Bug 11975 - qtatspi-plugin causes delay opening KDE GUI applications as root
Summary: qtatspi-plugin causes delay opening KDE GUI applications as root
Status: RESOLVED FIXED
Alias: None
Product: Mageia
Classification: Unclassified
Component: RPM Packages (show other bugs)
Version: Cauldron
Hardware: x86_64 Linux
Priority: Normal normal
Target Milestone: ---
Assignee: KDE maintainers
QA Contact:
URL:
Whiteboard:
Keywords: NEEDINFO
Depends on:
Blocks:
 
Reported: 2013-12-12 23:53 CET by Barry Jackson
Modified: 2016-10-16 19:09 CEST (History)
6 users (show)

See Also:
Source RPM:
CVE:
Status comment:


Attachments

Description Barry Jackson 2013-12-12 23:53:42 CET
Description of problem:
Modify a launcher e.g. dolphin using the Edit Launcher -> application tab -> Advanced options -> Run as different user

Enter "root" as user and OK the dialog.
Clicking the launcher correctly asks for root password, but then there is a delay of over a minute before the application opens.

Similarly if Kate is opened by clicking on a text file from within the root dolphin, it too opens after a long delay.

The delay is variable, there is nothing obviously wrong in top, cpu usage is low. Nothing in logs that seem related.

If I open dolphin as root from a terminal with:

kdialog --password 'Enter root password' | su - -c dolphin

or indeed just as root with

su -
dolphin

... there is no delay opening dolphin nor when opening kate from within it.

I have used dolphin like this for years without issues, even had it's own red icon, until the last few months in cauldron when this has become a problem.

Version-Release number of selected component (if applicable):


How reproducible:


Steps to Reproduce:
1.
2.
3.


Reproducible: 

Steps to Reproduce:
Comment 1 Barry Jackson 2013-12-13 00:24:15 CET
Testing in a VM the delay occurs for a real user (20 seconds) but for Xguest there is no delay.
Testing with a different real user on real hardware the delay is there (50 seconds).
I don't currently have Xguest on a real system to test.
Manuel Hiebel 2013-12-22 15:11:31 CET

CC: (none) => balcaen.john, lmenut, mageia
Blocks: (none) => 12079

Comment 2 Florian Hubold 2013-12-23 02:23:53 CET
Does the same delay occur when you run that application via kdesu?

CC: (none) => doktor5000

Comment 3 Barry Jackson 2013-12-23 12:45:23 CET
(In reply to Florian Hubold from comment #2)
> Does the same delay occur when you run that application via kdesu?

Yes

/OT
Why is kdesu not in $PATH? It took a bit of finding to do the test.

/usr/lib64/kde4/libexec/kdesu is not exactly intuitive :\
Comment 4 Florian Hubold 2013-12-23 15:50:12 CET
(In reply to Barry Jackson from comment #3)
> (In reply to Florian Hubold from comment #2)
> Why is kdesu not in $PATH? It took a bit of finding to do the test.
> 
> /usr/lib64/kde4/libexec/kdesu is not exactly intuitive :\

kdesu is probably primarily meant to be run via krunner (Alt+F2) where it is in the path. Otherwise, confirming the issue. Although it's not really critical overall, as running via su - or kdesu still works.

Will test with xguest on real hardware shortly.

FWIW, what graphics chip and what driver do you use? proprietary nvidia maybe?
Comment 5 Nicolas Lécureuil 2013-12-23 23:20:37 CET
(In reply to Barry Jackson from comment #3)
> (In reply to Florian Hubold from comment #2)
> > Does the same delay occur when you run that application via kdesu?
> 
> Yes
> 
> /OT
> Why is kdesu not in $PATH? It took a bit of finding to do the test.
> 
> /usr/lib64/kde4/libexec/kdesu is not exactly intuitive :\

kdesu is not aimed to be used in konsole like this. so this is not an issue.
Comment 6 Barry Jackson 2013-12-23 23:58:52 CET
(In reply to Florian Hubold from comment #4)
> (In reply to Barry Jackson from comment #3)
> > (In reply to Florian Hubold from comment #2)
> > Why is kdesu not in $PATH? It took a bit of finding to do the test.
> > 
> > /usr/lib64/kde4/libexec/kdesu is not exactly intuitive :\
> 
> kdesu is probably primarily meant to be run via krunner (Alt+F2) where it is
> in the path. Otherwise, confirming the issue. Although it's not really
> critical overall, as running via su - or kdesu still works.

