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 */ 21 struct 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 */ 86 struct 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 struct vxlan_metadata { 123 u32 gbp; 124 }; 125 126 /* per UDP socket information */ 127 struct vxlan_sock { 128 struct hlist_node hlist; 129 struct work_struct del_work; 130 struct socket *sock; 131 struct rcu_head rcu; 132 struct hlist_head vni_list[VNI_HASH_SIZE]; 133 atomic_t refcnt; 134 struct udp_offload udp_offloads; 135 u32 flags; 136 }; 137 138 union vxlan_addr { 139 struct sockaddr_in sin; 140 struct sockaddr_in6 sin6; 141 struct sockaddr sa; 142 }; 143 144 struct vxlan_rdst { 145 union vxlan_addr remote_ip; 146 __be16 remote_port; 147 __be32 remote_vni; 148 u32 remote_ifindex; 149 struct list_head list; 150 struct rcu_head rcu; 151 struct dst_cache dst_cache; 152 }; 153 154 struct vxlan_config { 155 union vxlan_addr remote_ip; 156 union vxlan_addr saddr; 157 __be32 vni; 158 int remote_ifindex; 159 int mtu; 160 __be16 dst_port; 161 u16 port_min; 162 u16 port_max; 163 u8 tos; 164 u8 ttl; 165 u32 flags; 166 unsigned long age_interval; 167 unsigned int addrmax; 168 bool no_share; 169 }; 170 171 /* Pseudo network device */ 172 struct vxlan_dev { 173 struct hlist_node hlist; /* vni hash table */ 174 struct list_head next; /* vxlan's per namespace list */ 175 struct vxlan_sock *vn4_sock; /* listening socket for IPv4 */ 176 #if IS_ENABLED(CONFIG_IPV6) 177 struct vxlan_sock *vn6_sock; /* listening socket for IPv6 */ 178 #endif 179 struct net_device *dev; 180 struct net *net; /* netns for packet i/o */ 181 struct vxlan_rdst default_dst; /* default destination */ 182 u32 flags; /* VXLAN_F_* in vxlan.h */ 183 184 struct timer_list age_timer; 185 spinlock_t hash_lock; 186 unsigned int addrcnt; 187 struct gro_cells gro_cells; 188 189 struct vxlan_config cfg; 190 191 struct hlist_head fdb_head[FDB_HASH_SIZE]; 192 }; 193 194 #define VXLAN_F_LEARN 0x01 195 #define VXLAN_F_PROXY 0x02 196 #define VXLAN_F_RSC 0x04 197 #define VXLAN_F_L2MISS 0x08 198 #define VXLAN_F_L3MISS 0x10 199 #define VXLAN_F_IPV6 0x20 200 #define VXLAN_F_UDP_ZERO_CSUM_TX 0x40 201 #define VXLAN_F_UDP_ZERO_CSUM6_TX 0x80 202 #define VXLAN_F_UDP_ZERO_CSUM6_RX 0x100 203 #define VXLAN_F_REMCSUM_TX 0x200 204 #define VXLAN_F_REMCSUM_RX 0x400 205 #define VXLAN_F_GBP 0x800 206 #define VXLAN_F_REMCSUM_NOPARTIAL 0x1000 207 #define VXLAN_F_COLLECT_METADATA 0x2000 208 209 /* Flags that are used in the receive path. These flags must match in 210 * order for a socket to be shareable 211 */ 212 #define VXLAN_F_RCV_FLAGS (VXLAN_F_GBP | \ 213 VXLAN_F_UDP_ZERO_CSUM6_RX | \ 214 VXLAN_F_REMCSUM_RX | \ 215 VXLAN_F_REMCSUM_NOPARTIAL | \ 216 VXLAN_F_COLLECT_METADATA) 217 218 struct net_device *vxlan_dev_create(struct net *net, const char *name, 219 u8 name_assign_type, struct vxlan_config *conf); 220 221 static inline __be16 vxlan_dev_dst_port(struct vxlan_dev *vxlan, 222 unsigned short