I'm so exited in finding this that I couldn't resist to post this link: http://stackoverflow.com/questions/199966/how-do-you-use-gcc-to-generate-assembly-code-in-intel-syntax

Example that worked for me: gcc main.c -S -O3 -masm=intel

Google, Thanks!

COOL: http://board.flatassembler.net/topic.php?p=84571#84571

What was in "main.c" ? How many lines ?

Anyway, I had tested (using CC1.EXE from MINGW, on DOS, no DGJ-PP, of course ) in the meantime:

- Seems to work somewhat

- Syntax is mostly fixed, not fully, still strange directives and MA$M (or rather TASM ideal ?) instructions ... too much of PTR and OFFSET for my taste ... OTOH no ASSUME or PROTON encountered yet. Maybe some simple S&K (Search&Kick) process could fix it finally.

- I've seen complaints about "ma$m=intel" being "broken" in GCC ... very valuable bug report

- Tested also the O-hack: compiler is very stupid by default, but seems to be incredibly smart with -O3 : detects constant "variables" as well as unused variables, rearranges calculations, ... the problem is that is brews very broken code in some cases Of course there have been many complaints about "-O2 and above is broken" or "-O3 and above is broken" around ... so is this a known and acceptable issue of "high quality compiler" that -O2 and above is an experimental kamikaze hack producing broken results ? The compiler doesn't warn at all

DOS386: just about everything on my gentoo linux server is compiled with -O3... some older GCC versions might be broken, and some (poorly written, usually) might break at high optimization levels (because they're making assumptions that C/C++ doesn't guarantee).

GCC might not always produce as good code as Intel or MS compilers, but it's not a bad product... and it's under constant development.

As for "OFFSET" and "PTR", it's the defacto standard on intel because of TASM and MASM. I'm not saying that's a good thing or that you should like it, but it's pretty obvious that GCC is going to use defacto syntax rather than output for a relatively niche assembler as FASM.

Not much, I've just done that to verify that what I wrote is what I wanted since it is safer than re-reading the documentation and believe you understood it correctly

I've used O3 because it was much clearer to read than the default setting.

Actually I am surprised you didn't preiously know that. Even I've known that for a long time and I don't even code in C. I'm not trying to boast here, just genuinely shocked . Quite a few other compilers can also generate Intel syntax.

I know, I have posted an example generated with Visual C++ 2005 for you when you claimed that the compilers are stupid for not detect unneeded volatile specifiers when the problem to solve don't need them.

The thing I didn't know was that GCC allowed to override the loved for no reason at all by Unix community AT&T syntax.

Oh, it is not for no reason. The reason is that it makes it uniquely Unix. Even though it is an abominable syntax it does have an air of Unix-ness to it whenever people read it.

I don't understand what makes it 'Unixy'. It's part of how the compiler works and isn't really for human consumption. Personally, I have no issues with ATT syntax.

Do you feel more comfortable with it than, say, fasm, nasm or similar Intel-like syntax?

I see ATT syntax as the pleasure of the masochistic, what are the advantages of ATT over Intel syntax anyway? The official architecture manuals don't use this syntax so you even have to force your mind into permanent translation, it even disagrees with the placement of the dest and src operands.

AT&T syntax is really just a relic of Bell Labs and UNIX.

It's not around because it's loved. It's still around because many people know it.

Not more so. Just the same. There are things I prefer, such as left to right reading.
What are the advantages of Intel syntax? Just fewer dollar signs? I don't see what the big deal is. I don't know what you mean by that but I don't translate anything when I use it. iow, I don't try and translate anything else into ATT. I only think in ATT. It does not but it is inconsistent in a couple areas and not everywhere. Just like MASM is inconcistent in a few areas.

680x0 had me thinking that way for a while. AT&T being so drastically different helps with shifting the mindset, imho. If it looked closer to TASM/NASM/MASM/FASM then I'd just stare lost +15% of the time while trying to read it.

EDIT by DOS386 : fixed broken quotes (is this like you tried to do ???)