1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef __NET_IP_TUNNELS_H
3 #define __NET_IP_TUNNELS_H 1
5 #include <linux/if_tunnel.h>
6 #include <linux/netdevice.h>
7 #include <linux/skbuff.h>
8 #include <linux/socket.h>
9 #include <linux/types.h>
10 #include <linux/u64_stats_sync.h>
11 #include <linux/bitops.h>
13 #include <net/dsfield.h>
14 #include <net/gro_cells.h>
15 #include <net/inet_ecn.h>
16 #include <net/netns/generic.h>
17 #include <net/rtnetlink.h>
18 #include <net/lwtunnel.h>
19 #include <net/dst_cache.h>
21 #if IS_ENABLED(CONFIG_IPV6)
23 #include <net/ip6_fib.h>
24 #include <net/ip6_route.h>
27 /* Keep error state on tunnel for 30 sec */
28 #define IPTUNNEL_ERR_TIMEO (30*HZ)
30 /* Used to memset ip_tunnel padding. */
31 #define IP_TUNNEL_KEY_SIZE offsetofend(struct ip_tunnel_key, tp_dst)
33 /* Used to memset ipv4 address padding. */
34 #define IP_TUNNEL_KEY_IPV4_PAD offsetofend(struct ip_tunnel_key, u.ipv4.dst)
35 #define IP_TUNNEL_KEY_IPV4_PAD_LEN \
36 (FIELD_SIZEOF(struct ip_tunnel_key, u) - \
37 FIELD_SIZEOF(struct ip_tunnel_key, u.ipv4))
39 struct ip_tunnel_key
{
52 u8 tos
; /* TOS for IPv4, TC for IPv6 */
53 u8 ttl
; /* TTL for IPv4, HL for IPv6 */
54 __be32 label
; /* Flow Label for IPv6 */
59 /* Flags for ip_tunnel_info mode. */
60 #define IP_TUNNEL_INFO_TX 0x01 /* represents tx tunnel parameters */
61 #define IP_TUNNEL_INFO_IPV6 0x02 /* key contains IPv6 addresses */
62 #define IP_TUNNEL_INFO_BRIDGE 0x04 /* represents a bridged tunnel id */
64 /* Maximum tunnel options length. */
65 #define IP_TUNNEL_OPTS_MAX \
66 GENMASK((FIELD_SIZEOF(struct ip_tunnel_info, \
67 options_len) * BITS_PER_BYTE) - 1, 0)
69 struct ip_tunnel_info
{
70 struct ip_tunnel_key key
;
71 #ifdef CONFIG_DST_CACHE
72 struct dst_cache dst_cache
;
78 /* 6rd prefix/relay information */
79 #ifdef CONFIG_IPV6_SIT_6RD
80 struct ip_tunnel_6rd_parm
{
81 struct in6_addr prefix
;
88 struct ip_tunnel_encap
{
95 struct ip_tunnel_prl_entry
{
96 struct ip_tunnel_prl_entry __rcu
*next
;
99 struct rcu_head rcu_head
;
103 /* A fan overlay /8 (250.0.0.0/8, for example) maps to exactly one /16
104 * underlay (10.88.0.0/16, for example). Multiple local addresses within
105 * the /16 may be used, but a particular overlay may not span
106 * multiple underlay subnets.
108 * We store one underlay, indexed by the overlay's high order octet.
