amaroK unstable

Fri Sep 2 15:45:49 BST 2005

Hi!
> 
> This is news to me. Could you provide a (valid) backtrace, make a bug report 
> and attach the backtrace, please?
There's the problem: My system doesn't like to generate backtraces from amarok
:-). I don't know why. Euen when there is a crash-mail generated, its backtrace
is useless.
  I compiled amarok with -g3 and run it from gdb. Even with that, the bt is
  1) either of the same quality as with the internal amarok bt collector
     (an example from frozen, not crashed amarok follows:)
Program received signal SIGINT, Interrupt.
[Switching to Thread -1208797856 (LWP 8838)]
0x013aac77 in ___newselect_nocancel () from /lib/libc.so.6
(gdb) bt
#0  0x013aac77 in ___newselect_nocancel () from /lib/libc.so.6
#1  0x007f93e6 in _XWaitForReadable () from /opt/X11/lib/libX11.so.6
#2  0xbfa5d670 in ?? ()
#3  0x00000000 in ?? ()
#4  0x00000000 in ?? ()
#5  0x00000000 in ?? ()
#6  0xbfa5d680 in ?? ()
#7  0x00c1c155 in fixup () from /lib/ld-linux.so.2
Previous frame inner to this frame (corrupt stack?)
  2) or cannot be printed at all (like this, from crash when loading playlist:)
(gdb) run
Starting program: /opt/kde3.4/bin/amarokapp
[Thread debugging using libthread_db enabled]
[New Thread -1208490656 (LWP 12105)]
[New Thread 57224112 (LWP 12117)]
[New Thread 38611888 (LWP 12118)]
QLayout: Adding KToolBar/mainToolBar (child of QVBox/unnamed) to layout for PlaylistWindow/PlaylistWindow
QPainter::begin: Cannot paint null pixmap
QColor::setRgb: RGB parameter(s) out of range
QColor::setRgb: RGB parameter(s) out of range
QColor::setRgb: RGB parameter(s) out of range
STARTUP
amarok: [CollectionDB] [ERROR!] [virtual QStringList SqliteConnection::query(const QString&)]  sqlite3_compile error:
amarok: [CollectionDB] [ERROR!] table amazon already exists
amarok: [CollectionDB] [ERROR!] on query: CREATE TABLE amazon ( asin VARCHAR(20), locale VARCHAR(2), filename VARCHAR(33), refetchdate INTEGER );
amarok: [Scrobbler] [WARNING!] KIO error! errno: 24
amarok:   [ThreadWeaver] [WARNING!] [void ThreadWeaver::Job::setProgressTotalSteps(unsigned int)] You can't set steps to 0!
QObject::connect: No such slot ArtsEngine::newStreamData(char*,int)
QObject::connect:  (sender name:   'unnamed')
QObject::connect:  (receiver name: 'unnamed')
Couldn't get registers: No such process.
(gdb) bt
Cannot fetch general-purpose registers for thread -1208490656: generic error
  Just once the bt showed something interesting (and I hope You will believe me
that I REALLY have amarok with debug info compiled in :-) ). It came from an
attempt to play a live stream. But again, the "previous frame inner..." symptom
is here as well.
