Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
tjh.dev
kernel
os
linux
1#ifndef __NET_VXLAN_H
2#define __NET_VXLAN_H 1
3
4#include <linux/ip.h>
5#include <linux/ipv6.h>
6#include <linux/if_vlan.h>
7#include <linux/skbuff.h>
8#include <linux/netdevice.h>
9#include <linux/udp.h>
10#include <net/dst_metadata.h>
11
12/* VXLAN protocol (RFC 7348) header:
13 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
14 * |R|R|R|R|I|R|R|R| Reserved |
15 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
16 * | VXLAN Network Identifier (VNI) | Reserved |
17 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
18 *
19 * I = VXLAN Network Identifier (VNI) present.
20 */
21struct vxlanhdr {
22 __be32 vx_flags;
23 __be32 vx_vni;
24};
25
26/* VXLAN header flags. */
27#define VXLAN_HF_VNI cpu_to_be32(BIT(27))
28
29#define VXLAN_N_VID (1u << 24)
30#define VXLAN_VID_MASK (VXLAN_N_VID - 1)
31#define VXLAN_VNI_MASK cpu_to_be32(VXLAN_VID_MASK << 8)
32#define VXLAN_HLEN (sizeof(struct udphdr) + sizeof(struct vxlanhdr))
33
34#define VNI_HASH_BITS 10
35#define VNI_HASH_SIZE (1<<VNI_HASH_BITS)
36#define FDB_HASH_BITS 8
37#define FDB_HASH_SIZE (1<<FDB_HASH_BITS)
38
39/* Remote checksum offload for VXLAN (VXLAN_F_REMCSUM_[RT]X):
40 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
41 * |R|R|R|R|I|R|R|R|R|R|C| Reserved |
42 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
43 * | VXLAN Network Identifier (VNI) |O| Csum start |
44 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
45 *
46 * C = Remote checksum offload bit. When set indicates that the
47 * remote checksum offload data is present.
48 *
49 * O = Offset bit. Indicates the checksum offset relative to
50 * checksum start.
51 *
52 * Csum start = Checksum start divided by two.
53 *
54 * http://tools.ietf.org/html/draft-herbert-vxlan-rco
55 */
56
57/* VXLAN-RCO header flags. */
58#define VXLAN_HF_RCO cpu_to_be32(BIT(21))
59
60/* Remote checksum offload header option */
61#define VXLAN_RCO_MASK cpu_to_be32(0x7f) /* Last byte of vni field */
62#define VXLAN_RCO_UDP cpu_to_be32(0x80) /* Indicate UDP RCO (TCP when not set *) */
63#define VXLAN_RCO_SHIFT 1 /* Left shift of start */
64#define VXLAN_RCO_SHIFT_MASK ((1 << VXLAN_RCO_SHIFT) - 1)
65#define VXLAN_MAX_REMCSUM_START (0x7f << VXLAN_RCO_SHIFT)
66
67/*
68 * VXLAN Group Based Policy Extension (VXLAN_F_GBP):
69 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
70 * |G|R|R|R|I|R|R|R|R|D|R|R|A|R|R|R| Group Policy ID |
71 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
72 * | VXLAN Network Identifier (VNI) | Reserved |
73 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
74 *
75 * G = Group Policy ID present.
76 *
77 * D = Don't Learn bit. When set, this bit indicates that the egress
78 * VTEP MUST NOT learn the source address of the encapsulated frame.
79 *
80 * A = Indicates that the group policy has already been applied to
81 * this packet. Policies MUST NOT be applied by devices when the
82 * A bit is set.
