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