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread -1208380064 (LWP 13610)]
0x06413ab2 in QPixmap::convertToImage () from /opt/qt3.2/lib/libqt-mt.so.3
(gdb) bt
#0  0x06413ab2 in QPixmap::convertToImage () from /opt/qt3.2/lib/libqt-mt.so.3
#1  0x064c5795 in QPixmap::createHeuristicMask () from /opt/qt3.2/lib/libqt-mt.so.3
#2  0x081041d5 in OSDWidget::render (this=0x8892948, M=17, size=@0xbfec2e38) at osd.cpp:284
#3  0x081034c8 in OSDWidget::show (this=0x8892948) at osd.cpp:104
#4  0x0810326a in OSDWidget::show (this=0x8892948, text=@0xbfec3220, newImage={data = 0xb69d2eb8}) at osd.cpp:69
#5  0x081050d1 in amaroK::OSD::show (this=0x8892948, bundle=@0x82542b8) at osd.cpp:562
#6  0x0809059a in App::engineNewMetaData (this=0xbfec4360, bundle=@0x82542b8) at app.cpp:747
#7  0x080e477f in EngineSubject::newMetaDataNotify (this=0x8254288, bundle=@0x82542b8, trackChanged=true) at engineobserver.cpp:81
#8  0x080e2614 in EngineController::playRemote (this=0x8254260, job=0xbfec3350) at enginecontroller.cpp:540
#9  0x080e2ff4 in EngineController::qt_invoke (this=0x8254260, _id=22, _o=0xa365de0) at enginecontroller.moc:246
#10 0x064b547a in QObject::activate_signal () from /opt/qt3.2/lib/libqt-mt.so.3
#11 0x0112bb1a in KIO::Job::result () from /opt/kde3.4/lib/libkio.so.4
#12 0x00000014 in ?? ()
Previous frame inner to this frame (corrupt stack?)
(gdb)
  When I break amarok during normal play, a very nice, correct backtrace
occurs:
Program received signal SIGINT, Interrupt.
0x011fdc77 in ___newselect_nocancel () from /lib/libc.so.6
(gdb) bt
#0  0x011fdc77 in ___newselect_nocancel () from /lib/libc.so.6
#1  0x03777fec in QEventLoop::processEvents () from /opt/qt3.2/lib/libqt-mt.so.3
#2  0x037dacdf in QEventLoop::enterLoop () from /opt/qt3.2/lib/libqt-mt.so.3
#3  0x037dab94 in QEventLoop::exec () from /opt/qt3.2/lib/libqt-mt.so.3
#4  0x037ca360 in QApplication::exec () from /opt/qt3.2/lib/libqt-mt.so.3
#5  0x08186a4f in main (argc=1, argv=0xbf82adc4) at main.cpp:94
(gdb)
  As a side effect, I've discovered a very frequent crash during amarok exiting,
which normally also doesn't show itself (no bomb, no mail):
Program received signal SIGSEGV, Segmentation fault.
0x011b545d in malloc_consolidate () from /lib/libc.so.6
(gdb) bt
#0  0x011b545d in malloc_consolidate () from /lib/libc.so.6
#1  0x011b52ed in _int_free () from /lib/libc.so.6
#2  0x011b4224 in free () from /lib/libc.so.6
#3  0x037f098d in QImage::freeBits () from /opt/qt3.2/lib/libqt-mt.so.3
#4  0x037f01ae in QImage::reset () from /opt/qt3.2/lib/libqt-mt.so.3
#5  0x037efbe9 in QImage::~QImage () from /opt/qt3.2/lib/libqt-mt.so.3
#6  0x080a2469 in ~CollectionDB (this=0x8254140) at collectiondb.cpp:113
#7  0x01180015 in exit () from /lib/libc.so.6
#8  0x0116bf46 in __libc_start_main () from /lib/libc.so.6
#9  0x08089021 in _start ()
(gdb)
  I think that all this indicates a huge memory/stack corruption.
  Please believe me that the system is otherwise very stable, average uptime
is in a magnitude of months and reboot is always caused by the kernel upgrade
or similar things :-).
  I'm sorry for such a lengthy mail. I think that it is not clear enough to
create a bug report from it :-(.
> >   Another weak moment is when the first song is to be played. The same as
> > above applies. AmaroK sometimes crashes, sometimes freezes, sometimes
> > starts to play.
> 
> Are you referring to Gst-Engine? (please always state the engine you are 
> using, it's crucial). If so, don't use alsasink, but use osssink instead. 
> Gst-Engine has just been rewritten, and alsasink support is not yet 
> functional.
I forgot to mention, sorry. I'm using Arts engine.
> 
                               With regards, Pavel Troller