Mapbox memory leak crashes app when LeakCanary is enabled

[cvance]

Not entirely sure if this is a Mapbox bug (or Mapzen or LeakCanary or a bug in my code) but I believe it is. Either way this seems to stem from a memory leak in Mapbox.

Platform: Android
Mapbox SDK version: 5.0.0

Steps to trigger behavior

1. Add and enable LeakCanary to a project using Mapbox 5.0.0
2. Open an activity with a mapview, location engine and location engine listener
3. Background the app

Expected behavior

LeakCanary catches a memory leak from Mapbox

Actual behavior

LeakCanary crashes the app. Disabling LeakCanary seems to resolve this crash.

java.lang.ClassCastException: android.os.BinderProxy cannot be cast to com.mapzen.android.lost.internal.FusedLocationProviderService$FusedLocationProviderBinder
                                                                                    at com.mapzen.android.lost.internal.FusedLocationProviderApiImpl.onServiceConnected(FusedLocationProviderApiImpl.java:45)
                                                                                    at com.mapzen.android.lost.internal.FusedLocationServiceConnectionManager.onServiceConnected(FusedLocationServiceConnectionManager.java:79)
                                                                                    at com.mapzen.android.lost.internal.FusedLocationProviderApiImpl.onServiceConnected(FusedLocationProviderApiImpl.java:65)
                                                                                    at android.app.LoadedApk$ServiceDispatcher.doConnected(LoadedApk.java:1453)
                                                                                    at android.app.LoadedApk$ServiceDispatcher$RunConnection.run(LoadedApk.java:1481)
                                                                                    at android.os.Handler.handleCallback(Handler.java:751)
                                                                                    at android.os.Handler.dispatchMessage(Handler.java:95)
                                                                                    at android.os.Looper.loop(Looper.java:154)
                                                                                    at android.app.ActivityThread.main(ActivityThread.java:6119)
                                                                                    at java.lang.reflect.Method.invoke(Native Method)
                                                                                    at com.android.internal.os.ZygoteInit$MethodAndArgsCaller.run(ZygoteInit.java:886)
                                                                                    at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:776)

[tobrun]

@cvance can you share the code of the related to mapview, location engine and location engine listener?

[cvance]

I'm able to reproduce the crash using this sample activity, LeakCanary 1.5 and Mapbox SDK 5.0 and Mapbox UI 2.0 on my Nexus 6P (7.1.2beta)

[Guardiola31337]

Same issue reported in MAS.

[Emeritus-DarranKelinske]

I fixed this by removing all location permission from the manifest. There is a suggestion to use tools:remove in the manifest in the upstream bug, but I have not figured that out yet.

[tobrun]

@cvance, would you be able to retest this with latest SNAPSHOT? I'm thinking that #8825 might have resolved the issue (not able to reproduce with that code).

Installation instructions for SNAPSHOT:

repositories {
    mavenCentral()
    maven { url "http://oss.sonatype.org/content/repositories/snapshots/" }
}

dependencies {
    compile ('com.mapbox.mapboxsdk:mapbox-android-sdk:5.1.0-SNAPSHOT@aar') {
        transitive=true
    }
}

[cvance]

I'm able to reproduce still with 5.1.0-SNAPSHOT. Side note, it also appears to crash Pixel XL on launch, somewhere in native code.

[tobrun]

In #8872, I'm updating LOST to the latest 2.30-SNAPSHOT.
This should include the fix that addresses this issue.

[tobrun]

I'm going to close this issue, so we have more clarity on the milestone related to the LOST issue. I'm opting to track this in #9139 as we have a PR tackling that in #9142. In this PR we working on resolving the current outstanding upstream issues (lostzen/lost#213). Thank you all for chipping in, continuing this issue in #9139.