GZip crash (found in Crashlytics)

[KennyGoers]

I'm starting to see more instances of the following crash:

Caused by java.lang.NullPointerException
at okio.GzipSource.updateCrc(GzipSource.java:189)
at okio.GzipSource.read(GzipSource.java:82)
at okio.RealBufferedSource$1.read(RealBufferedSource.java:430)
at java.io.InputStreamReader.read(InputStreamReader.java:231)
at java.io.BufferedReader.fillBuf(BufferedReader.java:145)
at java.io.BufferedReader.readLine(BufferedReader.java:397)

Seeking pointers as to why this might be happening to start with.

Seeing only in Android 4.4.2

Thanks,

[swankjesse]

All device manufacturers?

[KennyGoers]

Not all I think, but list includes Quanta, Acer, ZTE, alps and PCD and still only 4.4.2

[donatienthorez]

I also encounter the same bug on Android 4.4.2 (96%) and also 4.4.1 (4%).

Here are the list of devices manufacturers I get :
Acer,QMobile, InFocus, Panasonic,HTC, iBallMobile, TECNO, Micromax, Vodafone, bq, Gigabyte, Sony

[visiolink]

Also seeing the same on Android 4.4 using okhttp 3.9.0 with stacktrace:

Caused by: java.lang.NullPointerException:
at okio.GzipSource.updateCrc (GzipSource.java:189)
at okio.GzipSource.read (GzipSource.java:82)
at okio.RealBufferedSource.request (RealBufferedSource.java:67)
at okio.RealBufferedSource.rangeEquals (RealBufferedSource.java:408)
at okio.RealBufferedSource.rangeEquals (RealBufferedSource.java:392)
at okhttp3.internal.Util.bomAwareCharset (Util.java:449)

[swankjesse]

I wonder if it's caused by running out of memory in native code.

[visiolink]

I guess the root cause of this is the same as square/okhttp#3641

So probably this can also be worked around with:
android.enableD8=false

[JakeWharton]

s/worked around/temporarily avoided/

…

On Mon, Dec 18, 2017 at 7:50 AM visiolink-android-dev < ***@***.***> wrote: I guess the root cause of this is the same as square/okhttp#3641 <square/okhttp#3641> So probably this can also be worked around with: android.enableD8=false — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#323 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAEEEVvyk7uVo1Wm_hbfiPNH99SRndn9ks5tBl-MgaJpZM4QQlf9> .

[visiolink]

Confirmed to be temporarily avoided by disabling D8.

[swankjesse]

If anyone has a fix that prevents the problem, I’m receptive. Otherwise I think this is a problem in the platform.

[madsager]

Update from the D8 side on this one. Turns out the VM on these devices had yet another bug. We have landed a workaround for the VM bug in D8 which will be part of Android Studio 3.1 stable (and will make its way into android studio 3.2 over the next couple of weeks as well).

D8-side bug: https://issuetracker.google.com/75408029