dolfin team mailing list archive

[Johan Hake <johan.hake@xxxxxxxxx>]

On Monday 29 December 2008 11:59:26 Ilmar Wilbers wrote:
> You're right, I recompiled PETSc on the buildbots, and now they pass
> this step. There is a single Python demo that is failing with a PETSc
> error, not exactly sure what the problem is there.

Which demo is it, and are you sure it is a PETSc related error? I have one 
python demo that still fails on me. It is

   function/projection-interpolation

When PETSc is the linear backen PETSc debug messages kicks in, but when uBLAS 
is the backend I get a "normal" stack trace. 

The segfault only appears when DOLFIN_NOPLOTS is defined, so it is a strange 
bug, and seems to occur when the memory is released. Here is the full output 
when failing:

Checking mesh ordering (finished).
Assembling matrix over cells (finished).
Assembling vector over cells (finished).
Solving linear system of size 36 x 36 (uBLAS Krylov solver).
*** Warning: Conjugate-gradient method not programmed for uBLASKrylovSolver. 
Using GMRES.
Krylov solver converged in 7 iterations.
[hake-laptop:21122] *** Process received signal ***
[hake-laptop:21122] Signal: Segmentation fault (11)
[hake-laptop:21122] Signal code: Address not mapped (1)
[hake-laptop:21122] Failing at address: 0x3fc5bd62
[hake-laptop:21122] [ 0] [0xb7ee9440]
[hake-laptop:21122] [ 1] python [0x8088198]
[hake-laptop:21122] [ 2] python [0x80f4560]
[hake-laptop:21122] [ 3] python(PyGC_Collect+0x25) [0x80f4df5]
[hake-laptop:21122] [ 4] python(Py_Finalize+0xba) [0x80ea2da]
[hake-laptop:21122] [ 5] python(Py_Main+0x468) [0x8058d68]
[hake-laptop:21122] [ 6] python(main+0x22) [0x80587f2]
[hake-laptop:21122] [ 7] /lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xe0) 
[0xb7d3e450]
[hake-laptop:21122] [ 8] python [0x8058761]
[hake-laptop:21122] *** End of error message ***
Segmentation fault

It seems that the segfaults happens in the python library.

Am I the only one getting this error?

Johan

> ilmar
>
> Garth N. Wells wrote:
> > The problem could be that PETSc on the buildbot is not compiled with C++
> > support (configure option --with-clanguage=C++).
> >
> > Garth
> >
> > buildbot@xxxxxxxxxx wrote:
> >> The Buildbot has detected a new failure of linux_32.
> >> Full details are available at:
> >>  http://fenics.org:8010/dolfin/linux_32/builds/1160
> >>
> >> Buildbot URL: http://fenics.org:8010/dolfin/
> >>
> >> Buildslave for this Build: x86-linux
> >>
> >> Build Reason: Build Source Stamp: HEAD
> >> Blamelist: Anders Logg,Garth N. Wells,Johan Hake
> >>
> >> BUILD FAILED: failed compile_5
> >>
> >> sincerely,
> >>  -The Buildbot
> >>
> >> _______________________________________________
> >> DOLFIN-dev mailing list
> >> DOLFIN-dev@xxxxxxxxxx
> >> http://www.fenics.org/mailman/listinfo/dolfin-dev
>
> _______________________________________________
> DOLFIN-dev mailing list
> DOLFIN-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/dolfin-dev