83 *
84 * https://tools.ietf.org/html/draft-smith-vxlan-group-policy
85 */
86struct vxlanhdr_gbp {
87 u8 vx_flags;
88#ifdef __LITTLE_ENDIAN_BITFIELD
89 u8 reserved_flags1:3,
90 policy_applied:1,
91 reserved_flags2:2,
92 dont_learn:1,
93 reserved_flags3:1;
94#elif defined(__BIG_ENDIAN_BITFIELD)
95 u8 reserved_flags1:1,
96 dont_learn:1,
97 reserved_flags2:2,
98 policy_applied:1,
99 reserved_flags3:3;
100#else
101#error "Please fix <asm/byteorder.h>"
102#endif
103 __be16 policy_id;
104 __be32 vx_vni;
105};
106
107/* VXLAN-GBP header flags. */
108#define VXLAN_HF_GBP cpu_to_be32(BIT(31))
109
110#define VXLAN_GBP_USED_BITS (VXLAN_HF_GBP | cpu_to_be32(0xFFFFFF))
111
112/* skb->mark mapping
113 *
114 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
115 * |R|R|R|R|R|R|R|R|R|D|R|R|A|R|R|R| Group Policy ID |
116 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
117 */
118#define VXLAN_GBP_DONT_LEARN (BIT(6) << 16)
119#define VXLAN_GBP_POLICY_APPLIED (BIT(3) << 16)
120#define VXLAN_GBP_ID_MASK (0xFFFF)
121
122/*
123 * VXLAN Generic Protocol Extension (VXLAN_F_GPE):
124 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
125 * |R|R|Ver|I|P|R|O| Reserved |Next Protocol |
126 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
127 * | VXLAN Network Identifier (VNI) | Reserved |
128 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
129 *
130 * Ver = Version. Indicates VXLAN GPE protocol version.
131 *
132 * P = Next Protocol Bit. The P bit is set to indicate that the
133 * Next Protocol field is present.
134 *
135 * O = OAM Flag Bit. The O bit is set to indicate that the packet
136 * is an OAM packet.
137 *
138 * Next Protocol = This 8 bit field indicates the protocol header
139 * immediately following the VXLAN GPE header.
140 *
141 * https://tools.ietf.org/html/draft-ietf-nvo3-vxlan-gpe-01
142 */
143
144struct vxlanhdr_gpe {
145#if defined(__LITTLE_ENDIAN_BITFIELD)
146 u8 oam_flag:1,
147 reserved_flags1:1,
148 np_applied:1,
149 instance_applied:1,
150 version:2,
151reserved_flags2:2;
152#elif defined(__BIG_ENDIAN_BITFIELD)
153 u8 reserved_flags2:2,
154 version:2,
155 instance_applied:1,
156 np_applied:1,
157 reserved_flags1:1,
158 oam_flag:1;
159#endif
160 u8 reserved_flags3;
161 u8 reserved_flags4;
162 u8 next_protocol;
163 __be32 vx_vni;
164};
165
166/* VXLAN-GPE header flags. */
167#define VXLAN_HF_VER cpu_to_be32(BIT(29) | BIT(28))
168#define VXLAN_HF_NP cpu_to_be32(BIT(26))
169#define VXLAN_HF_OAM cpu_to_be32(BIT(24))
170
171#define VXLAN_GPE_USED_BITS (VXLAN_HF_VER | VXLAN_HF_NP | VXLAN_HF_OAM | \
172 cpu_to_be32(0xff))
173
174/* VXLAN-GPE header Next Protocol. */
175#define VXLAN_GPE_NP_IPV4 0x01
176#define VXLAN_GPE_NP_IPV6 0x02
177#define VXLAN_GPE_NP_ETHERNET 0x03
178#define VXLAN_GPE_NP_NSH 0x04
179
180struct vxlan_metadata {
181 u32 gbp;
182};
183
184/* per UDP socket information */
185struct vxlan_sock {
186 struct hlist_node hlist;
187 struct socket *sock;
188 struct hlist_head vni_list[VNI_HASH_SIZE];
