[petsc-users] relocation truncated to fit: R_X86_64_PC32 error

[Matthew Knepley]

On Mon, Sep 17, 2012 at 6:03 AM, Miguel Méndez <mendez8 at gmail.com> wrote:

> Hi everybody,
>
> I'm getting the infamous "relocation truncated to fit: R_X86_64_PC32"
> error when compiling a fortran program. I've tried many things, the last
> one of them being to recompile Petsc and Slepc libraries. This has solved
> all of the compile errors related to this relocation except for one:
>
>
>
>> /usr/local/software/slepc-3.0.0-p7-fpic/complex/build/lib/libslepc.a(cross.o):
>> In function `ShellMatGetDiagonal_CROSS':
>> cross.c:(.text+0x3ca): relocation truncated to fit: R_X86_64_PC32 against
>> symbol `MPIU_COMPLEX' defined in COMMON section in
>> /usr/local/software/petsc-3.0.0-p12-fpic/complex/build/lib/libpetsc.a(init.o)
>
>
> I don't even think I know where the error is, in Petsc or in Slepc? I've
> been trough Intel forums and done as they recommend: recompile the
> libraries with "--mcmodel=medium" and "shared-intel". I've tried too
> combinations with "-fpic", but I always end having the same error.
>
> I've searched the logs for messages about "libpetsc.a" building, but
> there's nothing about it. Who and when builds this exact library? Why it
> seems that compile options I set are not applying in the building of this
> file?
>

This comes from not building with -fPIC (or what ever the PIC flag is for
your compiler). Find that source file, build
it with PIC, and archive it to libslepc.a. Of course, upgrading would make
this automatic.

    Matt


> Thanks,
>
> Miguel
>



-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20120917/ebd0014d/attachment.html>