On 08/22/2012 01:42 PM, Felipe Balbi wrote:
If ccg needs a different setup method, it's completely non-standard. I don't think it should be using composite.c at all. Just make ccg a standalone gadget non-dependent on this composite.c and let's move on. If later ccg wants to continue to exist in the kernel, they need to adapt to composite.c properly, instead of hacking its own stuff on composite.c's setup method.
Okay. Just for the protocol: - the setup callback has a list_for_each() of a private ccg member which has _no_ users at all. - the state tracking of the udc/gadget which can be "connected, disconnected, configured, …" was discussed here before and it should be implemented in udc-core. So I will move it as it has been done for keucr driver. Sebastian _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/devel
