[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20230207215643.43f76bdd@kernel.org>
Date: Tue, 7 Feb 2023 21:56:43 -0800
From: Jakub Kicinski <kuba@...nel.org>
To: Tony Nguyen <anthony.l.nguyen@...el.com>
Cc: davem@...emloft.net, pabeni@...hat.com, edumazet@...gle.com,
Sebastian Czapla <sebastianx.czapla@...el.com>,
netdev@...r.kernel.org,
Grzegorz Szczurek <grzegorzx.szczurek@...el.com>,
Marek Szlosek <marek.szlosek@...el.com>
Subject: Re: [PATCH net-next 1/6] i40e: Add flag for disabling VF source
pruning
On Mon, 6 Feb 2023 15:56:30 -0800 Tony Nguyen wrote:
> Allow user to change source pruning for VF VSIs. This allows VFs to
> receive packets with MAC_SRC and MAC_DST equal to VFs mac.
>
> Added priv flag vf-source-pruning to allow user to change
> source pruning setting. Reset all VSIs to commit the setting.
> If vf-source-pruning is off and VF is trusted on with spoofchk off
> then disable source pruning on specific VF takes effect.
>
> Without this patch it is not possible to change source pruning
> setting on VF VSIs.
Intel keeps trickling in private flags to adjust the behavior of legacy
SR-IOV NDOs. No documentation, no well understood semantics.
This does not seem defensible upstream when we already started rejecting
any use of legacy SR-IOV NDOs in new drivers.
Quite sad this is bundled in the same series as the much appreciated
removal of magic status codes.
Powered by blists - more mailing lists