabled())
c316c629
NA		 885 return -EINVAL;
886 /* Special Purpose VIF in PIM
1da177e4
LT		 887 * All the packets will be sent to the daemon
888 */
0c12295a 889 if (mrt->mroute_reg_vif_num >= 0)
1da177e4 890 return -EADDRINUSE;
f0ad0860 891 dev = ipmr_reg_vif(net, mrt);
1da177e4
LT		 892 if (!dev)
893 return -ENOBUFS;
d607032d
WC		 894 err = dev_set_allmulti(dev, 1);
895 if (err) {
896 unregister_netdevice(dev);
7dc00c82 897 dev_put(dev);
d607032d
WC		 898 return err;
899 }
1da177e4 900 break;
e905a9ed 901 case VIFF_TUNNEL:
4feb88e5 902 dev = ipmr_new_tunnel(net, vifc);
1da177e4
LT		 903 if (!dev)
904 return -ENOBUFS;
d607032d
WC		 905 err = dev_set_allmulti(dev, 1);
906 if (err) {
907 ipmr_del_tunnel(dev, vifc);
7dc00c82 908 dev_put(dev);
d607032d
WC		 909 return err;
910 }
1da177e4 911 break;
ee5e81f0 912 case VIFF_USE_IFINDEX:
1da177e4 913 case 0:
ee5e81f0
I		 914 if (vifc->vifc_flags == VIFF_USE_IFINDEX) {
915 dev = dev_get_by_index(net, vifc->vifc_lcl_ifindex);
51456b29 916 if (dev && !__in_dev_get_rtnl(dev)) {
ee5e81f0
I		 917 dev_put(dev);
918 return -EADDRNOTAVAIL;
919 }
a8cb16dd 920 } else {
ee5e81f0 921 dev = ip_dev_find(net, vifc->vifc_lcl_addr.s_addr);
a8cb16dd 922 }
1da177e4
LT		 923 if (!dev)
924 return -EADDRNOTAVAIL;
d607032d 925 err = dev_set_allmulti(dev, 1);
7dc00c82
WC		 926 if (err) {
927 dev_put(dev);
d607032d 928 return err;
7dc00c82 929 }
1da177e4
LT		 930 break;
931 default:
932 return -EINVAL;
933 }
934
a8cb16dd
ED		 935 in_dev = __in_dev_get_rtnl(dev);
936 if (!in_dev) {
d0490cfd 937 dev_put(dev);
1da177e4 938 return -EADDRNOTAVAIL;
d0490cfd 939 }
42f811b8 940 IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)++;
3b022865
DA		 941 inet_netconf_notify_devconf(net, RTM_NEWNETCONF, NETCONFA_MC_FORWARDING,
942 dev->ifindex, &in_dev->cnf);
1da177e4
LT		 943 ip_rt_multicast_event(in_dev);
944
a8cb16dd 945 /* Fill in the VIF structures */
6853f21f
YM		 946 vif_device_init(v, dev, vifc->vifc_rate_limit,
947 vifc->vifc_threshold,
948 vifc->vifc_flags | (!mrtsock ? VIFF_STATIC : 0),
949 (VIFF_TUNNEL | VIFF_REGISTER));
a8cb16dd 950
5d8b3e69
YG		 951 attr.orig_dev = dev;
952 if (!switchdev_port_attr_get(dev, &attr)) {
953 memcpy(v->dev_parent_id.id, attr.u.ppid.id, attr.u.ppid.id_len);
954 v->dev_parent_id.id_len = attr.u.ppid.id_len;
955 } else {
956 v->dev_parent_id.id_len = 0;
957 }
6853f21f 958
c354e124
JK		 959 v->local = vifc->vifc_lcl_addr.s_addr;
960 v->remote = vifc->vifc_rmt_addr.s_addr;
1da177e4
LT		 961
962 /* And finish update writing critical data */
963 write_lock_bh(&mrt_lock);
c354e124 964 v->dev = dev;
a8cb16dd 965 if (v->flags & VIFF_REGISTER)
0c12295a 966 mrt->mroute_reg_vif_num = vifi;
0c12295a
PM		 967 if (vifi+1 > mrt->maxvif)
968 mrt->maxvif = vifi+1;
1da177e4 969 write_unlock_bh(&mrt_lock);
b362053a 970 call_ipmr_vif_entry_notifiers(net, FIB_EVENT_VIF_ADD, v, vifi, mrt->id);
1da177e4
LT		 971 return 0;
972}
973
a8c9486b 974/* called with rcu_read_lock() */
0c12295a 975static struct mfc_cache *ipmr_cache_find(struct mr_table *mrt,
4feb88e5
BT		 976 __be32 origin,
977 __be32 mcastgrp)
1da177e4 978{
8fb472c0
NA		 979 struct mfc_cache_cmp_arg arg = {
980 .mfc_mcastgrp = mcastgrp,
981 .mfc_origin = origin
982 };
983 struct rhlist_head *tmp, *list;
1da177e4
LT		 984 struct mfc_cache *c;
985
8fb472c0
NA		 986 list = rhltable_lookup(&mrt->mfc_hash, &arg, ipmr_rht_params);
987 rhl_for_each_entry_rcu(c, tmp, list, mnode)
988 return c;
989
862465f2 990 return NULL;
1da177e4
LT		 991}
992
660b26dc
ND		 993/* Look for a (*,*,oif) entry */
994static struct mfc_cache *ipmr_cache_find_any_parent(struct mr_table *mrt,
995 int vifi)
996{
8fb472c0
NA		 997 struct mfc_cache_cmp_arg arg = {
998 .mfc_mcastgrp = htonl(INADDR_ANY),
999 .mfc_origin = htonl(INADDR_ANY)
1000 };
1001 struct rhlist_head *tmp, *list;
660b26dc
ND		 1002 struct mfc_cache *c;
1003
8fb472c0
NA		 1004 list = rhltable_lookup(&mrt->mfc_hash, &arg, ipmr_rht_params);
1005 rhl_for_each_entry_rcu(c, tmp, list, mnode)
1006 if (c->mfc_un.res.ttls[vifi] < 255)
660b26dc
ND		 1007 return c;
1008
1009 return NULL;
1010}
1011
1012/* Look for a (*,G) entry */
1013static struct mfc_cache *ipmr_cache_find_any(struct mr_table *mrt,
1014 __be32 mcastgrp, int vifi)
1015{
8fb472c0
NA		 1016 struct mfc_cache_cmp_arg arg = {
1017 .mfc_mcastgrp = mcastgrp,
1018 .mfc_origin = htonl(INADDR_ANY)
1019 };
1020 struct rhlist_head *tmp, *list;
660b26dc
ND		 1021 struct mfc_cache *c, *proxy;
1022
360eb5da 1023 if (mcastgrp == htonl(INADDR_ANY))
660b26dc
ND		 1024 goto skip;
1025
8fb472c0
NA		 1026 list = rhltable_lookup(&mrt->mfc_hash, &arg, ipmr_rht_params);
1027 rhl_for_each_entry_rcu(c, tmp, list, mnode) {
1028 if (c->mfc_un.res.ttls[vifi] < 255)
1029 return c;
1030
1031 /* It's ok if the vifi is part of the static tree */
1032 proxy = ipmr_cache_find_any_parent(mrt, c->mfc_parent);
1033 if (proxy && proxy->mfc_un.res.ttls[vifi] < 255)
1034 return c;
1035 }
660b26dc
ND		 1036
1037skip:
1038 return ipmr_cache_find_any_parent(mrt, vifi);
1039}
1040
8fb472c0
NA		 1041/* Look for a (S,G,iif) entry if parent != -1 */
1042static struct mfc_cache *ipmr_cache_find_parent(struct mr_table *mrt,
1043 __be32 origin, __be32 mcastgrp,
1044 int parent)
1045{
1046 struct mfc_cache_cmp_arg arg = {
1047 .mfc_mcastgrp = mcastgrp,
1048 .mfc_origin = origin,
1049 };
1050 struct rhlist_head *tmp, *list;
1051 struct mfc_cache *c;
1052
1053 list = rhltable_lookup(&mrt->mfc_hash, &arg, ipmr_rht_params);
1054 rhl_for_each_entry_rcu(c, tmp, list, mnode)
1055 if (parent == -1 || parent == c->mfc_parent)
1056 return c;
1057
1058 return NULL;
1059}
1060
7ef8f65d 1061/* Allocate a multicast cache entry */
d658f8a0 1062static struct mfc_cache *ipmr_cache_alloc(void)
1da177e4 1063{
c354e124 1064 struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL);
a8c9486b 1065
70a0dec4
TG		 1066 if (c) {
1067 c->mfc_un.res.last_assert = jiffies - MFC_ASSERT_THRESH - 1;
a8c9486b 1068 c->mfc_un.res.minvif = MAXVIFS;
310ebbba 1069 refcount_set(&c->mfc_un.res.refcount, 1);
70a0dec4 1070 }
1da177e4
LT		 1071 return c;
1072}
1073
d658f8a0 1074static struct mfc_cache *ipmr_cache_alloc_unres(void)
1da177e4 1075{
c354e124 1076 struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC);
a8c9486b
ED		 1077
1078 if (c) {
1079 skb_queue_head_init(&c->mfc_un.unres.unresolved);
1080 c->mfc_un.unres.expires = jiffies + 10*HZ;
1081 }
1da177e4
LT		 1082 return c;
1083}
1084
7ef8f65d 1085/* A cache entry has gone into a resolved state from queued */
0c12295a
PM		 1086static void ipmr_cache_resolve(struct net *net, struct mr_table *mrt,
1087 struct mfc_cache *uc, struct mfc_cache *c)
1da177e4
LT		 1088{
1089 struct sk_buff *skb;
9ef1d4c7 1090 struct nlmsgerr *e;
1da177e4 1091
a8cb16dd 1092 /* Play the pending entries through our router */
c354e124 1093 while ((skb = __skb_dequeue(&uc->mfc_un.unres.unresolved))) {
eddc9ec5 1094 if (ip_hdr(skb)->version == 0) {
af72868b
JB		 1095 struct nlmsghdr *nlh = skb_pull(skb,
1096 sizeof(struct iphdr));
1da177e4 1097
573ce260 1098 if (__ipmr_fill_mroute(mrt, skb, c, nlmsg_data(nlh)) > 0) {
a8cb16dd
ED		 1099 nlh->nlmsg_len = skb_tail_pointer(skb) -
1100 (u8 *)nlh;
1da177e4
LT		 1101 } else {
1102 nlh->nlmsg_type = NLMSG_ERROR;
573ce260 1103 nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr));
1da177e4 1104 skb_trim(skb, nlh->nlmsg_len);
573ce260 1105 e = nlmsg_data(nlh);
9ef1d4c7
PM		 1106 e->error = -EMSGSIZE;
1107 memset(&e->msg, 0, sizeof(e->msg));
1da177e4 1108 }
2942e900 1109
15e47304 1110 rtnl_unicast(skb, net, NETLINK_CB(skb).portid);
a8cb16dd 1111 } else {
4b1f0d33 1112 ip_mr_forward(net, mrt, skb->dev, skb, c, 0);
a8cb16dd 1113 }
1da177e4
LT		 1114 }
1115}
1116
5a645dd8 1117/* Bounce a cache query up to mrouted and netlink.
1da177e4 1118 *
c316c629 1119 * Called under mrt_lock.
1da177e4 1120 */
0c12295a 1121static int ipmr_cache_report(struct mr_table *mrt,
4feb88e5 1122 struct sk_buff *pkt, vifi_t vifi, int assert)
1da177e4 1123{
c9bdd4b5 1124 const int ihl = ip_hdrlen(pkt);
c316c629 1125 struct sock *mroute_sk;
1da177e4
LT		 1126 struct igmphdr *igmp;
1127 struct igmpmsg *msg;
c316c629 1128 struct sk_buff *skb;
1da177e4
LT		 1129 int ret;
1130
1da177e4
LT		 1131 if (assert == IGMPMSG_WHOLEPKT)
1132 skb = skb_realloc_headroom(pkt, sizeof(struct iphdr));
1133 else
1da177e4
LT		 1134 skb = alloc_skb(128, GFP_ATOMIC);
1135
132adf54 1136 if (!skb)
1da177e4
LT		 1137 return -ENOBUFS;
1138
1da177e4
LT		 1139 if (assert == IGMPMSG_WHOLEPKT) {
1140 /* Ugly, but we have no choice with this interface.
a8cb16dd
ED		 1141 * Duplicate old header, fix ihl, length etc.
