Bug 308240 - active-documentviewer crashes by using zoom gestures
Summary: active-documentviewer crashes by using zoom gestures
Status: RESOLVED UNMAINTAINED
Alias: None
Product: Active
Classification: Plasma
Component: Applications (show other bugs)
Version: Master
Platform: Meego/Harmattan Linux
: NOR crash
Target Milestone: unscheduled
Assignee: active
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2012-10-11 13:37 UTC by Maurice de la Ferté
Modified: 2020-06-29 08:40 UTC (History)
8 users (show)

See Also:
Latest Commit:
Version Fixed In:


Attachments
Kcrash file (410 bytes, application/octet-stream)
2012-10-11 13:39 UTC, Maurice de la Ferté
Details
Okular active crash backtrace (157.40 KB, image/png)
2013-01-21 20:54 UTC, Shmerl
Details
Okular Active crash on x86 (13.73 KB, text/plain)
2013-02-05 18:13 UTC, Thomas Pfeiffer
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Maurice de la Ferté 2012-10-11 13:37:11 UTC
The active-documentviewer (books) crashes by using zoom gestures on basyskom-plasma-active-archos-gen9-omapfb-tablet-mer-testing-release-rc3 based system.

Reproducible: Always

Steps to Reproduce:
1. Start 'books' from the peek area
2. Open pre installed 'Berlin_Routes.pdf'
3. Use Muli-Touch zoom gestures
Actual Results:  
Segmentation fault

Expected Results:  
No segmentation fault and zoom in/out works.
Comment 1 Maurice de la Ferté 2012-10-11 13:39:40 UTC
Created attachment 74476 [details]
Kcrash file
Comment 2 Marco Martin 2012-10-11 14:03:50 UTC
Application: Reader (active-documentviewer), signal: Segmentation fault
[Current thread is 1 (process 910)]

Thread 1 (process 910):
#0  0x4c966d74 in nanosleep () from /lib/libc.so.6
#1  0x4c966b00 in sleep () from /lib/libc.so.6
#2  0x4e3a2408 in ?? () from /usr/lib/libkdeui.so.5
#3  0x4e3a2408 in ?? () from /usr/lib/libkdeui.so.5
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
Comment 3 Marco Martin 2012-10-11 14:11:10 UTC
would need a more detailed backtrace, but seems to be only an arm issue
Comment 4 Maurice de la Ferté 2012-10-11 15:08:14 UTC
Backtrace with installed kdelibs debuginfo package
<snip>
Application: Reader (active-documentviewer), signal: Segmentation fault
[KCrash Handler]
#6  0x404fc050 in PagePainter::scalePixmapOnImage (dest=..., src=<optimized out>, scaledWidth=<optimized out>, scaledHeight=434, cropRect=..., format=QImage::Format_ARGB32_Premultiplied) at ../../ui/pagepainter.cpp:818
#7  0x404fc920 in PagePainter::paintCroppedPageOnPainter (destPainter=0x0, page=0x0, pixID=<optimized out>, flags=41, scaledWidth=625, scaledHeight=433, limits=..., crop=..., viewPortPoint=0x0) at ../../ui/pagepainter.cpp:241
#8  0x40500854 in PagePainter::paintPageOnPainter (destPainter=0xbe9a4300, page=0x257ca8, pixID=3, flags=41, scaledWidth=625, scaledHeight=433, limits=...) at ../../ui/pagepainter.cpp:54
#9  0x405069d0 in PageItem::paint (this=<optimized out>, painter=0xbe9a4300, option=0x18a888, widget=<optimized out>) at pageitem.cpp:282
#10 0x4de33070 in ?? () from /usr/lib/libQtGui.so.4
#11 0x4de33070 in ?? () from /usr/lib/libQtGui.so.4
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
<snap>
Comment 5 Jean Cayron 2012-10-12 08:36:06 UTC
I have had such problems on Archos G9
Comment 6 Aaron J. Seigo 2012-10-12 10:27:54 UTC
I've seen similar crashes on ExoPC as well. I'm wondering if it's an overflow issue somewhere.

For zooming around at "normal" scales, I don't think to be a problem. Zooming in quite a bit, however, seems to trigger this more reliably. I notice that (desktop) Okular limits how far you can zoom in ... perhaps this would be a fix also for the touch version if it is really zooming in too far that causes the problem.
Comment 7 Maurice de la Ferté 2012-10-12 11:02:13 UTC
On Archos G9 it crashes immediately by using zoom gestures. I tried to zoom as minimal as posible.
Comment 8 Jean Cayron 2012-10-12 11:19:22 UTC
I didn't have it with all zoom gesture. I had it once when unzooming to much (I could seen the page as a "ribbon" with no end of page over en below the page).

