On Thu, 20 Feb 2014, Dan Williams wrote: > > You forgot to acquire the port lock in usb_reset_device. The lock and > > unlock calls should surround the call to usb_reset_and_verify_device. > > Both that routine and hub_port_init should be documented as requiring > > their caller to hold the port lock. > > This might be a reason to move the lock to the usb_device itself. That's not feasible, because we have to acquire the lock at times where there's no device plugged into the port. > Otherwise we'll end up with a conditional lock when resetting root hub > devices. Sparse will not be able to statically verify such a lock > [1]. I had assumed that calliing usb_reset_device() on a root hub was > not allowed. It isn't. When you think about it, there is no way to reset a root hub. (Or at least, not anything resembling the way you reset an external hub.) > Hmm, maybe better to just up-level the udev->parent check to > usb_reset_device directly. There is one place where usb_reset_and_verify_device is called directly, not through usb_reset_device. (But that place doesn't need the check, so moving it wouldn't hurt anything...) 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
