[K8M800] Maverick 10.10 live-CD login failure, ttyN screens don't work, on Averatec 3280 running Lucid

[Peter Belew]

Binary package hint: xorg-server

[ Comments added for Natty 11.04 alpha version }
[ Comments added for Maveric beta versions ]

It is now impossible to boot and log in on this system from the beta-2 live CD - see comment #111.

THIS NOW SEEMS TO BE 'CURED' by running kernels from Karmic - see comments past #100.

A Framebuffer driver seems to be involved - see comment #109.

I installed Lucid on the Averatec 3280 laptop from the Alternate CD on 10 December 2009. I then updated the system with apt-get several times. On booting this morning, the logical screen origin was shifted to the center of the screen. With some difficulty, I was able to log in, and to try to find a solution.

After the original installation, my login screen was too large, so that the login window was of center. It looks like some bugs have been reported about that issue, such as #231555 and @214704; I haven't looked at the details. I have had similar problems with earlier releases on this laptop. However, once logged in, the screen resolution was correct (1024x768 on this laptop's panel). Adding an xorg.conf file seemed to cure the login screen size problem.

The split-screen problem only appeared this morning, after an update the previous evening.

I will attach a photo of the screen and any further details a bit later - it's a bit difficult to edit this report with the screen split with the browser's left and right margins in the middle of the screen.

Also, I now get a blank screen when trying to ctrl-alt-F3 etc to a tty console - this did not happen with the original installation of Lucid, yesterday.

This problem makes the system virtually unusable!

DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=10.04
DISTRIB_CODENAME=lucid
DISTRIB_DESCRIPTION="Ubuntu lucid (development branch)"
Linux EROS 2.6.32-7-generic #10-Ubuntu SMP Sun Dec 6 13:43:20 UTC 2009 i686 GNU/Linux
model name : Mobile AMD Sempron(tm) Processor 28

ProblemType: Bug
Architecture: i386
Date: Fri Dec 11 08:43:04 2009
DistroRelease: Ubuntu 10.04
InstallationMedia: Error: [Errno 13] Permission denied: '/var/log/installer/media-info'
Package: gdm 2.29.1-0ubuntu4
ProcEnviron:
 PATH=(custom, user)
 LANG=en_US.UTF-8
 SHELL=/bin/bash
ProcVersionSignature: Ubuntu 2.6.32-7.10-generic
SourcePackage: gdm
Tags: lucid
Uname: Linux 2.6.32-7-generic i686

[Peter Belew (peterbe)]

I am attaching a photo of the screen. The red pointer on the left is at the physical left side of the screen, but the logical left of the screen is in the center, so the background lynx image is split. The hide buttons for the top and bottom panels appear in the center of the screen.

Note that the login screen is split the same way.

The system is using an openchrome driver.

[Peter Belew (peterbe)]

I am attaching /var/log/Xorg.0.log and /etc/X11/xorg.conf.

The Xorg.0.log was generated when booting WITHOUT an xorg.conf file.

[Peter Belew (peterbe)]

Here is an xorg.conf file which I used earlier when the issue was that the boot login screen was too large - this problem was similar to an earlier bug I reported on the same laptop, running earlier Ubuntu releases: Bug #296142. With the initial installation of lucid, only the login screen was the wrong size; then the screen was sized properly to 1024x768.

But now this split-screen bug has appeared. It appears that having xorg.conf installed doesn't affect the new bug, one way or the other.

BUT it should be noted that when I try to switch to a console TTY, the screen goes blank. Switching back to X with ALT-F7 DOES work however. Before the split-screen bug appeared this morning, console TTYs worked fine.

[Peter Belew (peterbe)]

I tried an experiment: I put the xorg.conf file back in /etc/X11, but changed 'openchrome' to 'vesa', and restarted. The screen ends up being split the same way, with the left-hand part of the logical screen showing in the middle, but the mouse pointer position is the same as the logical position: e.g. moving the mouse all the way to the left or right moves the pointer to the center of the screen, and I can actually click on icons etc where they appear, instead of on the opposite side of the screen.

Running the openchrome driver, the screen looks exactly the same, but in order to point at something on the screen, I have to move the pointer to the corresponding position on the opposite half of the screen.

In either case, the display has the left and right halves reversed, which looks like an address bit inverted. With the vesa driver, mouse position corresponds to screen position for other objects, while with the openchrome driver, mouse position corresponds to physical position on the display. Apparently cursors are handled differently by the two drivers.

