On Mon, Feb 21, 2022 at 05:20:05PM +0800, Slark Xiao wrote:
>> >> @@ -1063,6 +1065,10 @@ static const struct usb_device_id option_ids[] = {
> >> >> .driver_info = RSVD(0) | RSVD(1) | RSVD(6) },
> >> >> { USB_DEVICE(DELL_VENDOR_ID, DELL_PRODUCT_5821E_ESIM),
> >> >> .driver_info = RSVD(0) | RSVD(1) | RSVD(6) },
> >> >> + { USB_DEVICE_INTERFACE_CLASS(DELL_VENDOR_ID, DELL_PRODUCT_5829E, 0xff),
> >> >> + .driver_info = RSVD(6) },
> >> >> + { USB_DEVICE_INTERFACE_CLASS(DELL_VENDOR_ID, DELL_PRODUCT_5829E_ESIM, 0xff),
> >> >> + .driver_info = RSVD(6) },
> >> >
> >> >It looks like these entries will cause the driver to bind also to the
> >> >QMI port however.
> >> >
> >
> >> Actually not, currently RMNET port would load the qmi_wwan driver
> >> successfully even the class of QMI is also 0xff.
> >
> >That's not guaranteed so RMNET mode could break depending on probe
> >order with the above entries.
> >
> >> Do you mean I should add RSVD(0) to reduce confusion?
> >
> >You need to reserve it for correctness (or restructure the entries in
> >some other way to achieve the same result).
> >
> Yes, so I re-send a V2 version to reserve interface 0,1,6. Please help take a look on that.
There's no need to reserve interface 1 since it's not vendor class in
any configuration. I've already fixed up and applied your v2.
Also, in the future, please make sure to wrap the lines of your mails at
72 column or so.
Johan
