Re: [PATCH 2.6.27-rc3] led: driver for LEDs on PCEngines ALIX.2 and ALIX.3 boards

[Andrew Morton <akpm@linux-foundation.org>]

On Sat, 23 Aug 2008 02:08:15 +0500
Constantin Baranov <const@const.mimas.ru> wrote:

> >  static int alix_led_resume(struct platform_device *dev)
> >  {
> >  	int i;
> > +
> >  	for (i = 0; i < ARRAY_SIZE(alix_leds); i++)
> >  		led_classdev_resume(&alix_leds[i].cdev);
> >  	return 0;
> > @@ -92,7 +95,7 @@ static int __init alix_led_probe(struct 
> >  		ret = led_classdev_register(&pdev->dev, &alix_leds[i].cdev);
> >  
> >  	if (ret < 0) {
> > -		for (i = i - 2; i >= 0; i--)
> > +		while (--i >= 0)
> >  			led_classdev_unregister(&alix_leds[i].cdev);
> >  	}
> 
> At the first iteration this while-loop will attempt to unregister
> device that has failed and thus is not registered.
> My for-loop starts from device immediately before failed one.

ug, OK, the complex expression in that for-loop is to blame.

For maintinability we should aim for code which is as simple and as
straightfroward as possible and which adheres to oft-used and
well-understood kernel idioms.

For example, this?

static int __init alix_led_probe(struct platform_device *pdev)
{
	int i;
	int ret;

	for (i = 0; i < ARRAY_SIZE(alix_leds); i++) {
		ret = led_classdev_register(&pdev->dev, &alix_leds[i].cdev);
		if (ret < 0)
			goto fail;
	}
	return 0;

fail:
	while (--i >= 0)
		led_classdev_unregister(&alix_leds[i].cdev);
	return ret;
}

(note there's no longer a need for a fake initalisation of `ret')

> Note that probe routine originates from leds-ams-delta.c.

That's what you get for copying stuff :(