https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93217
Martin Liška <marxin at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |redi at gcc dot gnu.org
--- Comment #3 from Martin Liška <marxin at gcc dot gnu.org> ---
Ok, I screw up parenthesis in IPA inliner. I've got a patch for that.
But still, the test-case should not be sensitive to an inlining decision. The
problematic version does some inlining of:
Inlining _Tp std::__atomic_impl::__fetch_add_flt(_Tp*,
std::__atomic_impl::_Val<_Tp>, std::memory_order) [with _Tp = long double]/404
size 27.
Called once from void test03()/285 887 insns.
Inlined into void test03()/285 which now has 904 size
Inlining _Tp std::__atomic_impl::__fetch_add_flt(_Tp*,
std::__atomic_impl::_Val<_Tp>, std::memory_order) [with _Tp = long double]/404
size 27.
Called once from void test03()/285 904 insns.
Inlined into void test03()/285 which now has 921 size
Inlining _Tp std::__atomic_impl::__sub_fetch_flt(_Tp*,
std::__atomic_impl::_Val<_Tp>) [with _Tp = double]/393 size 21.
Called once from void test02()/281 733 insns.
Inlined into void test02()/281 which now has 747 size
Inlining _Tp std::__atomic_impl::__add_fetch_flt(_Tp*,
std::__atomic_impl::_Val<_Tp>) [with _Tp = double]/392 size 21.
Called once from void test02()/281 747 insns.
Inlined into void test02()/281 which now has 761 size
Can you please Jonathan take a look what happens?