[Peter Belew (peterbe)]

I changed the driver back to "openchrome", and added

  Option "SWCursor" "true"

to the "Device" section for the display. Now the cursor appears as with the vesa driver, making it easier to point at objects on the reversed screen. I suspected the problem with the cursor position was due to using a "hardware" cursor.

See: https://help.ubuntu.com/community/OpenChrome#Problems and solutions

After this bug is fixed, I'll revert to the hardware cursor.

[Peter Belew (peterbe)]

I notice that the right side of the screen is a pixel or so higher than the left side - that can be seen where the ends of the top and bottom panels meet in the middle of the screen.

[Sebastien Bacher (seb128)]

Thank you for your bug report, gdm didn't change recently so it could rather been an xorg than gdm issue, the number of comments doesn't make the issue easy to understand either especially than gdm is the login screen but first screenshot shows a desktop config and not the login banner

[Peter Belew (peterbe)]

Yes, it's likely to be an xorg issue. The driver is xserver-xorg-video-openchrome,
dpkg -s xserver-xorg-video-openchrome
Package: xserver-xorg-video-openchrome
Status: install ok installed
Priority: optional
Section: x11
Installed-Size: 556
Maintainer: Ubuntu X-SWAT <email address hidden>
Architecture: i386
Version: 1:0.2.904+svn812-1ubuntu1
Replaces: libchromexvmc1, libchromexvmcpro1
Provides: xserver-xorg-video-6
Depends: libc6 (>= 2.4), libdrm2 (>= 2.3.1), libx11-6, libxext6, libxv1, libxvmc1, xserver-xorg-core (>= 2:1.6.99.900)
Conflicts: libchromexvmc1, libchromexvmcpro1
Description: X.Org X server -- VIA display driver

When the resolution is 1024x768, the desktop appears as in the photo whether or not there is an xorg.conf file installed. The principal difference is whether or not the hardware cursor is enabled. With no xorg.conf disabling the hardware cursor, the cursor moves from the extreme left to the extreme right of the screen; in other words its range is not the same as the virtual X position of other objects on the screen. This applies to both the login window, and to the desktop after logging in.

If the hardware cursor is disabled in xorg.conf, then the cursor position coordinates are the same as the rest of the screen - shifted half a screen width from the left edge of the screen, with the right side of the screen wrapped to the left.

If resolutions other than 1024x768, such as 800x600, 640x480, etc, then the X origin is shifted to some position other than the center of the screen.

[Peter Belew (peterbe)]

The current openchrome driver is dated
-rw-r--r-- 1 root root 333172 2009-12-07 08:15 /usr/lib/xorg/modules/drivers/openchrome_drv.so

[Peter Belew (peterbe)]

I just booted from the original 10 December live CD. In this case, the login screen appears to be something like 1600x1200, clipped to the upper-left 1024x768 which this panel will display. The small login 'window' thus appears in the lower-right of the screen, but is accessible. Once logged in, I still see a 1600x1200 screen, but I can set the resolution to 1024x768. I checked the date of openchrome_drv.so, and it's the same (the date in the above comment is Pacific Standard Time). So I believe the change is not in that module, but rather in some other component.

Hopefully this time for the change (sometime between 04:00 and 16:00 UTC on 11 Dec) will be of help in tracking down what changed.

This bug appeared overnight (Pacific Time) after the initial CD alpha-1 release. Note that my installation was from the Alternate CD, while today's test was from the Desktop/Live CD.

Note that tty console displays are unusable - the screen is all black - but Alt-F7 does usually work to switch back to X.

[Peter Belew (peterbe)]

At this point, it is clear to me that part of the problem is in low-level xorg code - code handling the mapping between the internal X coordinate system and the UniChrome display's coordinate system. It looks, though, like the mapping is correct for the UniChrome display's hardware cursor, when that is used.

But the problem with text (ttyN) displays not working is something else.

[Peter Belew (peterbe)]

Note that after I installed from the alternate CD, while the login screen came up a 1600x1200, the Gnome display came up at 1024x768. Yet when I ran the live CD, the Gnome display initially came up at 1600x1200. Why the difference? After a bit of reflection, I realized that, since I had kept my home directories from the Karmic installation, resolution saved in my home directory (somewhere) would be maintained. BUT I don't know where that is - I'd like to know where the setting from System>Preferences>Display is saved.

