[Bug c++/62051] New: [4.9/4.10] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[Bug c++/62051] New: [4.9/4.10] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[raphael.kubo.da.costa at intel dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

            Bug ID: 62051
           Summary: [4.9/4.10] Undefined reference to vtable with -O2 and
                    -fdevirtualize-speculatively
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: raphael.kubo.da.costa at intel dot com

classes.h:
#ifndef CLASSES_H
#define CLASSES_H
struct Base {
  virtual ~Base() {}
  virtual void virt_func() = 0;
};
struct Derived : public Base {
 public:
  __attribute__((visibility("default"))) Derived();
  virtual ~Derived() { func(); }
  virtual void virt_func();
  __attribute__((visibility("default"))) void func();
};
#endif

classes.cc:
#include "classes.h"
Derived::Derived() {}
void Derived::virt_func() {}
void Derived::func() {}

main.cc:
#include "classes.h"
void f() {
  Derived* m = new Derived;
  delete m;
}
int main() {
  f();
  return 0;
}

Building it with:
  g++ -fvisibility=hidden -fPIC -shared -o libclasses.so classes.cc
  g++ -fPIC -O2 main.cc libclasses.so

Everything works fine with GCC 4.8, but both 4.9 and 4.10-snapshot fail like
this, unless I either use a lower optimization level or pass
-fno-devirtualize-speculatively:
/tmp/ccXRd3i4.o: In function `f()':
main.cc:(.text+0x2e): undefined reference to `vtable for Derived'
/tmp/ccXRd3i4.o: In function `Derived::~Derived()':
main.cc:(.text._ZN7DerivedD0Ev[_ZN7DerivedD5Ev]+0x3): undefined reference to
`vtable for Derived'
/tmp/ccXRd3i4.o: In function `Derived::~Derived()':
main.cc:(.text._ZN7DerivedD2Ev[_ZN7DerivedD5Ev]+0x7): undefined reference to
`vtable for Derived'

GCC 4.8:
Using built-in specs.
COLLECT_GCC=g++-4.8
COLLECT_LTO_WRAPPER=/usr/lib/gcc/x86_64-linux-gnu/4.8/lto-wrapper
Target: x86_64-linux-gnu
Configured with: ../src/configure -v --with-pkgversion='Debian 4.8.3-7'
--with-bugurl=file:///usr/share/doc/gcc-4.8/README.Bugs
--enable-languages=c,c++,java,go,d,fortran,objc,obj-c++ --prefix=/usr
--program-suffix=-4.8 --enable-shared --enable-linker-build-id
--libexecdir=/usr/lib --without-included-gettext --enable-threads=posix
--with-gxx-include-dir=/usr/include/c++/4.8 --libdir=/usr/lib --enable-nls
--with-sysroot=/ --enable-clocale=gnu --enable-libstdcxx-debug
--enable-libstdcxx-time=yes --enable-gnu-unique-object --disable-libmudflap
--enable-plugin --with-system-zlib --disable-browser-plugin
--enable-java-awt=gtk --enable-gtk-cairo
--with-java-home=/usr/lib/jvm/java-1.5.0-gcj-4.8-amd64/jre --enable-java-home
--with-jvm-root-dir=/usr/lib/jvm/java-1.5.0-gcj-4.8-amd64
--with-jvm-jar-dir=/usr/lib/jvm-exports/java-1.5.0-gcj-4.8-amd64
--with-arch-directory=amd64 --with-ecj-jar=/usr/share/java/eclipse-ecj.jar
--enable-objc-gc --enable-multiarch --with-arch-32=i586 --with-abi=m64
--with-multilib-list=m32,m64,mx32 --with-tune=generic --enable-checking=release
--build=x86_64-linux-gnu --host=x86_64-linux-gnu --target=x86_64-linux-gnu
Thread model: posix
gcc version 4.8.3 (Debian 4.8.3-7)

