Re: USB networking (cdce) with Linux host (g_ether)

[Nick Hudson <skrll%netbsd.org@localhost>]

On 04/05/15 01:52, Ryo ONODERA wrote:
> Hi,
>
> I believe that following patch is good for us because my USB modem device
> is bad information and it should be treated as special case.
>
> Index: sys/dev/usb/umodem.c
> ===================================================================
> RCS file: /cvsroot/src/sys/dev/usb/umodem.c,v
> retrieving revision 1.66
> diff -u -r1.66 umodem.c
> --- sys/dev/usb/umodem.c	6 Apr 2014 02:01:43 -0000	1.66
> +++ sys/dev/usb/umodem.c	5 Apr 2015 00:49:30 -0000
> @@ -81,6 +81,11 @@
>   	NULL,
>   };
>   
> +Static const struct usb_devno umodem_devs[] = {
> +	{ USB_VENDOR_CONEXANT, USB_PRODUCT_CONEXANT_MODEM_1},
> +};
> +#define umodem_lookup(v, p) usb_lookup(umodem_devs, v, p)
> +
>   int             umodem_match(device_t, cfdata_t, void *);
>   void            umodem_attach(device_t, device_t, void *);
>   int             umodem_detach(device_t, int);
> @@ -97,22 +102,24 @@
>   	struct usbif_attach_arg *uaa = aux;
>   	usb_interface_descriptor_t *id;
>   	int cm, acm;
> +	int ret = UMATCH_NONE;
>   
> -	id = usbd_get_interface_descriptor(uaa->iface);
> -	if (uaa->subclass != UISUBCLASS_ABSTRACT_CONTROL_MODEL &&
> -	    (id->bInterfaceClass == UICLASS_CDC_DATA &&
> -	     id->bInterfaceSubClass == UISUBCLASS_DATA))
> -		return (UMATCH_IFACECLASS_IFACESUBCLASS);
> -
> -	if (uaa->class != UICLASS_CDC ||
> -	    uaa->subclass != UISUBCLASS_ABSTRACT_CONTROL_MODEL ||
> -	    !(uaa->proto == UIPROTO_CDC_NOCLASS || uaa->proto == UIPROTO_CDC_AT))
> -		return (UMATCH_NONE);
> +	if (umodem_lookup(uaa->vendor, uaa->product))
> +		ret = UMATCH_VENDOR_PRODUCT;
>   
> +	if (uaa->class == UICLASS_CDC &&
> +	    uaa->subclass == UISUBCLASS_ABSTRACT_CONTROL_MODEL &&
> +	    uaa->proto == UIPROTO_CDC_AT)
> +		ret = UMATCH_IFACECLASS_IFACESUBCLASS_IFACEPROTO;

Missing  uaa->proto == UIPROTO_CDC_NOCLASS, I think.

That said, I think there's a bug somewhere else that's causing problems 
here.

Can you boot with -xv flags and an updated -current kernel with USB_DEBUG
and usbdebug set? I've add some more debug output to the kernel.

Thanks,
Nick