[Llvm-bgq-discuss] linker errors with llvm/current/bin/bgclang++

Sat Jan 19 21:43:26 CST 2013

Great, thanks.  Let me know if you'd like me to minimize it.  This
time I'll wait a bit to see if I can avoid work.

Geoffrey

On Sat, Jan 19, 2013 at 7:34 PM, Hal Finkel <hfinkel at anl.gov> wrote:
> Geoffrey,
>
> I've not seen that before. I'll rebase our local version next week; if it is still a problem after that I'll file a bug report upstream.
>
> Thanks again,
> Hal
>
> ----- Original Message -----
>> From: "Geoffrey Irving" <irving at naml.us>
>> To: "Hal Finkel" <hfinkel at anl.gov>
>> Cc: llvm-bgq-discuss at lists.alcf.anl.gov
>> Sent: Saturday, January 19, 2013 8:27:43 PM
>> Subject: Re: [Llvm-bgq-discuss] linker errors with llvm/current/bin/bgclang++
>>
>> Ah, that would certainly explain it.
>>
>> Next up, clang is crashing for me with the error shown below.  The
>> full stack trace, precompiled source, and run script are at
>>
>>     http://naml.us/random/clang
>>
>> Have you seen similar things before, or know any workarounds?
>>
>> Thanks,
>> Geoffrey
>>
>> vesta:pentago% cns -j 1
>> scons: building associated VariantDir targets: build/powerpc/release
>> /home/projects/llvm/current/bin/bgclang++ -o
>> build/powerpc/release/endgame.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_core
>> -I/bgsys/drivers/V1R1M2/ppc64/comm/sys/include
>> -I/bgsys/drivers/V1R1M2/ppc64
>> -I/bgsys/drivers/V1R1M2/ppc64/spi/include
>> -I/bgsys/drivers/V1R1M2/ppc64/spi/include/kernel/cnk
>> -I/bgsys/drivers/V1R1M2/ppc64/comm/gcc/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
>> endgame.cpp
>> clang:
>> /gpfs/vesta_home/hfinkel/src/llvm-trunk/tools/clang/lib/Sema/SemaTemplateInstantiate.cpp:2682:
>> llvm::PointerUnion<clang::Decl*,
>> clang::LocalInstantiationScope::DeclArgumentPack*>*
>> clang::LocalInstantiationScope::findInstantiationOf(const
>> clang::Decl*): Assertion `isa<LabelDecl>(D) && "declaration not
>> instantiated in this scope"' failed.
>>
>> On Sat, Jan 19, 2013 at 5:52 PM, Hal Finkel <hfinkel at anl.gov> wrote:
>> > Geoffrey,
>> >
>> > In general, the code compiled with bgclang will not run on the
>> > login nodes. At -O3 (which enables autovectorization), bgclang
>> > will generate QPX vector instructions; those QPX instructions are
>> > not supported by the Power 7 CPU in the login node. You'll need to
>> > submit a job and run on the compute nodes.
>> >
>> >  -Hal
>> >
>> > ----- Original Message -----
>> >> From: "Geoffrey Irving" <geoffrey.irving at gmail.com>
>> >> To: "Hal Finkel" <hfinkel at anl.gov>
>> >> Cc: "Geoffrey Irving" <irving at naml.us>,
>> >> llvm-bgq-discuss at lists.alcf.anl.gov
>> >> Sent: Saturday, January 19, 2013 6:17:56 PM
>> >> Subject: Re: [Llvm-bgq-discuss] linker errors with
>> >> llvm/current/bin/bgclang++
>> >>
>> >> No, this was on a login node.  Am I correct that compiling and
>> >> quick
>> >> unit tests (which are much faster than compiling) should happen on
>> >> login nodes?
>> >>
>> >> Geoffrey
>> >>
>> >> On Jan 19, 2013, at 3:57 PM, Hal Finkel <hfinkel at anl.gov> wrote:
>> >>
>> >> > Geoffrey,
>> >> >
>> >> > You're running on the compute nodes, right? Not the login nodes?
