[IGSTK-Developers] Fwd: dealing with unitialized value in vtkXOpenGLRenderWindow.cxx: looking for an expert

[Michel Audette]

Dear fellow IGSTK developers,

I have been discussing valgrind errors that relate to OpenGL with VTK
experts, and Francois Bertel and Lisa Avila in particular had relevant
comments on this issue. It certainly looks like these are common
errors that they suppress. Moreover, as Francois relates below, the
suppression requirements vary depending on the implementation of
OpenGL.

Best wishes,

Michel

---------- Forwarded message ----------
From: Francois Bertel <francois.bertel at kitware.com>
Date: Fri, Jul 17, 2009 at 10:19 AM
Subject: Re: dealing with unitialized value in
vtkXOpenGLRenderWindow.cxx: looking for an expert
To: kitware-developers at kitware.com


On the VTK dashboard, dash6 and faraway use Mesa. They both have
suppressions rules regarding Mesa (look at
DashboardScripts/dash6_VTKValgrindSuppressions.supp).

On VTKEdge, arkadia uses the nVidia OpenGL implementation. The
suppression file is the second script on this page:
https://www.kitware.com/CDash/viewNotes.php?buildid=44924

This kind of valgrind errors depends on the version of the X library
and implementation of OpenGL and version of valgrind. You cannot just
reuse and existing suppression file from another machine.



On Fri, Jul 17, 2009 at 10:12 AM, Lisa Avila<lisa.avila at kitware.com> wrote:
> We suppress lots of errors caused by OpenGL - most caused I think by some
> confusion on the main memory / GPU memory transition (for example something
> looks like an uninitialized memory read because it can't find where that
> memory was ever written - but that's off in some inaccessible device driver
> code perhaps).  I know in Purify there was a list of them somewhere, and it
> was fairly obvious to recognize which ones were OpenGL related. I assume we
> have the same sort of list somewhere in VTK for valgrind which I assume you
> can just copy over to IGSTK. Looking at the VTK source tree this seems to be
> in CMake/VTKValgringSuppressions.supp.
>
> Lisa
>
>
> Michel Audette wrote:
>>
>> Hi François,
>>
>> that one is typical. Can you confirm that this kind of valgrind error
>> has been suppressed from the VTK dashboard? I'm trying to get an idea
>> as to why IGSTK is getting those, and VTK is not.
>>
>> Best wishes,
>>
>> Michel
>>
>> On Thu, Jul 16, 2009 at 4:13 PM, Francois
>> Bertel<francois.bertel at kitware.com> wrote:
>>
>>>
>>> If the following case is one of them:
>>> http://www.cdash.org/CDash/viewDynamicAnalysisFile.php?id=378502
>>>
>>> just add a suppression rule to the valgrind submission.
>>>
>>> On Thu, Jul 16, 2009 at 6:24 PM, Michel
>>> Audette<michel.audette at kitware.com> wrote:
>>>
>>>>
>>>> Dear fellow developers,
>>>>
>>>> I am looking to discuss with a VTK expert, and particularly with
>>>> someone knowledgeable in vtkXOpenGLRenderWindow.cxx.
>>>>
>>>> IGSTK has a number of valgrind errors that all seem related to the
>>>> same issue, a call to glXSwapBuffers at line 460.
>>>>   glXSwapBuffers(this->DisplayId, this->WindowId);
>>>>
>>>> I would like to discuss with this expert what could possibly be
>>>> lacking in IGSTK in terms of providing the appropriate initialization.
>>>>
>>>> Best wishes,
>>>>
>>>> Michel
>>>>
>>>> --
>>>> Michel Audette, Ph.D.
>>>> R & D Engineer,
>>>> Kitware Inc.,
>>>> Chapel Hill, N.C.
>>>>
>>>>
>>>
>>> --
>>> François Bertel, PhD  | Kitware Inc. Suite 204
>>> 1 (518) 371 3971 x113 | 28 Corporate Drive
>>>                     | Clifton Park NY 12065, USA
>>>
>>>
>>
>>
>>
>>
>



--
François Bertel, PhD  | Kitware Inc. Suite 204
1 (518) 371 3971 x113 | 28 Corporate Drive
                     | Clifton Park NY 12065, USA



-- 
Michel Audette, Ph.D.
R & D Engineer,
Kitware Inc.,
Chapel Hill, N.C.