[Llvm-bgq-discuss] bad elf

[Hal Finkel]

----- Original Message -----
> From: "Geoffrey Irving" <irving at naml.us>
> To: llvm-bgq-discuss at lists.alcf.anl.gov
> Sent: Friday, February 1, 2013 1:31:46 AM
> Subject: [Llvm-bgq-discuss] bad elf
> 
> I switched to clang mpi wrapper compilers, but still get the bad elf
> error:
> 
>     ...
>     2013-02-01 07:06:51.870 (FATAL) [0xfff829da210]
> VST-02640-13751-32:22357:ibm.runjob.client.Job: could not start job:
> job failed to start
>     2013-02-01 07:06:51.871 (FATAL) [0xfff829da210]
> VST-02640-13751-32:22357:ibm.runjob.client.Job: Load failed on
> Q02-I7-J00: Application executable ELF header contains invalid value,
> errno 8 Exec format error

This almost always indicates that you'd somehow ended up linking to the wrong c-runtime-library startup routines. You've picked up the login-node versions and not the compute-node versions. I don't know whether or not those get included in the shared libraries as well.

If you run the 'file' utility on the various .so files and the main executable it should report a 'linux version' associated with the file. Which version does it say? Are they all the same?

 -Hal

>     ...
> 
> Any ideas what might be causing this?  I put the full output files
> (error, output, cobalt) and the program (though not the shared
> libraries it depends on) here:
> 
>     http://naml.us/random/clang/bad-1
> 
> The program in question is indeed an elf, according to file:
> 
>     vesta:bad-1% file
>     ~/otherlab/other/install/release/bin/endgame-mpi
>     /home/irving/otherlab/other/install/release/bin/endgame-mpi: ELF
> 64-bit MSB shared object, 64-bit PowerPC or cisco 7500, version 1
> (SYSV), dynamically linked, not stripped
> 
> The program was built via
> 
> /home/projects/llvm/mpi/clang.legacy/bin/mpiclang++ -o
> build/powerpc/release/mpi/main.os -c -U__GXX_EXPERIMENTAL_CXX0X__
> -mcmodel=small -O2 -Wall -Winit-self -Woverloaded-virtual
> -Wsign-compare -fno-strict-aliasing -std=c++11 -Werror
> -Wno-array-bounds -Wno-unknown-pragmas -Wno-invalid-offsetof -fPIC
> -DUSE_MPI -DOTHER_PYTHON -DNDEBUG -DOTHER_THREAD_SAFE=1
> -DBOOST_EXCEPTION_DISABLE -DBUILDING_pentago_mpi
> -I/home/irving/include -I/usr/include/python2.6
> -I/home/irving/.local/lib/python2.6/site-packages/numpy/core/include
> -I/home/irving/include -I/usr/local/include -Ibuild/include
> -Ibuild/powerpc/release -I. -I/home/irving/download/boost_1_52_0
> mpi/main.cpp
> /home/projects/llvm/mpi/clang.legacy/bin/mpiclang++ -o
> build/powerpc/release/mpi/endgame-mpi -shared
> -Wl,-rpath=/home/irving/lib
> -Wl,-rpath=/gpfs/vesta_home/irving/otherlab/other/install/release/lib
> build/powerpc/release/mpi/main.os -L/usr/lib64 -L/home/irving/lib
> -L/usr/local/lib
> -L/gpfs/vesta_home/irving/otherlab/other/install/release/lib
> -lpentago_mpi -lpentago_core -lother_core -llzma -lpython2.6 -lz
> -lsnappy
> Install file: "build/powerpc/release/mpi/endgame-mpi" as
> "/gpfs/vesta_home/irving/otherlab/other/install/release/bin/endgame-mpi"
> 
> Again, that's ignoring all the shared libraries.  Those could be the
> problem, but I'd hope an error in finding or parsing shared libraries
> would look different than this.
> 
> Close!  Thanks for the all the help!
> 
> Geoffrey
> _______________________________________________
> llvm-bgq-discuss mailing list
> llvm-bgq-discuss at lists.alcf.anl.gov
> https://lists.alcf.anl.gov/mailman/listinfo/llvm-bgq-discuss
>