1142 * And all this only to mangle msg->im_msgtype and
1143 * to set msg->im_mbz to "mbz" :-)
1da177e4 1144 */
878c8145
ACM		 1145 skb_push(skb, sizeof(struct iphdr));
1146 skb_reset_network_header(skb);
badff6d0 1147 skb_reset_transport_header(skb);
0272ffc4 1148 msg = (struct igmpmsg *)skb_network_header(skb);
d56f90a7 1149 memcpy(msg, skb_network_header(pkt), sizeof(struct iphdr));
1da177e4
LT		 1150 msg->im_msgtype = IGMPMSG_WHOLEPKT;
1151 msg->im_mbz = 0;
0c12295a 1152 msg->im_vif = mrt->mroute_reg_vif_num;
eddc9ec5
ACM		 1153 ip_hdr(skb)->ihl = sizeof(struct iphdr) >> 2;
1154 ip_hdr(skb)->tot_len = htons(ntohs(ip_hdr(pkt)->tot_len) +
1155 sizeof(struct iphdr));
c316c629
NA		 1156 } else {
1157 /* Copy the IP header */
1158 skb_set_network_header(skb, skb->len);
1159 skb_put(skb, ihl);
1160 skb_copy_to_linear_data(skb, pkt->data, ihl);
1161 /* Flag to the kernel this is a route add */
1162 ip_hdr(skb)->protocol = 0;
1163 msg = (struct igmpmsg *)skb_network_header(skb);
1164 msg->im_vif = vifi;
1165 skb_dst_set(skb, dst_clone(skb_dst(pkt)));
1166 /* Add our header */
4df864c1 1167 igmp = skb_put(skb, sizeof(struct igmphdr));
c316c629
NA		 1168 igmp->type = assert;
1169 msg->im_msgtype = assert;
1170 igmp->code = 0;
1171 ip_hdr(skb)->tot_len = htons(skb->len); /* Fix the length */
1172 skb->transport_header = skb->network_header;
e905a9ed 1173 }
1da177e4 1174
4c968709
ED		 1175 rcu_read_lock();
1176 mroute_sk = rcu_dereference(mrt->mroute_sk);
51456b29 1177 if (!mroute_sk) {
4c968709 1178 rcu_read_unlock();
1da177e4
LT		 1179 kfree_skb(skb);
1180 return -EINVAL;
1181 }
1182
5a645dd8
JG		 1183 igmpmsg_netlink_event(mrt, skb);
1184
a8cb16dd 1185 /* Deliver to mrouted */
4c968709
ED		 1186 ret = sock_queue_rcv_skb(mroute_sk, skb);
1187 rcu_read_unlock();
70a269e6 1188 if (ret < 0) {
e87cc472 1189 net_warn_ratelimited("mroute: pending queue full, dropping entries\n");
1da177e4
LT		 1190 kfree_skb(skb);
1191 }
1192
1193 return ret;
1194}
1195
7ef8f65d
NA		 1196/* Queue a packet for resolution. It gets locked cache entry! */
1197static int ipmr_cache_unresolved(struct mr_table *mrt, vifi_t vifi,
4b1f0d33 1198 struct sk_buff *skb, struct net_device *dev)
1da177e4 1199{
8fb472c0
NA		 1200 const struct iphdr *iph = ip_hdr(skb);
1201 struct mfc_cache *c;
862465f2 1202 bool found = false;
1da177e4 1203 int err;
1da177e4
LT		 1204
1205 spin_lock_bh(&mfc_unres_lock);
0c12295a 1206 list_for_each_entry(c, &mrt->mfc_unres_queue, list) {
e258beb2 1207 if (c->mfc_mcastgrp == iph->daddr &&
862465f2
PM		 1208 c->mfc_origin == iph->saddr) {
1209 found = true;
1da177e4 1210 break;
862465f2 1211 }
1da177e4
LT		 1212 }
1213
862465f2 1214 if (!found) {
a8cb16dd 1215 /* Create a new entry if allowable */
0c12295a 1216 if (atomic_read(&mrt->cache_resolve_queue_len) >= 10 ||
d658f8a0 1217 (c = ipmr_cache_alloc_unres()) == NULL) {
1da177e4
LT		 1218 spin_unlock_bh(&mfc_unres_lock);
1219
1220 kfree_skb(skb);
1221 return -ENOBUFS;
1222 }
1223
a8cb16dd 1224 /* Fill in the new cache entry */
eddc9ec5
ACM		 1225 c->mfc_parent = -1;
1226 c->mfc_origin = iph->saddr;
1227 c->mfc_mcastgrp = iph->daddr;
1da177e4 1228
a8cb16dd 1229 /* Reflect first query at mrouted. */
0c12295a 1230 err = ipmr_cache_report(mrt, skb, vifi, IGMPMSG_NOCACHE);
4feb88e5 1231 if (err < 0) {
e905a9ed 1232 /* If the report failed throw the cache entry
1da177e4
LT		 1233 out - Brad Parker
1234 */
1235 spin_unlock_bh(&mfc_unres_lock);
1236
5c0a66f5 1237 ipmr_cache_free(c);
1da177e4
LT		 1238 kfree_skb(skb);
1239 return err;
1240 }
1241
0c12295a
PM		 1242 atomic_inc(&mrt->cache_resolve_queue_len);
1243 list_add(&c->list, &mrt->mfc_unres_queue);
8cd3ac9f 1244 mroute_netlink_event(mrt, c, RTM_NEWROUTE);
1da177e4 1245
278554bd
DM		 1246 if (atomic_read(&mrt->cache_resolve_queue_len) == 1)
1247 mod_timer(&mrt->ipmr_expire_timer, c->mfc_un.unres.expires);
1da177e4
LT		 1248 }
1249
a8cb16dd 1250 /* See if we can append the packet */
a8cb16dd 1251 if (c->mfc_un.unres.unresolved.qlen > 3) {
1da177e4
LT		 1252 kfree_skb(skb);
1253 err = -ENOBUFS;
1254 } else {
4b1f0d33
DS		 1255 if (dev) {
1256 skb->dev = dev;
1257 skb->skb_iif = dev->ifindex;
1258 }
c354e124 1259 skb_queue_tail(&c->mfc_un.unres.unresolved, skb);
1da177e4
LT		 1260 err = 0;
1261 }
1262
1263 spin_unlock_bh(&mfc_unres_lock);
1264 return err;
1265}
1266
7ef8f65d 1267/* MFC cache manipulation by user space mroute daemon */
1da177e4 1268
660b26dc 1269static int ipmr_mfc_delete(struct mr_table *mrt, struct mfcctl *mfc, int parent)
1da177e4 1270{
b362053a 1271 struct net *net = read_pnet(&mrt->net);
8fb472c0 1272 struct mfc_cache *c;
1da177e4 1273
8fb472c0
NA		 1274 /* The entries are added/deleted only under RTNL */
1275 rcu_read_lock();
1276 c = ipmr_cache_find_parent(mrt, mfc->mfcc_origin.s_addr,
1277 mfc->mfcc_mcastgrp.s_addr, parent);
1278 rcu_read_unlock();
1279 if (!c)
1280 return -ENOENT;
1281 rhltable_remove(&mrt->mfc_hash, &c->mnode, ipmr_rht_params);
1282 list_del_rcu(&c->list);
b362053a 1283 call_ipmr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_DEL, c, mrt->id);
8fb472c0 1284 mroute_netlink_event(mrt, c, RTM_DELROUTE);
310ebbba 1285 ipmr_cache_put(c);
1da177e4 1286
8fb472c0 1287 return 0;
1da177e4
LT		 1288}
1289
0c12295a 1290static int ipmr_mfc_add(struct net *net, struct mr_table *mrt,
660b26dc 1291 struct mfcctl *mfc, int mrtsock, int parent)
1da177e4 1292{
862465f2 1293 struct mfc_cache *uc, *c;
8fb472c0
NA		 1294 bool found;
1295 int ret;
1da177e4 1296
a50436f2
PM		 1297 if (mfc->mfcc_parent >= MAXVIFS)
1298 return -ENFILE;
1299
8fb472c0
NA		 1300 /* The entries are added/deleted only under RTNL */
1301 rcu_read_lock();
1302 c = ipmr_cache_find_parent(mrt, mfc->mfcc_origin.s_addr,
1303 mfc->mfcc_mcastgrp.s_addr, parent);
1304 rcu_read_unlock();
1305 if (c) {
1da177e4
LT		 1306 write_lock_bh(&mrt_lock);
1307 c->mfc_parent = mfc->mfcc_parent;
0c12295a 1308 ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls);
1da177e4
LT		 1309 if (!mrtsock)
1310 c->mfc_flags |= MFC_STATIC;
1311 write_unlock_bh(&mrt_lock);
b362053a
YG		 1312 call_ipmr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_REPLACE, c,
1313 mrt->id);
8cd3ac9f 1314 mroute_netlink_event(mrt, c, RTM_NEWROUTE);
1da177e4
LT		 1315 return 0;
1316 }
1317
360eb5da 1318 if (mfc->mfcc_mcastgrp.s_addr != htonl(INADDR_ANY) &&
660b26dc 1319 !ipv4_is_multicast(mfc->mfcc_mcastgrp.s_addr))
1da177e4
LT		 1320 return -EINVAL;
1321
d658f8a0 1322 c = ipmr_cache_alloc();
51456b29 1323 if (!c)
1da177e4
LT		 1324 return -ENOMEM;
1325
c354e124
JK		 1326 c->mfc_origin = mfc->mfcc_origin.s_addr;
1327 c->mfc_mcastgrp = mfc->mfcc_mcastgrp.s_addr;
1328 c->mfc_parent = mfc->mfcc_parent;
0c12295a 1329 ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls);
1da177e4
LT		 1330 if (!mrtsock)
1331 c->mfc_flags |= MFC_STATIC;
1332
8fb472c0
NA		 1333 ret = rhltable_insert_key(&mrt->mfc_hash, &c->cmparg, &c->mnode,
1334 ipmr_rht_params);
1335 if (ret) {
1336 pr_err("ipmr: rhtable insert error %d\n", ret);
1337 ipmr_cache_free(c);
1338 return ret;
1339 }
1340 list_add_tail_rcu(&c->list, &mrt->mfc_cache_list);
7ef8f65d
NA		 1341 /* Check to see if we resolved a queued list. If so we
1342 * need to send on the frames and tidy up.
1da177e4 1343 */
b0ebb739 1344 found = false;
1da177e4 1345 spin_lock_bh(&mfc_unres_lock);
0c12295a 1346 list_for_each_entry(uc, &mrt->mfc_unres_queue, list) {
e258beb2 1347 if (uc->mfc_origin == c->mfc_origin &&
1da177e4 1348 uc->mfc_mcastgrp == c->mfc_mcastgrp) {
862465f2 1349 list_del(&uc->list);
0c12295a 1350 atomic_dec(&mrt->cache_resolve_queue_len);
b0ebb739 1351 found = true;
1da177e4
LT		 1352 break;
1353 }
1354 }
0c12295a
PM		 1355 if (list_empty(&mrt->mfc_unres_queue))
1356 del_timer(&mrt->ipmr_expire_timer);
1da177e4
LT		 1357 spin_unlock_bh(&mfc_unres_lock);
1358
b0ebb739 1359 if (found) {
0c12295a 1360 ipmr_cache_resolve(net, mrt, uc, c);
5c0a66f5 1361 ipmr_cache_free(uc);
1da177e4 1362 }
b362053a 1363 call_ipmr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_ADD, c, mrt->id);
8cd3ac9f 1364 mroute_netlink_event(mrt, c, RTM_NEWROUTE);
1da177e4
LT		 1365 return 0;
1366}
1367
7ef8f65d 1368/* Close the multicast socket, and clear the vif tables etc */
0e615e96 1369static void mroute_clean_tables(struct mr_table *mrt, bool all)
1da177e4 1370{
b362053a 1371 struct net *net = read_pnet(&mrt->net);
8fb472c0 1372 struct mfc_cache *c, *tmp;
d17fa6fa 1373 LIST_HEAD(list);
8fb472c0 1374 int i;
e905a9ed 1375
a8cb16dd 1376 /* Shut down all active vif entries */
0c12295a 1377 for (i = 0; i < mrt->maxvif; i++) {
0e615e96
NA		 1378 if (!all && (mrt->vif_table[i].flags & VIFF_STATIC))
1379 continue;
1380 vif_delete(mrt, i, 0, &list);
1da177e4 1381 }
d17fa6fa 1382 unregister_netdevice_many(&list);
1da177e4 1383
a8cb16dd 1384 /* Wipe the cache */
8fb472c0
NA		 1385 list_for_each_entry_safe(c, tmp, &mrt->mfc_cache_list, list) {
1386 if (!all && (c->mfc_flags & MFC_STATIC))
1387 continue;
1388 rhltable_remove(&mrt->mfc_hash, &c->mnode, ipmr_rht_params);
1389 list_del_rcu(&c->list);
b362053a
YG		 1390 call_ipmr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_DEL, c,
1391 mrt->id);
8fb472c0 1392 mroute_netlink_event(mrt, c, RTM_DELROUTE);
310ebbba 1393 ipmr_cache_put(c);
1da177e4
LT		 1394 }
1395
0c12295a 1396 if (atomic_read(&mrt->cache_resolve_queue_len) != 0) {
1da177e4 1397 spin_lock_bh(&mfc_unres_lock);
8fb472c0 1398 list_for_each_entry_safe(c, tmp, &mrt->mfc_unres_queue, list) {
862465f2 1399 list_del(&c->list);
8cd3ac9f 1400 mroute_netlink_event(mrt, c, RTM_DELROUTE);
0c12295a 1401 ipmr_destroy_unres(mrt, c);
1da177e4
LT		 1402 }
1403 spin_unlock_bh(&mfc_unres_lock);
1404 }
1405}
1406
4c968709
ED		 1407/* called from ip_ra_control(), before an RCU grace period,
1408 * we dont need to call synchronize_rcu() here
1409 */
1da177e4
LT		 1410static void mrtsock_destruct(struct sock *sk)
1411{
4feb88e5 1412 struct net *net = sock_net(sk);
f0ad0860 1413 struct mr_table *mrt;
4feb88e5 1414
1215e51e 1415 ASSERT_RTNL();
f0ad0860 1416 ipmr_for_each_table(mrt, net) {
4c968709 1417 if (sk == rtnl_dereference(mrt->mroute_sk)) {
f0ad0860 1418 IPV4_DEVCONF_ALL(net, MC_FORWARDING)--;
3b022865
DA		 1419 inet_netconf_notify_devconf(net, RTM_NEWNETCONF,
1420 NETCONFA_MC_FORWARDING,
d67b8c61
ND		 1421 NETCONFA_IFINDEX_ALL,
1422 net->ipv4.devconf_all);
a9b3cd7f 1423 RCU_INIT_POINTER(mrt->mroute_sk, NULL);
0e615e96 1424 mroute_clean_tables(mrt, false);
f0ad0860 1425 }
1da177e4 1426 }
1da177e4
LT		 1427}
1428
7ef8f65d
NA		 1429/* Socket options and virtual interface manipulation. The whole
1430 * virtual interface system is a complete heap, but unfortunately
1431 * that's how BSD mrouted happens to think. Maybe one day with a proper
1432 * MOSPF/PIM router set up we can clean this up.
