[Users] Stuck compiling dh.cc
Roland Haas
rhaas at illinois.edu
Tue Sep 29 15:36:44 CDT 2020
Hello Chris,
my best guess is that the MPI stack on your system got updated.
MPI_UB has been removed in newer MPI standards and OpenMPI 4 does no
longer offer it by default.
Carpet should have stopped using it in commit
5073af2caddf84b64d0aa475968cec4e9845ba73 "CarpetLib: Use
MPI_Type_create_struct instead of MPI_Type_struct" from 2016.
The best solution would be to update your Einstein Toolkit tree.
If that is not possible you can try and see if you can downgrade or use
an older MPI stack.
Finally you can (if this is your workstation or so) ask Cactus to build
the MPI library included in ExternalLibraries/MPI from source (which is
configured to provide the old MPI_UB symbol) by setting:
MPI_DIR=BUILD
in your option list and compiling from scratch (make sim-realclean).
Yours,
Roland
> Hi everyone,
>
> I've managed to get this compiling and have just completed the example BBH
> run simulating the system of the first detection, however now that I have
> left it for a few weeks and come back, naturally nothing works.
>
> The error I now have is with MPI_UB and dh.cc when building through
> simfactory. I get the following output:
>
> In file included from
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dist.hh:12,
> from
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/vect.hh:13,
> from
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/bbox.hh:11,
> from
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:10:
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc: In
> function ‘ompi_datatype_t* mpi_datatype(const dh::light_dboxes&)’:
> /opt/openmpi4/include/mpi.h:337:63: error: expected ‘}’ before ‘;’ token
> #define THIS_SYMBOL_WAS_REMOVED_IN_MPI30(symbol, new_symbol) 0;
> static_assert(0, #symbol " was removed in MPI-3.0. Use " #new_symbol "
> instead.")
> ^
> /opt/openmpi4/include/mpi.h:1131:24: note: in expansion of macro
> ‘THIS_SYMBOL_WAS_REMOVED_IN_MPI30’
> # define MPI_UB THIS_SYMBOL_WAS_REMOVED_IN_MPI30(MPI_UB,
> MPI_Type_create_resized);
> ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:2195:23:
> note: in expansion of macro ‘MPI_UB’
> {1, sizeof s, MPI_UB, "MPI_UB", "MPI_UB"}};
> ^~~~~~
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:2195:9:
> note: to match this ‘{’
> {1, sizeof s, MPI_UB, "MPI_UB", "MPI_UB"}};
> ^
> In file included from
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dist.hh:12,
> from
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/vect.hh:13,
> from
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/bbox.hh:11,
> from
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:10:
> /opt/openmpi4/include/mpi.h:337:63: error: expected ‘}’ before ‘;’ token
> #define THIS_SYMBOL_WAS_REMOVED_IN_MPI30(symbol, new_symbol) 0;
> static_assert(0, #symbol " was removed in MPI-3.0. Use " #new_symbol "
> instead.")
> ^
> /opt/openmpi4/include/mpi.h:1131:24: note: in expansion of macro
> ‘THIS_SYMBOL_WAS_REMOVED_IN_MPI30’
> # define MPI_UB THIS_SYMBOL_WAS_REMOVED_IN_MPI30(MPI_UB,
> MPI_Type_create_resized);
> ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:2195:23:
> note: in expansion of macro ‘MPI_UB’
> {1, sizeof s, MPI_UB, "MPI_UB", "MPI_UB"}};
> ^~~~~~
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:2188:46:
> note: to match this ‘{’
> dist::mpi_struct_descr_t const descr[] = {
> ^
> In file included from
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dist.hh:12,
> from
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/vect.hh:13,
> from
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/bbox.hh:11,
> from
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:10:
> /opt/openmpi4/include/mpi.h:337:79: error: static assertion failed: MPI_UB
> was removed in MPI-3.0. Use MPI_Type_create_resized instead.
> #define THIS_SYMBOL_WAS_REMOVED_IN_MPI30(symbol, new_symbol) 0;
> static_assert(0, #symbol " was removed in MPI-3.0. Use " #new_symbol "
> instead.")