GCC 4.9:
Using built-in specs.
COLLECT_GCC=/usr/bin/g++
COLLECT_LTO_WRAPPER=/usr/lib/gcc/x86_64-linux-gnu/4.9/lto-wrapper
Target: x86_64-linux-gnu
Configured with: ../src/configure -v --with-pkgversion='Debian 4.9.1-4'
--with-bugurl=file:///usr/share/doc/gcc-4.9/README.Bugs
--enable-languages=c,c++,java,go,d,fortran,objc,obj-c++ --prefix=/usr
--program-suffix=-4.9 --enable-shared --enable-linker-build-id
--libexecdir=/usr/lib --without-included-gettext --enable-threads=posix
--with-gxx-include-dir=/usr/include/c++/4.9 --libdir=/usr/lib --enable-nls
--with-sysroot=/ --enable-clocale=gnu --enable-libstdcxx-debug
--enable-libstdcxx-time=yes --enable-gnu-unique-object --disable-vtable-verify
--enable-plugin --with-system-zlib --disable-browser-plugin
--enable-java-awt=gtk --enable-gtk-cairo
--with-java-home=/usr/lib/jvm/java-1.5.0-gcj-4.9-amd64/jre --enable-java-home
--with-jvm-root-dir=/usr/lib/jvm/java-1.5.0-gcj-4.9-amd64
--with-jvm-jar-dir=/usr/lib/jvm-exports/java-1.5.0-gcj-4.9-amd64
--with-arch-directory=amd64 --with-ecj-jar=/usr/share/java/eclipse-ecj.jar
--enable-objc-gc --enable-multiarch --with-arch-32=i586 --with-abi=m64
--with-multilib-list=m32,m64,mx32 --enable-multilib --with-tune=generic
--enable-checking=release --build=x86_64-linux-gnu --host=x86_64-linux-gnu
--target=x86_64-linux-gnu
Thread model: posix
gcc version 4.9.1 (Debian 4.9.1-4)

GCC 4.10:
Using built-in specs.
COLLECT_GCC=/usr/lib/gcc-snapshot/bin/g++
COLLECT_LTO_WRAPPER=/usr/lib/gcc-snapshot/libexec/gcc/x86_64-linux-gnu/4.10.0/lto-wrapper
Target: x86_64-linux-gnu
Configured with: ../src/configure -v --with-pkgversion='Debian 20140802-1'
--with-bugurl=file:///usr/share/doc/gcc-snapshot/README.Bugs
--enable-languages=c,ada,c++,java,go,fortran,objc,obj-c++
--prefix=/usr/lib/gcc-snapshot --enable-shared --enable-linker-build-id
--disable-nls --with-sysroot=/ --enable-clocale=gnu --enable-libstdcxx-debug
--enable-libstdcxx-time=yes --enable-gnu-unique-object --disable-vtable-verify
--enable-plugin --with-system-zlib --disable-browser-plugin
--enable-java-awt=gtk --enable-gtk-cairo
--with-java-home=/usr/lib/jvm/java-1.5.0-gcj-4.10-snap-amd64/jre
--enable-java-home
--with-jvm-root-dir=/usr/lib/jvm/java-1.5.0-gcj-4.10-snap-amd64
--with-jvm-jar-dir=/usr/lib/jvm-exports/java-1.5.0-gcj-4.10-snap-amd64
--with-arch-directory=amd64 --with-ecj-jar=/usr/share/java/eclipse-ecj.jar
--enable-objc-gc --enable-multiarch --with-arch-32=i586 --with-abi=m64
--with-multilib-list=m32,m64,mx32 --enable-multilib --with-tune=generic
--disable-werror --enable-checking=yes --build=x86_64-linux-gnu
--host=x86_64-linux-gnu --target=x86_64-linux-gnu
Thread model: posix
gcc version 4.10.0 20140802 (experimental) [trunk revision 213518] (Debian
20140802-1)

[Bug c++/62051] [4.9/4.10] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.9.3

[Bug ipa/62051] [4.9/5 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |wrong-code
           Priority|P3                          |P1
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2015-03-19
          Component|c++                         |ipa
            Version|unknown                     |5.0
     Ever confirmed|0                           |1

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
Thus confirmed.

[Bug ipa/62051] [4.9/5 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[trippels at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

Markus Trippelsdorf <trippels at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |trippels at gcc dot gnu.org

--- Comment #3 from Markus Trippelsdorf <trippels at gcc dot gnu.org> ---
If you define "void Derived::func()" directly in the header if works
as expected. 
And since ~Derived() calls this function you must provide it in every
compilation unit.
So the devirtualization looks correct to me.

[Bug ipa/62051] [4.9/5 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[hubicka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

--- Comment #5 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
yep, this seems to be the usual situation where visibilities does not mix that
well with assumptions that program is linked with symbols defined in the
headers.

