fnccheck-devel

[Fnccheck-devel] functioncheck and stl

[Polly P. <ppo...@in...>]

Greetings,

I am using FunctionCheck 3.0.11 with gcc 2.95.2 on a Debian woody 
system, and I am getting a segfault in libfc. When my executable is 
relinked without libfc, it runs correctly.

Here is a snippet of my gdb 5.1 backtrace:

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1024 (runnable)]
0x4001fc46 in fc_com_enter () from /usr/local/lib/libfc.so.0
(gdb) bt
#0  0x4001fc46 in fc_com_enter () from /usr/local/lib/libfc.so.0
#1  0x4001ec11 in __cyg_profile_func_enter () from /usr/local/lib/libfc.so.0
#2  0x080fa0a9 in __default_alloc_template<true, 0>::_S_refill (__n=24)
     at 
/usr/lib/gcc-lib/i386-linux/2.95.2/../../../../include/g++-3/stl_alloc.h:529
#3  0x402debec in __static_initialization_and_destruction_0 
(__initialize_p=1,
     __priority=65535) at /usr/include/g++-3/stl_alloc.h:419

I compiled libfc with debugging enabled and set the appropriate 
environmental variable, but didn't seem to get any debugging output I 
could pass along.

This is a multithreaded process. Here are the environmental variables set:

FC_ALLOW_THREAD=1
FC_DEBUG=1
FC_HELP=1
LD_LIBRARY_PATH=/usr/local/lib

My libfc.so is in /usr/local/lib, and an ldd of my binary shows that it 
is properly linked. I doublechecked that my ccflags were set correctly. 
Fc_manager is in my path.

The 3.0.11 Release Notes indicate that it has a fix for C++ templates, 
and I haven't found any open bug reports that seem relevant.

Does anyone have any ideas about what I should do next? Any suggestions 
would be most welcome.

Thanks in advance,

Polly Powledge
ppo...@in...

Re: [Fnccheck-devel] functioncheck and stl

[Yannick P. <yp...@ba...>]

Polly Powledge wrote:

> Greetings,
>
> I am using FunctionCheck 3.0.11 with gcc 2.95.2 on a Debian woody
> system, and I am getting a segfault in libfc. When my executable is
> relinked without libfc, it runs correctly.
>

Hi. Do you use '-pg' option for compilation ?
I know that my profiler is not compatible with prof or gprof (don't know
why at all).

If not, I can't see a reason for that.

You may send an archive of your program (if not secret...) in order to
let me try it by my side. It's easier for my to debug functioncheck
directly than giving instructions to people to find the bug...

Regards
--
Yannick



>
> Here is a snippet of my gdb 5.1 backtrace:
>
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 1024 (runnable)]
> 0x4001fc46 in fc_com_enter () from /usr/local/lib/libfc.so.0
> (gdb) bt
> #0  0x4001fc46 in fc_com_enter () from /usr/local/lib/libfc.so.0
> #1  0x4001ec11 in __cyg_profile_func_enter () from /usr/local/lib/libfc.so.0
> #2  0x080fa0a9 in __default_alloc_template<true, 0>::_S_refill (__n=24)
>      at
> /usr/lib/gcc-lib/i386-linux/2.95.2/../../../../include/g++-3/stl_alloc.h:529
> #3  0x402debec in __static_initialization_and_destruction_0
> (__initialize_p=1,
>      __priority=65535) at /usr/include/g++-3/stl_alloc.h:419
>
> I compiled libfc with debugging enabled and set the appropriate
> environmental variable, but didn't seem to get any debugging output I
> could pass along.
>
> This is a multithreaded process. Here are the environmental variables set:
>
> FC_ALLOW_THREAD=1
> FC_DEBUG=1
> FC_HELP=1
> LD_LIBRARY_PATH=/usr/local/lib
>
> My libfc.so is in /usr/local/lib, and an ldd of my binary shows that it
> is properly linked. I doublechecked that my ccflags were set correctly.
> Fc_manager is in my path.
>
> The 3.0.11 Release Notes indicate that it has a fix for C++ templates,
> and I haven't found any open bug reports that seem relevant.
>
> Does anyone have any ideas about what I should do next? Any suggestions
> would be most welcome.
>
> Thanks in advance,
>
> Polly Powledge
> ppo...@in...
>
> _______________________________________________________________
>
> Hundreds of nodes, one monster rendering program.
> Now that?s a super model! Visit http://clustering.foundries.sf.net/
> _______________________________________________
> Fnccheck-devel mailing list
> Fnc...@li...
> https://lists.sourceforge.net/lists/listinfo/fnccheck-devel