[mvapich-discuss] Error building with PGI 19.x and MVAPICH2 (v 2.3.1) on Centos 7

[Subramoni, Hari]

Hi, Tim.

Please accept my sincere apologies for the delay in getting back to you on this. Can you please try the patch below and see if it works for you?

diff --git a/src/pm/mpirun/mpispawn.c b/src/pm/mpirun/mpispawn.c
index 9941977..75cfa7b 100644
--- a/src/pm/mpirun/mpispawn.c
+++ b/src/pm/mpirun/mpispawn.c
@@ -336,7 +336,7 @@ char *obtain_host_list_from_file()
     return host_list;
 }

-void find_children (int id, int degree, int size, int children[degree])
+void find_children (int id, int degree, int size, int child[degree])
 {
     size_t c_start = degree * id + 1;
     size_t i;
@@ -346,14 +346,14 @@ void find_children (int id, int degree, int size, int children[degree])
     }

     for (i = 0; i < degree; i++) {
-        children[i] = -1;
+        child[i] = -1;
     }

     if (c_start < size) {
         i = (degree > size - c_start) ? size - c_start : degree;

         while (i--) {
-            children[i] = c_start + i;
+            child[i] = c_start + i;
         }
     }
 }

Thx,
Hari.

-----Original Message-----
From: mvapich-discuss-bounces at cse.ohio-state.edu <mvapich-discuss-bounces at mailman.cse.ohio-state.edu> On Behalf Of Carlson, Timothy S
Sent: Friday, July 19, 2019 12:39 PM
To: mvapich-discuss at cse.ohio-state.edu <mvapich-discuss at mailman.cse.ohio-state.edu>
Subject: [mvapich-discuss] Error building with PGI 19.x and MVAPICH2 (v 2.3.1) on Centos 7

Hello,
I have found that a basic build of mvapich2 with the PGI compilers fails when building mvapich2 2.3.1. To reproduce

./configure --with-device=ch3:psm --prefix=/share/apps/mvapich2/2.3.1/pgi/19.1

With CC/CXX/FC/F77 set correctly for the PGI compilers.

The error comes when compiling mpispawn.c

make V=1

source='src/pm/mpirun/mpispawn.c' object='src/pm/mpirun/mpispawn.o' libtool=no \ DEPDIR=.deps depmode=pgcc /bin/sh ./confdb/depcomp \
pgcc -DHAVE_CONFIG_H -I. -I./src/include  -I./src/include -I./src/mpi/datatype -I./src/mpi/spawn -I./src/util/dbg -Isrc/util/logging/common -I./src/util/mem -I./src/util/thread -I./src/util/type -I./src/util/wrappers -I./src/util/wrappers -I./src/util/assert -I./src/util/timers -I./src/util/pointer -I./src/binding/cxx -I/root/mvapich2-2.3.1/src/binding/fortran/mpif_h -Isrc/binding/fortran/use_mpi  -I./src/mpid/ch3/include -I./src/mpid/ch3/include -I/root/mvapich2-2.3.1/src/mpid/ch3/channels/common/include          -I./src/mpid/ch3/channels/psm/include -I./src/mpi/coll -D_GNU_SOURCE    -I./src/mpid/ch3/channels/common/include    -I./src/mpid/common/datatype -I./src/mpid/common/datatype -I./src/mpid/common/sched   -I./src/mpid/common/thread -I./src/mpid/common/timers -I. -I./src/mpid/ch3/channels/common/include -I./src/pm/mpirun -I./src/mpid/ch3/channels/common/include -I./src/pm/mpirun -I./src/pm/mpirun/src -I./src/pm/mpirun/src/hostfile -I./src/mpid/ch3/channels/common/include -I./src/pm/mpirun -I./src/pm/mpirun/src -I./src/pm/mpirun/src/slurm -I./src/mpid/ch3/channels/common/include -I./src/pm/mpirun -I./src/pm/mpirun/src -I./src/mpid/ch3/channels/common/include   -I./src/pmi/simple -I./src/pmi/simple -Icontrib/hwloc/include/ -I./contrib/hwloc/include -I/root/mvapich2-2.3.1/src/mpid/ch3/channels/psm/include -I/root/mvapich2-2.3.1/src/mpid/ch3/channels/psm/include -I/root/mvapich2-2.3.1/src/util/wrappers -I/root/mvapich2-2.3.1/src/util/wrappers -I/root/mvapich2-2.3.1/src/mpl/include -I/root/mvapich2-2.3.1/src/mpl/include -I/root/mvapich2-2.3.1/src/openpa/src -I/root/mvapich2-2.3.1/src/openpa/src -D_REENTRANT -I/root/mvapich2-2.3.1/src/mpi/romio/include -I/include -I/include  -DNDEBUG -DNVALGRIND -O2 -c -o src/pm/mpirun/mpispawn.o src/pm/mpirun/mpispawn.c
/qfs/projects/ops/rh7/apps/pgi/19.1/linux86-64-llvm/19.1/share/llvm/bin/opt: /tmp/pgccHz0RbJC5eSYAu.ll:9974:27: error: forward reference and definition of global have different types @children = common global %astruct.dt27083* zeroinitializer, align 8

If I replace -O2 with -O it goes past the error without problem.  This is true for the PGI 19.1 and 19.5

I have not dug too deeply into this.. I'm a sys-admin not a developer 😊

Thanks

Tim

Tim Carlson
Team Lead - HPC, Research Computing
Communications & Information Technology Directorate Pacific Northwest National Laboratory | www.pnnl.gov
 
509.375.6435 | tim at pnnl.gov 








_______________________________________________
mvapich-discuss mailing list
mvapich-discuss at cse.ohio-state.edu
http://mailman.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss