Dear All,
I tried "mpic++” to make wxWidgets library and it doesn’t change anything.
I found the openmpi-1.10.0 on my Mac: OSX 10.9.5 with Apple Clang 6.0 always
fails to MPI_Finalize even with a very simple program (bottom of this mail).
[Venus:60708] [ 4] Assertion failed: (OPAL_OBJ_MAGIC_ID == ((opal_object_t *)
(&fl->fl_allocations))->obj_magic_id), function opal_free_list_destruct, file
../../opal/class/opal_free_list.c, line 70.
[Venus:60709] *** Process received signal ***
[Venus:60709] Signal: Abort trap: 6 (6)
[Venus:60709] Signal code: (0)
[Venus:60709] [ 0] 0 libsystem_platform.dylib 0x00007fff8e0445aa
_sigtramp + 26
[Venus:60709] [ 1] 0 ??? 0x0000000000000000
0x0 + 0
[Venus:60709] [ 2] 0 libsystem_c.dylib 0x00007fff9496db1a
abort + 125
[Venus:60709] [ 3] 0 libsystem_c.dylib 0x00007fff949379bf
basename + 0
[Venus:60709] [ 4] 0 libopen-pal.13.dylib 0x000000010c0a9213
opal_free_list_destruct + 515
[Venus:60709] [ 5] 0 mca_osc_rdma.so 0x000000010c542ad5
component_finalize + 101
[Venus:60709] [ 6] 0 libopen-pal.13.dylib 0x000000010f265213
opal_free_list_destruct + 515
[Venus:60708] [ 5] 0 mca_osc_rdma.so 0x000000010f6fead5
component_finalize + 101
[Venus:60708] [ 6] 0 libmpi.12.dylib 0x000000010bdecf8a
ompi_osc_base_finalize + 74
[Venus:60709] [ 7] 0 libmpi.12.dylib 0x000000010efaaf8a
ompi_osc_base_finalize + 74
[Venus:60708] [ 7] 0 libmpi.12.dylib 0x000000010bc7e57a
ompi_mpi_finalize + 2746
[Venus:60709] [ 8] 0 libmpi.12.dylib 0x000000010bcbb56d
MPI_Finalize + 125
Regarding my code I mentioned in my original mail, the behaviour is very weird.
MPI_Isend is called from the different named function, it works.
And I wrote a sample program to try to reproduce my problem but it works fine,
except the problem of MPI_Finalize.
So I decided to make gcc-5.2 and make openmpi on it, which seems to be a
recommendation of the FINK project.
2015/10/28 8:29、ABE Hiroshi <hab...@gmail.com> のメール：
> Dear Nathan and all,
>
> Thank you for your information. I tried it in this morning, it seems to get
> the same result. I will try another option. Thank you for the key to go in.
> And I found a statement in the FAQ ragarding PETSc which says you should use
> OpenMPI wrapper compiler. I use wxWidgets library and try to compile with the
> wrapper.
>
> 2015/10/27 23:56、Nathan Hjelm <hje...@lanl.gov> のメール：
>
>>
>> I have seen hangs when the tcp component is in use. If you are running
>> on a single machine running with mpirun -mca btl self,vader.
>>
>> -Nathan
>>
>> On Mon, Oct 26, 2015 at 09:17:20PM -0600, ABE Hiroshi wrote:
>>> Dear All,
>>>
>>> I have a multithread-ed program and as a next step it is reconstructing to
>>> MPI program. The code is to be MPI / Multithread hybrid one.
>>>
>>> The code proceeds MPI-routines as:
>>>
>>> 1. Send data by MPI_Isend with exlusive tag numbers to the other node.
>>> This is done in ONE master thread.
>>> 2. Receive the sent data by MPI_Irecv in several threads (usually the same
>>> as the number of CPU core) and do their jobs.
>>>
>>> There is one main thread (main process) and one master thread and several
>>> working threads in the code. MPI_Isend is called in the master thread.
>>> MPI_Irecv is called in the working threads.
>>>
>>> My problem is MPI_Wait stalls after calling MPI_Isend. MPI_Wait is called
>>> just after MPI_Isend. Several time the routines get through, but after
>>> sending several data MPI_Wait stalls.
>>>
>>> Using Xcode debugger, the loop with c->c_signaled at line 70 of
>>> opal_conidition_wait (opal/threads/condition.h) never escape.
>>>
>>> I guess I would make something wrong. I would like to know how to find the
>>> problem.
>>> I would be obliged if you'd point the solution or the next direction to be
>>> investigated for debugging.
>>>
>>> My Environment : OSX 10.9.5, Apple LLVM 6.0 (LLVM 3.5svn), OpenMPI 1.10.0
>>> The thread is wxThread from wxWidgets Library (3.0.2) which is a wrapper
>>> of pthread.
>>>
>>> OpenMPI is configure-ed : --enable-mpi-thread-multiple --enable-debug
>>> --enable-event-debug
>>> Please find the detail (config.log and ompi_info -all) attached in this
>>> mail.
>>>
>>> Thank you very much in advance.
>>>
>>> Sincerely,
ABE Hiroshi
from Tokorozawa, JAPAN
#include <iostream>
#include <mpi.h>
#define bufdim 128
int main(int argc, char *argv[])
{
char buffer[bufdim];
char id_str[32];
// mpi :
MPI::Init(argc,argv);
MPI::Status status;
int size;
int rank;
int tag;
size=MPI::COMM_WORLD.Get_size();
rank=MPI::COMM_WORLD.Get_rank();
tag=0;
if (rank==0) {
printf("%d: we have %d processors\n",rank,size);
int i;
i=1;
for ( ;i<size; ++i) {
sprintf(buffer,"hello %d! ",i);
MPI::COMM_WORLD.Send(buffer,bufdim,MPI::CHAR,i,tag);
}
i=1;
for ( ;i<size; ++i) {
MPI::COMM_WORLD.Recv(buffer,bufdim,MPI::CHAR,i,tag,status);
printf("%d: %s\n",rank,buffer);
}
}
else {
MPI::COMM_WORLD.Recv(buffer,bufdim,MPI::CHAR,0,tag,status);
sprintf(id_str,"processor %d ",rank);
strncat(buffer,id_str,bufdim-1);
strncat(buffer,"reporting for duty\n",bufdim-1);
MPI::COMM_WORLD.Send(buffer,bufdim,MPI::CHAR,0,tag);
}
MPI::Finalize();
return 0;
}
