From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-bugzilla@gcc.gnu.org>
Received: by sourceware.org (Postfix, from userid 48)
	id 2C6183858CDA; Fri, 17 Mar 2023 23:02:10 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 2C6183858CDA
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org;
	s=default; t=1679094130;
	bh=Uy3j/b1anjXU4j2lSV4Lf5VXKWqXb9s80vc+jYS3EKg=;
	h=From:To:Subject:Date:From;
	b=PRuudeVVucXQBTFvOKn9xiLqXu17h4GD/lnUav4G+XdhzY+6qlFNxLpXqPZXp0sPU
	 40wn8MfK5Q1MHBdly1ViRUVTf0uPkCnOkYvxe1cl/fbZN5o9nPqhrAKHI1xDglwh3S
	 NRBp3f6jjDmszwHTG1tT3Zs3hFJ225QbkSXGCZw4=
From: "idhameed at protonmail dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug c++/109180] New: Explicit template instantiation doesn't work
 for constexpr destructors on arm32 when optimizations are enabled
Date: Fri, 17 Mar 2023 23:02:09 +0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: new
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: c++
X-Bugzilla-Version: 12.2.0
X-Bugzilla-Keywords: 
X-Bugzilla-Severity: normal
X-Bugzilla-Who: idhameed at protonmail dot com
X-Bugzilla-Status: UNCONFIRMED
X-Bugzilla-Resolution: 
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: bug_id short_desc product version bug_status
 bug_severity priority component assigned_to reporter target_milestone
Message-ID: <bug-109180-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
List-Id: <gcc-bugs.sourceware.org>

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D109180

            Bug ID: 109180
           Summary: Explicit template instantiation doesn't work for
                    constexpr destructors on arm32 when optimizations are
                    enabled
           Product: gcc
           Version: 12.2.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: idhameed at protonmail dot com
  Target Milestone: ---

Example (compile with "g++ -std=3Dc++20 -O -c"):

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D

template <typename t>
struct record {
    constexpr ~record() { }
};

inline record<char>
return_record() { return {}; }

template record<char>::~record();

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D

Compiler Explorer: https://godbolt.org/z/j37rvr961

GCC 10+ targeting 32-bit ARM won't emit any symbols for record<char>::~reco=
rd
when using most -O levels (-O, -O2, -O3, -Ofast, -Og). Using -fno-inline--or
placing the explicit instantiation of record<char>::~record above the
definition of return_record--works around this bug and causes a symbol for
record<char>::~record to be emitted, as I'd expect. x86-64 GCC doesn't have
this problem.

The original, non-minimized problem I encountered: GCC 12.2.0 as a
cross-compiler on an x86-64 host targeting arm-linux-gnueabihf can't be use=
d to
compile and link LLVM 15's libc++, because libc++'s implementation of
std::basic_string has code that fits the aformentioned pattern--there's an
overload of `std::__1::literals::string_literals::operator "" s` that retur=
ns a
basic_string<char>. I tried the release/15.x branch of llvm-project at comm=
it
92e7ef99303f8f367f279ccfa2393e4b96db915a.=