Commit | Line | Data |
---|---|---|
b2441318 | 1 | /* SPDX-License-Identifier: GPL-2.0 */ |
5f256bec EB |
2 | /* |
3 | * Operations on the network namespace | |
4 | */ | |
5 | #ifndef __NET_NET_NAMESPACE_H | |
6 | #define __NET_NET_NAMESPACE_H | |
7 | ||
60063497 | 8 | #include <linux/atomic.h> |
c122e14d | 9 | #include <linux/refcount.h> |
5f256bec EB |
10 | #include <linux/workqueue.h> |
11 | #include <linux/list.h> | |
bee95250 | 12 | #include <linux/sysctl.h> |
fbdeaed4 | 13 | #include <linux/uidgid.h> |
5f256bec | 14 | |
6a662719 | 15 | #include <net/flow.h> |
8efa6e93 | 16 | #include <net/netns/core.h> |
852566f5 | 17 | #include <net/netns/mib.h> |
a0a53c8b | 18 | #include <net/netns/unix.h> |
2aaef4e4 | 19 | #include <net/netns/packet.h> |
8afd351c | 20 | #include <net/netns/ipv4.h> |
b0f159db | 21 | #include <net/netns/ipv6.h> |
633fc86f | 22 | #include <net/netns/ieee802154_6lowpan.h> |
4db67e80 | 23 | #include <net/netns/sctp.h> |
67019cc9 | 24 | #include <net/netns/dccp.h> |
f3c1a44a | 25 | #include <net/netns/netfilter.h> |
8d870052 | 26 | #include <net/netns/x_tables.h> |
dfdb8d79 AD |
27 | #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE) |
28 | #include <net/netns/conntrack.h> | |
29 | #endif | |
99633ab2 | 30 | #include <net/netns/nftables.h> |
d62ddc21 | 31 | #include <net/netns/xfrm.h> |
0189197f | 32 | #include <net/netns/mpls.h> |
8e8cda6d | 33 | #include <net/netns/can.h> |
1d0dc069 | 34 | #include <net/netns/xdp.h> |
435d5f4b | 35 | #include <linux/ns_common.h> |
04c52dec PNA |
36 | #include <linux/idr.h> |
37 | #include <linux/skbuff.h> | |
a0a53c8b | 38 | |
038e7332 | 39 | struct user_namespace; |
457c4cbc | 40 | struct proc_dir_entry; |
2774c7ab | 41 | struct net_device; |
97c53cac | 42 | struct sock; |
1597fbc0 | 43 | struct ctl_table_header; |
dec827d1 | 44 | struct net_generic; |
94e5e308 | 45 | struct uevent_sock; |
2553d064 | 46 | struct netns_ipvs; |
d58e468b | 47 | struct bpf_prog; |
1597fbc0 | 48 | |
7c28bd0b ED |
49 | |
50 | #define NETDEV_HASHBITS 8 | |
51 | #define NETDEV_HASHENTRIES (1 << NETDEV_HASHBITS) | |
52 | ||
5f256bec | 53 | struct net { |
c122e14d | 54 | refcount_t passive; /* To decided when the network |
a685e089 AV |
55 | * namespace should be freed. |
56 | */ | |
273c28bc | 57 | refcount_t count; /* To decided when the network |
a685e089 | 58 | * namespace should be shut down. |
5f256bec | 59 | */ |
8e602ce2 ED |
60 | spinlock_t rules_mod_lock; |
61 | ||
355b9855 | 62 | u32 hash_mix; |
33cf7c90 ED |
63 | atomic64_t cookie_gen; |
64 | ||
5f256bec | 65 | struct list_head list; /* list of network namespaces */ |
19efbd93 | 66 | struct list_head exit_list; /* To linked to call pernet exit |
4420bf21 KT |
67 | * methods on dead net ( |
68 | * pernet_ops_rwsem read locked), | |
69 | * or to unregister pernet ops | |
70 | * (pernet_ops_rwsem write locked). | |
19efbd93 | 71 | */ |
65b7b5b9 KT |
72 | struct llist_node cleanup_list; /* namespaces on death row */ |
