[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <724f69f0-7eab-40aa-84f0-07055f051175@prolan.hu>
Date: Mon, 25 Aug 2025 16:39:12 +0200
From: Csókás Bence <csokas.bence@...lan.hu>
To: "Russell King (Oracle)" <linux@...linux.org.uk>
CC: Andrew Lunn <andrew@...n.ch>, Heiner Kallweit <hkallweit1@...il.com>,
"David S. Miller" <davem@...emloft.net>, Eric Dumazet <edumazet@...gle.com>,
Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>, "Philipp
Zabel" <p.zabel@...gutronix.de>, <netdev@...r.kernel.org>,
<linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] net: mdiobus: Move all reset registration to
`mdiobus_register_reset()`
Hi,
On 2025. 08. 25. 16:16, Russell King (Oracle) wrote:
> On Mon, Aug 25, 2025 at 04:09:34PM +0200, Bence Csókás wrote:
>> Make `mdiobus_register_reset()` function handle both gpiod and
>> reset-controller-based reset registration.
>
> The commit description should include not only _what_ is being done but
> also _why_.
Well, my question was, when I saw this part of code: why have it
separate? Users shouldn't care whether a device uses gpiod or
reset-controller when they call `mdio_device_reset()`, so why should
they have to care here and call two separate register functions, one
after another? In fact, the whole thing should be moved to mdio_device.c
honestly. Along with the setting of mdiodev->reset_{,de}assert_delay.
The end goal is fixing this "Can't read PHY ID because the PHY was never
reset" bug that's been plaguing users for years.
Bence
Powered by blists - more mailing lists