[Paraview] Segfault in ParaView 4.4 RC2

[Dorier, Matthieu]

Bingo! My Catalyst-instrumented sample code now works like a charm on my VM as well.
Thanks a lot for your help.

Matthieu 
________________________________________
From: Utkarsh Ayachit [utkarsh.ayachit at kitware.com]
Sent: Tuesday, September 15, 2015 3:21 PM
To: Dorier, Matthieu
Cc: paraview at paraview.org
Subject: Re: [Paraview] Segfault in ParaView 4.4 RC2

Matthieu,

Simply set the CMake variable OPENGL_glu_LIBRARY to empty. I think
you're pointing it to non-existant
/home/mdorier/local/mesa/lib/libGLU.so and hence the issue. If I am
not mistaken, GLU is optional only required for certain tests.

Utkarsh

On Tue, Sep 15, 2015 at 3:26 PM, Dorier, Matthieu <mdorier at anl.gov> wrote:
> I compiled and installed Mesa locally (Gallium llvmpipe state-tracker), but when compiling ParaView I get this error:
>
> make[2]: *** No rule to make target `/home/mdorier/local/mesa/lib/libGLU.so', needed by `lib/libIceTGL.so'.  Stop.
>
> It appears that libGLU.so is not built by Mesa. Is this normal? (I'm not familiar with Mesa at all) Should GLU be downloaded and compiled from somewhere? Or is there an option in Mesa?
>
> (I've used Mesa 11.0.0)
>
> Thanks,
>
> Matthieu
>
> ________________________________________
> From: ParaView [paraview-bounces at paraview.org] on behalf of Dorier, Matthieu [mdorier at anl.gov]
> Sent: Tuesday, September 15, 2015 11:00 AM
> To: Utkarsh Ayachit
> Cc: paraview at paraview.org
> Subject: Re: [Paraview] Segfault in ParaView 4.4 RC2
>
> Thanks a lot Utkarsh!
>
> In the mean time I tried my code on the Blue Waters machine at NCSA and it works like a charm, so yes, it has to be the VM. I'll try with OSMesa. I don't need the ParaView GUI on the VM anyway.
>
> Thanks,
>
> Matthieu
>
> ________________________________________
> From: Utkarsh Ayachit [utkarsh.ayachit at kitware.com]
> Sent: Tuesday, September 15, 2015 8:31 AM
> To: Dorier, Matthieu
> Cc: Cory Quammen; paraview at paraview.org
> Subject: Re: [Paraview] Segfault in ParaView 4.4 RC2
>
> NVM...those were from your original tarball. I thought they were
> generated as an output from my run :). It's indeed an X issue in both
> cases. Couple of options:
>
> 1. Do a Mesa3D build locally on the VM. Then set LD_LIBRARY_PATH to
> point to the generated OpenGL libs and run ParaView/Catalyst. That'll
> avoid using the GL libraries provided by virtualbox.
> 2. Do a OSMesa build and then rebuild ParaView/Catalyst to use OSMesa.
> That'd mean you can't have the Qt UI in the same build, but Catalyst
> and other scripting executables will work. That would avoid the X
> issue all together. This Wiki [a] describes how to get a OSMesa
> enabled ParaView build.
>
> [a] http://www.paraview.org/Wiki/ParaView/ParaView_And_Mesa_3D
>
> On Tue, Sep 15, 2015 at 9:19 AM, Utkarsh Ayachit
> <utkarsh.ayachit at kitware.com> wrote:
>> Matthieu,
>>
>> Good news, I just saw the black window issue too on my build. I'll
>> track it down and get back to you.
>>
>> Utkarsh
>>
>> On Tue, Sep 15, 2015 at 9:12 AM, Utkarsh Ayachit
>> <utkarsh.ayachit at kitware.com> wrote:
>>> Matthieu,
>>>
>>>> - If I run the application from the VM's graphical interface, it runs fine. It only gives me the OpenGL warning I mentioned before. BUT the generated file has a big black area (see image_2.png attached). Also I see a window appear for a brief moment with the correct (full) view, then it disappears. I don't know why I get that.
>>>
>>> I've seen this happen with the view size is greater the X window size.
>>> Try changing the ViewSize to something small, say 200x200. Does that
>>> work? If so, you'd need to change the X window size. Other option
>>> would be build with OSMesa support. Since you're only using this build
>>> for Catalyst, OSMesa will circumvent all the X related issues you've
>>> been seeing on this VM.
>>>
>>>> - If I connect to the VM through SSH with -X (otherwise I get the bad display error) and run the application, Catalyst also outputs an image, but this image also has an artifact (see image_1.png attached), and then the application crashes in a segfault, outputing a backtrace (*** glibc detected *** ./Mandelbulb: munmap_chunk(): invalid pointer: 0x00007faffb503010 ***).
>>>> In the backtrace, the first line (so the last function call) is this:
>>>> /home/mdorier/ParaView-build/lib/libvtkCommonCore-pv4.4.so.1(_ZN20vtkDataArrayTemplateIhE11DeleteArrayEv+0x4b)[0x7fb028c257fd]
>>>
>>> I am going to try your code to see if i can reproduce this one. I'll
>>> get back to you shortly.
>>>
>>> Utkarsh
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
>
> Please keep messages on-topic and check the ParaView Wiki at: http://paraview.org/Wiki/ParaView
>
> Search the list archives at: http://markmail.org/search/?q=ParaView
>
> Follow this link to subscribe/unsubscribe:
> http://public.kitware.com/mailman/listinfo/paraview