konqueror die with Broken pipe

[Paul Chitescu]

Low Swee Leong-A16957 <Swee.leong.Low at motorola.com> wrote:
>     Connected to VFB server: 240 x 320 x 16
>     debug: KFlashFactory::KFlashFactory
>     Launcher::dispatchLoop()...

>     QGList::locate: Index -1 out of range
>     QGList::locate: Index -1 out of range

These two look really suspect... It may be the initial failure point.

>     debug: KHTMLPart::init this=0x85183f8 d=0x8516ba8

Here it should have been a line like this one:
debug: creating slave for file:/usr/share/apps/konq-embed/start.html

The fact there is none indicates things are already screwed up.

>     launcher: creating slave for protocol

This may be caused by the initial problem. It should say "for protocol file"
or whatever.

>     Segmentation fault

This is fatal, it occurs somewhere in the main, interactive, process. This
should be debugged first.

>     launcher: error sending launcherSlavePidCmd: Broken pipe

The pipe is broken to the main process (which segfaulted) when the launcher
tries to reply with the pid of the newly created KIO slave.

>     konqueror: launcher.cpp:313: bool KIO::Launcher::createSlaveInternal
(const char *, cmsghdr *): Assertion `slave' failed.

The SlaveBase::createSlave failure (normal, since there's no protocol name)
is noticed and the KIO slave process aborts since there's nothing to do.
This is normal and happens after konqueror already crashed anyway.

>
>     does anyone know whats that error means ?
>     what is launcher doing ?
>     is segmentation fault causing launcher to die or launcher causing
segmentation fault ?
>     it dies when the creating a internal slave when the protocol is
nothing. why is that happening ? why it call that function while starting ?

You should either run under gdb or run gdb on the generated core file. You
must be able at least to generate a backtrace.

You should also consider configuring with --enable-debug=full which will
create a large binary with full debugging info (including file names and
line numbers) and also enable almost all messages.

--
 Paul Chitescu
 pchitescu at null.ro   http://pchitescu.null.ro/   ICQ:22641673
 Any spammers will be painfully squeezed into /dev/null