110 #define FAN_OVERLAY_CNT 256
118 struct list_head list
;
122 struct ip_tunnel_fan
{
123 struct list_head fan_maps
;
127 struct ip_tunnel __rcu
*next
;
128 struct hlist_node hash_node
;
129 struct net_device
*dev
;
130 struct net
*net
; /* netns for packet i/o */
132 unsigned long err_time
; /* Time when the last ICMP error
134 int err_count
; /* Number of arrived ICMP errors */
136 /* These four fields used only by GRE */
137 u32 i_seqno
; /* The last seen seqno */
138 u32 o_seqno
; /* The last output seqno */
139 int tun_hlen
; /* Precalculated header length */
141 /* This field used only by ERSPAN */
142 u32 index
; /* ERSPAN type II index */
144 struct dst_cache dst_cache
;
146 struct ip_tunnel_parm parms
;
149 int encap_hlen
; /* Encap header length (FOU,GUE) */
150 int hlen
; /* tun_hlen + encap_hlen */
151 struct ip_tunnel_encap encap
;
154 #ifdef CONFIG_IPV6_SIT_6RD
155 struct ip_tunnel_6rd_parm ip6rd
;
157 struct ip_tunnel_prl_entry __rcu
*prl
; /* potential router list */
158 unsigned int prl_count
; /* # of entries in PRL */
159 struct ip_tunnel_fan fan
;
160 unsigned int ip_tnl_net_id
;
161 struct gro_cells gro_cells
;
167 #define TUNNEL_CSUM __cpu_to_be16(0x01)
168 #define TUNNEL_ROUTING __cpu_to_be16(0x02)
169 #define TUNNEL_KEY __cpu_to_be16(0x04)
170 #define TUNNEL_SEQ __cpu_to_be16(0x08)
171 #define TUNNEL_STRICT __cpu_to_be16(0x10)
172 #define TUNNEL_REC __cpu_to_be16(0x20)
173 #define TUNNEL_VERSION __cpu_to_be16(0x40)
174 #define TUNNEL_NO_KEY __cpu_to_be16(0x80)
175 #define TUNNEL_DONT_FRAGMENT __cpu_to_be16(0x0100)
176 #define TUNNEL_OAM __cpu_to_be16(0x0200)
177 #define TUNNEL_CRIT_OPT __cpu_to_be16(0x0400)
178 #define TUNNEL_GENEVE_OPT __cpu_to_be16(0x0800)
179 #define TUNNEL_VXLAN_OPT __cpu_to_be16(0x1000)
180 #define TUNNEL_NOCACHE __cpu_to_be16(0x2000)
181 #define TUNNEL_ERSPAN_OPT __cpu_to_be16(0x4000)
183 #define TUNNEL_OPTIONS_PRESENT \
184 (TUNNEL_GENEVE_OPT | TUNNEL_VXLAN_OPT | TUNNEL_ERSPAN_OPT)
186 static inline int fan_has_map(const struct ip_tunnel_fan
*fan
)
188 return !list_empty(&fan
->fan_maps
);
191 struct tnl_ptk_info
{
199 #define PACKET_RCVD 0
200 #define PACKET_REJECT 1
201 #define PACKET_NEXT 2
203 #define IP_TNL_HASH_BITS 7
204 #define IP_TNL_HASH_SIZE (1 << IP_TNL_HASH_BITS)
206 struct ip_tunnel_net
{
207 struct net_device
*fb_tunnel_dev
;
208 struct hlist_head tunnels
[IP_TNL_HASH_SIZE
];
209 struct ip_tunnel __rcu
*collect_md_tun
;
212 static inline void ip_tunnel_key_init(struct ip_tunnel_key
*key
,
213 __be32 saddr
, __be32 daddr
,
214 u8 tos
, u8 ttl
, __be32 label
,
215 __be16 tp_src
, __be16 tp_dst
,
216 __be64 tun_id
, __be16 tun_flags
)
218 key
->tun_id
= tun_id
;
219 key
->u
.ipv4
.src
= saddr
;
220 key
->u
.ipv4
.dst
= daddr
;
221 memset((unsigned char *)key
+ IP_TUNNEL_KEY_IPV4_PAD
,
222 0, IP_TUNNEL_KEY_IPV4_PAD_LEN
);
226 key
->tun_flags
= tun_flags
;
228 /* For the tunnel types on the top of IPsec, the tp_src and tp_dst of
229 * the upper tunnel are used.
230 * E.g: GRE over IPSEC, the tp_src and tp_port are zero.
