Re: [PATCH 2/2] maps/mtd-ram: add an of-platform driver

[Grant Likely <grant.likely@secretlab.ca>]

On Wed, May 20, 2009 at 6:46 PM, Wolfram Sang <w.sang@pengutronix.de> wrote=
:
>> > + =A0 =A0 =A0 name =3D of_get_property(op->node, "name", NULL);
>> > + =A0 =A0 =A0 if (!name) {
>> > + =A0 =A0 =A0 =A0 =A0 =A0 =A0 ret =3D -ENOENT;
>> > + =A0 =A0 =A0 =A0 =A0 =A0 =A0 dev_dbg(&op->dev, "could not get node na=
me\n");
>> > + =A0 =A0 =A0 =A0 =A0 =A0 =A0 goto bad1;
>> > + =A0 =A0 =A0 }
>
> Can I just use
>
> =A0 =A0 =A0 =A0name =3D op->node->name
>
> here? I wonder because of_device in asm/of_device.h states 'node' as "to =
be
> obsoleted"
It may be labeled as such, but it's been so for over 2 years now, and
op->node is used all over the place.  I think Ben would like to
eventually move to using the node pointer in archdata, but that will
require a fair bit of refactoring.

Ben will kick me if I'm wrong, but I'll go out on a limb and say that
I think you're okay to use it.

>. And could I safely assume that all architectures will have the node
> entry?

All three current users do.

>> > +static struct of_device_id of_ram_match[] =3D {
>> > + =A0 =A0 =A0 { .compatible =3D "mtd-ram", },
>> > + =A0 =A0 =A0 {},
>> > +};
>> > +MODULE_DEVICE_TABLE(of, of_ram_match);
>> > +
>> > +static struct of_platform_driver of_ram_driver =3D {
>>
>> __devinitdata
>
> I assume you mean the match_table and not the of_platform_driver. Shouldn=
't it
> even better be const and using __devinitconst?

correct on both counts.

g.

--=20
Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.