IssueTracker

This bug is critical! Please fix it ASAP!

Broke for me too

This should be set to High Priority. As other comments allude, this is critical and is causing our app (and I am sure others) to crash and potential workarounds are pretty ugly.

It's a silly JNI mistake, please fix it before 2.3 goes alive on devices

Everyone, tell everyone you know to star this issue!  This is a crucial, crucial bug to fix.

Please fix asap!

36 people starred this issue and may be notified of changes.  How do I star it?! I agree, this is a pretty big issue, although if you target for 2.2 or below you should be ok.

Hey yakkyjunk, targeting 2.2 does not resolve the issue.  Anything that uses that functionality and is running on 2.3 will encounter this and crash.

(Click on the star next to the word 'Issue' in the blue title bar)

This is a critical issue for us as well.

I can verify the crash using these other sample projects:

https://github.com/commonsguy/cw-advandroid/tree/master/WebView/GeoWeb1
https://github.com/commonsguy/cw-advandroid/tree/master/WebView/GeoWeb2
https://github.com/commonsguy/cw-abj/tree/master/PhoneGap/Checklist

Note that the latter is a PhoneGap project, suggesting that all PhoneGap-powered apps will fail on Android 2.3 as well.

All of these projects work fine on Android 2.2.

--
Mark Murphy (a Commons Guy)
mmurphy@commonsware.com

Actually, those samples I just posted are not good ones, as there may be a separate SIGSEGV related to mock GPS data sent via DDMS. Here is a really tiny sample project that does demonstrate the addJavascriptInterface() problem:

http://commonsware.com/misc/SIGSEGVOne.zip

--
Mark Murphy (a Commons Guy)
mmurphy@commonsware.com

I was just about to post very similar to Mark's example. My results are exactly the same - crash 100% of the time.

A workaround posted here:
http://quitenoteworthy.blogspot.com/2010/12/handling-android-23-webviews-broken.html

The above link is only a partial workaround.

Major issues:

This does not allow for a synchronous response from java. Any javascript code can be rewritten with return callbacks from java in a continuation passing style, but this generally requires a complete restructuring of the affected code base.

The javascript to java interface presented here does not persist across page loads. The original (broken) mechanism is available to any javascript executing within the webview, whether in the parent document or in any nested iframe, and is unaffected by content reloading.

Agreed on both points. That's why this is a workaround, and only for simple cases; the real solution is for the Android team to actually fix the root cause.

If you're really looking for a workaround, I think it'd be easier to override the WebChromeClient's onJSAlert method and write a simple parser.  You still can't return values, but you don't have to reload the page.

That said, this does not work for me at all.  I'd have to rewrite almost all of my code to get this working, which is pretty terrible.

A good news for you guys, the crash doesn't happen on real Nexus S device, so basically this is another bug on emulator as NIO on 2.2

Repro'd on the 2.3 emulator; verified that this is NOT reproducible on a Nexus S.

Great, that's good news.  Any idea when we might get a new emulator build?  Also, is there any way to get early device builds for the Nexus One?  Alternatively, is there a program we can apply to for early access in the future?

Please solve his.

I encountered this issue too.  I upgraded the SDK to include version 2.3 of the emulator and our app crashed, so I decided to check the WebViewDemo app to make sure it wasn't a problem with our app.  The demo app crashed right away, and that led me to this page.  Hopefully they can fix it soon, because this JavaScript functionality is a major part of our app.

learn to qa

please solve as soon as, my app depend on the javascript.

[Comment deleted]

Grrr, client project testing is not possible on 2.3 emulator because of this issue and 2.2 has issue with camera. Disappointed with emulator.

This is a critical issue for us as well, please fix this ASAP.

Thanks.

This is a big issue, please fix this quickly.

Thanks

I'm also experiencing this issue in the 2.3 emulator... when including js in the webview demo... the

Just want to confirm that are you guys getting the following like I do?

