Re: [PATCH 1/2] rust: usb: add basic USB abstractions

[Alan Stern <stern@rowland.harvard.edu>]

On Tue, Sep 23, 2025 at 04:38:34PM +0200, Danilo Krummrich wrote:
> >> @Greg: Can we guarantee that a struct usb_device is always bound as long as one
> >> of its interfaces is still bound?
> >
> > Bound to what?
> 
> Well, that's kinda my point. :)
> 
> Having a &usb::Device<Bound> would mean that for the lifetime of the reference
> it is guaranteed that the usb::Device is bound to its USB device driver
> (struct usb_device_driver).
> 
> The code above establishes that you can get a &usb::Device<Bound> from a
> &usb::Interface<Bound>, i.e. an interface that is bound to a USB driver
> (struct usb_driver).
> 
> It also does establish the same with other device contexts, such as the Core
> context.
> 
> Despite the question whether this is sematically useful, I doubt that this is
> a correct assumption to take.

The intention of the USB stack is that yes, a usb_device cannot have 
children if it isn't bound to a usb_device_driver.  However, we don't 
try to guarantee that this is true; a particular driver might not 
enforce this restriction.

There is a surprisingly large number of calls to 
usb_register_device_driver() in the kernel (four in addition to the 
standard one).  I suppose a little auditing could ensure that these 
drivers do deconfigure their devices when they unbind.

Alan Stern