No, running via kdesu does not work, it produces the delay as I confirmed above.
Running as root using "su -" is the only option that works.

I consider it critical, as a feature is broken which should work, and until mga4 has always worked.

> 
> Will test with xguest on real hardware shortly.
> 
> FWIW, what graphics chip and what driver do you use? proprietary nvidia
> maybe?

Basic onboard Intel - no fancy graphics card.

From hwinfo:

15: PCI 02.0: 0300 VGA compatible controller (VGA)
  [Created at pci.319]
  Unique ID: _Znp.TE0vqGgTKoB
  SysFS ID: /devices/pci0000:00/0000:00:02.0
  SysFS BusID: 0000:00:02.0
  Hardware Class: graphics card
  Model: "Intel VGA compatible controller"
  Vendor: pci 0x8086 "Intel Corporation"
  Device: pci 0x0162 
  SubVendor: pci 0x1458 "Giga-byte Technology"
  SubDevice: pci 0xd000 
  Revision: 0x09
  Driver: "i915"
  Driver Modules: "drm"
  Memory Range: 0xf7800000-0xf7bfffff (rw,non-prefetchable)
  Memory Range: 0xe0000000-0xefffffff (ro,non-prefetchable)
  I/O Ports: 0xf000-0xf03f (rw)
  IRQ: 41 (14705 events)
  I/O Ports: 0x3c0-0x3df (rw)
  Module Alias: "pci:v00008086d00000162sv00001458sd0000D000bc03sc00i00"
  Driver Info #0:
    Driver Status: i915 is active
    Driver Activation Cmd: "modprobe i915"
  Config Status: cfg=new, avail=yes, need=no, active=unknown
Comment 7 Barry Jackson 2014-01-26 13:13:16 CET
Raising priority/severity as this is a real PITA.

Still happening in real and VBox fully updated Cauldron installs.

Priority: Normal => High
Severity: normal => major

Comment 8 Florian Hubold 2014-02-02 18:49:46 CET
Seems we got sort of a reverse case of the original bug:

In Mageia 4, kwrite (as an example) does not start when launched from the menu
When changing the launcher to run as user and putting in your actual username there, it works as intended. That is with nvidia driver 325.15, so the nvidia driver should not be the cause, cf. https://bugs.mageia.org/show_bug.cgi?id=12079

Can someone from KDE team please take a look at this? Both issues should be fixed.
If required, I'll clone this bug into a new one for this issue.

CC: (none) => neoclust

Comment 9 Barry Jackson 2014-04-02 18:49:46 CEST
OK, after hours of testing I think I found the culprit.

It seems to be fixed by the removal of the qtatspi-plugin package which is required by onboard.

I have emailed the KDE developer of qtatspi-plugin in the hope that he may have some insight on this.


@Florian yes I think that's a different issue so another bug is best.
Comment 10 Barry Jackson 2014-04-03 18:26:10 CEST
I have opened a bug with Onboard upstream as they suggested qtatspi-plugin as require for some Onboard functionality.
https://bugs.launchpad.net/onboard/+bug/1302056

Summary: When running KDE apps as root from a modified launcher they take 1min+ to open => qtatspi-plugin causes delay opening KDE GUI applications as root

Comment 11 Barry Jackson 2014-04-09 00:36:34 CEST
This does seem to be a qtatspi-plugin problem related to dbus, which is confirmed by the author of qtatspi, but no solution has been suggested.

For now I have made qtatspi-plugin a 'suggest' of onboard so that it may be removed if required to work around this issue. (with some loss of features).