12-20 19:29:34.582: WARN/dalvikvm(725): JNI WARNING: jarray 0x4074f6e0 points to non-array object (Ljava/lang/String;)
12-20 19:29:34.622: INFO/dalvikvm(725): "WebViewCoreThread" prio=5 tid=9 NATIVE
12-20 19:29:34.622: INFO/dalvikvm(725):   | group="main" sCount=0 dsCount=0 obj=0x40521328 self=0x26bea8
12-20 19:29:34.622: INFO/dalvikvm(725):   | sysTid=734 nice=0 sched=0/0 cgrp=default handle=2539488
12-20 19:29:34.632: INFO/dalvikvm(725):   | schedstat=( 5341953614 5808331660 581 )
12-20 19:29:34.632: INFO/dalvikvm(725):   at android.webkit.WebViewCore.nativeTouchUp(Native Method)
12-20 19:29:34.632: INFO/dalvikvm(725):   at android.webkit.WebViewCore.nativeTouchUp(Native Method)
12-20 19:29:34.642: INFO/dalvikvm(725):   at android.webkit.WebViewCore.access$3300(WebViewCore.java:53)
12-20 19:29:34.652: INFO/dalvikvm(725):   at android.webkit.WebViewCore$EventHub$1.handleMessage(WebViewCore.java:1158)
12-20 19:29:34.652: INFO/dalvikvm(725):   at android.os.Handler.dispatchMessage(Handler.java:99)
12-20 19:29:34.652: INFO/dalvikvm(725):   at android.os.Looper.loop(Looper.java:123)
12-20 19:29:34.652: INFO/dalvikvm(725):   at android.webkit.WebViewCore$WebCoreThread.run(WebViewCore.java:629)
12-20 19:29:34.662: INFO/dalvikvm(725):   at java.lang.Thread.run(Thread.java:1019)
12-20 19:29:34.672: ERROR/dalvikvm(725): VM aborting
12-20 19:29:34.852: INFO/DEBUG(30): *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***
12-20 19:29:34.852: INFO/DEBUG(30): Build fingerprint: 'generic/sdk/generic:2.3/GRH55/79397:eng/test-keys'
12-20 19:29:34.862: INFO/DEBUG(30): pid: 725, tid: 734  >>> XXXXXXXX <<<
12-20 19:29:34.862: INFO/DEBUG(30): signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr deadd00d
12-20 19:29:34.872: INFO/DEBUG(30):  r0 fffffebc  r1 deadd00d  r2 00000026  r3 00000000
12-20 19:29:34.872: INFO/DEBUG(30):  r4 81da45c8  r5 4074f6e0  r6 81d8592c  r7 0026d2b8
12-20 19:29:34.872: INFO/DEBUG(30):  r8 44174b54  r9 43d72e4c  10 43d72e34  fp 41edd82c
12-20 19:29:34.891: INFO/DEBUG(30):  ip 81da4728  sp 44174348  lr afd19375  pc 81d45a02  cpsr 20000030
12-20 19:29:35.912: INFO/DEBUG(30):          #00  pc 00045a02  /system/lib/libdvm.so
12-20 19:29:35.912: INFO/DEBUG(30):          #01  pc 000376fc  /system/lib/libdvm.so
12-20 19:29:35.912: INFO/DEBUG(30):          #02  pc 000399c4  /system/lib/libdvm.so
12-20 19:29:35.922: INFO/DEBUG(30):          #03  pc 0003a4a0  /system/lib/libdvm.so
12-20 19:29:35.932: INFO/DEBUG(30):          #04  pc 0032b6d6  /system/lib/libwebcore.so
12-20 19:29:35.952: INFO/DEBUG(30):          #05  pc 002a4da4  /system/lib/libwebcore.so
12-20 19:29:35.952: INFO/DEBUG(30):          #06  pc 001a6136  /system/lib/libwebcore.so
12-20 19:29:35.952: INFO/DEBUG(30):          #07  pc 002a5870  /system/lib/libwebcore.so
12-20 19:29:35.962: INFO/DEBUG(30):          #08  pc 00359e36  /system/lib/libwebcore.so
12-20 19:29:35.972: INFO/DEBUG(30):          #09  pc 0035d18e  /system/lib/libwebcore.so
12-20 19:29:35.972: INFO/DEBUG(30):          #10  pc 0036b078  /system/lib/libwebcore.so
12-20 19:29:35.982: INFO/DEBUG(30):          #11  pc 003627ac  /system/lib/libwebcore.so
12-20 19:29:35.982: INFO/DEBUG(30):          #12  pc 0029fb24  /system/lib/libwebcore.so
12-20 19:29:36.002: INFO/DEBUG(30):          #13  pc 0023367e  /system/lib/libwebcore.so
12-20 19:29:36.002: INFO/DEBUG(30):          #14  pc 00239c0c  /system/lib/libwebcore.so
12-20 19:29:36.011: INFO/DEBUG(30):          #15  pc 00237a28  /system/lib/libwebcore.so
12-20 19:29:36.011: INFO/DEBUG(30):          #16  pc 00237b32  /system/lib/libwebcore.so
12-20 19:29:36.022: INFO/DEBUG(30):          #17  pc 0023adba  /system/lib/libwebcore.so
12-20 19:29:36.022: INFO/DEBUG(30):          #18  pc 0023af4e  /system/lib/libwebcore.so
12-20 19:29:36.042: INFO/DEBUG(30):          #19  pc 000f80c0  /system/lib/libwebcore.so
12-20 19:29:36.042: INFO/DEBUG(30):          #20  pc 000f8c66  /system/lib/libwebcore.so
12-20 19:29:36.052: INFO/DEBUG(30):          #21  pc 001b81fa  /system/lib/libwebcore.so
12-20 19:29:36.062: INFO/DEBUG(30):          #22  pc 001b82f2  /system/lib/libwebcore.so
12-20 19:29:36.062: INFO/DEBUG(30):          #23  pc 001b8318  /system/lib/libwebcore.so
12-20 19:29:36.072: INFO/DEBUG(30):          #24  pc 00017d34  /system/lib/libdvm.so
12-20 19:29:36.082: INFO/DEBUG(30): code around pc:
12-20 19:29:36.082: INFO/DEBUG(30): 81d459e0 447a4479 ed0cf7d1 20004c09 ee34f7d1
12-20 19:29:36.092: INFO/DEBUG(30): 81d459f0 447c4808 6bdb5823 d0002b00 49064798
12-20 19:29:36.092: INFO/DEBUG(30): 81d45a00 700a2226 eea0f7d1 0004355f 0004511d
12-20 19:29:36.102: INFO/DEBUG(30): 81d45a10 0005ebd2 fffffebc deadd00d b510b40e
12-20 19:29:36.102: INFO/DEBUG(30): 81d45a20 4c0a4b09 447bb083 aa05591b 6b5bca02
12-20 19:29:36.112: INFO/DEBUG(30): code around lr:
12-20 19:29:36.112: INFO/DEBUG(30): afd19354 b0834a0d 589c447b 26009001 686768a5
12-20 19:29:36.112: INFO/DEBUG(30): afd19364 220ce008 2b005eab 1c28d003 47889901
12-20 19:29:36.122: INFO/DEBUG(30): afd19374 35544306 d5f43f01 2c006824 b003d1ee
12-20 19:29:36.122: INFO/DEBUG(30): afd19384 bdf01c30 000281a8 ffffff88 1c0fb5f0
12-20 19:29:36.132: INFO/DEBUG(30): afd19394 43551c3d a904b087 1c16ac01 604d9004
12-20 19:29:36.132: INFO/DEBUG(30): stack:
12-20 19:29:36.132: INFO/DEBUG(30):     44174308  00000015  
12-20 19:29:36.132: INFO/DEBUG(30):     4417430c  afd18407  /system/lib/libc.so
12-20 19:29:36.142: INFO/DEBUG(30):     44174310  afd4270c  /system/lib/libc.so
12-20 19:29:36.142: INFO/DEBUG(30):     44174314  afd426b8  /system/lib/libc.so
12-20 19:29:36.152: INFO/DEBUG(30):     44174318  00000000  
12-20 19:29:36.152: INFO/DEBUG(30):     4417431c  afd19375  /system/lib/libc.so
12-20 19:29:36.162: INFO/DEBUG(30):     44174320  0026bea8  [heap]
12-20 19:29:36.162: INFO/DEBUG(30):     44174324  afd183d9  /system/lib/libc.so
12-20 19:29:36.162: INFO/DEBUG(30):     44174328  0026d2b8  [heap]
12-20 19:29:36.172: INFO/DEBUG(30):     4417432c  0005ebd2  [heap]
12-20 19:29:36.172: INFO/DEBUG(30):     44174330  4074f6e0  /dev/ashmem/dalvik-heap (deleted)
12-20 19:29:36.172: INFO/DEBUG(30):     44174334  81d8592c  /system/lib/libdvm.so
12-20 19:29:36.182: INFO/DEBUG(30):     44174338  0026d2b8  [heap]
12-20 19:29:36.182: INFO/DEBUG(30):     4417433c  afd18437  /system/lib/libc.so
12-20 19:29:36.191: INFO/DEBUG(30):     44174340  df002777  
12-20 19:29:36.191: INFO/DEBUG(30):     44174344  e3a070ad  
12-20 19:29:36.191: INFO/DEBUG(30): #00 44174348  00000001  
12-20 19:29:36.191: INFO/DEBUG(30):     4417434c  81d37701  /system/lib/libdvm.so
12-20 19:29:36.191: INFO/DEBUG(30): #01 44174350  00000001  
12-20 19:29:36.191: INFO/DEBUG(30):     44174354  81d399c9  /system/lib/libdvm.so

