spmd_utils.F90 compilation error for Serial compilation

[moguo_sun@nasa_gov]

Dear helper,       I got following error when I tried to make after I configured like this.  I can not figure out where I should set spmd issue right. Thanks for your help. $camcfg/configure -fc ifort -dyn fv -hgrid 10x15 -nospmd -nosmp -testmake -j2 >&!  make.out  --------ifort -c  -I. -I/homedir/msun/cam5/cesm1_1_2/models/atm/cam/src/chemistry/pp_trop_mam3 -I/homedir/msun/cam5/cesm1_1_2/models/atm/cam/
src/chemistry/modal_aero -I/homedir/msun/cam5/cesm1_1_2/models/atm/cam/src/chemistry/mozart -I/homedir/msun/cam5/cesm1_1_2/models/atm
/cam/src/chemistry/bulk_aero -I/homedir/msun/cam5/cesm1_1_2/models/atm/cam/src/chemistry/utils -I/homedir/msun/cam5/cesm1_1_2/models/
atm/cam/src/physics/rrtmg -I/homedir/msun/cam5/cesm1_1_2/models/atm/cam/src/physics/rrtmg/ext/rrt
mg_mcica -I/homedir/msun/cam5/cesm1_1_2/models/atm/cam/src/physics/rrtmg/ext/rrtmg_lw -I/homedir/msun/cam5/cesm1_1_2/models/atm/cam/s
rc/physics/rrtmg/ext/rrtmg_sw -I/homedir/msun/cam5/cesm1_1_2/models/atm/cam/src/physics/cam -I/homedir/msun/cam5/cesm1_1_2/models/atm
/cam/src/dynamics/fv -I/homedir/msun/cam5/cesm1_1_2/models/atm/cam/src/utils/pilgrim -I/homedir/msun/cam5/cesm1_1_2/models/atm/cam/sr
c/cpl_mct -I/homedir/msun/cam5/cesm1_1_2/models/atm/cam/src/cpl_share -I/homedir/msun/cam5/cesm1_1_2/models/atm/cam/src/control -I/ho
medir/msun/cam5/cesm1_1_2/models/atm/cam/src/utils -I/homedir/msun/cam5/cesm1_1_2/models/utils/timing -I/homedir/msun/cam5/cesm1_1_2/
models/utils/pio -I/homedir/msun/cam5/cesm1_1_2/models/utils/esmf_wrf_timemgr -I/homedir/msun/cam5/cesm1_1_2/models/drv/driver -I/hom
edir/msun/cam5/cesm1_1_2/models/drv/shr -I/homedir/msun/cam5/cesm1_1_2/models/ocn/docn -I/homedir/msun/cam5/cesm1_1_2/models/ocn/docn
/cpl_mct -I/homedir/msun/cam5/cesm1_1_2/models/lnd/clm/src/cpl_mct -I/homedir/msun/cam5/cesm1_1_2/models/lnd/clm/src/cpl_share -I/hom
edir/msun/cam5/cesm1_1_2/models/lnd/clm/src/main -I/homedir/msun/cam5/cesm1_1_2/models/lnd/clm/src/biogeophys -I/homedir/msun/cam5/ce
sm1_1_2/models/lnd/clm/src/biogeochem -I/homedir/msun/cam5/cesm1_1_2/models/ice/cice/src/drivers/cpl_mct -I/homedir/msun/cam5/cesm1_1
_2/models/ice/cice/src/drivers/cpl_share -I/homedir/msun/cam5/cesm1_1_2/models/ice/cice/src/mpi -I/homedir/msun/cam5/cesm1_1_2/models
