864339 - Crash on 'New Private Tab' @ org.mozilla.gecko.mozglue.NativeZip.createInputStream(NativeZip.java:71)

Status: RESOLVED FIXED

(Firefox for Android Graveyard :: General, defect)

Description

[Aaron Train [:aaronmt]]

E/GeckoAppShell( 9698): >>> REPORTING UNCAUGHT EXCEPTION FROM THREAD 512 ("GeckoBackgroundThread")
E/GeckoAppShell( 9698): java.lang.IllegalArgumentException: Got compression 8, but expected 0 (STORE)!
E/GeckoAppShell( 9698): 	at org.mozilla.gecko.mozglue.NativeZip.createInputStream(NativeZip.java:71)
E/GeckoAppShell( 9698): 	at org.mozilla.gecko.mozglue.NativeZip._getInputStream(Native Method)
E/GeckoAppShell( 9698): 	at org.mozilla.gecko.mozglue.NativeZip.getInputStream(NativeZip.java:61)
E/GeckoAppShell( 9698): 	at org.mozilla.gecko.util.GeckoJarReader.getStream(GeckoJarReader.java:114)
E/GeckoAppShell( 9698): 	at org.mozilla.gecko.util.GeckoJarReader.getBitmapDrawable(GeckoJarReader.java:44)
E/GeckoAppShell( 9698): 	at org.mozilla.gecko.util.GeckoJarReader.getBitmap(GeckoJarReader.java:31)
E/GeckoAppShell( 9698): 	at org.mozilla.gecko.Favicons$LoadFaviconTask.downloadFavicon(Favicons.java:271)
E/GeckoAppShell( 9698): 	at org.mozilla.gecko.Favicons$LoadFaviconTask.doInBackground(Favicons.java:364)
E/GeckoAppShell( 9698): 	at org.mozilla.gecko.Favicons$LoadFaviconTask.doInBackground(Favicons.java:230)
E/GeckoAppShell( 9698): 	at org.mozilla.gecko.util.UiAsyncTask$BackgroundTaskRunnable.run(UiAsyncTask.java:48)
E/GeckoAppShell( 9698): 	at android.os.Handler.handleCallback(Handler.java:725)
E/GeckoAppShell( 9698): 	at android.os.Handler.dispatchMessage(Handler.java:92)
E/GeckoAppShell( 9698): 	at android.os.Looper.loop(Looper.java:137)
E/GeckoAppShell( 9698): 	at org.mozilla.gecko.util.GeckoBackgroundThread.run(GeckoBackgroundThread.java:32)

STR: Open a 'New Private Tab'

bp-9c146cdd-fcac-4ec7-8b64-a31422130422

--
Nightly (04/22)
LG Nexus 4 (Android 4.2.2)

Comment 1

[Aaron Train [:aaronmt]]

Regression from bug 849589?

Comment 2

[Scoobidiver (away)]

More reports at: https://crash-stats.mozilla.com/report/list?signature=java.lang.IllegalArgumentException%3A+Got+compression+8%2C+but+expected+0+%28STORE%29!+at+org.mozilla.gecko.mozglue.NativeZip.createInputStream%28NativeZip.java%29

Comment 3

[Mike Hommey [:glandium]]

What landed for bug 849589 was not a fix, but sort of diagnostics, so this is actually the same bug, but with a different message, hopefully more helpful.

Comment 4

[Mike Hommey [:glandium]]

So, it appears it is not exactly the same thing ; that is, that bug 849589 did add a crashing case that didn't exist before.

Comment 5

[Aaron Train [:aaronmt]]

http://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=c5d8f52f4457&tochange=9ff5c0134bbf

Bug 849589

Comment 6

[Aaron Train [:aaronmt]]

This is blocking any testing of private browsing on Nightly for both QA and end-users.

Comment 7

[Robert Kaiser]

Assigning to Chris, as the bug 849589 patch is his.

Comment 8

[Chris Peterson [:cpeterson]]

Attachment: fix-private-tab-crash.patch

Add InflaterInputStream support back to NativeZip.createInputStream().

Comment 9

[Chris Peterson [:cpeterson]]

Attachment: part-2-remove-EmptyStackException.patch

Part 2: Replace GeckoJarReader EmptyStackException with an empty() check. Exceptions have a small performance overhead, plus we can easily avoid triggering and catching this EmptyStackException.

Comment 10

[Chris Peterson [:cpeterson]]

I was not able to reproduce this private browsing exception with my local builds, even though I could with Nightly builds. But I *was* able to reproduce it locally after I switched my local builds from using the "unofficial" (Fennec) branding to the "nightly" branding resources!

So there seems to be some jar difference between mobile/android/branding/nightly/content/favicon32.png and mobile/android/branding/unofficial/content/favicon32.png.

Some logcat:

  downloadFavicon: jar:jar:file:///data/app/org.mozilla.fennec_cpeterson-2.apk!/omni.ja!/chrome/chrome/content/branding/favicon32.png
  getStream: filename=omni.ja
  getStream: filename=chrome/chrome/content/branding/favicon32.png

which triggers the "Unexpected compression: 8" exception from NativeZip.createInputStream().

Comment 11

[Mike Hommey [:glandium]]

(In reply to Chris Peterson (:cpeterson) from comment #10)
> I was not able to reproduce this private browsing exception with my local
> builds, even though I could with Nightly builds. But I *was* able to
> reproduce it locally after I switched my local builds from using the
> "unofficial" (Fennec) branding to the "nightly" branding resources!
> 
> So there seems to be some jar difference between
> mobile/android/branding/nightly/content/favicon32.png and
> mobile/android/branding/unofficial/content/favicon32.png.
> 
> Some logcat:
> 
>   downloadFavicon:
> jar:jar:file:///data/app/org.mozilla.fennec_cpeterson-2.apk!/omni.ja!/chrome/
> chrome/content/branding/favicon32.png
>   getStream: filename=omni.ja
>   getStream: filename=chrome/chrome/content/branding/favicon32.png
> 
> which triggers the "Unexpected compression: 8" exception from
> NativeZip.createInputStream().

Ah, that makes sense. Reading the file would get an input stream, wouldn't it? That is supposed to be supported. What is not is using the (deflated) input stream again as a zip.

Comment 12

[Chris Peterson [:cpeterson]]

Patch 1 of 2:
https://hg.mozilla.org/integration/mozilla-inbound/rev/d7a0a407d1ce

Comment 13

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/d7a0a407d1ce

Comment 14

[Wesley Johnston (:wesj)]

Comment on attachment 741086 [details] [diff] [review]
part-2-remove-EmptyStackException.patch

Review of attachment 741086 [details] [diff] [review]:
-----------------------------------------------------------------

I don't really like using a try-catch to wallpaper this.

Comment 15

[Wesley Johnston (:wesj)]

Comment on attachment 741086 [details] [diff] [review]
part-2-remove-EmptyStackException.patch

Review of attachment 741086 [details] [diff] [review]:
-----------------------------------------------------------------

Oh wait. This isn't what I thought it was. Apologies!

Comment 16

[Chris Peterson [:cpeterson]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/30bc2a74313c

Comment 17

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/30bc2a74313c