On 06/30/2012 01:04 AM, Stephen Warren wrote:
 >> -	dev_dbg(&pdev->dev, "got pwm for backlight\n");
 >> -
 >
 > That seems like an unrelated change?

Dammit, I hoped that would have gone unnoticed. ;)

 >> +	pb->enable_gpio = -EINVAL;
 >> +	if (data->use_enable_gpio) {
 >> +		ret = devm_gpio_request_one(&pdev->dev, data->enable_gpio,
 >> +				GPIOF_OUT_INIT_HIGH, "backlight_enable");
 >> +		if (ret)
 >> +			dev_warn(&pdev->dev,
 >> +				"error %d requesting control gpio\n", ret);
 >
 > Shouldn't that be a hard error? If the user specified that some GPIO be
 > used, and the GPIO could not be requested, shouldn't the driver fail to
 > initialize?

Yes, you are right.

Thanks,
Alex.