Re: [PATCH v8 1/1] serial: core: Start managing serial controllers to enable runtime PM

[Tony Lindgren <tony@atomide.com>]

* Andy Shevchenko <andriy.shevchenko@intel.com> [230322 14:45]:
> On Wed, Mar 22, 2023 at 01:12:51PM +0200, Tony Lindgren wrote:
> > +	device_initialize(&sbd->dev);
> > +	sbd->dev.parent = parent_dev;
> > +	sbd->dev.bus = &serial_base_bus_type;
> > +	sbd->dev.release = &serial_base_release;
> > +
> > +	if (str_has_prefix(name, "ctrl")) {
> > +		id = port->ctrl_id;
> > +	} else {
> > +		id = port->line;
> > +		sbd->port = port;
> > +	}
> > +
> > +	err = dev_set_name(&sbd->dev, "%s.%s.%d", name, dev_name(port->dev), id);
> > +	if (err)
> > +		goto err_free_dev;
> > +
> > +	err = device_add(&sbd->dev);
> > +	if (err)
> > +		goto err_put_device;
> > +
> > +	return &sbd->dev;
> > +
> > +err_put_device:
> > +	put_device(&sbd->dev);
> 
> > +	kfree_const(sbd->dev.kobj.name);
> 
> This is double free if not const, right?
> At least that's how I read kobject_cleanup() implementation.
> 
> Sorry I haven't paid attention to this earlier.

Thanks for spotting that, will drop it. Looks like we have the name
allocated by kobject_set_name_vargs(), and then kobject_cleanup()
frees it.

> ...
> 
> > +/*
> > + * Serial core port device driver
> > + */
> 
> Put it on one line for now?

Sure, will do.

Thanks,

Tony