1 /* SPDX-License-Identifier: GPL-2.0 */
3 #define __NET_VXLAN_H 1
5 #include <linux/if_vlan.h>
6 #include <net/udp_tunnel.h>
7 #include <net/dst_metadata.h>
8 #include <net/rtnetlink.h>
9 #include <net/switchdev.h>
11 /* VXLAN protocol (RFC 7348) header:
12 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
13 * |R|R|R|R|I|R|R|R| Reserved |
14 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
15 * | VXLAN Network Identifier (VNI) | Reserved |
16 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
18 * I = VXLAN Network Identifier (VNI) present.
25 /* VXLAN header flags. */
26 #define VXLAN_HF_VNI cpu_to_be32(BIT(27))
28 #define VXLAN_N_VID (1u << 24)
29 #define VXLAN_VID_MASK (VXLAN_N_VID - 1)
30 #define VXLAN_VNI_MASK cpu_to_be32(VXLAN_VID_MASK << 8)
31 #define VXLAN_HLEN (sizeof(struct udphdr) + sizeof(struct vxlanhdr))
33 #define VNI_HASH_BITS 10
34 #define VNI_HASH_SIZE (1<<VNI_HASH_BITS)
35 #define FDB_HASH_BITS 8
36 #define FDB_HASH_SIZE (1<<FDB_HASH_BITS)
38 /* Remote checksum offload for VXLAN (VXLAN_F_REMCSUM_[RT]X):
39 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
40 * |R|R|R|R|I|R|R|R|R|R|C| Reserved |
41 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
42 * | VXLAN Network Identifier (VNI) |O| Csum start |
43 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
45 * C = Remote checksum offload bit. When set indicates that the
46 * remote checksum offload data is present.
48 * O = Offset bit. Indicates the checksum offset relative to
51 * Csum start = Checksum start divided by two.
53 * http://tools.ietf.org/html/draft-herbert-vxlan-rco
56 /* VXLAN-RCO header flags. */
57 #define VXLAN_HF_RCO cpu_to_be32(BIT(21))
59 /* Remote checksum offload header option */
60 #define VXLAN_RCO_MASK cpu_to_be32(0x7f) /* Last byte of vni field */
61 #define VXLAN_RCO_UDP cpu_to_be32(0x80) /* Indicate UDP RCO (TCP when not set *) */
62 #define VXLAN_RCO_SHIFT 1 /* Left shift of start */
63 #define VXLAN_RCO_SHIFT_MASK ((1 << VXLAN_RCO_SHIFT) - 1)
64 #define VXLAN_MAX_REMCSUM_START (0x7f << VXLAN_RCO_SHIFT)
67 * VXLAN Group Based Policy Extension (VXLAN_F_GBP):
68 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
69 * |G|R|R|R|I|R|R|R|R|D|R|R|A|R|R|R| Group Policy ID |
70 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
71 * | VXLAN Network Identifier (VNI) | Reserved |
72 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
74 * G = Group Policy ID present.
76 * D = Don't Learn bit. When set, this bit indicates that the egress
77 * VTEP MUST NOT learn the source address of the encapsulated frame.
79 * A = Indicates that the group policy has already been applied to
80 * this packet. Policies MUST NOT be applied by devices when the
83 * https://tools.ietf.org/html/draft-smith-vxlan-group-policy
87 #ifdef __LITTLE_ENDIAN_BITFIELD
93 #elif defined(__BIG_ENDIAN_BITFIELD)
100 #error "Please fix <asm/byteorder.h>"
106 /* VXLAN-GBP header flags. */
107 #define VXLAN_HF_GBP cpu_to_be32(BIT(31))
109 #define VXLAN_GBP_USED_BITS (VXLAN_HF_GBP | cpu_to_be32(0xFFFFFF))
113 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
114 * |R|R|R|R|R|R|R|R|R|D|R|R|A|R|R|R| Group Policy ID |
115 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
117 #define VXLAN_GBP_DONT_LEARN (BIT(6) << 16)
118 #define VXLAN_GBP_POLICY_APPLIED (BIT(3) << 16)
119 #define VXLAN_GBP_ID_MASK (0xFFFF)
122 * VXLAN Generic Protocol Extension (VXLAN_F_GPE):
123 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
124 * |R|R|Ver|I|P|R|O| Reserved |Next Protocol |
125 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
126 * | VXLAN Network Identifier (VNI) | Reserved |
127 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
129 * Ver = Version. Indicates VXLAN GPE protocol version.
