netdev - Re: [RFC net-next 0/5] Add ETS and TBF Qdisc offload for Airoha EN7581 SoC

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: <f8e74e29-f4b0-4e38-8701-a4364d68230f@lunn.ch>
Date: Tue, 17 Dec 2024 00:24:13 +0100
From: Andrew Lunn <andrew@...n.ch>
To: Lorenzo Bianconi <lorenzo@...nel.org>
Cc: Vladimir Oltean <olteanv@...il.com>,
	Oleksij Rempel <linux@...pel-privat.de>, netdev@...r.kernel.org,
	davem@...emloft.net, edumazet@...gle.com, kuba@...nel.org,
	pabeni@...hat.com, horms@...nel.org, nbd@....name,
	sean.wang@...iatek.com, Mark-MC.Lee@...iatek.com,
	lorenzo.bianconi83@...il.com
Subject: Re: [RFC net-next 0/5] Add ETS and TBF Qdisc offload for Airoha
 EN7581 SoC

> Considering patch [0], we are still offloading the Qdisc on the provided
> DSA switch port (e.g. LANx) via the port_setup_tc() callback available in
> dsa_user_setup_qdisc(), but we are introducing even the ndo_setup_tc_conduit()
> callback in order to use the hw Qdisc capabilities available on the mac chip
> (e.g. EN7581) for the routed traffic from WAN to LANx. We will still apply
> the Qdisc defined on LANx for L2 traffic from LANy to LANx. Agree?

I've not read all the details, so i could be getting something
wrong. But let me point out the basics. Offloading is used to
accelerate what Linux already supports in software. So forget about
your hardware. How would i configure a bunch of e1000e cards connected
to a software bridge to do what you want?

There is no conduit interface in this, so i would not expect to
explicitly configure a conduit interface. Maybe the offloading needs
to implicitly configure the conduit, but that should be all hidden
away from the user. But given the software bridge has no concept of a
conduit, i doubt it.

It could well be our model does not map to the hardware too well,
leaving some bits unusable, but there is not much you can do about
that, that is the Linux model, accelerate what Linux supports in
software.

	Andrew