Re: [PATCH 2/3] Maple bus support for the Sega Dreamcast - keyboard support

["Dmitry Torokhov" <dmitry.torokhov@gmail.com>]

On 9/10/07, Adrian McMenamin <dmitry.torokhov@gmail.com> wrote:
> On 10/09/2007, Dmitry Torokhov <dtor@insightbb.com> wrote:
> > Hi Adrian,
> >
>
> Thanks for the comments - will get on with this but....
>
>
> > > +     for (i = 0; i < NR_SCANCODES; i++)
> > > +             kbd->keycode[i] = dc_kbd_keycode[i];
> >
> > memcpy?
> >
>
> I see that other drivers use memcpy - and will happily convert over -
> but, out of interest, is there a reasopn why it is superior?
>

First and foremost it conveys the intent clearly - you are copying a
chunk of memory (namely a keymap) from one place to another - thus
memcpy. Plus it may produce smaller/faster code (but since it is not a
hot path this does not matter much).

> >
> > maple_device appears to be fully integrated in sysfs, please add:
> >        kbd->dev->dev.parent = &dev->dev;
> >
>
> The bus code already correctly ids the parent device (the above code
> would appear to assign the device as the device's parent
> incidentally). Is it wrong to make that assignment in the central bus
> code as opposed to the driver?
>

Here I want to set up sysfs relation for the newly created struct
input_dev which is a child of struct maple_dev. The central bus
(maple) code has no idea of input device existence, hasn't it? IOW we
need:

     input_dev->dev.parent = &maple_dev->dev;

-- 
Dmitry