How can I call a method with this method signature in C from JNA?
int open_device(context *ctx, device **dev, int index);
The last two lines of the C method look like this:
*dev = pdev;
return 0;
That's the only use of dev
in that method. That means that I have to pass a poiner to an empty pointer to the method, right? The method then fills the empty pointer with the address of a device
object and I can pass the pointer to the device to other methods.
My question is: Is this the right way to do that? If it is, how do I allocate a new pointer from Java?
Based on the accepted answer, I did this:
Memory p = new Memory(Pointer.SIZE);
Memory p2 = new Memory(Pointer.SIZE);
p.setPointer(0, p2);
nativeLib.open_device(ctx, p, index);
return p2;
It appears that the JNA Pointer
class has setPointer
and getPointer
methods to allow for multiple indirection, and the Memory
class to actually "allocate" native objects. So you should be able to do something like: (I'm just guessing from the JNA docs, I've not tested this)
Pointer pDev = new Memory(Pointer.SIZE); // allocate space to hold a pointer value
// pass pDev to open_device
Pointer dev = pDev.getPointer(0); // retrieve pointer stored at pDev
There are no pointers in Java, only references.
You cannot reassign a reference when you pass them to methods, because you pass them by value. Everything is passed by value in Java.
You can rewrite this method to instantiate a new instance of the device and return that instead of an int.
Even better answer. You may want to allocate (malloc etc) depending upon Java String length. Example below is unit test from JNA project.
public void testGetSetStringWithDefaultEncoding() throws Exception {
final String ENCODING = Native.DEFAULT_ENCODING;
String VALUE = getName();
int size = VALUE.getBytes(ENCODING).length+1;
Memory m = new Memory(size);
m.setString(0, VALUE);
assertEquals("Wrong decoded value", VALUE, m.getString(0));
}