dolfin team mailing list archive

[Joachim Berdal Haga <jobh@xxxxxxxxx>]

After merging with trunk, I get errors like the following (example:
demo/undocumented/block_matrix/cpp). Does anybody recognize the error,
or have a clue?

Build sparsity pattern
Create layout
End Create layout
[0]PETSC ERROR:
------------------------------------------------------------------------
[0]PETSC ERROR: Caught signal number 11 SEGV: Segmentation Violation,
probably memory access out of range

....

Program received signal SIGSEGV, Segmentation fault.
0x0805a2ed in dolfin::GenericVector::init (this=0x8100290,
    sparsity_pattern=...)
    at ../../../../../../../include/dolfin/la/GenericVector.h:59
59          { resize(tensor_layout.local_range(0)); zero(); }
(gdb) bt
#0  0x0805a2ed in dolfin::GenericVector::init (this=0x8100290,
    sparsity_pattern=...)
    at ../../../../../../../include/dolfin/la/GenericVector.h:59
#1  0xb7d86e2e in dolfin::AssemblerTools::init_global_tensor (A=..., a=...,
    reset_sparsity=false, add_values=false)
    at ../../dolfin/fem/AssemblerTools.cpp:196
#2  0xb7dacf01 in dolfin::Assembler::assemble (A=..., a=..., cell_domains=0x0,
    exterior_facet_domains=0x0, interior_facet_domains=0x0,
    reset_sparsity=true, add_values=false, finalize_tensor=true)
    at ../../dolfin/fem/Assembler.cpp:169
#3  0xb7dad9e6 in dolfin::Assembler::assemble (A=..., a=...,
    reset_sparsity=true, add_values=false, finalize_tensor=true)
    at ../../dolfin/fem/Assembler.cpp:58
#4  0xb7d5c76e in dolfin::assemble (A=..., a=..., reset_sparsity=true,
    add_values=false, finalize_tensor=true) at ../../dolfin/fem/assemble.cpp:41
#5  0x0805943e in main ()
    at ../../../../../demo/undocumented/block-matrix/cpp/main.cpp:45

-j.


On 28 February 2012 16:42, Anders Logg <logg@xxxxxxxxx> wrote:
> On Tue, Feb 28, 2012 at 02:48:07PM +0100, Joachim Berdal Haga wrote:
>> On 28 February 2012 14:10, Joachim Berdal Haga <jobh@xxxxxxxxx> wrote:
>> >> It seems that that line was added by Joachim in revision 6540:
>> >>
>> >> http://bazaar.launchpad.net/~jobh/dolfin/symmetric-assemble/annotate/head:/dolfin/la/GenericMatrix.cpp
>> >
>> > Aha, yes, it looks like it's a diagnostic that I'd forgotten about. In
>> > fact, that revision wasn't supposed to touch GenericMatrix.* at all.
>> > I'll revert those changes in my branch, just give me a few minutes.
>>
>> ... or rather a few hours, so I can re-run the tests as well...
>
> ok. :-)
>
> --
> Anders