Re: [PATCH WIP] parport: add device model

[Greg KH <gregkh@linuxfoundation.org>]

On Sat, Apr 11, 2015 at 01:41:34PM +0530, Sudip Mukherjee wrote:
> On Sat, Apr 11, 2015 at 09:27:16AM +0200, Greg KH wrote:
> > On Sat, Apr 11, 2015 at 10:56:51AM +0530, Sudip Mukherjee wrote:
> > > > I know that I have done a rubbish review.  I'm going to have to review
> > > > this properly later.
> > > main thing i wanted to know is if my approach is correct. since nothing
> > > on that so I hope I am on the correct track. Thanks.
> > > I will send in the next version in a day or two.
> > 
> > At quick glance, you are on the right track.  Writing a new bus is hard,
> > I know, the documentation is lacking and it's tricky in places and the
> > api is horrid in others.  I know this, just never had the time to make
> > it easier, so if you have any questions about it, please let me know.
> well, as of now one question. I am planning like this :
> 
>                        sys
>          _______________|_____________
>          |	|	|       |    | 
>         bus
> ________|______
>               |
>            parport
>        _______|_____
>        |            |
>     devices       drivers- lp, ppdev, panel etc.. 
>   _____|________
>   |            |
> parport0      parport1
> 
> 
> I can understand that drivers need to be binded to one device, so
> suppose ppdev wants to use parport0, how that binding will be done?

The driver core does the "binding".

> do i need mark the driver in the parport->ddev.driver ?

No, just have the probe function for the ppdev function return 0,
meaning it successfully bound to the driver that was passed to it and
all will be fine.

> I think that should have been automatically done if i have a probe
> function for the bus...

Yes.

> somehow I am missing this step in the other drivers codes.

Hope this helps,

greg k-h