NoSuchFieldException: unsubscribed

[yyfrankyy]

java.lang.NoSuchFieldException: unsubscribed
java.lang.Class.getDeclaredField(Class.java:929)
java.util.concurrent.atomic.AtomicIntegerFieldUpdater$AtomicIntegerFieldUpdaterImpl.<init>(AtomicIntegerFieldUpdater.java:251)
java.util.concurrent.atomic.AtomicIntegerFieldUpdater.newUpdater(AtomicIntegerFieldUpdater.java:49)
rx.subscriptions.BooleanSubscription.<clinit>(SourceFile:32)
rx.subscriptions.Subscriptions.create(SourceFile:75)
rx.subjects.SubjectSubscriptionManager.addUnsubscriber(SourceFile:68)
rx.subjects.SubjectSubscriptionManager.call(SourceFile:58)
rx.subjects.SubjectSubscriptionManager.call(SourceFile:35)
rx.Observable.subscribe(SourceFile:7803)
rx.Observable.subscribe(SourceFile:7531)

Happened randomly, I cannot reproduce this.

Machine Info

SM-N900;Android 5.0 LRX21V,level 21

[amitcse]

I too have seen this issue occurring randomly. In my case it was happening while using HystrixCollapser inside map/flatMap operator. On debugging I found that onCompleted was being called on the observable before onNext. I was not able to isolate if it was due to rx operator or Hystrix.

@yyfrankyy What is scenario in your use-case? Can you share your flow of composition ?

[akarnokd]

Hi. Do you use ProGuard by any chance? I'm not an expert on it but you might need to change the field retention policy.

[artem-zinnatullin]

ProGuard detects and keeps fields used by AtomicFieldUpdater since v4.6 (release happened 4 years ago), maybe @yyfrankyy uses another code obfuscator?

[yyfrankyy]

@akarnokd @artem-zinnatullin we do use proguard, but keep all the staff that rx needs by -keep class rx.** { * ; }

@amitcse It crashed at the very beginning since the Application#onCreate(), code is here:

https://github.com/WeTeX/Watchers/blob/master/src/main/java/com/github/QQMail/Watchers.java#L244

And here are some use cases, it works quite well at most of the time.

https://github.com/WeTeX/Watchers/blob/master/src/test/java/com/github/QQMail/WatchersTest.java

I grep all the crashes related to NoSuchFieldException: unsubscribed, all the crashes came from

• SM-G900I;Android 5.0,level 21
• SM-G9006V;Android 5.0,level 21

Some of the crashes look like this, from these two devices. they look so similar..

• GT-I9507V;Android 5.0.1,level 21
• GT-I9500;Android 5.0.1,level 21

java.lang.NoSuchFieldException: requested
java.lang.Class.getDeclaredField(Class.java:929)
java.util.concurrent.atomic.AtomicLongFieldUpdater$CASUpdater.<init>(AtomicLongFieldUpdater.java:251)
java.util.concurrent.atomic.AtomicLongFieldUpdater.newUpdater(AtomicLongFieldUpdater.java:50)
rx.internal.operators.OperatorObserveOn$ObserveOnSubscriber.<clinit>(SourceFile:86)
rx.internal.operators.OperatorObserveOn.rx.Subscriber call(rx.Subscriber)(SourceFile:64)
rx.internal.operators.OperatorObserveOn.java.lang.Object call(java.lang.Object)(SourceFile:44)

[artem-zinnatullin]

IDK, but I see 2 possible reasons:

1. Samsung: probably, they've patched the Runtime (they like to change everything in Android) and it tries to minify fields names for less memory usage (though I don't think that they're really doing such things, but, who knows).
2. ProGuard rules problem or ProGuard bug or Android Gradle Plugin bug when you're doing non-clean builds (though it'll break app on other devices too).

@yyfrankyy can you please wrap the subscribe() call into try-catch and then collect info about the fields of the BooleanSubscription class, it'll help us detect the problem.

try {
  obs.subscribe(...);
} catch (NoSuchFieldException e) {
  // collect info about fields of the BooleanSubscription class
  BooleanSubscription.class.getDeclaredFields(); // include it into the crash report
}

Also, recently I've shipped ProGuard rules for RxJava as aar https://github.com/artem-zinnatullin/RxJavaProGuardRules, you can try to use them instead of yours.

[ppiech]

It's definitely not pro-guard, since the decompiled class code looks correct. I can also try to capture the fields in the class (thank you for the suggestion), but since I haven't reproduced the issue myself it'll take a week before I have that data. In the mean time, I think I'll just replace Atomic_FieldUpdater with Atomic_ objects, since I confirmed that it works.

[yyfrankyy]

@artem-zinnatullin Sure, we can do this on our next release next month, will come back later.

