netdev - Re: mv88e6240 configuration broken for B850v3

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: <Ya6LAmB3GswfYqB7@shell.armlinux.org.uk>
Date:   Mon, 6 Dec 2021 22:13:22 +0000
From:   "Russell King (Oracle)" <linux@...linux.org.uk>
To:     Vladimir Oltean <olteanv@...il.com>
Cc:     Andrew Lunn <andrew@...n.ch>,
        Martyn Welch <martyn.welch@...labora.com>,
        Vivien Didelot <vivien.didelot@...il.com>,
        Florian Fainelli <f.fainelli@...il.com>,
        netdev@...r.kernel.org, kernel@...labora.com
Subject: Re: mv88e6240 configuration broken for B850v3

On Mon, Dec 06, 2021 at 11:44:28PM +0200, Vladimir Oltean wrote:
> On Mon, Dec 06, 2021 at 08:18:30PM +0000, Russell King (Oracle) wrote:
> > > If we're going to impersonate phylink we could at least provide the same
> > > arguments as phylink will.
> > 
> > What is going on here in terms of impersonation is entirely reasonable.
> > 
> > The only things in this respect that phylink guarantees are:
> > 
> > 1) The MAC/PCS configuration will not be substantially reconfigured
> >    unless a call to mac_link_down() was made if a call to mac_link_up()
> >    was previously made.
> 
> The wording here is unclear. Did you mean "When the MAC/PCS configuration
> is substantially reconfigured and the last call was a mac_link_up(), a
> follow-up call to mac_link_down() will also be made"?
> And what do you mean by "substantially reconfigured"?

I mean what the documentation refers to as a "full initialisation" of
the link, which happens if we have to change the interface mode or
MLO_AN mode. Only minor changes (advertisements or pause modes if under
manual control) will be changed without a prior call to mac_link_down().

For example, phylink will _never_ do:

mac_link_up()
mac_prepare()
mac_config()
pcs_config()
mac_finish()
mac_link_up()

However, with legacy (pre-March 2020) users, it _may_ do:

mac_link_up()
mac_config() (e.g. changing the in-band advertisement)
mac_an_restart()

The problem here is the legacy stuff which clouds the picture by making
extra calls to mac_config() when we're only changing things like the
in-band advert.

> phylink_major_config called from the paths that aren't phylink_mac_initial_config
> (because that happens with no preceding call to either mac_link_down or
> mac_link_up), right?

Where we call phylink_major_config(), we ensure that if we know the link
was previously up, we make a call to mac_link_down() first.

> > 2) The arguments to mac_link_down() will be the same as the preceeding
> >    mac_link_up() call - in other words, the "mode" and "interface".
> 
> Does this imply that "there will always be a preceding mac_link_up to
> every mac_link_down call"?

>From the design of phylink, yes it does, since phylink assumes that the
link is initially down.

> Because if it does imply that, DSA violates it.

Yes, DSA violates that, but DSA is free to do that if it makes sense,
and from what I understand of DSA, this is a special case. From what
I've seen with Marvell DSA, they start off auto-configuring the
inter-switch and CPU ports in link-up mode, whereas phylink assumes
that the link is down.

When DSA sets stuff up and brings up the CPU port, the very first
thing that phylink does is reconfigure the port - but the port is in
link-up mode, and that can mess up the port. So Andrew introduced
this to fix it. See commit 3be98b2d5fbc.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 40Mbps down 10Mbps up. Decent connectivity at last!