netdev - Re: [net-next PATCH 8/8] fbnic: Add phydev representing PMD to phylink setup

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] [day] [month] [year] [list]

Message-ID: <CAKgT0Uf6my+3EsaJdNZi-m7EGYo0izvO_TmfVbX_5taEkw-MCg@mail.gmail.com>
Date: Tue, 28 Oct 2025 16:07:50 -0700
From: Alexander Duyck <alexander.duyck@...il.com>
To: Andrew Lunn <andrew@...n.ch>
Cc: netdev@...r.kernel.org, kuba@...nel.org, kernel-team@...a.com, 
	andrew+netdev@...n.ch, hkallweit1@...il.com, linux@...linux.org.uk, 
	pabeni@...hat.com, davem@...emloft.net
Subject: Re: [net-next PATCH 8/8] fbnic: Add phydev representing PMD to
 phylink setup

On Tue, Oct 28, 2025 at 3:56 PM Andrew Lunn <andrew@...n.ch> wrote:
>
> > > > +     phylink_resume(fbn->phylink);
> > >
> > > When was is suspended?
> >
> > We don't use the start/stop calls. Instead we use the resume/suspend
> > calls in order to deal with the fact that we normally aren't fully
> > resetting the link. The first call automatically gets converted to a
> > phylink_start because the bit isn't set for the MAC_WOL, however all
> > subsequent setups it becomes a resume so that we aren't tearing the
> > link down fully in order to avoid blocking the BMC which is sharing
> > the link similar to how a WOL connection would.
>
> /**
>  * phylink_resume() - handle a network device resume event
>  * @pl: a pointer to a &struct phylink returned from phylink_create()
>  *
>  * Undo the effects of phylink_suspend(), returning the link to an
>  * operational state.
>  */
>
> There needs to be a call to phylink_suspend() before you call
> phylink_resume(). If there is a prior call to phylink_suspend() all is
> O.K.
>
> Russell gets unhappy if you don't follow the documentation. The
> documentation is part of the API, part of the contract. If you break
> the contract, don't be surprised is your driver breaks sometime in the
> future.

We have had some back and forth in the past about it. Basically we
need to work through and come up with a better way of handling the BMC
portion of this. As it stands there are still some side effects but it
can't really be helped as the resume logic will force the MAC down
briefly.

I am still also using wol_enabled and haven't come up with a good
alternative yet to say that a BMC is present on the link.