[mirror_ubuntu-bionic-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>

#define VNI_HASH_BITS	10
#define VNI_HASH_SIZE	(1<<VNI_HASH_BITS)

/*
 * 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(24)
#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))

struct vxlan_metadata {
	__be32		vni;
	u32		gbp;

	/* Temporary until vxlan_rcv() API is gone */
	struct metadata_dst *tun_dst;
};

struct vxlan_sock;
typedef void (vxlan_rcv_t)(struct vxlan_sock *vh, struct sk_buff *skb,
			   struct vxlan_metadata *md);

/* per UDP socket information */
struct vxlan_sock {
	struct hlist_node hlist;
	vxlan_rcv_t	 *rcv;
	void		 *data;
	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;
};

#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
#define VXLAN_F_FLOW_BASED		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_UDP_ZERO_CSUM6_RX |	\
					 VXLAN_F_REMCSUM_RX |		\
					 VXLAN_F_REMCSUM_NOPARTIAL |	\
					 VXLAN_F_COLLECT_METADATA |	\
					 VXLAN_F_FLOW_BASED)

struct vxlan_sock *vxlan_sock_add(struct net *net, __be16 port,
				  vxlan_rcv_t *rcv, void *data,
				  bool no_share, u32 flags);

void vxlan_sock_release(struct vxlan_sock *vs);

int vxlan_xmit_skb(struct rtable *rt, struct sock *sk, struct sk_buff *skb,
		   __be32 src, __be32 dst, __u8 tos, __u8 ttl, __be16 df,
		   __be16 src_port, __be16 dst_port, struct vxlan_metadata *md,
		   bool xnet, u32 vxflags);

