PDP settings for LTE only network

[Aleksander Morgado]

On Thu, Jun 16, 2022 at 4:21 PM Reinhard Speyerer <rspmn at arcor.de> wrote:
>
> On Wed, Jun 15, 2022 at 09:12:22PM +0200, Aleksander Morgado wrote:
>
> > > > > Is there a similar command or NM setting to set the modem UE Usage Setting to data-centric mode?
> > > > > Equivalent to: AT+QNVFW="/nv/item_files/modem/mmode/ue_usage_setting",01
> > > > >
> > > > > That command came from a Quectel document for AT&T 3G Sunset Guide.
> > > > >
> > > >
> > > > mmcli has --3gpp-set-eps-ue-mode-operation=ps2 for that, but it will
> > > > only work if the modem supports +CEMODE; not all modems I've found
> > > > support it. And even so, I'm not sure if the CEMODE command would be
> > > > fully equivalent to that change in the settings with QNVFW, maybe ask
> > > > Quectel about that?
> > >
> > > Hi Aleksander,
> > >
> > > on all Qualcomm firmwares I have seen so far AT+CEMODE only supports
> > > reading the setting but not changing it.
> >
> > Could be the case; when I implemented it I was using an Intel-based
> > TOBY-L4, which did support both reading and setting.
> >
> > > The TS 27.007 command
> > > corresponding to the above AT+QNVFW command seems to be AT+CEUS
> > > which typically isn't implemented at all.
> > >
> > > However NASSetSystemSelectionPref TLV 0x21 which is already part
> > > of the original Gobi API usually seems to work quite well.
> > >
> >
> > Oh, thanks for the hint!
> >
> > So that would be
> > qmi_message_nas_set_system_selection_preference_input_set_usage_preference()
> > with QMI_NAS_USAGE_PREFERENCE_DATA_CENTRIC.
> >
> > https://gitlab.freedesktop.org/mobile-broadband/libqmi/-/blob/main/data/qmi-service-nas.json#L1530
> >
> > I think we could update ModemManager to support that.
> >
> > Looks like AT+CEUS and the QMI variant are a subset of what AT+CEMODE
> > allows, right? Or am I missing something?
>
> Hi Aleksander,
>
> you would have to combine TLV 0x21 with TLV 0x18 ("Service Domain Preference")
> for switching between PS mode and CS/PS mode. However for QMI this TLV applies
> to all RATs and therefore the result might differ from AT+CEMODE which
> according to TS 27.007 only applies to EPS.
>

Hum; yeah, that's true, +CEMODE is only for EPS, which is why the
method/property in the MM API is
SetEpsUeModeOperation()/EpsUeModeOperation. The MM APIs were written
too bound to the +CEMODE command interpretation really, maybe not the
best thing.

A more generic API like the one in QMI with service domain preference
and usage preference could make more sense indeed, and I don't think
it would be very complex to do. It's true that we don't want to have
every possible QMI setting in the MM APIs, but this one seems simple
enough I think, and useful. What do you think?

-- 
Aleksander
https://aleksander.es