[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <fd36f18360b2800b37fe6b7466b7361afd43718b.camel@mellanox.com>
Date: Fri, 3 Apr 2020 02:59:57 +0000
From: Saeed Mahameed <saeedm@...lanox.com>
To: "wenxu@...oud.cn" <wenxu@...oud.cn>,
Paul Blakey <paulb@...lanox.com>,
Vlad Buslov <vladbu@...lanox.com>,
"pablo@...filter.org" <pablo@...filter.org>,
Oz Shlomo <ozsh@...lanox.com>
CC: "netdev@...r.kernel.org" <netdev@...r.kernel.org>
Subject: Re: [PATCH net-next] net/mlx5e: avoid check the hw_stats of
flow_action for FT flow
On Sun, 2020-03-29 at 14:56 +0800, wenxu@...oud.cn wrote:
> From: wenxu <wenxu@...oud.cn>
>
> The hw_stats in flow_action can't be supported in nftable
> flowtables. This check will lead the nft flowtable offload
> failed. So don't check the hw_stats of flow_action for FT
> flow.
>
This looks like a work around not a solution .. if the user requested a
hw_stats action that the hw can't support, no matter what the request
is, we should fail it even if it was for ft offloads.
if it is not support by nftable, then the caller shouldn't ask for
hw_stats action in first place.
> Signed-off-by: wenxu <wenxu@...oud.cn>
> ---
> drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
> b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
> index 901b5fa..4666015 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
> @@ -3703,7 +3703,8 @@ static int parse_tc_fdb_actions(struct
> mlx5e_priv *priv,
> if (!flow_action_has_entries(flow_action))
> return -EINVAL;
>
> - if (!flow_action_hw_stats_check(flow_action, extack,
> + if (!ft_flow &&
> + !flow_action_hw_stats_check(flow_action, extack,
> FLOW_ACTION_HW_STATS_DELAYED_BI
> T))
> return -EOPNOTSUPP;
>
Powered by blists - more mailing lists