Re: [i386/stabs] Arguments of main on gcc >= 4.1

[Joel Brobecker <brobecker@adacore.com>]

> gcc 3.4.4 doesn't emit the stack-alignment code.  Unless
> the comments in i386-tdep.c are wrong, any gcc < 4.1 doesn't emit
> the stack-alignment code, which makes them immune to the patch.
> I did regtest the patch on Cygwin.

That's good to know.

> >The fact that this has nobody before you reported that this is
> >broken since 4.1 shows that this is probably not an extremely
> >important issue. 
> 
> I've seen this reported at the mingw-users@ list with the
> new gcc4.2 binaries.  As you know, mingw support wasn't in the
> FSF tree, so problems seen on mingw weren't being reported here.
> 
> But sure, this isn't certainly my priority ;-)

Just to be clear, I didn't mean to brush off the patch. What I meant
was that, if we break compatibility with old compilers, it isn't
going to be extremely damaging (given that other people have lived
with the same problem for so long already).

> Even if gcc debug info isn't fixed, there will be several gcc
> releases with the problem and (pending confirmation), no
> release with stack alignment without the problem.

The problem if they fix GCC is that we would then have to support
both conventions. If we can't auto-detect, then perhaps we'll have
to introduce a set command. Ugh :-(.

> I'll try to break the patch further -- the setup for the
> workaround is mostly bugfixing, and I'll come back to the
> workaround later on if I'm still using a stabs outputting
> compiler by then :-)

I definitely recommend DWARF. We have been using DWARF on Windows
for a couple of years, I think and it works great!

-- 
Joel