189 atomic_t refcnt;
190 u32 flags;
191};
192
193union vxlan_addr {
194 struct sockaddr_in sin;
195 struct sockaddr_in6 sin6;
196 struct sockaddr sa;
197};
198
199struct vxlan_rdst {
200 union vxlan_addr remote_ip;
201 __be16 remote_port;
202 __be32 remote_vni;
203 u32 remote_ifindex;
204 struct list_head list;
205 struct rcu_head rcu;
206 struct dst_cache dst_cache;
207};
208
209struct vxlan_config {
210 union vxlan_addr remote_ip;
211 union vxlan_addr saddr;
212 __be32 vni;
213 int remote_ifindex;
214 int mtu;
215 __be16 dst_port;
216 u16 port_min;
217 u16 port_max;
218 u8 tos;
219 u8 ttl;
220 __be32 label;
221 u32 flags;
222 unsigned long age_interval;
223 unsigned int addrmax;
224 bool no_share;
225};
226
227/* Pseudo network device */
228struct vxlan_dev {
229 struct hlist_node hlist; /* vni hash table */
230 struct list_head next; /* vxlan's per namespace list */
231 struct vxlan_sock *vn4_sock; /* listening socket for IPv4 */
232#if IS_ENABLED(CONFIG_IPV6)
233 struct vxlan_sock *vn6_sock; /* listening socket for IPv6 */
234#endif
235 struct net_device *dev;
236 struct net *net; /* netns for packet i/o */
237 struct vxlan_rdst default_dst; /* default destination */
238 u32 flags; /* VXLAN_F_* in vxlan.h */
239
240 struct timer_list age_timer;
241 spinlock_t hash_lock;
242 unsigned int addrcnt;
243 struct gro_cells gro_cells;
244
245 struct vxlan_config cfg;
246
247 struct hlist_head fdb_head[FDB_HASH_SIZE];
248};
249
250#define VXLAN_F_LEARN 0x01
251#define VXLAN_F_PROXY 0x02
252#define VXLAN_F_RSC 0x04
253#define VXLAN_F_L2MISS 0x08
254#define VXLAN_F_L3MISS 0x10
255#define VXLAN_F_IPV6 0x20
256#define VXLAN_F_UDP_ZERO_CSUM_TX 0x40
257#define VXLAN_F_UDP_ZERO_CSUM6_TX 0x80
258#define VXLAN_F_UDP_ZERO_CSUM6_RX 0x100
259#define VXLAN_F_REMCSUM_TX 0x200
260#define VXLAN_F_REMCSUM_RX 0x400
261#define VXLAN_F_GBP 0x800
262#define VXLAN_F_REMCSUM_NOPARTIAL 0x1000
263#define VXLAN_F_COLLECT_METADATA 0x2000
264#define VXLAN_F_GPE 0x4000
265
266/* Flags that are used in the receive path. These flags must match in
267 * order for a socket to be shareable
268 */
269#define VXLAN_F_RCV_FLAGS (VXLAN_F_GBP | \
270 VXLAN_F_GPE | \
271 VXLAN_F_UDP_ZERO_CSUM6_RX | \
272 VXLAN_F_REMCSUM_RX | \
273 VXLAN_F_REMCSUM_NOPARTIAL | \
274 VXLAN_F_COLLECT_METADATA)
275
276/* Flags that can be set together with VXLAN_F_GPE. */
277#define VXLAN_F_ALLOWED_GPE (VXLAN_F_GPE | \
278 VXLAN_F_IPV6 | \
279 VXLAN_F_UDP_ZERO_CSUM_TX | \
280 VXLAN_F_UDP_ZERO_CSUM6_TX | \
281 VXLAN_F_UDP_ZERO_CSUM6_RX | \
282 VXLAN_F_COLLECT_METADATA)
283
284struct net_device *vxlan_dev_create(struct net *net, const char *name,
285 u8 name_assign_type, struct vxlan_config *conf);
286
287static inline __be16 vxlan_dev_dst_port(struct vxlan_dev *vxlan,
288 unsigned short family)
289{
290#if IS_ENABLED(CONFIG_IPV6)
291 if (family == AF_INET6)
292 return inet_sk(vxlan->vn6_sock->sock->sk)->inet_sport;
293#endif
294 return inet_sk(vxlan->vn4_sock->sock->sk)->inet_sport;
295}
296