232 key
->tp_src
= tp_src
;
233 key
->tp_dst
= tp_dst
;
235 /* Clear struct padding. */
236 if (sizeof(*key
) != IP_TUNNEL_KEY_SIZE
)
237 memset((unsigned char *)key
+ IP_TUNNEL_KEY_SIZE
,
238 0, sizeof(*key
) - IP_TUNNEL_KEY_SIZE
);
242 ip_tunnel_dst_cache_usable(const struct sk_buff
*skb
,
243 const struct ip_tunnel_info
*info
)
249 if (info
->key
.tun_flags
& TUNNEL_NOCACHE
)
255 static inline unsigned short ip_tunnel_info_af(const struct ip_tunnel_info
258 return tun_info
->mode
& IP_TUNNEL_INFO_IPV6
? AF_INET6
: AF_INET
;
261 static inline __be64
key32_to_tunnel_id(__be32 key
)
264 return (__force __be64
)key
;
266 return (__force __be64
)((__force u64
)key
<< 32);
270 /* Returns the least-significant 32 bits of a __be64. */
271 static inline __be32
tunnel_id_to_key32(__be64 tun_id
)
274 return (__force __be32
)tun_id
;
276 return (__force __be32
)((__force u64
)tun_id
>> 32);
282 int ip_tunnel_init(struct net_device
*dev
);
283 void ip_tunnel_uninit(struct net_device
*dev
);
284 void ip_tunnel_dellink(struct net_device
*dev
, struct list_head
*head
);
285 struct net
*ip_tunnel_get_link_net(const struct net_device
*dev
);
286 int ip_tunnel_get_iflink(const struct net_device
*dev
);
287 int ip_tunnel_init_net(struct net
*net
, unsigned int ip_tnl_net_id
,
288 struct rtnl_link_ops
*ops
, char *devname
);
290 void ip_tunnel_delete_nets(struct list_head
*list_net
, unsigned int id
,
291 struct rtnl_link_ops
*ops
);
293 void ip_tunnel_xmit(struct sk_buff
*skb
, struct net_device
*dev
,
294 const struct iphdr
*tnl_params
, const u8 protocol
);
295 void ip_md_tunnel_xmit(struct sk_buff
*skb
, struct net_device
*dev
,
297 int ip_tunnel_ioctl(struct net_device
*dev
, struct ip_tunnel_parm
*p
, int cmd
);
298 int __ip_tunnel_change_mtu(struct net_device
*dev
, int new_mtu
, bool strict
);
299 int ip_tunnel_change_mtu(struct net_device
*dev
, int new_mtu
);
301 void ip_tunnel_get_stats64(struct net_device
*dev
,
302 struct rtnl_link_stats64
*tot
);
303 struct ip_tunnel
*ip_tunnel_lookup(struct ip_tunnel_net
*itn
,
304 int link
, __be16 flags
,
305 __be32 remote
, __be32 local
,
308 int ip_tunnel_rcv(struct ip_tunnel
*tunnel
, struct sk_buff
*skb
,
309 const struct tnl_ptk_info
*tpi
, struct metadata_dst
*tun_dst
,
311 int ip_tunnel_changelink(struct net_device
*dev
, struct nlattr
*tb
[],
312 struct ip_tunnel_parm
*p
, __u32 fwmark
);
313 int ip_tunnel_newlink(struct net_device
*dev
, struct nlattr
*tb
[],
314 struct ip_tunnel_parm
*p
, __u32 fwmark
);
315 void ip_tunnel_setup(struct net_device
*dev
, unsigned int net_id
);
317 struct ip_tunnel_encap_ops
{
318 size_t (*encap_hlen
)(struct ip_tunnel_encap
*e
);
319 int (*build_header
)(struct sk_buff
*skb
, struct ip_tunnel_encap
*e
,
320 u8
*protocol
, struct flowi4
*fl4
);
323 #define MAX_IPTUN_ENCAP_OPS 8
325 extern const struct ip_tunnel_encap_ops __rcu
*
326 iptun_encaps
[MAX_IPTUN_ENCAP_OPS
];
328 int ip_tunnel_encap_add_ops(const struct ip_tunnel_encap_ops
*op
,
330 int ip_tunnel_encap_del_ops(const struct ip_tunnel_encap_ops
*op
,
333 int ip_tunnel_encap_setup(struct ip_tunnel
*t
,
334 struct ip_tunnel_encap
*ipencap
);
336 static inline int ip_encap_hlen(struct ip_tunnel_encap
*e
)
338 const struct ip_tunnel_encap_ops
*ops
;
341 if (e
->type
== TUNNEL_ENCAP_NONE
)
344 if (e
->type
>= MAX_IPTUN_ENCAP_OPS
)
348 ops
= rcu_dereference(iptun_encaps
[e
->type
]);
349 if (likely(ops
&& ops
->encap_hlen
))
350 hlen
= ops
->encap_hlen(e
);
356 static inline int ip_tunnel_encap(struct sk_buff
*skb
, struct ip_tunnel
*t
,
357 u8
*protocol
, struct flowi4
*fl4
)
