netdev - Re: [RFC net-next 1/3] netdev: add per-queue statistics

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <de03710a-8409-49c6-bc62-c49e8291cb73@intel.com>
Date: Fri, 23 Feb 2024 12:51:51 -0800
From: "Nambiar, Amritha" <amritha.nambiar@...el.com>
To: Jakub Kicinski <kuba@...nel.org>
CC: <davem@...emloft.net>, <netdev@...r.kernel.org>, <edumazet@...gle.com>,
	<pabeni@...hat.com>, <danielj@...dia.com>, <mst@...hat.com>,
	<michael.chan@...adcom.com>
Subject: Re: [RFC net-next 1/3] netdev: add per-queue statistics

On 2/22/2024 5:44 PM, Jakub Kicinski wrote:
> On Thu, 22 Feb 2024 16:29:08 -0800 Nambiar, Amritha wrote:
>> Thanks, this almost has all the bits to also lookup stats for a single
>> queue with --do stats-get with a queue id and type.
> 
> We could without the projection. The projection (BTW not a great name,
> couldn't come up with a better one.. split? dis-aggregation? view?
> un-grouping?) "splits" a single object (netdev stats) across components
> (queues). I was wondering if at some point we may add another
> projection, splitting a queue. And then a queue+id+projection would
> actually have to return multiple objects. So maybe it's more consistent
> to just not support do at all for this op, and only support dump?
> 

So I understand splitting a netdev object into component queues, but do 
you have anything in mind WRT to splitting a queue, what could be the 
components for a queue object?
Agree that we can avoid the 'do' support if there are multiple 
possibilities for the projection/scope/view.
"scope" or "view" LGTM.

> We can support filtered dump on ifindex + queue id + type, and expect
> it to return one object for now.
> 
Sounds good if we are doing away with the 'do' support.

> Not 100% sure so I went with the "keep it simple, we can add more later"
> approach.
>