Call Graph error: CCallTree::Build: No entries in CG raw file.

[plipcsei]

When I stop the call-graph collection ofan applicationwhich usesa lot of shared objects, I get the following error message:

CCallTree::Build: No entries in CG raw file.

The application starts and behaves as usual, but call-graph data is not collected. I suspect that some environment variable of Vtune is not set correctly, the VTUNE_GLOBAL_DIR and the cache directory is pointing to writeable locations. Is there anything else I could check. For smaller applications call-graph works.

[plipcsei]

Additional information:

I'm using vtserver on linux and analyzer 7.1 on windowswith RDC.

[David_A_Intel1]

From previous problem reports, this could be a mismatch of symbol data or bad debug info from the compiler/linker. Please see http://support.intel.com/support/performancetools/vtune/v71/sb/cs-006107-prd1391.htmfor information on troubleshooting call graph problems.

What development tools are you using to create your application?

[plipcsei]

Thanks DaveA for the reply. According to your opinion I think the problem is that there is no debug info in the libraries I profile. I'm using g++ 2.95.4 for C++ code. I didn't compile the release version with the -g option, this also could be a problem. I read the information at the link you provided, but this didn't help to much. Anyway I'll try to gather call-graph information on debug libraries, I hope this will help.

[jeffrey-gallagher]

Hey plipcsei,

Be sure to report back here, no matter what.

Also consider, if you are totally stuck, you should open a case at http://premier.intel.com and let those folks work their magic on it.

If you have to open the case, might not hurt to attach a pack file of the issue, so that support can see what you're seeing.

cheers

jdg

[plipcsei]

Hi,

Well, I'm still stuck with this problem. My company already contacted somebody from Intel, he asked for the packed file and version infos, unfortunately I can't reproduce the fault, because our servers are not available, but I let you know if there is some progress in this matter.

Br,

Peter

[jeffrey-gallagher]

Peter, very well, many thanks for reporting back. Should you get into a position to reproduce the experiment, folks here will be very eager to learn more about it.

cheers

jdg