[Grant Bowman (grantbow)]

I think succinctness is key to getting this problem looked at and solved by the right people. I suspect there are very few people with the necessary skills.

[Peter Belew (peterbe)]

A number of xorg modules were just updated. This has had no effect on this bug, unfortunately.

[Peter Belew (peterbe)]

To Sebastien Bacher - I've attached a photo of the login screen.

For the login screen, the X-origin is always the center of the screen, no matter what resolution was previously set in the Gnome preferences dialog.

I just noticed the image is blurry - I was aiming at the center of the screen and there isn't any bright object there to autofocus on - sorry!

[Peter Belew (peterbe)]

To Sebastien Bacher - I've attached a photo of the login screen.

For the login screen, the X-origin is always the center of the screen, no matter what resolution was previously set in the Gnome preferences dialog.

I just noticed the image is blurry - I was aiming at the center of the screen and there isn't any bright object there to autofocus on - sorry!

[Peter Belew (peterbe)]

Here is what the Gnome screen looks like at 800x600.

Here the screen x-axis origin is a couple of cm from the right side of the screen. Here I have placed the red mouse pointer at the edge of the screen, so that it wraps around to the other side of the screen. Likewise the terminal window wraps around to the other side of the screen.

This photo shows fairly clearly how the screen to the right of the dividing line is a pixel or two higher than the screen to the left - look at the position of the adjoining hide buttons. I suspect this will help give an idea about what is going on with screen address mapping.

[Peter Belew (peterbe)]

Here is the 800x600 screen.

[Peter Belew (peterbe)]

This is still happening as of today. I haven't tried building an openchrome driver for lucid, but that might be a good idea, even though I'm a bit skeptical as to whether this is a driver issue, because it looks like the problem started happening after updates were made within a day after the beta-1 release, and the driver file didn't change, as far as I could tell.

[Peter Belew (peterbe)]

I just rebuilt the openchrome driver on this system, without any patches, and there is no change - the x-origin of the screen coordinates is in the middle of the screen (for 1024x769) and at different horizontal positions for other resolutions. So I think it's likely this is elsewhere in xorg.

[Peter Belew (peterbe)]

Also the tty consoles still don't work - if I ctrl-alt-F2 for example, I get a blank, black screen, though I can alt-F7 to return to X.

[Peter Belew (peterbe)]

I'm attaching Xorg.0.log, after updating the system a few minutes ago, then trying to switch to a console with ctrl-alt-F3, then switching back with alt-f7.

[Peter Belew (peterbe)]

Here is an Xorg.0.log from about 3 minutes earlier.

[Bartosz Kosiorek (gang65)]

Please use new mode switch by adding to your xorg.conf file following lines:
   Option "modeSwitch" "new"

Is console screen works correctly after that?

[Peter Belew (peterbe)]

No, adding that option to xorg.conf does not help. The screen x-origin
is still shifted over (always in the center for the login screen, in
varying positions for different X screen modes). Also switching to a
ttyN console still results in a blank screen.

There's another really strange effect that sometimes happens while
booting - after any disk checking, gradually the text screen buffer is
clouded over - as if a dark cloud is moving up the screen. Finally the
screen goes black, when this happens. And once just a few minutes ago
there was a segfault while booting.

On Mon, Dec 21, 2009 at 2:38 PM, Bartosz <email address hidden> wrote:
> Please use new mode switch by adding to your xorg.conf file following lines:
>   Option "modeSwitch" "new"
>
> Is console screen works correctly after that?
>
> --
> [K8M800] Screen origin shifted to center, screen wraps, ttyN screens don't work, on Averatec 3280 with openchrome driver
> https://bugs.launchpad.net/bugs/495553
> You received this bug notification because you are a direct subscriber
> of the bug.
>

--
It's completely normal that you lose data on "Windows" platforms.
That's why you have a UNIX or Linux or BSD server for backups.

SMAUG: http://scruz.org/
My Web:http://littlegreenmen.armory.com/~peterbe/
UBUNTU 10.04 LTS April 2010
  https://wiki.ubuntu.com/LucidLynx

[Peter Belew (peterbe)]

Unfortunately, later today I updated the system, and much of the system was wiped out, and it will no longer boot into Lucid. The system still has Jaunty on it, but with the Grub 2 for Karmic and Lucid. (Lucid was installed over Karmic).

