[JDEV] icqv7 core dumps on external part ?

David Waite mass at akuma.org
Thu Mar 28 13:02:35 CST 2002

Most of the past GCC issues have been threading-related; so if this 
happens predictably it is probably an issue in the code of the external 
component (or the JECL libraries it is based on).

I guess I can put it on my long TODO list :-) The only memory debugging 
tool I know of for *nix which would work for debugging this sort of 
setup is dmalloc.

-David Waite

Wim Godden wrote:

> Sadly, I'm already using 3.0.4... I had to upgrade to 3.0.4 because my 
> previous gcc wouldn't work...
> So I guess I'm stuck ?
> David Waite wrote:
>> Without looking at the code, I can say there is a 99% chance this 
>> means that there has been memory corruption. Several things could 
>> have caused this corruption (array write out of bounds, double free, 
>> etc.) There is a good chance it could be gcc 3.0.1 - there are a ton 
>> of changes which happened to their C++ implementation between 2.95.3 
>> and 3.0.x. There have also been a lot of improvements since 3.0, so 
>> if you can upgrade to 3.0.4, I would recommend it. The STL 
>> implementation within versions of gcc before 3.0.x were 
>> thread-unsafe, especially for std:string (they were interally 
>> reference-counted, but the reference modification was not atomic, nor 
>> was the reference comparison)
>> -David Waite
>> Wim Godden wrote:
>>>This is the result :
>>>Core was generated by `icqv7ext'.
>>>Program terminated with signal 11, Segmentation fault.
>>>Reading symbols from /usr/local/lib/libsigc-1.0.so.0...done.
>>>Reading symbols from /lib/libpthread.so.0...done.
>>>Reading symbols from /lib/libresolv.so.2...done.
>>>Reading symbols from /usr/local/lib/libstdc++.so.3...done.
>>>Reading symbols from /lib/libm.so.6...done.
>>>Reading symbols from /usr/local/lib/libgcc_s.so.1...done.
>>>Reading symbols from /lib/libc.so.6...done.
>>>Reading symbols from /lib/ld-linux.so.2...done.
>>>#0  0x40126e88 in chunk_alloc (ar_ptr=0x401bbd40, nb=16) at malloc.c:2875
>>>2875    malloc.c: No such file or directory.
>>>(gdb) bt
>>>#0  0x40126e88 in chunk_alloc (ar_ptr=0x401bbd40, nb=16) at malloc.c:2875
>>>#1  0x401265ae in __libc_malloc (bytes=12) at malloc.c:2696
>>>#2  0x80de950 in _ZNSt23__malloc_alloc_templateILi0EE8allocateEj (__n=12) at
>>>#3  0x8
>>>06f6d1 in
>>>    this=0x8153010, watcher=@0xbffff81c, pool=@0xbffff7c4, key=0,
>>>socketfd=-1, karma=0x0, conn_limit=0x0)
>>>    at /usr/local/include/g++-v3/bits/stl_alloc.h:184
>>>#4  0x806e655 in
>>>    this=0xbffff81c, listener=0xbffff9f4, host=0x814fad4 "",
>>>service=0x814fab4 "_jabber._tcp", defaultport=7000,
>>>    thread_key=0, karma=0x0) at SocketWatcher.cpp:195
>>>#5  0x80db48e in
>>>    name=@0xbffff7a4, secret=@0xbffffac0, a=@0xbffff5d0) at
>>>#6  0x8059516 in _ZN8IcqTransC1ERKSsS1_S1_jb (this=0xbffff7a0,
>>>serviceid=@0xbffffad0, password=@0xbffffac0, hostname=@0xbffffab0,
>>>    port=
>>>7000, outgoing_dir=true) at IcqTrans.cpp:67
>>>#7  0x8069c17 in main (argc=1, argv=0xbffffb44) at main.cpp:113
>>>Current language:  auto; currently c
>>>David Waite wrote:
>>>>Wim Godden wrote:
>>>>>Trying to run it like this :
>>>>>icqv7ext -n icqv7ext -s secretpass -h -p 7000 -m
>>>>>I get :
>>>>>Starting component...
>>>>>       Component ID : icqv7ext
>>>>>       Secret       : secretpass
>>>>>       Jabberd IP   :
>>>>>       Jabberd Port : 7000
>>>>>Segmentation fault (core dumped)
>>>>>Any ideas ?
>>>>Sounds like it is core-dumping wihle trying to start the component :-)
>>>>Perhaps you could post a backtrace from the core? (perhaps "gdb icqv7ext
>>>>-c core", then "bt".) Cores from threaded components are much nicer if
>>>>they are generated from something other than Linux; if you are using
>>>>Linux try to use the newest kernel possible (as they reworked a bit of
>>>>this logic within the 2.4 series)
>>>>-David Waite
>>>>jdev mailing list
>>>>jdev at jabber.org
>>>FirstLink Networks - http://www.firstlinknetworks.com
>>>De leukste moppen op je scherm en in je mailbox : http://www.moppen.be !!!!
>>>jdev mailing list
>>>jdev at jabber.org
> -- 
> -----
> FirstLink Networks - http://www.firstlinknetworks.com
> -----
> De leukste moppen op je scherm en in je mailbox : http://www.moppen.be 
> !!!!

More information about the JDev mailing list