netdev - Re: [PATCH v4 2/2] virtio_net: Extend virtio to use VF datapath when available

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: <20180307043534-mutt-send-email-mst@kernel.org>
Date:   Wed, 7 Mar 2018 04:38:54 +0200
From:   "Michael S. Tsirkin" <mst@...hat.com>
To:     Alexander Duyck <alexander.duyck@...il.com>
Cc:     Jiri Pirko <jiri@...nulli.us>,
        Stephen Hemminger <stephen@...workplumber.org>,
        Sridhar Samudrala <sridhar.samudrala@...el.com>,
        David Miller <davem@...emloft.net>,
        Netdev <netdev@...r.kernel.org>, virtio-dev@...ts.oasis-open.org,
        "Brandeburg, Jesse" <jesse.brandeburg@...el.com>,
        "Duyck, Alexander H" <alexander.h.duyck@...el.com>,
        Jakub Kicinski <kubakici@...pl>
Subject: Re: [PATCH v4 2/2] virtio_net: Extend virtio to use VF datapath when
 available

On Tue, Mar 06, 2018 at 03:27:46PM -0800, Alexander Duyck wrote:
> > I definitelly vote for a separate common shared code for both netvsc and
> > virtio_net - even if you use 2 and 3 netdev model, you could share the
> > common code. Strict checks and limitation should be in place.
> 
> Noted. But as I also mentioned there isn't that much "common" code
> between the two models. I think if anything we could probably look at
> peeling out a few bits such as "get_<iface>_bymac" which really would
> become dev_get_by_mac_and_ops in order to find the device for the
> notifiers. I probably wouldn't even put that in our driver and would
> instead put it in the core code since it almost makes more sense
> there. Beyond that sharing becomes much more challenging due to the
> differences in the Rx and Tx paths that build out of the difference
> between the 2 driver and 3 driver models.

At this point it might be worth it to articulate the advantages
of the 3 netdev model.

If they are compelling, why wouldn't netvsc users want them?

Alex, I think you were one of the strongest proponents of this model,
you should be well placed to provide a summary.

-- 
MST