Re: [PATCH 7/8] sierra: Support CDMA modems

[Denis Kenzior <denkenz@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 2137 bytes --]

Hi Forest,

>>> I'm looking to pick this series back up now.  I understand the implementation is
>>> not ideal, but it at least provides for basic connectivity until we have support
>>> for the CnS protocol.  Will you accept the full series if I address the other
>>> issues with the code that you mentioned?
>>
>> If we accept something upstream we assume full responsibility for
>> the code, and everything that entails, including future maintenance.
>> Thus we do not have a habit of accepting functionality that is
>> inherently broken.  So unless you can convince me otherwise, my
>> answer is: "I'd rather not" ;)
>
> Well I think the argument to make is that we ought to be able to gracefully
> degrade to handle a device that we have limited support for.  We know how to
> bring up a connection on this device.  I think users would prefer to have a
> device that works with a limited feature set than one that doesn't work at all.

Don't get me wrong, we do want to enable as many devices as possible... 
The problem is that the oFono clients do not know when the device is in 
'degraded' mode.  Our assumption was that we always have multiple AT 
ports available or we are using high-speed interfaces.  If the AT chat 
is blocked by PPP, then we can't even get basic things like signal 
strength notifications while on a data connection.  To me that is 
unacceptable, especially since we know that any AT command based 
implementation is a long-term dead end.

>
> Anyway, I notice that one of the ports appears to be QCDM.  It appears that QCDM
> support is partially implemented in oFono?  But I don't know QCDM at all.  Can
> we use it to do netreg?

oFono QCDM support is quite bare, not much beyond a test tool. 
Contributions are always welcome!

>
> If not, I could probably implement (partial) CnS support at some point.  I'm
> just not sure I will have time to do it right away.
>

That would be great.

> I assume a proper netreg implementation on one of these other ports would
> address your primary concern?

Yep, correct.

>
> Thanks,
> Forest

Regards,
-Denis