1da177e4 1433 */
e905a9ed 1434
29e97d21
NA		 1435int ip_mroute_setsockopt(struct sock *sk, int optname, char __user *optval,
1436 unsigned int optlen)
1da177e4 1437{
4feb88e5 1438 struct net *net = sock_net(sk);
29e97d21 1439 int val, ret = 0, parent = 0;
f0ad0860 1440 struct mr_table *mrt;
29e97d21
NA		 1441 struct vifctl vif;
1442 struct mfcctl mfc;
1443 u32 uval;
f0ad0860 1444
29e97d21
NA		 1445 /* There's one exception to the lock - MRT_DONE which needs to unlock */
1446 rtnl_lock();
5e1859fb 1447 if (sk->sk_type != SOCK_RAW ||
29e97d21
NA		 1448 inet_sk(sk)->inet_num != IPPROTO_IGMP) {
1449 ret = -EOPNOTSUPP;
1450 goto out_unlock;
1451 }
5e1859fb 1452
f0ad0860 1453 mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT);
29e97d21
NA		 1454 if (!mrt) {
1455 ret = -ENOENT;
1456 goto out_unlock;
1457 }
132adf54 1458 if (optname != MRT_INIT) {
33d480ce 1459 if (sk != rcu_access_pointer(mrt->mroute_sk) &&
29e97d21
NA		 1460 !ns_capable(net->user_ns, CAP_NET_ADMIN)) {
1461 ret = -EACCES;
1462 goto out_unlock;
1463 }
1da177e4
LT		 1464 }
1465
132adf54
SH		 1466 switch (optname) {
1467 case MRT_INIT:
42e6b89c 1468 if (optlen != sizeof(int)) {
29e97d21 1469 ret = -EINVAL;
42e6b89c
NA		 1470 break;
1471 }
1472 if (rtnl_dereference(mrt->mroute_sk)) {
29e97d21 1473 ret = -EADDRINUSE;
29e97d21 1474 break;
42e6b89c 1475 }
132adf54
SH		 1476
1477 ret = ip_ra_control(sk, 1, mrtsock_destruct);
1478 if (ret == 0) {
cf778b00 1479 rcu_assign_pointer(mrt->mroute_sk, sk);
4feb88e5 1480 IPV4_DEVCONF_ALL(net, MC_FORWARDING)++;
3b022865
DA		 1481 inet_netconf_notify_devconf(net, RTM_NEWNETCONF,
1482 NETCONFA_MC_FORWARDING,
d67b8c61
ND		 1483 NETCONFA_IFINDEX_ALL,
1484 net->ipv4.devconf_all);
132adf54 1485 }
29e97d21 1486 break;
132adf54 1487 case MRT_DONE:
29e97d21
NA		 1488 if (sk != rcu_access_pointer(mrt->mroute_sk)) {
1489 ret = -EACCES;
1490 } else {
29e97d21 1491 ret = ip_ra_control(sk, 0, NULL);
1215e51e 1492 goto out_unlock;
29e97d21
NA		 1493 }
1494 break;
132adf54
SH		 1495 case MRT_ADD_VIF:
1496 case MRT_DEL_VIF:
29e97d21
NA		 1497 if (optlen != sizeof(vif)) {
1498 ret = -EINVAL;
1499 break;
1500 }
1501 if (copy_from_user(&vif, optval, sizeof(vif))) {
1502 ret = -EFAULT;
1503 break;
1504 }
1505 if (vif.vifc_vifi >= MAXVIFS) {
1506 ret = -ENFILE;
1507 break;
1508 }
c354e124 1509 if (optname == MRT_ADD_VIF) {
4c968709
ED		 1510 ret = vif_add(net, mrt, &vif,
1511 sk == rtnl_dereference(mrt->mroute_sk));
132adf54 1512 } else {
0c12295a 1513 ret = vif_delete(mrt, vif.vifc_vifi, 0, NULL);
132adf54 1514 }
29e97d21 1515 break;
7ef8f65d
NA		 1516 /* Manipulate the forwarding caches. These live
1517 * in a sort of kernel/user symbiosis.
1518 */
132adf54
SH		 1519 case MRT_ADD_MFC:
1520 case MRT_DEL_MFC:
660b26dc 1521 parent = -1;
fcfd6dfa 1522 /* fall through */
660b26dc
ND		 1523 case MRT_ADD_MFC_PROXY:
1524 case MRT_DEL_MFC_PROXY:
29e97d21
NA		 1525 if (optlen != sizeof(mfc)) {
1526 ret = -EINVAL;
1527 break;
1528 }
1529 if (copy_from_user(&mfc, optval, sizeof(mfc))) {
1530 ret = -EFAULT;
1531 break;
1532 }
660b26dc
ND		 1533 if (parent == 0)
1534 parent = mfc.mfcc_parent;
660b26dc
ND		 1535 if (optname == MRT_DEL_MFC || optname == MRT_DEL_MFC_PROXY)
1536 ret = ipmr_mfc_delete(mrt, &mfc, parent);
132adf54 1537 else
4c968709 1538 ret = ipmr_mfc_add(net, mrt, &mfc,
660b26dc
ND		 1539 sk == rtnl_dereference(mrt->mroute_sk),
1540 parent);
29e97d21 1541 break;
7ef8f65d 1542 /* Control PIM assert. */
132adf54 1543 case MRT_ASSERT:
29e97d21
NA		 1544 if (optlen != sizeof(val)) {
1545 ret = -EINVAL;
1546 break;
1547 }
1548 if (get_user(val, (int __user *)optval)) {
1549 ret = -EFAULT;
1550 break;
1551 }
1552 mrt->mroute_do_assert = val;
1553 break;
132adf54 1554 case MRT_PIM:
1973a4ea 1555 if (!ipmr_pimsm_enabled()) {
29e97d21
NA		 1556 ret = -ENOPROTOOPT;
1557 break;
1558 }
1559 if (optlen != sizeof(val)) {
1560 ret = -EINVAL;
1561 break;
1562 }
1563 if (get_user(val, (int __user *)optval)) {
1564 ret = -EFAULT;
1565 break;
1566 }
ba93ef74 1567
29e97d21
NA		 1568 val = !!val;
1569 if (val != mrt->mroute_do_pim) {
1570 mrt->mroute_do_pim = val;
1571 mrt->mroute_do_assert = val;
1da177e4 1572 }
29e97d21 1573 break;
f0ad0860 1574 case MRT_TABLE:
29e97d21
NA		 1575 if (!IS_BUILTIN(CONFIG_IP_MROUTE_MULTIPLE_TABLES)) {
1576 ret = -ENOPROTOOPT;
1577 break;
1578 }
1579 if (optlen != sizeof(uval)) {
1580 ret = -EINVAL;
1581 break;
1582 }
1583 if (get_user(uval, (u32 __user *)optval)) {
1584 ret = -EFAULT;
1585 break;
1586 }
f0ad0860 1587
4c968709
ED		 1588 if (sk == rtnl_dereference(mrt->mroute_sk)) {
1589 ret = -EBUSY;
1590 } else {
29e97d21 1591 mrt = ipmr_new_table(net, uval);
1113ebbc
NA		 1592 if (IS_ERR(mrt))
1593 ret = PTR_ERR(mrt);
5e1859fb 1594 else
29e97d21 1595 raw_sk(sk)->ipmr_table = uval;
4c968709 1596 }
29e97d21 1597 break;
7ef8f65d 1598 /* Spurious command, or MRT_VERSION which you cannot set. */
132adf54 1599 default:
29e97d21 1600 ret = -ENOPROTOOPT;
1da177e4 1601 }
29e97d21
NA		 1602out_unlock:
1603 rtnl_unlock();
29e97d21 1604 return ret;
1da177e4
LT		 1605}
1606
7ef8f65d 1607/* Getsock opt support for the multicast routing system. */
c354e124 1608int ip_mroute_getsockopt(struct sock *sk, int optname, char __user *optval, int __user *optlen)
1da177e4
LT		 1609{
1610 int olr;
1611 int val;
4feb88e5 1612 struct net *net = sock_net(sk);
f0ad0860
PM		 1613 struct mr_table *mrt;
1614
5e1859fb
ED		 1615 if (sk->sk_type != SOCK_RAW ||
1616 inet_sk(sk)->inet_num != IPPROTO_IGMP)
1617 return -EOPNOTSUPP;
1618
f0ad0860 1619 mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT);
51456b29 1620 if (!mrt)
f0ad0860 1621 return -ENOENT;
1da177e4 1622
fe9ef3ce
NA		 1623 switch (optname) {
1624 case MRT_VERSION:
1625 val = 0x0305;
1626 break;
1627 case MRT_PIM:
1973a4ea 1628 if (!ipmr_pimsm_enabled())
fe9ef3ce
NA		 1629 return -ENOPROTOOPT;
1630 val = mrt->mroute_do_pim;
1631 break;
1632 case MRT_ASSERT:
1633 val = mrt->mroute_do_assert;
1634 break;
1635 default:
1da177e4 1636 return -ENOPROTOOPT;
fe9ef3ce 1637 }
1da177e4
LT		 1638
1639 if (get_user(olr, optlen))
1640 return -EFAULT;
1da177e4
LT		 1641 olr = min_t(unsigned int, olr, sizeof(int));
1642 if (olr < 0)
1643 return -EINVAL;
c354e124 1644 if (put_user(olr, optlen))
1da177e4 1645 return -EFAULT;
c354e124 1646 if (copy_to_user(optval, &val, olr))
1da177e4
LT		 1647 return -EFAULT;
1648 return 0;
1649}
1650
7ef8f65d 1651/* The IP multicast ioctl support routines. */
1da177e4
LT		 1652int ipmr_ioctl(struct sock *sk, int cmd, void __user *arg)
1653{
1654 struct sioc_sg_req sr;
1655 struct sioc_vif_req vr;
1656 struct vif_device *vif;
1657 struct mfc_cache *c;
4feb88e5 1658 struct net *net = sock_net(sk);
f0ad0860
PM		 1659 struct mr_table *mrt;
1660
1661 mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT);
51456b29 1662 if (!mrt)
f0ad0860 1663 return -ENOENT;
e905a9ed 1664
132adf54
SH		 1665 switch (cmd) {
1666 case SIOCGETVIFCNT:
c354e124 1667 if (copy_from_user(&vr, arg, sizeof(vr)))
132adf54 1668 return -EFAULT;
0c12295a 1669 if (vr.vifi >= mrt->maxvif)
132adf54
SH		 1670 return -EINVAL;
1671 read_lock(&mrt_lock);
0c12295a
PM		 1672 vif = &mrt->vif_table[vr.vifi];
1673 if (VIF_EXISTS(mrt, vr.vifi)) {
c354e124
JK		 1674 vr.icount = vif->pkt_in;
1675 vr.ocount = vif->pkt_out;
1676 vr.ibytes = vif->bytes_in;
1677 vr.obytes = vif->bytes_out;
1da177e4 1678 read_unlock(&mrt_lock);
1da177e4 1679
c354e124 1680 if (copy_to_user(arg, &vr, sizeof(vr)))
132adf54
SH		 1681 return -EFAULT;
1682 return 0;
1683 }
1684 read_unlock(&mrt_lock);
1685 return -EADDRNOTAVAIL;
1686 case SIOCGETSGCNT:
c354e124 1687 if (copy_from_user(&sr, arg, sizeof(sr)))
132adf54
SH		 1688 return -EFAULT;
1689
a8c9486b 1690 rcu_read_lock();
0c12295a 1691 c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr);
132adf54
SH		 1692 if (c) {
1693 sr.pktcnt = c->mfc_un.res.pkt;
1694 sr.bytecnt = c->mfc_un.res.bytes;
1695 sr.wrong_if = c->mfc_un.res.wrong_if;
a8c9486b 1696 rcu_read_unlock();
132adf54 1697
c354e124 1698 if (copy_to_user(arg, &sr, sizeof(sr)))
132adf54
SH		 1699 return -EFAULT;
1700 return 0;
1701 }
a8c9486b 1702 rcu_read_unlock();
132adf54
SH		 1703 return -EADDRNOTAVAIL;
1704 default:
1705 return -ENOIOCTLCMD;
1da177e4
LT		 1706 }
1707}
1708
709b46e8
EB		 1709#ifdef CONFIG_COMPAT
1710struct compat_sioc_sg_req {
1711 struct in_addr src;
1712 struct in_addr grp;
1713 compat_ulong_t pktcnt;
1714 compat_ulong_t bytecnt;
1715 compat_ulong_t wrong_if;
1716};
1717
ca6b8bb0
DM		 1718struct compat_sioc_vif_req {
1719 vifi_t vifi; /* Which iface */
1720 compat_ulong_t icount;
1721 compat_ulong_t ocount;
1722 compat_ulong_t ibytes;
1723 compat_ulong_t obytes;
1724};
1725
709b46e8
EB		 1726int ipmr_compat_ioctl(struct sock *sk, unsigned int cmd, void __user *arg)
1727{
0033d5ad 1728 struct compat_sioc_sg_req sr;
ca6b8bb0
DM		 1729 struct compat_sioc_vif_req vr;
1730 struct vif_device *vif;
709b46e8
EB		 1731 struct mfc_cache *c;
1732 struct net *net = sock_net(sk);
1733 struct mr_table *mrt;
1734
1735 mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT);
51456b29 1736 if (!mrt)
709b46e8
EB		 1737 return -ENOENT;
1738
1739 switch (cmd) {
ca6b8bb0
DM		 1740 case SIOCGETVIFCNT:
1741 if (copy_from_user(&vr, arg, sizeof(vr)))
1742 return -EFAULT;
1743 if (vr.vifi >= mrt->maxvif)
1744 return -EINVAL;
1745 read_lock(&mrt_lock);
1746 vif = &mrt->vif_table[vr.vifi];
1747 if (VIF_EXISTS(mrt, vr.vifi)) {
1748 vr.icount = vif->pkt_in;
1749 vr.ocount = vif->pkt_out;
1750 vr.ibytes = vif->bytes_in;
1751 vr.obytes = vif->bytes_out;
1752 read_unlock(&mrt_lock);
1753
1754 if (copy_to_user(arg, &vr, sizeof(vr)))
1755 return -EFAULT;
1756 return 0;
1757 }
1758 read_unlock(&mrt_lock);
1759 return -EADDRNOTAVAIL;
709b46e8
EB		 1760 case SIOCGETSGCNT:
1761 if (copy_from_user(&sr, arg, sizeof(sr)))
1762 return -EFAULT;
1763
1764 rcu_read_lock();
1765 c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr);
1766 if (c) {
1767 sr.pktcnt = c->mfc_un.res.pkt;
1768 sr.bytecnt = c->mfc_un.res.bytes;
1769 sr.wrong_if = c->mfc_un.res.wrong_if;
1770 rcu_read_unlock();
1771
1772 if (copy_to_user(arg, &sr, sizeof(sr)))
1773 return -EFAULT;
1774 return 0;
1775 }
1776 rcu_read_unlock();
1777 return -EADDRNOTAVAIL;
1778 default:
1779 return -ENOIOCTLCMD;
1780 }
1781}
1782#endif
1783
1da177e4
LT		 1784static int ipmr_device_event(struct notifier_block *this, unsigned long event, void *ptr)
1785{
351638e7 1786 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
4feb88e5 1787 struct net *net = dev_net(dev);
f0ad0860 1788 struct mr_table *mrt;
1da177e4
LT		 1789 struct vif_device *v;
1790 int ct;
e9dc8653 1791
1da177e4
LT		 1792 if (event != NETDEV_UNREGISTER)
1793 return NOTIFY_DONE;
f0ad0860
PM		 1794
1795 ipmr_for_each_table(mrt, net) {
1796 v = &mrt->vif_table[0];
1797 for (ct = 0; ct < mrt->maxvif; ct++, v++) {
1798 if (v->dev == dev)
e92036a6 1799 vif_delete(mrt, ct, 1, NULL);
f0ad0860 1800 }
1da177e4
LT		 1801 }
1802 return NOTIFY_DONE;
1803}
1804
c354e124 1805static struct notifier_block ip_mr_notifier = {
1da177e4
LT		 1806 .notifier_call = ipmr_device_event,
1807};
1808
7ef8f65d
NA		 1809/* Encapsulate a packet by attaching a valid IPIP header to it.
