Re: [PATCH v3 2/2] gpio: sim: initialize a managed pointer when declaring it

[Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>]

On Sun, Sep 17, 2023 at 11:12:25AM +0200, Bartosz Golaszewski wrote:
> From: Bartosz Golaszewski <bartosz.golaszewski@xxxxxxxxxx>
> 
> Variables managed with __free() should typically be initialized where
> they are declared so that the __free() callback is paired with its
> counterpart resource allocator. Fix the second instance of using
> __free() in gpio-sim to follow this pattern.

...

>  {
> -	struct gpio_sim_device *dev __free(kfree) = NULL;
>  	int id;
>  
> -	dev = kzalloc(sizeof(*dev), GFP_KERNEL);
> +	struct gpio_sim_device *dev __free(kfree) = kzalloc(sizeof(*dev),
> +							    GFP_KERNEL);
>  	if (!dev)
>  		return ERR_PTR(-ENOMEM);

Aside: Oh, this might be a downside of the __free() sugar, as we can
theoretically end up with a code in the future like

	struct bar *foo;
	...
	struct baz *foo __free() = ...
	...

and I am not sure how it goes to work. Or relaxed variant with

	struct bar *foo;
	...
	{
		struct baz *foo __free() = ...
		...
	}

where we would have two variables with the same name, but different scope
(this, perhaps, would work, but I assume compiler should warn about shadowed
 name for the variable).

(Also what if in both cases bar == baz, i.e. same type?)

-- 
With Best Regards,
Andy Shevchenko