[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <4493521f-a241-ef1f-75d4-cd65e1576089@intel.com>
Date: Thu, 16 Feb 2023 02:34:37 -0800
From: "Nambiar, Amritha" <amritha.nambiar@...el.com>
To: Jakub Kicinski <kuba@...nel.org>,
Alexander H Duyck <alexander.duyck@...il.com>
CC: <netdev@...r.kernel.org>, <davem@...emloft.net>,
<edumazet@...gle.com>, <pabeni@...hat.com>,
Saeed Mahameed <saeed@...nel.org>,
"Samudrala, Sridhar" <sridhar.samudrala@...el.com>
Subject: Re: Kernel interface to configure queue-group parameters
On 2/8/2023 4:36 PM, Jakub Kicinski wrote:
> On Tue, 07 Feb 2023 08:28:56 -0800 Alexander H Duyck wrote:
>> I think much of this depends on exactly what functionality we are
>> talking about.
>
> Right, maybe we need to take a page out of the container's book and
> concede that best we can do is provide targeted APIs for slices of
> the problem. Which someone in user space would have to combine.
>
Agree, a common interface for various parameters for the queue-group
does not seem like a practical approach and the interface to use is
largely driven by the functionality itself.
>>> 4. Devlink:
>>> Pros:
>>> - New parameters can be added without any changes to the kernel or
>>> userspace.
>>>
>>> Cons:
>>> - Queue/Queue_group is a function-wide entity, Devlink is for
>>> device-wide stuff. Devlink being device centric is not suitable for
>>> queue parameters such as rates, NAPI etc.
>>
>> Yeah, I wouldn't expect something like this to be a good fit.
>
> Devlink has the hierarchical rate API for example.
> Maybe we should (re)consider adding top level nodes for RSS contexts
> there?
Powered by blists - more mailing lists