upstart failsafe incorrect screen resolution when running with 2.6.31 kernerls

Bug #492270 reported by pacrook
16
This bug affects 2 people
Affects Status Importance Assigned to Milestone
linux (Ubuntu)
Won't Fix
Undecided
Unassigned
xserver-xorg-video-intel (Ubuntu)
Invalid
Medium
Unassigned

Bug Description

Binary package hint: upstart

If Xorg is broken so that "upstart" tries to display a message that Ubuntu is running in low graphics mode, then with 2.6.31-14 and 2.6.31-15 kernels it fails to determine the correct screen resolution and the message is thus unreadable. Tried navigating the messed-up upstart menus to get to a console but just end up with black screens.

How to repeat this bug:

(a) break X (not sure how to do this bit) - In my case experimenting with skim/scim/ibus was enough to break hal.
       "(EE) config/hal: couldn't initialise context: unknown error (null)"

(b) reboot with a 2.6.31 kernel - every time ended up with a unreadable message.

(c) (optional) from knowledge of upstart menus; click "ok" and the select bottom button on next (messed up) menu which should be the go to a console option. Tab until "ok" button selected and hit return. Result completely bank screens.

Expectation: readable fall-back in the case that X fails.

If I booted with 2.6.28-16 then can read message "Ubuntu is running in low graphics mode" and can navigate to a console.

Possibly related warnings from /var/log/Xorg.failsafe.log following messed up display:

(WW) Open ACPI failed (/var/run/acpid.socket) (No such file or directory)
(WW) VESA(0): Unable to estimate virtual size
(WW) VESA(0): No valid modes left. Trying less strict filter...
(WW) VESA(0): Unable to estimate virtual size
(WW) VESA(0): No valid modes left. Trying aggressive sync range...
(WW) VESA(0): Unable to estimate virtual size
(WW) xf86CloseConsole: KDSETMODE failed: Input/output error
(WW) xf86CloseConsole: VT_GETMODE failed: Input/output error
(WW) xf86OpenConsole: VT_GETSTATE failed: Input/output error
(WW) xf86CloseConsole: VT_ACTIVATE failed: Input/output error
(WW) xf86CloseConsole: VT_WAITACTIVE failed: Input/output error

also see ubuntu forums tread:
http://ubuntuforums.org/showthread.php?t=1344322

Ubuntu Release 9.10

upstart:
  Installed: 0.6.3-10
  Candidate: 0.6.3-10
  Version table:
 *** 0.6.3-10 0
        500 http://gb.archive.ubuntu.com karmic/main Packages
        100 /var/lib/dpkg/status

machine: DELL XPS notebook M1210

ProblemType: Bug
Architecture: i386
Date: Fri Dec 4 11:07:47 2009
DistroRelease: Ubuntu 9.10
NonfreeKernelModules: openafs
Package: upstart 0.6.3-10
ProcEnviron:
 LANGUAGE=en_GB:en
 LANG=en_GB.UTF-8
 SHELL=/bin/bash
ProcVersionSignature: Ubuntu 2.6.31-15.50-generic
SourcePackage: upstart
Uname: Linux 2.6.31-15-generic i686
XsessionErrors: (polkit-gnome-authentication-agent-1:2534): GLib-CRITICAL **: g_once_init_leave: assertion `initialization_value != 0' failed

Revision history for this message
pacrook (paul-crook) wrote :
affects: upstart (Ubuntu) → xorg (Ubuntu)
Revision history for this message
pacrook (paul-crook) wrote :

I'm thinking this is related to bug #491483.

No-one mentions messed up upstart displays but the "cure" seems to be to roll back the X updates as set out in post#19 bug #491483.

Revision history for this message
Bryce Harrington (bryce) wrote :

Hi paul-crook,

Thanks for including the attached files. Could you also include your /var/log/Xorg.0.log (or Xorg.0.log.old) from after reproducing the issue?

Please attach the output of `lspci -vvnn` too.

[This is an automated message. Apologies if it has reached you inappropriately; please just reply to this message indicating so.]

tags: added: needs-xorglog
Changed in xorg (Ubuntu):
status: New → Incomplete
Revision history for this message
pacrook (paul-crook) wrote :

Hi Bryce,

having downgraded my Xorg as per post#19 of bug #491483 my system is currently okay. Unfortunately I don't have time to update X and reproduce it again until begining of next week. Hope that's okay.

Revision history for this message
pacrook (paul-crook) wrote :

Hi,

Unfortunately the fixing of bug #491483 means that its going to be difficult to reproduce. However I do suspect that the problem with failsafe selecting the wrong resolution (with 2.6.31 kernels) on my machine hasn't gone away. Its just hidden until the next time X / upstart breaks.

