Frank: the attachment is missing, could you try to use this link https://bugs.freedesktop.org/attachment.cgi?bugid=68333&action=enter ?

I'm not sure why I didn't see the error the first time, but using your link bugzilla complains that the file size (4706 MB) exceeds the 3000 MB limit.

If you like, I can mail it to you personally, or ftp it to a site of your choice.

Frank: would it be possible you make a copy and reduce the size of the file? (BTW, an attachment to the bugtracker is public).

Frank: you can also upload the file by using this link:
https://wiki.documentfoundation.org/QA/Bugzilla-Attachments

Created attachment 84654 [details]
Presentation causing error part 1/2

Part 1 of 2 files created from the presentation via "split -- bytes=2500K" to circumvent 3M maximum.

Created attachment 84657 [details]
Presentation causing error part 2/2

Part 2 of 2.

OK, Julien, just recombine the pieces via:

cat error.pps.1 error.pps.2 >error.pps

On pc Debian x86-64 with master sources updated today or 4.1 sources updated some days ago + brand new LO profile, I don't reproduce crash or hanging.

Now I had several logs:
warn:avmedia:5423:1:avmedia/source/viewer/mediawindowbase_impl.cxx:82: failed to create media player service com.sun.star.comp.avmedia.Manager_GStreamer
warn:legacy.osl:5423:1:slideshow/source/engine/shapes/drawshape.cxx:351: DrawShape::getActualUnitShapeBounds(): bounds noticeably larger than original shape - clipping!
warn:legacy.osl:5423:1:slideshow/source/engine/shapes/drawshape.cxx:1084: DrawShape::leaveAnimationMode(): called on DrawShape without views
warn:legacy.osl:5423:1:basegfx/source/range/b2drangeclipper.cxx:272: ImplPolygon::append(): added point violates 90 degree line angle constraint!

Could you try 2 things:
- rename your LO directory profile (see https://wiki.documentfoundation.org/UserProfile)
- test prerelease 4.1.1RC2 (see http://www.libreoffice.org/download/pre-releases/)

No go.  I renamed ./config/libreoffice to ./config/libreoffice-orig, ran, got the same hang and a new ./config/libreoffice.

Then I installed the beta version from your link and repeated the test with both a copy of my existing .config/libreoffice and with no ./config/libreoffice, and still got the hang.

I am having a problem with sound on my system at the moment, and LO puts out /stdout/stderr messages noting that it can't start the jack server because of problems with alsa:

[ftg@ftgme2 ~]$ soffice

(soffice:17988): GLib-GObject-WARNING **: type name 'avmux_stream_segment,ssegment' contains invalid characters

(soffice:17988): GLib-GObject-CRITICAL **: g_type_set_qdata: assertion 'node != NULL' failed

(soffice:17988): GLib-GObject-CRITICAL **: g_type_add_interface_static: assertion 'G_TYPE_IS_INSTANTIATABLE (instance_type)' failed

(soffice:17988): GStreamer-CRITICAL **: gst_element_register: assertion 'g_type_is_a (type, GST_TYPE_ELEMENT)' failed
AL lib: (EE) alsa_open_playback: Could not open playback device 'default': Connection refused
[ftg@ftgme2 ~]$ 

but all of these messages only show up after I ALT-F4 the window and tell KDE to terminate it.  The presentation does have an audio component, so possibly the problem occurs because of the sound failure.

Testing further, there may be two problems here.  

When I test on the machine having the sound problem, impress hangs after filling in the main window but before filling in the left-side thumbnails.  Clicking the X icon to close it has no effect, but multiple ALT-F4 will reliably trigger the KDE prompt to terminate the application.

If I test on a machine without sound problems, I get the solid hang which doesn't even respond to multiple ALT-F4.  However, CTRL-ALT-DEL will trigger the KDE logout sequence, and there is a standalone vertical slider on the right side of the screen from top to bottom, so impress may be in the process of resizing/redrawing itself, and it may be that the impress window does not have focus and whatever does does not respond to ALT-F4.

I had tested on both machines before filing this bug, which is why I reported that sometimes it crashed and sometimes it hung so bad that it had to be terminated from a tty, but it's possible that the sound system error is causing a "soft" lockup of its own, with the "hard" lockup only happening on a machine where the sound problem doesn't occur.

Michael: I'm a bit stuck here. Any idea?

