netdev - Re: [PATCH net-next] net: dsa: mv88e6xxx: support in-band signalling with external PHY for 1000BaseX/2500BaseX

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: <8f81bd4f-e4b5-7abd-5f20-3b812b55a9f5@gmail.com>
Date:   Sat, 2 Mar 2019 22:03:52 +0100
From:   Heiner Kallweit <hkallweit1@...il.com>
To:     Andrew Lunn <andrew@...n.ch>
Cc:     Vivien Didelot <vivien.didelot@...il.com>,
        Florian Fainelli <f.fainelli@...il.com>,
        David Miller <davem@...emloft.net>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>
Subject: Re: [PATCH net-next] net: dsa: mv88e6xxx: support in-band signalling
 with external PHY for 1000BaseX/2500BaseX

On 02.03.2019 21:45, Andrew Lunn wrote:
>> I briefly looked at the SFP connector and there don't seem to be pins for
>> out-of-band signalling. So there must be some inband signalling.
> 
> Hi Heiner
> 
> Optical SFPs have an i2c bus with an 'EEPROM' on it. The EEPROM
> contains information about the SFP, including its maximum
> bit-rate. The Linux SFP driver will look at this bitrate, and pick the
> link mode appropriate to it. It then calls mac_config with that link
> mode, and the link speed.
> 
I briefly looked at the sfp phy driver and it doesn't seem to use phylink.
So maybe the code in question is needed only if phylink is involved.

As Russell wrote phylink may not yet fully support use case
[Switch port MAC] <--> [SERDES PHY] <--> [External PHY]
He sketched a few ideas.

> The MAC then needs to configure the SERDES to that mode/speed. After a
> while, it will get sync and trigger an interrupt. At is then reported
> via phylink_mac_change() at which point the carrier is indicated as
> up. If one end is using 1000Base-X and the other 2500Base-X, they will
> fail to sync. There is no negotiation down to 1000Base-X. 
> 
> Optical SFPs are pretty passive devices, they just do electrical to
> optical, and not a lot more. All the 'intelligence' is in the SERDES
> layers, and they talk to each other end-to-end, unlike copper, where
> the MAC SERDES is talking to the PHY SERDES.
> 
> Copper SFPs are different. They use SGMII with inband signalling.
> There is also a mechanism to encapsulate MDIO over i2c.
> 
Good to know. Again something learnt.

So there may be cases where the code we talk about isn't needed.
But if let's say a 2.5Gbps copper PHY is connected to port 9 of
a 88E6390 (like the Aquantia PHY), then 2500BaseX is used and I
think the code is needed, or?

>        Andrew
> 
Heiner