[mirror_ubuntu-zesty-kernel.git] / include / net / vxlan.h

#ifndef __NET_VXLAN_H
#define __NET_VXLAN_H 1

#include <linux/ip.h>
#include <linux/ipv6.h>
#include <linux/if_vlan.h>
#include <linux/skbuff.h>
#include <linux/netdevice.h>
#include <linux/udp.h>
#include <net/dst_metadata.h>

/*
 * VXLAN Group Based Policy Extension:
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 * |1|-|-|-|1|-|-|-|R|D|R|R|A|R|R|R|        Group Policy ID        |
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 * |                VXLAN Network Identifier (VNI) |   Reserved    |
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 *
 * D = Don't Learn bit. When set, this bit indicates that the egress
 *     VTEP MUST NOT learn the source address of the encapsulated frame.
 *
 * A = Indicates that the group policy has already been applied to
 *     this packet. Policies MUST NOT be applied by devices when the
 *     A bit is set.
 *
 * [0] https://tools.ietf.org/html/draft-smith-vxlan-group-policy
 */
struct vxlanhdr_gbp {
	u8	vx_flags;
#ifdef __LITTLE_ENDIAN_BITFIELD
	u8	reserved_flags1:3,
		policy_applied:1,
		reserved_flags2:2,
		dont_learn:1,
		reserved_flags3:1;
#elif defined(__BIG_ENDIAN_BITFIELD)
	u8	reserved_flags1:1,
		dont_learn:1,
		reserved_flags2:2,
		policy_applied:1,
		reserved_flags3:3;
#else
#error	"Please fix <asm/byteorder.h>"
#endif
	__be16	policy_id;
	__be32	vx_vni;
};

#define VXLAN_GBP_USED_BITS (VXLAN_HF_GBP | 0xFFFFFF)

/* skb->mark mapping
 *
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 * |R|R|R|R|R|R|R|R|R|D|R|R|A|R|R|R|        Group Policy ID        |
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 */
#define VXLAN_GBP_DONT_LEARN		(BIT(6) << 16)
#define VXLAN_GBP_POLICY_APPLIED	(BIT(3) << 16)
#define VXLAN_GBP_ID_MASK		(0xFFFF)

/* VXLAN protocol header:
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 * |G|R|R|R|I|R|R|C|               Reserved                        |
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 * |                VXLAN Network Identifier (VNI) |   Reserved    |
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 *
 * G = 1	Group Policy (VXLAN-GBP)
 * I = 1	VXLAN Network Identifier (VNI) present
 * C = 1	Remote checksum offload (RCO)
 */
struct vxlanhdr {
	__be32 vx_flags;
	__be32 vx_vni;
};

/* VXLAN header flags. */
#define VXLAN_HF_RCO BIT(21)
#define VXLAN_HF_VNI BIT(27)
#define VXLAN_HF_GBP BIT(31)

/* Remote checksum offload header option */
#define VXLAN_RCO_MASK  0x7f    /* Last byte of vni field */
#define VXLAN_RCO_UDP   0x80    /* Indicate UDP RCO (TCP when not set *) */
#define VXLAN_RCO_SHIFT 1       /* Left shift of start */
#define VXLAN_RCO_SHIFT_MASK ((1 << VXLAN_RCO_SHIFT) - 1)
#define VXLAN_MAX_REMCSUM_START (VXLAN_RCO_MASK << VXLAN_RCO_SHIFT)

#define VXLAN_N_VID     (1u << 24)
#define VXLAN_VID_MASK  (VXLAN_N_VID - 1)
#define VXLAN_VNI_MASK  (VXLAN_VID_MASK << 8)
#define VXLAN_HLEN (sizeof(struct udphdr) + sizeof(struct vxlanhdr))

#define VNI_HASH_BITS	10
#define VNI_HASH_SIZE	(1<<VNI_HASH_BITS)
#define FDB_HASH_BITS	8
#define FDB_HASH_SIZE	(1<<FDB_HASH_BITS)

struct vxlan_metadata {
	u32		gbp;
};