This is a very bad! Please fix it ASAP

I also have problems with this. (I think at least)

My app crashes when run on the 2.3 emulator with the log printed above.
I don't use any webview myself in the app startup, but I guess some of the ad network SDKs might do, thus causing the crash (?).
If so, this will possibly break most apps with ads in them...

Why is this bug still rated only priority medium? To me it is a blocker.

Maybe it's only in the emulator? If so, could we please get an official confirmation on that? But even so, I won't be able to test the app on the 2.3 platform before publishing.

Agree, this is a very serious bug. Please raise the priority.

I get this error on my builds custom builds for my msm7227 cell phone device when running gmail.

Very serious bug, fix please! Raise the priority.

Please fix this ASAP. Thank you.

Experiencing the same problem here.  Please fix.

I have the same issue when interfacing on 2.3. please someone from google give some feedback.

Whole development is stuck!
Please fix asap

It shows the status of this has been reviewed, what was the outcome of that review??

I think this is being treated as quite a low priority since it only appears to be a bug in the emulator and not real devices. Of course a new release of the emulator (e.g. 2.3.0.1) wouldn't go amiss in fixing this.

I think it's high priority even it appears on emulator currently. Please note that most customized OS are built from the public source code of Android which is mostly the same as emulator. I've already seen the bug is transferred to devices with customized 2.3 rom.

Same issue. My app depends on Javascript. Please fix.

Pleas fix bug as soon as possible!My app can't run on 2.3

I have the similar issue and who can tell me the method of analyzing the LogCat log like this issue.

Please fix this quickly.

Please make this issue as critical and resolve it ASAP.

Same Issue , same Log, even with Phonegap Example Apps. Please fix this.

Same problem. This have to be fixed soon.. PLEASE...

Voted to be fixed as well. I think they'll do it next release but not until then.

Please fix, I use webview/JS alot....Ouch.

the new released 2.3.1 doesn't fix it...

I was so hopeful when I saw that there was a new release 2.3.1.  I downloaded it and the same problem is still there.

I have the same problem both in the 2.3 emulator and on ZTE Blade phone using 2.2. It works perfect on for example HTC Desire and some other Android phones we have tested. So it is actually relating both to real devices and to other released than 2.3. We can't make an official release of our application until this is corrected.

Amazon Marketplace says my Phonegap app can not run in their tests due to this error. They also say it fails on the Droid Incredible running Android OS 2.2

This is a critical bug and should be fixed innan out-of-band update, and nothing less.

Please fix this bug, this is critical, many apps based on webview are not working on Android 2.3.1

this is ctritical!

Fix please!

is there a workaround for this?

The only workaround that I know of is to test on the 2.2 or on the device. In my app, there is no 2.3 dependent functionality so it works fine, but I am trying to get my hands on a real device anyway because I understand that the emulator is quite slow.

it's critical...

just tried the brand-spanking-new 2.3.3 emulator image.  Same problem :-(

This is critical for me too. Testing all platforms for android apps is critical. Even if it only happnes on simulator, which is not yet confirmed, it's critical.

I compiled gingerbread for HTC Magic and it not work.
Is this not only emulator bug?

It should be fixed asap

Okay folks, other phonegap developer are finding the same bug on real devices,, its effing real serious problem now.

We need to start tracking this by android OS version for emulator, android device OS version , webkit version, etc

If I get the dev job I interview on Monday I am writing an effing parser to fix..

The way I see it, unless it's a super-critical _security_ flaw it will not get fixed in any current version of Android. Google is currently working on Gingerbread 2.4 and Honeycomb, so all we can do is pray that they have corrected this issue in those releases.

I'm not sure what's so drastically wrong with Gingerbread 2.3, but there is still only one device out that has it officially: the Nexus S. Google hasn't updated the Nexus 1 yet for some mysterious reason, and none of the manufacturers have released any official updates to it (no surprise there though).

Just target Froyo with your PhoneGap applications for the time being, and wait for Gingerbread 2.4 which will probably be the widely used version. This reminds me of the 2 Eclairs, although I can't remember what was broken in the first one...

Guys, this might be good news..

Try switching addJavascriptInterface to after loadURL

Its not fully tested yet..need to do a better example and fully test it..but Mark Murphy's example with that change is getting no errors.. Yeah!