family) 223 { 224 #if IS_ENABLED(CONFIG_IPV6) 225 if (family == AF_INET6) 226 return inet_sk(vxlan->vn6_sock->sock->sk)->inet_sport; 227 #endif 228 return inet_sk(vxlan->vn4_sock->sock->sk)->inet_sport; 229 } 230 231 static inline netdev_features_t vxlan_features_check(struct sk_buff *skb, 232 netdev_features_t features) 233 { 234 u8 l4_hdr = 0; 235 236 if (!skb->encapsulation) 237 return features; 238 239 switch (vlan_get_protocol(skb)) { 240 case htons(ETH_P_IP): 241 l4_hdr = ip_hdr(skb)->protocol; 242 break; 243 case htons(ETH_P_IPV6): 244 l4_hdr = ipv6_hdr(skb)->nexthdr; 245 break; 246 default: 247 return features;; 248 } 249 250 if ((l4_hdr == IPPROTO_UDP) && 251 (skb->inner_protocol_type != ENCAP_TYPE_ETHER || 252 skb->inner_protocol != htons(ETH_P_TEB) || 253 (skb_inner_mac_header(skb) - skb_transport_header(skb) != 254 sizeof(struct udphdr) + sizeof(struct vxlanhdr)))) 255 return features & ~(NETIF_F_CSUM_MASK | NETIF_F_GSO_MASK); 256 257 return features; 258 } 259 260 /* IP header + UDP + VXLAN + Ethernet header */ 261 #define VXLAN_HEADROOM (20 + 8 + 8 + 14) 262 /* IPv6 header + UDP + VXLAN + Ethernet header */ 263 #define VXLAN6_HEADROOM (40 + 8 + 8 + 14) 264 265 static inline struct vxlanhdr *vxlan_hdr(struct sk_buff *skb) 266 { 267 return (struct vxlanhdr *)(udp_hdr(skb) + 1); 268 } 269 270 static inline __be32 vxlan_vni(__be32 vni_field) 271 { 272 #if defined(__BIG_ENDIAN) 273 return vni_field >> 8; 274 #else 275 return (vni_field & VXLAN_VNI_MASK) << 8; 276 #endif 277 } 278 279 static inline __be32 vxlan_vni_field(__be32 vni) 280 { 281 #if defined(__BIG_ENDIAN) 282 return vni << 8; 283 #else 284 return vni >> 8; 285 #endif 286 } 287 288 static inline __be32 vxlan_tun_id_to_vni(__be64 tun_id) 289 { 290 #if defined(__BIG_ENDIAN) 291 return tun_id; 292 #else 293 return tun_id >> 32; 294 #endif 295 } 296 297 static inline __be64 vxlan_vni_to_tun_id(__be32 vni) 298 { 299 #if defined(__BIG_ENDIAN) 300 return (__be64)vni; 301 #else 302 return (__be64)vni << 32; 303 #endif 304 } 305 306 static inline size_t vxlan_rco_start(__be32 vni_field) 307 { 308 return be32_to_cpu(vni_field & VXLAN_RCO_MASK) << VXLAN_RCO_SHIFT; 309 } 310 311 static inline size_t vxlan_rco_offset(__be32 vni_field) 312 { 313 return (vni_field & VXLAN_RCO_UDP) ? 314 offsetof(struct udphdr, check) : 315 offsetof(struct tcphdr, check); 316 } 317 318 static inline __be32 vxlan_compute_rco(unsigned int start, unsigned int offset) 319 { 320 __be32 vni_field = cpu_to_be32(start >> VXLAN_RCO_SHIFT); 321 322 if (offset == offsetof(struct udphdr, check)) 323 vni_field |= VXLAN_RCO_UDP; 324 return vni_field; 325 } 326 327 #if IS_ENABLED(CONFIG_VXLAN) 328 void vxlan_get_rx_port(struct net_device *netdev); 329 #else 330 static inline void vxlan_get_rx_port(struct net_device *netdev) 331 { 332 } 333 #endif 334 335 static inline unsigned short vxlan_get_sk_family(struct vxlan_sock *vs) 336 { 337 return vs->sock->sk->sk_family; 338 } 339 340 #endif 341