Re: [PATCH v7 1/5] driver core: Introduce device_{add,remove}_of_node()

[Herve Codina <herve.codina@bootlin.com>]

Hi Jonathan,

On Wed, 19 Feb 2025 15:59:01 +0000
Jonathan Cameron <Jonathan.Cameron@huawei.com> wrote:

...

> > Signed-off-by: Herve Codina <herve.codina@bootlin.com>  
> A few passing comments. Not suggestions to actually change anything
> at this stage though. Maybe a potential follow up if you think it's
> a good idea.
> 
...

> > +void device_remove_of_node(struct device *dev)
> > +{
> > +	dev = get_device(dev);
> > +	if (!dev)
> > +		return;  
> Maybe use
> 	struct device *d __free(put_device) = get_device(dev);
> 
> 	if (!d->of_node);
> 		return;
> 
> Not a reason to respin though!
> 
> 
...

> > +int device_add_of_node(struct device *dev, struct device_node *of_node)
> > +{
> > +	int ret;
> > +
> > +	if (!of_node)
> > +		return -EINVAL;
> > +
> > +	dev = get_device(dev);  
> 
> Likewise could use __free() magic here as well for slight simpliciations.
> 

I see. Indeed, the __free(put_device) can be an improvement in core.c

I think that this has to be done out of this series in a more globally way
because put_device() is used in several place in this file and having a mix
between __free(put_device) and put_device() calls in a goto label is not the
best solution.

For this reason, as you proposed except if someone else pushes in the
__free(put_device) direction in functions introduced in this patch, I
prefer to keep this patch as it is.

Thanks for your feedback,
Hervé