/* per UDP socket information */
struct vxlan_sock {
	struct hlist_node hlist;
	struct work_struct del_work;
	struct socket	 *sock;
	struct rcu_head	  rcu;
	struct hlist_head vni_list[VNI_HASH_SIZE];
	atomic_t	  refcnt;
	struct udp_offload udp_offloads;
	u32		  flags;
};

union vxlan_addr {
	struct sockaddr_in sin;
	struct sockaddr_in6 sin6;
	struct sockaddr sa;
};

struct vxlan_rdst {
	union vxlan_addr	 remote_ip;
	__be16			 remote_port;
	u32			 remote_vni;
	u32			 remote_ifindex;
	struct list_head	 list;
	struct rcu_head		 rcu;
};

struct vxlan_config {
	union vxlan_addr	remote_ip;
	union vxlan_addr	saddr;
	u32			vni;
	int			remote_ifindex;
	int			mtu;
	__be16			dst_port;
	u16			port_min;
	u16			port_max;
	u8			tos;
	u8			ttl;
	u32			flags;
	unsigned long		age_interval;
	unsigned int		addrmax;
	bool			no_share;
};

/* Pseudo network device */
struct vxlan_dev {
	struct hlist_node hlist;	/* vni hash table */
	struct list_head  next;		/* vxlan's per namespace list */
	struct vxlan_sock *vn4_sock;	/* listening socket for IPv4 */
#if IS_ENABLED(CONFIG_IPV6)
	struct vxlan_sock *vn6_sock;	/* listening socket for IPv6 */
#endif
	struct net_device *dev;
	struct net	  *net;		/* netns for packet i/o */
	struct vxlan_rdst default_dst;	/* default destination */
	u32		  flags;	/* VXLAN_F_* in vxlan.h */

	struct timer_list age_timer;
	spinlock_t	  hash_lock;
	unsigned int	  addrcnt;
	struct gro_cells  gro_cells;

	struct vxlan_config	cfg;

	struct hlist_head fdb_head[FDB_HASH_SIZE];
};

#define VXLAN_F_LEARN			0x01
#define VXLAN_F_PROXY			0x02
#define VXLAN_F_RSC			0x04
#define VXLAN_F_L2MISS			0x08
#define VXLAN_F_L3MISS			0x10
#define VXLAN_F_IPV6			0x20
#define VXLAN_F_UDP_CSUM		0x40
#define VXLAN_F_UDP_ZERO_CSUM6_TX	0x80
#define VXLAN_F_UDP_ZERO_CSUM6_RX	0x100
#define VXLAN_F_REMCSUM_TX		0x200
#define VXLAN_F_REMCSUM_RX		0x400
#define VXLAN_F_GBP			0x800
#define VXLAN_F_REMCSUM_NOPARTIAL	0x1000
#define VXLAN_F_COLLECT_METADATA	0x2000

/* Flags that are used in the receive path. These flags must match in
 * order for a socket to be shareable
 */
#define VXLAN_F_RCV_FLAGS		(VXLAN_F_GBP |			\
					 VXLAN_F_UDP_ZERO_CSUM6_RX |	\
					 VXLAN_F_REMCSUM_RX |		\
					 VXLAN_F_REMCSUM_NOPARTIAL |	\
					 VXLAN_F_COLLECT_METADATA)

struct net_device *vxlan_dev_create(struct net *net, const char *name,
				    u8 name_assign_type, struct vxlan_config *conf);

static inline __be16 vxlan_dev_dst_port(struct vxlan_dev *vxlan,
					unsigned short family)
{
#if IS_ENABLED(CONFIG_IPV6)
	if (family == AF_INET6)
		return inet_sk(vxlan->vn6_sock->sock->sk)->inet_sport;
#endif
	return inet_sk(vxlan->vn4_sock->sock->sk)->inet_sport;
}

static inline netdev_features_t vxlan_features_check(struct sk_buff *skb,
						     netdev_features_t features)
{
	u8 l4_hdr = 0;

	if (!skb->encapsulation)
		return features;

	switch (vlan_get_protocol(skb)) {
	case htons(ETH_P_IP):
		l4_hdr = ip_hdr(skb)->protocol;
		break;
	case htons(ETH_P_IPV6):
		l4_hdr = ipv6_hdr(skb)->nexthdr;
		break;
	default:
		return features;;
	}

	if ((l4_hdr == IPPROTO_UDP) &&
	    (skb->inner_protocol_type != ENCAP_TYPE_ETHER ||
	     skb->inner_protocol != htons(ETH_P_TEB) ||
	     (skb_inner_mac_header(skb) - skb_transport_header(skb) !=
	      sizeof(struct udphdr) + sizeof(struct vxlanhdr))))
		return features & ~(NETIF_F_CSUM_MASK | NETIF_F_GSO_MASK);

	return features;
}

/* IP header + UDP + VXLAN + Ethernet header */
#define VXLAN_HEADROOM (20 + 8 + 8 + 14)
/* IPv6 header + UDP + VXLAN + Ethernet header */
#define VXLAN6_HEADROOM (40 + 8 + 8 + 14)