/ice/cice/src/source -I/homedir/msun/cam5/cesm1_1_2/models/glc/sglc/cpl_mct -I/homedir/msun/cam5/cesm1_1_2/models/rof/rtm/src/cpl_mct
 -I/homedir/msun/cam5/cesm1_1_2/models/rof/rtm/src/cpl_share -I/homedir/msun/cam5/cesm1_1_2/models/rof/rtm/src/riverroute -I/homedir/
msun/cam5/cesm1_1_2/models/csm_share/shr -I/homedir/msun/cam5/cesm1_1_2/models/csm_share/dshr -I/usr/local/include/  -I/usr/local/inc
lude/ -I/homedir/msun/cam5/work/bld/mct/mct -I/homedir/msun/cam5/work/bld/mct/mpeu -I/homedir/msun/cam5/work/bld/mct/mpi-serial -DNO_
SHR_VMATH -DSEQ_MCT -DNO_C_SIZEOF -DFORTRANUNDERSCORE -DCO2A -DMAXPATCH_PFT=numpft+1 -DLSMLAT=1 -DLSMLON=1 -DPLON=24 -DPLAT=19 -DNUM_
COMP_INST_ATM=1 -DNUM_COMP_INST_LND=1 -DNUM_COMP_INST_OCN=1 -DNUM_COMP_INST_ICE=1 -DNUM_COMP_INST_GLC=1 -DNUM_COMP_INST_ROF=1 -DPLEV=
30 -DPCNST=25 -DPCOLS=16 -DPTRM=1 -DPTRN=1 -DPTRK=1 -DSTAGGERED  -DCCSMCOUPLED -Dcoupled -Dncdf -DNCAT=1 -DNXGLOB=24 -DNYGLOB=19 -DNT
R_AERO=0 -DBLCKX=24 -DBLCKY=19 -DMXBLCKS=1  -DMODAL_AERO -DMODAL_AERO_3MODE  -D_USEBOX -D_NETCDF   -DLINUX -DCPRINTEL -static -ftz -c
onvert big_endian -fp-model precise  -O2 -DHIDE_MPI -D_MPISERIAL -DNO_MPIMOD -DNO_MPI2 -DNO_SIZEOF  /homedir/msun/cam5/cesm1_1_2/mode
ls/atm/cam/src/utils/spmd_utils.F90
/homedir/msun/cam5/cesm1_1_2/models/atm/cam/src/utils/spmd_utils.F90(51): error #6418: This name has already been assigned a data typ
e.   [MPI_STATUS_IGNORE]
   integer :: mpi_status_ignore     ! Needs to be defined in mpi-serial
--------------^
compilation aborted for /homedir/msun/cam5/cesm1_1_2/models/atm/cam/src/utils/spmd_utils.F90 (code 1)
gmake: *** [spmd_utils.o] Error 1
gmake: *** Waiting for unfinished jobs....
/homedir/msun/cam5/cesm1_1_2/models/ice/cice/src/source/ice_calendar.F90(374): (col. 15) remark: LOOP WAS VECTORIZED.
/homedir/msun/cam5/cesm1_1_2/models/ice/cice/src/source/ice_calendar.F90(376): (col. 15) remark: LOOP WAS VECTORIZED.
/homedir/msun/cam5/cesm1_1_2/models/ice/cice/src/source/ice_calendar.F90(385): (col. 12) remark: LOOP WAS VECTORIZED.
/homedir/msun/cam5/cesm1_1_2/models/ice/cice/src/source/ice_calendar.F90(390): (col. 18) remark: LOOP WAS VECTORIZED.
/homedir/msun/cam5/cesm1_1_2/models/ice/cice/src/source/ice_calendar.F90(392): (col. 18) remark: LOOP WAS VECTORIZED.
/homedir/msun/cam5/cesm1_1_2/models/ice/cice/src/source/ice_calendar.F90(403): (col. 18) remark: LOOP WAS VECTORIZED.
------------------------- Moguo

