On 06/15/2015 10:56 AM, Matt Ranostay wrote: > On Mon, Jun 15, 2015 at 1:15 AM, Jacek Anaszewski > <j.anaszewski@xxxxxxxxxxx> wrote: >>> + error = devm_led_classdev_register(dev, &led->cdev); >>> + if (error < 0) >>> + return -EINVAL; >>> + INIT_WORK(&led->work, cap11xx_led_work); >>> + schedule_work(&led->work); >> >> >> Work queue should be initialized before registration of the LED class >> device. > What is the reason for this? Just curious It's a race condition. The LED device is accessible once devm_led_classdev_register() returns. So a user could be calling into cap11xx_led_set() before the work queue is initialized, which would hence access a NULL pointer. Daniel -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html