The logcat output when switching addJavascriptInterface to after loadURL:


02-20 21:34:52.452: DEBUG/dalvikvm(486): GC_CONCURRENT freed 103K, 69% free 319K/1024K, external 0K/0K, paused 2ms+1ms
02-20 21:34:52.462: DEBUG/jdwp(486): Got wake-up signal, bailing out of select
02-20 21:34:52.462: DEBUG/dalvikvm(486): Debugger has detached; object registry had 1 entries
02-20 21:34:56.692: INFO/ActivityManager(75): Displayed com.mobilebytes.brokenaddjavascriptinterface/.MainActivity: +4s298ms
02-20 21:35:05.141: DEBUG/dalvikvm(367): GC_EXPLICIT freed 7K, 54% free 2543K/5511K, external 716K/1038K, paused 91ms
02-20 21:35:10.281: DEBUG/dalvikvm(423): GC_EXPLICIT freed 317K, 54% free 2538K/5511K, external 716K/1038K, paused 219ms
02-20 21:35:15.223: DEBUG/dalvikvm(438): GC_EXPLICIT freed 481K, 55% free 2595K/5703K, external 716K/1038K, paused 138ms
02-20 21:35:20.251: DEBUG/dalvikvm(148): GC_EXPLICIT freed 93K, 51% free 2934K/5895K, external 2048K/2461K, paused 158ms
02-20 21:39:33.452: DEBUG/SntpClient(75): request time failed: java.net.SocketException: Address family not supported by protocol
02-20 21:44:33.461: DEBUG/SntpClient(75): request time failed: java.net.SocketException: Address family not supported by protocol
02-20 21:49:33.472: DEBUG/SntpClient(75): request time failed: java.net.SocketException: Address family not supported by protocol


As you can see no errors...

That would mean something not being loaded that addJavascriptInterface needs until loadURL is called..in 2.2, 2.3 emulators and some devices

I'm unable to make addJavascriptInterface to work, even after calling it after loadUrl().  If I call addJavascriptInterface immediately after the loadUrl(), I get the same trap as if I had put it before.  If I make the call well after loadUrl() - I do it in a consoleMessage callback - then the "object" I'm trying to install never gets installed.

Perhaps it's a bug in my code.  Anyone else have any luck?

So far I've only had a chance to try moving addJavascriptInterface() to immediately after loadUrl(), but it didn't help.

Sorry folks I still get same locat errors.

However, if I do not touch the app no user input I am not seeing WebViewCore messages such as th AddJavascritInterface succeeded so is this a WebViewCore initialization and native lib interference?

Please fix this bug, this is critical.

The Honeycomb emulator was officially released today, and I think I noticed the Gingerbread emulator getting some updates. Anyone tried this in the new SDK yet?

Works in 3.0 emulator.

For the 2.3.1 and 2.3.3 emulators - no change.


02-22 22:47:55.529: WARN/dalvikvm(404): JNI WARNING: jarray 0x4051a6e8 points to non-array object (Ljava/lang/String;)
02-22 22:47:55.538: INFO/dalvikvm(404): "WebViewCoreThread" prio=5 tid=9 NATIVE
02-22 22:47:55.538: INFO/dalvikvm(404):   | group="main" sCount=0 dsCount=0 obj=0x40520330 self=0x1f5278
02-22 22:47:55.538: INFO/dalvikvm(404):   | sysTid=412 nice=0 sched=0/0 cgrp=default handle=2054032
02-22 22:47:55.548: INFO/dalvikvm(404):   | schedstat=( 1467256456 1411922893 266 )
02-22 22:47:55.548: INFO/dalvikvm(404):   at android.webkit.LoadListener.nativeFinished(Native Method)
02-22 22:47:55.558: INFO/dalvikvm(404):   at android.webkit.LoadListener.nativeFinished(Native Method)
02-22 22:47:55.570: INFO/dalvikvm(404):   at android.webkit.LoadListener.tearDown(LoadListener.java:1200)
02-22 22:47:55.570: INFO/dalvikvm(404):   at android.webkit.LoadListener.handleEndData(LoadListener.java:721)
02-22 22:47:55.577: INFO/dalvikvm(404):   at android.webkit.LoadListener.handleMessage(LoadListener.java:219)
02-22 22:47:55.577: INFO/dalvikvm(404):   at android.os.Handler.dispatchMessage(Handler.java:99)
02-22 22:47:55.577: INFO/dalvikvm(404):   at android.os.Looper.loop(Looper.java:123)
02-22 22:47:55.577: INFO/dalvikvm(404):   at android.webkit.WebViewCore$WebCoreThread.run(WebViewCore.java:629)
02-22 22:47:55.587: INFO/dalvikvm(404):   at java.lang.Thread.run(Thread.java:1019)
02-22 22:47:55.587: ERROR/dalvikvm(404): VM aborting
02-22 22:47:55.798: INFO/DEBUG(31): *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***
02-22 22:47:55.798: INFO/DEBUG(31): Build fingerprint: 'generic/sdk/generic:2.3.1/GSI11/93351:eng/test-keys'
02-22 22:47:55.798: INFO/DEBUG(31): pid: 404, tid: 412  >>> -------------------------------------- <<<
02-22 22:47:55.808: INFO/DEBUG(31): signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr deadd00d

