On Wed, Mar 28, 2012 at 4:00 PM, Jakub Jelinek <ja...@redhat.com> wrote:
> Hi!
>
> The builtin_decl_explicit_p changes broke the __builtin_va_start
> -> __builtin_next_arg folding, previously it was testing that
> built_in_decls[BUILT_IN_NEXT_ARG] == NULL (and giving up in that case),
> so the conversion missed !.
>
> Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux,
> ok for trunk/4.7?
Ok.
Thanks,
Richard.
> 2012-03-28 Jakub Jelinek <ja...@redhat.com>
>
> PR middle-end/52691
> * tree-ssa-ccp.c (optimize_stdarg_builtin): Optimize
> __builtin_va_start to __builtin_next_arg if the latter is
> builtin_decl_explicit_p rather than when it is not.
>
> * gcc.dg/pr52691.c: New test.
>
> --- gcc/tree-ssa-ccp.c.jj 2012-01-13 21:47:35.000000000 +0100
> +++ gcc/tree-ssa-ccp.c 2012-03-28 12:48:44.358615418 +0200
> @@ -2288,7 +2288,7 @@ optimize_stdarg_builtin (gimple call)
> case BUILT_IN_VA_START:
> if (!va_list_simple_ptr
> || targetm.expand_builtin_va_start != NULL
> - || builtin_decl_explicit_p (BUILT_IN_NEXT_ARG))
> + || !builtin_decl_explicit_p (BUILT_IN_NEXT_ARG))
> return NULL_TREE;
>
> if (gimple_call_num_args (call) != 2)
> --- gcc/testsuite/gcc.dg/pr52691.c.jj 2012-03-28 12:57:34.555364325 +0200
> +++ gcc/testsuite/gcc.dg/pr52691.c 2012-03-28 13:11:16.020328633 +0200
> @@ -0,0 +1,24 @@
> +/* PR middle-end/52691 */
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -fdump-tree-optimized" } */
> +
> +#include <stdarg.h>
> +
> +int
> +foo (int a, ...)
> +{
> + int b = 0, c = 0;
> + va_list ap;
> + va_start (ap, a);
> + if (a > 1)
> + b = va_arg (ap, double);
> + if (a > 2)
> + c = va_arg (ap, long long);
> + va_end (ap);
> + return a + b + c;
> +}
> +
> +/* { dg-final { scan-tree-dump "__builtin_next_arg" "optimized" { target { {
> i?86-*-* x86_64-*-* } && ia32 } } } } */
> +/* { dg-final { scan-tree-dump "__builtin_next_arg" "optimized" { target {
> powerpc*-*-darwin* powerpc*-*-aix* } } } } */
> +/* { dg-final { scan-tree-dump "__builtin_next_arg" "optimized" { target {
> powerpc*-*-linux* && lp64 } } } } */
> +/* { dg-final { cleanup-tree-dump "optimized" } } */
>
> Jakub