[plastec]

Hi,
I have the same issues. All only on Samsung devices with Android 5

1:
....
Caused by java.lang.NoSuchFieldException: producerIndex
at java.lang.Class.getDeclaredField(Class.java)
at rx.internal.util.unsafe.UnsafeAccess.addressOf(SourceFile)
at rx.internal.util.unsafe.SpmcArrayQueueProducerField.(SourceFile)
at rx.internal.util.RxRingBuffer$2.createObject(SourceFile)
at rx.internal.util.ObjectPool.borrowObject(SourceFile)
at rx.internal.util.RxRingBuffer.(SourceFile)
at rx.internal.util.RxRingBuffer.getSpmcInstance(SourceFile)
at rx.internal.operators.OnSubscribeCombineLatest$MultiSourceProducer.(SourceFile)
at rx.internal.operators.OnSubscribeCombineLatest.call(SourceFile)
at rx.Observable.unsafeSubscribe(SourceFile)
at rx.internal.operators.OperatorSubscribeOn$1$1.call(SourceFile)
at rx.internal.schedulers.ScheduledAction.run(SourceFile)
at rx.schedulers.ExecutorScheduler$ExecutorSchedulerWorker.run(SourceFile)
...

2:
...
Caused by java.lang.NoSuchFieldException: do
at java.lang.Class.getDeclaredField(Class.java)
at java.util.concurrent.atomic.AtomicIntegerFieldUpdater$AtomicIntegerFieldUpdaterImpl.(AtomicIntegerFieldUpdater.java)
at java.util.concurrent.atomic.AtomicIntegerFieldUpdater.newUpdater(AtomicIntegerFieldUpdater.java)
at rx.subscriptions.BooleanSubscription.(SourceFile)
at rx.subscriptions.Subscriptions.create(SourceFile)
at rx.subjects.SubjectSubscriptionManager.addUnsubscriber(SourceFile)
at rx.Observable.subscribe(SourceFile)
at rx.Observable.subscribe(SourceFile)
...

3:
...
Caused by java.lang.NoSuchFieldException: this
at java.lang.Class.getDeclaredField(Class.java)
at java.util.concurrent.atomic.AtomicLongFieldUpdater$CASUpdater.(AtomicLongFieldUpdater.java)
at java.util.concurrent.atomic.AtomicLongFieldUpdater.newUpdater(AtomicLongFieldUpdater.java)
at rx.internal.operators.OnSubscribeCombineLatest$MultiSourceProducer.(SourceFile)
at rx.internal.operators.OnSubscribeCombineLatest.call(SourceFile)
at rx.Observable.unsafeSubscribe(SourceFile)
at rx.internal.operators.OperatorSubscribeOn$1$1.call(SourceFile)
at rx.internal.schedulers.ScheduledAction.run(SourceFile)
at rx.schedulers.ExecutorScheduler$ExecutorSchedulerWorker.run(SourceFile)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java)
at java.lang.Thread.run(Thread.java)

4:
...
Caused by java.lang.NoSuchFieldException: do
at java.lang.Class.getDeclaredField(Class.java)
at java.util.concurrent.atomic.AtomicIntegerFieldUpdater$AtomicIntegerFieldUpdaterImpl.(AtomicIntegerFieldUpdater.java)
at java.util.concurrent.atomic.AtomicIntegerFieldUpdater.newUpdater(AtomicIntegerFieldUpdater.java)
at rx.schedulers.CachedThreadScheduler$EventLoopWorker.(SourceFile)
at rx.schedulers.CachedThreadScheduler.createWorker(SourceFile)
at rx.internal.operators.OperatorObserveOn$ObserveOnSubscriber.(SourceFile)
at rx.internal.operators.OperatorObserveOn.call(SourceFile)
at rx.Observable$2.call(SourceFile)
at rx.Observable.subscribe(SourceFile)
at rx.Observable.subscribe(SourceFile)
...

5:
...
Caused by java.lang.NoSuchFieldException: case
at java.lang.Class.getDeclaredField(Class.java)
at java.util.concurrent.atomic.AtomicLongFieldUpdater$CASUpdater.(AtomicLongFieldUpdater.java)
at java.util.concurrent.atomic.AtomicLongFieldUpdater.newUpdater(AtomicLongFieldUpdater.java)
at rx.internal.operators.OperatorObserveOn$ObserveOnSubscriber.(SourceFile)
at rx.internal.operators.OperatorObserveOn.call(SourceFile)
at rx.Observable$2.call(SourceFile)
at rx.Observable.subscribe(SourceFile)
at rx.Observable.subscribe(SourceFile)
....

Looks like Samsung introduce some specific problem on Android 5.

Do you plan to apply a workaround for that?

[digitalbuddha]

1000+ crashes past few days in New York Times app. Hoping for @markrietveld fix to be merged soon. We are reaching out to Samsung as well to find out exactly what happened on their end.

