On Sat, Jul 26, 2025 at 3:13 PM <pan2...@intel.com> wrote:
>
> From: Pan Li <pan2...@intel.com>
>
> Like widen_mul based pattern, we would like introduce the mul based
> pattern as well. The pattern is quite simple compares to the
> widen_mul, thus add new instead of the for loop in match.pd.
>
> gcc/ChangeLog:
>
> * match.pd: Add mul based unsigned SAT_MUL.
>
> Signed-off-by: Pan Li <pan2...@intel.com>
> ---
> gcc/match.pd | 12 ++++++++++++
> 1 file changed, 12 insertions(+)
>
> diff --git a/gcc/match.pd b/gcc/match.pd
> index 4903552c82a..f8d74fd52ad 100644
> --- a/gcc/match.pd
> +++ b/gcc/match.pd
> @@ -3611,6 +3611,18 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
> bool widen_mult_p = cvt5_prec == cvt6_prec && widen_prec == cvt6_prec
> * 2;
> }
> (if (widen_prec > prec && c2_is_max_p && widen_mult_p)))))
> + (match (unsigned_integer_sat_mul @0 @1)
> + (convert@4 (min (mult:c@3 (convert @0) (convert @1)) INTEGER_CST@2))
> + (if (types_match (type, @0, @1) && types_match (type, @4))
the types_match (type, @4) looks redundant. Do you want to
restrict this outer conversion to a sign-conversion? In that
case use (nop_convert ...)
> + (with
> + {
> + unsigned prec = TYPE_PRECISION (type);
> + unsigned widen_prec = TYPE_PRECISION (TREE_TYPE (@3));
> + wide_int c2 = wi::to_wide (@2);
> + wide_int max = wi::mask (prec, false, widen_prec);
> + bool c2_is_max_p = wi::eq_p (c2, max);
> + }
> + (if (widen_prec > prec && c2_is_max_p)))))
> )
>
> /* The boundary condition for case 10: IMM = 1:
> --
> 2.43.0
>