1810 * This avoids tunnel drivers and other mess and gives us the speed so
1811 * important for multicast video.
1da177e4 1812 */
b6a7719a
HFS		 1813static void ip_encap(struct net *net, struct sk_buff *skb,
1814 __be32 saddr, __be32 daddr)
1da177e4 1815{
8856dfa3 1816 struct iphdr *iph;
b71d1d42 1817 const struct iphdr *old_iph = ip_hdr(skb);
8856dfa3
ACM		 1818
1819 skb_push(skb, sizeof(struct iphdr));
b0e380b1 1820 skb->transport_header = skb->network_header;
8856dfa3 1821 skb_reset_network_header(skb);
eddc9ec5 1822 iph = ip_hdr(skb);
1da177e4 1823
a8cb16dd 1824 iph->version = 4;
e023dd64
ACM		 1825 iph->tos = old_iph->tos;
1826 iph->ttl = old_iph->ttl;
1da177e4
LT		 1827 iph->frag_off = 0;
1828 iph->daddr = daddr;
1829 iph->saddr = saddr;
1830 iph->protocol = IPPROTO_IPIP;
1831 iph->ihl = 5;
1832 iph->tot_len = htons(skb->len);
b6a7719a 1833 ip_select_ident(net, skb, NULL);
1da177e4
LT		 1834 ip_send_check(iph);
1835
1da177e4
LT		 1836 memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt));
1837 nf_reset(skb);
1838}
1839
0c4b51f0
EB		 1840static inline int ipmr_forward_finish(struct net *net, struct sock *sk,
1841 struct sk_buff *skb)
1da177e4 1842{
a8cb16dd 1843 struct ip_options *opt = &(IPCB(skb)->opt);
1da177e4 1844
73186df8
DM		 1845 IP_INC_STATS(net, IPSTATS_MIB_OUTFORWDATAGRAMS);
1846 IP_ADD_STATS(net, IPSTATS_MIB_OUTOCTETS, skb->len);
1da177e4
LT		 1847
1848 if (unlikely(opt->optlen))
1849 ip_forward_options(skb);
1850
13206b6b 1851 return dst_output(net, sk, skb);
1da177e4
LT		 1852}
1853
a5bc9294
YG		 1854#ifdef CONFIG_NET_SWITCHDEV
1855static bool ipmr_forward_offloaded(struct sk_buff *skb, struct mr_table *mrt,
1856 int in_vifi, int out_vifi)
1857{
1858 struct vif_device *out_vif = &mrt->vif_table[out_vifi];
1859 struct vif_device *in_vif = &mrt->vif_table[in_vifi];
1860
1861 if (!skb->offload_mr_fwd_mark)
1862 return false;
1863 if (!out_vif->dev_parent_id.id_len || !in_vif->dev_parent_id.id_len)
1864 return false;
1865 return netdev_phys_item_id_same(&out_vif->dev_parent_id,
1866 &in_vif->dev_parent_id);
1867}
1868#else
1869static bool ipmr_forward_offloaded(struct sk_buff *skb, struct mr_table *mrt,
1870 int in_vifi, int out_vifi)
1871{
1872 return false;
1873}
1874#endif
1875
7ef8f65d 1876/* Processing handlers for ipmr_forward */
1da177e4 1877
0c12295a 1878static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt,
a5bc9294
YG		 1879 int in_vifi, struct sk_buff *skb,
1880 struct mfc_cache *c, int vifi)
1da177e4 1881{
eddc9ec5 1882 const struct iphdr *iph = ip_hdr(skb);
0c12295a 1883 struct vif_device *vif = &mrt->vif_table[vifi];
1da177e4
LT		 1884 struct net_device *dev;
1885 struct rtable *rt;
31e4543d 1886 struct flowi4 fl4;
1da177e4
LT		 1887 int encap = 0;
1888
51456b29 1889 if (!vif->dev)
1da177e4
LT		 1890 goto out_free;
1891
1da177e4
LT		 1892 if (vif->flags & VIFF_REGISTER) {
1893 vif->pkt_out++;
c354e124 1894 vif->bytes_out += skb->len;
cf3677ae
PE		 1895 vif->dev->stats.tx_bytes += skb->len;
1896 vif->dev->stats.tx_packets++;
0c12295a 1897 ipmr_cache_report(mrt, skb, vifi, IGMPMSG_WHOLEPKT);
69ebbf58 1898 goto out_free;
1da177e4 1899 }
1da177e4 1900
a5bc9294
YG		 1901 if (ipmr_forward_offloaded(skb, mrt, in_vifi, vifi))
1902 goto out_free;
1903
a8cb16dd 1904 if (vif->flags & VIFF_TUNNEL) {
31e4543d 1905 rt = ip_route_output_ports(net, &fl4, NULL,
78fbfd8a
DM		 1906 vif->remote, vif->local,
1907 0, 0,
1908 IPPROTO_IPIP,
1909 RT_TOS(iph->tos), vif->link);
b23dd4fe 1910 if (IS_ERR(rt))
1da177e4
LT		 1911 goto out_free;
1912 encap = sizeof(struct iphdr);
1913 } else {
31e4543d 1914 rt = ip_route_output_ports(net, &fl4, NULL, iph->daddr, 0,
78fbfd8a
DM		 1915 0, 0,
1916 IPPROTO_IPIP,
1917 RT_TOS(iph->tos), vif->link);
b23dd4fe 1918 if (IS_ERR(rt))
1da177e4
LT		 1919 goto out_free;
1920 }
1921
d8d1f30b 1922 dev = rt->dst.dev;
1da177e4 1923
d8d1f30b 1924 if (skb->len+encap > dst_mtu(&rt->dst) && (ntohs(iph->frag_off) & IP_DF)) {
1da177e4 1925 /* Do not fragment multicasts. Alas, IPv4 does not
a8cb16dd
ED		 1926 * allow to send ICMP, so that packets will disappear
1927 * to blackhole.
1da177e4 1928 */
73186df8 1929 IP_INC_STATS(net, IPSTATS_MIB_FRAGFAILS);
1da177e4
LT		 1930 ip_rt_put(rt);
1931 goto out_free;
1932 }
1933
d8d1f30b 1934 encap += LL_RESERVED_SPACE(dev) + rt->dst.header_len;
1da177e4
LT		 1935
1936 if (skb_cow(skb, encap)) {
e905a9ed 1937 ip_rt_put(rt);
1da177e4
LT		 1938 goto out_free;
1939 }
1940
1941 vif->pkt_out++;
c354e124 1942 vif->bytes_out += skb->len;
1da177e4 1943
adf30907 1944 skb_dst_drop(skb);
d8d1f30b 1945 skb_dst_set(skb, &rt->dst);
eddc9ec5 1946 ip_decrease_ttl(ip_hdr(skb));
1da177e4
LT		 1947
1948 /* FIXME: forward and output firewalls used to be called here.
a8cb16dd
ED		 1949 * What do we do with netfilter? -- RR
1950 */
1da177e4 1951 if (vif->flags & VIFF_TUNNEL) {
b6a7719a 1952 ip_encap(net, skb, vif->local, vif->remote);
1da177e4 1953 /* FIXME: extra output firewall step used to be here. --RR */
2f4c02d4
PE		 1954 vif->dev->stats.tx_packets++;
1955 vif->dev->stats.tx_bytes += skb->len;
1da177e4
LT		 1956 }
1957
9ee6c5dc 1958 IPCB(skb)->flags |= IPSKB_FORWARDED;
1da177e4 1959
7ef8f65d 1960 /* RFC1584 teaches, that DVMRP/PIM router must deliver packets locally
1da177e4
LT		 1961 * not only before forwarding, but after forwarding on all output
1962 * interfaces. It is clear, if mrouter runs a multicasting
1963 * program, it should receive packets not depending to what interface
1964 * program is joined.
1965 * If we will not make it, the program will have to join on all
1966 * interfaces. On the other hand, multihoming host (or router, but
1967 * not mrouter) cannot join to more than one interface - it will
1968 * result in receiving multiple packets.
1969 */
29a26a56
EB		 1970 NF_HOOK(NFPROTO_IPV4, NF_INET_FORWARD,
1971 net, NULL, skb, skb->dev, dev,
1da177e4
LT		 1972 ipmr_forward_finish);
1973 return;
1974
1975out_free:
1976 kfree_skb(skb);
1da177e4
LT		 1977}
1978
0c12295a 1979static int ipmr_find_vif(struct mr_table *mrt, struct net_device *dev)
1da177e4
LT		 1980{
1981 int ct;
0c12295a
PM		 1982
1983 for (ct = mrt->maxvif-1; ct >= 0; ct--) {
1984 if (mrt->vif_table[ct].dev == dev)
1da177e4
LT		 1985 break;
1986 }
1987 return ct;
1988}
1989
1990/* "local" means that we should preserve one skb (for local delivery) */
c4854ec8 1991static void ip_mr_forward(struct net *net, struct mr_table *mrt,
4b1f0d33
DS		 1992 struct net_device *dev, struct sk_buff *skb,
1993 struct mfc_cache *cache, int local)
1da177e4 1994{
4b1f0d33 1995 int true_vifi = ipmr_find_vif(mrt, dev);
1da177e4
LT		 1996 int psend = -1;
1997 int vif, ct;
1998
1999 vif = cache->mfc_parent;
2000 cache->mfc_un.res.pkt++;
2001 cache->mfc_un.res.bytes += skb->len;
43b9e127 2002 cache->mfc_un.res.lastuse = jiffies;
1da177e4 2003
360eb5da 2004 if (cache->mfc_origin == htonl(INADDR_ANY) && true_vifi >= 0) {
660b26dc
ND		 2005 struct mfc_cache *cache_proxy;
2006
2007 /* For an (*,G) entry, we only check that the incomming
2008 * interface is part of the static tree.
2009 */
2010 cache_proxy = ipmr_cache_find_any_parent(mrt, vif);
2011 if (cache_proxy &&
2012 cache_proxy->mfc_un.res.ttls[true_vifi] < 255)
2013 goto forward;
2014 }
2015
7ef8f65d 2016 /* Wrong interface: drop packet and (maybe) send PIM assert. */
4b1f0d33 2017 if (mrt->vif_table[vif].dev != dev) {
c7537967 2018 if (rt_is_output_route(skb_rtable(skb))) {
1da177e4 2019 /* It is our own packet, looped back.
a8cb16dd
ED		 2020 * Very complicated situation...
2021 *
2022 * The best workaround until routing daemons will be
2023 * fixed is not to redistribute packet, if it was
2024 * send through wrong interface. It means, that
2025 * multicast applications WILL NOT work for
2026 * (S,G), which have default multicast route pointing
2027 * to wrong oif. In any case, it is not a good
2028 * idea to use multicasting applications on router.
1da177e4
LT		 2029 */
2030 goto dont_forward;
2031 }
2032
2033 cache->mfc_un.res.wrong_if++;
1da177e4 2034
0c12295a 2035 if (true_vifi >= 0 && mrt->mroute_do_assert &&
1da177e4 2036 /* pimsm uses asserts, when switching from RPT to SPT,
a8cb16dd
ED		 2037 * so that we cannot check that packet arrived on an oif.
2038 * It is bad, but otherwise we would need to move pretty
2039 * large chunk of pimd to kernel. Ough... --ANK
1da177e4 2040 */
0c12295a 2041 (mrt->mroute_do_pim ||
6f9374a9 2042 cache->mfc_un.res.ttls[true_vifi] < 255) &&
e905a9ed 2043 time_after(jiffies,
1da177e4
LT		 2044 cache->mfc_un.res.last_assert + MFC_ASSERT_THRESH)) {
2045 cache->mfc_un.res.last_assert = jiffies;
0c12295a 2046 ipmr_cache_report(mrt, skb, true_vifi, IGMPMSG_WRONGVIF);
1da177e4
LT		 2047 }
2048 goto dont_forward;
2049 }
2050
660b26dc 2051forward:
0c12295a
PM		 2052 mrt->vif_table[vif].pkt_in++;
2053 mrt->vif_table[vif].bytes_in += skb->len;
1da177e4 2054
7ef8f65d 2055 /* Forward the frame */
360eb5da
ND		 2056 if (cache->mfc_origin == htonl(INADDR_ANY) &&
2057 cache->mfc_mcastgrp == htonl(INADDR_ANY)) {
660b26dc
ND		 2058 if (true_vifi >= 0 &&
2059 true_vifi != cache->mfc_parent &&
2060 ip_hdr(skb)->ttl >
2061 cache->mfc_un.res.ttls[cache->mfc_parent]) {
2062 /* It's an (*,*) entry and the packet is not coming from
2063 * the upstream: forward the packet to the upstream
2064 * only.
2065 */
2066 psend = cache->mfc_parent;
2067 goto last_forward;
2068 }
2069 goto dont_forward;
2070 }
a8cb16dd
ED		 2071 for (ct = cache->mfc_un.res.maxvif - 1;
2072 ct >= cache->mfc_un.res.minvif; ct--) {
660b26dc 2073 /* For (*,G) entry, don't forward to the incoming interface */
360eb5da
ND		 2074 if ((cache->mfc_origin != htonl(INADDR_ANY) ||
2075 ct != true_vifi) &&
660b26dc 2076 ip_hdr(skb)->ttl > cache->mfc_un.res.ttls[ct]) {
1da177e4
LT		 2077 if (psend != -1) {
2078 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
a8cb16dd 2079
1da177e4 2080 if (skb2)
a5bc9294
YG		 2081 ipmr_queue_xmit(net, mrt, true_vifi,
2082 skb2, cache, psend);
1da177e4 2083 }
c354e124 2084 psend = ct;
1da177e4
LT		 2085 }
2086 }
660b26dc 2087last_forward:
1da177e4
LT		 2088 if (psend != -1) {
2089 if (local) {
2090 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
a8cb16dd 2091
1da177e4 2092 if (skb2)
a5bc9294
YG		 2093 ipmr_queue_xmit(net, mrt, true_vifi, skb2,
2094 cache, psend);
1da177e4 2095 } else {
a5bc9294 2096 ipmr_queue_xmit(net, mrt, true_vifi, skb, cache, psend);
c4854ec8 2097 return;
1da177e4
LT		 2098 }
2099 }
2100
2101dont_forward:
2102 if (!local)
2103 kfree_skb(skb);
1da177e4
LT		 2104}
2105
417da66f 2106static struct mr_table *ipmr_rt_fib_lookup(struct net *net, struct sk_buff *skb)
ee3f1aaf 2107{
417da66f
DM		 2108 struct rtable *rt = skb_rtable(skb);
2109 struct iphdr *iph = ip_hdr(skb);
da91981b 2110 struct flowi4 fl4 = {
417da66f
DM		 2111 .daddr = iph->daddr,
2112 .saddr = iph->saddr,
b0fe4a31 2113 .flowi4_tos = RT_TOS(iph->tos),
4fd551d7
DM		 2114 .flowi4_oif = (rt_is_output_route(rt) ?
