Please try the patch from this post: http://www.open-mpi.org/community/lists/users/2015/11/28030.php
2015-11-10 9:27 GMT+01:00 Lisandro Dalcin <dalc...@gmail.com>: > This is with 1.10.1, configured with flags --enable-debug > --enable-mem-debug. Use the attached test cases to reproduce yourself. > > * Iallgatherv() prints a warning, but otherwise seems to succeed (for > some reason, running under valgrind hangs, so I cannot test further): > > $ mpicc iallgatherv.c > $ ./a.out > MPI Error in MPI_Pack_size() (0:0) > > * Iscatterv() and Igatherv() both prints the pack-size error and fail: > > $ mpicc iscatterv.c > $ ./a.out > MPI Error in MPI_Pack_size() (0:0) > Error in NBC_Copy() (49) > [kw2060:25204] *** An error occurred in MPI_Iscatterv > [kw2060:25204] *** reported by process [140736503742465,0] > [kw2060:25204] *** on communicator MPI_COMM_SELF > [kw2060:25204] *** MPI_ERR_SIZE: invalid size > [kw2060:25204] *** MPI_ERRORS_ARE_FATAL (processes in this > communicator will now abort, > [kw2060:25204] *** and potentially your MPI job) > > $ mpicc igatherv.c > $ ./a.out > MPI Error in MPI_Pack_size() (0:0) > Error in NBC_Copy() (49) > [kw2060:25212] *** An error occurred in MPI_Igatherv > [kw2060:25212] *** reported by process [140727914332161,0] > [kw2060:25212] *** on communicator MPI_COMM_SELF > [kw2060:25212] *** MPI_ERR_SIZE: invalid size > [kw2060:25212] *** MPI_ERRORS_ARE_FATAL (processes in this > communicator will now abort, > [kw2060:25212] *** and potentially your MPI job) > > > > -- > Lisandro Dalcin > ============ > Research Scientist > Computer, Electrical and Mathematical Sciences & Engineering (CEMSE) > Numerical Porous Media Center (NumPor) > King Abdullah University of Science and Technology (KAUST) > http://numpor.kaust.edu.sa/ > > 4700 King Abdullah University of Science and Technology > al-Khawarizmi Bldg (Bldg 1), Office # 4332 > Thuwal 23955-6900, Kingdom of Saudi Arabia > http://www.kaust.edu.sa > > Office Phone: +966 12 808-0459 > > _______________________________________________ > devel mailing list > de...@open-mpi.org > Subscription: http://www.open-mpi.org/mailman/listinfo.cgi/devel > Link to this post: > http://www.open-mpi.org/community/lists/devel/2015/11/18351.php > -- Kind regards Nick
