netdev - Re: [Intel-wired-lan] [PATCH iwl-next v4 4/5] ice: Add tx_scheduling

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: <20240222150717.627209a9@kernel.org>
Date: Thu, 22 Feb 2024 15:07:17 -0800
From: Jakub Kicinski <kuba@...nel.org>
To: Jiri Pirko <jiri@...nulli.us>
Cc: Mateusz Polchlopek <mateusz.polchlopek@...el.com>,
 <intel-wired-lan@...ts.osuosl.org>, <netdev@...r.kernel.org>,
 <horms@...nel.org>, <przemyslaw.kitszel@...el.com>, Lukasz Czapnik
 <lukasz.czapnik@...el.com>
Subject: Re: [Intel-wired-lan] [PATCH iwl-next v4 4/5] ice: Add
 tx_scheduling_layers devlink param

On Thu, 22 Feb 2024 14:25:21 +0100 Mateusz Polchlopek wrote:
> >> This is kind of proprietary param similar to number of which were shot
> >> down for mlx5 in past. Jakub?  
> > 
> > I remain somewhat confused about what this does.
> > Specifically IIUC the problem is that the radix of each node is
> > limited, so we need to start creating multi-layer hierarchies
> > if we want a higher radix. Or in the "5-layer mode" the radix
> > is automatically higher?  
> 
> Basically, switching from 9 to 5 layers topology allows us to have 512 
> leaves instead of 8 leaves which improves performance. I will add this 
> information to the commit message and Documentation too, when we get an 
> ACK for devlink parameter.

Sounds fine. Please update the doc to focus on the radix, rather than
the layers. Layers are not so important to the user. And maybe give an
example of things which won't be possible with 5-layer config.

Jiri, I'm not aware of any other devices with this sort of trade off.
We shouldn't add the param if either:
 - this can be changed dynamically as user instantiates rate limiters;
 - we know other devices have similar needs.
If neither of those is true, param seems fine to me..