02-22 22:53:51.301: WARN/dalvikvm(418): JNI WARNING: jarray 0x40518c48 points to non-array object (Ljava/lang/String;)
02-22 22:53:51.310: INFO/dalvikvm(418): "WebViewCoreThread" prio=5 tid=9 NATIVE
02-22 22:53:51.310: INFO/dalvikvm(418):   | group="main" sCount=0 dsCount=0 obj=0x40520988 self=0x1f4ec8
02-22 22:53:51.310: INFO/dalvikvm(418):   | sysTid=426 nice=0 sched=0/0 cgrp=default handle=2054608
02-22 22:53:51.310: INFO/dalvikvm(418):   | schedstat=( 1769120282 2152704812 298 )
02-22 22:53:51.310: INFO/dalvikvm(418):   at android.webkit.LoadListener.nativeFinished(Native Method)
02-22 22:53:51.310: INFO/dalvikvm(418):   at android.webkit.LoadListener.nativeFinished(Native Method)
02-22 22:53:51.310: INFO/dalvikvm(418):   at android.webkit.LoadListener.tearDown(LoadListener.java:1200)
02-22 22:53:51.310: INFO/dalvikvm(418):   at android.webkit.LoadListener.handleEndData(LoadListener.java:721)
02-22 22:53:51.310: INFO/dalvikvm(418):   at android.webkit.LoadListener.handleMessage(LoadListener.java:219)
02-22 22:53:51.310: INFO/dalvikvm(418):   at android.os.Handler.dispatchMessage(Handler.java:99)
02-22 22:53:51.310: INFO/dalvikvm(418):   at android.os.Looper.loop(Looper.java:123)
02-22 22:53:51.310: INFO/dalvikvm(418):   at android.webkit.WebViewCore$WebCoreThread.run(WebViewCore.java:629)
02-22 22:53:51.310: INFO/dalvikvm(418):   at java.lang.Thread.run(Thread.java:1019)
02-22 22:53:51.310: ERROR/dalvikvm(418): VM aborting
02-22 22:53:51.322: DEBUG/dalvikvm(277): GC_EXPLICIT freed 86K, 54% free 2718K/5831K, external 716K/1038K, paused 7092ms
02-22 22:53:51.431: INFO/DEBUG(31): *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***
02-22 22:53:51.431: INFO/DEBUG(31): Build fingerprint: 'generic/sdk/generic:2.3.3/GRI34/101070:eng/test-keys'
02-22 22:53:51.431: INFO/DEBUG(31): pid: 418, tid: 426  >>> -------------------------------------- <<<
02-22 22:53:51.431: INFO/DEBUG(31): signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr deadd00d

Works for me in the 3.0 emulator as well.  It appears that the 3.0 emulator still uses JSC (try eval("\") and look at the exception).

[Comment deleted]

Frankly, the lack of response (or acknowledgment) by Google regarding this issue is a cataclysmic failure and a black eye for the Android platform. By not addressing this, Google is making it clear that they do not care about their developers nor do they care about the operating system. Pathetic. This makes Apple seem open.

This is ctritical...

Ahah yes, I'm also amazed by the lack of response by google on this one.

I'd put good money on the lack of response being that it's fixed in Honeycomb, and the fix will be back-ported into 2.4 (which is still Gingerbread), and they'll never bother to fix 2.3 because 2.3 won't actually end up on devices other than the Nexus S.

Of course, they should've come out and actually said that officially.....

I just realized that no one announced this in this thread: the issue appears to be fixed in the version of Gingerbread that got pushed to the Nexus One last week (2.3.3).

The lack of official response is still a bummer, but it looks like disaster has been averted.

The issue has never shown itself on the phone.  It has always been on the emulator.

This is problem with 2.3 emulator and Google Apis version 9

Yes its a confirmed bug in android 2.3.3 level 9..I tested it on  i get the same log mentioned above....

[Comment deleted]

The Nexus S never has suffered from this problem and does not now (2.3.3). Please try the sample apps from comment #10.

the same problem on nexus one 2.3.3!

the same problem.

Same problem Galaxy S I9000XWJV1

the same problem on emutor

Apparently, a PhoneGap developer named Bryce Curtis has addressed this issue as it affects PhoneGap by implementing a creative workaround that eliminates PhoneGap's dependency upon addJavascriptInterface(). From what I can tell, his fix is likely to go into the next (0.9.5) release of PhoneGap.

In particular, this fix would seem to restore PhoneGap's plugin functionality. It may fix some or all of the PhoneGap problems reported above that were due to its dependency upon addJavascriptInterface().

Since PhoneGap plugins provide an alternative (for PhoneGap apps) to using addJavascriptInterface() for crossing the Javascript to Java boundary, once Bryce's fix eliminating PhoneGap's own dependency upon addJavascriptInterface goes in, it seems possible that applications using PhoneGap that are presently dependent upon addJavascriptInterface() might be able to resolve the problems caused by the addJavascriptInterface() anomalies reported here by recoding their Javascript-to-Java bridge code to use PhoneGap plugins instead of using addJavascriptInterface(). And it seems possible that apps already using the plugin facility will start to work better in Android 2.3 once they update to a PhoneGap version that includes this fix.

It also seems possible that non-PhoneGap Android apps that need to communicate between a WebView and Java might be able to adapt Bryce's approach to their own needs, although since it was designed primarily as a workaround to restore PhoneGap's own functionality it will not necessarily be as pretty an approach as using the plugin facility itself.

Note that I'm reporting this second hand, as something I came across while searching for an alternative to addJavascriptInterface() for my own development. I'm relatively new to PhoneGap, and can't claim to fully understand Bryce's code myself. However, the code itself can be found here:

https://github.com/phonegap/phonegap-android/commit/2bc7bd6768be3d82eab2e1b4f295abebffa79359

and a discussion of the problem that includes Bryce can be found here:

http://phonegap.lighthouseapp.com/projects/20118/tickets/99-phonegap-apps-rejected-by-amazon-marketplace#ticket-99-12

Once again, I can't be sure that I have accurately reported the above, so if I've misread this discussion and development, somebody please post a correction. But since no one else had mentioned this so far on this issue list, and since it appears to me, given what I do understand of it, to be highly pertinent to this issue, I wanted to make it known here. When in doubt, please refer to the two links above and to those actually involved in addressing this very significant issue.

Also, just found the following thread with a recent note from Bryce Curtis at the end regarding his fix:

http://groups.google.com/group/phonegap/browse_thread/thread/a2b6ac44ea798ce4

the same problem.

this seems to be a bug in the default JS engine(JSC) on emulator
in mydroid emulator build
if you enable v8 as default JS engine
this crashes is gone

this is also the reason why this crash will NOT happen on the device(nexus S)

Its been a while, it should have been fixed by now. Serious issue for developers.

Has someone found a workaround, for developers not using PhoneGap.
Also, enabling v8 engine is a solution that cannot be used for devices. Because it is required to rebuild image. I am beginner, dont mind if this question is awfully noob. Thanks.

the issue exists now.

