d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

mlx5: Implement offloads flowtable namespace.

Browse Source 
This namespace will be used for TCP offloads, like hardware decryption
of TLS TCP data.

MFC after:	1 week
Sponsored by:	NVIDIA Networking
This commit is contained in:
Hans Petter Selasky 2022-02-01 16:20:13 +01:00
parent e059c120b4
commit 0ee1b09eaa
2 changed files with 25 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
sys/dev/mlx5/fs.h
View File

@ -54,6 +54,7 @@ enum {
enum mlx5_flow_namespace_type {
MLX5_FLOW_NAMESPACE_BYPASS,
MLX5_FLOW_NAMESPACE_OFFLOADS,
MLX5_FLOW_NAMESPACE_KERNEL,
MLX5_FLOW_NAMESPACE_LEFTOVERS,
MLX5_FLOW_NAMESPACE_SNIFFER_RX,

35
sys/dev/mlx5/mlx5_core/mlx5_fs_tree.c
View File

@ -1,5 +1,5 @@
/*-
* Copyright (c) 2013-2017, Mellanox Technologies, Ltd. All rights reserved.
* Copyright (c) 2013-2021, Mellanox Technologies, Ltd. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
@ -67,13 +67,19 @@
#define FS_REQUIRED_CAPS(...) {.arr_sz = INIT_CAPS_ARRAY_SIZE(__VA_ARGS__), \
.caps = (long[]) {__VA_ARGS__}}
#define BYPASS_MAX_FT 5
#define BYPASS_PRIO_MAX_FT 1
#define KERNEL_MAX_FT 5
#define LEFTOVER_MAX_FT 1
#define KERNEL_MIN_LEVEL 3
#define LEFTOVER_MIN_LEVEL KERNEL_MIN_LEVEL + 1
#define BYPASS_MIN_LEVEL MLX5_NUM_BYPASS_FTS + LEFTOVER_MIN_LEVEL
/* Flowtable sizes: */
#define BYPASS_MAX_FT 5
#define BYPASS_PRIO_MAX_FT 1
#define OFFLOADS_MAX_FT 2
#define KERNEL_MAX_FT 5
#define LEFTOVER_MAX_FT 1
/* Flowtable levels: */
#define OFFLOADS_MIN_LEVEL 3
#define KERNEL_MIN_LEVEL (OFFLOADS_MIN_LEVEL + 1)
#define LEFTOVER_MIN_LEVEL (KERNEL_MIN_LEVEL + 1)
#define BYPASS_MIN_LEVEL (MLX5_NUM_BYPASS_FTS + LEFTOVER_MIN_LEVEL)
struct node_caps {
size_t arr_sz;
long *caps;
@ -92,7 +98,7 @@ struct init_tree_node {
} root_fs = {
.type = FS_TYPE_NAMESPACE,
.name = "root",
.ar_size = 3,
.ar_size = 4,
.children = (struct init_tree_node[]) {
ADD_PRIO("by_pass_prio", 0, BYPASS_MIN_LEVEL, 0,
FS_REQUIRED_CAPS(FS_CAP(flow_table_properties_nic_receive.flow_modify_en),
@ -116,6 +122,10 @@ struct init_tree_node {
BYPASS_PRIO_MAX_FT),
ADD_FT_PRIO("prio-mcast", 0,
BYPASS_PRIO_MAX_FT))),
ADD_PRIO("offloads_prio", 0, OFFLOADS_MIN_LEVEL, 0, {},
ADD_NS("offloads_ns",
ADD_FT_PRIO("prio_offloads-0", 0,
OFFLOADS_MAX_FT))),
ADD_PRIO("kernel_prio", 0, KERNEL_MIN_LEVEL, 0, {},
ADD_NS("kernel_ns",
ADD_FT_PRIO("prio_kernel-0", 0,
@ -2375,12 +2385,15 @@ struct mlx5_flow_namespace *mlx5_get_flow_namespace(struct mlx5_core_dev *dev,
case MLX5_FLOW_NAMESPACE_BYPASS:
prio = 0;
break;
case MLX5_FLOW_NAMESPACE_KERNEL:
case MLX5_FLOW_NAMESPACE_OFFLOADS:
prio = 1;
break;
case MLX5_FLOW_NAMESPACE_LEFTOVERS:
case MLX5_FLOW_NAMESPACE_KERNEL:
prio = 2;
break;
case MLX5_FLOW_NAMESPACE_LEFTOVERS:
prio = 3;
break;
case MLX5_FLOW_NAMESPACE_FDB:
if (dev->fdb_root_ns)
return &dev->fdb_root_ns->ns;