1mlx5dv_flow_action_esp(3) mlx5dv_flow_action_esp(3)
2
3
4
6 mlx5dv_flow_action_esp - Flow action esp for mlx5 provider
7
9 #include <infiniband/mlx5/mlx5dv.h>
10
11 struct ibv_flow_action *
12 mlx5dv_create_flow_action_esp(struct ibv_context *ctx,
13 struct ibv_flow_action_esp_attr *esp,
14 struct mlx5dv_flow_action_esp *mlx5_attr);
15
17 Create an IPSEC ESP flow steering action.
18 This verb is identical to ibv_create_flow_action_esp verb, but allows
19 mlx5 specific flags.
20
22 Please see ibv_flow_action_esp(3) man page for ctx and esp.
23
24 mlx5_attr argument
25 struct mlx5dv_flow_action_esp {
26 uint64_t comp_mask; /* Use enum mlx5dv_flow_action_esp_mask */
27 uint32_t action_flags; /* Use enum mlx5dv_flow_action_flags */
28 };
29
30 comp_mask
31 Bitmask specifying what fields in the structure are valid (enum
32 mlx5dv_flow_action_esp_mask).
33
34 action_flags
35 A bitwise OR of the various values described below.
36
37 MLX5DV_FLOW_ACTION_FLAGS_REQUIRE_METADATA:
38 Each received and transmitted packet using offload is expected
39 to carry metadata in the form of a L2 header
40 with ethernet type 0x8CE4, followed by 6 bytes of data and the
41 original packet ethertype.
42
44 The ESN is expected to be placed in the IV field for egress packets.
45 The 64 bit sequence number is written in big-endian over the 64 bit IV
46 field.
47 There is no need to call modify to update the ESN window on egress when
48 this DV is used.
49
51 ibv_flow_action_esp(3), RFC 4106
52
53
54
55 mlx5dv_flow_action_esp(3)