Bin Liu <binmlist@xxxxxxxxx> writes: > Hi, > > Sorry if I missed any previous discussion on this topic. > > The msdn link [1] says that > > The host does not continuously poll the USB Control endpoint for input > control messages. Upon placing a control message on its Control > endpoint, the device must return a notification on the Communication > Class interface's Interrupt IN endpoint, which is polled by the host > whenever the device can return control messages. > > Which means the following transfer sequence from the host: > > - SETUP xfer: SET_ENCAPSULATED_COMMAND > - INT xfer > - SETUP xfer: GET_ENCAPSULATED_RESPONSE > > But in rndis_command() in rndis_host.c, the INT xfer between > SET_ENCAPSULATED_COMMAND and GET_ENCAPSULATED_RESPONSE has a if > condition check, so the INT xfer is not sent for some (most) rndis > devices. > > Is there any reason why the INT xfer has this if condition? I don't know the previous discussions either, but I have always assumed this is just due to the usbnet restrictions at the time the code was written. You had to to the sync send + recv commands then, and the interrupt transfer was probably considered unnecessary. Until some devices actually requiring it started showing up and the flag was added. Guessing a lot here... Anyway, I believe this code could be rewritten to do the async commands now that Dan has added the ability to start the usbnet status URB at driver probe time. Bjørn -- 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
