By the way, petsc did compile and install metis and parmetis succesfully before the make error. However, running the newly compiled gpmetis program gives the same segmentation fault! So the original problem was not solved by recompiling, unfortunately.
Chris ________________________________ From: Klaij, Christiaan Sent: Wednesday, January 04, 2017 3:53 PM To: Matthew Knepley Cc: petsc-users; Satish Balay Subject: Re: [petsc-users] problems after glibc upgrade to 2.17-157 So how would I do that? Does LIBS=<string> accept spaces in the string? Something like this perhaps: LIBS="-L/cm/shared/apps/intel/compilers_and_libraries_2016.3.210/linux/compiler/lib/intel64_lin -lifcore" But I'm starting to believe that my intel install is somehow broken. I'm getting these intel compilers from rpm's provided by our cluster vendor. On a workstation I can try yum remove and install of the intel packages. Not so easy on a production cluster. Is this worth a try? Or will it just copy/paste the same broken (?) stuff in the same place? Chris ________________________________ From: Matthew Knepley <knep...@gmail.com> Sent: Wednesday, January 04, 2017 3:13 PM To: Klaij, Christiaan Cc: petsc-users; Satish Balay Subject: Re: [petsc-users] problems after glibc upgrade to 2.17-157 On Wed, Jan 4, 2017 at 7:37 AM, Klaij, Christiaan <c.kl...@marin.nl<mailto:c.kl...@marin.nl>> wrote: I've tried with: --LIBS=/cm/shared/apps/intel/compilers_and_libraries_2016.3.210/linux/compiler/lib/intel64_lin/libifcore.a -lstdc++\\ This is likely connected to the problem below, but I would have to see the log. but that doesn't seem to make a difference. With the option --with-cxx=0 the configure part does work(!), but then I get **************************ERROR************************************* Error during compile, check Linux-x86_64-Intel/lib/petsc/conf/make.log Send it and Linux-x86_64-Intel/lib/petsc/conf/configure.log to petsc-ma...@mcs.anl.gov<mailto:petsc-ma...@mcs.anl.gov> ******************************************************************* Here is the problem: CLINKER /projects/developers/cklaij/ReFRESCO/Dev/trunk/Libs/build/petsc-3.7.4/Linux-x86_64-Intel/lib/libpetsc.so.3.7.4 ld: /cm/shared/apps/intel/compilers_and_libraries_2016.3.210/linux/compiler/lib/intel64_lin/libifcore.a(for_init.o): relocation R_X86_64_32 against `.rodata.str1.4' can not be used when making a shared object; recompile with -fPIC /cm/shared/apps/intel/compilers_and_libraries_2016.3.210/linux/compiler/lib/intel64_lin/libifcore.a: could not read symbols: Bad value Clearly there is something wrong with the compiler install. However, can you put a libifcore.so in LIBS instead? Matt See the attached log files. Chris dr. ir. Christiaan Klaij | CFD Researcher | Research & Development MARIN | T +31 317 49 33 44<tel:+31%20317%20493%20344> | c.kl...@marin.nl<mailto:c.kl...@marin.nl> | www.marin.nl<http://www.marin.nl> [LinkedIn]<https://www.linkedin.com/company/marin> [YouTube] <http://www.youtube.com/marinmultimedia> [Twitter] <https://twitter.com/MARIN_nieuws> [Facebook] <https://www.facebook.com/marin.wageningen> MARIN news: MARIN Report 119: highlighting naval research projects<http://www.marin.nl/web/News/News-items/MARIN-Report-119-highlighting-naval-research-projects.htm> ________________________________ From: Matthew Knepley <knep...@gmail.com<mailto:knep...@gmail.com>> Sent: Wednesday, January 04, 2017 1:43 PM To: Klaij, Christiaan Cc: petsc-users; Satish Balay Subject: Re: [petsc-users] problems after glibc upgrade to 2.17-157 On Wed, Jan 4, 2017 at 4:32 AM, Klaij, Christiaan <c.kl...@marin.nl<mailto:c.kl...@marin.nl>> wrote: Satish, I tried your suggestion: --with-clib-autodetect=0 --with-fortranlib-autodetect=0 --with-cxxlib-autodetect=0 LIBS=LIBS=/path_to/libifcore.a I guess I don't really need "LIBS= " twice (?) so I've used this line: LIBS=/cm/shared/apps/intel/compilers_and_libraries_2016.3.210/linux/compiler/lib/intel64_lin/libifcore.a Unfortunately, this approach also fails (attached log): Ah, this error is much easier: Executing: mpif90 -o /tmp/petsc-3GfeyZ/config.compilers/conftest -fPIC -g -O3 /tmp/petsc-3GfeyZ/config.compilers/conftest.o /tmp/petsc-3GfeyZ/config.compilers/cxxobj.o /tmp/petsc-3GfeyZ/config.compilers/confc.o -ldl /cm/shared/apps/intel/compilers_and_libraries_2016.3.210/linux/compiler/lib/intel64_lin/libifcore.a Possible ERROR while running linker: exit code 256 stderr: /tmp/petsc-3GfeyZ/config.compilers/cxxobj.o:(.gnu.linkonce.d.DW.ref.