636602 - Android crash reporter crashes

Status: RESOLVED FIXED

(Toolkit :: Crash Reporting, defect)

Description

[Josh Matthews [:jdm]]

I've seen Fennec crash and bring up the reporter app, but then the reporter app appears to crash.  From logcat:

I/GeckoCrashReport(12465): sendReport: /data/data/org.mozilla.fennec/mozilla/Crash Reports/pending/0510b74f-54cc-2b9f-51ffd321-55589558.dmp
I/GeckoCrashReport(12465): server url: https://crash-reports.mozilla.com/submit
W/dalvikvm(12465): threadid=17: thread exiting with uncaught exception (group=0x4001dc20)
E/AndroidRuntime(12465): Uncaught handler: thread Thread-9 exiting due to uncaught exception
E/AndroidRuntime(12465): java.lang.NoSuchFieldError: android.os.Build.CPU_ABI2
E/AndroidRuntime(12465): 	at org.mozilla.fennec.CrashReporter.sendReport(CrashReporter.java:258)
E/AndroidRuntime(12465): 	at org.mozilla.fennec.CrashReporter$1.run(CrashReporter.java:134)
E/AndroidRuntime(12465): 	at java.lang.Thread.run(Thread.java:1096)

Comment 1

[Josh Matthews [:jdm]]

I think at minimum investigation of this should block, because I'm unsure if any of the crashes I'm trying to send are making it to the crash-stats servers.

Comment 2

[Doug Turner (:dougt)]

we need to catch these things.

http://mxr.mozilla.org/mozilla-central/source/embedding/android/CrashReporter.java.in#259

http://hg.mozilla.org/mozilla-central/rev/3369bc46ff9c

I can steal...

Comment 3

[Doug Turner (:dougt)]

Attachment: patch v.1

not sure I really like this, but try try try.

I was thinking about doing something in the catch, but figured if the field isn't there in the report, we know an exception occurred.

Comment 4

[Brad Lassey [:blassey] (use needinfo?)]

Comment on attachment 515832 [details] [diff] [review]
patch v.1

doug, can you just put a try catch in sendPart, rather than.... this?

Comment 5

[Doug Turner (:dougt)]

Comment on attachment 515832 [details] [diff] [review]
patch v.1

brad and I spoke.  we can be a little tighter with the try{}'s

Comment 6

[Brad Lassey [:blassey] (use needinfo?)]

Attachment: patch

Comment 7

[Doug Turner (:dougt)]

Comment on attachment 515978 [details] [diff] [review]
patch



>       sendPart(os, boundary, "Android_Model", Build.MODEL);
>       sendPart(os, boundary, "Android_Board", Build.BOARD);
>       sendPart(os, boundary, "Android_Brand", Build.BRAND);
>-      sendPart(os, boundary, "Android_CPU_ABI", Build.CPU_ABI); 
>-      sendPart(os, boundary, "Android_CPU_ABI2", Build.CPU_ABI2);
>       sendPart(os, boundary, "Android_Device", Build.DEVICE);
>       sendPart(os, boundary, "Android_Display", Build.DISPLAY);
>       sendPart(os, boundary, "Android_Fingerprint", Build.FINGERPRINT);

So we require at least version 4.

>-      if (Build.VERSION.SDK_INT >= 8)
>-        sendPart(os, boundary, "Android_Hardware", Build.HARDWARE);
>+      sendPart(os, boundary, "Android_CPU_ABI", Build.CPU_ABI); 
>+      if (Build.VERSION.SDK_INT >= 8) {
>+        try {
>+          sendPart(os, boundary, "Android_CPU_ABI2", Build.CPU_ABI2);
>+          sendPart(os, boundary, "Android_Hardware", Build.HARDWARE);
>+        } catch (Exception ex) {
>+          Log.e("GeckoCrashReporter", "Exception while sending SDK version 8 keys", ex);
>+        }

Why do we need this try?  Lets remove it, and r+

Comment 8

[Brad Lassey [:blassey] (use needinfo?)]

(In reply to comment #7)
> 
> So we require at least version 4.
and that's enforced by our manifest https://mxr.mozilla.org/mozilla-central/source/embedding/android/AndroidManifest.xml.in#9

Comment 9

[Brad Lassey [:blassey] (use needinfo?)]

pushed http://hg.mozilla.org/mozilla-central/rev/efbce8140bb4