This will hopefully fix the retry timeout occurring with simultaneous SMB
writes (subsequent threads were starving the initial thread that owned the
lock).