I guess we need to get a version with debugging symbols installed on Frank's machine, and when it hangs attach gdb:

ps ax | grep soffice.bin
gdb
attach <pid-of-soffice.bin>
thread apply all backtrace

and paste the output here [~useless without debugging symbols for gstreamer & libreoffice ].

Much of the multimedia playback is synchronous in LibreOffice; so - I can well believe it hangs the suite if something bad happens.

Also a list of installed gstreamer packages would be rather helpful.

Michael: badfully, http://dev-builds.libreoffice.org/daily/master/Linux-Fedora17-x86_64@4-gcc-4.7-dbgutil/ is empty

Frank: perhaps you may give a try to master sources (= future 4.2.0) by using one of the non empty+Linux daily master build? (see http://dev-builds.libreoffice.org/daily/master/)
If you still reproduce the crash with master sources + have some time (+ feel like of course:-)), I would suggest you build LO by using this link (see https://wiki.documentfoundation.org/Development/How_to_build)
Notice: after "./autogen.sh", add this line in "autogen.input" file
--enable-dbgutil

I've confined my testing to the system without the sound problem, which is now running the Mageia 4.1.1.2 RPM.  In this environment, if I run soffice with --backtrace, the bug hangs X.  If I run it without, it gets a segfault and closes.

So what I did was to install the debug package, start soffice, let it get to the "Finish" prompt, switch to a tty, attach gdb, and then click Finish.  Switching to the tty, I could see that a SIGSEGV had occurred.  Here is the backtrace:

#0  0x00007fb3f6ec23f7 in GetGLXScreenConfigs () from /lib64/libGL.so.1
#1  0x00007fb3f6ec271e in glXCreateContext () from /lib64/libGL.so.1
#2  0x00007fb3cb4447e4 in (anonymous namespace)::OGLTransitionerImpl::initWindowFromSlideShowView (this=this@entry=0x7fb3e403e9c0, xView=...)
    at /usr/src/debug/libreoffice-4.1.1.2/slideshow/source/engine/OGLTrans/generic/OGLTrans_TransitionerImpl.cxx:683
#3  0x00007fb3cb4488dc in initialize (xEnteringSlide=..., xLeavingSlide=..., xView=..., this=0x7fb3e403e9c0)
    at /usr/src/debug/libreoffice-4.1.1.2/slideshow/source/engine/OGLTrans/generic/OGLTrans_TransitionerImpl.cxx:380
#4  (anonymous namespace)::OGLTransitionFactoryImpl::createTransition (this=<optimized out>, transitionType=<optimized out>, transitionSubType=104, view=..., 
    leavingBitmap=..., enteringBitmap=...) at /usr/src/debug/libreoffice-4.1.1.2/slideshow/source/engine/OGLTrans/generic/OGLTrans_TransitionerImpl.cxx:1847
#5  0x00007fb3cb7bcdd6 in slideshow::internal::(anonymous namespace)::PluginSlideChange::addTransition (this=this@entry=0x3e45400, rView=...)
    at /usr/src/debug/libreoffice-4.1.1.2/slideshow/source/engine/transitions/slidetransitionfactory.cxx:180
#6  0x00007fb3cb7be4bb in PluginSlideChange (rEventMultiplexer=..., pSoundPlayer=..., xFactory=..., rScreenUpdater=..., rViewContainer=..., pEnteringSlide=..., 
    leavingSlide_=..., nTransitionSubType=104, nTransitionType=37, this=0x3e45400)
    at /usr/src/debug/libreoffice-4.1.1.2/slideshow/source/engine/transitions/slidetransitionfactory.cxx:149
#7  createPluginTransition (rEventMultiplexer=..., pSoundPlayer=..., xFactory=..., rScreenUpdater=..., rViewContainer=..., pEnteringSlide=..., pLeavingSlide=..., 
    nTransitionSubType=104, nTransitionType=37) at /usr/src/debug/libreoffice-4.1.1.2/slideshow/source/engine/transitions/slidetransitionfactory.cxx:919
#8  slideshow::internal::TransitionFactory::createSlideTransition (pLeavingSlide=..., pEnteringSlide=..., rViewContainer=..., rScreenUpdater=..., rEventMultiplexer=..., 
    xOptionalFactory=..., nTransitionType=37, nTransitionSubType=104, bTransitionDirection=true, rTransitionFadeColor=..., pSoundPlayer=...)
    at /usr/src/debug/libreoffice-4.1.1.2/slideshow/source/engine/transitions/slidetransitionfactory.cxx:975
#9  0x00007fb3cb78eeeb in (anonymous namespace)::SlideShowImpl::displaySlide (this=0x7fb40ea85608, xSlide=..., xDrawPages=..., xRootNode=..., rProperties=...)
    at /usr/src/debug/libreoffice-4.1.1.2/slideshow/source/engine/slideshowimpl.cxx:828
#10 0x00007fb3dcc08119 in sd::AnimationSlideController::displayCurrentSlide (this=0x3d70820, xShow=..., xDrawPages=..., 
    bSkipAllMainSequenceEffects=bSkipAllMainSequenceEffects@entry=false) at /usr/src/debug/libreoffice-4.1.1.2/sd/source/ui/slideshow/slideshowimpl.cxx:492
#11 0x00007fb3dcc09fce in sd::SlideshowImpl::displayCurrentSlide (this=this@entry=0x7fb3e459ddc0, bSkipAllMainSequenceEffects=bSkipAllMainSequenceEffects@entry=false)
    at /usr/src/debug/libreoffice-4.1.1.2/sd/source/ui/slideshow/slideshowimpl.cxx:1390
#12 0x00007fb3dcc0a809 in sd::SlideshowImpl::displaySlideIndex (this=this@entry=0x7fb3e459ddc0, nSlideIndex=<optimized out>)
    at /usr/src/debug/libreoffice-4.1.1.2/sd/source/ui/slideshow/slideshowimpl.cxx:1702
#13 0x00007fb3dcc0acd2 in sd::SlideshowImpl::startShowImpl (this=this@entry=0x7fb3e459ddc0, aProperties=...)
    at /usr/src/debug/libreoffice-4.1.1.2/sd/source/ui/slideshow/slideshowimpl.cxx:1159
#14 0x00007fb3dcc0da80 in sd::SlideshowImpl::startShow (this=0x7fb3e459ddc0, pPresSettings=<optimized out>)
    at /usr/src/debug/libreoffice-4.1.1.2/sd/source/ui/slideshow/slideshowimpl.cxx:1096
#15 0x00007fb3dcbfd0de in sd::SlideShow::activate (this=0x7fb3d5ce78c8, rBase=...) at /usr/src/debug/libreoffice-4.1.1.2/sd/source/ui/slideshow/slideshow.cxx:1051
#16 0x00007fb3dcd8cc6b in sd::PresentationViewShell::Activate (this=0x3d22f30, bIsMDIActivate=<optimized out>)
    at /usr/src/debug/libreoffice-4.1.1.2/sd/source/ui/view/presvish.cxx:144
#17 0x00007fb40c019e81 in ?? ()
#18 0x00000000035d3948 in ?? ()
#19 0x0000000000000000 in ?? ()
A debugging session is active.