2115 skb->dev->ifindex : 0),
2116 .flowi4_iif = (rt_is_output_route(rt) ?
1fb9489b 2117 LOOPBACK_IFINDEX :
4fd551d7 2118 skb->dev->ifindex),
b4869889 2119 .flowi4_mark = skb->mark,
ee3f1aaf
DM		 2120 };
2121 struct mr_table *mrt;
2122 int err;
2123
da91981b 2124 err = ipmr_fib_lookup(net, &fl4, &mrt);
ee3f1aaf
DM		 2125 if (err)
2126 return ERR_PTR(err);
2127 return mrt;
2128}
1da177e4 2129
7ef8f65d
NA		 2130/* Multicast packets for forwarding arrive here
2131 * Called with rcu_read_lock();
1da177e4 2132 */
1da177e4
LT		 2133int ip_mr_input(struct sk_buff *skb)
2134{
2135 struct mfc_cache *cache;
4feb88e5 2136 struct net *net = dev_net(skb->dev);
511c3f92 2137 int local = skb_rtable(skb)->rt_flags & RTCF_LOCAL;
f0ad0860 2138 struct mr_table *mrt;
bcfc7d33
TW		 2139 struct net_device *dev;
2140
2141 /* skb->dev passed in is the loX master dev for vrfs.
2142 * As there are no vifs associated with loopback devices,
2143 * get the proper interface that does have a vif associated with it.
2144 */
2145 dev = skb->dev;
2146 if (netif_is_l3_master(skb->dev)) {
2147 dev = dev_get_by_index_rcu(net, IPCB(skb)->iif);
2148 if (!dev) {
2149 kfree_skb(skb);
2150 return -ENODEV;
2151 }
2152 }
1da177e4
LT		 2153
2154 /* Packet is looped back after forward, it should not be
a8cb16dd 2155 * forwarded second time, but still can be delivered locally.
1da177e4 2156 */
4c968709 2157 if (IPCB(skb)->flags & IPSKB_FORWARDED)
1da177e4
LT		 2158 goto dont_forward;
2159
417da66f 2160 mrt = ipmr_rt_fib_lookup(net, skb);
ee3f1aaf
DM		 2161 if (IS_ERR(mrt)) {
2162 kfree_skb(skb);
2163 return PTR_ERR(mrt);
e40dbc51 2164 }
1da177e4 2165 if (!local) {
4c968709
ED		 2166 if (IPCB(skb)->opt.router_alert) {
2167 if (ip_call_ra_chain(skb))
2168 return 0;
2169 } else if (ip_hdr(skb)->protocol == IPPROTO_IGMP) {
2170 /* IGMPv1 (and broken IGMPv2 implementations sort of
2171 * Cisco IOS <= 11.2(8)) do not put router alert
2172 * option to IGMP packets destined to routable
2173 * groups. It is very bad, because it means
2174 * that we can forward NO IGMP messages.
2175 */
2176 struct sock *mroute_sk;
2177
2178 mroute_sk = rcu_dereference(mrt->mroute_sk);
2179 if (mroute_sk) {
2180 nf_reset(skb);
2181 raw_rcv(mroute_sk, skb);
2182 return 0;
2183 }
1da177e4
LT		 2184 }
2185 }
2186
a8c9486b 2187 /* already under rcu_read_lock() */
0c12295a 2188 cache = ipmr_cache_find(mrt, ip_hdr(skb)->saddr, ip_hdr(skb)->daddr);
51456b29 2189 if (!cache) {
bcfc7d33 2190 int vif = ipmr_find_vif(mrt, dev);
660b26dc
ND		 2191
2192 if (vif >= 0)
2193 cache = ipmr_cache_find_any(mrt, ip_hdr(skb)->daddr,
2194 vif);
2195 }
1da177e4 2196
7ef8f65d 2197 /* No usable cache entry */
51456b29 2198 if (!cache) {
1da177e4
LT		 2199 int vif;
2200
2201 if (local) {
2202 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
2203 ip_local_deliver(skb);
51456b29 2204 if (!skb2)
1da177e4 2205 return -ENOBUFS;
1da177e4
LT		 2206 skb = skb2;
2207 }
2208
a8c9486b 2209 read_lock(&mrt_lock);
bcfc7d33 2210 vif = ipmr_find_vif(mrt, dev);
1da177e4 2211 if (vif >= 0) {
4b1f0d33 2212 int err2 = ipmr_cache_unresolved(mrt, vif, skb, dev);
1da177e4
LT		 2213 read_unlock(&mrt_lock);
2214
0eae88f3 2215 return err2;
1da177e4
LT		 2216 }
2217 read_unlock(&mrt_lock);
2218 kfree_skb(skb);
2219 return -ENODEV;
2220 }
2221
a8c9486b 2222 read_lock(&mrt_lock);
4b1f0d33 2223 ip_mr_forward(net, mrt, dev, skb, cache, local);
1da177e4
LT		 2224 read_unlock(&mrt_lock);
2225
2226 if (local)
2227 return ip_local_deliver(skb);
2228
2229 return 0;
2230
2231dont_forward:
2232 if (local)
2233 return ip_local_deliver(skb);
2234 kfree_skb(skb);
2235 return 0;
2236}
2237
b1879204 2238#ifdef CONFIG_IP_PIMSM_V1
7ef8f65d 2239/* Handle IGMP messages of PIMv1 */
a8cb16dd 2240int pim_rcv_v1(struct sk_buff *skb)
b1879204
IJ		 2241{
2242 struct igmphdr *pim;
4feb88e5 2243 struct net *net = dev_net(skb->dev);
f0ad0860 2244 struct mr_table *mrt;
b1879204
IJ		 2245
2246 if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr)))
2247 goto drop;
2248
2249 pim = igmp_hdr(skb);
2250
417da66f 2251 mrt = ipmr_rt_fib_lookup(net, skb);
ee3f1aaf
DM		 2252 if (IS_ERR(mrt))
2253 goto drop;
0c12295a 2254 if (!mrt->mroute_do_pim ||
b1879204
IJ		 2255 pim->group != PIM_V1_VERSION || pim->code != PIM_V1_REGISTER)
2256 goto drop;
2257
f0ad0860 2258 if (__pim_rcv(mrt, skb, sizeof(*pim))) {
b1879204
IJ		 2259drop:
2260 kfree_skb(skb);
2261 }
1da177e4
LT		 2262 return 0;
2263}
2264#endif
2265
2266#ifdef CONFIG_IP_PIMSM_V2
a8cb16dd 2267static int pim_rcv(struct sk_buff *skb)
1da177e4
LT		 2268{
2269 struct pimreghdr *pim;
f0ad0860
PM		 2270 struct net *net = dev_net(skb->dev);
2271 struct mr_table *mrt;
1da177e4 2272
b1879204 2273 if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr)))
1da177e4
LT		 2274 goto drop;
2275
9c70220b 2276 pim = (struct pimreghdr *)skb_transport_header(skb);
56245cae 2277 if (pim->type != ((PIM_VERSION << 4) | (PIM_TYPE_REGISTER)) ||
a8cb16dd 2278 (pim->flags & PIM_NULL_REGISTER) ||
e905a9ed 2279 (ip_compute_csum((void *)pim, sizeof(*pim)) != 0 &&
d3bc23e7 2280 csum_fold(skb_checksum(skb, 0, skb->len, 0))))
1da177e4
LT		 2281 goto drop;
2282
417da66f 2283 mrt = ipmr_rt_fib_lookup(net, skb);
ee3f1aaf
DM		 2284 if (IS_ERR(mrt))
2285 goto drop;
f0ad0860 2286 if (__pim_rcv(mrt, skb, sizeof(*pim))) {
b1879204
IJ		 2287drop:
2288 kfree_skb(skb);
2289 }
1da177e4
LT		 2290 return 0;
2291}
2292#endif
2293
cb6a4e46
PM		 2294static int __ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb,
2295 struct mfc_cache *c, struct rtmsg *rtm)
1da177e4 2296{
adfa85e4 2297 struct rta_mfc_stats mfcs;
43b9e127
NA		 2298 struct nlattr *mp_attr;
2299 struct rtnexthop *nhp;
b5036cd4 2300 unsigned long lastuse;
43b9e127 2301 int ct;
1da177e4 2302
7438189b 2303 /* If cache is unresolved, don't try to parse IIF and OIF */
1708ebc9
NA		 2304 if (c->mfc_parent >= MAXVIFS) {
2305 rtm->rtm_flags |= RTNH_F_UNRESOLVED;
7438189b 2306 return -ENOENT;
1708ebc9 2307 }
7438189b 2308
92a395e5 2309 if (VIF_EXISTS(mrt, c->mfc_parent) &&
6853f21f
YM		 2310 nla_put_u32(skb, RTA_IIF,
2311 mrt->vif_table[c->mfc_parent].dev->ifindex) < 0)
92a395e5 2312 return -EMSGSIZE;
1da177e4 2313
c7c0bbea
YG		 2314 if (c->mfc_flags & MFC_OFFLOAD)
2315 rtm->rtm_flags |= RTNH_F_OFFLOAD;
2316
92a395e5
TG		 2317 if (!(mp_attr = nla_nest_start(skb, RTA_MULTIPATH)))
2318 return -EMSGSIZE;
1da177e4
LT		 2319
2320 for (ct = c->mfc_un.res.minvif; ct < c->mfc_un.res.maxvif; ct++) {
0c12295a 2321 if (VIF_EXISTS(mrt, ct) && c->mfc_un.res.ttls[ct] < 255) {
6853f21f
YM		 2322 struct vif_device *vif;
2323
92a395e5
TG		 2324 if (!(nhp = nla_reserve_nohdr(skb, sizeof(*nhp)))) {
2325 nla_nest_cancel(skb, mp_attr);
2326 return -EMSGSIZE;
2327 }
2328
1da177e4
LT		 2329 nhp->rtnh_flags = 0;
2330 nhp->rtnh_hops = c->mfc_un.res.ttls[ct];
6853f21f
YM		 2331 vif = &mrt->vif_table[ct];
2332 nhp->rtnh_ifindex = vif->dev->ifindex;
1da177e4
LT		 2333 nhp->rtnh_len = sizeof(*nhp);
2334 }
2335 }
92a395e5
TG		 2336
2337 nla_nest_end(skb, mp_attr);
2338
b5036cd4
NA		 2339 lastuse = READ_ONCE(c->mfc_un.res.lastuse);
2340 lastuse = time_after_eq(jiffies, lastuse) ? jiffies - lastuse : 0;
2341
adfa85e4
ND		 2342 mfcs.mfcs_packets = c->mfc_un.res.pkt;
2343 mfcs.mfcs_bytes = c->mfc_un.res.bytes;
2344 mfcs.mfcs_wrong_if = c->mfc_un.res.wrong_if;
43b9e127 2345 if (nla_put_64bit(skb, RTA_MFC_STATS, sizeof(mfcs), &mfcs, RTA_PAD) ||
b5036cd4 2346 nla_put_u64_64bit(skb, RTA_EXPIRES, jiffies_to_clock_t(lastuse),
43b9e127 2347 RTA_PAD))
adfa85e4
ND		 2348 return -EMSGSIZE;
2349
1da177e4
LT		 2350 rtm->rtm_type = RTN_MULTICAST;
2351 return 1;
1da177e4
LT		 2352}
2353
9a1b9496
DM		 2354int ipmr_get_route(struct net *net, struct sk_buff *skb,
2355 __be32 saddr, __be32 daddr,
9f09eaea 2356 struct rtmsg *rtm, u32 portid)
1da177e4 2357{
1da177e4 2358 struct mfc_cache *cache;
9a1b9496
DM		 2359 struct mr_table *mrt;
2360 int err;
1da177e4 2361
f0ad0860 2362 mrt = ipmr_get_table(net, RT_TABLE_DEFAULT);
51456b29 2363 if (!mrt)
f0ad0860
PM		 2364 return -ENOENT;
2365
a8c9486b 2366 rcu_read_lock();
9a1b9496 2367 cache = ipmr_cache_find(mrt, saddr, daddr);
51456b29 2368 if (!cache && skb->dev) {
660b26dc 2369 int vif = ipmr_find_vif(mrt, skb->dev);
1da177e4 2370
660b26dc
ND		 2371 if (vif >= 0)
2372 cache = ipmr_cache_find_any(mrt, daddr, vif);
2373 }
51456b29 2374 if (!cache) {
72287490 2375 struct sk_buff *skb2;
eddc9ec5 2376 struct iphdr *iph;
1da177e4 2377 struct net_device *dev;
a8cb16dd 2378 int vif = -1;
1da177e4 2379
1da177e4 2380 dev = skb->dev;
a8c9486b 2381 read_lock(&mrt_lock);
a8cb16dd
ED		 2382 if (dev)
2383 vif = ipmr_find_vif(mrt, dev);
2384 if (vif < 0) {
1da177e4 2385 read_unlock(&mrt_lock);
a8c9486b 2386 rcu_read_unlock();
1da177e4
LT		 2387 return -ENODEV;
2388 }
72287490
AK		 2389 skb2 = skb_clone(skb, GFP_ATOMIC);
2390 if (!skb2) {
2391 read_unlock(&mrt_lock);
a8c9486b 2392 rcu_read_unlock();
72287490
AK		 2393 return -ENOMEM;
2394 }
2395
2cf75070 2396 NETLINK_CB(skb2).portid = portid;
e2d1bca7
