> > Why was this changed? They should return a pointer to their device, > > instead of relying on their internal table of devices. > > No idea about the reason, but remember that the probe() return can't be a > pointer to the device itself, only to one of the two interfaces. I think that > the control interface is best, but I haven't done the changes yet.
Usually what's returned is the internal device handle; no reason it can't be used, since the first thing disconnect() needs to do is get that to see what device state needs updating. That handle can hook back to the interface struct(s) if needed. The USB subsystem only treats it as a flag indicating whether the device is claimed or not. Also, maybe it'd be better to start thinking of "function" or "interface" drivers not "device" drivers ... since some devices (I'm looking at one right now :) need more than one to work. - Dave _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
