[Bug libstdc++/66262] New: [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[Bug libstdc++/66262] New: [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[costamagnagianfranco at yahoo dot it]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

            Bug ID: 66262
           Summary: [REGRESSION] testsuite failure with libstdc++ (gcc-5)
                    and g++-4.9 together
           Product: gcc
           Version: 5.1.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: libstdc++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: costamagnagianfranco at yahoo dot it
  Target Milestone: ---

Hi, sorry in advance for the low quality bug report.

I'm reporting this upstream, since I suspect the ABI broke in some bad way
between 4.9 and 5.0.

note: I tried to explain as much as possible on Debian bug [1]

Firstly I wasn't able to make it work with g++-5 and libstdc++ from gcc-5
package

After the Debian Maintainer rebuilt gcc-5 with 

   * Configure with
       --disable-libstdcxx-dual-abi --with-default-libstdcxx-abi=c++98
     While libstdc++ provides a dual ABI to support both the c++98 and c++11
     ABI, there is no committment on compatibility of the old experimental
     c++11 ABI from GCC 4.9 and the stable c++11 ABI in GCC 5.


and now I can build them with both g++-5 and libstdc++ from gcc-5.

the problem actually is that now Debian has libstdc++ from gcc-5, and the
default compiler still set to gcc-4.9.

With that combination (g++-4.9 and libstdc++ from gcc-5) the bug is there and I
can't upload in Debian the new release until I figure this out.

I can try to provide more informations, just feel free to ask me.

[1] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=784655

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[trippels at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING
   Last reconfirmed|                            |2015-05-23
                 CC|                            |trippels at gcc dot gnu.org
     Ever confirmed|0                           |1

--- Comment #1 from Markus Trippelsdorf <trippels at gcc dot gnu.org> ---
Please attach a testcase that fails with libstdc++ from gcc-5 
and succeeds with libstdc++ from gcc-4.9.

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[costamagnagianfranco at yahoo dot it]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

--- Comment #2 from Gianfranco <costamagnagianfranco at yahoo dot it> ---
I mentioned a testcase in the Debian bug report.
Unfortunately I'm not the upstream developer, I might find difficult to extract
what you exactly need...

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[trippels at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

--- Comment #3 from Markus Trippelsdorf <trippels at gcc dot gnu.org> ---
(In reply to Gianfranco from comment #2)
> I mentioned a testcase in the Debian bug report.
> Unfortunately I'm not the upstream developer, I might find difficult to
> extract what you exactly need...

Please see https://gcc.gnu.org/bugs/ for a list of what is needed.

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[trippels at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |RESOLVED
         Resolution|---                         |INVALID

--- Comment #4 from Markus Trippelsdorf <trippels at gcc dot gnu.org> ---
No feedback, no testcase. Closing.

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[costamagnagianfranco at yahoo dot it]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

Gianfranco <costamagnagianfranco at yahoo dot it> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |NEW
         Resolution|INVALID                     |---

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[costamagnagianfranco at yahoo dot it]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

--- Comment #5 from Gianfranco <costamagnagianfranco at yahoo dot it> ---
I'm still trying to provide what requested, as you can see there
I'm working right now on it
https://buildd.debian.org/status/package.php?p=casablanca&suite=experimental

however it is not trivial by me :)

please keep it open for a little longer!

thanks

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[costamagnagianfranco at yahoo dot it]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

--- Comment #6 from Gianfranco <costamagnagianfranco at yahoo dot it> ---
Created attachment 35662
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35662&action=edit
"Reduced" testcase

I tried in the last few days to reduce as much as possible the testcase.
I really hope it is enough for you to understand something more :)

to reproduce:
pbuilder-dist sid create
pbuilder-dist sid login
apt-get install vim cmake
apt-get install libboost-chrono-dev libboost-system-dev libboost-locale-dev
libboost-filesystem-dev 
apt-get install g++-5
cd casablanca/
mkdir build
cd build/
cmake .. && make test_runner streams_test -j 9 && cd Binaries/ && ./test_runner
libstreams_test.so
#FAILURE
cd ../..
mkdir build-gcc5
cd build-gcc5
export CXX=g++-5
cmake .. && make test_runner streams_test -j 9 && cd Binaries/ && ./test_runner
libstreams_test.so
#SUCCESS


the build needs unittestpp to create the unittest, and needs the "test_runner"
to open the library .so to run tests inside.

The bug is for sure in the libstreams_test.so, created from this source file
tests/functional/streams/fstreambuf_tests.cpp

I guess this because copying the .so built in gcc-5 directory in the gcc-4 one
makes the testsuite pass the tests.