ACM		 2397 skb_push(skb2, sizeof(struct iphdr));
2398 skb_reset_network_header(skb2);
eddc9ec5
ACM		 2399 iph = ip_hdr(skb2);
2400 iph->ihl = sizeof(struct iphdr) >> 2;
9a1b9496
DM		 2401 iph->saddr = saddr;
2402 iph->daddr = daddr;
eddc9ec5 2403 iph->version = 0;
4b1f0d33 2404 err = ipmr_cache_unresolved(mrt, vif, skb2, dev);
1da177e4 2405 read_unlock(&mrt_lock);
a8c9486b 2406 rcu_read_unlock();
1da177e4
LT		 2407 return err;
2408 }
2409
a8c9486b 2410 read_lock(&mrt_lock);
cb6a4e46 2411 err = __ipmr_fill_mroute(mrt, skb, cache, rtm);
1da177e4 2412 read_unlock(&mrt_lock);
a8c9486b 2413 rcu_read_unlock();
1da177e4
LT		 2414 return err;
2415}
2416
cb6a4e46 2417static int ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb,
65886f43
ND		 2418 u32 portid, u32 seq, struct mfc_cache *c, int cmd,
2419 int flags)
cb6a4e46
PM		 2420{
2421 struct nlmsghdr *nlh;
2422 struct rtmsg *rtm;
1eb99af5 2423 int err;
cb6a4e46 2424
65886f43 2425 nlh = nlmsg_put(skb, portid, seq, cmd, sizeof(*rtm), flags);
51456b29 2426 if (!nlh)
cb6a4e46
PM		 2427 return -EMSGSIZE;
2428
2429 rtm = nlmsg_data(nlh);
2430 rtm->rtm_family = RTNL_FAMILY_IPMR;
2431 rtm->rtm_dst_len = 32;
2432 rtm->rtm_src_len = 32;
2433 rtm->rtm_tos = 0;
2434 rtm->rtm_table = mrt->id;
f3756b79
DM		 2435 if (nla_put_u32(skb, RTA_TABLE, mrt->id))
2436 goto nla_put_failure;
cb6a4e46
PM		 2437 rtm->rtm_type = RTN_MULTICAST;
2438 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
9a68ac72
ND		 2439 if (c->mfc_flags & MFC_STATIC)
2440 rtm->rtm_protocol = RTPROT_STATIC;
2441 else
2442 rtm->rtm_protocol = RTPROT_MROUTED;
cb6a4e46
PM		 2443 rtm->rtm_flags = 0;
2444
930345ea
JB		 2445 if (nla_put_in_addr(skb, RTA_SRC, c->mfc_origin) ||
2446 nla_put_in_addr(skb, RTA_DST, c->mfc_mcastgrp))
f3756b79 2447 goto nla_put_failure;
1eb99af5
ND		 2448 err = __ipmr_fill_mroute(mrt, skb, c, rtm);
2449 /* do not break the dump if cache is unresolved */
2450 if (err < 0 && err != -ENOENT)
cb6a4e46
PM		 2451 goto nla_put_failure;
2452
053c095a
JB		 2453 nlmsg_end(skb, nlh);
2454 return 0;
cb6a4e46
PM		 2455
2456nla_put_failure:
2457 nlmsg_cancel(skb, nlh);
2458 return -EMSGSIZE;
2459}
2460
8cd3ac9f
ND		 2461static size_t mroute_msgsize(bool unresolved, int maxvif)
2462{
2463 size_t len =
2464 NLMSG_ALIGN(sizeof(struct rtmsg))
2465 + nla_total_size(4) /* RTA_TABLE */
2466 + nla_total_size(4) /* RTA_SRC */
2467 + nla_total_size(4) /* RTA_DST */
2468 ;
2469
2470 if (!unresolved)
2471 len = len
2472 + nla_total_size(4) /* RTA_IIF */
2473 + nla_total_size(0) /* RTA_MULTIPATH */
2474 + maxvif * NLA_ALIGN(sizeof(struct rtnexthop))
2475 /* RTA_MFC_STATS */
a9a08042 2476 + nla_total_size_64bit(sizeof(struct rta_mfc_stats))
8cd3ac9f
ND		 2477 ;
2478
2479 return len;
2480}
2481
2482static void mroute_netlink_event(struct mr_table *mrt, struct mfc_cache *mfc,
2483 int cmd)
2484{
2485 struct net *net = read_pnet(&mrt->net);
2486 struct sk_buff *skb;
2487 int err = -ENOBUFS;
2488
2489 skb = nlmsg_new(mroute_msgsize(mfc->mfc_parent >= MAXVIFS, mrt->maxvif),
2490 GFP_ATOMIC);
51456b29 2491 if (!skb)
8cd3ac9f
ND		 2492 goto errout;
2493
65886f43 2494 err = ipmr_fill_mroute(mrt, skb, 0, 0, mfc, cmd, 0);
8cd3ac9f
ND		 2495 if (err < 0)
2496 goto errout;
2497
2498 rtnl_notify(skb, net, 0, RTNLGRP_IPV4_MROUTE, NULL, GFP_ATOMIC);
2499 return;
2500
2501errout:
2502 kfree_skb(skb);
2503 if (err < 0)
2504 rtnl_set_sk_err(net, RTNLGRP_IPV4_MROUTE, err);
2505}
2506
5a645dd8
JG		 2507static size_t igmpmsg_netlink_msgsize(size_t payloadlen)
2508{
2509 size_t len =
2510 NLMSG_ALIGN(sizeof(struct rtgenmsg))
2511 + nla_total_size(1) /* IPMRA_CREPORT_MSGTYPE */
2512 + nla_total_size(4) /* IPMRA_CREPORT_VIF_ID */
2513 + nla_total_size(4) /* IPMRA_CREPORT_SRC_ADDR */
2514 + nla_total_size(4) /* IPMRA_CREPORT_DST_ADDR */
2515 /* IPMRA_CREPORT_PKT */
2516 + nla_total_size(payloadlen)
2517 ;
2518
2519 return len;
2520}
2521
2522static void igmpmsg_netlink_event(struct mr_table *mrt, struct sk_buff *pkt)
2523{
2524 struct net *net = read_pnet(&mrt->net);
2525 struct nlmsghdr *nlh;
2526 struct rtgenmsg *rtgenm;
2527 struct igmpmsg *msg;
2528 struct sk_buff *skb;
2529 struct nlattr *nla;
2530 int payloadlen;
2531
2532 payloadlen = pkt->len - sizeof(struct igmpmsg);
2533 msg = (struct igmpmsg *)skb_network_header(pkt);
2534
2535 skb = nlmsg_new(igmpmsg_netlink_msgsize(payloadlen), GFP_ATOMIC);
2536 if (!skb)
2537 goto errout;
2538
2539 nlh = nlmsg_put(skb, 0, 0, RTM_NEWCACHEREPORT,
2540 sizeof(struct rtgenmsg), 0);
2541 if (!nlh)
2542 goto errout;
2543 rtgenm = nlmsg_data(nlh);
2544 rtgenm->rtgen_family = RTNL_FAMILY_IPMR;
2545 if (nla_put_u8(skb, IPMRA_CREPORT_MSGTYPE, msg->im_msgtype) ||
2546 nla_put_u32(skb, IPMRA_CREPORT_VIF_ID, msg->im_vif) ||
2547 nla_put_in_addr(skb, IPMRA_CREPORT_SRC_ADDR,
2548 msg->im_src.s_addr) ||
2549 nla_put_in_addr(skb, IPMRA_CREPORT_DST_ADDR,
2550 msg->im_dst.s_addr))
2551 goto nla_put_failure;
2552
2553 nla = nla_reserve(skb, IPMRA_CREPORT_PKT, payloadlen);
2554 if (!nla || skb_copy_bits(pkt, sizeof(struct igmpmsg),
2555 nla_data(nla), payloadlen))
2556 goto nla_put_failure;
2557
2558 nlmsg_end(skb, nlh);
2559
2560 rtnl_notify(skb, net, 0, RTNLGRP_IPV4_MROUTE_R, NULL, GFP_ATOMIC);
2561 return;
2562
2563nla_put_failure:
2564 nlmsg_cancel(skb, nlh);
2565errout:
2566 kfree_skb(skb);
2567 rtnl_set_sk_err(net, RTNLGRP_IPV4_MROUTE_R, -ENOBUFS);
2568}
2569
4f75ba69
DS		 2570static int ipmr_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh,
2571 struct netlink_ext_ack *extack)
2572{
2573 struct net *net = sock_net(in_skb->sk);
2574 struct nlattr *tb[RTA_MAX + 1];
2575 struct sk_buff *skb = NULL;
2576 struct mfc_cache *cache;
2577 struct mr_table *mrt;
2578 struct rtmsg *rtm;
2579 __be32 src, grp;
2580 u32 tableid;
2581 int err;
2582
2583 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX,
2584 rtm_ipv4_policy, extack);
2585 if (err < 0)
2586 goto errout;
2587
2588 rtm = nlmsg_data(nlh);
2589
2590 src = tb[RTA_SRC] ? nla_get_in_addr(tb[RTA_SRC]) : 0;
2591 grp = tb[RTA_DST] ? nla_get_in_addr(tb[RTA_DST]) : 0;
2592 tableid = tb[RTA_TABLE] ? nla_get_u32(tb[RTA_TABLE]) : 0;
2593
2594 mrt = ipmr_get_table(net, tableid ? tableid : RT_TABLE_DEFAULT);
2e3d232e
DC		 2595 if (!mrt) {
2596 err = -ENOENT;
4f75ba69
DS		 2597 goto errout_free;
2598 }
2599
2600 /* entries are added/deleted only under RTNL */
2601 rcu_read_lock();
2602 cache = ipmr_cache_find(mrt, src, grp);
2603 rcu_read_unlock();
2604 if (!cache) {
2605 err = -ENOENT;
2606 goto errout_free;
2607 }
2608
2609 skb = nlmsg_new(mroute_msgsize(false, mrt->maxvif), GFP_KERNEL);
2610 if (!skb) {
2611 err = -ENOBUFS;
2612 goto errout_free;
2613 }
2614
2615 err = ipmr_fill_mroute(mrt, skb, NETLINK_CB(in_skb).portid,
2616 nlh->nlmsg_seq, cache,
2617 RTM_NEWROUTE, 0);
2618 if (err < 0)
2619 goto errout_free;
2620
2621 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
2622
2623errout:
2624 return err;
2625
2626errout_free:
2627 kfree_skb(skb);
2628 goto errout;
2629}
2630
cb6a4e46
PM		 2631static int ipmr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb)
2632{
2633 struct net *net = sock_net(skb->sk);
2634 struct mr_table *mrt;
2635 struct mfc_cache *mfc;
2636 unsigned int t = 0, s_t;
cb6a4e46
PM		 2637 unsigned int e = 0, s_e;
2638
2639 s_t = cb->args[0];
8fb472c0 2640 s_e = cb->args[1];
cb6a4e46 2641
a8c9486b 2642 rcu_read_lock();
cb6a4e46
PM		 2643 ipmr_for_each_table(mrt, net) {
2644 if (t < s_t)
2645 goto next_table;
8fb472c0
NA		 2646 list_for_each_entry_rcu(mfc, &mrt->mfc_cache_list, list) {
2647 if (e < s_e)
2648 goto next_entry;
2649 if (ipmr_fill_mroute(mrt, skb,
2650 NETLINK_CB(cb->skb).portid,
2651 cb->nlh->nlmsg_seq,
2652 mfc, RTM_NEWROUTE,
2653 NLM_F_MULTI) < 0)
2654 goto done;
cb6a4e46 2655next_entry:
8fb472c0 2656 e++;
cb6a4e46 2657 }
8fb472c0
NA		 2658 e = 0;
2659 s_e = 0;
2660
1eb99af5
ND		 2661 spin_lock_bh(&mfc_unres_lock);
2662 list_for_each_entry(mfc, &mrt->mfc_unres_queue, list) {
2663 if (e < s_e)
2664 goto next_entry2;
2665 if (ipmr_fill_mroute(mrt, skb,
2666 NETLINK_CB(cb->skb).portid,
2667 cb->nlh->nlmsg_seq,
65886f43
ND		 2668 mfc, RTM_NEWROUTE,
2669 NLM_F_MULTI) < 0) {
1eb99af5
ND		 2670 spin_unlock_bh(&mfc_unres_lock);
2671 goto done;
2672 }
2673next_entry2:
2674 e++;
2675 }
2676 spin_unlock_bh(&mfc_unres_lock);
8fb472c0
NA		 2677 e = 0;
2678 s_e = 0;
cb6a4e46
PM		 2679next_table:
2680 t++;
2681 }
2682done:
a8c9486b 2683 rcu_read_unlock();
cb6a4e46 2684
8fb472c0 2685 cb->args[1] = e;
cb6a4e46
PM		 2686 cb->args[0] = t;
2687
2688 return skb->len;
2689}
2690
ccbb0aa6
NA		 2691static const struct nla_policy rtm_ipmr_policy[RTA_MAX + 1] = {
2692 [RTA_SRC] = { .type = NLA_U32 },
2693 [RTA_DST] = { .type = NLA_U32 },
2694 [RTA_IIF] = { .type = NLA_U32 },
2695 [RTA_TABLE] = { .type = NLA_U32 },
2696 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
2697};
2698
2699static bool ipmr_rtm_validate_proto(unsigned char rtm_protocol)
2700{
2701 switch (rtm_protocol) {
2702 case RTPROT_STATIC:
2703 case RTPROT_MROUTED:
2704 return true;
