Strange instructions in compiler output

[Greg 'groggy' Lehey]

On Friday,  5 March 2004 at 18:43:11 -0500, Chungwei Hsiung wrote:
> Greg 'groggy' Lehey wrote:
>
>> On Friday,  5 March 2004 at 13:43:04 -0500, Chungwei Hsiung wrote:
>>
>>
>>> Hello..
>>> I am super new to this list, and I have a simple question that I don't
>>> know why it does that. I have a simple test program. I compile it, and
>>> gdb to disassemble main. I got the following..
>>>
>>> 0x8048201 <main+9>:     mov    $0x0,%eax
>>> 0x8048206 <main+14>:    sub    %eax,%esp
>>> ...
>>>
>>> I don't know if at line 5, we move zero to %eax. why do we need to sub
>>>> eax, %esp? why do we need to substract 0 from the stack pointer??
>>> Any help is really appreciated.
>>
>> This is probably because you didn't optimize the output.  You'd be
>> surprised how many redundant instructions the compiler puts in under
>> these circumstances.  Try optimizing and see what the code looks like.
>>
>> If this *was* done with optimization, let's see the source code.
>
> Hello.. thank you very much for the reply
> I actually don't know how to use the optimization. 

Use the gcc command line options.  See below.

>I just compile it with gcc 3.2.2, and use gdb to disassemble main to
>get this assembly. Is it possible I can get the non-redundent output?
>here is the code I compile..
>
> ...

The best way to look at the assembly output is to generate it directly
from the compiler.  I get:

$ cc -O -pipe -mcpu=pentiumpro -S exec.c
$ cat exec.s
.LC0:
        .string "/bin/sh"
...
main:
        pushl   %ebp
        movl    %esp, %ebp
        subl    $24, %esp
        andl    $-16, %esp
        movl    $.LC0, -8(%ebp)
        leal    -8(%ebp), %edx
        movl    $0, 4(%edx)
        movl    -8(%ebp), %eax
        movl    %eax, (%esp)
        movl    %edx, 4(%esp)
        movl    $0, 8(%esp)
        call    execve
        movl    $0, %eax
        movl    %ebp, %esp
        popl    %ebp
        ret

This doesn't look that much like your code.  Without the -O (optimize)
flag  I get:

$ cc  -pipe -mcpu=pentiumpro -S exec.c
$ cat exec.s
...
main:
        pushl   %ebp
        movl    %esp, %ebp
        subl    $24, %esp
        andl    $-16, %esp
        movl    $0, %eax
        subl    %eax, %esp
        movl    $.LC0, -8(%ebp)

So yes, it looks as if you're not optimizing.

Greg
--
Note: I discard all HTML mail unseen.
Finger grog at FreeBSD.org for PGP public key.
See complete headers for address and phone numbers.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-hackers/attachments/20040306/be2d077c/attachment.bin