[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20240711161615.123a5008@kernel.org>
Date: Thu, 11 Jul 2024 16:16:15 -0700
From: Jakub Kicinski <kuba@...nel.org>
To: Saeed Mahameed <saeed@...nel.org>
Cc: davem@...emloft.net, netdev@...r.kernel.org, edumazet@...gle.com,
pabeni@...hat.com, tariqt@...dia.com, rrameshbabu@...dia.com,
saeedm@...dia.com, yuehaibing@...wei.com, horms@...nel.org,
jacob.e.keller@...el.com, afaris@...dia.com
Subject: Re: [PATCH net-next v2] eth: mlx5: expose NETIF_F_NTUPLE when ARFS
is compiled out
On Thu, 11 Jul 2024 16:05:29 -0700 Saeed Mahameed wrote:
> >+#if IS_ENABLED(CONFIG_MLX5_EN_ARFS)
> > netdev->hw_features |= NETIF_F_NTUPLE;
> >+#elif IS_ENABLED(CONFIG_MLX5_EN_RXNFC)
> >+ netdev->features |= NETIF_F_NTUPLE;
>
> Why default ON when RXNFC and OFF when ARFS ?
> Default should be off always, and this needs to be advertised in
> hw_features in both cases.
>
> I think this should be
> #if IS_ENABLED(ARFS) || IS_ENABLED(RXFNC)
> netdev->hw_features |= NTUPLE;
That's what I thought, but on reflection since the filters can be
added, and disable doesn't actually do anything - "fixed on" started
to sound more appropriate. The additional "[fixed]" could also be useful
for troubleshooting, to signal that this is a different situation than
ARFS=y. No hard preference tho.
> Otherwise LGTM
>
> Acked-by: Saeed Mahameed <saeedm@...dia.com>
Thanks!
Powered by blists - more mailing lists