[ITK] Win64 Visual Studio 12 segfaults

[Floris Berendsen]

Hi,

My colleague noticed a similar bug report to Visual Studio:
https://connect.microsoft.com/VisualStudio/feedbackdetail/view/879286/incorrect-code-generated-in-x64-build-results-in-access-violation
He wrote a minimal example to reproduce that issue, see the link.
A segfault appears to occur in in IsInside(const Self & region) const of
itkImageRegion.h as well, but we could not yet make a minimal example of
that.

Thanks,
Floris


On Wed, Jan 7, 2015 at 8:40 PM, Matt McCormick <matt.mccormick at kitware.com>
wrote:

> +1
>
> Also, note that builds should be against ITK master -- a patch merged
> on 2015-01-03 has already addressed failing registration tests:
>
>   https://open.cdash.org/buildSummary.php?buildid=3637621
>
> Thanks,
> Matt
>
> On Wed, Jan 7, 2015 at 1:00 PM, Bradley Lowekamp <blowekamp at mail.nih.gov>
> wrote:
> > If the problem can be reproduced with just using itkFixedArray.h allow
> that
> > would be significant progress. As that class does not depend on any other
> > compiled code/library.
> >
> > Brad
> >
> >
> >
> > On Jan 7, 2015, at 10:33 AM, Мар'ян Климов <nekto1989 at gmail.com> wrote:
> >
> > Hi Floris,
> >
> > It would be good if you can reproduce this compiler problem without ITK.
> > I've tried but no success.
> >
> >
> https://connect.microsoft.com/VisualStudio/feedback/details/1060166/compiler-generates-divpd-instruction-on-non-aligned-data
> >
> > As workaround I suggest adding something like this
> >
> > #ifdef _MSC_VER
> > #if (_MSC_VER >= 1800)
> >     #ifdef _WIN64
> >      __declspec(align(16))
> >     #endif
> > #endif
> > #endif
> >
> > for classes that cause problems (FixedArray, Index etc.)
> >
> > Marian
> >
> > 2015-01-07 18:44 GMT+02:00 Floris Berendsen <floris at isi.uu.nl>:
> >>
> >> Hi all,
> >>
> >> I was wondering what the progress is of ITK with respect to Win64 Visual
> >> Studio 12 (2013). The dashboard has a Win64-VS12-Release under Expected
> >> Nightly Exotic, but it reports many segfault failures (I can confirm the
> >> same behaviour).
> >> All communications I could find related to this topic seem rather old:
> >>
> >> http://public.kitware.com/pipermail/community/2013-November/004255.html
> >>
> >>
> http://insight-developers.itk.narkive.com/niWueQGp/4-5-release-candidate-approaching
> >>
> >>
> http://insight-developers.itk.narkive.com/E6RszEvx/regression-errors-with-visual-studio-2013
> >>
> >> Does anyone know more about this? Would it help if I try to make a
> minimal
> >> example that reproduces the error?
> >>
> >> Thanks,
> >> Floris
> >>
> >> _______________________________________________
> >> Community mailing list
> >> Community at itk.org
> >> http://public.kitware.com/mailman/listinfo/community
> >>
> >
> > _______________________________________________
> > Community mailing list
> > Community at itk.org
> > http://public.kitware.com/mailman/listinfo/community
> >
> >
> >
> > _______________________________________________
> > Community mailing list
> > Community at itk.org
> > http://public.kitware.com/mailman/listinfo/community
> >
> _______________________________________________
> Community mailing list
> Community at itk.org
> http://public.kitware.com/mailman/listinfo/community
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/community/attachments/20150108/5c5e84dd/attachment-0001.html>