I suppose we could make C++ FE to track if a type has methods with visibility
default declared (I can't track that from symbol table as they may be
unreachable). In such case we may want to consider methods of that type with
non-default visibility and no resolution info to be
!can_refer_decl_in_current_unit_p

Jason, would that make sense to you?

[Bug ipa/62051] [4.9/5 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P1                          |P2
                 CC|                            |jakub at gcc dot gnu.org

[Bug ipa/62051] [4.9/5 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[jason at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

--- Comment #6 from Jason Merrill <jason at gcc dot gnu.org> ---
(In reply to Jan Hubicka from comment #5)
> I suppose we could make C++ FE to track if a type has methods with
> visibility default declared (I can't track that from symbol table as they
> may be unreachable). In such case we may want to consider methods of that
> type with non-default visibility and no resolution info to be
> !can_refer_decl_in_current_unit_p
> 
> Jason, would that make sense to you?

Yes, though I think for such a class we probably want to consider all virtual
methods unreachable unless they have explicit default visibility; in the
testcase the main program isn't being compiled with -fvisibility=hidden, so
~Derived has implicit default visibility.

[Bug ipa/62051] [4.9/5 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[hubicka at ucw dot cz]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

--- Comment #7 from Jan Hubicka <hubicka at ucw dot cz> ---
> 
> Yes, though I think for such a class we probably want to consider all virtual
> methods unreachable unless they have explicit default visibility; in the
> testcase the main program isn't being compiled with -fvisibility=hidden, so
> ~Derived has implicit default visibility.

Yes, I think this makes sense.  Do you think you can implement the C++ FE part
that will mark classes having methods with visibility specified?
We do not LTO stream TYPE_METHODS so this is bit hard to determine at
gimple-fold
time.  At some point I tried to enable TYPE_METHODS streaming but it had bad
effect on firefox LTO linktimes by significantly increasing strongly connected
component size.

[Bug ipa/62051] [4.9/5/6 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
GCC 4.9.3 has been released.

[Bug ipa/62051] [4.9/5/6 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.9.3                       |4.9.4

[Bug ipa/62051] [4.9/5/6 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[jason at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

--- Comment #9 from Jason Merrill <jason at gcc dot gnu.org> ---
Created attachment 36480
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=36480&action=edit
visibility flag patch

Is this the sort of thing you had in mind?

[Bug ipa/62051] [8/9/10 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[hubicka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

Jan Hubicka <hubicka at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|8.5                         |11.0

--- Comment #23 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
This is bit of a grey area of what we can/can not refer in presence of
visibilities and I hope codebases are now adopted for GCC behaviour.  I think
we could delay this post GCC10, so re-taretting.

[Bug ipa/62051] [8/9/10/11/12 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|11.0                        |11.2

--- Comment #24 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
GCC 11.1 has been released, retargeting bugs to GCC 11.2.

[Bug ipa/62051] [9/10/11/12 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|11.2                        |11.3

--- Comment #25 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 11.2 is being released, retargeting bugs to GCC 11.3

[Bug ipa/62051] [9/10/11/12 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

--- Comment #26 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Is this even valid?

That is Base and Derived in the shared library and the main program are
considered two different classes because of -fvisibility=hidden and the classes
are not marked as visibility default.

[Bug ipa/62051] [9/10/11/12 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[jason at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

--- Comment #27 from Jason Merrill <jason at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #26)
> That is Base and Derived in the shared library and the main program are
> considered two different classes because of -fvisibility=hidden and the
> classes are not marked as visibility default.

Symbol visibility is outside the language, and doesn't affect mangling, so I
don't think it's accurate to say they are considered two different classes.  It
seems to me they are the same classes, with controlled access to the
implementation.

[Bug ipa/62051] [9/10/11/12 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|11.3                        |11.4

--- Comment #28 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 11.3 is being released, retargeting bugs to GCC 11.4.

[Bug ipa/62051] [10/11/12/13 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|                            |13.0
   Target Milestone|11.4                        |13.0
   Last reconfirmed|2015-03-19 00:00:00         |2022-12-20

--- Comment #29 from Richard Biener <rguenth at gcc dot gnu.org> ---
Re-confirmed on trunk.

[Bug ipa/62051] [10/11/12/13/14 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|13.0                        |13.2

--- Comment #30 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 13.1 is being released, retargeting bugs to GCC 13.2.

[Bug ipa/62051] [11/12/13/14 Regression] Undefined reference to vtable with -O2 and -fdevirtualize-speculatively

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62051

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|13.2                        |11.5