On Fri, 6 Sep 2019 Charles.Hyde@xxxxxxxxxxxx wrote: > <snipped> > > > + ret = usbnet_read_cmd(dev, USB_CDC_GET_NET_ADDRESS, > > > + USB_DIR_IN | USB_TYPE_CLASS > > > + | USB_RECIP_INTERFACE, 0, > > > + USB_REQ_SET_ADDRESS, buf, ETH_ALEN); > > > > Where did that USB_REQ_SET_ADDRESS come from? Did you just look up an > > arbutrary macro that happened to match your device config? How do you > > expect this to work with a generic NCM device? Or even your own device, > > when the next firmware revision moves the function to a different interface? > <snipped> > > https://wiki.osdev.org/Universal_Serial_Bus#SET_ADDRESS > > https://www.usb.org/document-library/network-control-model-devices-specification-v10-and-errata-and-adopters-agreement > Download and view the NCM specification v1.0 (Errata 1), dated November 24, 2010. See table 6.2 on page 30. Also see sections 6.2.2 and 6.2.3 on page 32. > > USB_REQ_SET_ADDRESS came from include/uapi/linux/usb/ch9.h. This matches the SET_ADDRESS definition from the osdev wiki page, so I used it because the name and numeric values both matched. It further matches what the Windows driver issues. The names and values may match, but the meanings do not. USB_REQ_SET_ADDRESS refers to a USB bus address, not a MAC address. Alan Stern
