Re: [Ekiga-list] Debugging information

[Eugen Dedu <Eugen Dedu pu-pm univ-fcomte fr>]

Hamlet wrote:
> On Monday 08 March 2010 03:16:56 Eugen Dedu wrote:
>>> I have noticed that the amount of debug information printed with the
>>> option "-d N" (even N=5) has dramatcally decreased passing from 3.0.1 to
>>> 3.2.5/6 (including OPAL/PTlib updates).
>>>
>>> Now I get only messages on screen from dyna.cxx and a few CODEC/protocol
>>> related sources.
>>> And I need connection information, badly, since it's not working and
>>> that's where all my diagnostics ends now.
> [...]
>> I think there is a debug option, which is turned on by default.  What
>> are the options to build ptlib, opal and ekiga in gentoo?
> 
> About 40. For each one of them!!!
> There is a "debug" option, which I thought just meant the symbols are not 
> stripped away. Wrong. In this case, it seems to mean something more.

Well, I wanted to explain it to the wiki, but seems to me that it's
sufficiently clear:
  --disable-tracing     Remove PTRACE and all trace logging

> If I enable it (the system requires it to be enabled either for all the three 
> packages or for none of them) I get a rich crash (regardless the -d option):
> 
> $ ekiga
> *** glibc detected *** ekiga: munmap_chunk(): invalid pointer: 
> 0x00000000008d3fb0 ***
> ======= Backtrace: =========                                                         
> /lib/libc.so.6[0x7ff6dedb4808]                                                       
> /usr/lib/libpt_d.so.2.6.6(_ZN13PAbstractList13RemoveElementEP12PListElement+0x6b)
> [0x7ff6e48bb87b]
> /usr/lib/libpt_d.so.2.6.6(_ZN13PAbstractList8RemoveAtEi+0x2a)[0x7ff6e48bb96a]                    
> /usr/lib/libpt_d.so.2.6.6(_ZN11PCollection9RemoveAllEv+0x1e)[0x7ff6e48b9fce]                     
> /usr/lib/libpt_d.so.2.6.6(_ZN13PAbstractList15DestroyContentsEv+0xd)
> [0x7ff6e48bab9d]             
> /usr/lib/libpt_d.so.2.6.6(_ZN10PContainer8DestructEv+0x2e)[0x7ff6e48c145e]                       
> /usr/lib/libpt_d.so.2.6.6(_ZN14PPluginManager19LoadPluginDirectoryERK10PDirectory+0x432)
> [0x7ff6e488f992]
> /usr/lib/libpt_d.so.2.6.6(_ZN19PluginLoaderStartup9OnStartupEv+0x72)
> [0x7ff6e488fc22]                    
> /usr/lib/libpt_d.so.2.6.6(_ZN8PProcessC2EPKcS1_ttNS_10CodeStatusEtb+0x70c)
> [0x7ff6e48b14ec]              
> ekiga(_ZN12GnomeMeetingC1Ev+0x4a)[0x4cec5a]                                                             
> ekiga(main+0x504)[0x4c0a34]                                                                             
> /lib/libc.so.6(__libc_start_main+0xe6)[0x7ff6ded5fa26]                                                  
> ekiga(_ZN13PArrayObjects6InsertERK7PObjectPS0_+0x41)[0x4a60a9]                                          
> ======= Memory map: ========                                                                            
> 00400000-00667000 r-xp 00000000 fd:00 58597                              
> /usr/bin/ekiga                 
> [ ... and a long memory map follows, available on demand ]
> Aborted
> 
> If I run it with gdb, it keeps one CPU 100% busy and does nothing apparent. 
> After 2 hours I interrupted it and asked for a backtrace, and the answer was 
> (starting from my Ctrl+C): 
> 
> Program received signal SIGINT, Interrupt.
> 0x00007ffff09d6ac0 in ?? () from //usr/lib64/opengl/nvidia/lib/libGL.so.1
> (gdb) backtrace
> #0  0x00007ffff09d6ac0 in ?? () from //usr/lib64/opengl/nvidia/lib/libGL.so.1
> #1  0x00007ffff09d75c2 in ?? () from //usr/lib64/opengl/nvidia/lib/libGL.so.1
> #2  0x00007ffff7e90b1b in ?? () from //usr/lib64/opengl/nvidia/lib/libnvidia-
> tls.so.1
> #3  0x00007fffed7e087b in PAbstractList::RemoveElement(PListElement*) () from 
> /usr/lib/libpt_d.so.2.6.6
> #4  0x00007fffed7e096a in PAbstractList::RemoveAt(int) () from 
> /usr/lib/libpt_d.so.2.6.6
> #5  0x00007fffed7defce in PCollection::RemoveAll() () from 
> /usr/lib/libpt_d.so.2.6.6
> #6  0x00007fffed7dfb9d in PAbstractList::DestroyContents() () from 
> /usr/lib/libpt_d.so.2.6.6
> #7  0x00007fffed7e645e in PContainer::Destruct() () from 
> /usr/lib/libpt_d.so.2.6.6
> #8  0x00007fffed7b4992 in PPluginManager::LoadPluginDirectory(PDirectory 
> const&) () from /usr/lib/libpt_d.so.2.6.6
> #9  0x00007fffed7b4c22 in PluginLoaderStartup::OnStartup() () from 
> /usr/lib/libpt_d.so.2.6.6
> #10 0x00007fffed7d64ec in PProcess::PProcess(char const*, char const*, 
> unsigned short, unsigned short, PProcess::CodeStatus, unsigned short, bool) ()
>    from /usr/lib/libpt_d.so.2.6.6
> #11 0x00000000004cec5a in GnomeMeeting::GnomeMeeting() ()
> #12 0x00000000004c0a34 in main ()
> 
> Looks the same.

I think that such a crash has been fixed since 3.2.6.  Could you please
test the latest stable branch:
git clone git://git.gnome.org/ekiga; cd ekiga; git checkout -b
gnome-2-26 origin/gnome-2-26
and use ./autogen.sh instead of ./configure?

> If I don't enable the Gentoo debug flag, it passes through this without 
> problems -- then it fails the echo test, but that's not the subject of these 

The echo test sometimes stops, a bug in the SIP server very probably.
Right now it works for ex.

> e-mails. What do you think about this?
> 
> 
> Gentoo ebuild for:
> nVidia drivers: nvidia-drivers 190.53-r1
> X.org server: xorg-server 1.7.5.901 (1.7.6 RC 1)

-- 
Eugen