Dealing with 16 bit characters using sun.misc.Unsafe

Question

Help me understand something.

Using .. import sun.misc.Unsafe;

If i need to put a character 0-127 to some memory address, in order to prevent out of range chars, i do this

if (0 != (c & 0xFF80)) {
    throw new RuntimeException("Only Ascii characters are supported. 0-127.");
} else {
    // Since Java's chars are 16 bits long, i cast it 
    // to byte (8 bits) and then `putByte()`
    UNSAFE.putByte(address, (byte) c);
}

When my test checks what's in that memory

    char c1 = Ascii.a;
    asciiEncoder.encode(address, c1);
    assertThat(unsafe.getChar(address), is(c1));

This test seems to pass and fail on some machines. Let me repeat that - this works on some MACs and not the others and some Windows and not the others.

What is failing? Character different to 'a' is returned.

As i understand it, this is possible if upon allocation, memory is not cleaned up, but does not this take care of this?

@Before
public void setUp() {
    address = unsafe.allocateMemory(64); // Isn't this mem clean?
} 

If i change this to 8 bits, test passes always

@Before
public void setUp() {
    address = unsafe.allocateMemory(8);
} 

Please help me understand

Answer 1

JavaDoc for Unsafe#allocationMemory says:

Allocates a new block of native memory, of the given size in bytes. The contents of the memory are uninitialized; they will generally be garbage. The resulting native pointer will never be zero, and will be aligned for all value types. Dispose of this memory by calling #freeMemory , or resize it with #reallocateMemory .

Which means when you put byte there and read char back you get extra byte of garbage.