As usual, this screws small developers who can't afford to buy every phone on the market to test. Can I test my app on 2.3 now? No, because the emulator is buggy. Note this isn't the first issue - the LVL tools also don't work on 2.3.

Agreed! And even for smaller developers like me who cant even figure out how the fix works for webview only. This is a pain in the ass.

I'm work on the framework,I've resolved it on our tablet and written it here.
my English is poor, so I hope you can understand.
http://blog.csdn.net/a345017062/archive/2011/05/04/6394864.aspx

2.2 seems destined to become Google's XP, while 2.3 is Vista. Or maybe 2.3 is Windows Me and 3.0 is Vista. Anyway my point is please fix this.

is it fixed in 2.3.4?

[Comment deleted]

This problem makes me a big trouble .Has Anyone known how to fix it yet ?

I am not sure if this bug is real on devices or just in the emulator as I hear some rumors, but it still is crashing my code on the 2.3.3 emulator which I thought had received a fix.

Any updates on this issue....any workaround solution which is completely feasible

If we make a debug build for libwebcore and then use addJavaScriptInterface api it do not crash, strange, hard to make out what is the issue

The issue is when the API returns.  I've been banging my head on this issue for months, but if you look at Comment 101, you can see that the issue is related to issues with the JNI bridge.  However, I don't see any fix to this.  After looking at the CTS, it SHOULD (and probably has) cause the CTS to fail, which may explain why there are very few low-end phones that run Android 2.3, since they would never pass that test.

We have a similar feature which needs to be supported on 2.3.3, but it seems to be failing. When will this be fixed.

 This issue seems clearly to be a crucial one and there was no change since...December 2010?
Are you kidding me?

Wait, JS to Java breaks on Gingerbread?? Not having Gingerbread, this is the first I've heard of this, and the app I am developing fundamentally relies on this!

Are people only seeing this on devices?  There is some thrash in the comments and it is hard to tell if it is only the emulator or the phones as well.

If you are seeing it on a real phone, can you send us which phone model?   We are trying to reproduce?

Also, how are people handling this crash bug in their apps?  Targeting around it?

Thx everyone for the help!

This is terrible. I am trying to be a responsible developer by testing my code on Gingerbread (I only got a 2.2 device on hand), yet I am punished and wasted my whole morning to trace my code to discover this bug. Even if the bug just occurred on the emulator, it should be fixed asap!

I am learning [Hello Android] with simulator of version 2.3.3 and get crashed.
I changed the version of simulator to 2.2 after reading comments above and passed the test.
However I want the question be solved ASAP, just like everyone wishes.

May i know when will this be fixed? in 2.3.5 ?
Do google provide any emergency workaround ??

I have done some investigation into this bug, and I believe it is a problem introduced in the JSC javascript engine in Gingerbread.

First, some background:

Versions of Android 2.2 and greater include the v8 javascript engine in addition to the original WebKit Javascript engine (JavaScriptCore, or JSC).  The vast majority of devices use v8, but not necessarily all - some low-memory devices may apparently use JSC.  From the WebKit makefile:

# The default / alternative engine depends on the device class.
# On devices with a lot of memory (e.g. Passion/Sholes), the
# default is V8. On everything else, the only choice is JSC.
# TODO: use ARCH_ARM_HAVE_ARMV7 once that variable is added to
# the build system.

However, emulator images always use JSC.  You can determine which JS engine is in use at runtime through the following javascript snippet:


if (window.devicePixelRatio) { //If WebKit browser
   var st = escape(navigator.javaEnabled.toString());
   if (st === 'function%20javaEnabled%28%29%20%7B%20%5Bnative%20code%5D%20%7D') {
      document.write('V8 detected');
   } else {
      document.write('JSC detected');
   }
} else {
  document.write("Not a WebKit browser");
}


My Nexus one has 2.3.4 on it, and when I ran this piece of javascript it reported 'v8', and I wasn't able to repro this bug.  

I then downloaded and build the android source with the JS_ENGINE environment variable set to 'jsc', then installed the custom image onto my Nexus One.  Now my phone reports having 2.3.4 on it, and the JS snippet above reports 'jsc', and I can consistently repro this JNI warning.  So this bug _can_ happen on devices, and likely is happening in the wild on certain low-memory devices.

Unfortunately, I have not found any workaround other than detecting at runtime which JS engine is in use and conditionally disabling JavaScript-to-Java behavior in our app.

I tried in HTC desire with a custom rom of 2.3, it's not crashed but the javascriptinterface is not called. The same is in the emultor. My app relies on this feature. please fix it asap.

The addJavascriptInterface seems to work fine, the error seems to be raised once a method is invoked through the interface.
Leaving off any method invokation and keeping the (this way useless, so should be fixed) addJavascriptInterface seems to avoid the crash.

Anyone else got this same result?

As real device I have a Galaxy S I9000 updated on Gingerbread via official channels, it works fine (I suppose it uses the V8 interpreter).

 

Just to confirm comment 117. I compiled android 2.3.4 using JS_ENGINE=v8 to get a emulator image using v8 Javascript (defaults to jsc). This problem is not reproducible in the emulator. (though this configuration has other problems: most Javascript dies with ILL_ILLOPC exception, due to v8 requiring a later instruction set than the emulator provides).

I guess older phones should only upgrade to 2.2.

this problem seems  caused by template callJNIMethod<>, replace it with direct function define and replace all call in JNIBridge.cpp JavaClassJSC.cpp JavaInstanceJSC.cpp, it works fine on emulator,but it is strange  why v8 engine doesn't have this problem, it also call callJNIMethod<> func

Very serious bug, fix please!

This is a total NFC showstopper. Is this politically intentional? Please fix this ASAP.

This is frustrating.  Please fix, Google.

Droid2 testing device running 2.3.3 runs fine. Please fix emu.

We're using phoneGap 1.0 and an app using file:/// based html files.
we had the problem of unwanted javscript "gap_poll:" prompts on Android phones and emulators.

