This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Fix x86intrin.h with -Wsystem-headers
- From: Jakub Jelinek <jakub at redhat dot com>
- To: Uros Bizjak <ubizjak at gmail dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Fri, 6 Nov 2009 11:03:31 +0100
- Subject: Re: [PATCH] Fix x86intrin.h with -Wsystem-headers
- References: <5787cf470911060158m7bb972der63a32146d92a3c7d@mail.gmail.com>
- Reply-to: Jakub Jelinek <jakub at redhat dot com>
On Fri, Nov 06, 2009 at 10:58:02AM +0100, Uros Bizjak wrote:
> > Both fma4intrin.h and avxintrin.h define the same 4 types, which is an error
> > except for system headers.
> > My understanding is that -mfma4 implies -mavx and -mfma4 -mno-avx is the
> > same as -mno-fma4, plus fma4intrin.h can be only included from x86intrin.h.
> > Thus, in all cases when fma4intrin.h is included, avxintrin.h is also
> > included and defines the types too.
> > This patch makes sure that fma4intrin.h comes after avxintrin.h and uses
> > the avxintrin.h definitions instead of providing its own ones.
> >
> > Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
>
> Is there a reason that you also moved mm3dnow.h ?
I've actually moved the fma4/xop/lwp headers to the end of file, rather than
moving the other headers. My thought was that newer extensions should come
later than older ones, and 3dnow is certainly older than fma4, etc.
But it is not something I care too much about, just that the version I've
posted has been tested already.
Jakub