__gxx_personality_v0+0x0): undefined reference to `__gxx_personality_v0' Intel as lazy writing its C++ compiler, so it uses some of g++. If you want to use C++, you will need to add -lstdc++ to your LIBS variable (I think). Otherwise, please turn it off using --with-cxx=0. Thanks, Matt ******************************************************************************* UNABLE to CONFIGURE with GIVEN OPTIONS (see configure.log for details): ------------------------------------------------------------------------------- Fortran could not successfully link C++ objects ******************************************************************************* There are multiple libifcore.a in the intel compiler lib: one in intel64_lin and one in intel64_lin_mic. Tried both, got same error. Chris dr. ir. Christiaan Klaij | CFD Researcher | Research & Development MARIN | T +31 317 49 33 44<tel:%2B31%20317%2049%2033%2044> | mailto:c.kl...@marin.nl<mailto:c.kl...@marin.nl> | http://www.marin.nl MARIN news: http://www.marin.nl/web/News/News-items/Comparison-of-uRANS-and-BEMBEM-for-propeller-pressure-pulse-prediction.htm ________________________________________ From: Satish Balay <ba...@mcs.anl.gov<mailto:ba...@mcs.anl.gov>> Sent: Tuesday, January 03, 2017 4:37 PM To: Klaij, Christiaan Cc: petsc-users@mcs.anl.gov<mailto:petsc-users@mcs.anl.gov> Subject: Re: [petsc-users] problems after glibc upgrade to 2.17-157 Do you have similar issues with gnu compilers? It must be some incompatibility with intel compilers with this glibc change. >>>>>>>>> compilers: Check that C libraries can be used from Fortran Pushing language FC Popping language FC Pushing language FC Popping language FC Pushing language FC Popping language FC **** Configure header /tmp/petsc-rOjdnN/confdefs.h **** <<<<<<<<<< Thre is a bug in configure [Matt?] that eats away some of the log - so I don't see the exact error you are getting. If standalone micc/mpif90 etc work - then you can try the following additional options: --with-clib-autodetect=0 --with-fortranlib-autodetect=0 --with-cxxlib-autodetect=0 LIBS=LIBS=/path_to/libifcore.a [replace "path_to" with the correct path to the ifort lubifcore.a library] Note: I have a RHEL7 box with this glibc - and I don't see this issue. >>>> -bash-4.2$ cat /etc/redhat-release Red Hat Enterprise Linux Server release 7.3 (Maipo) -bash-4.2$ rpm -q glibc glibc-2.17-157.el7_3.1.x86_64 glibc-2.17-157.el7_3.1.i686 -bash-4.2$ mpiicc --version icc (ICC) 17.0.0 20160721 Copyright (C) 1985-2016 Intel Corporation. All rights reserved. -bash-4.2$ <<<< Satish On Tue, 3 Jan 2017, Klaij, Christiaan wrote: > > I've been using petsc-3.7.4 with intel mpi and compilers, > superlu_dist, metis and parmetis on a cluster running > SL7. Everything was working fine until SL7 got an update where > glibc was upgraded from 2.17-106 to 2.17-157. > > This update seemed to have broken (at least) parmetis: the > standalone binary gpmetis started to give a segmentation > fault. The core dump shows this: > > Core was generated by `gpmetis'. > Program terminated with signal 11, Segmentation fault. > #0 0x00002aaaac6b865e in memmove () from /lib64/libc.so.6 > > That's when I decided to recompile, but to my surprise I cannot > even get past the configure stage (log attached)! > > ******************************************************************************* > UNABLE to EXECUTE BINARIES for ./configure > ------------------------------------------------------------------------------- > Cannot run executables created with FC. If this machine uses a batch system > to submit jobs you will need to configure using ./configure with the > additional option --with-batch. > Otherwise there is problem with the compilers. Can you compile and run code > with your compiler 'mpif90'? > See http://www.mcs.anl.gov/petsc/documentation/faq.html#libimf > ******************************************************************************* > > Note the following: > > 1) Configure was done with the exact same options that worked > fine before the update of SL7. > > 2) The intel mpi and compilers are exactly the same as before the > update of SL7. > > 3) The cluster does not require a batch system to run code. > > 4) I can compile and run code with mpif90 on this cluster. > > 5) The problem also occurs on a workstation running SL7. > > Any clues on how to proceed? > Chris > > > dr. ir. Christiaan Klaij | CFD Researcher | Research & Development > MARIN | T +31 317 49 33 44<tel:%2B31%20317%2049%2033%2044> | > mailto:c.kl...@marin.nl<mailto:c.kl...@marin.nl> | http://www.marin.nl > > MARIN news: > http://www.marin.nl/web/News/News-items/Comparison-of-uRANS-and-BEMBEM-for-propeller-pressure-pulse-prediction.htm > > -- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener -- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener