On Thu, 03 Sep 2020 02:45:53 +0200, Alan Stern wrote: > > In a few places here this check is completely unnecessary. All it does > is verify that the device does have an endpoint 0 and the the type of > the endpoint matches the type of the pipe. Well, every USB device > always has an endpoint 0, and it is always a bidirectional control > endpoint. Therefore a simple static check is all you need: There's no > point calling usb_pipe_type_check() when the pipe is of the form > usb_{snd|rcv}ctrlpipe(dev, 0). > > In short, this check should be removed completely; it does nothing. Fair enough, but I think those removals should be in another patch. thanks, Takashi