netdev - Re: [PATCH net-next 1/2] net: phy: don't stop state machine in case of MDIO error

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: <4dfc05d9-9f4f-101f-e026-b57b0c194e1d@gmail.com>
Date:   Sun, 16 Dec 2018 10:40:54 +0100
From:   Heiner Kallweit <hkallweit1@...il.com>
To:     Andrew Lunn <andrew@...n.ch>
Cc:     Florian Fainelli <f.fainelli@...il.com>,
        David Miller <davem@...emloft.net>,
        Fugang Duan <fugang.duan@....com>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>
Subject: Re: [PATCH net-next 1/2] net: phy: don't stop state machine in case
 of MDIO error

On 16.12.2018 09:42, Andrew Lunn wrote:
> On Sat, Dec 15, 2018 at 05:18:33PM +0100, Heiner Kallweit wrote:
>> If we detect a MDIO error, it seems to be a little bit too aggressive
>> to stop the state machine and bring down the PHY completely.
> 
> Hi Heiner
> 
> My assumption is, if we get one MDIO error, we will gets lots more
> MDIO errors. This should be the sort of bus which either works, or it
> does not. In that situation, i think it does make sense to stop
> everything.
> 
> I would like to know more about the FEC. Unfortunately, the commit
> adding the timeout handling code does not explain why it is needed.
> We might need to dig further backwards in time to figure it out.
> 
After checking the fec history it seems that at the time this
workaround was added as part of phylib support (8 years ago),
the MDIO access timeout value was too low and therefore sometimes
MDIO access failed. Later timeout was set to a higher value and
driver switched to an event-driven mechanism to signal end of
MDIO access. So it should be safe to remove the workaround.

One issue with phy_error() is that it silently stops the PHY.
We should at least add a phydev_err() to let the user know.

>    Andrew
> 
Heiner