>> >> >
>> >> > -Hal
>> >> >
>> >> > ----- Original Message -----
>> >> >> From: "Geoffrey Irving" <irving at naml.us>
>> >> >> To: "Hal Finkel" <hfinkel at anl.gov>
>> >> >> Cc: llvm-bgq-discuss at lists.alcf.anl.gov
>> >> >> Sent: Friday, January 18, 2013 9:35:18 PM
>> >> >> Subject: Re: [Llvm-bgq-discuss] linker errors with
>> >> >> llvm/current/bin/bgclang++
>> >> >>
>> >> >> Cool, that gets me through linking and unit tests in debug mode
>> >> >> and
>> >> >> optimized with -O2.  However, I get an illegal instruction in
>> >> >> some
>> >> >> static initialization code if I use -O3.  I'll see how easy it
>> >> >> is
>> >> >> to
>> >> >> reduce my test case.
>> >> >>
>> >> >> In any case, progress!
>> >> >>
>> >> >> Thanks,
>> >> >> Geoffrey
>> >> >>
>> >> >> On Fri, Jan 18, 2013 at 7:18 PM, Hal Finkel <hfinkel at anl.gov>
>> >> >> wrote:
>> >> >>> Geoffrey,
>> >> >>>
>> >> >>> I believe this bug has been fixed in trunk; I'll updated our
>> >> >>> install next week. In the mean time, try building with
>> >> >>> -mcmodel=small
>> >> >>>
>> >> >>> -Hal
>> >> >>>
>> >> >>> ----- Original Message -----
>> >> >>>> From: "Geoffrey Irving" <irving at naml.us>
>> >> >>>> To: llvm-bgq-discuss at lists.alcf.anl.gov
>> >> >>>> Sent: Friday, January 18, 2013 9:10:59 PM
>> >> >>>> Subject: [Llvm-bgq-discuss] linker errors with
>> >> >>>> llvm/current/bin/bgclang++
>> >> >>>>
>> >> >>>> I'm trying to get my C++11 app running on vesta.  Compiling
>> >> >>>> with
>> >> >>>> clang
>> >> >>>> gives me the following linker errors (summarized below, with
>> >> >>>> full
>> >> >>>> build output attached):
>> >> >>>>
>> >> >>>> /home/projects/llvm/current/bin/bgclang++ -o
>> >> >>>> build/powerpc/release/core/array/RawArray.os -c
>> >> >>>> -U__GXX_EXPERIMENTAL_CXX0X__ -O3 -funroll-loops -Wall
>> >> >>>> -Winit-self
>> >> >>>> -Woverloaded-virtual -Wsign-compare -fno-strict-aliasing
>> >> >>>> -std=c++0x
>> >> >>>> -Werror -Wno-array-bounds -Wno-unknown-pragmas -fPIC
>> >> >>>> -DOTHER_PYTHON
>> >> >>>> -DNDEBUG -DOTHER_THREAD_SAFE=1 -DBOOST_EXCEPTION_DISABLE
>> >> >>>> -DBUILDING_other_core -I/usr/include/python2.6
>> >> >>>> -I/home/irving/.local/lib/python2.6/site-packages/numpy/core/include
>> >> >>>> -I/usr/local/include -Ibuild/include
>> >> >>>> -I/home/irving/download/boost_1_52_0 core/array/RawArray.cpp
>> >> >>>> ...MANY OTHER COMPILE COMMANDS...
