790199 - RIL: mozMobileConnection.data.shortName is empty with "Free Mobile"

Status: RESOLVED DUPLICATE of bug 785988

(Core :: DOM: Device Interfaces, defect)

Description

[Fabien Cazenave [:kaze]]

Attachment: this is not a fix!

Latest Gecko + Gaia.
Carrier = Free Mobile w/ data plan (mcc/mnc = 208/15).
DEBUG_ALL = true in ril_consts.js

An « adb logcat | grep Operator » shows that 'shortName' is empty:

  Free,,20815

Here’s an excerpt of |adb logcat -b radio|:

D/RILC    (  110): UI --- RIL_REQUEST_OPERATOR (22) ---> RIL [RID 0, token id 169, data len 0]
D/RILC    (  110): RID 0 voice srv: modem id=0, ma=Multimode(0), net_pref=GSM WCDMA preferred(0)
D/RILC    (  110): voip_supported = 0
D/RILC    (  110): Lookup NITZ info
D/RILC    (  110): whh  :current plmn 2 f8 51 
D/RILC    (  110): whh  :manual plmn 2 f8 51 
D/RILC    (  110): plmn1 mcc = 2 0 8, mnc = 1 5 15
D/RILC    (  110): plmn2 mcc = 2 0 8, mnc = 1 5 15
D/RILC    (  110): ONS from saved NITZ
D/RILC    (  110): Reply to RIL --> Current operator full name : Free
D/RILC    (  110): Reply to RIL --> Current operator short name : 
D/RILC    (  110): Reply to RIL --> Current operator MCCMNC : 20815
D/RILC    (  110): UI <--- RIL_REQUEST_OPERATOR (22) Complete --- RIL [RID 0, Token 169, Success, Len 12 ]

FTR, sometimes I even get '20815' as longName.

Reproduced on several Otoro devices, with several "Free Mobile" SIM cards.
The problem does not appear when I test my SIM card with :lissyx’s Nexus S device.

The problem is, if 'shortName' is empty:
 * there’s no carrier name to display in Gaia;
 * mozMobileConnection.data.mcc and .mnc are '0', because that's how it works with _processOperator() in ril_worker.js:

    if (this.operator.longName !== longName ||
        this.operator.shortName !== shortName ||
        thisTuple !== networkTuple) {

      this.operator.longName = longName;
      this.operator.shortName = shortName;
      this.operator.mcc = 0;
      this.operator.mnc = 0;

With the dirty patch (attached) most symptoms are “fixed” in Gaia: the carrier name is properly displayed, and mcc/mnc have non-null values. Of course, this is nowhere near a proper fix.

Comment 1

[Fabien Cazenave [:kaze]]

FWIW, the 'shortName' is properly retrieved on an HTC Desire Z (Android 2.3 GB).

  D/RILJ    ( 1455): [0072]< OPERATOR {Free, Free, 20815, 0}

Comment 2

[Fabien Cazenave [:kaze]]

Note: without a fix, it’s almost impossible to debug the 3G/Data connection on my device — at least with my current operator.

Comment 3

[:gerard-majax]

If it can help, I found the Android part that seems to handle his:
src/java/com/android/internal/telephony/cdma/CdmaServiceStateTracker.java, in handlePollStateResultMessage(). Code is available as a git clone from https://android.googlesource.com/platform/frameworks/opt/telephony

Comment 4

[:gerard-majax]

Maybe also having a look at src/java/com/android/internal/telephony/OperatorInfo.java and the "public static final Creator<OperatorInfo> CREATOR" variable.

Comment 5

[Philipp von Weitershausen [:philikon]]

(In reply to Fabien Cazenave (:kaze) from comment #0)
> An « adb logcat | grep Operator » shows that 'shortName' is empty:
> 
>   Free,,20815

That's something the operator (most likely) is responsible for, not us.

> FTR, sometimes I even get '20815' as longName.

That's worrying. This may or may not be a separate bug. Let's keep this in mind and file a separate bug when appropriate.

> The problem is, if 'shortName' is empty:
>  * there’s no carrier name to display in Gaia;

IMHO Gaia should do the "non-fix" that you attached here:

  var carrierName = shortName || longName;

>  * mozMobileConnection.data.mcc and .mnc are '0', because that's how it
> works with _processOperator() in ril_worker.js:
> 
>     if (this.operator.longName !== longName ||
>         this.operator.shortName !== shortName ||
>         thisTuple !== networkTuple) {
> 
>       this.operator.longName = longName;
>       this.operator.shortName = shortName;
>       this.operator.mcc = 0;
>       this.operator.mnc = 0;

I'm probably missing something, but I fail to see how this code prevents us from setting mcc and mnc correctly, since further down [1] we end up processing the network tuple for mcc and mnc. Can you provide more debugging info?

