Bug 3439 - kde dies when using minimal install with autologin
Summary: kde dies when using minimal install with autologin
Status: RESOLVED OLD
Alias: None
Product: Mageia
Classification: Unclassified
Component: RPM Packages (show other bugs)
Version: Cauldron
Hardware: i586 Linux
Priority: Normal normal
Target Milestone: ---
Assignee: Mageia Bug Squad
QA Contact:
URL:
Whiteboard:
Keywords: NEEDINFO
Depends on:
Blocks:
 
Reported: 2011-11-24 22:18 CET by Dave Hodgins
Modified: 2012-06-12 11:47 CEST (History)
4 users (show)

See Also:
Source RPM: gdm-3.2.1.1-1.mga2.src.rpm
CVE:
Status comment:


Attachments
.xsession-errors from first (failed) attempt to start kde (74.45 KB, text/plain)
2011-11-24 22:19 CET, Dave Hodgins
Details
.xsession-errors from second (good) start of kde (57.50 KB, text/plain)
2011-11-24 22:20 CET, Dave Hodgins
Details
service -s output when kde fails (6.00 KB, text/plain)
2011-11-25 02:17 CET, Dave Hodgins
Details
service -s output when kde works (5.98 KB, text/plain)
2011-11-25 02:18 CET, Dave Hodgins
Details

Description Dave Hodgins 2011-11-24 22:18:10 CET
On a minimal install, with gdm configured to autologin, kde starts,
but then dies, returning the system to the gdm login screen.

A subsequent login to kde is ok.  It appears kde is starting before
something that it requires has started.

I expect this would also happen if using the autologin package, or kdm,
instead of gdm.

In the .xsession-errors file on the first start of kde, the failures start
with ...
kglobalaccel: Fatal IO error: client killed
kactivitymanagerd: Fatal IO error: client killed
kdeinit4: Fatal IO error: client killed
kdeinit4: sending SIGHUP to children.

I'll attach the .xsession-errors files from both the first start of kde,
where it fails, and the second start, where the startup is good.
Comment 1 Dave Hodgins 2011-11-24 22:19:27 CET
Created attachment 1108 [details]
.xsession-errors from first (failed) attempt to start kde
Comment 2 Dave Hodgins 2011-11-24 22:20:26 CET
Created attachment 1109 [details]
.xsession-errors from second (good) start of kde
Comment 3 Dave Hodgins 2011-11-24 22:34:01 CET
Looking through the first .xsession-errors file, I think the line
plasma-desktop(3990)/libakonadi Akonadi::AgentManagerPrivate::createDBusInterface: AgentManager failed to get a val
id AgentManager DBus interface. Error is: 1 "org.freedesktop.DBus.Error.NameHasNoOwner" "Could not get owner of nam
e 'org.freedesktop.Akonadi.Control': no such name"

is the critical one.  It looks to me like kde is starting before dbus.
Comment 4 Manuel Hiebel 2011-11-24 22:58:34 CET
Mikala, any idea ?

Assignee: bugsquad => balcaen.john

Comment 5 Dave Hodgins 2011-11-24 23:53:05 CET
Also, just fyi, I currently have two cauldron installs, and this only
happens on one of them.

The one where it doesn't happen, is an install that started as a Mdv 2009.1
install, udated via urpmi to 2010.2, then to Mageia 1, alpha 2 (by
mistake), which has since been kept as a cauldron install.
It has about 4800 packages installed.

On the install where it does happen, this was a clean install of mga 1,
updated to cauldron using the pre-release alpha iso.  I've installed
all of the updates since then.  It only has around 2200 packages installed,
and only default services enabled.

It seems that the extra services in the full install, slow down the staring
of kde long enough that autologin works in that install.
Comment 6 Dave Hodgins 2011-11-25 02:17:34 CET
Created attachment 1112 [details]
service -s output when kde fails

Now I know it is not dbus that is causing the problem, but now I have no
idea what is causing the problem.

I created  ... $ cat /home/dave/.kde4/Autostart/status
#!/bin/bash
/sbin/service -s > /home/dave/service.status

Comparing the output when kde fails with the output when kde works shows
no obvious differences. The dbus service has already started.

I'm attaching the output from when kde fails.
Comment 7 Dave Hodgins 2011-11-25 02:18:20 CET
Created attachment 1113 [details]
service -s output when kde works
Dave Hodgins 2011-11-25 02:18:52 CET

Attachment 1112 mime type: application/octet-stream => text/plain

Comment 8 John Balcaen 2011-11-26 01:13:10 CET
if i'm not wrong the issue is only happening with systemd & not sysvinit.
@ david > you're speaking about the autologin package ? or the autologin functionnality of kdm/gdm ? (not sure if gdm provides such functionnality but if i'm not wrong it was also broken )

CC: (none) => balcaen.john

D Morgan 2011-11-28 02:04:21 CET

CC: (none) => dmorganec

Comment 9 Dave Hodgins 2011-11-28 03:55:58 CET
I realize I was wrong about both cauldron installs using gdm.  The one
where autologin is working properly was using kdm.  Switching it to gdm
causes the same failure there, so I've changed the rpm for this bug to gdm.

This is with systemd.  I haven't tried sysvinit in cauldron, since systemd
became the default.

Some more info that may help to track down the problem.

The crash doesn't happen until I do something that involves pressing the
enter key.

I can play pysol, open konsole and type in a command. Everything seems
to be fine, until I press the enter key.  Also happens with gnome.
All I have to do to crash it, after autologin, is press the enter key.
After normal login, everything is ok.

Additionally, I noticed that when the desktop manager crashes (I.E. when
I press the enter key), there are some garbage characters added to VT 7.

It's as if the enter key is still going to vt7, while other keys pressed
on the keyboard are going to vt8.

Source RPM: systemd-37-11.mga2.src.rpm => gdm-3.2.1.1-1.mga2.src.rpm

Comment 10 John Balcaen 2011-11-28 06:06:02 CET
Ok the problem seems related to gdm here & not kde at all.
Do you have the .xsession-errors available during the crash ?
(& removing me as assignee :p )

Assignee: balcaen.john => bugsquad

Comment 11 Marja Van Waes 2012-01-23 07:14:02 CET
(In reply to comment #10)
> Ok the problem seems related to gdm here & not kde at all.
> Do you have the .xsession-errors available during the crash ?
> (& removing me as assignee :p )

@ mikala

You mean there is something missing in attachment 1108 [details]?

cc'ing two more gdm committers

CC: (none) => jani.valimaa, mageia, marja11

Comment 12 John Balcaen 2012-01-23 10:36:07 CET
I did ?
Comment 13 Marja Van Waes 2012-04-13 17:05:36 CEST
@ Dave

We have gdm-3.4.0.1-3.mga2 now, is this bug still valid?
Jani Välimaa 2012-04-28 10:03:37 CEST

CC: jani.valimaa => (none)

Comment 14 Marja Van Waes 2012-05-26 13:04:55 CEST
Hi,

This bug was filed against cauldron, but we do not have cauldron at the moment.

Please report whether this bug is still valid for Mageia 2.

Thanks :)

Cheers,
marja

Keywords: (none) => NEEDINFO

Comment 15 Marja Van Waes 2012-06-12 11:47:51 CEST
Closing, because this bug was filed in november 2011, there is nothing indicating the bug is still valid and there was no reply to the last two comments

Please reopen if it is still valid for Mga2 or current cauldron

Status: NEW => RESOLVED
Resolution: (none) => OLD


Note You need to log in before you can comment on or make changes to this bug.