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_ALL_CSUM | 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
#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 PS	11
	12	#define VNI_HASH_BITS 10
	13	#define VNI_HASH_SIZE (1<<VNI_HASH_BITS)
	14
3511494c TG	15	/*
	16	* VXLAN Group Based Policy Extension:
	17	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	18	* \|1\|-\|-\|-\|1\|-\|-\|-\|R\|D\|R\|R\|A\|R\|R\|R\| Group Policy ID \|
	19	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	20	* \| VXLAN Network Identifier (VNI) \| Reserved \|
	21	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	22	*
	23	* D = Don't Learn bit. When set, this bit indicates that the egress
	24	* VTEP MUST NOT learn the source address of the encapsulated frame.
	25	*
	26	* A = Indicates that the group policy has already been applied to
	27	* this packet. Policies MUST NOT be applied by devices when the
	28	* A bit is set.
	29	*
	30	* [0] https://tools.ietf.org/html/draft-smith-vxlan-group-policy
	31	*/
	32	struct vxlanhdr_gbp {
	33	__u8 vx_flags;
	34	#ifdef __LITTLE_ENDIAN_BITFIELD
	35	__u8 reserved_flags1:3,
	36	policy_applied:1,
	37	reserved_flags2:2,
	38	dont_learn:1,
	39	reserved_flags3:1;
	40	#elif defined(__BIG_ENDIAN_BITFIELD)
	41	__u8 reserved_flags1:1,
	42	dont_learn:1,
	43	reserved_flags2:2,
	44	policy_applied:1,
	45	reserved_flags3:3;
	46	#else
	47	#error "Please fix <asm/byteorder.h>"
	48	#endif
	49	__be16 policy_id;
	50	__be32 vx_vni;
	51	};
	52
	53	#define VXLAN_GBP_USED_BITS (VXLAN_HF_GBP \| 0xFFFFFF)
	54
	55	/* skb->mark mapping
	56	*
	57	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	58	* \|R\|R\|R\|R\|R\|R\|R\|R\|R\|D\|R\|R\|A\|R\|R\|R\| Group Policy ID \|
	59	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	60	*/
	61	#define VXLAN_GBP_DONT_LEARN (BIT(6) << 16)
	62	#define VXLAN_GBP_POLICY_APPLIED (BIT(3) << 16)
	63	#define VXLAN_GBP_ID_MASK (0xFFFF)
	64
	65	/* VXLAN protocol header:
	66	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	67	* \|G\|R\|R\|R\|I\|R\|R\|C\| Reserved \|
	68	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	69	* \| VXLAN Network Identifier (VNI) \| Reserved \|
	70	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	71	*
	72	* G = 1 Group Policy (VXLAN-GBP)
	73	* I = 1 VXLAN Network Identifier (VNI) present
	74	* C = 1 Remote checksum offload (RCO)
	75	*/
11bf7828 JS	76	struct vxlanhdr {
	77	__be32 vx_flags;
	78	__be32 vx_vni;
	79	};
	80
3bf39475	81	/* VXLAN header flags. */
3511494c TG	82	#define VXLAN_HF_RCO BIT(24)
	83	#define VXLAN_HF_VNI BIT(27)
	84	#define VXLAN_HF_GBP BIT(31)
dfd8645e TH	85
	86	/* Remote checksum offload header option */
	87	#define VXLAN_RCO_MASK 0x7f /* Last byte of vni field */
	88	#define VXLAN_RCO_UDP 0x80 /* Indicate UDP RCO (TCP when not set ) /
	89	#define VXLAN_RCO_SHIFT 1 /* Left shift of start */
	90	#define VXLAN_RCO_SHIFT_MASK ((1 << VXLAN_RCO_SHIFT) - 1)
	91	#define VXLAN_MAX_REMCSUM_START (VXLAN_RCO_MASK << VXLAN_RCO_SHIFT)
3bf39475 TH	92
	93	#define VXLAN_N_VID (1u << 24)
	94	#define VXLAN_VID_MASK (VXLAN_N_VID - 1)
40fb70f3	95	#define VXLAN_VNI_MASK (VXLAN_VID_MASK << 8)
3bf39475 TH	96	#define VXLAN_HLEN (sizeof(struct udphdr) + sizeof(struct vxlanhdr))
3bf39475 TH	97
3511494c TG	98	struct vxlan_metadata {
	99	__be32 vni;
	100	u32 gbp;
ee122c79 TG	101
	102	/* Temporary until vxlan_rcv() API is gone */
	103	struct metadata_dst *tun_dst;
3511494c TG	104	};
3511494c TG	105
012a5729	106	struct vxlan_sock;
3511494c TG	107	typedef void (vxlan_rcv_t)(struct vxlan_sock vh, struct sk_buff skb,
3511494c TG	108	struct vxlan_metadata *md);
012a5729 PS	109
	110	/* per UDP socket information */
	111	struct vxlan_sock {
	112	struct hlist_node hlist;
	113	vxlan_rcv_t *rcv;
	114	void *data;
	115	struct work_struct del_work;
	116	struct socket *sock;
	117	struct rcu_head rcu;
	118	struct hlist_head vni_list[VNI_HASH_SIZE];
	119	atomic_t refcnt;
dc01e7d3	120	struct udp_offload udp_offloads;
dfd8645e	121	u32 flags;
012a5729 PS	122	};
012a5729 PS	123
359a0ea9 TH	124	#define VXLAN_F_LEARN 0x01
	125	#define VXLAN_F_PROXY 0x02
	126	#define VXLAN_F_RSC 0x04
	127	#define VXLAN_F_L2MISS 0x08
	128	#define VXLAN_F_L3MISS 0x10
	129	#define VXLAN_F_IPV6 0x20
	130	#define VXLAN_F_UDP_CSUM 0x40
	131	#define VXLAN_F_UDP_ZERO_CSUM6_TX 0x80
	132	#define VXLAN_F_UDP_ZERO_CSUM6_RX 0x100
dfd8645e TH	133	#define VXLAN_F_REMCSUM_TX 0x200
dfd8645e TH	134	#define VXLAN_F_REMCSUM_RX 0x400
3511494c	135	#define VXLAN_F_GBP 0x800
0ace2ca8	136	#define VXLAN_F_REMCSUM_NOPARTIAL 0x1000
ee122c79 TG	137	#define VXLAN_F_COLLECT_METADATA 0x2000
ee122c79 TG	138	#define VXLAN_F_FLOW_BASED 0x4000
359a0ea9	139
d299ce14	140	/* Flags that are used in the receive path. These flags must match in
af33c1ad TH	141	* order for a socket to be shareable
	142	*/
	143	#define VXLAN_F_RCV_FLAGS (VXLAN_F_GBP \| \
	144	VXLAN_F_UDP_ZERO_CSUM6_RX \| \
0ace2ca8	145	VXLAN_F_REMCSUM_RX \| \
ee122c79 TG	146	VXLAN_F_REMCSUM_NOPARTIAL \| \
	147	VXLAN_F_COLLECT_METADATA \| \
	148	VXLAN_F_FLOW_BASED)
ac5132d1	149
012a5729 PS	150	struct vxlan_sock vxlan_sock_add(struct net net, __be16 port,
012a5729 PS	151	vxlan_rcv_t rcv, void data,
359a0ea9	152	bool no_share, u32 flags);
012a5729 PS	153
012a5729 PS	154	void vxlan_sock_release(struct vxlan_sock *vs);
49560532	155
79b16aad	156	int vxlan_xmit_skb(struct rtable rt, struct sock sk, struct sk_buff *skb,
49560532	157	__be32 src, __be32 dst, __u8 tos, __u8 ttl, __be16 df,
3511494c	158	__be16 src_port, __be16 dst_port, struct vxlan_metadata *md,
af33c1ad	159	bool xnet, u32 vxflags);
49560532	160
5f35227e JG	161	static inline netdev_features_t vxlan_features_check(struct sk_buff *skb,
5f35227e JG	162	netdev_features_t features)
11bf7828	163	{
5f35227e JG	164	u8 l4_hdr = 0;
	165
	166	if (!skb->encapsulation)
	167	return features;
	168
	169	switch (vlan_get_protocol(skb)) {
	170	case htons(ETH_P_IP):
	171	l4_hdr = ip_hdr(skb)->protocol;
	172	break;
	173	case htons(ETH_P_IPV6):
	174	l4_hdr = ipv6_hdr(skb)->nexthdr;
	175	break;
	176	default:
	177	return features;;
	178	}
	179
	180	if ((l4_hdr == IPPROTO_UDP) &&
11bf7828 JS	181	(skb->inner_protocol_type != ENCAP_TYPE_ETHER \|\|
	182	skb->inner_protocol != htons(ETH_P_TEB) \|\|
	183	(skb_inner_mac_header(skb) - skb_transport_header(skb) !=
	184	sizeof(struct udphdr) + sizeof(struct vxlanhdr))))
5f35227e	185	return features & ~(NETIF_F_ALL_CSUM \| NETIF_F_GSO_MASK);
11bf7828	186
5f35227e	187	return features;
11bf7828	188	}
23e62de3	189
e6cd988c JG	190	/* IP header + UDP + VXLAN + Ethernet header */
	191	#define VXLAN_HEADROOM (20 + 8 + 8 + 14)
	192	/* IPv6 header + UDP + VXLAN + Ethernet header */
	193	#define VXLAN6_HEADROOM (40 + 8 + 8 + 14)
	194
	195	#if IS_ENABLED(CONFIG_VXLAN)
53cf5275	196	void vxlan_get_rx_port(struct net_device *netdev);
e6cd988c JG	197	#else
	198	static inline void vxlan_get_rx_port(struct net_device *netdev)
	199	{
	200	}
	201	#endif
012a5729	202	#endif