On the Archos image there are few problems of crash also of the desktop itself, often when using the document-viewer : the systems comes to be very slow and then unresponsive (unless maybe you wait an hour, what I didn't do). Maybe is it related (maybe not). I'll see if I get such crashes without use of active-document-viewer.
Comment 9 Jean Cayron 2012-10-13 12:51:34 UTC
Maurice,

You probably tested the "Berlin routes" pdf. It crashes directly for me as well. But have you tried PDF's with just text, for example? For me, a PDF with only text takes more time to make the thing crash. I can zoom with it. 

Maybe related with the size of the content to load.
Comment 10 Thomas Pfeiffer 2012-12-06 23:09:43 UTC
This still happens reproducibly on Master with Berlin_Routes.pdf
Comment 11 Shmerl 2012-12-24 20:56:23 UTC
I have the same problem with various PDFs, with Plasma Active 3 on Nexus 7. It crashes way too often to be usable.
Comment 12 Marco Martin 2013-01-21 16:31:07 UTC
is this still the case on a current build?
 (also would be interesting to see backtraces from the nexus, since there is the possibility it may be an architecture specific thing)
Comment 13 Shmerl 2013-01-21 20:54:31 UTC
Created attachment 76615 [details]
Okular active crash backtrace
Comment 14 Shmerl 2013-01-21 20:55:03 UTC
I can't save the result from the crash handler, since current devel on Nexus 7 is way unstable - crashing on most actions. I managed to take a screenshot with ksnapshot though. See attachment:

https://bugs.kde.org/attachment.cgi?id=76615
Comment 15 Anders Lund 2013-02-04 11:12:07 UTC
After an update, I am able to zoom in text only documents, while more complex pdfs still crash. PA/mer on nexus 7
Comment 16 Anders Lund 2013-02-04 11:12:27 UTC
After an update, I am able to zoom in text only documents, while more complex pdfs still crash. PA/mer on nexus 7
Comment 17 Thomas Pfeiffer 2013-02-05 18:13:50 UTC
Created attachment 76931 [details]
Okular Active crash on x86

I can still make Okular Active crash with zooming in and out a little on Berlin_Routes.pdf on x86 (see attached crash report)
Comment 18 Thomas Pfeiffer 2013-03-26 20:34:26 UTC
I could not ae it crash on wetab with the currenti image. Can skmebody please retest on ARM?
Comment 19 Anders Lund 2013-03-26 20:36:00 UTC
Can I get that version on my nexus 7 to test?
Comment 20 Anders Lund 2013-03-27 08:20:55 UTC
Updating the nexus 7 (zypper ref && zypper up) does not get me anything new, and the current okular is reliably crashy.
Comment 21 Shmerl 2013-04-30 01:03:16 UTC
It still crashes for me. Nexus 7 / PA current devel image. I'm not sure if it's Nexus 7 or ARM specific - I don't have other hardware. This really should be  a blocker for PA 4, since bookreader is a very important application. At least for me, it's 95% of all use cases of PA at present, and it being completely crashy makes PA of very little use.
Comment 22 Thomas Pfeiffer 2013-04-30 22:30:20 UTC
I have not been able to reproduce it for a while on Wetab, so it seems to be specific to ARM or Nexus 7.
Now the question is: Do we release with architecture-specific blockers or not?
Comment 23 Aaron J. Seigo 2013-05-01 09:22:14 UTC
In this case, as our primary (as in: upstream supported) image will be for x86 and PA4 is rather overdue already *and* this bug was in previous releases, I'm fine with shipping with this flaw. It would be nicer to see it fixed, but the cause is not yet known (though some of us have looked into it...)
Comment 24 Thomas Pfeiffer 2013-08-06 15:37:13 UTC
I could reproduce the crash with the latest image on x86 :(
Comment 25 Marco Martin 2013-08-07 10:04:10 UTC
with which file it happens? if is not huge could be attached here?
Comment 26 Anders Lund 2013-08-07 10:07:43 UTC
In my case it happened with any file. I don't use PA at the moment though, so I have no idea about potential improvements though :)
Comment 27 Thomas Pfeiffer 2013-08-07 10:35:17 UTC
I could reproduce it with this one: 
http://www.gastonsanchez.com/PLS_Path_Modeling_with_R.pdf
Not immediately, but after zooming in pretty far, zooming out again and repeating that a few times it always crashed at some point
Comment 28 Marco Martin 2013-08-09 11:39:04 UTC
the file crashes in desktop okular as well, slightly different bt (this is from master):

