GCC Optimization?

GCC Optimization?

[asmfreshmeat]

Hello,

where is one question I didnt found a good answer. 
When I'm disassembling the main function with gdb or objdump
I see always code like this:

0x08048354 <main+0>:    push   %ebp
0x08048355 <main+1>:    mov    %esp,%ebp
0x08048357 <main+3>:    sub    $0x8,%esp
0x0804835a <main+6>:    and    $0xfffffff0,%esp
0x0804835d <main+9>:    mov    $0x0,%eax
0x08048362 <main+14>:   sub    %eax,%esp

I understand what the first 3 lines are good for. Pushing the 
Basepointer, moving the Basepointer to the Stackpointer,
reserving 8 Bytes on the stack for variables.

But why do they use and .... I think and does only even %esp
so that the last 4 Bits are set to Zero. 
But if %esp was 0xcccccccc before it would now be 0xccccccc0
what happens to the 4 Bits from 0xccccccc0 to 0xcccccccc?
And what is this good for?

Second question:
What does mov $0x0,%eax? Why dont they use xor %eax and what is it good for?

Third question: 
What is sub %eax,%esp good for? We subtract 0 from esp... mmmhh... for what?

Best regards

-- 
Geschenkt: 3 Monate GMX ProMail + 3 Top-Spielfilme auf DVD
++ Jetzt kostenlos testen http://www.gmx.net/de/go/mail ++

Re: GCC Optimization?

[Brian Raiter]

> But why do they use and .... I think and does only even %esp
> so that the last 4 Bits are set to Zero. 
> But if %esp was 0xcccccccc before it would now be 0xccccccc0
> what happens to the 4 Bits from 0xccccccc0 to 0xcccccccc?
> And what is this good for?

The x86 push and pop instructions are much, much less efficient if the
stack pointer is not dword-aligned. So if one function (say) pushes a
segment register onto the stack, it can affect the performance of
every function that is called from there.

> Second question:
> What does mov $0x0,%eax? Why dont they use xor %eax and what is it
> good for?

It zeros the accumulator, of course. As for why they don't use xor,
why should they? What would be the advantage?

> Third question: 
> What is sub %eax,%esp good for? We subtract 0 from
> esp... mmmhh... for what?

Did you try building the same code with gcc optimizations turned on?

b

Re: GCC Optimization?

[an0nym]

>> But why do they use and .... I think and does only even %esp
>> so that the last 4 Bits are set to Zero. 
>> But if %esp was 0xcccccccc before it would now be 0xccccccc0
>> what happens to the 4 Bits from 0xccccccc0 to 0xcccccccc?
>> And what is this good for?
>
>The x86 push and pop instructions are much, much less efficient if the
>stack pointer is not dword-aligned. So if one function (say) pushes a
>segment register onto the stack, it can affect the performance of
>every function that is called from there.
Alright. Thanks!


>> Second question:
>> What does mov $0x0,%eax? Why dont they use xor %eax and what is it
>> good for?
>
>It zeros the accumulator, of course. As for why they don't use xor,
>why should they? What would be the advantage?
xor is imho a little bit faster than mov and xor %eax,%eax does the same
like mov $0x0,%eax. But I still don't know why the accumulator has to be
zeroed.


>> Third question: 
>> What is sub %eax,%esp good for? We subtract 0 from
>> esp... mmmhh... for what?
>
>Did you try building the same code with gcc optimizations turned on?
No I didn't use any optimiziation, I just thought of 'default'
optimiziations without turning -o2 or -o3 on.
So what is this code good for? I subtracts zero from %esp and saves the
result in %esp. So this code doesn't do anything with %esp. But I think
there should be a purpose?!

Best regards

-- 
Geschenkt: 3 Monate GMX ProMail + 3 Top-Spielfilme auf DVD
++ Jetzt kostenlos testen http://www.gmx.net/de/go/mail ++

Re: GCC Optimization?

[Mateusz Marzantowicz]

> Second question:
> What does mov $0x0,%eax? Why dont they use xor %eax and what is it good for?
> 

It's because GCC was designed for more processors than only x86 based 
and in some cases it won't work. With optimization enabled you will find 
there xorl %eax, %eax instead of mov $0x0,%ea

Regards