Since GLX seems to be involved, it may be relevant to know that I'm running with the Mageia lib64xcb-glx0 package at level 1.9.1.  I recall that the glx requirement can be satisfied by at least one other package, so that may be the difference in our systems.

ah - that crash is a dup of an existing OpenGL crasher; remove the 3D/GL transitions - if it doesn't crash, then it's a dup of an existing bug with similar trace you could hunt for and mark this bug a dup of that one please.

Thanks.

(In reply to comment #14)
> remove the 3D/GL
> transitions 

Please translate :0  Exactly how do I do this ?

No repro using LO 4.2.0.0.beta2 + Ubuntu 12.04.3 (x86_64)

Entire presentation plays through.

I'm confused.

Both files seem to be damaged. They don't open with MS Office 2011 on OS X 10.9. Says "damaged".

I read "recombine" the attachments? Can someone able to, merge those to a proper ppt file and attach that (pls remove the unusable files in that case).

(In reply to comment #17)
> I'm confused.
> 
> Both files seem to be damaged. They don't open with MS Office 2011 on OS X
> 10.9. Says "damaged".
> 
> I read "recombine" the attachments? Can someone able to, merge those to a
> proper ppt file and attach that (pls remove the unusable files in that case).

This bugzilla doesn't allow attachments as large as the full file.  Follow the directions in comment#7 to recombine the two fragments into a usable file.

I confirm that this is fixed in 4.1.3.2; closing

Franks, thanks a lot for the update on this.

Also +1 from me for great bug title!

guess WORKSFORME is the correct state, since there's no specific commit fixing this we are aware of.