popa

This instruction pops all the general purpose registers off the stack in the reverse order of PUSHA. That is, EDI, ESI, EBP, ESP, EBX, EDX, ECX, EAX. Used to restore state after a call to PUSHA.

Code:

IF OperandSize = 32 (* instruction = POPAD *)
THEN
EDI <- Pop();
ESI <- Pop();
EBP <- Pop();
increment ESP by 4 (* skip next 4 bytes of stack *)
EBX <- Pop();
EDX <- Pop();
ECX <- Pop();
EAX <- Pop();
ELSE (* Oper    

... does that mean on some old cpus (E)SP is poped and some new ones doesn't?

xor eax, eax
xor ebx, ebx
xor ecx, ecx
xor edx, edx
xor edi, edi
xor esi, esi
xor ebp, ebp
    

xor eax,eax
mov ebx,eax
mov ecx,eax
mov edx,eax
mov edi,eax
mov esi,eax
mov ebp,eax
    

I think this could be maybe executed faster:

Code:

xor eax,eax
mov ebx,eax
mov ecx,eax
mov edx,eax
mov edi,eax
mov esi,eax
mov ebp,eax
    

So depending on how pipeline execution is optimized it could be faster with a mov instruction than always "calculating" and using/blocking the internal ALU for an XOR operation. I would say an XOR operation is more "expensive" than a simple mov (copy of data).

I doubt it would make any difference except for the extra dependency you have from eax propagating to the other registers.

revolution wrote:

I doubt it would make any difference except for the extra dependency you have from eax propagating to the other registers.

In fact this is a very untypical piece of code. I never saw a situation where all or so many registers have to be preloaded with 0. So at the end this is one of the endless academic approaches.

I was going to jump to another process. Some registers may contain private value so I had to do this