Reputation: 2867
Hoping there are more knowledgeable people around, here I go. The attempts to call a java instance method from C almost succeed, but on a veery frail ground...
Problem #1 - Calling into a java method such as this one:
public void publishResult(String fromV8)
{
try
{
//final String backupBuffer = "tututu";
final String backupBuffer = String.valueOf(fromV8);
runOnUiThread(new Runnable() {
@Override
public void run()
{
((TextView)findViewById(46)).setText(backupBuffer);
}
});
} catch (Exception ex)
{
ex.printStackTrace();
}
}
works, but as soon as I try to do anything with the 'fromV8' parameter the JVM terminates with this kind of error message in LogCat:
04-26 00:16:10.714: INFO/DEBUG(14398): *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***
04-26 00:16:10.714: INFO/DEBUG(14398): Build fingerprint: 'LGE/thunderg/thunderg/thunderg:2.2/FRF91/eng.swapan.pati.20101012.153426:user/release-keys'
04-26 00:16:10.714: INFO/DEBUG(14398): >>> com.example.hellojni <<<
04-26 00:16:10.714: INFO/DEBUG(14398): signal 11 (SIGSEGV), fault addr 6c6c65bc
04-26 00:16:10.714: INFO/DEBUG(14398): r0 00000001 r1 00000001 r2 42f8bf90 r3 0022e1a8
[......plenty of registries and stack hexdump.....stripped]
04-26 00:16:10.814: INFO/DEBUG(14398): 46b5ed04 aca508fb /system/lib/libdvm.so
04-26 00:16:11.124: INFO/ActivityManager(1363): Process com.example.hellojni (pid 16632) has died.
04-26 00:16:11.134: INFO/WindowManager(1363): WIN DEATH: Window{452be390 com.example.hellojni/com.example.hellojni.HelloJni paused=false}
04-26 00:16:11.134: INFO/WindowManager(1363): Setting rotation to 0, animFlags=1
04-26 00:16:11.144: INFO/BootReceiver(1363): Copying /data/tombstones/tombstone_09 to DropBox (SYSTEM_TOMBSTONE)
04-26 00:16:11.154: INFO/ActivityManager(1363): Config changed: { scale=1.0 imsi=226/5
loc=en_GB touch=3 keys=1/1/2 nav=1/1 orien=1 layout=18 uiMode=17 seq=65}
And problem #2 - even if I do not do anything with the parameter but set a breakpoint inside the java method the JVM aborts again. Results are again in LogCat and it resembles the one above but has some extra two lines:
04-26 00:26:29.821: WARN/dalvikvm(16864): JDWP: slot 4 expected to hold object, 0x22e7b0 invalid
04-26 00:26:29.821: ERROR/dalvikvm(16864): VM aborting
[...+ SAME AS ABOVE...]
At no times have there shown up any exception traces in Console. Are these common symptoms or what ?
Thank you!
Upvotes: 0
Views: 1281
Reputation: 21
I've seen that problem before when I got lazy on object parameters passed FROM c/c++ TO Java (for me the bugbear seems to be Strings).
env->CallStaticVoidMethod(interfaceClass, sendaString2Java, "Send me to Java");
Compiles just fine but throws a runtime error:
11-26 13:08:54.770: W/dalvikvm(1818): JDWP: slot 1 expected to hold object, 0xbea7a38c invalid
If you make the object you want to pass back a "Java Object" before returning it - all is well
env->CallStaticVoidMethod(interfaceClass, sendaString2Java, env->NewStringUTF("Send me to Java"));
works just fine.
I really lean hard on the example here: http://android.wooyd.org/JNIExample/ but never really figured out why the cpp file is in the assets folder instead of a JNI folder.
Upvotes: 2
Reputation: 11040
the SIGSEGV at fault addr 6c6c65bc indicates an invalid memory reference or segmentation fault (SIGSEGV).
For exceptions from the native code, you must explicit call on C/C++ side the method ExceptionDescribe()
. More details you can find here Exceptions.
Upvotes: 0