I can provide the binaries if needed, unfortunately I don't know how to debug
the library inside, since it is called with dlopen.


g++ -v
Using built-in specs.
COLLECT_GCC=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 4.9.2-19'
--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 4.9.2 (Debian 4.9.2-19) 


g++-5 -v
Using built-in specs.
COLLECT_GCC=g++-5
COLLECT_LTO_WRAPPER=/usr/lib/gcc/x86_64-linux-gnu/5/lto-wrapper
Target: x86_64-linux-gnu
Configured with: ../src/configure -v --with-pkgversion='Debian 5.1.1-8'
--with-bugurl=file:///usr/share/doc/gcc-5/README.Bugs
--enable-languages=c,ada,c++,java,go,d,fortran,objc,obj-c++ --prefix=/usr
--program-suffix=-5 --enable-shared --enable-linker-build-id
--libexecdir=/usr/lib --without-included-gettext --enable-threads=posix
--libdir=/usr/lib --enable-nls --with-sysroot=/ --enable-clocale=gnu
--enable-libstdcxx-debug --enable-libstdcxx-time=yes
--with-default-libstdcxx-abi=c++98 --disable-libstdcxx-dual-abi
--enable-gnu-unique-object --disable-vtable-verify --enable-libmpx
--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-5-amd64/jre --enable-java-home
--with-jvm-root-dir=/usr/lib/jvm/java-1.5.0-gcj-5-amd64
--with-jvm-jar-dir=/usr/lib/jvm-exports/java-1.5.0-gcj-5-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 5.1.1 20150528 (Debian 5.1.1-8) 


copying libstdc++6 from an ubuntu machine makes the test successfully pass on
the gcc-4 build.
(don't know how much of this is helpful for you)

thanks a lot,

Gianfranco

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[trippels at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

--- Comment #7 from Markus Trippelsdorf <trippels at gcc dot gnu.org> ---
Thanks.

I get with gcc-4.9:
...
Linking CXX executable ../../../Binaries/test_runner
CMakeFiles/test_runner.dir/test_runner.cpp.o:test_runner.cpp:function
boost::filesystem::path::path<boost::filesystem::directory_entry>(boost::filesystem::directory_entry
const&,
boost::enable_if<boost::filesystem::path_traits::is_pathable<boost::decay<boost::filesystem::directory_entry>::type>,
void>::type*): error: undefined reference to
'boost::filesystem::path_traits::dispatch(boost::filesystem::directory_entry
const&, std::string&)'
collect2: error: ld returned 1 exit status
tests/common/TestRunner/CMakeFiles/test_runner.dir/build.make:113: recipe for
target 'Binaries/test_runner' failed

Note that I'm using the new C++ ABI already and I don't have a testbox
that uses the old ABI.

So someone that still uses the old ABI needs to debug this further...

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[costamagnagianfranco at yahoo dot it]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

--- Comment #8 from Gianfranco <costamagnagianfranco at yahoo dot it> ---
Thanks to you Markus, I spent almost one entire week in removing code in order
to have something "useful" for you, and this is the maximum I could achieve.

I don't know how much I can help you, your build failure is caused from the
fact that you didn't rebuilt your boost framework with the new ABI?

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[trippels at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

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

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

--- Comment #9 from Markus Trippelsdorf <trippels at gcc dot gnu.org> ---
(In reply to Gianfranco from comment #8)
> Thanks to you Markus, I spent almost one entire week in removing code in
> order to have something "useful" for you, and this is the maximum I could
> achieve.
> 
> I don't know how much I can help you, your build failure is caused from the
> fact that you didn't rebuilt your boost framework with the new ABI?

No. It is caused by building my whole system (including Boost) with the 
new ABI. std::string parameters are mangled different in this case.

So you'll have to wail until a developer, who still uses the old ABI,
picks up this issue.

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[doko at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

Matthias Klose <doko at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #10 from Matthias Klose <doko at gcc dot gnu.org> ---
duplicate of PR66145

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|FIXED                       |DUPLICATE

--- Comment #11 from H.J. Lu <hjl.tools at gmail dot com> ---
Dup.

*** This bug has been marked as a duplicate of bug 66145 ***

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

--- Comment #12 from Jonathan Wakely <redi at gcc dot gnu.org> ---
I'm not convinced this is a duplicate, the test tries to catch a
std::system_error not a std::ios::failure.

On Fedora 22 using the old ABI the test passes.

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|DUPLICATE                   |---

--- Comment #13 from Jonathan Wakely <redi at gcc dot gnu.org> ---
Apparently the error is:

/casablanca/Release/tests/functional/streams/fstreambuf_tests.cpp:148: error:
Failure in OpenForReadDoesntCreateFile1:
CHECK_EQUAL(static_cast<int>(std::errc::no_such_file_or_directory),
_condFound.value()) where
static_cast<int>(std::errc::no_such_file_or_directory)=2 and
_condFound.value()=0

Which should not depend on the ABI.

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|---                         |INVALID

--- Comment #14 from Jonathan Wakely <redi at gcc dot gnu.org> ---
However, it's never been supported to combine objects built with different
major versions if using C++11.

(In reply to Gianfranco from comment #0)
> the problem actually is that now Debian has libstdc++ from gcc-5, and the
> default compiler still set to gcc-4.9.

That combination cannot be used for C++11 code.

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[costamagnagianfranco at yahoo dot it]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

--- Comment #15 from Gianfranco <costamagnagianfranco at yahoo dot it> ---
Thanks to all for the suggestions! Apparently this is what I presumed from the
begin (I read a lot about ABI changes in gcc-5)

I'll try to figure out with doko what is the best approach for fixing this!
thanks to all again,

G.

[Bug libstdc++/66262] [REGRESSION] testsuite failure with libstdc++ (gcc-5) and g++-4.9 together

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66262

--- Comment #16 from Jonathan Wakely <redi at gcc dot gnu.org> ---
PR66438 has a proper analysis of this issue (and it isn't related to PR66145)