Re: [PATCH 7/9] clk: Allow parents to be specified without string names

From: Stephen Boyd
Date: Fri Feb 15 2019 - 16:13:40 EST


Quoting Jerome Brunet (2019-02-13 01:32:23)
> On Tue, 2019-01-29 at 10:56 -0800, Stephen Boyd wrote:
> > > > +/**
> > > > + * struct clk_parent_data - clk parent information
> > > > + * @hw: parent clk_hw pointer (used for clk providers with internal
> > > > clks)
> > > > + * @name: parent name local to provider registering clk
> > > > + * @fallback: globally unique parent name (used as a fallback)
> > > > + */
> > > > +struct clk_parent_data {
> > > > + struct clk_hw *hw;
> > > > + const char *name;
> > > > + const char *fallback;
>
> One last nitpick about this structure, because I did not figure it out at
> first.
>
> 'fallback' is what we known as 'name' in CCF so far.
>
> What do you think about renaming 'fallback' to 'name' and 'name' to something
> more obvious, like 'of_name' or 'fw_name' or something else ?

Ok. I'm not super fond of assuming it's the DT specific, so maybe
fw_name is good, or ext_name for external name? Or con_id to match
clkdev?

>
> > >
> > > If I understand correctly, .name and .fallback will be ignored if hw is
> > > provided ? Maybe this should be documented somehow ?
> >
> > Sure. I'll add some documentation to the long portion of the kernel-doc
> > here describing priority order.
>
> Anyway, with this patch, I should be able to remove a lot of (ugly) code I
> have been writting lately. I'll be happy to test it when you have a v2 ready.
>

Great!