2705 }
2706 return false;
2707}
2708
2709static int ipmr_nla_get_ttls(const struct nlattr *nla, struct mfcctl *mfcc)
2710{
2711 struct rtnexthop *rtnh = nla_data(nla);
2712 int remaining = nla_len(nla), vifi = 0;
2713
2714 while (rtnh_ok(rtnh, remaining)) {
2715 mfcc->mfcc_ttls[vifi] = rtnh->rtnh_hops;
2716 if (++vifi == MAXVIFS)
2717 break;
2718 rtnh = rtnh_next(rtnh, &remaining);
2719 }
2720
2721 return remaining > 0 ? -EINVAL : vifi;
2722}
2723
2724/* returns < 0 on error, 0 for ADD_MFC and 1 for ADD_MFC_PROXY */
2725static int rtm_to_ipmr_mfcc(struct net *net, struct nlmsghdr *nlh,
2726 struct mfcctl *mfcc, int *mrtsock,
c21ef3e3
DA		 2727 struct mr_table **mrtret,
2728 struct netlink_ext_ack *extack)
ccbb0aa6
NA		 2729{
2730 struct net_device *dev = NULL;
2731 u32 tblid = RT_TABLE_DEFAULT;
2732 struct mr_table *mrt;
2733 struct nlattr *attr;
2734 struct rtmsg *rtm;
2735 int ret, rem;
2736
fceb6435 2737 ret = nlmsg_validate(nlh, sizeof(*rtm), RTA_MAX, rtm_ipmr_policy,
c21ef3e3 2738 extack);
ccbb0aa6
NA		 2739 if (ret < 0)
2740 goto out;
2741 rtm = nlmsg_data(nlh);
2742
2743 ret = -EINVAL;
2744 if (rtm->rtm_family != RTNL_FAMILY_IPMR || rtm->rtm_dst_len != 32 ||
2745 rtm->rtm_type != RTN_MULTICAST ||
2746 rtm->rtm_scope != RT_SCOPE_UNIVERSE ||
2747 !ipmr_rtm_validate_proto(rtm->rtm_protocol))
2748 goto out;
2749
2750 memset(mfcc, 0, sizeof(*mfcc));
2751 mfcc->mfcc_parent = -1;
2752 ret = 0;
2753 nlmsg_for_each_attr(attr, nlh, sizeof(struct rtmsg), rem) {
2754 switch (nla_type(attr)) {
2755 case RTA_SRC:
2756 mfcc->mfcc_origin.s_addr = nla_get_be32(attr);
2757 break;
2758 case RTA_DST:
2759 mfcc->mfcc_mcastgrp.s_addr = nla_get_be32(attr);
2760 break;
2761 case RTA_IIF:
2762 dev = __dev_get_by_index(net, nla_get_u32(attr));
2763 if (!dev) {
2764 ret = -ENODEV;
2765 goto out;
2766 }
2767 break;
2768 case RTA_MULTIPATH:
2769 if (ipmr_nla_get_ttls(attr, mfcc) < 0) {
2770 ret = -EINVAL;
2771 goto out;
2772 }
2773 break;
2774 case RTA_PREFSRC:
2775 ret = 1;
2776 break;
2777 case RTA_TABLE:
2778 tblid = nla_get_u32(attr);
2779 break;
2780 }
2781 }
2782 mrt = ipmr_get_table(net, tblid);
2783 if (!mrt) {
2784 ret = -ENOENT;
2785 goto out;
2786 }
2787 *mrtret = mrt;
2788 *mrtsock = rtm->rtm_protocol == RTPROT_MROUTED ? 1 : 0;
2789 if (dev)
2790 mfcc->mfcc_parent = ipmr_find_vif(mrt, dev);
2791
2792out:
2793 return ret;
2794}
2795
2796/* takes care of both newroute and delroute */
c21ef3e3
DA		 2797static int ipmr_rtm_route(struct sk_buff *skb, struct nlmsghdr *nlh,
2798 struct netlink_ext_ack *extack)
ccbb0aa6
NA		 2799{
2800 struct net *net = sock_net(skb->sk);
2801 int ret, mrtsock, parent;
2802 struct mr_table *tbl;
2803 struct mfcctl mfcc;
2804
2805 mrtsock = 0;
2806 tbl = NULL;
c21ef3e3 2807 ret = rtm_to_ipmr_mfcc(net, nlh, &mfcc, &mrtsock, &tbl, extack);
ccbb0aa6
NA		 2808 if (ret < 0)
2809 return ret;
2810
2811 parent = ret ? mfcc.mfcc_parent : -1;
2812 if (nlh->nlmsg_type == RTM_NEWROUTE)
2813 return ipmr_mfc_add(net, tbl, &mfcc, mrtsock, parent);
2814 else
2815 return ipmr_mfc_delete(tbl, &mfcc, parent);
2816}
2817
772c344d
NA		 2818static bool ipmr_fill_table(struct mr_table *mrt, struct sk_buff *skb)
2819{
2820 u32 queue_len = atomic_read(&mrt->cache_resolve_queue_len);
2821
2822 if (nla_put_u32(skb, IPMRA_TABLE_ID, mrt->id) ||
2823 nla_put_u32(skb, IPMRA_TABLE_CACHE_RES_QUEUE_LEN, queue_len) ||
2824 nla_put_s32(skb, IPMRA_TABLE_MROUTE_REG_VIF_NUM,
2825 mrt->mroute_reg_vif_num) ||
2826 nla_put_u8(skb, IPMRA_TABLE_MROUTE_DO_ASSERT,
2827 mrt->mroute_do_assert) ||
2828 nla_put_u8(skb, IPMRA_TABLE_MROUTE_DO_PIM, mrt->mroute_do_pim))
2829 return false;
2830
2831 return true;
2832}
2833
2834static bool ipmr_fill_vif(struct mr_table *mrt, u32 vifid, struct sk_buff *skb)
2835{
2836 struct nlattr *vif_nest;
2837 struct vif_device *vif;
2838
2839 /* if the VIF doesn't exist just continue */
2840 if (!VIF_EXISTS(mrt, vifid))
2841 return true;
2842
2843 vif = &mrt->vif_table[vifid];
2844 vif_nest = nla_nest_start(skb, IPMRA_VIF);
2845 if (!vif_nest)
2846 return false;
2847 if (nla_put_u32(skb, IPMRA_VIFA_IFINDEX, vif->dev->ifindex) ||
2848 nla_put_u32(skb, IPMRA_VIFA_VIF_ID, vifid) ||
2849 nla_put_u16(skb, IPMRA_VIFA_FLAGS, vif->flags) ||
2850 nla_put_u64_64bit(skb, IPMRA_VIFA_BYTES_IN, vif->bytes_in,
2851 IPMRA_VIFA_PAD) ||
2852 nla_put_u64_64bit(skb, IPMRA_VIFA_BYTES_OUT, vif->bytes_out,
2853 IPMRA_VIFA_PAD) ||
2854 nla_put_u64_64bit(skb, IPMRA_VIFA_PACKETS_IN, vif->pkt_in,
2855 IPMRA_VIFA_PAD) ||
2856 nla_put_u64_64bit(skb, IPMRA_VIFA_PACKETS_OUT, vif->pkt_out,
2857 IPMRA_VIFA_PAD) ||
2858 nla_put_be32(skb, IPMRA_VIFA_LOCAL_ADDR, vif->local) ||
2859 nla_put_be32(skb, IPMRA_VIFA_REMOTE_ADDR, vif->remote)) {
2860 nla_nest_cancel(skb, vif_nest);
2861 return false;
2862 }
2863 nla_nest_end(skb, vif_nest);
2864
2865 return true;
2866}
2867
2868static int ipmr_rtm_dumplink(struct sk_buff *skb, struct netlink_callback *cb)
2869{
2870 struct net *net = sock_net(skb->sk);
2871 struct nlmsghdr *nlh = NULL;
2872 unsigned int t = 0, s_t;
2873 unsigned int e = 0, s_e;
2874 struct mr_table *mrt;
2875
2876 s_t = cb->args[0];
2877 s_e = cb->args[1];
2878
2879 ipmr_for_each_table(mrt, net) {
2880 struct nlattr *vifs, *af;
2881 struct ifinfomsg *hdr;
2882 u32 i;
2883
2884 if (t < s_t)
2885 goto skip_table;
2886 nlh = nlmsg_put(skb, NETLINK_CB(cb->skb).portid,
2887 cb->nlh->nlmsg_seq, RTM_NEWLINK,
2888 sizeof(*hdr), NLM_F_MULTI);
2889 if (!nlh)
2890 break;
2891
2892 hdr = nlmsg_data(nlh);
2893 memset(hdr, 0, sizeof(*hdr));
2894 hdr->ifi_family = RTNL_FAMILY_IPMR;
2895
2896 af = nla_nest_start(skb, IFLA_AF_SPEC);
2897 if (!af) {
2898 nlmsg_cancel(skb, nlh);
2899 goto out;
2900 }
2901
2902 if (!ipmr_fill_table(mrt, skb)) {
2903 nlmsg_cancel(skb, nlh);
2904 goto out;
2905 }
2906
2907 vifs = nla_nest_start(skb, IPMRA_TABLE_VIFS);
2908 if (!vifs) {
2909 nla_nest_end(skb, af);
2910 nlmsg_end(skb, nlh);
2911 goto out;
2912 }
2913 for (i = 0; i < mrt->maxvif; i++) {
2914 if (e < s_e)
2915 goto skip_entry;
2916 if (!ipmr_fill_vif(mrt, i, skb)) {
2917 nla_nest_end(skb, vifs);
2918 nla_nest_end(skb, af);
2919 nlmsg_end(skb, nlh);
2920 goto out;
2921 }
2922skip_entry:
2923 e++;
2924 }
2925 s_e = 0;
2926 e = 0;
2927 nla_nest_end(skb, vifs);
2928 nla_nest_end(skb, af);
2929 nlmsg_end(skb, nlh);
2930skip_table:
2931 t++;
2932 }
2933
2934out:
2935 cb->args[1] = e;
2936 cb->args[0] = t;
2937
2938 return skb->len;
2939}
2940
e905a9ed 2941#ifdef CONFIG_PROC_FS
7ef8f65d
NA		 2942/* The /proc interfaces to multicast routing :
2943 * /proc/net/ip_mr_cache & /proc/net/ip_mr_vif
1da177e4
LT		 2944 */
2945struct ipmr_vif_iter {
f6bb4514 2946 struct seq_net_private p;
f0ad0860 2947 struct mr_table *mrt;
1da177e4
LT		 2948 int ct;
2949};
2950
f6bb4514 2951static struct vif_device *ipmr_vif_seq_idx(struct net *net,
6853f21f
YM		 2952 struct ipmr_vif_iter *iter,
2953 loff_t pos)
1da177e4 2954{
f0ad0860 2955 struct mr_table *mrt = iter->mrt;
0c12295a
PM		 2956
2957 for (iter->ct = 0; iter->ct < mrt->maxvif; ++iter->ct) {
2958 if (!VIF_EXISTS(mrt, iter->ct))
1da177e4 2959 continue;
e905a9ed 2960 if (pos-- == 0)
0c12295a 2961 return &mrt->vif_table[iter->ct];
1da177e4
LT		 2962 }
2963 return NULL;
2964}
2965
2966static void *ipmr_vif_seq_start(struct seq_file *seq, loff_t *pos)
ba93ef74 2967 __acquires(mrt_lock)
1da177e4 2968{
f0ad0860 2969 struct ipmr_vif_iter *iter = seq->private;
f6bb4514 2970 struct net *net = seq_file_net(seq);
f0ad0860
PM		 2971 struct mr_table *mrt;
2972
2973 mrt = ipmr_get_table(net, RT_TABLE_DEFAULT);
51456b29 2974 if (!mrt)
f0ad0860
PM		 2975 return ERR_PTR(-ENOENT);
2976
2977 iter->mrt = mrt;
f6bb4514 2978
1da177e4 2979 read_lock(&mrt_lock);
f6bb4514 2980 return *pos ? ipmr_vif_seq_idx(net, seq->private, *pos - 1)
1da177e4
LT		 2981 : SEQ_START_TOKEN;
2982}
2983
2984static void *ipmr_vif_seq_next(struct seq_file *seq, void *v, loff_t *pos)
2985{
2986 struct ipmr_vif_iter *iter = seq->private;
f6bb4514 2987 struct net *net = seq_file_net(seq);
f0ad0860 2988 struct mr_table *mrt = iter->mrt;
1da177e4
LT		 2989
2990 ++*pos;
2991 if (v == SEQ_START_TOKEN)
f6bb4514 2992 return ipmr_vif_seq_idx(net, iter, 0);
e905a9ed 2993
0c12295a
PM		 2994 while (++iter->ct < mrt->maxvif) {
2995 if (!VIF_EXISTS(mrt, iter->ct))
1da177e4 2996 continue;
0c12295a 2997 return &mrt->vif_table[iter->ct];
1da177e4
LT		 2998 }
2999 return NULL;
3000}
3001
3002static void ipmr_vif_seq_stop(struct seq_file *seq, void *v)
ba93ef74 3003 __releases(mrt_lock)
1da177e4
LT		 3004{
3005 read_unlock(&mrt_lock);
3006}
3007
3008static int ipmr_vif_seq_show(struct seq_file *seq, void *v)
3009{
f0ad0860
PM		 3010 struct ipmr_vif_iter *iter = seq->private;
3011 struct mr_table *mrt = iter->mrt;
f6bb4514 3012
1da177e4 3013 if (v == SEQ_START_TOKEN) {
e905a9ed 3014 seq_puts(seq,
1da177e4
LT		 3015 "Interface BytesIn PktsIn BytesOut PktsOut Flags Local Remote\n");
3016 } else {
3017 const struct vif_device *vif = v;
6853f21f
YM		 3018 const char *name = vif->dev ?
