[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <8729016553E3654398EA69218DA29EEF15A81CEB@cnshjmbx02>
Date: Thu, 23 Feb 2017 00:10:27 +0000
From: YUAN Linyu <Linyu.Yuan@...atel-sbell.com.cn>
To: Andrew Lunn <andrew@...n.ch>
CC: Florian Fainelli <f.fainelli@...il.com>,
"David S . Miller" <davem@...emloft.net>,
"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
"cugyly@....com" <cugyly@....com>
Subject: RE: create drivers/net/mdio and move mdio drivers into it
> -----Original Message-----
> From: Andrew Lunn [mailto:andrew@...n.ch]
> Sent: Wednesday, February 22, 2017 6:21 PM
> To: YUAN Linyu
> Cc: Florian Fainelli; David S . Miller; netdev@...r.kernel.org; cugyly@....com
> Subject: Re: create drivers/net/mdio and move mdio drivers into it
>
> On Wed, Feb 22, 2017 at 05:38:49AM +0000, YUAN Linyu wrote:
> > Hi Florian,
> >
> > 1.
> > Let's go back to original topic,
> > Can we move all mdio dirvers into drivers/net/mdio ?
>
> Hi Yuan
>
> Please could you explain what benefit this brings. Please also list
> all the downsides for such a move. As Florian said, we need to ensure
> such a move adds more value than it removes.
At beginning I think mdio and phy are two different things, mdio should have it's home.
>
> > Per may understanding,
> > I don't know why create a struct mii_bus instance to represent a mdio device
> in current mdio driver.
> > Why not create a struct mdio_device instance, it's easy to understand.
> > (We can move part of member of mii_bus to mdio_device).
>
> Please take a step back. What are you trying to achieve. What is the
> big picture. What cannot you do with the current design?
Big picture is we can remove struct mii_bus, and use struct mdio_device/driver for mdio controller.
>
> Andrew
Powered by blists - more mailing lists