哈希表：不进行转换的转换？

Question

I am implementing a specialized hashtable. I'm trying store a lot of data in a single 64-bit int key, for space usage and performance reasons.

Each key should have this structure:

// Key structure, from LSB
// eval result (16 bits)
// move (16 bits)
// age (16 bits): the move of the game on which this position would have occurred
// depth (8 bits)
// node type (8 bits): from the three constants above

Here is a simple implementation:

var keys [1000]uint64
var values [1000]uint64

func Put(b *dragontoothmg.Board, m dragontoothmg.Move, eval int16, depth uint8, ntype uint8) {
    var value uint64 = uint64(eval) | (uint64(m) << 16) | (uint64(b.Fullmoveno) << 32) |
        (uint64(depth) << 48) | (uint64(ntype) << 56)
    hash := b.Hash()
    key := hash ^ value
    index := hash % uint64(len(keys))
    keys[index] = key
    values[index] = value
}

func Get(b *dragontoothmg.Board) (found bool, move dragontoothmg.Move,
    eval int16, depth uint8, ntype uint8) {
    hash := b.Hash()
    index := hash % uint64(len(keys))
    key := keys[index]
    value := values[index]
    found = (hash == (key ^ value))
    if !found {
        return false, 0, 0, 0, 0
    }
    eval = int16(value & 0xFFFF)
    move = dragontoothmg.Move((value >> 16) & 0xFFFF)
    depth = uint8((value >> 48) & 0xFF)
    ntype = uint8((value >> 56) & 0xFF)
    return
}

However, when I try to Get() the data, it comes back corrupted. I suspect this might be related to the fact that eval is a signed int, and the cast converts it to a signed uint64. What have I done wrong, and how can I fix it?

This is failing test result:

--- FAIL: TestSimpleTt (0.10s)
    transtable_test.go:37: Simple ttable test failed. 
        Put data: (board: rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 0) e2e4 -30 6 2 
         Fetched data: true h8h8 -30 255 255
FAIL

h8h8 is the maximum value of the field, for what it's worth.

Answer 1

Thanks to everyone who responded. This was the solution to my issue:

uint64(eval) for negative values was moving the sign bit to the most significant bit of the 64-bit int, which then interfered with the data that was supposed to be stored at that bit. I solved it by first converting to uint16:

uint64(uint16(eval))