#if IS_ENABLED(CONFIG_VXLAN)
void vxlan_get_rx_port(struct net_device *netdev);
#else
static inline void vxlan_get_rx_port(struct net_device *netdev)
{
}
#endif

static inline unsigned short vxlan_get_sk_family(struct vxlan_sock *vs)
{
	return vs->sock->sk->sk_family;
}

#endif
Commit	Line	Data
012a5729 PS	1	#ifndef __NET_VXLAN_H
	2	#define __NET_VXLAN_H 1
	3
5f35227e JG	4	#include <linux/ip.h>
	5	#include <linux/ipv6.h>
	6	#include <linux/if_vlan.h>
012a5729 PS	7	#include <linux/skbuff.h>
	8	#include <linux/netdevice.h>
	9	#include <linux/udp.h>
ee122c79	10	#include <net/dst_metadata.h>
012a5729	11
3511494c TG	12	/*
	13	* VXLAN Group Based Policy Extension:
	14	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	15	* \|1\|-\|-\|-\|1\|-\|-\|-\|R\|D\|R\|R\|A\|R\|R\|R\| Group Policy ID \|
	16	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	17	* \| VXLAN Network Identifier (VNI) \| Reserved \|
	18	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	19	*
	20	* D = Don't Learn bit. When set, this bit indicates that the egress
	21	* VTEP MUST NOT learn the source address of the encapsulated frame.
	22	*
	23	* A = Indicates that the group policy has already been applied to
	24	* this packet. Policies MUST NOT be applied by devices when the
	25	* A bit is set.
	26	*
	27	* [0] https://tools.ietf.org/html/draft-smith-vxlan-group-policy
	28	*/
	29	struct vxlanhdr_gbp {
0e715d6f	30	u8 vx_flags;
3511494c	31	#ifdef __LITTLE_ENDIAN_BITFIELD
0e715d6f	32	u8 reserved_flags1:3,
3511494c TG	33	policy_applied:1,
	34	reserved_flags2:2,
	35	dont_learn:1,
	36	reserved_flags3:1;
	37	#elif defined(__BIG_ENDIAN_BITFIELD)
0e715d6f	38	u8 reserved_flags1:1,
3511494c TG	39	dont_learn:1,
	40	reserved_flags2:2,
	41	policy_applied:1,
	42	reserved_flags3:3;
	43	#else
	44	#error "Please fix <asm/byteorder.h>"
	45	#endif
	46	__be16 policy_id;
	47	__be32 vx_vni;
	48	};
	49
	50	#define VXLAN_GBP_USED_BITS (VXLAN_HF_GBP \| 0xFFFFFF)
	51
	52	/* skb->mark mapping
	53	*
	54	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	55	* \|R\|R\|R\|R\|R\|R\|R\|R\|R\|D\|R\|R\|A\|R\|R\|R\| Group Policy ID \|
	56	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	57	*/
	58	#define VXLAN_GBP_DONT_LEARN (BIT(6) << 16)
	59	#define VXLAN_GBP_POLICY_APPLIED (BIT(3) << 16)
	60	#define VXLAN_GBP_ID_MASK (0xFFFF)
	61
	62	/* VXLAN protocol header:
	63	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	64	* \|G\|R\|R\|R\|I\|R\|R\|C\| Reserved \|
	65	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	66	* \| VXLAN Network Identifier (VNI) \| Reserved \|
	67	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	68	*
	69	* G = 1 Group Policy (VXLAN-GBP)
	70	* I = 1 VXLAN Network Identifier (VNI) present
	71	* C = 1 Remote checksum offload (RCO)
	72	*/
11bf7828 JS	73	struct vxlanhdr {
	74	__be32 vx_flags;
	75	__be32 vx_vni;
	76	};
	77
3bf39475	78	/* VXLAN header flags. */
c5fb8caa	79	#define VXLAN_HF_RCO BIT(21)
