[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: 20040220: "make test" problem on Linux x86_64 system



Jan,

>To: address@hidden
>From: Jan Fischer <address@hidden>
>Subject: netcdf problems.....
>Organization: UCAR/Unidata
>Keywords: 200402201510.i1KFAhrV012277

The above message contained the following:

> I do have problems compiling and installing netCDF on my Dual Opteron 
> with SuSE Linux 9.0 professional, using Portland compiler.
> Perhaps you can help me....

We do not have that system here, so we can't duplicate your problem.

> The first errors I found are in 'config.log',

That's to be expected.  Errors in that file are only used to explain an
error in the file "configure.log".

> the next are in 'test.log'
> It was possible to create binaries, but they didn't work...
> 
> Here is the report according to the "install.html":
> 
> Best regards and many thanks, Jan.
> 
> -- 
> ____________________________________________
> Jan Fischer
> IMPULSE
> Institut für Chemie und Biologie des Meeres (ICBM)
> Carl-von-Ossietzky-Str. 9-11
> Postfach 25 03
> 26111 Oldenburg
> 
> Tel.:  0441-36116-258
> Fax:  0441-36116-255
> Mail: address@hidden
> Web: http://www.icbm.de/impulse
> 
> 
> 
> 
> A.
> _______________________________________________________________________________________________
> Linux jafi 2.4.21-102-smp #1 SMP Wed Sep 24 13:55:59 UTC 2003 x86_64 
> x86_64 x86_64 GNU/Linux
> 
> B.
> _______________________________________________________________________________________________
> 3.5.1-beta11

> C.
...

> D.
> __________________________________________________________________________
> /usr/pgi/linux86-64/5.1/bin
> 
> E.
...

> F.
...

> G.
...

> H.
...
> Making `test' in directory /home/jan/netcdf/src/nc_test
...
> Total number of failures: 0
> *** Success ***
>
> make[2]: Leaving directory `/home/jan/netcdf/src/nc_test'
> 
> Returning to directory /home/jan/netcdf/src
> 
> make[1]: Leaving directory `/home/jan/netcdf/src'
> make[1]: Entering directory `/home/jan/netcdf/src'
> 
> Making `test' in directory /home/jan/netcdf/src/nctest
> 
> make[2]: Entering directory `/home/jan/netcdf/src/nctest'
> pgcc -o nctest -fPIC  varget.o vargetg.o varput.o varputg.o vardef.o 
> vartests.o vputget.o vputgetg.o driver.o cdftests.o dimtests.o rec.o 
> atttests.o misctest.o add.o error.o emalloc.o val.o slabs.o -L../libsrc 
> -lnetcdf
> ./nctest
> ./nctest: Symbol `exit' causes overflow in R_X86_64_PC32 relocation
> *** Testing nccreate ...      ok ***
> *** Testing ncopen ...                ok ***
> *** Testing ncredef ...               ok ***
> *** Testing ncendef ...               ok ***
> *** Testing ncclose ...               ok ***
> *** Testing ncinquire ...     ok ***
> *** Testing ncsync ...                ok ***
> *** Testing ncabort ...               ok ***
> *** Testing ncdimdef ...      ok ***
> *** Testing ncdimid ...               ok ***
> *** Testing ncdiminq ...      ok ***
> *** Testing ncdimrename ...   ok ***
> *** Testing ncvardef ...      ok ***
> *** Testing ncvarid ...               ok ***
> *** Testing ncvarinq ...      ok ***
> *** Testing ncvarput1 ...     ok ***
> *** Testing ncvarget1 ...     ok ***
> *** Testing ncvarput ...      ok ***
> *** Testing ncvarget ...      *** test_slabs: ncvarget got wrong value for 
> point
> *** test_slabs: ncvarget got wrong value for vector
> *** test_slabs: ncvarget got wrong value for vector
> *** test_slabs: ncvarget got wrong value for vector

I've seen this problem before, but only with the Portland Group version
5 compiler on a Linux x86_64 system.  See, for example,

    http://www.unidata.ucar.edu/cgi-bin/msgout?/glimpse/netcdf/5417

This doesn't happen with the same version of the PG compiler on non-x86_64
systems or with other compilers (e.g., gcc, g77) on a Linux x86_64
system.

We suspect that Portland Group still has some work to do to correctly
support the AMD64 CPU.

You could either install the netCDF package as it currently exists or
try rebuilding it using non-PG compilers.  If you do install the
existing netCDF package, then you should test it with netCDF files and
programs of your own.

Good luck.

Regards,
Steve Emmerson