On Wed, Mar 9, 2022 at 12:20 PM Xi Ruoyao via Gcc-patches
<gcc-patches@gcc.gnu.org> wrote:
>
> Bootstrapped and tested on mips64el-linux-gnuabi64, and MIPS is the only
> port with a non-zero targetm.const_anchor. Ok for trunk?
>
> -- >8 --
>
> With a non-zero const_anchor, the behavior of this function relied on
> signed overflow.
>
> gcc/
>
> PR rtl-optimization/104843
> * cse.cc (compute_const_anchors): Cast to unsigned HOST_WIDE_INT
> to perform overflow arithmetics safely.
> ---
> gcc/cse.cc | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/gcc/cse.cc b/gcc/cse.cc
> index a18b599d324..7c39a009449 100644
> --- a/gcc/cse.cc
> +++ b/gcc/cse.cc
> @@ -1169,10 +1169,10 @@ compute_const_anchors (rtx cst,
> HOST_WIDE_INT *lower_base, HOST_WIDE_INT *lower_offs,
> HOST_WIDE_INT *upper_base, HOST_WIDE_INT *upper_offs)
> {
> - HOST_WIDE_INT n = INTVAL (cst);
> + unsigned HOST_WIDE_INT n = INTVAL (cst);
UINTVAL (cst)?
>
> *lower_base = n & ~(targetm.const_anchor - 1);
isn't it better to make targetm.const_anchor unsigned?
The & and ~ are not subject to overflow rules.
> - if (*lower_base == n)
> + if (*lower_base == INTVAL (cst))
duplicating this here is definitely ugly.
> return false;
>
> *upper_base =
> --
> 2.35.1
>
>
