gdb - nop with extra data, why?

问题

I am currently writing simple tests to understand how gdb compiles from C++ to asm using extern "C" and option -O0 and my asm object is compiled with nasm -g -f elf64

Here is my disassembled concerned stack frame:

   0x0000000000400570 <+0>: push   rbp
   0x0000000000400571 <+1>: mov    rbp,rsp
   0x0000000000400574 <+4>: push   r10
=> 0x0000000000400576 <+6>: mov    r10,QWORD PTR [rbp-0x8]
   0x000000000040057a <+10>:    pop    r10
   0x000000000040057c <+12>:    mov    rsp,rbp
   0x000000000040057f <+15>:    pop    rbp
   0x0000000000400580 <+16>:    ret    
   0x0000000000400581 <+17>:    nop    WORD PTR cs:[rax+rax*1+0x0] ; this instruction
   0x000000000040058b <+27>:    nop    DWORD PTR [rax+rax*1+0x0]   ; and this one

The two last instructions are padded with nop, I can get this because of alignment.

Now my question is why nop are being along WORD PTR cs:[rax+rax*1+0x0] and DWORD PTR [rax+rax*1+0x0] ?

回答1:

Those are multi-byte nop instructions, used as padding for alignment purposes. In this case they won't ever be executed, so the compiler could have used anything but under other circumstances it may need to be executable. See also the intel optimization manual, section 3.5.1.9 Using NOPs, as well as the instruction set reference entry for NOP, of course.

来源：https://stackoverflow.com/questions/22486415/gdb-nop-with-extra-data-why