131 * P = Next Protocol Bit. The P bit is set to indicate that the
132 * Next Protocol field is present.
134 * O = OAM Flag Bit. The O bit is set to indicate that the packet
137 * Next Protocol = This 8 bit field indicates the protocol header
138 * immediately following the VXLAN GPE header.
140 * https://tools.ietf.org/html/draft-ietf-nvo3-vxlan-gpe-01
143 struct vxlanhdr_gpe
{
144 #if defined(__LITTLE_ENDIAN_BITFIELD)
151 #elif defined(__BIG_ENDIAN_BITFIELD)
152 u8 reserved_flags2
:2,
165 /* VXLAN-GPE header flags. */
166 #define VXLAN_HF_VER cpu_to_be32(BIT(29) | BIT(28))
167 #define VXLAN_HF_NP cpu_to_be32(BIT(26))
168 #define VXLAN_HF_OAM cpu_to_be32(BIT(24))
170 #define VXLAN_GPE_USED_BITS (VXLAN_HF_VER | VXLAN_HF_NP | VXLAN_HF_OAM | \
173 struct vxlan_metadata
{
177 /* per UDP socket information */
179 struct hlist_node hlist
;
181 struct hlist_head vni_list
[VNI_HASH_SIZE
];
187 struct sockaddr_in sin
;
188 struct sockaddr_in6 sin6
;
193 union vxlan_addr remote_ip
;
198 struct list_head list
;
200 struct dst_cache dst_cache
;
203 struct vxlan_config
{
204 union vxlan_addr remote_ip
;
205 union vxlan_addr saddr
;
216 unsigned long age_interval
;
217 unsigned int addrmax
;
219 enum ifla_vxlan_df df
;
222 struct vxlan_dev_node
{
223 struct hlist_node hlist
;
224 struct vxlan_dev
*vxlan
;
227 /* Pseudo network device */
229 struct vxlan_dev_node hlist4
; /* vni hash table for IPv4 socket */
230 #if IS_ENABLED(CONFIG_IPV6)
231 struct vxlan_dev_node hlist6
; /* vni hash table for IPv6 socket */
233 struct list_head next
; /* vxlan's per namespace list */
234 struct vxlan_sock __rcu
*vn4_sock
; /* listening socket for IPv4 */
235 #if IS_ENABLED(CONFIG_IPV6)
236 struct vxlan_sock __rcu
*vn6_sock
; /* listening socket for IPv6 */
238 struct net_device
*dev
;
239 struct net
*net
; /* netns for packet i/o */
240 struct vxlan_rdst default_dst
; /* default destination */
242 struct timer_list age_timer
;
243 spinlock_t hash_lock
;
244 unsigned int addrcnt
;
245 struct gro_cells gro_cells
;
247 struct vxlan_config cfg
;
249 struct hlist_head fdb_head
[FDB_HASH_SIZE
];
252 #define VXLAN_F_LEARN 0x01
253 #define VXLAN_F_PROXY 0x02
254 #define VXLAN_F_RSC 0x04
255 #define VXLAN_F_L2MISS 0x08
256 #define VXLAN_F_L3MISS 0x10
257 #define VXLAN_F_IPV6 0x20
258 #define VXLAN_F_UDP_ZERO_CSUM_TX 0x40
259 #define VXLAN_F_UDP_ZERO_CSUM6_TX 0x80
260 #define VXLAN_F_UDP_ZERO_CSUM6_RX 0x100
261 #define VXLAN_F_REMCSUM_TX 0x200
262 #define VXLAN_F_REMCSUM_RX 0x400
263 #define VXLAN_F_GBP 0x800
264 #define VXLAN_F_REMCSUM_NOPARTIAL 0x1000
