JNI直接缓冲区。(JNI direct buffers. Who is responsible for native buffer freeing?)

 假设一方面我们有一个用env->NewDirectByteBuffer()创建的直接字节缓冲区。 另一方面，我们有类似的直接缓冲区，但是使用ByteBuffer.allocateDirect()创建。 显然，这两个对象都应该由JVM以相同的方式管理，包括后台本机缓冲区的管理，第一种情况是由用户提供，第二种情况是由JVM从本机堆分配的。 
 
 当然，JVM必须在第二个对象的GC期间释放后备缓冲区（用ByteBuffer.allocateDirect()实例化）。 
 
 我的问题是：JVM是否会尝试在第一个对象的GC期间解除分配缓冲区（用env->NewDirectByteBuffer()实例化）？ 
 
 PS我也没有在JNI医生处找到明确的答案。 最有用的信息是http://www.ibm.com/developerworks/library/j-nativememory-linux/index.html ： 
 
 
 
 Direct ByteBuffer对象自动清理它们的本机缓冲区，但只能作为Java堆GC的一部分来执行 - 因此它们不会自动响应本机堆上的压力。 
 
 
 所以看起来JVM会在GC传递期间释放后备缓冲区，但是没有任何关于deallocator的提及。 是free()或其他东西。 

Assume in one hand we have a direct byte buffer created with env->NewDirectByteBuffer(). In other hand we have similar direct buffer, but created with ByteBuffer.allocateDirect(). Obviously both these objects should be managed by JVM in the same manner, including management of the backing native buffer, which in first case was provided by user, and in second case was allocated by JVM from native heap.
 
Of course JVM must free backing buffer during GC'ing of the second object (instantiated with ByteBuffer.allocateDirect()).
 
And my question: Will JVM try to deallocate buffer during GC'ing of the first object (instantiated with env->NewDirectByteBuffer()) ?
 
P.S. I've not found clear answer neither at JNI docs neither at SO. The most helpful info was here http://www.ibm.com/developerworks/library/j-nativememory-linux/index.html :
 
 
 
Direct ByteBuffer objects clean up their native buffers automatically but can only do so as part of Java heap GC — so they do not automatically respond to pressure on the native heap.
 
 
So it seems like JVM will free backing buffers during GC pass, but there is no any mentions about deallocator. Is it plain free() or something else.

原文：https://stackoverflow.com/questions/35363486