mpich hanging on startup
I'm got a strange situation - I'm trying to build hdf5 1.8.16 for Fedora. The Fedora builders do not have network access (/etc/resolv.conf set to "nameserver 127.0.0.1" and no local nameserver) for security purposes. The hdf5 t_mpi parallel test hangs on launch with mpich 3.2, but only on the arm builders. The processes seem to deadlock in the MPI code, each blocking in poll(), presumably waiting for communication from the other process. here are gdb stack traces of a two process t_mpi job: (gdb) bt #0 0xb69176ec in poll () from /lib/libc.so.6 #1 0xb6b4b178 in poll (__timeout=-1, __nfds=<optimized out>, __fds=<optimized out>) at /usr/include/bits/poll2.h:46 #2 MPIDU_Sock_wait (sock_set=0x7f57b3e0, millisecond_timeout=millisecond_timeout@entry=-1, eventp=0xb6fc6000, eventp@entry=0xbeffe054) at src/mpid/common/sock/poll/sock_wait.i:123 #3 0xb6b30960 in MPIDI_CH3i_Progress_wait (progress_state=0xbeffe0a0) at src/mpid/ch3/channels/sock/src/ch3_progress.c:221 #4 MPIDI_CH3I_Progress (blocking=blocking@entry=1, state=state@entry=0xbeffe0a0) at src/mpid/ch3/channels/sock/src/ch3_progress.c:962 #5 0xb6abfeac in MPIC_Wait (request_ptr=0xb6c0a180 <MPID_Request_direct>, errflag=errflag@entry=0xbeffe330) at src/mpi/coll/helper_fns.c:225 #6 0xb6ac01cc in MPIC_Recv (buf=buf@entry=0x7f5a0820, count=count@entry=65, datatype=datatype@entry=1275069445, source=<optimized out>, tag=tag@entry=11, comm_ptr=0xb6bf0908 <MPID_Comm_direct>, comm_ptr@entry=0x7f586250, status=0xbeffe190, status@entry=0xb6ffec88 <__stack_chk_guard>, errflag=0xbeffe330, errflag@entry=0x7f586c80) at src/mpi/coll/helper_fns.c:355 #7 0xb6a0326c in MPIR_Reduce_binomial (errflag=0x7f586c80, comm_ptr=0x7f586250, root=-1231016932, op=-1090526356, datatype=1275069445, count=65, recvbuf=0xb6bf0908 <MPID_Comm_direct>, sendbuf=<optimized out>) at src/mpi/coll/reduce.c:181 #8 MPIR_Reduce_intra (sendbuf=sendbuf@entry=0xffffffff, recvbuf=recvbuf@entry=0xbeffe36c, count=count@entry=65, datatype=datatype@entry=1275069445, op=1476395014, op@entry=-1228891236, root=root@entry=0, comm_ptr=0xb6bf0908 <MPID_Comm_direct>, comm_ptr@entry=0x58000006, errflag=errflag@entry=0xbeffe330) at src/mpi/coll/reduce.c:874 #9 0xb6a02c1c in MPIR_Reduce_impl (sendbuf=0xffffffff, sendbuf@entry=0xb6bf1288 <MPID_Comm_builtin>, recvbuf=recvbuf@entry=0xbeffe36c, count=count@entry=65, datatype=datatype@entry=1275069445, op=1476395014, op@entry=-1228991864, root=root@entry=0, comm_ptr=comm_ptr@entry=0xb6bf0908 <MPID_Comm_direct>, errflag=0xbeffe330, errflag@entry=0xb6ffec88 <__stack_chk_guard>) at src/mpi/coll/reduce.c:1068 #10 0xb69f7b64 in MPIR_Allreduce_intra (sendbuf=0xb6bf1288 <MPID_Comm_builtin>, sendbuf@entry=0xffffffff, recvbuf=recvbuf@entry=0xbeffe36c, count=count@entry=65, datatype=datatype@entry=1275069445, op=op@entry=1476395014, comm_ptr=0xb6bf1288 <MPID_Comm_builtin>, comm_ptr@entry=0xb6acdd88 <MPIR_Get_contextid_sparse_group+824>, errflag=errflag@entry=0xbeffe330) at src/mpi/coll/allreduce.c:234 #11 0xb69f9414 in MPIR_Allreduce_impl (sendbuf=sendbuf@entry=0xffffffff, recvbuf=recvbuf@entry=0xbeffe36c, count=count@entry=65, datatype=1275069445, op=1476395014, comm_ptr=comm_ptr@entry=0xb6bf1288 <MPID_Comm_builtin>, errflag=0xbeffe330, errflag@entry=0xbeffe4c8) at src/mpi/coll/allreduce.c:763 #12 0xb6acdd88 in MPIR_Get_contextid_sparse_group ( comm_ptr=comm_ptr@entry=0xb6bf1288 <MPID_Comm_builtin>, group_ptr=0xb6998228 <__libc_malloc_initialized>, group_ptr@entry=0x0, tag=-1090526416, context_id=0xfffff000, context_id@entry=0xbeffe4c0, ignore_id=ignore_id@entry=0) at src/mpi/comm/contextid.c:496 #13 0xb6ace5a4 in MPIR_Get_contextid_sparse ( comm_ptr=comm_ptr@entry=0xb6bf1288 <MPID_Comm_builtin>, context_id=context_id@entry=0xbeffe4c0, ignore_id=ignore_id@entry=0) at src/mpi/comm/contextid.c:298 #14 0xb6acc14c in MPIR_Comm_copy ( comm_ptr=comm_ptr@entry=0xb6bf1288 <MPID_Comm_builtin>, size=2, outcomm_ptr=outcomm_ptr@entry=0xbeffe538) at src/mpi/comm/commutil.c:736 #15 0xb6a50a48 in MPIR_Comm_dup_impl ( comm_ptr=comm_ptr@entry=0xb6bf1288 <MPID_Comm_builtin>, newcomm_ptr=0xbeffe538, newcomm_ptr@entry=0xbeffe530) at src/mpi/comm/comm_dup.c:56 #16 0xb6a50cf0 in PMPI_Comm_dup (comm=1140850688, newcomm=0xbeffe574, newcomm@entry=0xbeffe56c) at src/mpi/comm/comm_dup.c:161 #17 0xb6daa0dc in H5FD_mpi_comm_info_dup (comm=<optimized out>, info=469762048, comm_new=comm_new@entry=0x7f5a0810, info_new=info_new@entry=0x7f5a0814) at ../../src/H5FDmpi.c:271 #18 0xb6daaa80 in H5FD_mpio_fapl_copy (_old_fa=0xbeffe64c) at ../../src/H5FDmpio.c:789 #19 0xb6d9e0ec in H5FD_pl_copy (copied_pl=0xbeffe5d0, old_pl=0xbeffe64c, pl_size=<optimized out>, copy_func=<optimized out>) at ../../src/H5FD.c:625 #20 H5FD_fapl_copy (copied_fapl=0xbeffe5d0, old_fapl=0xbeffe64c, driver_id=<optimized out>) at ../../src/H5FD.c:800 #21 H5FD_fapl_open (plist=plist@entry=0x7f5a06c8, driver_id=-1224955460, driver_id@entry=134217729, driver_info=driver_info@entry=0xbeffe64c) at ../../src/H5FD.c:753 #22 0xb6e6aea4 in H5P_set_driver (plist=plist@entry=0x7f5a06c8, new_driver_id=134217729, new_driver_info=0xbeffe64c, new_driver_info@entry=0xbeffe644) at ../../src/H5Pfapl.c:633 #23 0xb6dacec4 in H5Pset_fapl_mpio (fapl_id=<optimized out>, comm=1140850688, info=469762048) at ../../src/H5FDmpio.c:353 #24 0x7f5570e8 in parse_options (argv=<optimized out>, argc=<optimized out>) at ../../testpar/t_mpi.c:1041 #25 main (argc=0, argv=0x0) at ../../testpar/t_mpi.c:1106 (gdb) up 2 #2 MPIDU_Sock_wait (sock_set=0x7f57b3e0, millisecond_timeout=millisecond_timeout@entry=-1, eventp=0xb6fc6000, eventp@entry=0xbeffe054) at src/mpid/common/sock/poll/sock_wait.i:123 123 n_fds = poll(sock_set->pollfds, sock_set->poll_array_elems, (gdb) print millisecond_timeout $1 = -1 (gdb) print *sock_set $2 = {id = 0, starting_elem = 0, poll_array_sz = 32, poll_array_elems = 1, pollfds = 0x7f57b588, pollinfos = 0x7f57b690, eventq_head = 0x0, eventq_tail = 0x0, pollfds_active = 0x0, pollfds_updated = 0, wakeup_posted = 0, intr_sock = 0x0, intr_fds = {-1, -1}} (gdb) print *sock_set->pollfds $3 = {fd = 10, events = 1, revents = 0} #0 0xb69176ec in poll () from /lib/libc.so.6 #1 0xb6b4b178 in poll (__timeout=-1, __nfds=<optimized