LinuxQuestions.org - polkit-kde-auth and libglib-2.0.so.0.2800.6

- Slackware (https://www.linuxquestions.org/questions/slackware-14/)

- - polkit-kde-auth and libglib-2.0.so.0.2800.6 (https://www.linuxquestions.org/questions/slackware-14/polkit-kde-auth-and-libglib-2-0-so-0-2800-6-a-899229/)

polkit-kde-auth and libglib-2.0.so.0.2800.6

So I have done some research regarding the following error:

Code:

oscuridad kernel: [155226.902615] polkit-kde-auth[30098] general protection ip:7f7ec1dce698 sp:7fff39343d80 error:0 in libglib-2.0.so.0.2800.6[7f7ec1da7000+115000]

And apparently there is an upgraded pkg for polkit-kde-auth under 13.37 current that fixes the error listed above...

Is there an upgrade pkg for polkit-kde-auth on the 13.37?

I can seem to find it at all for 13.37

Thank You,
-

This issue has been happening for me as well on Slackware 13.37 too.

On top of that, my harddrive makes a lot of read/writes sounds at the time this happens.

And this is on every Slackware 13.37 box that I have (e.g. different machines) with a clean install and and a clean home.

So this is some sort of bug/snafu/glitch/what-have-you in Slackware, imo and I'd *LOVE* to know how to fix this.

On one PC I get this:

[ 189.803657] start_kdeinit (2203): /proc/2203/oom_adj is deprecated, please use /proc/2203/oom_score_adj instead.
[ 196.568039] ata1.00: configured for UDMA/133
[ 196.568049] ata1: EH complete
[ 218.678574] polkit-kde-auth[2337]: segfault at c ip b60ac524 sp bffe8560 error 4 in libc-2.13.so[b603a000+15c000]

But like I said this happens on all Slack boxes for me.

What's worse is that I have no idea what causes it. It seems to be random.

This happens on the huge-smp kernel and on the intrd-generic-smp kernels.

Hopefully some one will comment on a fix for this :)

There is no "polkit-kde-auth" however there is a "polkit-kde-1" package in Slackware 13.37. This software is abandoned by its creators, and its functionality has been split-off into two new packages: "polkit-kde-kcmodules-1" and "polkit-kde-agent-1". The old software does not receive maintenance, that is why these crash bugs still exist.

However these two new packages are not compatible with the KDE 4.5.x like you find in Slackware 13.37. You will have to upgrade to KDE 4.6.x or 4.7.x (packages for which you find in my own "ktown" repository, http://alien.slackbook.org/ktown/), the new polkit related packages are included there.

Eric

Hello Alien Bob ! :) Thanks for the reply. I'll give that a shot on my test box to make sure I get the directions correctly. I'll post back to let you know how I make out with it. Thanks again. Fogie

Eric,

I am still seeing the same errors (yet in a different order) on a completely clean default full installation.

Here is what I'm seeing:

Boot up in run level 3, the tail of dmesg:

[ 10.246136] EXT4-fs (sda7): re-mounted. Opts: (null)
[ 10.353347] EXT4-fs (sda7): re-mounted. Opts: (null)
[ 10.728691] lp0: using parport0 (interrupt-driven).
[ 10.730784] lp0: console ready
[ 15.155692] eth0: link up, 100Mbps, full-duplex, lpa 0x45E1
[ 26.374991] cx18-0: Failed to initialize on vbi0
[ 26.383401] cx18-0: Failed to initialize on video0
[ 26.389967] cx18-0: Failed to initialize on video24
[ 26.393303] cx18-0: Failed to initialize on video32

Please ignore the messages about the cx18, as that simply means that I don't have the firmatware installed for my tv tuner, which as I said, this is a default full installation.

Now, as root I issue 'telinit 4' and here is the tail of dmesg:

[ 10.246136] EXT4-fs (sda7): re-mounted. Opts: (null)
[ 10.353347] EXT4-fs (sda7): re-mounted. Opts: (null)
[ 10.728691] lp0: using parport0 (interrupt-driven).
[ 10.730784] lp0: console ready
[ 15.155692] eth0: link up, 100Mbps, full-duplex, lpa 0x45E1
[ 26.374991] cx18-0: Failed to initialize on vbi0
[ 26.383401] cx18-0: Failed to initialize on video0
[ 26.389967] cx18-0: Failed to initialize on video24
[ 26.393303] cx18-0: Failed to initialize on video32

So no errors yet.