3511494c TG	80	#define VXLAN_HF_VNI BIT(27)
3511494c TG	81	#define VXLAN_HF_GBP BIT(31)
dfd8645e TH	82
	83	/* Remote checksum offload header option */
	84	#define VXLAN_RCO_MASK 0x7f /* Last byte of vni field */
	85	#define VXLAN_RCO_UDP 0x80 /* Indicate UDP RCO (TCP when not set ) /
	86	#define VXLAN_RCO_SHIFT 1 /* Left shift of start */
	87	#define VXLAN_RCO_SHIFT_MASK ((1 << VXLAN_RCO_SHIFT) - 1)
	88	#define VXLAN_MAX_REMCSUM_START (VXLAN_RCO_MASK << VXLAN_RCO_SHIFT)
3bf39475 TH	89
	90	#define VXLAN_N_VID (1u << 24)
	91	#define VXLAN_VID_MASK (VXLAN_N_VID - 1)
40fb70f3	92	#define VXLAN_VNI_MASK (VXLAN_VID_MASK << 8)
3bf39475 TH	93	#define VXLAN_HLEN (sizeof(struct udphdr) + sizeof(struct vxlanhdr))
3bf39475 TH	94
0dfbdf41 TG	95	#define VNI_HASH_BITS 10
	96	#define VNI_HASH_SIZE (1<<VNI_HASH_BITS)
	97	#define FDB_HASH_BITS 8
	98	#define FDB_HASH_SIZE (1<<FDB_HASH_BITS)
	99
3511494c	100	struct vxlan_metadata {
3511494c TG	101	u32 gbp;
	102	};
	103
012a5729 PS	104	/* per UDP socket information */
	105	struct vxlan_sock {
	106	struct hlist_node hlist;
012a5729 PS	107	struct work_struct del_work;
	108	struct socket *sock;
	109	struct rcu_head rcu;
	110	struct hlist_head vni_list[VNI_HASH_SIZE];
	111	atomic_t refcnt;
dc01e7d3	112	struct udp_offload udp_offloads;
dfd8645e	113	u32 flags;
012a5729 PS	114	};
012a5729 PS	115
0dfbdf41 TG	116	union vxlan_addr {
	117	struct sockaddr_in sin;
	118	struct sockaddr_in6 sin6;
	119	struct sockaddr sa;
	120	};
	121
	122	struct vxlan_rdst {
	123	union vxlan_addr remote_ip;
	124	__be16 remote_port;
	125	u32 remote_vni;
	126	u32 remote_ifindex;
	127	struct list_head list;
	128	struct rcu_head rcu;
	129	};
	130
	131	struct vxlan_config {
	132	union vxlan_addr remote_ip;
	133	union vxlan_addr saddr;
	134	u32 vni;
	135	int remote_ifindex;
	136	int mtu;
	137	__be16 dst_port;
0e715d6f JB	138	u16 port_min;
	139	u16 port_max;
	140	u8 tos;
	141	u8 ttl;
0dfbdf41 TG	142	u32 flags;
	143	unsigned long age_interval;
	144	unsigned int addrmax;
	145	bool no_share;
	146	};
	147
	148	/* Pseudo network device */
	149	struct vxlan_dev {
	150	struct hlist_node hlist; /* vni hash table */
	151	struct list_head next; /* vxlan's per namespace list */
b1be00a6 JB	152	struct vxlan_sock vn4_sock; / listening socket for IPv4 */
	153	#if IS_ENABLED(CONFIG_IPV6)
	154	struct vxlan_sock vn6_sock; / listening socket for IPv6 */
	155	#endif
0dfbdf41 TG	156	struct net_device *dev;
	157	struct net net; / netns for packet i/o */
	158	struct vxlan_rdst default_dst; /* default destination */
	159	u32 flags; /* VXLAN_F_* in vxlan.h */
	160
	161	struct timer_list age_timer;
	162	spinlock_t hash_lock;
	163	unsigned int addrcnt;
58ce31cc	164	struct gro_cells gro_cells;
0dfbdf41 TG	165
	166	struct vxlan_config cfg;
	167
	168	struct hlist_head fdb_head[FDB_HASH_SIZE];
	169	};
	170
359a0ea9 TH	171	#define VXLAN_F_LEARN 0x01
	172	#define VXLAN_F_PROXY 0x02