73 | ||
038e7332 | 74 | struct user_namespace *user_ns; /* Owning user namespace */ |
70328660 | 75 | struct ucounts *ucounts; |
de133464 | 76 | spinlock_t nsid_lock; |
0c7aecd4 | 77 | struct idr netns_ids; |
038e7332 | 78 | |
435d5f4b | 79 | struct ns_common ns; |
98f842e6 | 80 | |
457c4cbc EB |
81 | struct proc_dir_entry *proc_net; |
82 | struct proc_dir_entry *proc_net_stat; | |
881d966b | 83 | |
73455092 AV |
84 | #ifdef CONFIG_SYSCTL |
85 | struct ctl_table_set sysctls; | |
86 | #endif | |
95bdfccb | 87 | |
8e602ce2 ED |
88 | struct sock *rtnl; /* rtnetlink socket */ |
89 | struct sock *genl_sock; | |
2774c7ab | 90 | |
94e5e308 CB |
91 | struct uevent_sock *uevent_sock; /* uevent socket */ |
92 | ||
881d966b EB |
93 | struct list_head dev_base_head; |
94 | struct hlist_head *dev_name_head; | |
95 | struct hlist_head *dev_index_head; | |
4e985ada | 96 | unsigned int dev_base_seq; /* protected by rtnl_mutex */ |
aa79e66e | 97 | int ifindex; |
50624c93 | 98 | unsigned int dev_unreg_count; |
97c53cac | 99 | |
5fd30ee7 DL |
100 | /* core fib_rules */ |
101 | struct list_head rules_ops; | |
5fd30ee7 | 102 | |
4420bf21 KT |
103 | struct list_head fib_notifier_ops; /* Populated by |
104 | * register_pernet_subsys() | |
105 | */ | |
8e602ce2 | 106 | struct net_device *loopback_dev; /* The loopback */ |
8efa6e93 | 107 | struct netns_core core; |
852566f5 | 108 | struct netns_mib mib; |
2aaef4e4 | 109 | struct netns_packet packet; |
a0a53c8b | 110 | struct netns_unix unx; |
8afd351c | 111 | struct netns_ipv4 ipv4; |
dfd56b8b | 112 | #if IS_ENABLED(CONFIG_IPV6) |
b0f159db DL |
113 | struct netns_ipv6 ipv6; |
114 | #endif | |
633fc86f AA |
115 | #if IS_ENABLED(CONFIG_IEEE802154_6LOWPAN) |
116 | struct netns_ieee802154_lowpan ieee802154_lowpan; | |
117 | #endif | |
4db67e80 EB |
118 | #if defined(CONFIG_IP_SCTP) || defined(CONFIG_IP_SCTP_MODULE) |
119 | struct netns_sctp sctp; | |
120 | #endif | |
67019cc9 PE |
121 | #if defined(CONFIG_IP_DCCP) || defined(CONFIG_IP_DCCP_MODULE) |
122 | struct netns_dccp dccp; | |
123 | #endif | |
8d870052 | 124 | #ifdef CONFIG_NETFILTER |
f3c1a44a | 125 | struct netns_nf nf; |
8d870052 | 126 | struct netns_xt xt; |
dfdb8d79 AD |
127 | #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE) |
128 | struct netns_ct ct; | |
c038a767 | 129 | #endif |
99633ab2 PNA |
130 | #if defined(CONFIG_NF_TABLES) || defined(CONFIG_NF_TABLES_MODULE) |
131 | struct netns_nftables nft; | |
132 | #endif | |
c038a767 AW |
133 | #if IS_ENABLED(CONFIG_NF_DEFRAG_IPV6) |
134 | struct netns_nf_frag nf_frag; | |
9ce7bc03 | 135 | struct ctl_table_header *nf_frag_frags_hdr; |
dfdb8d79 | 136 | #endif |
cd8c20b6 AD |
137 | struct sock *nfnl; |
138 | struct sock *nfnl_stash; | |
3499abb2 AS |
139 | #if IS_ENABLED(CONFIG_NETFILTER_NETLINK_ACCT) |
140 | struct list_head nfnl_acct_list; | |
141 | #endif | |
19576c94 PN |
142 | #if IS_ENABLED(CONFIG_NF_CT_NETLINK_TIMEOUT) |
143 | struct list_head nfct_timeout_list; | |
