> On 22 Jul 2026, at 08:50, Philipp Tomsich <[email protected]> wrote:
>
> noce_emit_store_flag feeds the possibly-reversed comparison code to a
> store-flag insn, and to emit_store_flag, without checking that the
> reversal succeeded. reversed_comparison_code returns UNKNOWN for
> UNLT/UNLE/UNGT/UNGE, and for a MODE_CC comparison it cannot trace back
> to its COMPARE -- common for a floating-point condition on targets that
> compare into a condition-code register. emit_store_flag then reaches
> its floating-point path and calls swap_condition (UNKNOWN), which aborts.
>
> This was latent until r17-2519-ga33f26607eb4f3 (noce_try_shifted_store_flag),
> the first caller to reach noce_emit_store_flag with REVERSEP set for such
> a condition.
>
> Bail out early when the reversal failed.
>
> Bootstrapped and regression tested on aarch64-unknown-linux-gnu with no
> regressions.
>
> PR rtl-optimization/126347
> gcc/
> * ifcvt.cc (noce_emit_store_flag): Return NULL_RTX when the
> comparison code is UNKNOWN.
LGTM (though I’m not a maintainer) but could you add a creduced/cvised test
case from the ICE in the PR?
Thanks,
Kyrill
> ---
> gcc/ifcvt.cc | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/gcc/ifcvt.cc b/gcc/ifcvt.cc
> index 7c50ab06e2f7..6b169e69741d 100644
> --- a/gcc/ifcvt.cc
> +++ b/gcc/ifcvt.cc
> @@ -865,6 +865,11 @@ noce_emit_store_flag (struct noce_if_info *if_info, rtx
> x, bool reversep,
> else
> code = GET_CODE (cond);
>
> + /* reversed_comparison_code returns UNKNOWN for an unordered code, or a
> + CC-mode compare it cannot trace; neither path below can use that. */
> + if (code == UNKNOWN)
> + return NULL_RTX;
> +
> if ((if_info->cond_earliest == if_info->jump || cond_complex)
> && (normalize == 0 || STORE_FLAG_VALUE == normalize))
> {
> --
> 2.34.1
>
