problems with QMI dual stack connections from MM - works with qmicli

Tue Apr 27 15:01:50 UTC 2021

On Tue, Apr 27, 2021 at 02:10:12PM +0200, Bjørn Mork wrote:
> Reinhard Speyerer <rspmn at arcor.de> writes:
> 
> > On Mon, Apr 26, 2021 at 06:23:01PM +0200, Bjørn Mork wrote:
> >> Bjørn Mork <bjorn at mork.no> writes:
> >> 
> >> > Sorry, i do not think these issues are related to ModemManager at all.
> >> 
> >> This is a test from freshly booted modem, MM not running:
> >> 
> >> [...]
> >> root at finn:~# qmicli -p -d /dev/cdc-wdm0 --client-no-release-cid --client-cid=17 --wds-start-network=apn=telenor,ip-type=4
> >> error: couldn't start network: QMI protocol error (14): 'CallFailed'
> >> call end reason (1): generic-unspecified
> >> verbose call end reason (2,236): [internal] call-already-present
> >> [/dev/cdc-wdm0] Client ID not released:
> >>         Service: 'wds'
> >>             CID: '17'
> >> root at finn:~# qmicli -p -d /dev/cdc-wdm0 --client-no-release-cid --client-cid=17 --wds-start-network=apn=internet.public,ip-type=4
> >> error: couldn't start network: QMI protocol error (14): 'CallFailed'
> >> call end reason (1): generic-unspecified
> >> verbose call end reason (2,236): [internal] call-already-present
> >> [/dev/cdc-wdm0] Client ID not released:
> >>         Service: 'wds'
> >>             CID: '17'
> >> [...]
> >> 
> >> I don't think there is much MM can do here.  This modem firmware does
> >> not work properly with QMAP muxing.  Sorry about the confusion I've
> >> caused by assuming this would work.
> >> 
> >> Unless I am missing something in the sequence above?
> >> 
> >
> > Hi Bjørn,
> >
> > I'm not sure about the qmap5 vs. qmap part but assuming that none of
> > the "problematic" APNs matches the APN assigned to the default bearer
> > for EPS attach I would expect for this to be the correct QMI sequence
> > to use.
> >
> > Can you crosscheck with AT+CGDCONT? and AT+CGACT? whether the corresponding
> > contexts are really active on the UE as the QMI error suggests? If they are
> > the next step would perhaps be to check AT+QMBNCFG="List" for a Telenor
> > profile and the AT+QMBNCFG="AutoSel" setting in case you are still using the
> > Quectel RM502Q for testing this.
> 
> Thanks.  Yes, same modem.  I've seen the discussions wrt those Quectel
> carrier config files, but sort of ignored it as "doesn't affect me".
> But you're of course right that I should verify that assumtion.
> 
> This is the AT command output after repeating the qmicli connections up
> to and including the failed "internet.public" connection quoted above:
> 
> 
> AT+CGDCONT?
> +CGDCONT: 1,"IPV4V6","telenor.smart","0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0",0,0,0,0
> +CGDCONT: 2,"IPV4V6","ims","0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0",0,0,0,0
> +CGDCONT: 3,"IPV4V6","sos","0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0",0,0,0,1
> +CGDCONT: 4,"IP","internet.public","0.0.0.0",0,0,0,0
> 
> OK
> AT+CGACT?
> +CGACT: 1,1
> +CGACT: 2,1
> +CGACT: 3,0
> +CGACT: 4,0
> 
> OK
> AT+QMBNCFG="List"
> +QMBNCFG: "List",0,0,1,"ROW_Commercial",a010809,202001061
> +QMBNCFG: "List",1,0,0,"Telia_Sweden",a012400,201908071
> +QMBNCFG: "List",2,0,0,"TIM_Italy_Commercial",a012b00,201908071
> +QMBNCFG: "List",3,0,0,"France-Commercial-Orange",a010b21,201908071
> +QMBNCFG: "List",4,0,0,"Commercial-DT-VOLTE",a011f1f,201908071
> +QMBNCFG: "List",5,0,0,"UK-VoLTE-Vodafone",a010426,201908071
> +QMBNCFG: "List",6,0,0,"Commercial-EE",a01220b,201908071
> +QMBNCFG: "List",7,0,0,"Optus_Australia_Commercial",a014400,201908071
> +QMBNCFG: "List",8,0,0,"Telstra_Australia_Commercial",a010f00,202007081
> +QMBNCFG: "List",9,0,0,"Commercial-LGU",a012608,201908071
> +QMBNCFG: "List",10,0,0,"Commercial-KT",a01280b,201908071
> +QMBNCFG: "List",11,0,0,"Commercial-SKT",a01270a,201908071
> +QMBNCFG: "List",12,0,0,"Commercial-Reliance",a011b0c,202006011
> +QMBNCFG: "List",13,0,0,"Commercial-SBM",a011c0b,201908071
> +QMBNCFG: "List",14,0,0,"Commercial-KDDI",a010709,201908071
> +QMBNCFG: "List",15,0,0,"Commercial-DCM",a010d0d,201908071
> +QMBNCFG: "List",16,0,0,"VoLTE-CU",a011561,202001061
> +QMBNCFG: "List",17,0,0,"VoLTE_OPNMKT_CT",a0113e0,202001061
> +QMBNCFG: "List",18,0,0,"Volte_OpenMkt-Commercial-CMCC",a012010,202001061
> 
> OK
> AT+QMBNCFG="AutoSel"
> +QMBNCFG: "AutoSel",1
> 
> OK
> 
> 
> So I think the carrier configuration is "generic" and therefore OK?
> 
> Note that the "internet.public" profile (4) was added after the last
> connection test, as a side effect of testing Aleksanders profile
> management branch.  It doesn't seem to affect the end result though.
> 
> Other than that, I notice that the "ims" profile is connected.  That's a
> bit surprising to me, as I've never touched that.  I guess it could be
> an automatic thing from the "ROW_Commercial" config?  How would we know?
> Or it can be a leftover being configured by the OEM firmware on the
> router this modem is stuck in.  Although I don't find that very likely.
> I like to believe I know which APNs that firmware connects to.
> 

Hi Bjørn,

from the observed behaviour AT+QMBNCFG="AutoSel",1 seems to have a similar
semantics as the Sierra Wireless PRI "AUTO-SIM" feature discussed recently
without the need for a device reboot. Depending on the IMSI and/or other
properties of your SIM AT+CGDCONT entries may be created/overwritten
and certain contexts e.g. for IMS may automatically be activated.

Obtaining S1AP traces in the network or QCDM logs on the device and checking
the NAS EPS signalling might be a good way to validate your assumption about
the firmware vs APNs above.

If that doesn't explain the effects observed at the QMI level you could try
AT+QCFG="volte_disable",1 (if it is set to 0)
AT+QMBNCFG="AutoSel",0
AT+QMBNCFG="Deactivate"
power cycle the modem and check if anything changes.

Regards,
Reinhard