144 | #endif | |
d62ddc21 | 145 | #endif |
3d23e349 | 146 | #ifdef CONFIG_WEXT_CORE |
b333b3d2 | 147 | struct sk_buff_head wext_nlevents; |
8d870052 | 148 | #endif |
1c87733d | 149 | struct net_generic __rcu *gen; |
8e602ce2 | 150 | |
d58e468b PP |
151 | struct bpf_prog __rcu *flow_dissector_prog; |
152 | ||
8e602ce2 ED |
153 | /* Note : following structs are cache line aligned */ |
154 | #ifdef CONFIG_XFRM | |
155 | struct netns_xfrm xfrm; | |
156 | #endif | |
8b4d14d8 | 157 | #if IS_ENABLED(CONFIG_IP_VS) |
61b1ab45 | 158 | struct netns_ipvs *ipvs; |
0189197f EB |
159 | #endif |
160 | #if IS_ENABLED(CONFIG_MPLS) | |
161 | struct netns_mpls mpls; | |
8e8cda6d MK |
162 | #endif |
163 | #if IS_ENABLED(CONFIG_CAN) | |
164 | struct netns_can can; | |
1d0dc069 BT |
165 | #endif |
166 | #ifdef CONFIG_XDP_SOCKETS | |
167 | struct netns_xdp xdp; | |
8b4d14d8 | 168 | #endif |
51d7cccf | 169 | struct sock *diag_nlsk; |
5aad1de5 | 170 | atomic_t fnhe_genid; |
3859a271 | 171 | } __randomize_layout; |
5f256bec | 172 | |
c0f39322 DL |
173 | #include <linux/seq_file_net.h> |
174 | ||
4fabcd71 | 175 | /* Init's network namespace */ |
5f256bec | 176 | extern struct net init_net; |
a4aa834a | 177 | |
d727abcb | 178 | #ifdef CONFIG_NET_NS |
e67e16ea JP |
179 | struct net *copy_net_ns(unsigned long flags, struct user_namespace *user_ns, |
180 | struct net *old_net); | |
225c0a01 | 181 | |
fbdeaed4 TH |
182 | void net_ns_get_ownership(const struct net *net, kuid_t *uid, kgid_t *gid); |
183 | ||
7866cc57 | 184 | void net_ns_barrier(void); |
d727abcb EB |
185 | #else /* CONFIG_NET_NS */ |
186 | #include <linux/sched.h> | |
187 | #include <linux/nsproxy.h> | |
038e7332 EB |
188 | static inline struct net *copy_net_ns(unsigned long flags, |
189 | struct user_namespace *user_ns, struct net *old_net) | |
9dd776b6 | 190 | { |
d727abcb EB |
191 | if (flags & CLONE_NEWNET) |
192 | return ERR_PTR(-EINVAL); | |
193 | return old_net; | |
9dd776b6 | 194 | } |
7866cc57 | 195 | |
fbdeaed4 TH |
196 | static inline void net_ns_get_ownership(const struct net *net, |
197 | kuid_t *uid, kgid_t *gid) | |
198 | { | |
199 | *uid = GLOBAL_ROOT_UID; | |
200 | *gid = GLOBAL_ROOT_GID; | |
201 | } | |
202 | ||
7866cc57 | 203 | static inline void net_ns_barrier(void) {} |
d727abcb | 204 | #endif /* CONFIG_NET_NS */ |
225c0a01 DL |
205 | |
206 | ||
207 | extern struct list_head net_namespace_list; | |
9dd776b6 | 208 | |
e67e16ea | 209 | struct net *get_net_ns_by_pid(pid_t pid); |
0f5258cd | 210 | struct net *get_net_ns_by_fd(int fd); |
30ffee84 | 211 | |
535d3ae9 RK |
212 | #ifdef CONFIG_SYSCTL |
213 | void ipx_register_sysctl(void); | |
214 | void ipx_unregister_sysctl(void); | |
215 | #else | |
216 | #define ipx_register_sysctl() | |
217 | #define ipx_unregister_sysctl() | |
218 | #endif | |
219 | ||
d4655795 | 220 | #ifdef CONFIG_NET_NS |
e67e16ea | 221 | void __put_net(struct net *net); |
5f256bec EB |
222 | |
223 | static inline struct net *get_net(struct net *net) | |
