[petsc-dev] Configure problem on Titan

Mark Adams mfadams at lbl.gov
Wed Mar 11 21:50:14 CDT 2015


FYI, this is with -lstdc++



On Wed, Mar 11, 2015 at 5:05 PM, Barry Smith <bsmith at mcs.anl.gov> wrote:

>
>    Satish,
>
>     Please enhance
>
>   def checkFortranLinkingCxx(self):
>     '''Check that Fortran can be linked against C++'''
>     link = 0
>     cinc, cfunc, ffunc = self.manglerFuncs[self.fortranMangling]
>     cinc = 'extern "C" '+cinc+'\n'
>
>     cxxCode = 'void foo(void){'+self.mangleFortranFunction('d1chk')+'();}'
>     cxxobj  = os.path.join(self.tmpDir, 'cxxobj.o')
>     self.pushLanguage('Cxx')
>     if not self.checkCompile(cinc+cxxCode, None, cleanup = 0):
>
> >>>  try again here with additional library.
>
>       self.logPrint('Cannot compile Cxx function: '+cfunc, 3, 'compilers')
>       raise RuntimeError('Fortran could not successfully link C++ objects')
>     if not os.path.isfile(self.compilerObj):
>       self.logPrint('Cannot locate object file:
> '+os.path.abspath(self.compilerObj), 3, 'compilers')
>       raise RuntimeError('Fortran could not successfully link C++ objects')
>     os.rename(self.compilerObj, cxxobj)
>     self.popLanguage()
>
> so that if the basic link fails it automatically tries again here after
> adding the -lstdc++ library.
>
>    Thanks
>
>     Barry
>
> It is insane to expect users to know they have to add -lstdc++  themselves.
>
>
>
> > On Mar 11, 2015, at 5:10 PM, Satish Balay <balay at mcs.anl.gov> wrote:
> >
> > A more suitable way to specify -lstdc++ would be via option LIBS
> > [this way  its used by clinker,cxxlinker,flinker - as required]
> >
> > LIBS="-lstdc++"
> >
> > Satish
> >
> > On Wed, 11 Mar 2015, Eric Bavier wrote:
> >
> >> It looks like you are building with the PGI compilers.  Try adding:
> >>
> >> --CXX_LINKER_FLAGS="-Meh_frame -lstdc++"
> >>
> >> to your petsc configure flags.
> >>
> >> Hope that helps,
> >> `~Eric Bavier, Scientific Libraries, Cray Inc.
> >> ________________________________________
> >> From: petsc-dev-bounces at mcs.anl.gov [petsc-dev-bounces at mcs.anl.gov] on
> behalf of Satish Balay [balay at mcs.anl.gov]
> >> Sent: Wednesday, March 11, 2015 16:55
> >> To: Mark Adams
> >> Cc: For users of the development version of PETSc
> >> Subject: Re: [petsc-dev] Configure problem on Titan
> >>
> >>>>>>>>>>>
> >> Executing: ftn  -o /tmp/scratch/petsc-5_JlCR/config.compilers/conftest
>   -mp -fast  /tmp/scratch/petsc-5_JlCR/config.compilers/conftest.o
> /tmp/scratch/petsc-5_JlCR/config.compilers/cxxobj.o
> /tmp/scratch/petsc-5_JlCR/config.compilers/confc.o  -ldl
> >> Possible ERROR while running linker: exit code 512
> >> stderr:
> >> /tmp/scratch/petsc-5_JlCR/config.compilers/cxxobj.o:(.eh_frame+0x12):
> undefined reference to `__gxx_personality_v0'
> >> /tmp/scratch/petsc-5_JlCR/config.compilers/confc.o:(.eh_frame+0x12):
> undefined reference to `__gxx_personality_v0'
> >> /usr/bin/ld: link errors found, deleting executable
> `/tmp/scratch/petsc-5_JlCR/config.compilers/conftest'
> >> <<<<<<
> >>
> >> Perhaps the compilers are not setup correctly after the update?
> >>
> >> If you don't need c++ - you can try building with --with-cxx=0
> configure option..
> >>
> >> satish
> >>
> >> On Wed, 11 Mar 2015, Mark Adams wrote:
> >>
> >>> I am getting this error on Titan.  They just had a big PM and it was
> >>> recommended that we rebuild everything
> >>>
> >>
> >>
> >
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20150311/26c72e1c/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: configure.log
Type: application/octet-stream
Size: 79463 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20150311/26c72e1c/attachment.obj>


More information about the petsc-dev mailing list