On Apr 15, 2010, at 7:10, Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote: On Thu, 15 Apr 2010, Luben Tuikov wrote: If it was a whole new driver, would there be any difficulty with attaching to either a BOT device or a UAS device, depending on the device's subclass code? Would the correct mass-storage driver get invoked somehow? Absolutely not. Each interface and each alternative of the interface, maps to a USBD, given the interface class, subclass and protocol. The usb-core will either perform a dynamic mapping by advertising the triple to each USBD and then whomever "returned" the highest weight gets to handle the IF (less preferable), or the more preferable, each USBD declares the IF class, subclass and protocol it handles and its relative weight. Where did you get this idea? Not from reading the code -- this looks more like how you _imagine_ it might work than how it really does work. You catch up pretty quick. Thus, when the device presents IF0 alternative 0 as BOT (BBB) with two EPs, and IF0 alternative 1 as UAS with 4 EPs, if there is a UAS USBD in the system, the UAS IF0:1 should be preferred over the BOT one, IF0:0 (USBD is the BOT driver, a la "usb-storage"). An abstraction layer (usb-core) sitting between the HCD and the USBD would then correctly configure the device and interface and hand off the pipes to the USBD (in this case implementing UAS). No. usbcore does not configure interfaces. That is left up to the individual drivers. Thanks for pointing the obvious! 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