[1] https://mxr.mozilla.org/mozilla-central/source/dom/system/gonk/ril_worker.js#2679

Comment 6

[Philipp von Weitershausen [:philikon]]

(In reply to Philipp von Weitershausen [:philikon] from comment #5)
> >  * mozMobileConnection.data.mcc and .mnc are '0', because that's how it
> > works with _processOperator() in ril_worker.js:
> > 
> >     if (this.operator.longName !== longName ||
> >         this.operator.shortName !== shortName ||
> >         thisTuple !== networkTuple) {
> > 
> >       this.operator.longName = longName;
> >       this.operator.shortName = shortName;
> >       this.operator.mcc = 0;
> >       this.operator.mnc = 0;
> 
> I'm probably missing something, but I fail to see how this code prevents us
> from setting mcc and mnc correctly, since further down [1] we end up
> processing the network tuple for mcc and mnc. Can you provide more debugging
> info?

Oh yes, I think I know what I was missing: bug 785988 just landed. It should fix that particular problem for you.

Comment 7

[Vicamo Yang [:vicamo][:vyang]]

(In reply to Philipp von Weitershausen [:philikon] from comment #5)
> (In reply to Fabien Cazenave (:kaze) from comment #0)
> > FTR, sometimes I even get '20815' as longName.
> 
> That's worrying. This may or may not be a separate bug. Let's keep this in
> mind and file a separate bug when appropriate.

I didn't see this happening, but if it does, though I think it can't be considering the underlying AT command/Parcel parsing logic, please file another bug for it :)

> > The problem is, if 'shortName' is empty:
> >  * there’s no carrier name to display in Gaia;
> 
> IMHO Gaia should do the "non-fix" that you attached here:
> 
>   var carrierName = shortName || longName;

Already done in Gaia: https://github.com/mozilla-b2g/gaia/commit/33a9529f
However, there are still several places, cost control service and statusbar, do not handle empty short names well. Anyway, I think that's beyond Gecko's reach and I'll mark this as duplicate.

Comment 8

[Vicamo Yang [:vicamo][:vyang]]

(In reply to Vicamo Yang [:vicamo][:vyang] from comment #7)
> (In reply to Philipp von Weitershausen [:philikon] from comment #5)
> > IMHO Gaia should do the "non-fix" that you attached here:
> > 
> >   var carrierName = shortName || longName;
> 
> Already done in Gaia: https://github.com/mozilla-b2g/gaia/commit/33a9529f
> However, there are still several places, cost control service and statusbar,
> do not handle empty short names well.

Gaia issue opened for above problems: https://github.com/mozilla-b2g/gaia/issues/4599

Comment 9

[:gerard-majax]

(In reply to Philipp von Weitershausen [:philikon] from comment #5)
> (In reply to Fabien Cazenave (:kaze) from comment #0)
> > An « adb logcat | grep Operator » shows that 'shortName' is empty:
> > 
> >   Free,,20815
> 
> That's something the operator (most likely) is responsible for, not us.
> 

As far as I remember of the experiments we did with Kazé, and as stated in comment #1, with his SIM in my HTC Desire Z,on the same network, the RIL returned the correct shortName.

Comment 10

[Fabien Cazenave [:kaze]]

(In reply to Alexandre LISSY :gerard-majax from comment #9)
> As far as I remember of the experiments we did with Kazé, and as stated in
> comment #1, with his SIM in my HTC Desire Z,on the same network, the RIL
> returned the correct shortName.

I confirm.

Comment 11

[Vicamo Yang [:vicamo][:vyang]]

(In reply to Fabien Cazenave (:kaze) from comment #10)
> (In reply to Alexandre LISSY :gerard-majax from comment #9)
> > As far as I remember of the experiments we did with Kazé, and as stated in
> > comment #1, with his SIM in my HTC Desire Z,on the same network, the RIL
> > returned the correct shortName.
> 
> I confirm.

The Android rild parses the response of AT command COPS and separates it into three strings, long name, short name and mcc/mnc. The three strings are then passed to Gecko directly. As a result, if rild doesn't have correct info, neither does Gecko. There is probably a bug in rild or modem firmware of the device under test.

Comment 12

[Fabien Cazenave [:kaze]]

Vicamo: of course bug 785988 will reduce the symptoms of the empty 'shortName' property, but I don’t think it’s a duplicate.

I tested my SIM card in other Android devices, and I could check with ADB that 'shortName' was properly set.

Comment 13

[Fabien Cazenave [:kaze]]

Ooops, I didn’t see your last comment before posting mine.

Yes, it looks like a rild bug rather than a Gecko one. Should we open another bug?