[deliverable/linux.git] / include / net / vxlan.h

#ifndef __NET_VXLAN_H
#define __NET_VXLAN_H 1

#include <linux/if_vlan.h>
#include <net/udp_tunnel.h>
#include <net/dst_metadata.h>
#include <net/udp_tunnel.h>

/* VXLAN protocol (RFC 7348) header:
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 * |R|R|R|R|I|R|R|R|               Reserved                        |
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 * |                VXLAN Network Identifier (VNI) |   Reserved    |
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 *
 * I = VXLAN Network Identifier (VNI) present.
 */
struct vxlanhdr {
	__be32 vx_flags;
	__be32 vx_vni;
};

/* VXLAN header flags. */
#define VXLAN_HF_VNI	cpu_to_be32(BIT(27))

#define VXLAN_N_VID     (1u << 24)
#define VXLAN_VID_MASK  (VXLAN_N_VID - 1)
#define VXLAN_VNI_MASK	cpu_to_be32(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)

/* Remote checksum offload for VXLAN (VXLAN_F_REMCSUM_[RT]X):
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 * |R|R|R|R|I|R|R|R|R|R|C|              Reserved                   |
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 * |           VXLAN Network Identifier (VNI)      |O| Csum start  |
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 *
 * C = Remote checksum offload bit. When set indicates that the
 *     remote checksum offload data is present.
 *
 * O = Offset bit. Indicates the checksum offset relative to
 *     checksum start.
 *
 * Csum start = Checksum start divided by two.
 *
 * http://tools.ietf.org/html/draft-herbert-vxlan-rco
 */

/* VXLAN-RCO header flags. */
#define VXLAN_HF_RCO	cpu_to_be32(BIT(21))

/* Remote checksum offload header option */
#define VXLAN_RCO_MASK	cpu_to_be32(0x7f)  /* Last byte of vni field */
#define VXLAN_RCO_UDP	cpu_to_be32(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 (0x7f << VXLAN_RCO_SHIFT)

/*
 * VXLAN Group Based Policy Extension (VXLAN_F_GBP):
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 * |G|R|R|R|I|R|R|R|R|D|R|R|A|R|R|R|        Group Policy ID        |
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 * |                VXLAN Network Identifier (VNI) |   Reserved    |
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 *
 * G = Group Policy ID present.
 *
 * 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.
 *
 * 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;
};

/* VXLAN-GBP header flags. */
#define VXLAN_HF_GBP	cpu_to_be32(BIT(31))

#define VXLAN_GBP_USED_BITS (VXLAN_HF_GBP | cpu_to_be32(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 Generic Protocol Extension (VXLAN_F_GPE):
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 * |R|R|Ver|I|P|R|O|       Reserved                |Next Protocol  |
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 * |                VXLAN Network Identifier (VNI) |   Reserved    |
 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 *
 * Ver = Version. Indicates VXLAN GPE protocol version.
 *
 * P = Next Protocol Bit. The P bit is set to indicate that the
 *     Next Protocol field is present.
 *
 * O = OAM Flag Bit. The O bit is set to indicate that the packet
 *     is an OAM packet.
 *
 * Next Protocol = This 8 bit field indicates the protocol header
 * immediately following the VXLAN GPE header.
 *
 * https://tools.ietf.org/html/draft-ietf-nvo3-vxlan-gpe-01
 */

struct vxlanhdr_gpe {
#if defined(__LITTLE_ENDIAN_BITFIELD)
	u8	oam_flag:1,
		reserved_flags1:1,
		np_applied:1,
		instance_applied:1,
		version:2,
reserved_flags2:2;
#elif defined(__BIG_ENDIAN_BITFIELD)
	u8	reserved_flags2:2,
		version:2,
		instance_applied:1,
		np_applied:1,
		reserved_flags1:1,
		oam_flag:1;
#endif
	u8	reserved_flags3;
	u8	reserved_flags4;
	u8	next_protocol;
	__be32	vx_vni;
};

/* VXLAN-GPE header flags. */
#define VXLAN_HF_VER	cpu_to_be32(BIT(29) | BIT(28))
#define VXLAN_HF_NP	cpu_to_be32(BIT(26))
#define VXLAN_HF_OAM	cpu_to_be32(BIT(24))

#define VXLAN_GPE_USED_BITS (VXLAN_HF_VER | VXLAN_HF_NP | VXLAN_HF_OAM | \
			     cpu_to_be32(0xff))

