maxele.63 [adcirc to netcdf] segmentation error about asgs HOT 5 CLOSED

Which type of compiler are you using? I've encountered this error with Intel on certain machines, but it appears to be a compiler issue an not necessarily a code issue. The fix with the Intel compilers on one of my machines was accomplished by the addition of the compiler flag -heap-arrays 8192 where 8192 is the machine's stack size (ulimit -s). This seems to clear up the issue because Intel does not automatically move implicitly created arrays to the heap, leading to a segfault. These implicit arrays are created inside the netCDF routines. The fix needed to be applied to both the netCDF Fortran library and the adcirc2netcdf.f90 code itself.

from asgs.

I am using gfortran compiler , I tried compiling with intel but it gives me a error and I couldn't figure out yet.

here is the error

:~/Documents/KalpanaTesting_DONE/script/asgs/output$ make adcirc2netcdf compiler=intel MACHINENAME=poseidon NETCDF=enable NETCDF4=enable NETCDF4_COMPRESSION=enable
ifort -cpp -i-dynamic -DASGSNETCDF -DASGSNETCDF -DHAVE_NETCDF4 -DNETCDF_CAN_DEFLATE -I. -I/home/fhrl/Documents/Surge/scripts/netcdf/include -L/home/fhrl/Documents/Surge/scripts/netcdf/lib -c logging.f90 -lnetcdff
ifort: command line remark #10148: option '-i-dynamic' not supported
logging.f90(222): error #7013: This module file was not generated by any release of this compiler. [NETCDF]
use netcdf
----^
logging.f90(420): error #7013: This module file was not generated by any release of this compiler. [NETCDF]
use netcdf
----^
logging.f90(425): error #6404: This name does not have a type, and must have an explicit type. [NF90_NOERR]
if(ncstatus.ne.nf90_noerr)then
---------------^
logging.f90(426): error #6404: This name does not have a type, and must have an explicit type. [NF90_STRERROR]
write(,'(a,a)') "ERROR: ",trim(nf90_strerror(ncstatus))
-----------------------------------^
logging.f90(426): error #6362: The data types of the argument(s) are invalid. [TRIM]
write(,'(a,a)') "ERROR: ",trim(nf90_strerror(ncstatus))
-----------------------------------^
compilation aborted for logging.f90 (code 1)
makefile:170: recipe for target 'logging.mod' failed
make: *** [logging.mod] Error 1

from asgs.

Unfortunately, I'm not familiar enough with gfortran to know if it has some of the same heap allocation issues as Intel, though I don't think it does. The error from your intel compile is because your netCDF library was compiled with gfortran. You may try to debug your original error using gdb, which is the GNU debugger. It should give you some additional information that can help you see exactly where the code is going wrong. You should include -g -fbacktrace when compiling both netCDF and adcirc2netcdf to generate debugging information that can be used by gdb.

from asgs.

Thanks Zach
I will try to recompile fortran library with intel and will check if the problem persists

from asgs.

Hello @ArslaanK I concur with @zcobell about the intel compile. On the issue of the segmentation fault, I think I found and fixed this a while back in the 2014stable branch. I just tried to reproduce this issue with some test files from the shinnecock inlet test problem, and everything went as expected. I will close this issue, but we can reopen it if things are still not running as expected for you.

from asgs.