We found the problem went away when we put all the files in one folder (the www folder).

unblievable

[Comment deleted]

Fix it please, google.
It's a very serious bug.

It's really a critical bug, please fix it.

This critical bug has been there for almost a year and google is not showing any interest of fixing.  frustrating indeed

fuck google, it is real a serious bug.

Come on man, I don't believe it's unrepairable...

This bug interacts badly with the ability to open ParcelFileDescriptor pipes (which requires Gingerbread) to provide additional content:// URL's for custom WebView content. The workaround isn't realistic for anything other than the most trivial triggers (i.e. you can't pass complex data structures back and forth).

this bug is blocking, not just medium priority. fix this ASAP, google. your sells depends on customers being able to buy recent apps

This is still an issue almost 1 year on.
I have an app I'd like to publish.
It will not work on 80-90% of devices. 4.0 is not out for updates yet and even then a small percentage of users with that version for the next year as the up_X work thru the system.

Experiencing the same problem.  This is a critical issue and needs to be fixed!

Try mobilelite framework: http://code.google.com/p/mobilelite/
We have already solved this problem in the lightweight framework.

Hmmm, Android is a nightmare for developers, NICE!  Things just don't get fixed I guess.  

I have just implemented the fully synchronous workaround using WebViewChrome::onJsPrompt().

@Override
public boolean onJsPrompt(WebView view, String url,
                String message, String defaultValue,
                JsPromptResult result) {
       
        if(TextUtils.isEmpty(message) || !message.startsWith(SIGNATURE))
                return false;
       
        String json = message.substring(SIGNATURE.length());
        try{
                //It is our object
                JSONObject info = new JSONObject(json);
                JSONArray params = info.getJSONArray("params");
                int len = params.length();
                Object []paramValues = new Object[len];
                for(int i=0; i<len; ++i){
                        Object obj = params.opt(i);
                        paramValues[i] = obj;
                }
       
                Method method = ReflectionUtils.findOnlyMethod(api.getClass(), info.getString("method"));
                Object ret = method.invoke(api, paramValues);
               
                JSONObject res = new JSONObject();
                res.put("result", ret);
                result.confirm(res.toString());
        }catch(JSONException e){
                api.m_error_message = "Couldn't parse rpc call: " + json;
                result.confirm(null);
                return true;
        }catch(NoSuchMethodException e){
                api.m_error_message = "Can not find method: " + e.getMessage() + " on data " + json;
                result.confirm(null);
                return true;
        }catch(InvocationTargetException e){
                api.m_error_message = "Can not invoke method: " + e.getMessage() + " on data " + json;
                result.confirm(null);
                return true;
        }catch(IllegalAccessException e){
                api.m_error_message = "Can not access method: " + e.getMessage() + " on data " + json;
                result.confirm(null);
                return true;
        }
        return true;
}



On the javascript side it needs some changes as well, for example:

function callApiWorkaround(name, args){
  if(!args) args = [];
  var ret = prompt(g_api_signature + stringify({method: name, params: args}));
  if(ret && typeof ret == 'string'){
    ret = parse(ret);
    return ret.result;
  }
  return ret;
}

function api_trace(msg, callee){
  var method = 'trace';
  if(api)
    return api[method](msg, callee);
  return callApiWorkaround(method, [msg, callee]);
}

With the workaround in comment 140 the only problem left is how to determine when it should be used instead of addJavaScriptInterface.

The comment 117 gives us a clue that it should be used when there is JSC on Gingerbread, but I don't think it's enough, because in comments 53, 54 they mentioned bugs on 2.2 phones.

Hi
I am facing the same issue . Has anyone found a fix or workaround for this yet? Any help is appreciated.

Thanks

Hi,

I'm facing with this issue but i got some what different info in my logcat i'm attaching here..For the total project i have used 2.3.3,so for maps also i used this but now i'm facing this issue.How can i over come this .please help me in solving this issue.

For those who lacks detailed solution, I present the more detailed one.
Say for calling simple toast function :

wb.addJavascriptInterface(new JavaScriptInterface(this), "Android");

In class JavaScriptInterface :
public void showToast(String toast) {
       Toast.makeText(mContext, toast, Toast.LENGTH_SHORT).show();
}

suppose api is JavaScriptInterface implementation
wb is WebView
SIGNATURE is string constant defined by you to identify JSON RPC calls

First you need to implement WebChromeClient interface:

wb.setWebChromeClient(new WebChromeClient(){
    @Override
    public boolean onJsPrompt(WebView view, String url,
                    String message, String defaultValue,
                    JsPromptResult result) {

            if(TextUtils.isEmpty(message) || !message.startsWith(SIGNATURE))
                    return false;

            String json = message.substring(SIGNATURE.length());
            try{
                    // Это наш объект
                    JSONObject info = new JSONObject(json);
                    JSONArray params = info.getJSONArray("params");
                    int len = params.length();
                    Object []paramValues = new Object[len];
                    for(int i=0; i<len; ++i){
                            Object obj = params.opt(i);
                            paramValues[i] = obj;
                    }

                    Method method = ReflectionUtils.findOnlyMethod(api.getClass(),
                                info.getString("method"));
                    Object ret = method.invoke(api, paramValues);

                    JSONObject res = new JSONObject();
                    res.put("result", ret);
                    result.confirm(res.toString());
            }catch(JSONException e){
                    api.m_error_message = "Couldn't parse rpc call: " + json;
                    result.confirm(null);
                    return true;
            }catch(NoSuchMethodException e){
                    api.m_error_message = "Can not find method: " + e.getMessage()
                        + " on data " + json;
                    result.confirm(null);
                    return true;
            }catch(InvocationTargetException e){
                    api.m_error_message = "Can not invoke method: " +
                        e.getMessage() + " on data " + json;
                    result.confirm(null);
                    return true;
            }catch(IllegalAccessException e){
                    api.m_error_message = "Can not access method: " +
                        e.getMessage() + " on data " + json;
                    result.confirm(null);
                    return true;
            }
            return true;
    }
});