>> >> >>>> /home/projects/llvm/current/bin/bgclang++ -o
>> >> >>>> build/powerpc/release/core/libother_core.so -shared
>> >> >>>> -Wl,-rpath=/gpfs/vesta_home/irving/otherlab/other/install/release/lib
>> >> >>>> build/powerpc/release/core/array/RawArray.os
>> >> >>>> ...MANY_OTHER_OBJECT_FILES... -L/usr/lib64 -L/usr/local/lib
>> >> >>>> -Linstall/release/lib -lpython2.6
>> >> >>>> /bgsys/drivers/ppcfloor/gnu-linux/lib/gcc/powerpc64-bgq-linux/4.4.6/../../../../powerpc64-bgq-linux/bin/ld:
>> >> >>>> build/powerpc/release/core/array/RawArray.os(.text+0x63e):
>> >> >>>> unresolvable R_PPC64_TOC16_HA relocation against symbol
>> >> >>>> `_ZTVSt9bad_alloc@@GLIBCXX_3.4'
>> >> >>>> /bgsys/drivers/ppcfloor/gnu-linux/lib/gcc/powerpc64-bgq-linux/4.4.6/../../../../powerpc64-bgq-linux/bin/ld:
>> >> >>>> build/powerpc/release/core/array/RawArray.os(.text+0x642):
>> >> >>>> unresolvable R_PPC64_TOC16_LO relocation against symbol
>> >> >>>> `_ZTVSt9bad_alloc@@GLIBCXX_3.4'
>> >> >>>> /bgsys/drivers/ppcfloor/gnu-linux/lib/gcc/powerpc64-bgq-linux/4.4.6/../../../../powerpc64-bgq-linux/bin/ld:
>> >> >>>> build/powerpc/release/core/array/RawArray.os(.text+0x652):
>> >> >>>> unresolvable R_PPC64_TOC16_HA relocation against symbol
>> >> >>>> `_ZTISt9bad_alloc@@GLIBCXX_3.4'
>> >> >>>> /bgsys/drivers/ppcfloor/gnu-linux/lib/gcc/powerpc64-bgq-linux/4.4.6/../../../../powerpc64-bgq-linux/bin/ld:
>> >> >>>> build/powerpc/release/core/array/RawArray.os(.text+0x656):
>> >> >>>> unresolvable R_PPC64_TOC16_LO relocation against symbol
>> >> >>>> `_ZTISt9bad_alloc@@GLIBCXX_3.4'
>> >> >>>> /bgsys/drivers/ppcfloor/gnu-linux/lib/gcc/powerpc64-bgq-linux/4.4.6/../../../../powerpc64-bgq-linux/bin/ld:
>> >> >>>> final link failed: Nonrepresentable section on output
>> >> >>>> clang: error: linker command failed with exit code 1 (use -v
>> >> >>>> to
>> >> >>>> see
>> >> >>>> invocation)
>> >> >>>> scons: *** [build/powerpc/release/core/libother_core.so]
>> >> >>>> Error 1
>> >> >>>>
>> >> >>>> Any ideas what might be going wrong?  A google search turned
>> >> >>>> up
>> >> >>>> many
>> >> >>>> similar bugs in the past, but I haven't found a specific
>> >> >>>> reference.
>> >> >>>> Notes:
>> >> >>>>
>> >> >>>> 1. The only library I'm linking to is Python.
>> >> >>>> 2. I needed to undef __GXX_EXPERIMENTAL_CXX0X__ in order to
>> >> >>>> get
>> >> >>>> -std=c++0x to compile.  Otherwise it complains about stuff
>> >> >>>> like
>> >> >>>> this
>> >> >>>> (more details on request):
>> >> >>>>
>> >> >>>> /bgsys/drivers/ppcfloor/gnu-linux/powerpc64-bgq-linux/include/c++/4.4.6/bits/stl_pair.h:106:10:
>> >> >>>> error: no matching function for call to 'move'
>> >> >>>>        : first(std::move(__p.first)),
>> >> >>>>                ^~~~~~~~~
>> >> >>>>
>> >> >>>> Thanks,
>> >> >>>> Geoffrey
>> >> >>>>
>> >> >>>> _______________________________________________
>> >> >>>> llvm-bgq-discuss mailing list
>> >> >>>> llvm-bgq-discuss at lists.alcf.anl.gov
>> >> >>>> https://lists.alcf.anl.gov/mailman/listinfo/llvm-bgq-discuss
>> >> >>>>
>> >> >>>
>> >> >>> --
>> >> >>> Hal Finkel
>> >> >>> Postdoctoral Appointee
>> >> >>> Leadership Computing Facility
>> >> >>> Argonne National Laboratory
>> >> >>
>> >> >
>> >> > --
>> >> > Hal Finkel
>> >> > Postdoctoral Appointee
>> >> > Leadership Computing Facility
>> >> > Argonne National Laboratory
>> >>
>> >
>> > --
>> > Hal Finkel
>> > Postdoctoral Appointee
>> > Leadership Computing Facility
>> > Argonne National Laboratory
>>
>
> --
> Hal Finkel
> Postdoctoral Appointee
> Leadership Computing Facility
> Argonne National Laboratory
> _______________________________________________
> llvm-bgq-discuss mailing list
> llvm-bgq-discuss at lists.alcf.anl.gov
> https://lists.alcf.anl.gov/mailman/listinfo/llvm-bgq-discuss