297static inline netdev_features_t vxlan_features_check(struct sk_buff *skb,
298 netdev_features_t features)
299{
300 u8 l4_hdr = 0;
301
302 if (!skb->encapsulation)
303 return features;
304
305 switch (vlan_get_protocol(skb)) {
306 case htons(ETH_P_IP):
307 l4_hdr = ip_hdr(skb)->protocol;
308 break;
309 case htons(ETH_P_IPV6):
310 l4_hdr = ipv6_hdr(skb)->nexthdr;
311 break;
312 default:
313 return features;;
314 }
315
316 if ((l4_hdr == IPPROTO_UDP) &&
317 (skb->inner_protocol_type != ENCAP_TYPE_ETHER ||
318 skb->inner_protocol != htons(ETH_P_TEB) ||
319 (skb_inner_mac_header(skb) - skb_transport_header(skb) !=
320 sizeof(struct udphdr) + sizeof(struct vxlanhdr)) ||
321 (skb->ip_summed != CHECKSUM_NONE &&
322 !can_checksum_protocol(features, inner_eth_hdr(skb)->h_proto))))
323 return features & ~(NETIF_F_CSUM_MASK | NETIF_F_GSO_MASK);
324
325 return features;
326}
327
328/* IP header + UDP + VXLAN + Ethernet header */
329#define VXLAN_HEADROOM (20 + 8 + 8 + 14)
330/* IPv6 header + UDP + VXLAN + Ethernet header */
331#define VXLAN6_HEADROOM (40 + 8 + 8 + 14)
332
333static inline struct vxlanhdr *vxlan_hdr(struct sk_buff *skb)
334{
335 return (struct vxlanhdr *)(udp_hdr(skb) + 1);
336}
337
338static inline __be32 vxlan_vni(__be32 vni_field)
339{
340#if defined(__BIG_ENDIAN)
341 return (__force __be32)((__force u32)vni_field >> 8);
342#else
343 return (__force __be32)((__force u32)(vni_field & VXLAN_VNI_MASK) << 8);
344#endif
345}
346
347static inline __be32 vxlan_vni_field(__be32 vni)
348{
349#if defined(__BIG_ENDIAN)
350 return (__force __be32)((__force u32)vni << 8);
351#else
352 return (__force __be32)((__force u32)vni >> 8);
353#endif
354}
355
356static inline __be32 vxlan_tun_id_to_vni(__be64 tun_id)
357{
358#if defined(__BIG_ENDIAN)
359 return (__force __be32)tun_id;
360#else
361 return (__force __be32)((__force u64)tun_id >> 32);
362#endif
363}
364
365static inline __be64 vxlan_vni_to_tun_id(__be32 vni)
366{
367#if defined(__BIG_ENDIAN)
368 return (__force __be64)vni;
369#else
370 return (__force __be64)((u64)(__force u32)vni << 32);
371#endif
372}
373
374static inline size_t vxlan_rco_start(__be32 vni_field)
375{
376 return be32_to_cpu(vni_field & VXLAN_RCO_MASK) << VXLAN_RCO_SHIFT;
377}
378
379static inline size_t vxlan_rco_offset(__be32 vni_field)
380{
381 return (vni_field & VXLAN_RCO_UDP) ?
382 offsetof(struct udphdr, check) :
383 offsetof(struct tcphdr, check);
384}
385
386static inline __be32 vxlan_compute_rco(unsigned int start, unsigned int offset)
387{
388 __be32 vni_field = cpu_to_be32(start >> VXLAN_RCO_SHIFT);
389
390 if (offset == offsetof(struct udphdr, check))
391 vni_field |= VXLAN_RCO_UDP;
392 return vni_field;
393}
394
395static inline void vxlan_get_rx_port(struct net_device *netdev)
396{
397 ASSERT_RTNL();
398 call_netdevice_notifiers(NETDEV_OFFLOAD_PUSH_VXLAN, netdev);
399}
400
401static inline unsigned short vxlan_get_sk_family(struct vxlan_sock *vs)
402{
403 return vs->sock->sk->sk_family;
404}
405
406#endif