Re: [PATCH v1 1/1] device property: Clarify description on returned value in some functions

[Sakari Ailus]

On Fri, Feb 17, 2023 at 01:55:43PM +0200, Andy Shevchenko wrote:
> On Fri, Feb 17, 2023 at 01:44:25PM +0200, Sakari Ailus wrote:
> > On Fri, Feb 17, 2023 at 01:28:55PM +0200, Andy Shevchenko wrote:
> > > On Fri, Feb 17, 2023 at 01:18:31PM +0200, Sakari Ailus wrote:
> > > > On Fri, Feb 17, 2023 at 01:01:39PM +0200, Andy Shevchenko wrote:
> > > > > On Fri, Feb 17, 2023 at 12:27:53PM +0200, Sakari Ailus wrote:
> > > > > > On Thu, Feb 16, 2023 at 10:57:08PM +0200, Andy Shevchenko wrote:
> 
> ...
> 
> > > > > > >   * fwnode_get_next_child_node - Return the next child node handle for a node
> > > > > > >   * @fwnode: Firmware node to find the next child node for.
> > > > > > >   * @child: Handle to one of the node's child nodes or a %NULL handle.
> > > > > > > + *
> > > > > > > + * Caller is responsible to call fwnode_handle_put() on the returned fwnode
> > > > > > > + * pointer.
> > > > > > 
> > > > > > The loop itself will also put the child node, so this is only relevant
> > > > > > outside the loop.
> > > > > 
> > > > > Yes and this is exactly what people stumbled over. Hence this note.
> > > > > This call per se doesn't loop, so I didn't get how your comment can
> > > > > be transformed to anything here. Care to elaborate a bit more on
> > > > > what I have to add here or reword?
> > > > 
> > > > Ah, indeed. This is achieved by putting the previous child. Generally this
> > > > function is used via the loop helper macro and not called directly, hence
> > > > the documentation there matters the most. Those functions appear to be
> > > > without any documentation though.
> > > 
> > > So, what should I do?
> > 
> > Good question.
> > 
> > How about this text:
> > 
> > The caller is responsible for calling fwnode_handle_put() put on the
> > returned fwnode. Note that this function also puts a reference to @child
> > unconditionally.
> 
> Fine. Does it mean I have to change existing wording of the first sentence
> everywhere?

Up to you.

> 
> > This is actually done by the firmware specific implementation, namely on OF
> > and at least should be done on swnode.
> 
> Yes, that's. But it's not needed to be added.
> 
> > A second patch to document the fwnode iterator macros would be nice.
> 
> Not the reported problem. Maybe someone else can do the job?

It's perhaps not the most pressing issue at the moment. Let's see.

-- 
Sakari Ailus