netdev
[Top] [All Lists]

Re: netdev_ops retraction

To: Arnaldo Carvalho de Melo <acme@xxxxxxxxxxxxxxxx>
Subject: Re: netdev_ops retraction
From: "David S. Miller" <davem@xxxxxxxxxx>
Date: Wed, 30 Jul 2003 16:41:08 -0700
Cc: willy@xxxxxxxxxx, netdev@xxxxxxxxxxx
In-reply-to: <20030730233541.GB7057@xxxxxxxxxxxxxxxx>
References: <20030730184416.GI22222@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <20030730233541.GB7057@xxxxxxxxxxxxxxxx>
Sender: netdev-bounce@xxxxxxxxxxx
On Wed, 30 Jul 2003 20:35:42 -0300
Arnaldo Carvalho de Melo <acme@xxxxxxxxxxxxxxxx> wrote:

> Em Wed, Jul 30, 2003 at 07:44:16PM +0100, Matthew Wilcox escreveu:
> > I think there's still scope for a netdev_ops patch, but it's of dubious 
> > value
> > and more of a 2.7 project.
> 
> OK with me, I mentioned this in a brainstorm and thought of it as a 2.7 thing
> anyway.

I'm ok with the simplified ethtool-only version too.

Although I'm confused about what kind of problem there is with
netdev_ops being such a "large structure".

This is the kind of thing there'd be _ONE_ copy of in each driver,
ala.

struct netdev_ops tg3_netdev_ops {
 ...
        .foo    =       tg3_foo,
 ...
};

...
        tp->dev->netdev_ops = &tg3_netdev_ops;
...

Right?

<Prev in Thread] Current Thread [Next in Thread>