/* VXLAN-GPE header Next Protocol. */
#define VXLAN_GPE_NP_IPV4      0x01
#define VXLAN_GPE_NP_IPV6      0x02
#define VXLAN_GPE_NP_ETHERNET  0x03
#define VXLAN_GPE_NP_NSH       0x04

struct vxlan_metadata {
	u32		gbp;
};

/* per UDP socket information */
struct vxlan_sock {
	struct hlist_node hlist;
	struct socket	 *sock;
	struct hlist_head vni_list[VNI_HASH_SIZE];
	atomic_t	  refcnt;
	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;
	__be32			 remote_vni;
	u32			 remote_ifindex;
	struct list_head	 list;
	struct rcu_head		 rcu;
	struct dst_cache	 dst_cache;
};

struct vxlan_config {
	union vxlan_addr	remote_ip;
	union vxlan_addr	saddr;
	__be32			vni;
	int			remote_ifindex;
	int			mtu;
	__be16			dst_port;
	u16			port_min;
	u16			port_max;
	u8			tos;
	u8			ttl;
	__be32			label;
	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_ZERO_CSUM_TX	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
#define VXLAN_F_GPE			0x4000

/* 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_GPE |			\
					 VXLAN_F_UDP_ZERO_CSUM6_RX |	\
					 VXLAN_F_REMCSUM_RX |		\
					 VXLAN_F_REMCSUM_NOPARTIAL |	\
					 VXLAN_F_COLLECT_METADATA)

/* Flags that can be set together with VXLAN_F_GPE. */
#define VXLAN_F_ALLOWED_GPE		(VXLAN_F_GPE |			\
					 VXLAN_F_IPV6 |			\
					 VXLAN_F_UDP_ZERO_CSUM_TX |	\
					 VXLAN_F_UDP_ZERO_CSUM6_TX |	\
					 VXLAN_F_UDP_ZERO_CSUM6_RX |	\
					 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)) ||
	     (skb->ip_summed != CHECKSUM_NONE &&
	      !can_checksum_protocol(features, inner_eth_hdr(skb)->h_proto))))
		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)

static inline struct vxlanhdr *vxlan_hdr(struct sk_buff *skb)
{
	return (struct vxlanhdr *)(udp_hdr(skb) + 1);
}

static inline __be32 vxlan_vni(__be32 vni_field)
{
#if defined(__BIG_ENDIAN)
	return (__force __be32)((__force u32)vni_field >> 8);
#else
	return (__force __be32)((__force u32)(vni_field & VXLAN_VNI_MASK) << 8);
#endif
}

static inline __be32 vxlan_vni_field(__be32 vni)
{
#if defined(__BIG_ENDIAN)
	return (__force __be32)((__force u32)vni << 8);
#else
	return (__force __be32)((__force u32)vni >> 8);
#endif
}

static inline size_t vxlan_rco_start(__be32 vni_field)
{
	return be32_to_cpu(vni_field & VXLAN_RCO_MASK) << VXLAN_RCO_SHIFT;
}

static inline size_t vxlan_rco_offset(__be32 vni_field)
{
	return (vni_field & VXLAN_RCO_UDP) ?
		offsetof(struct udphdr, check) :
		offsetof(struct tcphdr, check);
}

static inline __be32 vxlan_compute_rco(unsigned int start, unsigned int offset)
{
	__be32 vni_field = cpu_to_be32(start >> VXLAN_RCO_SHIFT);

	if (offset == offsetof(struct udphdr, check))
		vni_field |= VXLAN_RCO_UDP;
	return vni_field;
}