Here is tail of dmesg after using logs into kdm and launches kde 4.6.5 without only the files in their home issued by Slackware's /etc/skeleton files:

[ 10.246136] EXT4-fs (sda7): re-mounted. Opts: (null)
[ 10.353347] EXT4-fs (sda7): re-mounted. Opts: (null)
[ 10.728691] lp0: using parport0 (interrupt-driven).
[ 10.730784] lp0: console ready
[ 15.155692] eth0: link up, 100Mbps, full-duplex, lpa 0x45E1
[ 26.374991] cx18-0: Failed to initialize on vbi0
[ 26.383401] cx18-0: Failed to initialize on video0
[ 26.389967] cx18-0: Failed to initialize on video24
[ 26.393303] cx18-0: Failed to initialize on video32
[ 169.652832] ata1.00: configured for UDMA/133
[ 169.652843] ata1: EH complete
[ 170.133156] EXT4-fs (sda7): re-mounted. Opts: commit=0
[ 187.565128] NET: Registered protocol family 10
[ 187.565555] lo: Disabled Privacy Extensions
[ 198.346019] eth0: no IPv6 routers present

You'll notice that libata is issuing a error code for the hard drive, and resets the drive.

Now the tail of dmesg on logout:

[ 10.246136] EXT4-fs (sda7): re-mounted. Opts: (null)
[ 10.353347] EXT4-fs (sda7): re-mounted. Opts: (null)
[ 10.728691] lp0: using parport0 (interrupt-driven).
[ 10.730784] lp0: console ready
[ 15.155692] eth0: link up, 100Mbps, full-duplex, lpa 0x45E1
[ 26.374991] cx18-0: Failed to initialize on vbi0
[ 26.383401] cx18-0: Failed to initialize on video0
[ 26.389967] cx18-0: Failed to initialize on video24
[ 26.393303] cx18-0: Failed to initialize on video32
[ 169.652832] ata1.00: configured for UDMA/133
[ 169.652843] ata1: EH complete
[ 170.133156] EXT4-fs (sda7): re-mounted. Opts: commit=0
[ 187.565128] NET: Registered protocol family 10
[ 187.565555] lo: Disabled Privacy Extensions
[ 198.346019] eth0: no IPv6 routers present
[ 300.232473] akonadi_agent_l[2478]: segfault at b485eaf8 ip b70b74e2 sp bfc39ce0 error 4 in libQtGui.so.4.7.0[b6d30000+a80000]
[ 302.343905] akonadi_agent_l[2569]: segfault at 0 ip b6bee360 sp bf86ded0 error 4 in libQtCore.so.4.7.0[b6a6f000+293000]

Here akonadi segfaults.

So it appears that this version of KDE still is whacking my hard drive and, instead of segfaulting at the initial login to KDE, now it segfaults on the exit of KDE :)

Perhaps I need to move up to Slackware current and see if the newer 4.7.X series of KDE will fix this for me or not.

This computer I tested on is a relatively new hard drive, 2 terbytes in size and smartctl is reporting no issues on the drive at all.

And like I had said previously, I see the libata/reinitiliazing on a few other Slack boxes as well. So I don't suspect it's my hardware. On my netbook, that hard drive is reinitialized/erroring every time we even resume from suspend to disk. But the computer feels flawless, and Slackware sleep/hibernates perfectly IMO, except this matter (which maybe is libata?? or KDE ??) I can't explain it.

Perhaps when I get some time over the next few days I'll give it a go at --current and see how that goes.

Other than that, I must say that despite the hard drive matter, this kde 4.6 is very nice, and may of the fixes that they did are very welcome here for sure :)

Thanks again for you help.

Fogie

Hi Fogie

On Slackware 13.37 you can safely install my KDE 4.6.5 packages (http://alien.slackbook.org/ktown/4.6.5/), they were built on that version. In my opinion 4.6.5 is very stable, and the polkit issues are gone. And even 4.7.3 will work according to the people who tried it on 13.37.

Eric

I'm still running KDE 4.6.5 and also experienced the akonadi segfault at logout.
This patch cleared up the issue for me:
https://projects.kde.org/projects/kd...9a22b91eb9a02c

Code:

commit c8e45b10c32679bf070af5b7ad9a22b91eb9a02c

Author: Sergio Martins <iamsergio[at]gmail[dot]com>

Date:  Mon Sep 12 13:48:33 2011 +0100



    Fix crash on agent_launcher exit.

    

    Don't unload plugins here. They will be unloaded on application exit,

    so it's redundant.

