On Fri, 27 Feb 2004, Greg KH wrote:

> On Wed, Feb 25, 2004 at 10:05:37AM -0500, Alan Stern wrote:
>
> > Why would anyone want to do this, you ask?  Well the USB subsystem does it
> > already.  Each USB device can have several configurations, only one of
> > which is active at any time.  Corresponding to each configuration is a set
> > of struct devices, and they (together with their embedded kobjects) are
> > allocated and initialized when the USB device is first detected.  The
> > struct devices are add()'ed and del()'ed as configurations are activated
> > and deactivated, leading to just the sort of call sequence shown above.
>
> Then we need to fix this.

The driver model does not support repeated device_add(), device_del(),
device_add(), device_del(), ... calls for the same device.  But that's
what happens to an interface's embedded struct device when we change
configurations.

Accordingly, this patch changes the device_add()/device_del() calls for
interfaces to device_register()/device_unregister().  When the interface
is unregistered the new code waits for the release method to run, so that
it will be safe to re-register the interface should the former
configuration be reinstated.

Greg, please check this out to make sure I haven't made some dumb mistake.
It works on my system and it fixes a memory leak in the USB system.