//Now we can add javascript interface. In order to deal with gingerbread bug we can
//check current API level and if it is Gingerbread just not add the interface.
//Or as in this case we can pass api level to javascript for it to determine if it should call
//prompt instead of javascript interface
wb.addJavascriptInterface(new JavaScriptInterface(this), "Android");


onJsPrompt uses ReflectionUtils.findOnlyMethod:

/**
 * Utilities for reflection.
 */
public class ReflectionUtils {
       private static Map<String, Set<Method>> METHOD_CACHE
               = new HashMap<String, Set<Method>>();

       private ReflectionUtils() { }
       /**
        * Finds methods with the given name on the given class.
        * @param clazz the class
        * @param name the method name
        * @return the methods
        */
       public static Set<Method> findMethods(Class<?> clazz, String name) {
               String cacheKey = clazz.getName().concat("::").concat(name);
               if (METHOD_CACHE.containsKey(cacheKey)) {
                       return METHOD_CACHE.get(cacheKey);
               }
               Set<Method> methods = new HashSet<Method>();
               for (Method method : clazz.getMethods()) {
                       if (method.getName().equals(name)) {
                               methods.add(method);
                       }
               }
               methods = Collections.unmodifiableSet(methods);
               METHOD_CACHE.put(cacheKey, methods);
               return methods;
       }

       /**
        * Finds methods with the given name on the given class.
        * @param clazz the class
        * @param name the method name
        * @return the methods
        * @throws NoSuchMethodException
        */
       public static Method findOnlyMethod(Class<?> clazz, String
name) throws NoSuchMethodException {
               Set<Method> set = findMethods(clazz, name);
               if(set.size() != 1)
                       throw new NoSuchMethodException(name + " zero or not
exactly one");
               return set.iterator().next();
       }
}

So the java part is done. Now let's make changes to the client.
Instead of just calling
Android.showToast('something') make the following:

var g_api_signature = 'SIGNATURE value from java part';
var g_api_androidAPI = 9; //This should be set to current
//android API level somehow from java part

var AndroidAPI = function(){
   function isJSC(){
       //You can determine which JS engine is in use at runtime through
       //the following javascript snippet:
       //(from http://code.google.com/p/android/issues/detail?id=12987#c117)
               return window.devicePixelRatio &&
                       escape(navigator.javaEnabled.toString()) !==
                       'function%20javaEnabled%28%29%20%7B%20%5Bnative%20code%5D%20%7D';
   }

   function isGingerbread(){
       var androidAPI = g_api_androidAPI;
               return 9 <= androidAPI && androidAPI <= 10;
   }

   function shouldUseGingerbreadWorkaround(){
       //The addJavaScriptInterface is encountered on gingerbread with
       //JSC javascript engine
       //So use workaround
       return isGingerbread() && isJSC();
   }

   function ApiSystemError(msg){
       this.name = 'AndroidApiError';
       this.message = msg;
   }

   var api = shouldUseGingerbreadWorkaround() ? null : window.Android,
       stringify = JSON.stringify, //You should also include json2.js
       parse = JSON.parse; //from https://github.com/douglascrockford/JSON-js

   function callAndroid(name, args){
       if(!args) args = [];
       var ret = promptfunc(g_api_signature + stringify({method: name,
                            params: args}));
       if(ret && typeof ret == 'string'){
               if(ret.charAt(0) == '{'){
                       ret = parse(ret);
                       return ret.result;
               }else{
                       //Error, a JSON object should have returned
                       throw new ApiSystemError(ret);
               }
       }else{
               throw new ApiSystemError('Api returned nothing!');
       }
   }

   /* the api_xxx functions are workaround for gingerbread bug *
    * http://code.google.com/p/android/issues/detail?id=12987 */
   function api_showToast(str){
       var method = 'showToast';
       //Please note that it is not even necessary for javascript
       //interface to be really added. In such case we just call
       //Java through prompt
       if(api)
               return api[method](str);
       return callAndroid(method, [str]);
   }

   return {
       showToast: function(str){
           return api_showToast(str);
       }
   };

}();

//Safely show Toast:
AndroidAPI.showToast('something');

Well, something like that. Enjoy.

@ comment 144:
I have tried the code you have posted above, i am getting some other errors..
 and what is "api" in java code..
Can you please attach a sample project where you have enabled this??

Thanks,
Deepak K Garg
gargengineer.blogspot.com

Here's a workaround that has worked for me with minimal pain: http://stackoverflow.com/questions/7424510/uncaught-typeerror-when-using-a-javascriptinterface 

So, I assume this has still not been fixed judging by the recent activity.

In summary...

I still get this error on 2.3.1 and 2.3.3 emulators.

However does it actually affect 2.3.3 devices? There seems to be some dispute in the thread about whether it does or not.

What a Terrible Failure (WTF)
Please fix asap

fatality... mortality... bugality!
Please fix it

Any news about this issue?
I'm experiencing the same problem with 3.1 and 4.0 version.

I also get this error on android 2.3.3 emulators.

Just got this error on android 2.3.3 emulators too.

[2012-03-22 11:02:48 - LogCatFilter] Ignoring invalid text regex.
[2012-03-22 11:02:48 - LogCatFilter] Dangling meta character '*' near index 0
*:S
^
I also get this error on android 2.3.3 emulators and htc wildfire S

I also get this error on android 2.3.3 emulator on Windows 7 Japanese.
(BTW, my au/KDDI IS03 (Android 2.2.1) do not get this error)

i wonder whether the issue was fixed

FIXED!!!PLS!!! Google!!!

[Comment deleted]

[Comment deleted]

I can reproduce this on Galaxy Nexus

I wasted half a day trying to figure out why code that ran fine in 1.6 was crashing in 2.3.3 emulator.

I re-ran into this issue today. What is shameful is that it has been 18 months since the issue was reported with apparently no fix or response from the Android team.  (If there is a response, please link to it here).  If only a small fraction of Android devices were still running 2.3.x then this problem would be a non-issue.  This obviously isn't the case.