[patch, 4.3] Fix PR middle-end/38616

[H.J. Lu]

On Sun, Jan 11, 2009 at 7:18 AM, Matthias Klose <doko@cs.tu-berlin.de> wrote:
> The backport was suggested in the bug report; checked on
> i486-linux-gnu with a biarch compiler without regressions.
>
> Ok for the branch (and the testcase for the trunk)?
>
>  Matthias
>
> gcc/
>
> 2009-01-11  Matthias Klose  <doko@ubuntu.com>
>
>        PR middle-end/38616, backport from mainline:
>        2008-05-04  Uros Bizjak  <ubizjak@gmail.com>
>
>        * config/i386/i386.md (*strmovsi_1): Simplify asm alternatives.
>        (*strmovsi_rex_1): Ditto.
>        (*strsetsi_1): Ditto.
>        (*strsetsi_rex_1): Ditto.
>
>        (add<mode>cc): Macroize expander from addqicc, addhicc, addsicc and
>        adddicc expanders using SWI mode iterator.
>
> gcc/testsuite/
>
> 2009-01-11  Matthias Klose  <doko@ubuntu.com>
>
>        PR middle-end/38616
>        * gcc.dg/pr38616.c: New test.
>
>
> Index: gcc/testsuite/gcc.dg/pr38616.c
> ===================================================================
> --- gcc/testsuite/gcc.dg/pr38616.c      (revision 0)
> +++ gcc/testsuite/gcc.dg/pr38616.c      (revision 0)
> @@ -0,0 +1,13 @@
> +/* PR middle-end/38616 */
> +/* { dg-do run } */
> +/* { dg-options "-O2 -fstack-protector" } */
> +
> +#include <stdio.h>
> +
> +#define BUFFER "1234567890abcdefghijklmno"
> +int main (void)
> +{
> +  char buffer[1024]="";
> +  sprintf (buffer, "%s", BUFFER);
> +  return strcmp (buffer, BUFFER);
> +}

This test failed on Linux/ia64. Is there a particular reason why
strcmp is used without prototype?


-- 
H.J.