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	4	#include <linux/if_vlan.h>
86a98057	5	#include <net/udp_tunnel.h>
ee122c79	6	#include <net/dst_metadata.h>
7c46a640	7	#include <net/udp_tunnel.h>
012a5729	8
828788ac JB	9	/* VXLAN protocol (RFC 7348) header:
	10	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	11	* \|R\|R\|R\|R\|I\|R\|R\|R\| Reserved \|
	12	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	13	* \| VXLAN Network Identifier (VNI) \| Reserved \|
	14	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	15	*
	16	* I = VXLAN Network Identifier (VNI) present.
	17	*/
	18	struct vxlanhdr {
	19	__be32 vx_flags;
	20	__be32 vx_vni;
	21	};
	22
	23	/* VXLAN header flags. */
54bfd872	24	#define VXLAN_HF_VNI cpu_to_be32(BIT(27))
828788ac JB	25
	26	#define VXLAN_N_VID (1u << 24)
	27	#define VXLAN_VID_MASK (VXLAN_N_VID - 1)
54bfd872	28	#define VXLAN_VNI_MASK cpu_to_be32(VXLAN_VID_MASK << 8)
828788ac JB	29	#define VXLAN_HLEN (sizeof(struct udphdr) + sizeof(struct vxlanhdr))
	30
	31	#define VNI_HASH_BITS 10
	32	#define VNI_HASH_SIZE (1<<VNI_HASH_BITS)
	33	#define FDB_HASH_BITS 8
	34	#define FDB_HASH_SIZE (1<<FDB_HASH_BITS)
	35
	36	/* Remote checksum offload for VXLAN (VXLAN_F_REMCSUM_[RT]X):
	37	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	38	* \|R\|R\|R\|R\|I\|R\|R\|R\|R\|R\|C\| Reserved \|
	39	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	40	* \| VXLAN Network Identifier (VNI) \|O\| Csum start \|
	41	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	42	*
	43	* C = Remote checksum offload bit. When set indicates that the
	44	* remote checksum offload data is present.
	45	*
	46	* O = Offset bit. Indicates the checksum offset relative to
	47	* checksum start.
	48	*
	49	* Csum start = Checksum start divided by two.
	50	*
	51	* http://tools.ietf.org/html/draft-herbert-vxlan-rco
	52	*/
	53
	54	/* VXLAN-RCO header flags. */
54bfd872	55	#define VXLAN_HF_RCO cpu_to_be32(BIT(21))
828788ac JB	56
828788ac JB	57	/* Remote checksum offload header option */
54bfd872 JB	58	#define VXLAN_RCO_MASK cpu_to_be32(0x7f) /* Last byte of vni field */
	59	#define VXLAN_RCO_UDP cpu_to_be32(0x80) /* Indicate UDP RCO (TCP when not set ) /
	60	#define VXLAN_RCO_SHIFT 1 /* Left shift of start */
828788ac	61	#define VXLAN_RCO_SHIFT_MASK ((1 << VXLAN_RCO_SHIFT) - 1)
54bfd872	62	#define VXLAN_MAX_REMCSUM_START (0x7f << VXLAN_RCO_SHIFT)
828788ac	63
3511494c	64	/*
828788ac	65	* VXLAN Group Based Policy Extension (VXLAN_F_GBP):
3511494c	66	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
828788ac	67	* \|G\|R\|R\|R\|I\|R\|R\|R\|R\|D\|R\|R\|A\|R\|R\|R\| Group Policy ID \|
3511494c TG	68	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	69	* \| VXLAN Network Identifier (VNI) \| Reserved \|
	70	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	71	*
828788ac JB	72	* G = Group Policy ID present.
828788ac JB	73	*
3511494c TG	74	* D = Don't Learn bit. When set, this bit indicates that the egress
	75	* VTEP MUST NOT learn the source address of the encapsulated frame.
	76	*
	77	* A = Indicates that the group policy has already been applied to
	78	* this packet. Policies MUST NOT be applied by devices when the
	79	* A bit is set.
	80	*
