ROOT, an OO framework for large-scale scientific data analysis and data mining

Suvayu Ali | 4 Oct 2009 18:57

[ROOT] Building ROOT from source on Fedora 11

Hi ROOT users,

I tried building 5.24 and ran into this problem. So I tried the latest
trunk from svn for 5.25, but no luck. :(

It builds just fine without any errors (but a few warnings) with the
following options:

> Enabled support for asimage, astiff, builtin_afterimage,
> builtin_ftgl, builtin_freetype, builtin_pcre, builtin_zlib, cint5,
> cintex, exceptions, fftw3, genvector, gsl_shared, mathmore, memstat,
> minuit2, opengl, python, reflex, roofit, shadowpw, shared, soversion,
> ssl, tmva, unuran, xft, xml, xrootd.

When I open a TBrowser and browse a ROOT file with histograms everything
is fine. But after I save one of those histograms say as an image or a
ROOT macro and then try to close the TBrowser, ROOT segfaults. The
TBrowser window freezes, and the whole terminal becomes non-responsive.
The only way to kill every thing is to do `kill -9 <PID_for_root
PID_for_root.exe>'.

I have attached the error ROOT spits out on segfaulting as a text file.

My system specifications:
2.6.30.8-64.fc11.x86_64
ROOT 5.25/03
g++ (GCC) 4.4.1 20090725 (Red Hat 4.4.1-2)

Thank you for any insights into solving this.

(Continue reading)

headers

Rene Brun | 5 Oct 2009 08:47

Re: [ROOT] Building ROOT from source on Fedora 11

I cannot reproduce this problem.
Could you compile ROOT in debug mode, then run the valgrind 
memory-checker on root.exe?

Rene Brun

Suvayu Ali wrote:
> Hi ROOT users,
>
> I tried building 5.24 and ran into this problem. So I tried the latest
> trunk from svn for 5.25, but no luck. :(
>
> It builds just fine without any errors (but a few warnings) with the
> following options:
>
>> Enabled support for asimage, astiff, builtin_afterimage,
>> builtin_ftgl, builtin_freetype, builtin_pcre, builtin_zlib, cint5,
>> cintex, exceptions, fftw3, genvector, gsl_shared, mathmore, memstat,
>> minuit2, opengl, python, reflex, roofit, shadowpw, shared, soversion,
>> ssl, tmva, unuran, xft, xml, xrootd.
>
> When I open a TBrowser and browse a ROOT file with histograms everything
> is fine. But after I save one of those histograms say as an image or a
> ROOT macro and then try to close the TBrowser, ROOT segfaults. The
> TBrowser window freezes, and the whole terminal becomes non-responsive.
> The only way to kill every thing is to do `kill -9 <PID_for_root
> PID_for_root.exe>'.
>
> I have attached the error ROOT spits out on segfaulting as a text file.
>

(Continue reading)

headers

Suvayu Ali | 6 Oct 2009 08:46

Re: [ROOT] Building ROOT from source on Fedora 11

Hi Rene,

Sorry for the delay, I am on the west coast of Canada. :\

On Sunday 04 October 2009 11:47 PM, Rene Brun wrote:
> I cannot reproduce this problem.
> Could you compile ROOT in debug mode, then run the valgrind
> memory-checker on root.exe?
>
I have attached a file with all the particulars about the build 
environment and the options I used to build ROOT. It also includes the 
backtrace from the segfault. I have also attached the log file generated 
by valgrind. I used the following command to attach valgrind to root.

>$ valgrind -v --log-file=valgrind-root.log root -l MonitorESD_91890_1.root

As reported earlier, after the segfault I had to kill ROOT with
$ kill -9 <PID_for_root PID_for_root.exe>

I also noticed that the segfault occurs only when I change something on 
the canvas using the histogram editor and try to close the TBrowser 
after saving it.

I hope this will help you track things down. Hopefully its not some 
stupid mistake on my part :-p

PS: If this makes any difference, the ROOT file was generated by ATHENA 
15.4.0.1 (ROOT version 5.22). It is full of TH1Fs, TH2Fs and TProfiles.

My system specifications:

(Continue reading)

headers

Axel Naumann | 12 Oct 2009 11:27

Re: [ROOT] Building ROOT from source on Fedora 11

Hi Suvayu,

you'll have to run valgrind on root.exe (yes, really, even on linux); 
"root" (without ".exe") is a tiny wrapper that will not show any issues.

Cheers, Axel.

Suvayu Ali wrote on 10/06/2009 08:46 AM:
> Hi Rene,
> 
> Sorry for the delay, I am on the west coast of Canada. :\
> 
> On Sunday 04 October 2009 11:47 PM, Rene Brun wrote:
>> I cannot reproduce this problem.
>> Could you compile ROOT in debug mode, then run the valgrind
>> memory-checker on root.exe?
>>
> I have attached a file with all the particulars about the build 
> environment and the options I used to build ROOT. It also includes the 
> backtrace from the segfault. I have also attached the log file generated 
> by valgrind. I used the following command to attach valgrind to root.
> 
>> $ valgrind -v --log-file=valgrind-root.log root -l 
>> MonitorESD_91890_1.root
> 
> As reported earlier, after the segfault I had to kill ROOT with
> $ kill -9 <PID_for_root PID_for_root.exe>
> 
> I also noticed that the segfault occurs only when I change something on 
> the canvas using the histogram editor and try to close the TBrowser

(Continue reading)