Extracts from an email from the author of qtatspi:

"So it is due to the dbus session being started. I'd be interested to know how Gnome handles this - does accessibility work for Gnome apps running as root?
I don't see a way to connect to the user's session bus. Of course Gnome might use xatoms to find the bus address - I'm not fond of using that way since it's not wayland compatible. "

"Yes, qtatspi is using the dbus session to discover the accessibility bus address and is thus to be blamed."

@Marmuta: maybe you can answer this?
Comment 12 Anne Nicolas 2014-12-15 13:17:18 CET
Any progress on that bug ? is it still valid for current cauldron?

CC: (none) => ennael1

Comment 13 Barry Jackson 2014-12-15 19:44:10 CET
No - qtatspi-plugin has only had one unrelated upstream commit this year.

As workaround I uninstall it in KDE, to avoid the annoying delay.

For most 'normal' users (meaning those who are unlikely to want to run KDE GUI applications as root) this is not an issue.

Maybe should be mentioned in errata?
Comment 14 Rémi Verschelde 2015-02-03 21:57:32 CET
I think a mention in the Errata would be nice yes. Can you handle it Barry?

CC: (none) => remi

Comment 15 Barry Jackson 2015-02-03 22:02:09 CET
(In reply to Rémi Verschelde from comment #14)
> I think a mention in the Errata would be nice yes. Can you handle it Barry?

Yes - will do.
Comment 16 Barry Jackson 2015-02-04 00:28:14 CET
https://wiki.mageia.org/en/Mageia_5_Errata#KDE
Comment 17 Rémi Verschelde 2015-02-17 21:25:16 CET
Should we mark this one as WONTFIX, or is there something that we can do to improve the situation? Is qtatspi-plugin doing its job properly apart from causing this bug, or could it be removed from KDE recommended packages?
Comment 18 Barry Jackson 2015-02-17 21:50:20 CET
(In reply to Rémi Verschelde from comment #17)
> Should we mark this one as WONTFIX, or is there something that we can do to
> improve the situation?

Nothing more I can do, so either leave it open or close it, I don't have a preference.

> Is qtatspi-plugin doing its job properly apart from
> causing this bug, 

Yes perfectly.

> or could it be removed from KDE recommended packages?

I don't understand - it is only 'Recommended' by onboard which works without it, with reduced functionality.

The bug *only* affects the *startup* of KDE GUI packages that are started as *ROOT* when qtatspi-plugin is installed.

Since this is a corner use case I think the errata is adequate and we can live with it.
Comment 19 Rémi Verschelde 2015-02-17 21:53:35 CET
(In reply to Barry Jackson from comment #18)
> (In reply to Rémi Verschelde from comment #17)
> 
> > or could it be removed from KDE recommended packages?
> 
> I don't understand - it is only 'Recommended' by onboard which works without
> it, with reduced functionality.

Ok this was a misunderstanding on my side, I forgot the onboard middle man :)
 
> Since this is a corner use case I think the errata is adequate and we can
> live with it.

Let's leave the bug report open, since it's not fixed and we would still like to fix it (so it's not strictly speaking a WONTFIX). I'll just remove it from the release_blocker KDE bugs tracker.
Rémi Verschelde 2015-02-17 21:53:46 CET

Blocks: 12079 => (none)

Barry Jackson 2015-03-08 22:12:05 CET

Priority: High => Normal
Severity: major => normal

Luc Menut 2016-08-25 16:42:55 CEST

CC: lmenut => (none)

Comment 20 Samuel Verschelde 2016-10-16 15:33:57 CEST
Is this bug still present in current cauldron?

Keywords: (none) => NEEDINFO
Assignee: bugsquad => kde

Comment 21 Barry Jackson 2016-10-16 19:09:09 CEST
I thought that I did not have qtatspi-plugin installed as I have not seen this issue for a while now.

Anyway - on checking, it *is* installed, so maybe recent plasma5 has somehow fixed this.

This Cauldron has been installed for a few months now, so the plugin must have been there since install and during that time I have not had the issue or I would have removed it.

Guess we can close it \o/

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


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