In the original fix for this PR, I'd made sure that including <arm_sme.h> didn't reach the final return in simulate_builtin_function_decl (which would indicate duplicate function definitions). But it seems I forgot to do the same thing for C++, which defines all of its overloads directly.
This patch fixes a case where we still recorded duplicate functions for C++. Thanks to Iain for reporting the resulting GC ICE and for help with reproducing it. Tested on aarch64-linux-gnu & pushed. Richard gcc/ PR target/112989 * config/aarch64/aarch64-sve-builtins-shapes.cc (build_one): Skip MODE_single variants of functions that don't take tuple arguments. --- gcc/config/aarch64/aarch64-sve-builtins-shapes.cc | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/gcc/config/aarch64/aarch64-sve-builtins-shapes.cc b/gcc/config/aarch64/aarch64-sve-builtins-shapes.cc index 2692b086ac5..f190770250f 100644 --- a/gcc/config/aarch64/aarch64-sve-builtins-shapes.cc +++ b/gcc/config/aarch64/aarch64-sve-builtins-shapes.cc @@ -338,6 +338,14 @@ build_one (function_builder &b, const char *signature, unsigned int ti, unsigned int gi, unsigned int pi, bool force_direct_overloads) { + /* For simplicity, function definitions are allowed to use the group + suffix lists vg2 and vg4 for shapes that have _single forms, + even though the _single form applies only to vgNx2 and vgNx4, + not to vgNx1. */ + if (mode_suffix_id == MODE_single + && group_suffixes[group.groups[gi]].vectors_per_tuple == 1) + return; + /* Byte forms of svdupq take 16 arguments. */ auto_vec<tree, 16> argument_types; function_instance instance (group.base_name, *group.base, *group.shape, -- 2.25.1