From: Vu Pham <vuhuong@mellanox.com>
Date: Mon, 2 Mar 2020 16:15:19 -0800
Subject: net/mlx5: Introduce egress acl forward-to-vport capability
Patch-mainline: v5.7-rc1
Git-commit: 86f5d0f3d4995b8b8909da7eea235e5bc1ceefbe
References: jsc#SLE-15172
Add HCA_CAP.egress_acl_forward_to_vport field to check whether HW
supports e-switch vport's egress acl to forward packets to other
e-switch vport or not.
By default E-Switch egress ACL forwards eswitch vports egress packets
to their corresponding NIC/VF vports.
With this cap enabled, the driver is allowed to alter this behavior
and forward packets to arbitrary NIC/VF vports with the following
limitations:
a. Multiple processing paths are supported if all of the following
conditions are met:
- HCA_CAP.egress_acl_forward_to_vport is set ==1.
- A destination of type Flow Table only appears once, as the
last destination in the list.
- Vport destination is supported if
HCA_CAP.egress_acl_forward_to_vport==1. Vport must not be
the Uplink.
b. Flow_tag not supported.
c. This table is only applicable after an FDB table is created.
d. Push VLAN action is not supported.
e. Pop VLAN action cannot be added concurrently to this table and
FDB table.
This feature will be used during port failover in bonding scenario
where two VFs representors are bonded to handle failover egress traffic
(VM's ingress/receive traffic).
Signed-off-by: Vu Pham <vuhuong@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Acked-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
include/linux/mlx5/mlx5_ifc.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- a/include/linux/mlx5/mlx5_ifc.h
+++ b/include/linux/mlx5/mlx5_ifc.h
@@ -741,7 +741,7 @@ struct mlx5_ifc_flow_table_eswitch_cap_b
u8 flow_source[0x1];
u8 reserved_at_18[0x2];
u8 multi_fdb_encap[0x1];
- u8 reserved_at_1b[0x1];
+ u8 egress_acl_forward_to_vport[0x1];
u8 fdb_multi_path_to_table[0x1];
u8 reserved_at_1d[0x3];