From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-return-170013-listarch-gcc=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 29904 invoked by alias); 26 Aug 2011 12:55:55 -0000
Received: (qmail 29895 invoked by uid 22791); 26 Aug 2011 12:55:55 -0000
X-SWARE-Spam-Status: No, hits=-2.3 required=5.0	tests=AWL,BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,FREEMAIL_FROM,RCVD_IN_DNSWL_LOW
X-Spam-Check-By: sourceware.org
Received: from mail-gy0-f175.google.com (HELO mail-gy0-f175.google.com) (209.85.160.175)    by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Fri, 26 Aug 2011 12:55:40 +0000
Received: by gyg4 with SMTP id 4so2878215gyg.20        for <gcc@gcc.gnu.org>; Fri, 26 Aug 2011 05:55:39 -0700 (PDT)
MIME-Version: 1.0
Received: by 10.150.61.12 with SMTP id j12mr2228268yba.22.1314363338227; Fri, 26 Aug 2011 05:55:38 -0700 (PDT)
Received: by 10.150.57.5 with HTTP; Fri, 26 Aug 2011 05:55:38 -0700 (PDT)
In-Reply-To: <CAK1BsWpuYaPPoetu96e=2H8NesLZrSFwGnZGow88vph-Ay+4kA@mail.gmail.com>
References: <CAK1BsWpuYaPPoetu96e=2H8NesLZrSFwGnZGow88vph-Ay+4kA@mail.gmail.com>
Date: Fri, 26 Aug 2011 12:55:00 -0000
Message-ID: <CAFiYyc1BD1+k=nxrp1YZtHYPXMeE9nRsh9S5y1bPY_SGEDsy_w@mail.gmail.com>
Subject: Re: linking time eroor with -fast-math -O0
From: Richard Guenther <richard.guenther@gmail.com>
To: Vladimir Yakovlev <vbyakovl23@gmail.com>
Cc: gcc@gcc.gnu.org
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
X-IsSubscribed: yes
Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc/>
List-Post: <mailto:gcc@gcc.gnu.org>
List-Help: <http://gcc.gnu.org/ml/>
Sender: gcc-owner@gcc.gnu.org
X-SW-Source: 2011-08/txt/msg00420.txt.bz2

On Fri, Aug 26, 2011 at 2:52 PM, Vladimir Yakovlev <vbyakovl23@gmail.com> w=
rote:
> Hi,
>
> Following test fails in linking if compiled with ffast-math and O0,
> but it compiled successfully with ffast-math and O2. Also no problem
> if -lm is added.
>
> $ cat t.c
> #include <stdio.h>
>
> float foo(float x)
> {
> =A0 float y =3D 0;
> =A0 while (x > 0.00000001) {
> =A0 =A0 y +=3D x*x*x*x*x*x*x*x*x*x*x*x*x;
> =A0 =A0 x =3D x/2;
> =A0 }
> =A0 return y;
> }
>
> int main (int argc, char =A0 =A0*argv[])
> {
> =A0float y =3D atoi(argv[1]);
> =A0printf("%f\n", foo(y));
> =A0return 0;
> }
>
>
> $ gcc =A0-ffast-math -O0 =A0 t.c
> /tmp/cccA1sUB.o: In function `foo':
> t.c:(.text+0x2c): undefined reference to `powf'
> collect2: error: ld returned 1 exit status
> $ gcc =A0-ffast-math -O2 =A0 t.c
> $ ./a.out 5
> 1220852096.000000
>
>
> FE with -ffast-math replaced x*x*...*x with __builtin_powf. Later with
> -O2 this call is replaced back into multiplications in sincos phase.
> The stability with -O0 is because sincos phase doesn't work on -O0.
>
> I think we must avoid doing this optimization in FE and turn off
> -ffast-math if -O0 is used. Your opinion.

No, I think we should avoid most of the builtin related folding at -O0.

Can you open a bugreport on gcc.gnu.org/bugzilla please?

Thanks,
Richard.

> Thanks,
> Vladimir
>