On Tue, 17 Jan 2012, Oliver Neukum wrote: > Am Samstag, 14. Januar 2012, 17:39:04 schrieb Alan Stern: > > On Sat, 14 Jan 2012, Oliver Neukum wrote: > > > > > True. Taking this to the logical conclusion they should have two delays, > > > one for internal events and one for external events. > > > > Sometimes it's not so easy to make the distinction. Sure, you can tell > > why the hub woke up originally, but other things may happen while the > > hub is awake. For example, a hub may get woken up because of an > > autoresume call for a child, and while it's awake there may be a port > > connect change. > > True, but do we care? Surely the reaction to child nodes is the far > more common case and we are talking about a heuristic anyway. Which case is more common is not at all obvious. It depends entirely on what sort of devices are plugged into the hub and how they are used. Still, Sarah may well have a point. It may well make sense to reduce the autosuspend delay for hubs down to 100 ms or so. Maybe even 0, but that would have unpleasant consequences for programs like lsusb. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
