Understanding Assembly

Question

I've got some assembly of a sorting algorithm and I want to figure out how exactly it functions.

I'm a little confused on some of the instructions, particularly the cmp and jle instructions, so I'm looking for help. This assembly sorts an array of three elements.

0.00 :        4009f8:       48 8b 07                mov    (%rdi),%rax
0.00 :        4009fb:       48 8b 57 08             mov    0x8(%rdi),%rdx
0.00 :        4009ff:       48 8b 4f 10             mov    0x10(%rdi),%rcx
0.00 :        400a03:       48 39 d0                cmp    %rdx,%rax
0.00 :        400a06:       7e 2b                   jle    400a33 <b+0x3b>
0.00 :        400a08:       48 39 c8                cmp    %rcx,%rax
0.00 :        400a0b:       7e 1a                   jle    400a27 <b+0x2f>
0.00 :        400a0d:       48 39 ca                cmp    %rcx,%rdx
0.00 :        400a10:       7e 0c                   jle    400a1e <b+0x26>
0.00 :        400a12:       48 89 0f                mov    %rcx,(%rdi)
0.00 :        400a15:       48 89 57 08             mov    %rdx,0x8(%rdi)
0.00 :        400a19:       48 89 47 10             mov    %rax,0x10(%rdi)
0.00 :        400a1d:       c3                      retq
0.00 :        400a1e:       48 89 17                mov    %rdx,(%rdi)
0.00 :        400a21:       48 89 4f 08             mov    %rcx,0x8(%rdi)
0.00 :        400a25:       eb f2                   jmp    400a19 <b+0x21>
0.00 :        400a27:       48 89 17                mov    %rdx,(%rdi)
0.00 :        400a2a:       48 89 47 08             mov    %rax,0x8(%rdi)
0.00 :        400a2e:       48 89 4f 10             mov    %rcx,0x10(%rdi)
0.00 :        400a32:       c3                      retq
0.00 :        400a33:       48 39 ca                cmp    %rcx,%rdx
0.00 :        400a36:       7e 1d                   jle    400a55 <b+0x5d>
0.00 :        400a38:       48 39 c8                cmp    %rcx,%rax
0.00 :        400a3b:       7e 0c                   jle    400a49 <b+0x51>
0.00 :        400a3d:       48 89 0f                mov    %rcx,(%rdi)
0.00 :        400a40:       48 89 47 08             mov    %rax,0x8(%rdi)
0.00 :        400a44:       48 89 57 10             mov    %rdx,0x10(%rdi)
0.00 :        400a48:       c3                      retq
0.00 :        400a49:       48 89 07                mov    %rax,(%rdi)
0.00 :        400a4c:       48 89 4f 08             mov    %rcx,0x8(%rdi)
0.00 :        400a50:       48 89 57 10             mov    %rdx,0x10(%rdi)
0.00 :        400a54:       c3                      retq
0.00 :        400a55:       48 89 07                mov    %rax,(%rdi)
0.00 :        400a58:       48 89 57 08             mov    %rdx,0x8(%rdi)
0.00 :        400a5c:       48 89 4f 10             mov    %rcx,0x10(%rdi)
0.00 :        400a60:       c3                      retq
0.00 :        400a61:       90                      nop

If someone can walk me through it, it'd be very helpful. I kind of get confused around the operands like 0x8(%rdi) and the cmp and jle instructions. Thanks.

Answer 1

Here are what the instructions mean:

mov : move
cmp : compare
jle : jump if less or equal (branch)
ret : return from procedure
nop : no-op

%r** are registers. They are usually %e** (eg: %eax, %edx, ...), but these are 64-bit registers.

As far as de-compiling the whole thing, that will take some more work.

See this: http://www.x86-64.org/documentation/assembly

Answer 2

It helps to replace the register names with proper names to trace the flow of data, and add branch labels for the control flow.

0.00 :        4009f8:       48 8b 07                mov    (%argptr),%var1
0.00 :        4009fb:       48 8b 57 08             mov    0x8(%argptr),%var2
0.00 :        4009ff:       48 8b 4f 10             mov    0x10(%argptr),%var3
0.00 :        400a03:       48 39 d0                cmp    %var2,%var1
0.00 :        400a06:       7e 2b                   jle    @v2le1
0.00 :        400a08:       48 39 c8                cmp    %var3,%var1
0.00 :        400a0b:       7e 1a                   jle    @v3le1
0.00 :        400a0d:       48 39 ca                cmp    %var3,%var2
0.00 :        400a10:       7e 0c                   jle    @v3le2

   # Now we know that 2 > 1 and 3 > 1 and 3 > 2. Write them to memory in order.

      etc