I'm currently working on Mangler's Android implementation.
I have a java class that looks like so:
public class VentriloEventData {
public short type;
public class _pcm {
public int length;
public short send_type;
public int rate;
public byte channels;
};
_pcm pcm;
}
The signature for my pcm object:
$ javap -s -p VentriloEventData
...
org.mangler.VentriloEventData$_pcm pcm;
Signature: Lorg/mangler/VentriloEventData$_pcm;
I am implementing a native JNI function called getevent, which will write to the fields in an instance of the VentriloEventData class. For what it's worth, it's defined and called in Java like so:
public static native int getevent(VentriloEventData data);
VentriloEventData data = new VentriloEventData();
getevent(data);
And my JNI implementation of getevent:
JNIEXPORT jint JNICALL Java_org_mangler_VentriloInterface_getevent(JNIEnv* env, jobject obj, jobject eventdata) {
v3_event *ev = v3_get_event(V3_BLOCK);
if(ev != NULL) {
jclass event_class = (*env)->GetObjectClass(env, eventdata);
// Event type.
jfieldID type_field = (*env)->GetFieldID(env, event_class, "type", "S");
(*env)->SetShortField(
env,
eventdata,
type_field,
1234
);
// Get PCM class.
jfieldID pcm_field = (*env)->GetFieldID(env, event_class, "pcm", "Lorg/mangler/VentriloEventData$_pcm;");
jobject pcm =
(*env)->GetObjectField(
env,
eventdata,
pcm_field
);
jclass pcm_class = (*env)->GetObjectClass(env, pcm);
// Set PCM fields.
jfieldID pcm_length_field = (*env)->GetFieldID(env, pcm_class, "length", "I");
(*env)->SetIntField(
env,
pcm,
pcm_length_field,
1337
);
free(ev);
}
return 0;
}
The code above works fine for writing into the type field (that is not wrapped by the _pcm class). Once getevent is called, data.type is verified to be 1234 at the java side :)
My problem is that the assertion "pcm != NULL" will fail. Note that pcm_field != NULL, which probably indicates that the signature to that field is correct... so there must be something wrong with my call to GetObjectField. It looks fine though if I compare it to the official JNI docs.
Been bashing my head on this problem for the past 2 hours and I'm getting a little desperate.. hoping a different perspective will help me out on this one.
