Re: [RFC 1/3] OLPC: Use device tree for platform identification

[Grant Likely <grant.likely-s3s/WqlpOiPyB63q8FvJNQ@public.gmane.org>]

On Tue, Mar 08, 2011 at 07:35:46PM -0800, Andres Salomon wrote:
> On Tue, 8 Mar 2011 21:13:00 +0000
> Daniel Drake <dsd-2X9k7bc8m7Mdnm+yROfE0A@public.gmane.org> wrote:
> 
> > On 4 March 2011 18:01, Andres Salomon <dilinger-pFFUokh25LWsTnJN9+BGXg@public.gmane.org> wrote:
> > >> +static int __init declare_of_platform_devices(void)
> > >> +{
> > >> +     return of_platform_bus_probe(NULL, of_ids, NULL);
> > >> +}
> > >> +device_initcall(declare_of_platform_devices);
> > >
> > > How about 'olpc_create_platform_devices' or some such name?  That's
> > > what's really happening here, right?
> > 
> > The reason I called it declare_of_platform_devices is because thats
> > what all the other platforms do (do a git grep). do you still think it
> > should be renamed?
> > 
> 
> Yes, unless there's some reason why it should be named "declare".  I'm
> assuming the names used in ppc platform device code are like that for
> historical reasons?

Probably, but they probably aren't very good reasons.  Name it what
makes sense for OLPC.  An olpc_ prefix certainly makes sense to me.

BTW, using a device_initcall() isn't very multiplatform friendly.  It
will get called regardless of the machine that the kernel is booted on
(OLPC or no) which probably isn't what you want.

g.