out>, __fds=<optimized out>) at /usr/include/bits/poll2.h:46 #2 MPIDU_Sock_wait (sock_set=0x7f57b3e0, millisecond_timeout=millisecond_timeout@entry=-1, eventp=0x7f5e5700, eventp@entry=0xbeffe074) at src/mpid/common/sock/poll/sock_wait.i:123 #3 0xb6b30960 in MPIDI_CH3i_Progress_wait (progress_state=0xbeffe0c0) at src/mpid/ch3/channels/sock/src/ch3_progress.c:221 #4 MPIDI_CH3I_Progress (blocking=blocking@entry=1, state=state@entry=0xbeffe0c0) at src/mpid/ch3/channels/sock/src/ch3_progress.c:962 #5 0xb6abfeac in MPIC_Wait (request_ptr=0xb6c0a180 <MPID_Request_direct>, errflag=errflag@entry=0xbeffe330) at src/mpi/coll/helper_fns.c:225 #6 0xb6ac000c in MPIC_Send (buf=buf@entry=0xb6bf0908 <MPID_Comm_direct>, count=count@entry=65, datatype=datatype@entry=1275069445, dest=<optimized out>, tag=tag@entry=11, comm_ptr=0xb6bf0908 <MPID_Comm_direct>, comm_ptr@entry=0x7f586250, errflag=0xbeffe330, errflag@entry=0x7f586c80) at src/mpi/coll/helper_fns.c:302 #7 0xb6a03438 in MPIR_Reduce_binomial (errflag=0x7f586c80, comm_ptr=0x7f586250, root=-1231016932, op=-1090526356, datatype=1275069445, count=65, recvbuf=0xb6bf0908 <MPID_Comm_direct>, sendbuf=<optimized out>) at src/mpi/coll/reduce.c:210 #8 MPIR_Reduce_intra (sendbuf=sendbuf@entry=0xbeffe36c, recvbuf=recvbuf@entry=0x0, count=count@entry=65, datatype=datatype@entry=1275069445, op=1476395014, op@entry=-1228891236, root=root@entry=0, comm_ptr=0xb6bf0908 <MPID_Comm_direct>, comm_ptr@entry=0x58000006, errflag=errflag@entry=0xbeffe330) at src/mpi/coll/reduce.c:874 #9 0xb6a02c1c in MPIR_Reduce_impl (sendbuf=sendbuf@entry=0xbeffe36c, recvbuf=recvbuf@entry=0x0, count=count@entry=65, datatype=datatype@entry=1275069445, op=1476395014, op@entry=-1228991864, root=root@entry=0, comm_ptr=comm_ptr@entry=0xb6bf0908 <MPID_Comm_direct>, errflag=0xbeffe330, errflag@entry=0xb6ffec88 <__stack_chk_guard>) at src/mpi/coll/reduce.c:1068 #10 0xb69f8b94 in MPIR_Allreduce_intra ( sendbuf=0xb6bf1288 <MPID_Comm_builtin>, sendbuf@entry=0xffffffff, recvbuf=recvbuf@entry=0xbeffe36c, count=count@entry=65, datatype=datatype@entry=1275069445, op=op@entry=1476395014, comm_ptr=0xb6bf1288 <MPID_Comm_builtin>, comm_ptr@entry=0xb6acdd88 <MPIR_Get_contextid_sparse_group+824>, errflag=errflag@entry=0xbeffe330) at src/mpi/coll/allreduce.c:226 #11 0xb69f9414 in MPIR_Allreduce_impl (sendbuf=sendbuf@entry=0xffffffff, recvbuf=recvbuf@entry=0xbeffe36c, count=count@entry=65, datatype=1275069445, op=1476395014, comm_ptr=comm_ptr@entry=0xb6bf1288 <MPID_Comm_builtin>, errflag=0xbeffe330, errflag@entry=0xbeffe4c8) at src/mpi/coll/allreduce.c:763 #12 0xb6acdd88 in MPIR_Get_contextid_sparse_group ( comm_ptr=comm_ptr@entry=0xb6bf1288 <MPID_Comm_builtin>, group_ptr=0xb6998228 <__libc_malloc_initialized>, group_ptr@entry=0x0, tag=-1090526416, context_id=0xfffff000, context_id@entry=0xbeffe4c0, ignore_id=ignore_id@entry=0) at src/mpi/comm/contextid.c:496 #13 0xb6ace5a4 in MPIR_Get_contextid_sparse ( comm_ptr=comm_ptr@entry=0xb6bf1288 <MPID_Comm_builtin>, context_id=context_id@entry=0xbeffe4c0, ignore_id=ignore_id@entry=0) at src/mpi/comm/contextid.c:298 #14 0xb6acc14c in MPIR_Comm_copy ( comm_ptr=comm_ptr@entry=0xb6bf1288 <MPID_Comm_builtin>, size=2, outcomm_ptr=outcomm_ptr@entry=0xbeffe538) at src/mpi/comm/commutil.c:736 #15 0xb6a50a48 in MPIR_Comm_dup_impl ( comm_ptr=comm_ptr@entry=0xb6bf1288 <MPID_Comm_builtin>, newcomm_ptr=0xbeffe538, newcomm_ptr@entry=0xbeffe530) at src/mpi/comm/comm_dup.c:56 #16 0xb6a50cf0 in PMPI_Comm_dup (comm=1140850688, newcomm=0xbeffe574, newcomm@entry=0xbeffe56c) at src/mpi/comm/comm_dup.c:161 #17 0xb6daa0dc in H5FD_mpi_comm_info_dup (comm=<optimized out>, info=469762048, comm_new=comm_new@entry=0x7f5a0810, info_new=info_new@entry=0x7f5a0814) at ../../src/H5FDmpi.c:271 #18 0xb6daaa80 in H5FD_mpio_fapl_copy (_old_fa=0xbeffe64c) at ../../src/H5FDmpio.c:789 #19 0xb6d9e0ec in H5FD_pl_copy (copied_pl=0xbeffe5d0, old_pl=0xbeffe64c, pl_size=<optimized out>, copy_func=<optimized out>) at ../../src/H5FD.c:625 #20 H5FD_fapl_copy (copied_fapl=0xbeffe5d0, old_fapl=0xbeffe64c, driver_id=<optimized out>) at ../../src/H5FD.c:800 #21 H5FD_fapl_open (plist=plist@entry=0x7f5a06c8, driver_id=-1224955460, driver_id@entry=134217729, driver_info=driver_info@entry=0xbeffe64c) at ../../src/H5FD.c:753 #22 0xb6e6aea4 in H5P_set_driver (plist=plist@entry=0x7f5a06c8, new_driver_id=134217729, new_driver_info=0xbeffe64c, new_driver_info@entry=0xbeffe644) at ../../src/H5Pfapl.c:633 #23 0xb6dacec4 in H5Pset_fapl_mpio (fapl_id=<optimized out>, comm=1140850688, info=469762048) at ../../src/H5FDmpio.c:353 #24 0x7f5570e8 in parse_options (argv=<optimized out>, argc=<optimized out>) at ../../testpar/t_mpi.c:1041 #25 main (argc=0, argv=0x0) at ../../testpar/t_mpi.c:1106 (gdb) up #2 MPIDU_Sock_wait (sock_set=0x7f57b3e0, millisecond_timeout=millisecond_timeout@entry=-1, eventp=0x7f5e5700, eventp@entry=0xbeffe074) at src/mpid/common/sock/poll/sock_wait.i:123 123 n_fds = poll(sock_set->pollfds, sock_set->poll_array_elems, (gdb) list 118 just use the same code as above. Otherwise, use 119 multithreaded code (and we don't then need the 120 MPIU_THREAD_CHECK_BEGIN/END macros) */ 121 if (!MPIR_ThreadInfo.isThreaded) { 122 MPIDI_FUNC_ENTER(MPID_STATE_POLL); 123 n_fds = poll(sock_set->pollfds, sock_set->poll_array_elems, 124 millisecond_timeout); 125 MPIDI_FUNC_EXIT(MPID_STATE_POLL); 126 } 127 else (gdb) print *sock_set $2 = {id = 0, starting_elem = 0, poll_array_sz = 32, poll_array_elems = 1, pollfds = 0x7f57b588, pollinfos = 0x7f57b690, eventq_head = 0x0, eventq_tail = 0x0, pollfds_active = 0x0, pollfds_updated = 0, wakeup_posted = 0, intr_sock = 0x0, intr_fds = {-1, -1}} (gdb) print millisecond_timeout $3 = -1 (gdb) print *sock_set->pollfds $4 = {fd = 6, events = 1, revents = 0} Is there anything else that might be useful in tracking this down? -- Orion Poplawski Technical Manager 303-415-9701 x222 NWRA, Boulder/CoRA Office FAX: 303-415-9702 3380 Mitchell Lane [email protected] Boulder, CO 80301 http://www.nwra.com _______________________________________________ discuss mailing list [email protected] To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss
participants (1)
-
Orion Poplawski