On Fri, Jul 25, 2025 at 10:53:38AM +0800, Yang Yujie wrote:
> A widening cast from a signed _BitInt operand to an unsigned _BitInt
> type involves filling the extra limb(s) with sign extension.
> On a target that wants _BitInts extended in memory, if this unsigned
> type has a partial limb, the unused part of it should be zeroed.
> 
> e.g. Assuming limb_mode == E_DImode, at the end of
> 
>   void
>   test (unsigned _BitInt(519) *t, _BitInt(512) x)
>     {
>       *t = -x;
>     }

Are you going to add this as a testcase with the BEXTC framework?
> 
> the most significant limb of *t should be masked with 0x7f.
> 
> This patch also fixes gcc.dg/torture/bitint-16.c, which aborts at -O2
> when the extension on load is optimized away.
> 
> gcc/ChangeLog:
> 
>       * gimple-lower-bitint.cc (bitint_large_huge::lower_mergeable_stmt):
>       zero-extend the partial limb of any unsigned _BitInt LHS assigned
>       with a widening sign-extension.
> ---
>  gcc/gimple-lower-bitint.cc | 8 ++++++++
>  1 file changed, 8 insertions(+)
> 
> diff --git a/gcc/gimple-lower-bitint.cc b/gcc/gimple-lower-bitint.cc
> index 631a7844790..d80191c6761 100644
> --- a/gcc/gimple-lower-bitint.cc
> +++ b/gcc/gimple-lower-bitint.cc
> @@ -3161,6 +3161,14 @@ bitint_large_huge::lower_mergeable_stmt (gimple *stmt, 
> tree_code &cmp_code,
>           {
>             tree l = limb_access (nlhs ? NULL_TREE : lhs_type,
>                                   nlhs ? nlhs : lhs, idx, true);
> +
> +           if (bitint_extended && sext && TYPE_UNSIGNED (lhs_type)
> +               && tree_fits_uhwi_p (idx) && !nlhs)

If condition doesn't fit on a single line, then every toplevel && or ||
should be on a separate line.  So
              if (bitint_extended
                  && sext
                  && TYPE_UNSIGNED (lhs_type)
                  && tree_fits_uhwi_p (idx)
                  && !nlhs)
> +             {
> +               rhs1 = add_cast (limb_access_type (lhs_type, idx), rhs1);
> +               rhs1 = add_cast (TREE_TYPE (l), rhs1);
> +             }
> +
>             g = gimple_build_assign (l, rhs1);
>           }
>         insert_before (g);
> -- 
> 2.46.0

        Jakub

Reply via email to