[Insight-developers] LSMIO test failing on 6 platforms

[Luis Ibanez]

On Wed, Sep 14, 2011 at 5:20 PM, Johnson, Hans J <hans-johnson at uiowa.edu> wrote:
> The LSMIO tests are failing on 6 platforms:
> http://www.cdash.org/CDash/testSummary.php?project=2&name=itkLSMImageIOTest&date=2011-09-14
> Hans
> --


Hi Hans,

I'm running this in GDB and found some interesting hints.


Program received signal SIGSEGV, Segmentation fault.
0x000000000055080a in itk::SmartPointer<itk::ImageIOBase>::operator->
(this=0xe8)
    at /home/ibanez/Dashboards/src/ITK/Modules/Core/Common/include/itkSmartPointer.h:71
71	  { return m_Pointer; }
(gdb) bt
#0  0x000000000055080a in
itk::SmartPointer<itk::ImageIOBase>::operator-> (this=0xe8)
    at /home/ibanez/Dashboards/src/ITK/Modules/Core/Common/include/itkSmartPointer.h:71
#1  0x00000000005bd290 in
itk::ImageFileReader<itk::Image<itk::RGBPixel<unsigned char>, 2u>,
itk::DefaultConvertPixelTraits<itk::RGBPixel<unsigned char> >
>::GenerateOutputInformation (this=0x0)
    at /home/ibanez/Dashboards/src/ITK/Modules/IO/Base/include/itkImageFileReader.hxx:162

Backtrace stopped: previous frame inner to this frame (corrupt stack?)

-----------------------------------

It looks like memory corruption to me...

It happens during the Update() method,
but that's probably not much of a surprise.

At the time it seg faults, the "this" points
to the reader is 0x00.

So, somehow the reader is being destroyed
in transit...

It could be that we are overriding a buffer ? maybe ?

I'll keep looking...


    Luis