No location provided (how to debug)

[AllesMeins]

Unfortunately I can't get the location backend to work and I'm a bit lost to find the reason. I've installed the latest UnifiedNlp (GAPPS Version, since I've playstore installed), the XposedUnifiedNlp Module and the IchnaeaNlpBackend 1.4.0 configured and activated. According to the SelfTest everything is in order and should work but "No UnifiedNlp location was provided by the system within 10 seconds".

Now I'm a bit lost on how to find the problem. There is no error message and logcat doesn't show any related errors. Are there any other ways to debug this problem? Or is this a known behaviour?

Im using Lineage 16.0 on an Sony Xperia XZ1Compact

[baldrianbandit]

I'm running Lineage 16 w/o Gapps on a XZ1C.

There is a known bug that makes IchnaeaNlpBackend find no cells, see [https://github.com//issues/38](this issue). With wifi disabled and with an app in foreground that queries network location, i.e. SatStat
adb logcat | grep Ichnaea
results in
IchnaeaBackendService: Cells: 0

The backend does resolve location by wifi, though. With wifi enabled, you should get IchnaeaBackendService: request: {"wifiAccessPoints":[<wifi access points around you>])
IchnaeaBackendService: response: {"location": {"lat": <lat>, "lng": <lng>}, "accuracy": <acc>}

[AllesMeins]

Thanks for the reply, but this doesn't seem to be the problem. I checked the log and there are access points around me in there and even my accurate location. But somehow this location does get lost somewhere along the way. I've added two logs - one is filtered by "nlp" and the other one by "Ichnaea". In both I removed my location and replaced it with XX,XXXXXX and YY,YYYYYY
logcat_nlp.txt
logcat_Ichnaea.txt

[domkrm]

I have the same problem and after some analysing, testing and searching I found the following issue tracker entry: https://issuetracker.google.com/issues/113560852
It seems that this is a bug since Android 9 in combination with 0-prefixed MNCs (e.g. "01").
The last comment is "The issue has been fixed and it will become be available in a future Android release". I looked into the Android 10 source code and the bug still exists.

[Iey4iej3]

I have the same problem and after some analysing, testing and searching I found the following issue tracker entry: https://issuetracker.google.com/issues/113560852
It seems that this is a bug since Android 9 in combination with 0-prefixed MNCs (e.g. "01").
The last comment is "The issue has been fixed and it will become be available in a future Android release". I looked into the Android 10 source code and the bug still exists.

This seems to be fixed upstream but not fixed in the source code of LineageOS 17 where I have just committed.

[domkrm]

This seems to be fixed upstream but not fixed in the source code of LineageOS 17 where I have just committed.

Yes, I see this too. But I don't think this is fixed correctly. It's still checked if length of MNC is between 2 and 3.

private static final int MNC_MIN_LENGTH = 2;
private static final int MNC_MAX_LENGTH = 3;

if (mnc.length() < MNC_MIN_LENGTH || mnc.length() > MNC_MAX_LENGTH) return false;

I know that MNC must be 2- or 3-digits. But since the MNC is passed as INT to the constructors of the subclasses the 0-prefixed MNCs will be translated into 1-digit MNCs.

public CellIdentityLte(int mcc, int mnc, int ci, int pci, int tac) {
    this(ci, pci, tac, CellInfo.UNAVAILABLE, CellInfo.UNAVAILABLE, String.valueOf(mcc),
            String.valueOf(mnc), null, null);
}

[Iey4iej3]

This seems to be fixed upstream but not fixed in the source code of LineageOS 17 where I have just committed.

Yes, I see this too. But I don't think this is fixed correctly. It's still checked if length of MNC is between 2 and 3.

private static final int MNC_MIN_LENGTH = 2;
private static final int MNC_MAX_LENGTH = 3;

if (mnc.length() < MNC_MIN_LENGTH || mnc.length() > MNC_MAX_LENGTH) return false;

I know that MNC must be 2- or 3-digits. But since the MNC is passed as INT to the constructors of the subclasses the 0-prefixed MNCs will be translated into 1-digit MNCs.

public CellIdentityLte(int mcc, int mnc, int ci, int pci, int tac) {
    this(ci, pci, tac, CellInfo.UNAVAILABLE, CellInfo.UNAVAILABLE, String.valueOf(mcc),
            String.valueOf(mnc), null, null);
}

Maybe you need to add a comment there. By the way, I don't know what @UnsupportedAppUsage(maxTargetSdk = Build.VERSION_CODES.P, trackingBug = 115609023) means (the line preceding the construction method that you have quoted).

You can also commit at the LOS repo.

[domkrm]

Maybe you need to add a comment there. By the way, I don't know what @UnsupportedAppUsage(maxTargetSdk = Build.VERSION_CODES.P, trackingBug = 115609023) means (the line preceding the construction method that you have quoted).

You can also commit at the LOS repo.

Sorry, I overlooked the second constructor in the subclasses.

public CellIdentityLte(int ci, int pci, int tac, int earfcn, int bandwidth, String mccStr,
        String mncStr, String alphal, String alphas) {
    super(TAG, CellInfo.TYPE_LTE, mccStr, mncStr, alphal, alphas);
...
}

Here the MNC is passed as STRING to the constructor.

But I don't get it where these constructors get called.

@UnsupportedAppUsage means that App developers shouldn't rely on the following method, because it's only for internal use and could be changed anytime.

[Iey4iej3]

There seems a lot of constructors. By the way, do you know how often these codes are merged into LOS repo?

[domkrm]

There seems a lot of constructors. By the way, do you know how often these codes are merged into LOS repo?

No, sorry. It's the first time I deal with the Android/LOS internal source code.

[kurt-by]

I have the same problem and after some analysing, testing and searching I found the following issue tracker entry: https://issuetracker.google.com/issues/113560852
It seems that this is a bug since Android 9 in combination with 0-prefixed MNCs (e.g. "01").
The last comment is "The issue has been fixed and it will become be available in a future Android release". I looked into the Android 10 source code and the bug still exists.

This seems to be fixed upstream but not fixed in the source code of LineageOS 17 where I have just committed.

Sorry, that's not how pull requests for LineageOS are done. They have ther code review in https://review.lineageos.org/q/status:open

[anonimno1]

Hi people,
anything new concerning this bug?
I am on LineageOS 16 for uG and none of the NLP backends (Mozilla, DejaVu) work with cell identification.
Due to the fact that this seems to be an Andoroid 9 bug switching to another ROM will not get this working, right?
Regards
K

[baldrianbandit]

Hi people,
Regards
K

Switching to microG and it's shipped UnifiedNlp can do the trick, at least when using LocalGSM network provider.
Refer to this post in the Sony XZ1C LineageOS thread.

[domkrm]

Hi people,
Regards
K

Switching to microG and it's shipped UnifiedNlp can do the trick, at least when using LocalGSM network provider.
Refer to this post in the Sony XZ1C LineageOS thread.

LocalGSM provider should always work.
I take a look in the source code.
If it can't get a valid cell information then another approach is used to get a valid cell information.