224 | { | |
273c28bc | 225 | refcount_inc(&net->count); |
5f256bec EB |
226 | return net; |
227 | } | |
228 | ||
077130c0 EB |
229 | static inline struct net *maybe_get_net(struct net *net) |
230 | { | |
231 | /* Used when we know struct net exists but we | |
232 | * aren't guaranteed a previous reference count | |
233 | * exists. If the reference count is zero this | |
234 | * function fails and returns NULL. | |
235 | */ | |
273c28bc | 236 | if (!refcount_inc_not_zero(&net->count)) |
077130c0 EB |
237 | net = NULL; |
238 | return net; | |
239 | } | |
240 | ||
5f256bec EB |
241 | static inline void put_net(struct net *net) |
242 | { | |
273c28bc | 243 | if (refcount_dec_and_test(&net->count)) |
5f256bec EB |
244 | __put_net(net); |
245 | } | |
246 | ||
878628fb YH |
247 | static inline |
248 | int net_eq(const struct net *net1, const struct net *net2) | |
249 | { | |
250 | return net1 == net2; | |
251 | } | |
a685e089 | 252 | |
4ee806d5 DS |
253 | static inline int check_net(const struct net *net) |
254 | { | |
3e3ab9cc | 255 | return refcount_read(&net->count) != 0; |
4ee806d5 DS |
256 | } |
257 | ||
e67e16ea | 258 | void net_drop_ns(void *); |
a685e089 | 259 | |
d4655795 | 260 | #else |
b9f75f45 | 261 | |
d4655795 PE |
262 | static inline struct net *get_net(struct net *net) |
263 | { | |
264 | return net; | |
265 | } | |
266 | ||
267 | static inline void put_net(struct net *net) | |
268 | { | |
269 | } | |
270 | ||
5d1e4468 DL |
271 | static inline struct net *maybe_get_net(struct net *net) |
272 | { | |
273 | return net; | |
274 | } | |
275 | ||
276 | static inline | |
277 | int net_eq(const struct net *net1, const struct net *net2) | |
278 | { | |
279 | return 1; | |
280 | } | |
a685e089 | 281 | |
4ee806d5 DS |
282 | static inline int check_net(const struct net *net) |
283 | { | |
284 | return 1; | |
285 | } | |
286 | ||
a685e089 | 287 | #define net_drop_ns NULL |
5d1e4468 DL |
288 | #endif |
289 | ||
290 | ||
0c5c9fb5 | 291 | typedef struct { |
8f424b5f | 292 | #ifdef CONFIG_NET_NS |
0c5c9fb5 EB |
293 | struct net *net; |
294 | #endif | |
295 | } possible_net_t; | |
8f424b5f | 296 | |
0c5c9fb5 | 297 | static inline void write_pnet(possible_net_t *pnet, struct net *net) |
8f424b5f | 298 | { |
0c5c9fb5 EB |
299 | #ifdef CONFIG_NET_NS |
300 | pnet->net = net; | |
301 | #endif | |
8f424b5f ED |
302 | } |
303 | ||
0c5c9fb5 | 304 | static inline struct net *read_pnet(const possible_net_t *pnet) |
8f424b5f | 305 | { |
0c5c9fb5 EB |
306 | #ifdef CONFIG_NET_NS |
307 | return pnet->net; | |
8f424b5f | 308 | #else |
0c5c9fb5 | 309 | return &init_net; |
8f424b5f | 310 | #endif |
0c5c9fb5 | 311 | } |
5d1e4468 | 312 | |
f0b07bb1 | 313 | /* Protected by net_rwsem */ |
5f256bec EB |
314 | #define for_each_net(VAR) \ |
315 | list_for_each_entry(VAR, &net_namespace_list, list) | |
316 | ||
11a28d37 JB |
317 | #define for_each_net_rcu(VAR) \ |
318 | list_for_each_entry_rcu(VAR, &net_namespace_list, list) | |