828788ac	81	* https://tools.ietf.org/html/draft-smith-vxlan-group-policy
3511494c TG	82	*/
3511494c TG	83	struct vxlanhdr_gbp {
0e715d6f	84	u8 vx_flags;
3511494c	85	#ifdef __LITTLE_ENDIAN_BITFIELD
0e715d6f	86	u8 reserved_flags1:3,
3511494c TG	87	policy_applied:1,
	88	reserved_flags2:2,
	89	dont_learn:1,
	90	reserved_flags3:1;
	91	#elif defined(__BIG_ENDIAN_BITFIELD)
0e715d6f	92	u8 reserved_flags1:1,
3511494c TG	93	dont_learn:1,
	94	reserved_flags2:2,
	95	policy_applied:1,
	96	reserved_flags3:3;
	97	#else
	98	#error "Please fix <asm/byteorder.h>"
	99	#endif
	100	__be16 policy_id;
	101	__be32 vx_vni;
	102	};
	103
828788ac	104	/* VXLAN-GBP header flags. */
54bfd872	105	#define VXLAN_HF_GBP cpu_to_be32(BIT(31))
828788ac	106
54bfd872	107	#define VXLAN_GBP_USED_BITS (VXLAN_HF_GBP \| cpu_to_be32(0xFFFFFF))
3511494c TG	108
	109	/* skb->mark mapping
	110	*
	111	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	112	* \|R\|R\|R\|R\|R\|R\|R\|R\|R\|D\|R\|R\|A\|R\|R\|R\| Group Policy ID \|
	113	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	114	*/
	115	#define VXLAN_GBP_DONT_LEARN (BIT(6) << 16)
	116	#define VXLAN_GBP_POLICY_APPLIED (BIT(3) << 16)
	117	#define VXLAN_GBP_ID_MASK (0xFFFF)
	118
e1e5314d JB	119	/*
	120	* VXLAN Generic Protocol Extension (VXLAN_F_GPE):
	121	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	122	* \|R\|R\|Ver\|I\|P\|R\|O\| Reserved \|Next Protocol \|
	123	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	124	* \| VXLAN Network Identifier (VNI) \| Reserved \|
	125	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	126	*
	127	* Ver = Version. Indicates VXLAN GPE protocol version.
	128	*
	129	* P = Next Protocol Bit. The P bit is set to indicate that the
	130	* Next Protocol field is present.
	131	*
	132	* O = OAM Flag Bit. The O bit is set to indicate that the packet
	133	* is an OAM packet.
	134	*
	135	* Next Protocol = This 8 bit field indicates the protocol header
	136	* immediately following the VXLAN GPE header.
	137	*
	138	* https://tools.ietf.org/html/draft-ietf-nvo3-vxlan-gpe-01
	139	*/
	140
	141	struct vxlanhdr_gpe {
	142	#if defined(__LITTLE_ENDIAN_BITFIELD)
	143	u8 oam_flag:1,
	144	reserved_flags1:1,
	145	np_applied:1,
	146	instance_applied:1,
	147	version:2,
	148	reserved_flags2:2;
	149	#elif defined(__BIG_ENDIAN_BITFIELD)
	150	u8 reserved_flags2:2,
	151	version:2,
	152	instance_applied:1,
	153	np_applied:1,
	154	reserved_flags1:1,
	155	oam_flag:1;
	156	#endif
	157	u8 reserved_flags3;
	158	u8 reserved_flags4;
	159	u8 next_protocol;
	160	__be32 vx_vni;
	161	};
	162
	163	/* VXLAN-GPE header flags. */
	164	#define VXLAN_HF_VER cpu_to_be32(BIT(29) \| BIT(28))
	165	#define VXLAN_HF_NP cpu_to_be32(BIT(26))
	166	#define VXLAN_HF_OAM cpu_to_be32(BIT(24))
	167
	168	#define VXLAN_GPE_USED_BITS (VXLAN_HF_VER \| VXLAN_HF_NP \| VXLAN_HF_OAM \| \
	169	cpu_to_be32(0xff))
	170
	171	/* VXLAN-GPE header Next Protocol. */
	172	#define VXLAN_GPE_NP_IPV4 0x01
	173	#define VXLAN_GPE_NP_IPV6 0x02
	174	#define VXLAN_GPE_NP_ETHERNET 0x03
	175	#define VXLAN_GPE_NP_NSH 0x04
	176
3511494c	177	struct vxlan_metadata {
3511494c TG	178	u32 gbp;
	179	};
	180
012a5729 PS	181	/* per UDP socket information */
	182	struct vxlan_sock {
	183	struct hlist_node hlist;
