khtml::CachedScript crash

[Koos Vriezen]

On Mon, 27 Jan 2003, Daniel Engelschalt wrote:

> visiting www.ufa-kino.de and javascript window open police set to "ask"
> leads
> to the following (reproducable) crash (kdelibs cvs from today):

This bug was reported earlier (same site, same crash).

[..]
> #5  <signal handler called>
> #6  khtml::CachedScript::isLoaded (this=0x0) at loader.h:238
> #7  0x01cc9027 in khtml::HTMLTokenizer::notifyFinished (this=0x83f4cd8)
>     at htmltokenizer.cpp:1613
> #8  0x01d5ac1e in khtml::CachedScript::checkNotify (this=0x8382678)
>     at loader.cpp:287
> #9  0x01d5ab82 in khtml::CachedScript::data (this=0x8382678,
>     buffer=@0x83a7274, eof=true) at loader.cpp:279
> #10 0x01d5e0a9 in khtml::Loader::slotFinished (this=0x82801d0,
> job=0x83fb400)
>     at loader.cpp:1111
> #11 0x01d60a69 in khtml::Loader::qt_invoke (this=0x82801d0, _id=2,
>     _o=0xbfffe944) at loader.moc:266
> #12 0x00b62114 in QObject::activate_signal () from
> /usr/qt/3/lib/libqt-mt.so.3
> #13 0x002b2112 in KIO::Job::result (this=0x83fb400, t0=0x83fb400)
>     at jobclasses.moc:157
> #14 0x002a1875 in KIO::Job::emitResult (this=0x83fb400) at job.cpp:183
> #15 0x002a2bb3 in KIO::SimpleJob::slotFinished (this=0x83fb400) at
> job.cpp:501
> #16 0x002a4799 in KIO::TransferJob::slotFinished (this=0x83fb400)
>     at job.cpp:782

It's one of those 'Blocking JS crashes. This is the bt with the ask dialog
still up:
#2  0x409b7833 in QEventLoop::enterLoop() ()
   from /mnt/hda7/local/koos/qt3/lib/libqt-mt.so.3
#4  0x40b557b8 in QDialog::exec() ()
   from /mnt/hda7/local/koos/qt3/lib/libqt-mt.so.3
#8  0x41cbc2c0 in KJS::Window::openWindow(KJS::ExecState*, KJS::List const&) ()
   from /mnt/hda3/usr/local/kde3/lib/libkhtml.so.4
#19 0x41bba2eb in KHTMLPart::executeScript(QString, int, DOM::Node const&,
QString const&) () from /mnt/hda3/usr/local/kde3/lib/libkhtml.so.4
#20 0x41bfbcd5 in khtml::HTMLTokenizer::scriptExecution(QString const&,
QString, int) () from /mnt/hda3/usr/local/kde3/lib/libkhtml.so.4
#21 0x41c00035 in khtml::HTMLTokenizer::notifyFinished(khtml::CachedObject*) ()

Probably another kio job event in the dialog's eventloop causes the
destruction of the tokenizer. Leaving a black hole when the dialog
finished.

IIRC, Dirk was looking in to this one.

Koos