I have built Kopete 0.12.2 on my CS 4 Linux (a clone of RHEL 4) system. It
is working well, except I get a segfault when I am connected to Yahoo!, and I
then connect to Yahoo! from another system. The backtrace is unusable.
I would like to run Kopete under GDB to see if I can find where the fault
occurs. But it appears that the install process strips the debug symbols. What
is the best way to build and install Kopete with debug symbols intact so it can
be run from GDB?

You sould get the SVN version here :
and compile it with the option

Le jeudi 26 octobre 2006 23:28, Matt Seitz a *:
I have built Kopete 0.12.2 on my CS 4 Linux (a clone of RHEL 4) system.
It is working well, except I get a segfault when I am connected to Yahoo!,
and I then connect to Yahoo! from another system. The backtrace is
unusable.

I would like to run Kopete under GDB to see if I can find where the fault
occurs. But it appears that the install process strips the debug symbols.
What is the best way to build and install Kopete with debug symbols intact
so it can be run from GDB?

Michel Hermier wrote:
When configuring the package, use
And when running it starting it withing gdb use

Thanks, that worked. I found the problem is a NULL pointer dereference in
"":

void ClientStream::cp_outgoingData( const QByteArray& outgoingBytes )
{
// take formatted bytes from CoreProtocol and put them on the wire
kdDebug(YAHRAW_DEBUG) << k_funcinfo << "[data size: " <<
outgoingBytes.size() << "]" << endl;
//cs_dump( outgoingBytes );
d->bs->write( outgoingBytes );
}

The segfault happens on the "d->bs->write( outgoingBytes );" statement. At this
point, "d->bs" is NULL.

I would file a report in Bugzilla. But it refuses to take my report, because
the version of KDE I am running is too old. I am running CS 4 (an RHEL 4
clone), which ships with a patched version of KDE 3.3.1.

Matt Seitz wrote:
Thanks, that worked. I found the problem is a NULL pointer dereference
in "":

I should probably provide some context. Here was my original problem description:

"I have built Kopete 0.12.2 on my CS 4 Linux (a clone of RHEL 4) system. It
is working well, except I get a segfault when I am connected to Yahoo!, and I
then connect to Yahoo! from another system."

So with some help from the list, I compiled a debug version and ran it under
GDB. That's how I found this problem.

Looking at the call stack, it appears this originated with a call to
"Client::sendPing()".

void ClientStream::cp_outgoingData( const QByteArray& outgoingBytes )
{
// take formatted bytes from CoreProtocol and put them on the wire
kdDebug(YAHRAW_DEBUG) << k_funcinfo << "[data size: " <<
outgoingBytes.size() << "]" << endl;
//cs_dump( outgoingBytes );
d->bs->write( outgoingBytes );
}

The segfault happens on the "d->bs->write( outgoingBytes );" statement.
At this point, "d->bs" is NULL.

I would file a report in Bugzilla. But it refuses to take my report,
because the version of KDE I am running is too old. I am running CS
4 (an RHEL 4 clone), which ships with a patched version of KDE 3.3.1.

kopete-devel mailing list
kopete-devel (AT) kde (DOT) org

Friday 27 2006 01:35, Matt Seitz wrote:
I would file a report in Bugzilla. *But it refuses to take my report,
because the version of KDE I am running is too old. *I am running CS 4
(an RHEL 4 clone), which ships with a patched version of KDE 3.3.1.

Are you able to confirm that the bug is still present in more recent Kopete
versions? If not, then you may want to update to a new KDE/Kopete on your
system.

The Kopete developers unfortunately don't have enough manpower to maintain the
old version, which is why Bugzilla refused the report, but the CS people
may be doing their own support. As a last resort you can perhaps patch the
code yourself.

In case the problem is still there with 3.5.x, things get more interesting.
The maintainers of the Yahoo plugin may be interested in a backtrace from
Valgrind (instead of GDB), as it's usually more accurate with memory
debugging.

Friday 27 2006 03:55, Martijn Klingens wrote:
Friday 27 2006 01:35, Matt Seitz wrote:
I would file a report in Bugzilla. *But it refuses to take my report,
because the version of KDE I am running is too old. *I am running CS
4 (an RHEL 4 clone), which ships with a patched version of KDE 3.3.1.

Are you able to confirm that the bug is still present in more recent Kopete
versions? If not, then you may want to update to a new KDE/Kopete on your
system.

The Kopete developers unfortunately don't have enough manpower to maintain
the old version, which is why Bugzilla refused the report, but the CS
people may be doing their own support. As a last resort you can perhaps
patch the code yourself.

In case the problem is still there with 3.5.x, things get more interesting.
The maintainers of the Yahoo plugin may be interested in a backtrace from
Valgrind (instead of GDB), as it's usually more accurate with memory
debugging.

He is talking about Kopete 0.12.2 (which is compilable with KDE 3.3.x) so I
don't think there's a problem with Kopete being too old. :)

Martijn Klingens wrote:
Are you able to confirm that the bug is still present in more recent Kopete
versions?

I am running Kopete 0.12.2. I thought that was the most recent version.

If not, then you may want to update to a new KDE/Kopete on your
system.

KDE 3.3.1-6 is the latest version of KDE that Red Hat has released for
Enterprise Linux 4. I'm reluctant to replace the entire RHEL version of the KDE
system with my own build.

The Kopete developers unfortunately don't have enough manpower to maintain the
old version, which is why Bugzilla refused the report

That's understandable, although in this case I am running the current version of
Kopete, just on an older version of KDE.

As a last resort you can perhaps patch the
code yourself.

I'd be happy to, if someone could point me to what the correct fix is.

In case the problem is still there with 3.5.x, things get more interesting.

I'll see if I can set up another machine with 3.5.x for testing.

The maintainers of the Yahoo plugin may be interested in a backtrace from
Valgrind (instead of GDB), as it's usually more accurate with memory
debugging.

K, I'll give that a try.

Friday 27 2006 15:57, Matt Seitz wrote:
Martijn Klingens wrote:
Are you able to confirm that the bug is still present in more recent
Kopete versions?

I am running Kopete 0.12.2. I thought that was the most recent version.

, sorry, I missed that :)

The Kopete developers unfortunately don't have enough manpower to
maintain the old version, which is why Bugzilla refused the report

That's understandable, although in this case I am running the current
version of Kopete, just on an older version of KDE.

In that case, just cheat the wizard, specify the newest KDE version and
mention in the report that the kopete version and the kdelibs version are
different.

In case the problem is still there with 3.5.x, things get more
interesting.

I'll see if I can set up another machine with 3.5.x for testing.

Well, your version should be good enough, probably, it doesn't *seem* to be
kdelibs-related, I entirely missed that you are running 0.12 :P

The maintainers of the Yahoo plugin may be interested in a backtrace from
Valgrind (instead of GDB), as it's usually more accurate with memory
debugging.

K, I'll give that a try.

Please do. I'm sure the Yahoo ppl can use the extra data it produces.