[Pkg-corba-devel] Bug#460419: Bug#460419: omniorb4: FTBFS on arm: segmentation fault

[Thomas Girard]

tags 460419 + confirmed help

Selon Martin Michlmayr <tbm at cyrius.com>:

> * Philipp Kern <pkern at debian.org> [2008-01-12 16:41]:
> > omniorb4 fails to build from source since 4.1.0-1 on arm and causes a
> > compiler segmentation fault.
> >
> > omniidl: 'cxx' imported from
> '../../../../src/lib/omniORB/omniidl_be/cxx/__init__.py'
> > omniidl: Preprocessing '../../../../idl/Naming.idl' with
> '/build/buildd/omniorb4-4.1.1/build/lib/omnicpp -lang-c++ -undef
> -D__OMNIIDL__=0x2630 -D__OMNIIDL_CXX__ "../../../../idl/Naming.idl"'
> > omniidl: Running front end
> > make[4]: *** [omniORB4/Naming.hh] Segmentation fault
> > make[4]: Leaving directory
> `/build/buildd/omniorb4-4.1.1/build/src/lib/omniORB'
>
> Our Netwinder buildds often show segfaults when compiling stuff and a
> rebuild helps.  I'm not sure what type of machine the "hedges" buildd
> is.  Aurelien, do you know?

Hello,

I can reproduce this on leisner. A first analysis with gdb makes me
think the stack gets corrupt. The segfault occurs after stepping into
a function, but before reaching first lines. Besides the very same
function is called twice without segfaulting, and when inkoked a third
time the crash occurs.

I believe the code uses alloca(), does that raise something?

Thanks for your help,

Regards,
Thomas