Moving values around
The same program in M68K, MIPS, RISC-V, Z80.
This is the first program of the ladder, and it exists to show where a value can come from. There are
three places: an immediate, which is a number written into the instruction itself; another
register; and memory, named by a label in square brackets. One instruction, mov, reaches all
three, which is why most lines of most x86 programs are a mov.
Type 402000 into the memory panel and the eight bytes there read 69 00 00 00 00 00 00 00, which
is the answer stored little endian, lowest byte first.
rax is the one register that does not end where the program left it. It holds 10 for two
instructions and then the exit at the bottom takes it for the call number, so what you read in the
panel at the end is 3C, which is 60. A register belongs to whoever wrote it most recently.
Look down the six working lines and count the square brackets. There is never more than one pair on a
line, and that is not an accident: two memory operands are not an instruction, so copying value
into a second variable would take two lines and a register in between. One pair is allowed anywhere,
though, so imul rcx, [value] would have multiplied straight out of memory and saved a line.
Change mov rcx, [value] to lea rcx, [value] and the program still runs, still multiplies, and
produces nonsense. rcx now holds 402000, the address the variable lives at rather than the 7
inside it, and the multiplication that follows is arithmetic on an address. The brackets are the whole
difference between the two lines.