On Fri, 8 Jun 2012, Greg KH wrote: > > Took a quick peek at drivers/usb/serial/bus.c and became more confused. > > usb_serial_device_remove() will just unregister the tty device, and > > usb_serial_device_probe() will the let you register it again. So you > > save a tty device, but unbinding from the usb-serial driver this way > > doesn't free any other resources. In particular not the precious ttyUSB > > minors. > > > > I find that behaviour a bit confusing. I did expect that unbinding > > would cause all resources related to the unbound port to be freed and > > made available for other drivers and/or devices. > > > > But I don't see any way this can be changed either. If we propagate the > > unbind to the usb driver, then there is no way to keep usb-serial "bind" > > working like before. And I guess that's not acceptable... > > Yeah, I don't think we can really do this properly, as we allow multiple > minors to a single device, and sometimes, we control them together > through the same interface. If the device has multiple interfaces, it > should work by handling the bind/unbind from the USB driver layer, not > the USB-Serial layer. It's not just that... Many of the serial drivers were written before the driver model was well established, and most likely they _can't_ unbind from individual ports. They are all-or-nothing. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
