Re: Predefine _M_IX86 in gcc

Cosmin Truta <[email protected]>
Newsgroups gmane.comp.gnu.mingw.patches
Message-ID <[email protected]>
On Sat, 21 Dec 2002, Danny Smith wrote:

> > Cosmin Truta wrote:
> > >
> > > I've just looked into the mingw specs, and I saw in the long line
> > >   "... march=i486:-D__i486 -D__i486__ ..."
> > > which can go into
> > >   "... march=i486:-D__i486 -D__i486__ -D_M_IX86=400 ..."


Earnie Boyd wrote:

> [...] However, should the macro definition be based on mcpu or march?


Danny Smith wrote:

> -march, if it is to be useful  It would then provide information on what
> insn set is available, not just how scheduling of the i386 insn set is
> tuned.  The __i386__, __i486__  predefines  are based on -march.


I wondered about this myself. Now I see: it should be -march.


Danny again:

> If this is high priority for you, you will need to provide patchset
> against GCC trunk (which will be 3.4 in about 8-9 months time).   Those
> lines in the specs file are generated from defines in i386.h in GCC
> source. I don't know if Linux and other i386 targets will will want them
> so the _M_IX86 definitions may have to go in cygwin.h or mingw.h

My original interest was actually in having GLUT working under MinGW.
_M_IX86 is a low-priority issue for me, too, but I am willing to take
care of it.

First, I plan to ask the Cygwin folks what do they think about it.
I expect they will not want it, for the same reason they don't want
-DWIN32 and -D_WIN32, but I'll ask, anyway.

I am a little busy right now. Still, I am not afraid of being late with
the submission, because gcc-3.4 is just in Stage 1, and this is a minor
patch.


Danny again:

> I like the existing GCC predifines better because they provide
> infornation on newer chips: pentium2/3/4, k6, k6-2,k6-3, athlon,
> athlon-4.  I don't know how they all equate to the _M_IX86 defines.

I don't think they do.

The existing predefines should remain of course, because they exist in
the other GCC incarnations, such as Linux, and MinGW is a GCC compiler.
_M_IX86 should be added to MinGW because it belongs to the Wintel32
platform (just as WIN32 and _WIN32 are), and MinGW is a Wintel32 compiler.


Cosmin



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.