On Mon, 15 Dec 2014, Marek Polacek wrote:
> m68k revealed another missing check before TYPE_OVERFLOW_WRAPS.
> I think we should use INTEGRAL_TYPE_P, and not ANY_INTEGRAL_TYPE_P
> because the switch handles VECTOR_CST and COMPLEX_CST.
>
> Bootstrapped/regtested on x86_64-linux and ppc64-linux, ok for trunk?
Ok.
Thanks,
Richard.
> 2014-12-15 Marek Polacek <pola...@redhat.com>
>
> PR middle-end/64292
> * fold-const.c (negate_expr_p): Add INTEGRAL_TYPE_P check.
>
> diff --git gcc/fold-const.c gcc/fold-const.c
> index d71fa94..07da71a 100644
> --- gcc/fold-const.c
> +++ gcc/fold-const.c
> @@ -400,7 +400,7 @@ negate_expr_p (tree t)
> switch (TREE_CODE (t))
> {
> case INTEGER_CST:
> - if (TYPE_OVERFLOW_WRAPS (type))
> + if (INTEGRAL_TYPE_P (type) && TYPE_OVERFLOW_WRAPS (type))
> return true;
>
> /* Check that -CST will not overflow type. */
>
> Marek
>
>
--
Richard Biener <rguent...@suse.de>
SUSE LINUX GmbH, GF: Felix Imendoerffer, Jane Smithard, Jennifer Guild,
Dilip Upmanyu, Graham Norton HRB 21284 (AG Nuernberg)
