usbatm : modprobe & rmmod

[Duncan Sands]

Hi Roman,

> Please consider me convinced.  Indeed the usbatm2 framework seems to
> suffice for everything I can think of, and is much safer by vurtue of
> making calls into the subdriver, and not vice versa.
> 
> So I give up trying with usb_atm (this time for good).

:)

> However, I think Matthieu's desire to use .driver_data to handle the
> chipset pecularities is a valid argument, so usbatm should provide some
> other means to pass usbatm_driver to the core.  It can be done either at
> subdriver probe time or in usb-serial fashion, by maintaining a list of
> subdrivers updated at subdriver register time, and choosing the
> appropriate one using usb_match_id in the core probe routine.

I agree, and will try to come up with something today.

All the best,

Duncan.