012a5729	184	struct socket *sock;
012a5729 PS	185	struct hlist_head vni_list[VNI_HASH_SIZE];
012a5729 PS	186	atomic_t refcnt;
dfd8645e	187	u32 flags;
012a5729 PS	188	};
012a5729 PS	189
0dfbdf41 TG	190	union vxlan_addr {
	191	struct sockaddr_in sin;
	192	struct sockaddr_in6 sin6;
	193	struct sockaddr sa;
	194	};
	195
	196	struct vxlan_rdst {
	197	union vxlan_addr remote_ip;
	198	__be16 remote_port;
54bfd872	199	__be32 remote_vni;
0dfbdf41 TG	200	u32 remote_ifindex;
	201	struct list_head list;
	202	struct rcu_head rcu;
0c1d70af	203	struct dst_cache dst_cache;
0dfbdf41 TG	204	};
	205
	206	struct vxlan_config {
	207	union vxlan_addr remote_ip;
	208	union vxlan_addr saddr;
54bfd872	209	__be32 vni;
0dfbdf41 TG	210	int remote_ifindex;
	211	int mtu;
	212	__be16 dst_port;
0e715d6f JB	213	u16 port_min;
	214	u16 port_max;
	215	u8 tos;
	216	u8 ttl;
e7f70af1	217	__be32 label;
0dfbdf41 TG	218	u32 flags;
	219	unsigned long age_interval;
	220	unsigned int addrmax;
	221	bool no_share;
	222	};
	223
	224	/* Pseudo network device */
	225	struct vxlan_dev {
	226	struct hlist_node hlist; /* vni hash table */
	227	struct list_head next; /* vxlan's per namespace list */
b1be00a6 JB	228	struct vxlan_sock vn4_sock; / listening socket for IPv4 */
	229	#if IS_ENABLED(CONFIG_IPV6)
	230	struct vxlan_sock vn6_sock; / listening socket for IPv6 */
	231	#endif
0dfbdf41 TG	232	struct net_device *dev;
	233	struct net net; / netns for packet i/o */
	234	struct vxlan_rdst default_dst; /* default destination */
	235	u32 flags; /* VXLAN_F_* in vxlan.h */
	236
	237	struct timer_list age_timer;
	238	spinlock_t hash_lock;
	239	unsigned int addrcnt;
58ce31cc	240	struct gro_cells gro_cells;
0dfbdf41 TG	241
	242	struct vxlan_config cfg;
	243
	244	struct hlist_head fdb_head[FDB_HASH_SIZE];
	245	};
	246
359a0ea9 TH	247	#define VXLAN_F_LEARN 0x01
	248	#define VXLAN_F_PROXY 0x02
	249	#define VXLAN_F_RSC 0x04
	250	#define VXLAN_F_L2MISS 0x08
	251	#define VXLAN_F_L3MISS 0x10
	252	#define VXLAN_F_IPV6 0x20
6ceb31ca	253	#define VXLAN_F_UDP_ZERO_CSUM_TX 0x40
359a0ea9 TH	254	#define VXLAN_F_UDP_ZERO_CSUM6_TX 0x80
359a0ea9 TH	255	#define VXLAN_F_UDP_ZERO_CSUM6_RX 0x100
dfd8645e TH	256	#define VXLAN_F_REMCSUM_TX 0x200
dfd8645e TH	257	#define VXLAN_F_REMCSUM_RX 0x400
3511494c	258	#define VXLAN_F_GBP 0x800
0ace2ca8	259	#define VXLAN_F_REMCSUM_NOPARTIAL 0x1000
ee122c79	260	#define VXLAN_F_COLLECT_METADATA 0x2000
e1e5314d	261	#define VXLAN_F_GPE 0x4000
359a0ea9	262
d299ce14	263	/* Flags that are used in the receive path. These flags must match in
af33c1ad TH	264	* order for a socket to be shareable
	265	*/
	266	#define VXLAN_F_RCV_FLAGS (VXLAN_F_GBP \| \
e1e5314d	267	VXLAN_F_GPE \| \
af33c1ad	268	VXLAN_F_UDP_ZERO_CSUM6_RX \| \
0ace2ca8	269	VXLAN_F_REMCSUM_RX \| \
ee122c79	270	VXLAN_F_REMCSUM_NOPARTIAL \| \
da8b43c0	271	VXLAN_F_COLLECT_METADATA)
ac5132d1	272
e1e5314d JB	273	/* Flags that can be set together with VXLAN_F_GPE. */
	274	#define VXLAN_F_ALLOWED_GPE (VXLAN_F_GPE \| \
	275	VXLAN_F_IPV6 \| \
	276	VXLAN_F_UDP_ZERO_CSUM_TX \| \
	277	VXLAN_F_UDP_ZERO_CSUM6_TX \| \
	278	VXLAN_F_UDP_ZERO_CSUM6_RX \| \
	279	VXLAN_F_COLLECT_METADATA)
	280
0dfbdf41 TG	281	struct net_device vxlan_dev_create(struct net net, const char *name,
