package xorg-driver-fglrx 2:8.543-0ubuntu1 failed to install/upgrade: EOF on stdin at conffile prompt

Hello:

It looks like atiogl.xml was modified. Did you do this yourself, or was it like that?

It's the automatic bug report (if it's the question..) and i didn't modified something.

Mario, am I correct in assuming this is a packaging issue rather than something needing upstreamed?

Bryce:

I'm thinking so, but I don't know for sure. I've not been able to reproduce
it yet.

On Mon, Nov 3, 2008 at 14:24, Bryce Harrington <email address hidden>wrote:

> Mario, am I correct in assuming this is a packaging issue rather than
> something needing upstreamed?
>
> ** Changed in: fglrx-installer (Ubuntu)
> Importance: Undecided => Medium
> Status: Confirmed => Triaged
>
> --
> package xorg-driver-fglrx 2:8.543-0ubuntu1 failed to install/upgrade: EOF
> on stdin at conffile prompt
> https://bugs.launchpad.net/bugs/283765
> You received this bug notification because you are subscribed to fglrx-
> installer in ubuntu.
>

--
Mario Limonciello
<email address hidden>

Confirm.

Just tried to activate fglrx in Hardware Drivers, it's downloaded and then crashed.

$ uname -a
Linux plankton 2.6.27-7-generic #1 SMP Tue Nov 4 19:33:20 UTC 2008 i686 GNU/Linux

I had this problem as well. After trying to install the driver, I uninstalled all the FGLRX components in gnome-terminal and tried using Hardware Drivers to install again. It worked (the driver appears as activated). I restarted and got the proper screen resolution, etc.

That said, I still can't get Compiz working, though that might be a separate issue.

The problem arises only when trying to install fglrx from the Hardware Drivers applet. If the packages are installed from Synaptic, the installation succeeds and after reboot the driver works perfectly, including compiz. The Hardware Drivers applet shows a gray bullet indicating a different version of the recommended driver is being used, the 'Activate' button supposedly corrects this, indicating a reboot is required, however, the status doesn't change after reboot.

If this problem is only happening when installed from hardware drivers - there is another caveat:

It's likely only happening if you previously had fglrx installed at some point with a configuration file in /etc/ait still present. Jockey isn't presenting a UI to show the user there was a configuration file change that they need to acknowledge.

The effect of this can be reproduced with:

sudo dpkg -P pmount
echo hello | sudo tee /etc/pmount.allow
sudo nohup python -c "import apt, os; os.environ['DEBIAN_FRONTEND'] = 'noninteractive'; c=apt.Cache(); c['pmount'].markInstall(); c.commit()"

I'll ask Michael about some advice.

Option 1 is to convince python-apt to use the dpkg option --force-confnew. Option 2 is a workaround in the fglrx jockey handler to move away old fglrx configuration files (to .dpkg-old). However, for those we'd need to keep the handler up to date with possible configuration files.

Option 3 is to make fglrx not use any configuration files at all, of course :)

Nominating for intrepid, too.

http://bazaar.launchpad.net/~ubuntu-core-dev/jockey/ubuntu/revision/284

http://bazaar.launchpad.net/~ubuntu-core-dev/jockey/intrepid/revision/284

This bug was fixed in the package jockey - 0.5~beta3-0ubuntu8

---------------
jockey (0.5~beta3-0ubuntu8) jaunty; urgency=low

  * Merge bug fixes from trunk:
    - Replace enabled/disabled icons with more decent variant, thanks Kenneth
      Wimer! (LP: #290239)
    - backend.py: Rewrite timeout behaviour for more robustness; do not time
      out right after a long method call.
    - Intercept crashes of the backend (which manifest as D-BUS NoReply
      error), present an error message, and restart backend. (LP: #273600)
    - jockey-gtk: Fix crash if nothing is selected in the tree view after an
      operation. (LP: #283887)
    - fglrx.py: Unconfigured driver defaults to ati, which already provides
      compositing. (LP: #285879)
  * jockey/oslib.py, install_package(): Use --force-confnew to fix hang/crash
    on "EOF on stdin at conffile prompt", since the jockey backend does not
    have any terminal or interactivity. This fixes fglrx install failure when
    the upstream installer was previously used. This still keeps .dpkg-old
    configuration backups, and using a tool like Jockey should "make it work".
    (LP: #283765)
  * data/handlers/b43.py: Add missing OSLib import. (LP: #295530)

 -- Martin Pitt <email address hidden> Mon, 24 Nov 2008 20:31:53 +0100

Is there any chance that jockey would be called to install a package more than once (say, the package is installed, then removed, then installed again)? If so, there's a possibility of --force-confnew moving a conffile to .dpkg-old twice, thereby clobbering the earliest instance.

