[ncl-install] Building NCAR Graphics from Source: Bus Error on SiCortex SC648

[Mary Haley]

Hi Steve,

This is a new system to me, so I'm treading on unfamiliar ground here.

In the $NCARG/common/src/libncarg_c directory are the gbytes/sbytes
routines.  There are two ways to get the code for these routines:
through "gbytes.f"/"sbytes.f" or through "gsbytes.c".

For a 64-bit system, you want to use gbytes.f and sbytes.f and *not*
gsbytes.c.

So, hopefully the solution to your problem is as simple as editing the
yMakefile in $NCARG/common/src/libncarg_c, and adding recognition of
your machine to the "#elif" commands you see in here.

For example (and I don't know the correction preprocessor word to use
here; I'm just using "mips64" as an example):

#elif defined(linux) && defined(mips64)
EXCSRCS = logic32.c
EXFSRCS = gbytes.f sbytes.f
EXOBJS  = sbytes.o gbytes.o logic32.o

You will need to know what to use for "mips64". It might be
"mips64", or even something like "_mips64".

Once you do this, run "make Everything" again to make sure these
changes take effect, and also make sure that sbytes.o and gbytes.o are
getting compiled into the "libncarg_c.a" library, and not
"gsbytes.c". This is important!

I assume this is a big endian machine? If not, then there may
be some additional steps you need. Let me know.

--Mary

On Fri, 4 Apr 2008, Steven G Decker wrote:

> Hi there,
>
> I am attempting to build NCAR Graphics from source on the SiCortex SC648 
> using the Pathscale suite of compilers.  I get most of the files listed at 
> http://www.ncl.ucar.edu/Download/files/NCARG_files.txt, but not all, and I 
> get a slew of undefined references when trying ncargex cpex08.
>
> Looking through the output of the build process, I see the first error here:
> pathf95  -O    -o fontc cfaamn.o  cfrdln.o  cfwrit.o  ffgttk.o  ffinfo.o 
> ffphol.o  ffppkt.o  ffprcf.o ffprsa.o  fftbkd.o  fftkin.o  sffndc.o sfgtin.o 
> sfgtkw.o  sfprcf.o  sfskbk.o sftbkd.o -L../../.././common/src/libncarg_c 
> -lncarg_c -L/home/decker/local/ncarg/lib -L/usr/lib64
>        Processing fontcap font1
> make[4]: *** [font1] Bus error
>
> When I run the tlocal test program, I get this output:
> 1TEST PACKAGE EXECUTING.  THIS LINE SHOULD BE PRINTED.
> 0WE WILL FIRST TEST THE INFORMATION YOU HAVE PROVIDED IN CERTAIN DATA
> STATEMENTS AT THE BEGINNING OF THIS PROGRAM.  IF ANY ERROR IS FOUND,
> EXECUTION WILL TERMINATE--NO ROUTINES WILL BE TESTED.
> 0YOU HAVE PROVIDED THE FOLLOWING CONSTANTS.
>
>      NUMBER OF BITS IN A WORD                                   32
>      NUMBER OF BITS IN AN INTEGER ASSIGNMENT                    32
>      NUMBER OF BITS IN AN INTEGER COMPARISON                    32
>      THE DIMENSION OF THE ARRAY MASK0                           32
>      THE DIMENSION OF THE ARRAY MASK1                           33
>      THE DIMENSION OF THE ARRAY ISCR                           190
> 0THE ABOVE DATA APPEARS TO BE CORRECT.  WE WILL NOW BEGIN TESTING THE
> SUPPORT ROUTINES YOU HAVE PROVIDED FOR USE WITH THE PLOT PACKAGE.
> 0A MESSAGE WILL BE PRINTED BEFORE EACH TEST BEGINS, WHENEVER AN ERROR IS
> FOUND, AND AT THE SUCCESSFUL CONCLUSION OF A TEST.  A SUMMARY OF THE
> TEST RESULTS WILL BE PRINTED AT THE END.  NOTE THAT AN ERROR FOUND IN
> ONE ROUTINE MAY SUSPEND TESTING OF ONE OR MORE SUBSEQUENT ROUTINES.
> 1NOW TESTING IOR.
> 0FINISHED TESTING IOR.  IOR APPARENTLY WORKS.
> 1NOW TESTING IAND.
> 0FINISHED TESTING IAND.  IAND APPARENTLY WORKS.
> 1NOW TESTING ISHIFT.
> 0FINISHED TESTING ISHIFT.  ISHIFT APPARENTLY WORKS.
> 1NOW TESTING GBYTES.
>
>
> **** Segmentation fault!  Fault address: (nil)
>
> This is likely to have been caused by either a null pointer dereference or a 
> general protection fault.
> Aborted
>
> So it appears something is wrong with GBYTES.  Checking the output of the 
> i1mach and r1mach functions showed no obvious errors, as the values match 
> what I get from the Fortran intrinsics like HUGE, TINY, etc.  I also thought 
> I modified the master yMakefile appropriately as well. (The system was 
> undetected at first, so I had to make modifications.)
>
> Thus, I'm at a loss at this point.  Please let me know if you can offer any 
> help or if you need more information.
>
> Here's the output of uname -a:
> Linux sca-m0n8.scsystem 2.6.15-sc-lustre-perfmon #1 SMP Thu Dec 20 18:10:56 
> EST 2007 mips64 SiCortex ICE9A V1.0 FPU V0.1 SiCortex SC-1000 GNU/Linux
>
> Thanks,
> Steve
>
> _______________________________________________
> ncl-install mailing list
> ncl-install at ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/ncl-install
>