Bug 212305

Summary: Okular creates disk activity which DoS the computer
Product: [Applications] okular Reporter: Nicolas Bigaouette <nbigaouette>
Component: generalAssignee: Okular developers <okular-devel>
Status: RESOLVED NOT A BUG    
Severity: normal CC: aacid, chaofeng111
Priority: NOR    
Version: 0.9.2   
Target Milestone: ---   
Platform: Arch Linux   
OS: Linux   
Latest Commit: Version Fixed In:
Sentry Crash Report:

Description Nicolas Bigaouette 2009-10-29 18:08:09 UTC
Version:           0.9.2 (using KDE 4.3.2)
Compiler:          gcc version 4.4.2 (GCC) ./configure --prefix=/usr --enable-shared --enable-languages=c,c++,fortran,objc,obj-c++,ada --enable-threads=posix --mandir=/usr/share/man --infodir=/usr/share/info --enable-__cxa_atexit --disable-multilib --libdir=/usr/lib --libexecdir=/usr/lib --enable-clocale=gnu --disable-libstdcxx-pch --with-tune=generic
OS:                Linux
Installed from:    Archlinux Packages

I have a recuring problem on my laptop since maybe a year. Today I was able to reproduce the problem.

The problem is: Everything runs fine, but at some point, the HD will go crazy and nothing is responsive anymore. The only way to save it is to try to detect when it starts and before the computer is completely down and reboot it by pressing the power button (controled by acpid I think) or switching to a virtual terminal and hitting ctrl+alt+del for a reboot. Most of the time I am not fast enough and the computer does not respond anymore, with the harddrive activity led going crazy.

Now to reproduce the problem, I did the following, leading to think the problem might be with okular:

I have suspended and resumed my computer once yesterday evening. While suspended, I docked it. I then realized my keyboard did not have battery anymore, so I hot-undocked it. Now I have been working on a latex file compiled to pdf. To reproduce the problem, what I need to do is to view the pdflatex generated file in okular. I then open up a second pdf file (Chapter 2 of http://www.springerlink.com/content/q1377k/ ) and search for "spherical ha". I can't remember if I need to search the next occurence or if the first one is enough, but that triggers the disk activity. Everything seems to freeze, as if the system was swapping like crazy, even though I don't have swap enabled at all:
free -m
             total       used       free     shared    buffers     cached
Mem:          2000       1433        567          0         14        517
-/+ buffers/cache:        900       1099
Swap:            0          0          0

The output of "dstat -cdnpmgsf -M topcpu,topbio" is located at http://pastebin.ca/1648014
(I started that command, in a virtual terminal, went back to X to toggle the bug, the came back in the console while checking the output of the command. Then switched to a second console where "ps aux | grep -i okular && pkill okular" was already entered and was waiting for the "enter" key to run.)

Hardware:
Dell Latitude D830
Core2 Duo T7300 @ 2GHz
2 GB ram
120 GB drive partitioned like this:
  sda1: Dell utility
  sda2: Old, unused XP partition
  sda3: boot ~100MB
  sda4: extended
  sda5: 2GB swap (disabled)
  sda6: ~10GB ext3 /
  sda7: ~90GB ext3 /home
Intel X3100 (965 something, never understood the codes after)

Software:
x86_64
kernel26 2.6.31.5
xf86-video-intel 2.8.1
KDE 4.3.2 with kwin effects on
KMS with "i915.modeset=1" kernel option
Everything else up to date... Note that this problem is happening since a couple of months, at least. So its not only these versions that are affected...

See also my post on the ArchLinux forum:
http://bbs.archlinux.org/viewtopic.php?pid=646148

Thank you
Comment 1 Albert Astals Cid 2009-10-29 23:30:48 UTC
Can you always reproduce the problem doing that? Or was it only just one time?
Comment 2 Nicolas Bigaouette 2009-10-29 23:51:20 UTC
(In reply to comment #1)
> Can you always reproduce the problem doing that? Or was it only just one time?
I was able to reproduce it three times this morning by doing exactly that. I think I tried a fourth time when I was searching for the exact way but did not triggered the problem because I did not do exactly the steps I described.

I remember also that the problem happened when I was reading pdf and scrolling them as fast as I could do. I think it was easier to trigger the problem when the pdf was a scan (so quite big). My feeling is that okular is trying to cache the file or something like that and for some reason goes crazy...
Comment 3 Albert Astals Cid 2009-10-30 00:35:03 UTC
The problem is that we don't have that file to try to reproduce
Comment 4 Nicolas Bigaouette 2009-10-30 00:38:09 UTC
Ok.

I'll try to find another pattern with publicly available files. I will report back in the next few days.
Comment 5 Chao Feng 2013-05-04 13:50:23 UTC
User not feedback for a long time. Set status to Needs info.
Comment 6 Albert Astals Cid 2014-05-08 09:50:19 UTC
No answer in more than 6 months