This is also a different standard of "make it work" than what's used in update-manager. I'm not convinced this is a good change to make in an SRU, until we've had time to see how it's received in jaunty.

Steve Langasek [2008-11-29 5:10 -0000]:
> Is there any chance that jockey would be called to install a package
> more than once (say, the package is installed, then removed, then
> installed again)?

In that case the conffile wouldn't be modified the second time, and
thus not touched at all, no?

> This is also a different standard of "make it work" than what's used in
> update-manager. I'm not convinced this is a good change to make in an
> SRU, until we've had time to see how it's received in jaunty.

Okay, I can back it out and reupload the other fixes, if you think
that's more appropriate?

I discussed that problem with Michael Vogt, and we both agreed that
this was the best solution so far.

Thanks for review!

On Sat, Nov 29, 2008 at 12:46:13PM -0000, Martin Pitt wrote:
> Steve Langasek [2008-11-29 5:10 -0000]:
> > Is there any chance that jockey would be called to install a package
> > more than once (say, the package is installed, then removed, then
> > installed again)?

> In that case the conffile wouldn't be modified the second time, and
> thus not touched at all, no?

Unless the user had touched that conffile in between.

It's unlikely - but not impossible.

> > This is also a different standard of "make it work" than what's used in
> > update-manager. I'm not convinced this is a good change to make in an
> > SRU, until we've had time to see how it's received in jaunty.

> Okay, I can back it out and reupload the other fixes, if you think
> that's more appropriate?

I do. The rest of the changes in 0.5~beta3-0ubuntu6.1 look fine to me for
SRU.

BTW, there was also a stray 3kloc jockey-debug.log included in the last
upload, would be nice if the next upload will drop that. :)

Thanks,
--
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.
Ubuntu Developer http://www.debian.org/
<email address hidden> <email address hidden>

I backed out this change and reuploaded (jockey-debug.log dropped, too, sorry for that).

Accepted into intrepid-proposed, please test and give feedback here. Please see https://wiki.ubuntu.com/Testing/EnableProposed for documentation how to enable and use -proposed. Thank you in advance!

Hallo Martin,
Leider kann ich kein Englisch.
Aber mittlerweile funktioniert alles.
Vielen Dank
Mit freundlichen Grüßen
Hans
Am Sonntag, den 07.12.2008, 20:17 +0100 schrieb Martin Pitt:
> Accepted into intrepid-proposed, please test and give feedback here.
> Please see https://wiki.ubuntu.com/Testing/EnableProposed for
> documentation how to enable and use -proposed. Thank you in advance!
>
> ** Tags added: verification-needed
>

Previous message is German and means "I can't speak English, but now everything works".

This bug was fixed in the package jockey - 0.5~beta3-0ubuntu6.1

---------------
jockey (0.5~beta3-0ubuntu6.1) intrepid-proposed; urgency=low

  * debian/control: Update Vcs-Bzr: for intrepid branch.
    (LP: #283765)
  * fglrx.py: Unconfigured driver defaults to ati, which already provides
    compositing. This fixes gnome-appearance-properties to endlessly ask to
    install the fglrx driver on systems where ati works fine. (LP: #285879)
  * data/handlers/b43.py: Add missing OSLib import. (LP: #295530)
  * data/handlers/broadcom_wl.py: If b44 is loaded, don't disable the entire
    handler, but instead install a "wl" modprobe handler which removes ssb and
    b44 first, then loads wl, and reloads b44 again, which makes both devices
    work. (LP: #289845)
  * tests/detection.py: Fix yum printer test for changed package name on
    openprinting.org. This only affects the test suite, so that it succeeds
    again. (Cherrypicked from trunk r501)

 -- Martin Pitt <email address hidden> Fri, 28 Nov 2008 11:43:06 +0100

Thanks for the replay, but a newer ati driver fixed it.

gretings,

A. al Mourabit

> Date: Sat, 23 Jan 2010 13:24:09 +0000
> From: <email address hidden>
> To: <email address hidden>
> Subject: [Bug 283765] Re: package xorg-driver-fglrx 2:8.543-0ubuntu1 failed to install/upgrade: EOF on stdin at conffile prompt
>
> ** Changed in: jockey (Ubuntu)
> Status: Fix Committed => Fix Released
>
> --
> package xorg-driver-fglrx 2:8.543-0ubuntu1 failed to install/upgrade: EOF on stdin at conffile prompt
> https://bugs.launchpad.net/bugs/283765
> You received this bug notification because you are a direct subscriber
> of the bug.

_________________________________________________________________
Alles over Windows
http://www.windows.nl/About.aspx