Re: Quectel EC200T USB: Problems with context activation

[Denis Kenzior <denkenz@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 3433 bytes --]

Hi Sergei,

On 5/19/21 6:35 PM, Sergei Golubtsov wrote:
> Hi Denis,
> 
> Thank you very much for your quick and helpful reply.
> 
> The problem is a bit vulgar actually. That's a shame that I took your
> time for that but I have another question as I am not sure about the
> findings below.

no worries.

> 
> I had tried to get dump with gsmdial and gsmdial successfully
> activated the context. So the problem is that we have authentication
> method set to NONE in ofono. But gsmdial uses CHAP by default.
> I have manually set the auth to CHAP for the context in ofono and
> ofono successfully activated the context.

Yes, oFono used to use CHAP by default, but this was changed a few years ago.

> There is the provider info from mobile-broadband-provider-info about
> my provider:
> 
>      <provider>
>      <name>China Unicom</name>
>      <gsm>
>      <network-id mcc="460" mnc="01"/>
>      <apn value="3gnet">
>      <plan type="postpaid"/>
>      <usage type="internet"/>
>      <username>uninet</username>
>      </apn>
>      <apn value="3gwap">
>      <usage type="mms"/>
>      <name>联通彩信</name>
>      <mmsc>http://mmsc.myuni.com.cn</mmsc>
>      <mmsproxy>10.0.0.172:80</mmsproxy>
>      </apn>
>      </gsm>
>      </provider>
> 
> 
> Am I correct that ofono must use CHAP if the auth method is not
> specified in the db?

We used to do this, then around Sep-Oct 2018 there was a strong push to add an 
explicit 'No Authentication' option for oFono from one of the modem hardware 
vendors.  I don't recall the exact arguments, search the archives, but the 
fallout was that an empty username / password implied no authentication.  Which 
I think does make a certain amount of sense.

git show 6cf24fe1f9cfa2a61422ad84abfdd32e7ea2cf78.  Look around at the commits 
from the same author.  I do seem to recall that 3GPP mandated CHAP to be used as 
the default, even in the no username / password case.  But my memory is fuzzy 
now, and there hasn't been any problems reported since then (until now).

> 
> I see that the technology used by the modem is HSPA. And I see the
> following in ofono/drivers/atmodem/lte.c:
> 
>      /* change the authentication method if the  parameters are invalid */
>      if (!*ldd->pending_info.username || !*ldd->pending_info.password)
>         auth_method = OFONO_GPRS_AUTH_METHOD_NONE;
> 
> 
> And in ofono/plugins/mbpi.c:
> 
>      /* select authentication method NONE if fit */
>      if (!ap->username || !ap->password)
>      ap->auth_method = OFONO_GPRS_AUTH_METHOD_NONE;
> 
> 
> As well as ofono/plugins/fileprovision.c::
> 
>      /* select default authentication method */
>      (*settings)[0].auth_method = OFONO_GPRS_AUTH_METHOD_NONE;
> 
> 
> ofono/src/lte.c:
> 
>      /* this must have a valid default */
>     if (!gprs_auth_method_from_string(auth_method_str,
>     &lte->info.auth_method))
>     lte->info.auth_method = OFONO_GPRS_AUTH_METHOD_NONE;
> 
> 
> I am not sure about the standards which may be relevant here. Sorry
> for asking this but I thought that we should use CHAP by default,
> don't we?

The question is really whether it is the network or the modem at fault here. 
Does AUTH_METHOD_NONE work on other networks with this hardware?

> 
> Thank you again and have a nice day.
> 
> Yours sincerely,
> Sergei Golubtsov.
> 

Regards,
-Denis