[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20220706125616.0a853dfc@kernel.org>
Date: Wed, 6 Jul 2022 12:56:16 -0700
From: Jakub Kicinski <kuba@...nel.org>
To: "Wilczynski, Michal" <michal.wilczynski@...el.com>
Cc: <netdev@...r.kernel.org>, Dima Chumak <dchumak@...dia.com>,
"Maxim Mikityanskiy" <maximmi@...dia.com>,
"Knitter, Konrad" <konrad.knitter@...el.com>,
Jiri Pirko <jiri@...nulli.us>,
Simon Horman <simon.horman@...igine.com>
Subject: Re: [RFC] ice: Reconfigure tx scheduling for SR-IOV
Reminder: please don't top post on the Linux lists.
On Wed, 6 Jul 2022 12:54:12 +0200 Wilczynski, Michal wrote:
> Hi,
>
> Thank you for your e-mail.
>
> I considered using devlink-rate, and it seems like a good fit. However
> we would also
>
> need support for rate-limiting for individual queues on the VF.
> Currently we have
>
> two types of rate objects in devlink-rate: leaf and node. Would adding a
> third one - queue be accepted ?
Something along those lines. IIUC htb offload as admission control for
VF representors is not a thing today, so since devlink rate exists the
lowest amount of duplication would be teaching it about queues.
> Also we might want to add some other object rate parameters to currently
> existing ones, for example 'priority'.
Presumably you can't admission control at a granularity higher than
a queue, so grouping queues should cover all use cases.
> If this sounds acceptable I will work on the patch and submit it as
> soon, as it's ready.
I'd be curious to hear from nVidia and Corigine folks as well.
We can revive the switchdev call if talking over VC helps with
the alignment between vendors.
Powered by blists - more mailing lists