319 | ||
4665079c PE |
320 | #ifdef CONFIG_NET_NS |
321 | #define __net_init | |
322 | #define __net_exit | |
022cbae6 | 323 | #define __net_initdata |
04a6f82c | 324 | #define __net_initconst |
4665079c PE |
325 | #else |
326 | #define __net_init __init | |
bd721ea7 | 327 | #define __net_exit __ref |
022cbae6 | 328 | #define __net_initdata __initdata |
04a6f82c | 329 | #define __net_initconst __initconst |
4665079c | 330 | #endif |
5f256bec | 331 | |
7a0877d4 | 332 | int peernet2id_alloc(struct net *net, struct net *peer); |
59324cf3 ND |
333 | int peernet2id(struct net *net, struct net *peer); |
334 | bool peernet_has_id(struct net *net, struct net *peer); | |
0c7aecd4 ND |
335 | struct net *get_net_ns_by_id(struct net *net, int id); |
336 | ||
5f256bec EB |
337 | struct pernet_operations { |
338 | struct list_head list; | |
6056415d KT |
339 | /* |
340 | * Below methods are called without any exclusive locks. | |
341 | * More than one net may be constructed and destructed | |
342 | * in parallel on several cpus. Every pernet_operations | |
343 | * have to keep in mind all other pernet_operations and | |
344 | * to introduce a locking, if they share common resources. | |
345 | * | |
8518e9bb KT |
346 | * The only time they are called with exclusive lock is |
347 | * from register_pernet_subsys(), unregister_pernet_subsys() | |
348 | * register_pernet_device() and unregister_pernet_device(). | |
349 | * | |
6056415d KT |
350 | * Exit methods using blocking RCU primitives, such as |
351 | * synchronize_rcu(), should be implemented via exit_batch. | |
352 | * Then, destruction of a group of net requires single | |
353 | * synchronize_rcu() related to these pernet_operations, | |
354 | * instead of separate synchronize_rcu() for every net. | |
355 | * Please, avoid synchronize_rcu() at all, where it's possible. | |
356 | */ | |
5f256bec EB |
357 | int (*init)(struct net *net); |
358 | void (*exit)(struct net *net); | |
72ad937a | 359 | void (*exit_batch)(struct list_head *net_exit_list); |
c7d03a00 | 360 | unsigned int *id; |
f875bae0 | 361 | size_t size; |
5f256bec EB |
362 | }; |
363 | ||
17edde52 EB |
364 | /* |
365 | * Use these carefully. If you implement a network device and it | |
366 | * needs per network namespace operations use device pernet operations, | |
367 | * otherwise use pernet subsys operations. | |
368 | * | |
4edf547b JB |
369 | * Network interfaces need to be removed from a dying netns _before_ |
370 | * subsys notifiers can be called, as most of the network code cleanup | |
371 | * (which is done from subsys notifiers) runs with the assumption that | |
372 | * dev_remove_pack has been called so no new packets will arrive during | |
373 | * and after the cleanup functions have been called. dev_remove_pack | |
374 | * is not per namespace so instead the guarantee of no more packets | |
375 | * arriving in a network namespace is provided by ensuring that all | |