 

[eaton]

Check whether /homedir/msun/cam5/work/bld/mct/mpi-serial/mpif.h  contains the name mpi_status_ignore.  If not (which is what I expect) then you need to figure out where the mpif.h that does contain mpi_status_ignore is coming from.  The CAM serial build expects to use mct/mpi-serial/mpif.h which should not contain that name.

 

[moguo_sun@nasa_gov]

Eaton,
       Thanks. You are right. But I still can't find the problem. I search all the *.h files and I found no mpi_status_ignore. Could you suggest where it could come from? Thanks. --------grep -i mpi_status_ignore ./cesm1_1_2/models/atm/cam/src/advection/slt/hadvtest.h
grep -i mpi_status_ignore ./cesm1_1_2/models/atm/cam/src/dynamics/eul/comsta.h
grep -i mpi_status_ignore ./cesm1_1_2/models/atm/cam/src/dynamics/eul/parslt.h
grep -i mpi_status_ignore ./cesm1_1_2/models/atm/cam/src/dynamics/fv/comsta.h
grep -i mpi_status_ignore ./cesm1_1_2/models/atm/cam/src/dynamics/homme/share/interface.h
grep -i mpi_status_ignore ./cesm1_1_2/models/atm/cam/src/dynamics/homme/share/timer.h
grep -i mpi_status_ignore ./cesm1_1_2/models/atm/cam/src/dynamics/sld/comsta.h
grep -i mpi_status_ignore ./cesm1_1_2/models/atm/cam/src/physics/carma/base/carma_globaer.h
grep -i mpi_status_ignore ./cesm1_1_2/models/atm/cam/src/physics/cosp/cosp_defs.h
grep -i mpi_status_ignore ./cesm1_1_2/models/atm/cam/src/utils/dtypes.h
grep -i mpi_status_ignore ./cesm1_1_2/models/atm/cam/src/utils/pilgrim/debug.h
grep -i mpi_status_ignore ./cesm1_1_2/models/atm/cam/src/utils/pilgrim/mlp_ptr.h
grep -i mpi_status_ignore ./cesm1_1_2/models/atm/cam/src/utils/pilgrim/pilgrim.h
grep -i mpi_status_ignore ./cesm1_1_2/models/csm_share/shr/shr_isnan.h
grep -i mpi_status_ignore ./cesm1_1_2/models/csm_share/unit_testers/config.h
grep -i mpi_status_ignore ./cesm1_1_2/models/lnd/clm/tools/cprnc/dtypes.h
grep -i mpi_status_ignore ./cesm1_1_2/models/utils/mct/mpi-serial/list.h
grep -i mpi_status_ignore ./cesm1_1_2/models/utils/mct/mpi-serial/listops.h
grep -i mpi_status_ignore ./cesm1_1_2/models/utils/mct/mpi-serial/listP.h
grep -i mpi_status_ignore ./cesm1_1_2/models/utils/mct/mpi-serial/mpi.h
grep -i mpi_status_ignore ./cesm1_1_2/models/utils/mct/mpi-serial/mpif.master.h
grep -i mpi_status_ignore ./cesm1_1_2/models/utils/mct/mpi-serial/mpif.real4double8.h
grep -i mpi_status_ignore ./cesm1_1_2/models/utils/mct/mpi-serial/mpif.real8double16.h
grep -i mpi_status_ignore ./cesm1_1_2/models/utils/mct/mpi-serial/mpif.real8double8.h
grep -i mpi_status_ignore ./cesm1_1_2/models/utils/mct/mpi-serial/mpiP.h
grep -i mpi_status_ignore ./cesm1_1_2/models/utils/pio/dtypes.h
grep -i mpi_status_ignore ./cesm1_1_2/models/utils/timing/gptl.h
grep -i mpi_status_ignore ./cesm1_1_2/models/utils/timing/private.h
grep -i mpi_status_ignore ./work/bld/mct/mpi-serial/mpif.h
grep -i mpi_status_ignore ./work/bld/mct/config.h
grep -i mpi_status_ignore ./work/bld/config.h
----------------- Thanks,Moguo    

 

[moguo_sun@nasa_gov]

Hi, Eaton,     I searched all files , not only *.h and I still could not find where it is defined. So, I go ahead to commented that line out in smpd_utils.F90 where it causes problem. The compilation goes well beyond that point and shows no error. I don't like this idea, but it seems only way to avoid the error right now.     Please let me know if you have concerns for my doing. Thanks.  Moguo 

 

[eaton]

My guess is that the compiler is being invoked via a wrapper script which is responsible for adding mpi related paths. 

 

[eaton]

This is a reasonable workaround.