	282	u8 name_assign_type, struct vxlan_config *conf);
	283
b1be00a6 JB	284	static inline __be16 vxlan_dev_dst_port(struct vxlan_dev *vxlan,
b1be00a6 JB	285	unsigned short family)
614732ea	286	{
b1be00a6 JB	287	#if IS_ENABLED(CONFIG_IPV6)
	288	if (family == AF_INET6)
	289	return inet_sk(vxlan->vn6_sock->sock->sk)->inet_sport;
	290	#endif
	291	return inet_sk(vxlan->vn4_sock->sock->sk)->inet_sport;
614732ea	292	}
49560532	293
5f35227e JG	294	static inline netdev_features_t vxlan_features_check(struct sk_buff *skb,
5f35227e JG	295	netdev_features_t features)
11bf7828	296	{
5f35227e JG	297	u8 l4_hdr = 0;
	298
	299	if (!skb->encapsulation)
	300	return features;
	301
	302	switch (vlan_get_protocol(skb)) {
	303	case htons(ETH_P_IP):
	304	l4_hdr = ip_hdr(skb)->protocol;
	305	break;
	306	case htons(ETH_P_IPV6):
	307	l4_hdr = ipv6_hdr(skb)->nexthdr;
	308	break;
	309	default:
	310	return features;;
	311	}
	312
	313	if ((l4_hdr == IPPROTO_UDP) &&
11bf7828 JS	314	(skb->inner_protocol_type != ENCAP_TYPE_ETHER \|\|
	315	skb->inner_protocol != htons(ETH_P_TEB) \|\|
	316	(skb_inner_mac_header(skb) - skb_transport_header(skb) !=
af67eb9e AD	317	sizeof(struct udphdr) + sizeof(struct vxlanhdr)) \|\|
	318	(skb->ip_summed != CHECKSUM_NONE &&
	319	!can_checksum_protocol(features, inner_eth_hdr(skb)->h_proto))))
a188222b	320	return features & ~(NETIF_F_CSUM_MASK \| NETIF_F_GSO_MASK);
11bf7828	321
5f35227e	322	return features;
11bf7828	323	}
23e62de3	324
e6cd988c JG	325	/* IP header + UDP + VXLAN + Ethernet header */
	326	#define VXLAN_HEADROOM (20 + 8 + 8 + 14)
	327	/* IPv6 header + UDP + VXLAN + Ethernet header */
	328	#define VXLAN6_HEADROOM (40 + 8 + 8 + 14)
	329
d4ac05ff JB	330	static inline struct vxlanhdr vxlan_hdr(struct sk_buff skb)
	331	{
	332	return (struct vxlanhdr *)(udp_hdr(skb) + 1);
	333	}
	334
54bfd872 JB	335	static inline __be32 vxlan_vni(__be32 vni_field)
	336	{
	337	#if defined(__BIG_ENDIAN)
5692d7ea	338	return (__force __be32)((__force u32)vni_field >> 8);
54bfd872	339	#else
5692d7ea	340	return (__force __be32)((__force u32)(vni_field & VXLAN_VNI_MASK) << 8);
54bfd872 JB	341	#endif
	342	}
	343
	344	static inline __be32 vxlan_vni_field(__be32 vni)
	345	{
	346	#if defined(__BIG_ENDIAN)
5692d7ea	347	return (__force __be32)((__force u32)vni << 8);
54bfd872	348	#else
5692d7ea	349	return (__force __be32)((__force u32)vni >> 8);
54bfd872 JB	350	#endif
	351	}
	352
54bfd872 JB	353	static inline size_t vxlan_rco_start(__be32 vni_field)
	354	{
	355	return be32_to_cpu(vni_field & VXLAN_RCO_MASK) << VXLAN_RCO_SHIFT;
	356	}
	357
	358	static inline size_t vxlan_rco_offset(__be32 vni_field)
	359	{
	360	return (vni_field & VXLAN_RCO_UDP) ?
	361	offsetof(struct udphdr, check) :
	362	offsetof(struct tcphdr, check);
	363	}
	364
	365	static inline __be32 vxlan_compute_rco(unsigned int start, unsigned int offset)
	366	{
	367	__be32 vni_field = cpu_to_be32(start >> VXLAN_RCO_SHIFT);
	368
	369	if (offset == offsetof(struct udphdr, check))
	370	vni_field \|= VXLAN_RCO_UDP;
	371	return vni_field;
	372	}
	373
705cc62f JB	374	static inline unsigned short vxlan_get_sk_family(struct vxlan_sock *vs)
	375	{
	376	return vs->sock->sk->sk_family;
	377	}
48e92c44 JB	378
48e92c44 JB	379	#endif