Package: com.nytimes.android
Version Code: 9486
Version Name: 5.9.2
Android: 5.0.2
Manufacturer: samsung
Model: SM-T805Y
Date: Mon Oct 26 06:31:34 AEDT 2015

java.lang.IllegalStateException: Fatal Exception thrown on Scheduler.Worker thread.
    at rx.internal.schedulers.ScheduledAction.run(SourceFile:62)
    at rx.schedulers.ExecutorScheduler$ExecutorSchedulerWorker.run(SourceFile:98)
    at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1112)
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:587)
    at java.lang.Thread.run(Thread.java:818)
Caused by: java.lang.ExceptionInInitializerError
    at rx.internal.operators.OperatorZip.call(SourceFile:113)
    at rx.internal.operators.OperatorZip.call(SourceFile:58)
    at rx.Observable$2.call(SourceFile:158)
    at rx.Observable$2.call(SourceFile:154)
    at rx.Observable.subscribe(SourceFile:7804)
    at rx.Observable.subscribe(SourceFile:7772)
    at com.nytimes.android.notification.DailyRichNotificationHelper.displayDailyRichNotification(SourceFile:108)
    at com.nytimes.android.notification.DailyRichNotificationHelper.displayDailyRichNotification(SourceFile:99)
    at com.nytimes.android.notification.DailyRichNotificationHelper$1.call(SourceFile:86)
    at com.nytimes.android.notification.DailyRichNotificationHelper$1.call(SourceFile:83)
    at rx.Observable$28.onNext(SourceFile:7588)
    at rx.observers.SafeSubscriber.onNext(SourceFile:130)
    at rx.internal.operators.OperatorMerge$MergeSubscriber.emitScalar(SourceFile:364)
    at rx.internal.operators.OperatorMerge$MergeSubscriber.tryEmit(SourceFile:326)
    at rx.internal.operators.OperatorMerge$InnerSubscriber.onNext(SourceFile:802)
    at rx.internal.operators.OperatorDoOnEach$1.onNext(SourceFile:84)
    at rx.subjects.SubjectSubscriptionManager$SubjectObserver.onNext(SourceFile:224)
    at rx.internal.operators.NotificationLite.accept(SourceFile:150)
    at rx.subjects.ReplaySubject$UnboundedReplayState.accept(SourceFile:466)
    at rx.subjects.ReplaySubject$UnboundedReplayState.replayObserverFromIndex(SourceFile:515)
    at rx.subjects.ReplaySubject$UnboundedReplayState.replayObserver(SourceFile:503)
    at rx.subjects.ReplaySubject.caughtUp(SourceFile:423)
    at rx.subjects.ReplaySubject.onNext(SourceFile:369)
    at rx.Observable$30.onNext(SourceFile:7681)
    at rx.observers.SafeSubscriber.onNext(SourceFile:130)
    at rx.internal.operators.OperatorDoOnEach$1.onNext(SourceFile:84)
    at rx.internal.operators.OperatorMerge$MergeSubscriber.emitScalar(SourceFile:364)
    at rx.internal.operators.OperatorMerge$MergeSubscriber.tryEmit(SourceFile:326)
    at rx.internal.operators.OperatorMerge$InnerSubscriber.onNext(SourceFile:802)
    at rx.internal.operators.OperatorMerge$MergeSubscriber.emitScalar(SourceFile:364)
    at rx.internal.operators.OperatorMerge$MergeSubscriber.tryEmit(SourceFile:326)
    at rx.internal.operators.OperatorMerge$InnerSubscriber.onNext(SourceFile:802)
    at rx.subjects.SubjectSubscriptionManager$SubjectObserver.onNext(SourceFile:224)
    at rx.subjects.AsyncSubject.onCompleted(SourceFile:101)
    at rx.util.async.Async$1$1.call(SourceFile:533)
    at rx.internal.schedulers.ScheduledAction.run(SourceFile:55)
    ... 4 more
Caused by: java.lang.RuntimeException: java.lang.NoSuchFieldException: counter
    at java.util.concurrent.atomic.AtomicLongFieldUpdater$CASUpdater.<init>(AtomicLongFieldUpdater.java:269)
    at java.util.concurrent.atomic.AtomicLongFieldUpdater.newUpdater(AtomicLongFieldUpdater.java:50)
    at rx.internal.operators.OperatorZip$Zip.<clinit>(SourceFile:187)
    ... 40 more
Caused by: java.lang.NoSuchFieldException: counter
    at java.lang.Class.getDeclaredField(Class.java:929)
    at java.util.concurrent.atomic.AtomicLongFieldUpdater$CASUpdater.<init>(AtomicLongFieldUpdater.java:251)
    ... 42 more