>
> ^
> /opt/openmpi4/include/mpi.h:1131:24: note: in expansion of macro
> ‘THIS_SYMBOL_WAS_REMOVED_IN_MPI30’
> # define MPI_UB THIS_SYMBOL_WAS_REMOVED_IN_MPI30(MPI_UB,
> MPI_Type_create_resized);
> ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:2195:23:
> note: in expansion of macro ‘MPI_UB’
> {1, sizeof s, MPI_UB, "MPI_UB", "MPI_UB"}};
> ^~~~~~
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:2195:29:
> error: expected primary-expression before ‘,’ token
> {1, sizeof s, MPI_UB, "MPI_UB", "MPI_UB"}};
> ^
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:2188:36:
> warning: unused variable ‘descr’ [-Wunused-variable]
> dist::mpi_struct_descr_t const descr[] = {
> ^~~~~
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:2177:23:
> warning: unused variable ‘newtype’ [-Wunused-variable]
> static MPI_Datatype newtype;
> ^~~~~~~
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:2195:50:
> warning: no return statement in function returning non-void [-Wreturn-type]
> {1, sizeof s, MPI_UB, "MPI_UB", "MPI_UB"}};
> ^
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc: At
> global scope:
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:2197:5:
> error: ‘newtype’ does not name a type; did you mean ‘iswctype’?
> newtype = dist::create_mpi_datatype(sizeof descr / sizeof descr[0],
> descr,
> ^~~~~~~
> iswctype
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:2208:5:
> error: ‘initialised’ does not name a type; did you mean ‘MPI_Initialized’?
> initialised = true;
> ^~~~~~~~~~~
> MPI_Initialized
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:2209:3:
> error: expected declaration before ‘}’ token
> }
> ^
> /maybehome/cstevens/Cactus/arrangements/Carpet/CarpetLib/src/dh.cc:2177:23:
> warning: ‘newtype’ defined but not used [-Wunused-variable]
> static MPI_Datatype newtype;
> ^~~~~~~
> make[3]: ***
> [/maybehome/cstevens/Cactus/configs/CTGD_Testing/config-data/make.config.rules:281:
> dh.cc.o] Error 1
> make[2]: *** [/maybehome/cstevens/Cactus/lib/make/make.thornlib:113:
> make.checked] Error 2
> make[1]: *** [/maybehome/cstevens/Cactus/lib/make/make.configuration:181:
> /maybehome/cstevens/Cactus/configs/CTGD_Testing/lib/libthorn_CarpetLib.a]
> Error 2
> make: *** [Makefile:260: CTGD_Testing] Error 2
>
> I did find the following thread:
>
> http://lists.einsteintoolkit.org/pipermail/users/2018-October/006566.html
>
> and I've done a clean compilation but same deal. I'm using OpenMPI 4.0.5
> and gcc/g++ etc 8.3.1. I've confirmed which mpicc etc points to the right
> place, and what I think the relevant lines in the option list file are:
>
> CPPFLAGS = -DOMPI_SKIP_MPICXX
> FPPFLAGS = -traditional
> CFLAGS = -g3 -std=gnu11 -Wunused-variable
> CXXFLAGS = -g3 -std=gnu++11 -Wunused-variable
> F77FLAGS = -g3 -fcray-pointer -ffixed-line-length-none
> F90FLAGS = -g3 -fcray-pointer -ffixed-line-length-none# -Wextra
> -Wimplicit-interface#-fallow-argument-mismatch
>
> MPI_DIR = /opt/openmpi4/
> MPI_LIB_DIRS = /opt/openmpi4/lib
> MPI_INC_DIRS = /opt/openmpi4/include
> MPI_LIBS = mpi open-rte open-pal
>
> I have added CPP = ... CXX = ... etc which point to the appropriate places.
>
> I also have
>
> cstevens at skynet:~/Cactus$ mpic++ --showme
> g++ -I/opt/openmpi4/include -pthread -Wl,-rpath -Wl,/opt/openmpi4/lib
> -Wl,--enable-new-dtags -L/opt/openmpi4/lib -lmpi
>
> Any help would be greatly appreciated.
>
> Thanks,
>
> Chris
>
--
My email is as private as my paper mail. I therefore support encrypting
and signing email messages. Get my PGP key from http://pgp.mit.edu .
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
Url : http://lists.einsteintoolkit.org/pipermail/users/attachments/20200929/e9b4a22c/attachment-0001.bin
More information about the Users
mailing list