[Paraview] Program received signal SIGFPE, Arithmetic exception.

Burlen Loring burlen.loring at gmail.com
Fri Mar 17 15:11:25 EDT 2017


Thank you guys. I poked around this morning and found the following:

The "BlueObeliskElements" is the only cmap in the default presets that 
triggers the issue. It has more values than can be displayed on a single 
line, and this is what triggers the FPE in the logic.

Enforcing a minimum swatch size(terminology from the code) prevents the 
FPE and seems to produce the desired result of displaying all of the 
cmap values in one line, although they are quite small.

a moment ago I pushed a patch onto gitlab demonstrating. Perhaps it or 
something like it could be back ported to the 5.3.0 branch?

On 03/17/2017 07:09 AM, Cory Quammen wrote:
> Burlen,
>
> I've created an issue for your report. It sure looks like a bug.
>
> https://gitlab.kitware.com/paraview/paraview/issues/17305
>
> Ufuk,
>
> That fix turns out to already be in v5.3.0:
>
> commit 90e710d5792116e640d25caa8ff455ae4e65d718
> Author: Sergey Sindeev <ssindeev at yandex.ru>
> Date:   Sun Dec 18 13:21:27 2016 +0100
>
>      fix a compilation crash when using the latest ICC compiler
>
> diff --git a/Qt/Components/pqPresetToPixmap.cxx
> b/Qt/Components/pqPresetToPixmap.cxx
> index 13f6b2e..b80fb80 100644
> --- a/Qt/Components/pqPresetToPixmap.cxx
> +++ b/Qt/Components/pqPresetToPixmap.cxx
> @@ -254,15 +254,12 @@ QPixmap
> pqPresetToPixmap::renderIndexedColorTransferFunction(
>       // Now determine best value for Nh in [Nh/2,Nh-1]
>       double bestQ = vtkMath::Inf();
>       int best = -1;
> -    for (int i = Nh / 2; i < Nh; ++i)
> +    double ar = Nv * wmp / static_cast<double>(hmp * Nh);
> +    double q = (ar >= 1.0) ? ar : 1. / ar;
> +    if (q < bestQ)
>       {
> -      double ar = Nv * wmp / static_cast<double>(hmp * Nh);
> -      double q = (ar >= 1.0) ? ar : 1. / ar;
> -      if (q < bestQ)
> -      {
> -        bestQ = q;
> -        best = i;
> -      }
> +      bestQ = q;
> +      best = Nh-1;
>       }
>       Nh = best;
>     }
>
> Thanks to you and Sergey for pursuing this Intel compiler issue and
> ParaView issue!
>
> - Cory
>
> On Fri, Mar 17, 2017 at 3:18 AM,  <u.utku.turuncoglu at be.itu.edu.tr> wrote:
>> Hi Burlen,
>>
>> Your issue could be related with following,
>>
>> https://software.intel.com/en-us/forums/intel-c-compiler/topic/702934
>>
>> I opened a bug report in Intel form and they found a bug in that
>> particular source file (pqPresetToPixmap.cxx). So, if you apply the fix
>> you might solve the problem but i am not sure.
>>
>> I hope it helps,
>> Regards,
>>
>> --ufuk
>>
>>
>>> I tried out ParaView 5.3.0 today, and I encountered the above crash.
>>>
>>> steps to reproduce: load data (sphere source will work), set color by
>>> var, open color map dialog, change to categorical, open choose presets
>>> dialog, scroll down, when you get near the bottom you will get the crash.
>>>
>>> Program received signal SIGFPE, Arithmetic exception.
>>> 0x00007ffff660a833 in
>>> pqPresetToPixmap::renderIndexedColorTransferFunction (this=0x5e22560,
>>> stc=0x647f460, size=...) at
>>> /home/bloring/work/ParaView/Qt/Components/pqPresetToPixmap.cxx:273
>>> 273      Nh = wmp / (ss + PQ_SWATCH_PAD);
>>> (gdb) p ss
>>> $5 = -2
>>>
>>> of course PQ_SWATCH_PAD == 2
>>> _______________________________________________
>>> 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
>>>
>>
>> _______________________________________________
>> 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
>
>



More information about the ParaView mailing list