Re: [PATCH] option: Improve Quectel EP06 detection

[Lars Melin <larsm17@gmail.com>]

On 9/10/2018 18:39, Kristian Evensen wrote:
> Hi,
> 
> On Mon, Sep 10, 2018 at 12:30 PM Johan Hovold <johan@kernel.org> wrote:
>> Please provide the output of usb-devices (or lsusb -v) for both
>> "configurations". How do you update the configuration by the way?
> 
> The configuration is updated using a proprietary AT-command
> (AT+QCFG="usbcfg"). The format of the command is as follows:
> AT+QCFG="usbcfg",<vid>,<pid>,<diag>,<nmea>,<at_port>,<modem>,<rmnet>,<adb>.
> In other words, you set which interfaces to enable/disable. Based on
> my testing, it is only possible to enable/disable diag, rmnet (QMI)
> and adb, as well as nmea, at_port and modem together. I.e., it is not
> possible to only disable for example nmea.
> 

> If I for example disable diag, then the bInterfaceNumber of nmea
> changes from 1 to 0, at from 2 to 1, etc., etc.
> 
> BR,
> Kristian
>

This also becomes a mess for the qmi-wwan driver which has the rmnet/qmi 
interface hardcoded to 4 so that driver will also need a workaround.
Quectel seems to have completely missed the reason why usb id's should 
be unique and not reused  for a different product or a different 
interface layout, there is already a workaround in qmi-wwan for their 
previous EC-20 card...
My opinion is that the option and qmi-wwan drivers should support EP06 
in the factory delivery configuration and not in a configuration the 
user has selected with a Quectel proprietary AT cmd.
Can you give some good reason for disabling an interface instead of 
letting it stay but not use it if you don't need it?


Thanks
/Lars