aleksander/cdma-at-activation review

[Aleksander Morgado]

Hey,

Took some time to get back to this, sorry.

>
> - The Novatel manual activation process looks like it's Sprint-only at
> the moment (eg, uses IOTA)?  Was that intentional?  Is Verizon
> activation supposed to use the AT+CDV bits?
>

Novatel is manual-activation-only (e.g. for Sprint), Sierra is
automatic-activation-only (e.g. for Verizon). I'm sure you can also
use manual activation for Verizon, but I didn't implement that as
automatic activation seems to do the job.


> - For the Sierra manual activation, after issuing IOTASTART the Encore
> networks PDF states you need to listen for both OK and then NO CARRIER,
> and only after NO CARRIER do you proceed with the namval.  I can't see
> anything that waits for NO CARRIER, did you find that waiting was not
> required?
>

All in all, I implemented the Sprint activation support in the Sierra
modem, but never got around to test it, because it requires to call
Sprint support asking for the SID/NID and all those parameters...

> - Does anything verify that the 'carrier_code' is not zero-length?  If
> that's an issue, I know it existed before this branch, but just
> checking.
>

I added a new commit to check for that.

> - For Novatel manual, does the "IOTA Enabled" string really mean
> everything is fine?  What does it change to if the activation fails?
>

Cannot tell, as said earlier I couldn't really test it properly.

> - activation_context_retry_step() - FALSE -> G_SOURCE_REMOVE perhaps?
> I'm a big fan of G_SOURCE_* these days :)
>

There's no real need for that; actually I've never seen that done.
Returning FALSE in the callback already removes the GSource from the
mainloop.

> The code itself looks fine to me; I'm happy to double-check on a couple
> Sierra and Novatel devices if somebody wants to send along the codes.
>

Can someone out there provide some Sprint SID/NID/... info to test this?

-- 
Aleksander
https://aleksander.es