    

    Seems they are being unloaded too early, before QThreadStorage

    cleanup occurs:

    

    ==18802== Invalid read of size 8

    ==18802==    at 0x53178AE: QThreadStorage<QFontCache*>::deleteData(void*) (in

    /data/installation/qt/qt-4.7/lib/libQtGui.so.4.7.1)

    ==18802==    by 0x5E8F8CF: QThreadStorageData::set(void*)

    (qthreadstorage.cpp:165)

    ==18802==    by 0x5317940: void

    qThreadStorage_setLocalData<QFontCache>(QThreadStorageData&, QFontCache**)

    (qthreadstorage.h:92)

    ==18802==    by 0x53165BA:

    QThreadStorage<QFontCache*>::setLocalData(QFontCache*) (qthreadstorage.h:148)

    ==18802==    by 0x531407A: QFontCache::cleanup() (qfont.cpp:2632)

    ==18802==    by 0x5406536: QFont::cleanup() (qfont_x11.cpp:182)

    ==18802==    by 0x5104E35: qt_cleanup() (qapplication_x11.cpp:2647)

    ==18802==    by 0x50603B2: QApplication::~QApplication()

    (qapplication.cpp:1172)

    ==18802==    by 0x41B1F6: main (agentlauncher.cpp:28)

    ==18802==  Address 0x11a090b0 is not stack'd, malloc'd or (recently) free'd

    

    CCBUG: 261788



diff --git a/agentserver/agentpluginloader.cpp b/agentserver/agentpluginloader.cpp

index 2d62d3b..456a124 100644

--- a/agentserver/agentpluginloader.cpp

+++ b/agentserver/agentpluginloader.cpp

@@ -25,13 +25,8 @@ using namespace Akonadi;

 

 AgentPluginLoader::~AgentPluginLoader()

 {

-  QHash<QString, QPluginLoader*>::iterator it = m_pluginLoaders.begin();

-  while ( it != m_pluginLoaders.end() ) {

-    it.value()->unload();

-    ++it;

-  }

-

  qDeleteAll( m_pluginLoaders );

+  m_pluginLoaders.clear();

 }

 

 QPluginLoader *AgentPluginLoader::load( const QString &pluginName )

Well that's pretty awesome; Eric applied the patch I found.
Many thanks Eric.

If this helps any KDE 4.6.5 users (or not), please leave feedback. :)

Interesting patch. It is included in the akonadi 1.6.2 package which is part of my KDE 4.7.3 "deps set". However the akonadi 1.5.3 included with my KDE 4.7.3 packages did not have this patch.

I applied the patch to the akonadi package in my KDE 4 6.5 repository and I am curious to hear if that solves your issues (for those who are running my KDE 4.6.5).
See also my RSS feed http://alien.slackbook.org/ktown/ChangeLog.rss to which you can subscribe if you want to be informed of future ktown updates.

Eric

Alien Bob,

Just wanted to report back to you on my experiences.

1. The new akonadi (patched) for 4.6.5 is a success for me here, after about 20 log in/outs and a few reboots as well.

Thank you so much !

2. Regarding the hard drives being reset over and over by KDE I now understand what's happening, and have a fix that I can live with.

KDE on log in/out, resume from suspend or hibernate, no matter what, pokes ACPI daemon, which in turn runs the

/usr/lib/pm-utils/power.d/

scripts of:

harddrive

and

journal-commit.

These are indeed what keeps reseting the hard drives.

Now interestingly enough, if you disable KDE's powermanagement daemon, the drives still error (with these two files still in place).

So disabling KDE power daemon isn't the culprit.

Turning off acpi in /etc/rc.d/ as a test doesn't even allow KDE to run, which is what pointed me to KDE being wonky here.

A nice google on kde resetting hard drives turned up some suggestongs such as "touch /etc/pm/power.d/journal-commit"

But that didn't work here.

Reviewing this journal-commit file we see it's supposed to only do ext3 , then later well they do ext4. But here's the thing, most of my machines are jfs, and it does it on them as well. So pm-utils package seems to be doing something it's not supposed to.

Anyway, I've removed these two files, harddrive and journal-commit permanently.

The /var/log/pm* logs, syslog, messages files show no complaints of these functions being gone.

What's more, I personally want top performance of my hard drive, especially on my netbook, they need all help they can get.

Regarding journal commits, that just scares me, there's gotta be a reason the file system defaults to $VALUE, and I really don't want to monkey with that.

As far as I'm concerned, adding "relatime" to /etc/fstab for jfs, or ext4 == way better battery savings. I mean why write to the hard drive every time you read something.

So the fact that I rm'd these two functions don't even seem to affect the battery life on my netbook that I can even monitor to be honest.

As to why KDE pokes ACPI and pm-utils on every log in, log out even on desktop computers is beyond me.

And why I cannot get the functionality in the main pm-functions script to hibernate the compuer after it's been suspended for 15 minutes as well seems to be another issue with pm-utils package, but I'll have to live with that :)

Anyway that's my fix for now :) I'm a happy fogie here.

Thanks again Eric.

- Fogie

---------- Post added 11-28-11 at 12:11 PM ----------

@wadsworth

Oh my, I forgot to thank you as well! Thank you to :)