376 | * network devices and all sockets have left the network namespace | |
377 | * before the cleanup methods are called. | |
17edde52 EB |
378 | * |
379 | * For the longest time the ipv4 icmp code was registered as a pernet | |
380 | * device which caused kernel oops, and panics during network | |
381 | * namespace cleanup. So please don't get this wrong. | |
382 | */ | |
e67e16ea JP |
383 | int register_pernet_subsys(struct pernet_operations *); |
384 | void unregister_pernet_subsys(struct pernet_operations *); | |
385 | int register_pernet_device(struct pernet_operations *); | |
386 | void unregister_pernet_device(struct pernet_operations *); | |
f875bae0 | 387 | |
95bdfccb EB |
388 | struct ctl_table; |
389 | struct ctl_table_header; | |
d62c612e | 390 | |
2ca794e5 | 391 | #ifdef CONFIG_SYSCTL |
e67e16ea JP |
392 | int net_sysctl_init(void); |
393 | struct ctl_table_header *register_net_sysctl(struct net *net, const char *path, | |
394 | struct ctl_table *table); | |
395 | void unregister_net_sysctl_table(struct ctl_table_header *header); | |
48c74958 EB |
396 | #else |
397 | static inline int net_sysctl_init(void) { return 0; } | |
398 | static inline struct ctl_table_header *register_net_sysctl(struct net *net, | |
399 | const char *path, struct ctl_table *table) | |
400 | { | |
401 | return NULL; | |
402 | } | |
403 | static inline void unregister_net_sysctl_table(struct ctl_table_header *header) | |
404 | { | |
405 | } | |
406 | #endif | |
407 | ||
ca4c3fc2 | 408 | static inline int rt_genid_ipv4(struct net *net) |
b42664f8 | 409 | { |
ca4c3fc2 | 410 | return atomic_read(&net->ipv4.rt_genid); |
b42664f8 ND |
411 | } |
412 | ||
ca4c3fc2 | 413 | static inline void rt_genid_bump_ipv4(struct net *net) |
b42664f8 | 414 | { |
ca4c3fc2 | 415 | atomic_inc(&net->ipv4.rt_genid); |
416 | } | |
417 | ||
705f1c86 | 418 | extern void (*__fib6_flush_trees)(struct net *net); |
ca4c3fc2 | 419 | static inline void rt_genid_bump_ipv6(struct net *net) |
420 | { | |
705f1c86 HFS |
421 | if (__fib6_flush_trees) |
422 | __fib6_flush_trees(net); | |
ca4c3fc2 | 423 | } |
ca4c3fc2 | 424 | |
599018a7 LR |
425 | #if IS_ENABLED(CONFIG_IEEE802154_6LOWPAN) |
426 | static inline struct netns_ieee802154_lowpan * | |
427 | net_ieee802154_lowpan(struct net *net) | |
428 | { | |
429 | return &net->ieee802154_lowpan; | |
430 | } | |
599018a7 LR |
431 | #endif |
432 | ||
ca4c3fc2 | 433 | /* For callers who don't really care about whether it's IPv4 or IPv6 */ |
434 | static inline void rt_genid_bump_all(struct net *net) | |
435 | { | |
436 | rt_genid_bump_ipv4(net); | |
437 | rt_genid_bump_ipv6(net); | |
b42664f8 | 438 | } |
95bdfccb | 439 | |
5aad1de5 TT |
440 | static inline int fnhe_genid(struct net *net) |
441 | { | |
442 | return atomic_read(&net->fnhe_genid); | |
443 | } | |
444 | ||
445 | static inline void fnhe_genid_bump(struct net *net) | |
446 | { | |
447 | atomic_inc(&net->fnhe_genid); | |
448 | } | |
449 | ||
5f256bec | 450 | #endif /* __NET_NET_NAMESPACE_H */ |