3019 vif->dev->name : "none";
1da177e4
LT		 3020
3021 seq_printf(seq,
91e6dd82 3022 "%2td %-10s %8ld %7ld %8ld %7ld %05X %08X %08X\n",
0c12295a 3023 vif - mrt->vif_table,
e905a9ed 3024 name, vif->bytes_in, vif->pkt_in,
1da177e4
LT		 3025 vif->bytes_out, vif->pkt_out,
3026 vif->flags, vif->local, vif->remote);
3027 }
3028 return 0;
3029}
3030
f690808e 3031static const struct seq_operations ipmr_vif_seq_ops = {
1da177e4
LT		 3032 .start = ipmr_vif_seq_start,
3033 .next = ipmr_vif_seq_next,
3034 .stop = ipmr_vif_seq_stop,
3035 .show = ipmr_vif_seq_show,
3036};
3037
3038static int ipmr_vif_open(struct inode *inode, struct file *file)
3039{
f6bb4514
BT		 3040 return seq_open_net(inode, file, &ipmr_vif_seq_ops,
3041 sizeof(struct ipmr_vif_iter));
1da177e4
LT		 3042}
3043
9a32144e 3044static const struct file_operations ipmr_vif_fops = {
1da177e4
LT		 3045 .open = ipmr_vif_open,
3046 .read = seq_read,
3047 .llseek = seq_lseek,
f6bb4514 3048 .release = seq_release_net,
1da177e4
LT		 3049};
3050
3051struct ipmr_mfc_iter {
f6bb4514 3052 struct seq_net_private p;
f0ad0860 3053 struct mr_table *mrt;
862465f2 3054 struct list_head *cache;
1da177e4
LT		 3055};
3056
f6bb4514
BT		 3057static struct mfc_cache *ipmr_mfc_seq_idx(struct net *net,
3058 struct ipmr_mfc_iter *it, loff_t pos)
1da177e4 3059{
f0ad0860 3060 struct mr_table *mrt = it->mrt;
1da177e4
LT		 3061 struct mfc_cache *mfc;
3062
a8c9486b 3063 rcu_read_lock();
8fb472c0
NA		 3064 it->cache = &mrt->mfc_cache_list;
3065 list_for_each_entry_rcu(mfc, &mrt->mfc_cache_list, list)
3066 if (pos-- == 0)
3067 return mfc;
a8c9486b 3068 rcu_read_unlock();
1da177e4 3069
1da177e4 3070 spin_lock_bh(&mfc_unres_lock);
0c12295a 3071 it->cache = &mrt->mfc_unres_queue;
862465f2 3072 list_for_each_entry(mfc, it->cache, list)
e258beb2 3073 if (pos-- == 0)
1da177e4
LT		 3074 return mfc;
3075 spin_unlock_bh(&mfc_unres_lock);
3076
3077 it->cache = NULL;
3078 return NULL;
3079}
3080
3081
3082static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos)
3083{
3084 struct ipmr_mfc_iter *it = seq->private;
f6bb4514 3085 struct net *net = seq_file_net(seq);
f0ad0860 3086 struct mr_table *mrt;
f6bb4514 3087
f0ad0860 3088 mrt = ipmr_get_table(net, RT_TABLE_DEFAULT);
51456b29 3089 if (!mrt)
f0ad0860 3090 return ERR_PTR(-ENOENT);
f6bb4514 3091
f0ad0860 3092 it->mrt = mrt;
1da177e4 3093 it->cache = NULL;
f6bb4514 3094 return *pos ? ipmr_mfc_seq_idx(net, seq->private, *pos - 1)
1da177e4
LT		 3095 : SEQ_START_TOKEN;
3096}
3097
3098static void *ipmr_mfc_seq_next(struct seq_file *seq, void *v, loff_t *pos)
3099{
1da177e4 3100 struct ipmr_mfc_iter *it = seq->private;
f6bb4514 3101 struct net *net = seq_file_net(seq);
f0ad0860 3102 struct mr_table *mrt = it->mrt;
8fb472c0 3103 struct mfc_cache *mfc = v;
1da177e4
LT		 3104
3105 ++*pos;
3106
3107 if (v == SEQ_START_TOKEN)
f6bb4514 3108 return ipmr_mfc_seq_idx(net, seq->private, 0);
1da177e4 3109
862465f2
PM		 3110 if (mfc->list.next != it->cache)
3111 return list_entry(mfc->list.next, struct mfc_cache, list);
e905a9ed 3112
0c12295a 3113 if (it->cache == &mrt->mfc_unres_queue)
1da177e4
LT		 3114 goto end_of_list;
3115
1da177e4 3116 /* exhausted cache_array, show unresolved */
a8c9486b 3117 rcu_read_unlock();
0c12295a 3118 it->cache = &mrt->mfc_unres_queue;
e905a9ed 3119
1da177e4 3120 spin_lock_bh(&mfc_unres_lock);
862465f2
PM		 3121 if (!list_empty(it->cache))
3122 return list_first_entry(it->cache, struct mfc_cache, list);
1da177e4 3123
a8cb16dd 3124end_of_list:
1da177e4
LT		 3125 spin_unlock_bh(&mfc_unres_lock);
3126 it->cache = NULL;
3127
3128 return NULL;
3129}
3130
3131static void ipmr_mfc_seq_stop(struct seq_file *seq, void *v)
3132{
3133 struct ipmr_mfc_iter *it = seq->private;
f0ad0860 3134 struct mr_table *mrt = it->mrt;
1da177e4 3135
0c12295a 3136 if (it->cache == &mrt->mfc_unres_queue)
1da177e4 3137 spin_unlock_bh(&mfc_unres_lock);
8fb472c0 3138 else if (it->cache == &mrt->mfc_cache_list)
a8c9486b 3139 rcu_read_unlock();
1da177e4
LT		 3140}
3141
3142static int ipmr_mfc_seq_show(struct seq_file *seq, void *v)
3143{
3144 int n;
3145
3146 if (v == SEQ_START_TOKEN) {
e905a9ed 3147 seq_puts(seq,
1da177e4
LT		 3148 "Group Origin Iif Pkts Bytes Wrong Oifs\n");
3149 } else {
3150 const struct mfc_cache *mfc = v;
3151 const struct ipmr_mfc_iter *it = seq->private;
f0ad0860 3152 const struct mr_table *mrt = it->mrt;
e905a9ed 3153
0eae88f3
ED		 3154 seq_printf(seq, "%08X %08X %-3hd",
3155 (__force u32) mfc->mfc_mcastgrp,
3156 (__force u32) mfc->mfc_origin,
1ea472e2 3157 mfc->mfc_parent);
1da177e4 3158
0c12295a 3159 if (it->cache != &mrt->mfc_unres_queue) {
1ea472e2
BT		 3160 seq_printf(seq, " %8lu %8lu %8lu",
3161 mfc->mfc_un.res.pkt,
3162 mfc->mfc_un.res.bytes,
3163 mfc->mfc_un.res.wrong_if);
132adf54 3164 for (n = mfc->mfc_un.res.minvif;
a8cb16dd 3165 n < mfc->mfc_un.res.maxvif; n++) {
0c12295a 3166 if (VIF_EXISTS(mrt, n) &&
cf958ae3
BT		 3167 mfc->mfc_un.res.ttls[n] < 255)
3168 seq_printf(seq,
e905a9ed 3169 " %2d:%-3d",
1da177e4
LT		 3170 n, mfc->mfc_un.res.ttls[n]);
3171 }
1ea472e2
BT		 3172 } else {
3173 /* unresolved mfc_caches don't contain
3174 * pkt, bytes and wrong_if values
3175 */
3176 seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul);
1da177e4
LT		 3177 }
3178 seq_putc(seq, '\n');
3179 }
3180 return 0;
3181}
3182
f690808e 3183static const struct seq_operations ipmr_mfc_seq_ops = {
1da177e4
LT		 3184 .start = ipmr_mfc_seq_start,
3185 .next = ipmr_mfc_seq_next,
3186 .stop = ipmr_mfc_seq_stop,
3187 .show = ipmr_mfc_seq_show,
3188};
3189
3190static int ipmr_mfc_open(struct inode *inode, struct file *file)
3191{
f6bb4514
BT		 3192 return seq_open_net(inode, file, &ipmr_mfc_seq_ops,
3193 sizeof(struct ipmr_mfc_iter));
1da177e4
LT		 3194}
3195
9a32144e 3196static const struct file_operations ipmr_mfc_fops = {
1da177e4
LT		 3197 .open = ipmr_mfc_open,
3198 .read = seq_read,
3199 .llseek = seq_lseek,
f6bb4514 3200 .release = seq_release_net,
1da177e4 3201};
e905a9ed 3202#endif
1da177e4
LT		 3203
3204#ifdef CONFIG_IP_PIMSM_V2
32613090 3205static const struct net_protocol pim_protocol = {
1da177e4 3206 .handler = pim_rcv,
403dbb97 3207 .netns_ok = 1,
1da177e4
LT		 3208};
3209#endif
3210
4d65b948
YG		 3211static unsigned int ipmr_seq_read(struct net *net)
3212{
3213 ASSERT_RTNL();
3214
3215 return net->ipv4.ipmr_seq + ipmr_rules_seq_read(net);
3216}
3217
3218static int ipmr_dump(struct net *net, struct notifier_block *nb)
3219{
3220 struct mr_table *mrt;
3221 int err;
3222
3223 err = ipmr_rules_dump(net, nb);
3224 if (err)
3225 return err;
3226
3227 ipmr_for_each_table(mrt, net) {
3228 struct vif_device *v = &mrt->vif_table[0];
3229 struct mfc_cache *mfc;
3230 int vifi;
3231
3232 /* Notifiy on table VIF entries */
3233 read_lock(&mrt_lock);
3234 for (vifi = 0; vifi < mrt->maxvif; vifi++, v++) {
3235 if (!v->dev)
3236 continue;
3237
3238 call_ipmr_vif_entry_notifier(nb, net, FIB_EVENT_VIF_ADD,
3239 v, vifi, mrt->id);
3240 }
3241 read_unlock(&mrt_lock);
3242
3243 /* Notify on table MFC entries */
3244 list_for_each_entry_rcu(mfc, &mrt->mfc_cache_list, list)
3245 call_ipmr_mfc_entry_notifier(nb, net,
3246 FIB_EVENT_ENTRY_ADD, mfc,
3247 mrt->id);
3248 }
3249
3250 return 0;
3251}
3252
3253static const struct fib_notifier_ops ipmr_notifier_ops_template = {
3254 .family = RTNL_FAMILY_IPMR,
3255 .fib_seq_read = ipmr_seq_read,
3256 .fib_dump = ipmr_dump,
3257 .owner = THIS_MODULE,
3258};
3259
ef739d8a 3260static int __net_init ipmr_notifier_init(struct net *net)
4d65b948
YG		 3261{
3262 struct fib_notifier_ops *ops;
3263
3264 net->ipv4.ipmr_seq = 0;
3265
3266 ops = fib_notifier_ops_register(&ipmr_notifier_ops_template, net);
3267 if (IS_ERR(ops))
3268 return PTR_ERR(ops);
3269 net->ipv4.ipmr_notifier_ops = ops;
3270
3271 return 0;
3272}
3273
3274static void __net_exit ipmr_notifier_exit(struct net *net)
3275{
3276 fib_notifier_ops_unregister(net->ipv4.ipmr_notifier_ops);
3277 net->ipv4.ipmr_notifier_ops = NULL;
3278}
3279
7ef8f65d 3280/* Setup for IP multicast routing */
cf958ae3
BT		 3281static int __net_init ipmr_net_init(struct net *net)
3282{
f0ad0860 3283 int err;
cf958ae3 3284
4d65b948
YG		 3285 err = ipmr_notifier_init(net);
3286 if (err)
3287 goto ipmr_notifier_fail;
3288
f0ad0860
PM		 3289 err = ipmr_rules_init(net);
3290 if (err < 0)
4d65b948 3291 goto ipmr_rules_fail;
f6bb4514
BT		 3292
3293#ifdef CONFIG_PROC_FS
3294 err = -ENOMEM;
d4beaa66 3295 if (!proc_create("ip_mr_vif", 0, net->proc_net, &ipmr_vif_fops))
f6bb4514 3296 goto proc_vif_fail;
d4beaa66 3297 if (!proc_create("ip_mr_cache", 0, net->proc_net, &ipmr_mfc_fops))
f6bb4514
BT		 3298 goto proc_cache_fail;
3299#endif
2bb8b26c
BT		 3300 return 0;
3301
f6bb4514
BT		 3302#ifdef CONFIG_PROC_FS
3303proc_cache_fail:
ece31ffd 3304 remove_proc_entry("ip_mr_vif", net->proc_net);
f6bb4514 3305proc_vif_fail:
f0ad0860 3306 ipmr_rules_exit(net);
f6bb4514 3307#endif
4d65b948
YG		 3308ipmr_rules_fail:
3309 ipmr_notifier_exit(net);
3310ipmr_notifier_fail:
cf958ae3
BT		 3311 return err;
3312}
3313
3314static void __net_exit ipmr_net_exit(struct net *net)
3315{
f6bb4514 3316#ifdef CONFIG_PROC_FS
ece31ffd
G		 3317 remove_proc_entry("ip_mr_cache", net->proc_net);
3318 remove_proc_entry("ip_mr_vif", net->proc_net);
f6bb4514 3319#endif
4d65b948 3320 ipmr_notifier_exit(net);
f0ad0860 3321 ipmr_rules_exit(net);
cf958ae3
BT		 3322}
3323
3324static struct pernet_operations ipmr_net_ops = {
3325 .init = ipmr_net_init,
3326 .exit = ipmr_net_exit,
f84c6821 3327 .async = true,
cf958ae3 3328};
e905a9ed 3329
03d2f897 3330int __init ip_mr_init(void)
1da177e4 3331{
03d2f897
WC		 3332 int err;
3333
1da177e4
LT		 3334 mrt_cachep = kmem_cache_create("ip_mrt_cache",
3335 sizeof(struct mfc_cache),
a8c9486b 3336 0, SLAB_HWCACHE_ALIGN | SLAB_PANIC,
20c2df83 3337 NULL);
03d2f897 3338
cf958ae3
BT		 3339 err = register_pernet_subsys(&ipmr_net_ops);
3340 if (err)
3341 goto reg_pernet_fail;
3342
03d2f897
WC		 3343 err = register_netdevice_notifier(&ip_mr_notifier);
3344 if (err)
3345 goto reg_notif_fail;
403dbb97
TG		 3346#ifdef CONFIG_IP_PIMSM_V2
3347 if (inet_add_protocol(&pim_protocol, IPPROTO_PIM) < 0) {
058bd4d2 3348 pr_err("%s: can't add PIM protocol\n", __func__);
403dbb97
TG		 3349 err = -EAGAIN;
3350 goto add_proto_fail;
3351 }
3352#endif
c7ac8679 3353 rtnl_register(RTNL_FAMILY_IPMR, RTM_GETROUTE,
b97bac64 3354 ipmr_rtm_getroute, ipmr_rtm_dumproute, 0);
ccbb0aa6 3355 rtnl_register(RTNL_FAMILY_IPMR, RTM_NEWROUTE,
b97bac64 3356 ipmr_rtm_route, NULL, 0);
ccbb0aa6 3357 rtnl_register(RTNL_FAMILY_IPMR, RTM_DELROUTE,
b97bac64 3358 ipmr_rtm_route, NULL, 0);
772c344d
NA		 3359
3360 rtnl_register(RTNL_FAMILY_IPMR, RTM_GETLINK,
b97bac64 3361 NULL, ipmr_rtm_dumplink, 0);
03d2f897 3362 return 0;
f6bb4514 3363
403dbb97
TG		 3364#ifdef CONFIG_IP_PIMSM_V2
3365add_proto_fail:
3366 unregister_netdevice_notifier(&ip_mr_notifier);
3367#endif
c3e38896 3368reg_notif_fail:
cf958ae3
BT		 3369 unregister_pernet_subsys(&ipmr_net_ops);
3370reg_pernet_fail:
c3e38896 3371 kmem_cache_destroy(mrt_cachep);
03d2f897 3372 return err;
1da177e4 3373}
Ubuntu Jammy Linux kernel mirror