265 #define VXLAN_F_COLLECT_METADATA 0x2000
266 #define VXLAN_F_GPE 0x4000
267 #define VXLAN_F_IPV6_LINKLOCAL 0x8000
268 #define VXLAN_F_TTL_INHERIT 0x10000
270 /* Flags that are used in the receive path. These flags must match in
271 * order for a socket to be shareable
273 #define VXLAN_F_RCV_FLAGS (VXLAN_F_GBP | \
275 VXLAN_F_UDP_ZERO_CSUM6_RX | \
276 VXLAN_F_REMCSUM_RX | \
277 VXLAN_F_REMCSUM_NOPARTIAL | \
278 VXLAN_F_COLLECT_METADATA)
280 /* Flags that can be set together with VXLAN_F_GPE. */
281 #define VXLAN_F_ALLOWED_GPE (VXLAN_F_GPE | \
283 VXLAN_F_IPV6_LINKLOCAL | \
284 VXLAN_F_UDP_ZERO_CSUM_TX | \
285 VXLAN_F_UDP_ZERO_CSUM6_TX | \
286 VXLAN_F_UDP_ZERO_CSUM6_RX | \
287 VXLAN_F_COLLECT_METADATA)
289 struct net_device
*vxlan_dev_create(struct net
*net
, const char *name
,
290 u8 name_assign_type
, struct vxlan_config
*conf
);
292 static inline netdev_features_t
vxlan_features_check(struct sk_buff
*skb
,
293 netdev_features_t features
)
297 if (!skb
->encapsulation
)
300 switch (vlan_get_protocol(skb
)) {
301 case htons(ETH_P_IP
):
302 l4_hdr
= ip_hdr(skb
)->protocol
;
304 case htons(ETH_P_IPV6
):
305 l4_hdr
= ipv6_hdr(skb
)->nexthdr
;
311 if ((l4_hdr
== IPPROTO_UDP
) &&
312 (skb
->inner_protocol_type
!= ENCAP_TYPE_ETHER
||
313 skb
->inner_protocol
!= htons(ETH_P_TEB
) ||
314 (skb_inner_mac_header(skb
) - skb_transport_header(skb
) !=
315 sizeof(struct udphdr
) + sizeof(struct vxlanhdr
)) ||
316 (skb
->ip_summed
!= CHECKSUM_NONE
&&
317 !can_checksum_protocol(features
, inner_eth_hdr(skb
)->h_proto
))))
318 return features
& ~(NETIF_F_CSUM_MASK
| NETIF_F_GSO_MASK
);
323 /* IP header + UDP + VXLAN + Ethernet header */
324 #define VXLAN_HEADROOM (20 + 8 + 8 + 14)
325 /* IPv6 header + UDP + VXLAN + Ethernet header */
326 #define VXLAN6_HEADROOM (40 + 8 + 8 + 14)
328 static inline struct vxlanhdr
*vxlan_hdr(struct sk_buff
*skb
)
330 return (struct vxlanhdr
*)(udp_hdr(skb
) + 1);
333 static inline __be32
vxlan_vni(__be32 vni_field
)
335 #if defined(__BIG_ENDIAN)
336 return (__force __be32
)((__force u32
)vni_field
>> 8);
338 return (__force __be32
)((__force u32
)(vni_field
& VXLAN_VNI_MASK
) << 8);
342 static inline __be32
vxlan_vni_field(__be32 vni
)
344 #if defined(__BIG_ENDIAN)
345 return (__force __be32
)((__force u32
)vni
<< 8);
347 return (__force __be32
)((__force u32
)vni
>> 8);
351 static inline size_t vxlan_rco_start(__be32 vni_field
)
353 return be32_to_cpu(vni_field
& VXLAN_RCO_MASK
) << VXLAN_RCO_SHIFT
;
356 static inline size_t vxlan_rco_offset(__be32 vni_field
)
358 return (vni_field
& VXLAN_RCO_UDP
) ?
