Intel assembly syntax with GCC

[Ktulu]

Greetings,

GCC uses AT&T/UNIX assembly syntax which I really detest. Fortunately GCC has provided us with the '-masm=intel' switch to use Intel assembly syntax instead. The only problem that I have is to compile even the most simple inline assembly code when using the Intel assembly syntax. Does anyone has any experience with the Intel assembly syntax when compiling code with GCC?

For example, the following two code snippets do not compile:

DWORD dwTest;

main()
{

    asm( ".intel_syntax noprefix;" );
    asm( "mov ds:_dwTest, 0xDEADBEEF;" );

    return 0;
}

main()
{

    DWORD dwTest;

    asm( ".intel_syntax noprefix;" );
    asm( "mov eax, _dwTest;" );

    return 0;
}

Could someone enlighten me or tell me how I should interpret the Intel assembly syntax with GCC?

Thanks for your time,
Ktulu

[GReaper]

The GCC produces AT&T syntax by default, therefore when using the inline switch to intel, the assembler is expecting Intel while GCC gives AT&T.

EDIT: Therefore, when switching like that, in inline assembly, you must always switch back after your assembly code finishes.

[_Mike]

Could someone enlighten me or tell me how I should interpret the Intel assembly syntax with GCC?

Thanks for your time,
Ktulu

I believe what you're looking for is one of

mov dword ptr ds:dwTest, 0xDEADBEEF
mov dword ptr [dwTest], 0xDEADBEEF

You need to specify the operand size, and get rid of the underscore.

The GCC produces AT&T syntax by default, therefore when using the inline switch to intel, the assembler is expecting Intel while GCC gives AT&T.

EDIT: Therefore, when switching like that, in inline assembly, you must always switch back after your assembly code finishes.

Maybe I'm misunderstanding what you're trying to say; But if you mean that you have to switch back to AT&T or the compiler will output incorrect assembly code from C code, then no that's not true.

[GReaper]

Maybe I'm misunderstanding what you're trying to say; But if you mean that you have to switch back to AT&T or the compiler will output incorrect assembly code from C code, then no that's not true.

Not incorrect code, just the wrong type. And yes, it's true!

[_Mike]

Not incorrect code, just the wrong type. And yes, it's true!

'gcc -masm=intel -save-temps test.c'
test.c

#include <stdio.h>

unsigned int i = 0;

int main()
{
    printf("%d\n", i);
    asm(".intel_syntax noprefix\n");
    asm("mov dword ptr [i], 1\n");
    printf("%d\n", i);

    return 0;
}

test.s

        .file   "test.c"
        .intel_syntax noprefix
.globl i
        .bss
        .align 4
        .type   i, @object
        .size   i, 4
i:
        .zero   4
        .section        .rodata
.LC0:
        .string "%d\n"
        .text
.globl main
        .type   main, @function
main:
.LFB0:
        .cfi_startproc
        push    rbp
        .cfi_def_cfa_offset 16
        mov     rbp, rsp
        .cfi_offset 6, -16
        .cfi_def_cfa_register 6
        mov     edx, DWORD PTR i[rip]
        mov     eax, OFFSET FLAT:.LC0
        mov     esi, edx
        mov     rdi, rax
        mov     eax, 0
        call    printf
#APP
# 8 "test.c" 1
        .intel_syntax noprefix

# 0 "" 2
# 9 "test.c" 1
        mov dword ptr [i], 1

# 0 "" 2
#NO_APP
        mov     edx, DWORD PTR i[rip]
        mov     eax, OFFSET FLAT:.LC0
        mov     esi, edx
        mov     rdi, rax
        mov     eax, 0
        call    printf
        mov     eax, 0
        leave
        ret
        .cfi_endproc
.LFE0:
        .size   main, .-main
        .ident  "GCC: (GNU) 4.4.3"

Compiles and runs just fine. Notice the lack of AT&T assembly.
Btw, apparently the '-masm=intel' switch makes '.intel_syntax noprefix' implicitly specified. I didn't know that until now

[GReaper]

Try the exast same code without the command -masm=intel and see what happens.