Any suggestions on how I can force upstart to drop into "warning using low resolution mode" to see if the display is still messed up?

Attached is the output of lspci -vvnn as requested.

Paul

Revision history for this message
pacrook (paul-crook) wrote :

Just thought that the Xorg.failsafe.logs might be more pertinent to this issue that Xorg.0.logs.

find attached Xorg.failsafe.log and Xorg.failsale.log.old. These date from when I was having problems seeing the "low resolution" warning. 'Xorg.failsafe.old' is a boot with a '2.6.28' where the resolution was correct idetified and the warning was readable. 'Xorg.failsafe' is a boot with a '2.6.31' where the screen was unreadable.

Hope they help,

Paul

Revision history for this message
pacrook (paul-crook) wrote :
Revision history for this message
Bryce Harrington (bryce) wrote :

Thanks; you can synthetically cause X to go into failsafe mode by simply doing something like:

 echo "foobar" >> /etc/X11/xorg.conf

and then restarting X or gdm (or restarting your computer).

Changed in xorg (Ubuntu):
importance: Undecided → Medium
status: Incomplete → Triaged
Revision history for this message
pacrook (paul-crook) wrote :

Perfect! That reproduces (or probably should say unveils) the bug.

I'd had a go at breaking things but nothing quit so blunt as simply over-writing xorg.conf with nonsense ;-)

Have done

  sudo su
  cp /etc/X11/xorg.conf /etc/X11/xorg.conf.bak
  echo foobar >> /etc/X11/xorg.conf

I get the following symptoms:

 - If I boot with a 2.6.31-15 generic kernel then I get an unreadable failsafe screen ("running in low resolution mode ....").
 - If I boot with a 2.6.28-16 generic kernel then I get a reabable failsafe screen ("running in low resolution mode ....").

The situation is slightly better than before because "2.6.31-15 (recovery)" now gets me to a readable menu and and I can fix my system.

Attached are Xorg.failsafe.log and Xorg.0.log

let me know if you need any more information.

Revision history for this message
pacrook (paul-crook) wrote :

and Xorg.0.log

Revision history for this message
Bryce Harrington (bryce) wrote :

Hi pacrook,

As I've been experimenting with KMS and failsafe mode lately for Lucid, we've found that issues like this one crop up. It seems that the -vesa video driver (which failsafeX mode uses) doesn't work well with KMS. We've converted it to use fbdev instead. I'm >pretty< sure that'll solve this issue. Basically, you can test it by substituting "fbdev" instead of "vesa" in the xorg.conf.failsafe (or your xorg.conf).

Alternatively, just boot the Ubuntu Lucid livecd, edit the xorg.conf to make it invalid (append 'foobar' or whatever), and then log out and back in (or do 'sudo service gdm restart' from the command line). Hopefully it should pop you into the failsafeX session properly.

If you get a chance, please give it a try and let us know. If we don't hear back from you in a few weeks we'll assume it's no longer an issue.

Changed in xorg (Ubuntu):
status: Triaged → Incomplete
Bryce Harrington (bryce)
affects: xorg (Ubuntu) → xserver-xorg-video-intel (Ubuntu)
Bryce Harrington (bryce)
tags: added: karmic
Bryce Harrington (bryce)
tags: added: resolution
Revision history for this message
Jeremy Foshee (jeremyfoshee) wrote :

added a kernel task to this bug.

pacrook,
      Have you been able to verify Bryce's recommendation above?

Thanks!

~JFo

Changed in xserver-xorg-video-intel (Ubuntu):
status: Incomplete → Invalid
Revision history for this message
pacrook (paul-crook) wrote :

Hi,

Sorry been a bit busy.

Just tried Bryce's recommendation and it doesn't really work. I don't end up with a mashed up display which is an improvement but fbdev gives up with a 'mode initialisation failed' so I don't get the GUI just a plain old text TTY waiting for me to log in (which is fine for me) but I suspect is not the intended behaviour.

I'll attach the logs once I've restored the system.

Paul

Revision history for this message
pacrook (paul-crook) wrote :

xorg.conf.failsafe; so you can double check I didn't make a mistake

Revision history for this message
pacrook (paul-crook) wrote :

/var/log/Xorg.failsafe.log

Brad Figg (brad-figg)
tags: added: kj-triage
Brad Figg (brad-figg)
tags: added: b73a1py79
Revision history for this message
Brad Figg (brad-figg) wrote :

This bug was filed against a series that is no longer supported and so is being marked as Won't Fix. If this issue still exists in supported series, please file a new bug.

Changed in linux (Ubuntu):
status: New → Won't Fix
Brad Figg (brad-figg)
tags: removed: b73a1py79
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.