359 const struct ip_tunnel_encap_ops
*ops
;
362 if (t
->encap
.type
== TUNNEL_ENCAP_NONE
)
365 if (t
->encap
.type
>= MAX_IPTUN_ENCAP_OPS
)
369 ops
= rcu_dereference(iptun_encaps
[t
->encap
.type
]);
370 if (likely(ops
&& ops
->build_header
))
371 ret
= ops
->build_header(skb
, &t
->encap
, protocol
, fl4
);
377 /* Extract dsfield from inner protocol */
378 static inline u8
ip_tunnel_get_dsfield(const struct iphdr
*iph
,
379 const struct sk_buff
*skb
)
381 if (skb
->protocol
== htons(ETH_P_IP
))
383 else if (skb
->protocol
== htons(ETH_P_IPV6
))
384 return ipv6_get_dsfield((const struct ipv6hdr
*)iph
);
389 /* Propogate ECN bits out */
390 static inline u8
ip_tunnel_ecn_encap(u8 tos
, const struct iphdr
*iph
,
391 const struct sk_buff
*skb
)
393 u8 inner
= ip_tunnel_get_dsfield(iph
, skb
);
395 return INET_ECN_encapsulate(tos
, inner
);
398 int __iptunnel_pull_header(struct sk_buff
*skb
, int hdr_len
,
399 __be16 inner_proto
, bool raw_proto
, bool xnet
);
401 static inline int iptunnel_pull_header(struct sk_buff
*skb
, int hdr_len
,
402 __be16 inner_proto
, bool xnet
)
404 return __iptunnel_pull_header(skb
, hdr_len
, inner_proto
, false, xnet
);
407 void iptunnel_xmit(struct sock
*sk
, struct rtable
*rt
, struct sk_buff
*skb
,
408 __be32 src
, __be32 dst
, u8 proto
,
409 u8 tos
, u8 ttl
, __be16 df
, bool xnet
);
410 struct metadata_dst
*iptunnel_metadata_reply(struct metadata_dst
*md
,
413 int iptunnel_handle_offloads(struct sk_buff
*skb
, int gso_type_mask
);
415 static inline int iptunnel_pull_offloads(struct sk_buff
*skb
)
417 if (skb_is_gso(skb
)) {
420 err
= skb_unclone(skb
, GFP_ATOMIC
);
423 skb_shinfo(skb
)->gso_type
&= ~(NETIF_F_GSO_ENCAP_ALL
>>
427 skb
->encapsulation
= 0;
431 static inline void iptunnel_xmit_stats(struct net_device
*dev
, int pkt_len
)
434 struct pcpu_sw_netstats
*tstats
= get_cpu_ptr(dev
->tstats
);
436 u64_stats_update_begin(&tstats
->syncp
);
437 tstats
->tx_bytes
+= pkt_len
;
438 tstats
->tx_packets
++;
439 u64_stats_update_end(&tstats
->syncp
);
442 struct net_device_stats
*err_stats
= &dev
->stats
;
445 err_stats
->tx_errors
++;
446 err_stats
->tx_aborted_errors
++;
448 err_stats
->tx_dropped
++;
453 static inline void *ip_tunnel_info_opts(struct ip_tunnel_info
*info
)
458 static inline void ip_tunnel_info_opts_get(void *to
,
459 const struct ip_tunnel_info
*info
)
461 memcpy(to
, info
+ 1, info
->options_len
);
464 static inline void ip_tunnel_info_opts_set(struct ip_tunnel_info
*info
,
465 const void *from
, int len
)
467 memcpy(ip_tunnel_info_opts(info
), from
, len
);
468 info
->options_len
= len
;
471 static inline struct ip_tunnel_info
*lwt_tun_info(struct lwtunnel_state
*lwtstate
)
473 return (struct ip_tunnel_info
*)lwtstate
->data
;
476 extern struct static_key ip_tunnel_metadata_cnt
;
478 /* Returns > 0 if metadata should be collected */
479 static inline int ip_tunnel_collect_metadata(void)
481 return static_key_false(&ip_tunnel_metadata_cnt
);
484 void __init
ip_tunnel_core_init(void);
486 void ip_tunnel_need_metadata(void);
487 void ip_tunnel_unneed_metadata(void);
489 #else /* CONFIG_INET */
491 static inline struct ip_tunnel_info
*lwt_tun_info(struct lwtunnel_state
*lwtstate
)
496 static inline void ip_tunnel_need_metadata(void)
500 static inline void ip_tunnel_unneed_metadata(void)
504 static inline void ip_tunnel_info_opts_get(void *to
,
505 const struct ip_tunnel_info
*info
)
509 static inline void ip_tunnel_info_opts_set(struct ip_tunnel_info
*info
,
510 const void *from
, int len
)
512 info
->options_len
= 0;
515 #endif /* CONFIG_INET */
517 #endif /* __NET_IP_TUNNELS_H */