Application: Okular (okular), signal: Aborted
Using host libthread_db library "/lib/libthread_db.so.1".
[KCrash Handler]
#7  0xb7767424 in __kernel_vsyscall ()
#8  0xb57c931f in raise () from /lib/libc.so.6
#9  0xb57cac03 in abort () from /lib/libc.so.6
#10 0xb5b76e5e in qt_message_output(QtMsgType, char const*) () from /usr/lib/libQtCore.so.4
#11 0xb5b77059 in ?? () from /usr/lib/libQtCore.so.4
#12 0xb5b77178 in qFatal(char const*, ...) () from /usr/lib/libQtCore.so.4
#13 0xb5b771f5 in qt_assert(char const*, char const*, int) () from /usr/lib/libQtCore.so.4
#14 0xb365ac97 in Okular::DocumentPrivate::sendGeneratorPixmapRequest (this=0x88a51c8) at /home/diau/git/kdegit/KDE/kdegraphics/okular/core/document.cpp:1104
#15 0xb366ba79 in Okular::Document::qt_static_metacall (_o=0x883f240, _c=QMetaObject::InvokeMetaMethod, _id=22, _a=0xbfcf3cdc) at /home/diau/git/kdegit/KDE/kdegraphics/okular/build/document.moc:133
#16 0xb5cacac2 in QMetaObject::activate(QObject*, QMetaObject const*, int, void**) () from /usr/lib/libQtCore.so.4
#17 0xb5cb3cb5 in ?? () from /usr/lib/libQtCore.so.4
#18 0xb5cb3d7c in ?? () from /usr/lib/libQtCore.so.4
#19 0xb5caa894 in QObject::event(QEvent*) () from /usr/lib/libQtCore.so.4
#20 0xb6381b9c in QApplicationPrivate::notify_helper(QObject*, QEvent*) () from /usr/lib/libQtGui.so.4
#21 0xb638659f in QApplication::notify(QObject*, QEvent*) () from /usr/lib/libQtGui.so.4
#22 0xb6f1b6b1 in KApplication::notify (this=0xbfcf44e8, receiver=0x89ce368, event=0xbfcf4220) at /home/diau/kdesrc/kde/kdelibs/kdeui/kernel/kapplication.cpp:311
#23 0xb5c9476e in QCoreApplication::notifyInternal(QObject*, QEvent*) () from /usr/lib/libQtCore.so.4
#24 0xb5cc9047 in ?? () from /usr/lib/libQtCore.so.4
#25 0xb5cc60c8 in ?? () from /usr/lib/libQtCore.so.4
#26 0xb4dc47d3 in g_main_context_dispatch () from /usr/lib/libglib-2.0.so.0
#27 0xb4dc4b70 in ?? () from /usr/lib/libglib-2.0.so.0
#28 0xb4dc4c51 in g_main_context_iteration () from /usr/lib/libglib-2.0.so.0
#29 0xb5cc6741 in QEventDispatcherGlib::processEvents(QFlags<QEventLoop::ProcessEventsFlag>) () from /usr/lib/libQtCore.so.4
#30 0xb643348a in ?? () from /usr/lib/libQtGui.so.4
#31 0xb5c9323c in QEventLoop::processEvents(QFlags<QEventLoop::ProcessEventsFlag>) () from /usr/lib/libQtCore.so.4
#32 0xb5c93531 in QEventLoop::exec(QFlags<QEventLoop::ProcessEventsFlag>) () from /usr/lib/libQtCore.so.4
#33 0xb5c987ea in QCoreApplication::exec() () from /usr/lib/libQtCore.so.4
#34 0xb637fa14 in QApplication::exec() () from /usr/lib/libQtGui.so.4
#35 0x0804fd22 in main (argc=2, argv=0xbfcf4634) at /home/diau/git/kdegit/KDE/kdegraphics/okular/shell/main.cpp:94
Comment 29 Albert Astals Cid 2013-08-12 21:00:33 UTC
Marco which assert are you getting? Because on that line you have in your bt i can only see a call to warnLimitedAnnotSupport();