netdev - Re: [PATCH net-next] net: mdio: access c22 registers via debugfs

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: <6eadc811-783d-2fc9-60c5-b765ecc87f5a@simerda.eu>
Date:   Tue, 19 Jan 2021 01:55:56 +0100
From:   Pavel Šimerda <code@...erda.eu>
To:     Tobias Waldekranz <tobias@...dekranz.com>, netdev@...r.kernel.org
Subject: Re: [PATCH net-next] net: mdio: access c22 registers via debugfs

On 1/19/21 12:27 AM, Tobias Waldekranz wrote:
> On Sat, Jan 16, 2021 at 22:19, Pavel Šimerda <code@...erda.eu> wrote:
>> Provide a debugging interface to read and write MDIO registers directly
>> without the need for a device driver.
>>
>> This is extremely useful when debugging switch hardware and phy hardware
>> issues. The interface provides proper locking for communication that
>> consists of a sequence of MDIO read/write commands.
>>
>> The interface binds directly to the MDIO bus abstraction in order to
>> provide support for all devices whether there's a hardware driver for
>> them or not. Registers are written by writing address, offset, and
>> value in hex, separated by colon. Registeres are read by writing only
>> address and offset, then reading the value.
>>
>> It can be easily tested using `socat`:
>>
>>      # socat - /sys/kernel/debug/mdio/f802c000.ethernet-ffffffff/control
>>
>> Example: Reading address 0x00 offset 0x00, value is 0x3000
>>
>>      Input: 00:00
>>      Output: 3000
>>
>> Example: Writing address 0x00 offset 0x00, value 0x2100
>>
>>      Input: 00:00:2100
>>
>> Signed-off-by: Pavel Šimerda <code@...erda.eu>
> 
> Hi Pavel,
> 
> I also tried my luck at adding an MDIO debug interface to the kernel a
> while back:
> 
> https://lore.kernel.org/netdev/C42DZQLTPHM5.2THDSRK84BI3T@wkz-x280

Hey Tobias,

nice to meet you!

> The conclusion was that, while nice to have, it makes it too easy for
> shady vendors to write out-of-tree drivers.

That was exactly what I was afraid of.

> You might want to have a look at https://github.com/wkz/mdio-tools. It
> solves the same issue that your debugfs interface does, and also some
> other nice things like clause 45 addressing and atomic read/mask/write
> operations.

Thank you very much!

Cheers,

Pavel