359 offsetof(struct udphdr
, check
) :
360 offsetof(struct tcphdr
, check
);
363 static inline __be32
vxlan_compute_rco(unsigned int start
, unsigned int offset
)
365 __be32 vni_field
= cpu_to_be32(start
>> VXLAN_RCO_SHIFT
);
367 if (offset
== offsetof(struct udphdr
, check
))
368 vni_field
|= VXLAN_RCO_UDP
;
372 static inline unsigned short vxlan_get_sk_family(struct vxlan_sock
*vs
)
374 return vs
->sock
->sk
->sk_family
;
377 #if IS_ENABLED(CONFIG_IPV6)
379 static inline bool vxlan_addr_any(const union vxlan_addr
*ipa
)
381 if (ipa
->sa
.sa_family
== AF_INET6
)
382 return ipv6_addr_any(&ipa
->sin6
.sin6_addr
);
384 return ipa
->sin
.sin_addr
.s_addr
== htonl(INADDR_ANY
);
387 static inline bool vxlan_addr_multicast(const union vxlan_addr
*ipa
)
389 if (ipa
->sa
.sa_family
== AF_INET6
)
390 return ipv6_addr_is_multicast(&ipa
->sin6
.sin6_addr
);
392 return IN_MULTICAST(ntohl(ipa
->sin
.sin_addr
.s_addr
));
395 #else /* !IS_ENABLED(CONFIG_IPV6) */
397 static inline bool vxlan_addr_any(const union vxlan_addr
*ipa
)
399 return ipa
->sin
.sin_addr
.s_addr
== htonl(INADDR_ANY
);
402 static inline bool vxlan_addr_multicast(const union vxlan_addr
*ipa
)
404 return IN_MULTICAST(ntohl(ipa
->sin
.sin_addr
.s_addr
));
407 #endif /* IS_ENABLED(CONFIG_IPV6) */
409 static inline bool netif_is_vxlan(const struct net_device
*dev
)
411 return dev
->rtnl_link_ops
&&
412 !strcmp(dev
->rtnl_link_ops
->kind
, "vxlan");
415 struct switchdev_notifier_vxlan_fdb_info
{
416 struct switchdev_notifier_info info
; /* must be first */
417 union vxlan_addr remote_ip
;
421 u8 eth_addr
[ETH_ALEN
];
427 #if IS_ENABLED(CONFIG_VXLAN)
428 int vxlan_fdb_find_uc(struct net_device
*dev
, const u8
*mac
, __be32 vni
,
429 struct switchdev_notifier_vxlan_fdb_info
*fdb_info
);
430 int vxlan_fdb_replay(const struct net_device
*dev
, __be32 vni
,
431 struct notifier_block
*nb
,
432 struct netlink_ext_ack
*extack
);
433 void vxlan_fdb_clear_offload(const struct net_device
*dev
, __be32 vni
);
437 vxlan_fdb_find_uc(struct net_device
*dev
, const u8
*mac
, __be32 vni
,
438 struct switchdev_notifier_vxlan_fdb_info
*fdb_info
)
443 static inline int vxlan_fdb_replay(const struct net_device
*dev
, __be32 vni
,
444 struct notifier_block
*nb
,
445 struct netlink_ext_ack
*extack
)
451 vxlan_fdb_clear_offload(const struct net_device
*dev
, __be32 vni
)
456 static inline void vxlan_flag_attr_error(int attrtype
,
457 struct netlink_ext_ack
*extack
)
459 #define VXLAN_FLAG(flg) \
460 case IFLA_VXLAN_##flg: \
461 NL_SET_ERR_MSG_MOD(extack, \
462 "cannot change " #flg " flag"); \
465 VXLAN_FLAG(TTL_INHERIT
);
466 VXLAN_FLAG(LEARNING
);
471 VXLAN_FLAG(COLLECT_METADATA
);
472 VXLAN_FLAG(UDP_ZERO_CSUM6_TX
);
473 VXLAN_FLAG(UDP_ZERO_CSUM6_RX
);
474 VXLAN_FLAG(REMCSUM_TX
);
475 VXLAN_FLAG(REMCSUM_RX
);
478 VXLAN_FLAG(REMCSUM_NOPARTIAL
);
480 NL_SET_ERR_MSG_MOD(extack
, \
481 "cannot change flag");