BytePointer creates leak

[nickbernz]

In BytePointer::BytePointer(ByteBuffer buffer), hasArray() is used to check whether or not buffer is direct. But since it is possible for a direct buffer to return true on that method, this leads to a memory leak.

public BytePointer(ByteBuffer buffer) {
        super(buffer);
        if (buffer != null && buffer.hasArray()) {
            byte[] array = buffer.array();
            allocateArray(array.length - buffer.arrayOffset());
            put(array, buffer.arrayOffset(), array.length - buffer.arrayOffset());
            position(buffer.position());
            limit(buffer.limit());
        }
    }

I would suggest to using instead is isDirect().

public BytePointer(ByteBuffer buffer) {
        super(buffer);
        if (buffer != null && !buffer.isDirect()) {
            byte[] array = buffer.array();
            allocateArray(array.length - buffer.arrayOffset());
            put(array, buffer.arrayOffset(), array.length - buffer.arrayOffset());
            position(buffer.position());
            limit(buffer.limit());
        }
    }

[saudet]

Do you have an example of a direct buffer with an array?

There can also be non-direct buffers without array, so it's not safe to call array() unless hasArray() returns true...

[nickbernz]

On my Pixel, the following creates direct buffers with array.

ByteBuffer buffer = ByteBuffer.allocateDirect((1024*1024));
boolean hasArray = buffer.hasArray();
boolean isDirect = buffer.isDirect();

Maybe you could check both conditions?

if (buffer != null && !buffer.isDirect() && buffer.hasArray()) {

[saudet]

Interesting. Yes, that would be a good idea. Could you send a PR for all the primitive Pointer classes?

[nickbernz]

I am sorry, I am not sure what you are asking for.

[saudet]

A pull request to send your patch?

[saudet]

Please do send a patch to fix this issue! Thanks a lot

[saudet]

I've included in JavaCPP 1.3.2 the changes you proposed. Thanks for reporting!