[Bug c++/94184] New: Global variable inline constructor elision.

[Bug c++/94184] New: Global variable inline constructor elision.

[maxim.yegorushkin at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94184

            Bug ID: 94184
           Summary: Global variable inline constructor elision.
           Product: gcc
           Version: 9.2.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: maxim.yegorushkin at gmail dot com
  Target Milestone: ---

The following code:

    struct A { 
        int a_; 
        A(int a) : a_(a) {}
    };

    struct B { 
        int a_; 
        constexpr B(int a) : a_(a) {}
    };

    A a{1};
    B b{1};

When compiled with `gcc-9.2 -O3 -march=skylake -mtune=skylake -std=gnu++17`
generates a call to `A` constructor for variable `a`:

    _GLOBAL__sub_I_a:
        mov    DWORD PTR [rip+0x200bd6],0x1        # 601030 <__TMC_END__>
        ret    

And no call for `B` constructor for variable `b`, as expected of `constexpr`.

`clang-9.0 -O3 -march=skylake -mtune=skylake -std=gnu++17`, however, elides the
constructor calls for both `a` and `b` global variables, regardless of
`constexpr`.

Why can't `gcc` elide the call to `A` constructor for `a`, since `A` constuctor
is an inline function with a clear side-effect?

https://gcc.godbolt.org/z/5y3cD5

[Bug c++/94184] Global variable inline constructor elision.

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94184

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement
           Keywords|                            |missed-optimization

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
There is another bug about this already.

[Bug c++/94184] Global variable inline constructor elision.

[trass3r at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94184

Trass3r <trass3r at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |trass3r at gmail dot com

--- Comment #2 from Trass3r <trass3r at gmail dot com> ---
Even simpler example:

const struct A
{
    A() {}
    int a = 0;
} a;

The global ctor doesn't get eliminated while clang removes it:
https://godbolt.org/z/9bjzMbx85

        .section        .text.startup,"ax",@progbits
        .type   _GLOBAL__sub_I_example.cpp, @function
_GLOBAL__sub_I_example.cpp:
        ret

It works if you remove the member a.

[Bug c++/94184] Global variable inline constructor elision.

[ppalka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94184

Patrick Palka <ppalka at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |DUPLICATE
             Status|UNCONFIRMED                 |RESOLVED
                 CC|                            |ppalka at gcc dot gnu.org

--- Comment #3 from Patrick Palka <ppalka at gcc dot gnu.org> ---
dup of PR102876 I think

*** This bug has been marked as a duplicate of bug 102876 ***