	173	#define VXLAN_F_RSC 0x04
	174	#define VXLAN_F_L2MISS 0x08
	175	#define VXLAN_F_L3MISS 0x10
	176	#define VXLAN_F_IPV6 0x20
	177	#define VXLAN_F_UDP_CSUM 0x40
	178	#define VXLAN_F_UDP_ZERO_CSUM6_TX 0x80
	179	#define VXLAN_F_UDP_ZERO_CSUM6_RX 0x100
dfd8645e TH	180	#define VXLAN_F_REMCSUM_TX 0x200
dfd8645e TH	181	#define VXLAN_F_REMCSUM_RX 0x400
3511494c	182	#define VXLAN_F_GBP 0x800
0ace2ca8	183	#define VXLAN_F_REMCSUM_NOPARTIAL 0x1000
ee122c79	184	#define VXLAN_F_COLLECT_METADATA 0x2000
359a0ea9	185
d299ce14	186	/* Flags that are used in the receive path. These flags must match in
af33c1ad TH	187	* order for a socket to be shareable
	188	*/
	189	#define VXLAN_F_RCV_FLAGS (VXLAN_F_GBP \| \
	190	VXLAN_F_UDP_ZERO_CSUM6_RX \| \
0ace2ca8	191	VXLAN_F_REMCSUM_RX \| \
ee122c79	192	VXLAN_F_REMCSUM_NOPARTIAL \| \
da8b43c0	193	VXLAN_F_COLLECT_METADATA)
ac5132d1	194
0dfbdf41 TG	195	struct net_device vxlan_dev_create(struct net net, const char *name,
	196	u8 name_assign_type, struct vxlan_config *conf);
	197
b1be00a6 JB	198	static inline __be16 vxlan_dev_dst_port(struct vxlan_dev *vxlan,
b1be00a6 JB	199	unsigned short family)
614732ea	200	{
b1be00a6 JB	201	#if IS_ENABLED(CONFIG_IPV6)
	202	if (family == AF_INET6)
	203	return inet_sk(vxlan->vn6_sock->sock->sk)->inet_sport;
	204	#endif
	205	return inet_sk(vxlan->vn4_sock->sock->sk)->inet_sport;
614732ea	206	}
49560532	207
5f35227e JG	208	static inline netdev_features_t vxlan_features_check(struct sk_buff *skb,
5f35227e JG	209	netdev_features_t features)
11bf7828	210	{
5f35227e JG	211	u8 l4_hdr = 0;
	212
	213	if (!skb->encapsulation)
	214	return features;
	215
	216	switch (vlan_get_protocol(skb)) {
	217	case htons(ETH_P_IP):
	218	l4_hdr = ip_hdr(skb)->protocol;
	219	break;
	220	case htons(ETH_P_IPV6):
	221	l4_hdr = ipv6_hdr(skb)->nexthdr;
	222	break;
	223	default:
	224	return features;;
	225	}
	226
	227	if ((l4_hdr == IPPROTO_UDP) &&
11bf7828 JS	228	(skb->inner_protocol_type != ENCAP_TYPE_ETHER \|\|
	229	skb->inner_protocol != htons(ETH_P_TEB) \|\|
	230	(skb_inner_mac_header(skb) - skb_transport_header(skb) !=
	231	sizeof(struct udphdr) + sizeof(struct vxlanhdr))))
a188222b	232	return features & ~(NETIF_F_CSUM_MASK \| NETIF_F_GSO_MASK);
11bf7828	233
5f35227e	234	return features;
11bf7828	235	}
23e62de3	236
e6cd988c JG	237	/* IP header + UDP + VXLAN + Ethernet header */
	238	#define VXLAN_HEADROOM (20 + 8 + 8 + 14)
	239	/* IPv6 header + UDP + VXLAN + Ethernet header */
	240	#define VXLAN6_HEADROOM (40 + 8 + 8 + 14)
	241
	242	#if IS_ENABLED(CONFIG_VXLAN)
53cf5275	243	void vxlan_get_rx_port(struct net_device *netdev);
e6cd988c JG	244	#else
	245	static inline void vxlan_get_rx_port(struct net_device *netdev)
	246	{
	247	}
	248	#endif
705cc62f JB	249
	250	static inline unsigned short vxlan_get_sk_family(struct vxlan_sock *vs)
	251	{
	252	return vs->sock->sk->sk_family;
	253	}
48e92c44 JB	254
48e92c44 JB	255	#endif