I might be able to rescue the /var/log/Xorg.0.log file from Jaunty.

MAYbe I'll try the beta-2 next month.

[Peter Belew (peterbe)]

Here is a tar file with all the Xorg.*.log* files in it that are still on the system after lucid was made unbootable. This was obtainable by running Jaunty on the same dual-boot system and mounting the volume with "/" on it.

[Peter Belew (peterbe)]

This archive contains dmesg, kern.log, and daemon.log

[Peter Belew (peterbe)]

I think I'll wait a couple of days, re-install Lucid, update, and see what happens. I keep my /home files in a separate partition for each OS, so it's easy to do that.

[Peter Belew (peterbe)]

Today I re-installed Lucid alpha-1 from the alternate CD. Grub was properly updated so I can boot into either Jaunty or Lucid.

Initially, the login screen appeared to be around 1600x1200. When I logged in, the screen size was set to 1024 x 768, but the hardware cursor only ranged from X=0 to X=800 or so on the screen. However on observing icons in the upper panel, I could see the system was mapping the cursor position to the full screen size, so I could actually click on the various icons (including a logout icon)

Adding

  Option "SWCursor" "true"

to an xorg.conf file I created solved the cursor position problem, at the expense of the flickering of the cursor as it was moved.

Note that I observed a similar discrepancy in position with the hardware cursor, when I originally reported the bug after updating alpha-1.

The only package I've added since reinstalling has been openssh-server, so I can log in to the laptop when there are screen problems.

The state of tty switching is: using Alt-F3 to switch to a console gives me a multicolored screen; using Alt-F7 will switch back to Gnome.

I don't plan to update for a few days, unless there is likely to be a fix for some of these problems.

[Peter Belew (peterbe)]

The attached archive contains a couple of copies of Xorg.0.log, Xorg.0.log and Xorg.0.log.old

The later one was created after I disabled the hardware cursor.

[Peter Belew (peterbe)]

I updated Lucid again with apt-get today, starting with Lucid alpha-1. This brings back the "Screen origin shifted to center" problem described in the original report above. So this problem is in one of the many packages updated today. Note that in any event the hardware cursor coordinates do not correspond to the screen coordinates, so it is necessary to disable the hardware cursor, as described above.

I am attaching a list of all the modules which were updated today at 20:52 Pacific Standard Time.

[Peter Belew (peterbe)]

(Continuing comment #33) As before, switching to a tty with Ctrl Alt Fn results in a blank screen, but I can switch back to X with Alt-F7.

I am appending a list of xorg packages which are installed, obtained by executing

  apt-show-versions | grep xorg

Also, for gdm I show

$ apt-show-versions | grep gdm
gdm/lucid uptodate 2.29.1-0ubuntu9
gdm-guest-session/lucid uptodate 0.14

[Peter Belew (peterbe)]

I have temporarily interrupted looking at this bug in order to install Karmic, to check for any progress on bug # 436341 on the same laptop. I'll re-install Lucid again soon, or wait for Lucid beta-2.

[Peter Belew (peterbe)]

I have again upgraded from Karmic to Lucid Alpha, starting the upgrade around 10 am PST today. The upgrade was done online (not a fresh install this time).

 I rebooted the system after the installation at 12:21 pm PST.

The problem with the screen X-origin being offset has not changed. I'm using the same xorg.conf, specifying a software cursor for the openchrome driver.

I'm attaching an archive of the 4 versions of Xorg.n.log on the system now. I believe the Xorg.1.log* files are from Karmic.

[Peter Belew (peterbe)]

No change as of Mon Jan 4 12:28:58 PST 2010 - bug still exists. Screen still off-center, trying to switch to tty1 through tty6 results in black screen.

[Peter Belew (peterbe)]

No change as of Tue Jan 5 18:32:37 PST 2010.

[Peter Belew (peterbe)]

I have filed a separate bug #504381 regarding related but worse behavior when trying to run Xubuntu beta1, with updates, on the same computer.

[Peter Belew (peterbe)]

A difference between this bug and the behavior reported on bug #504381 is that this bug behaves the same way whether kernel 2.6.32-9-generic or 2.6.32-7-generic (the one on the install CD) is running, while for bug #504381 it is different.

[Peter Belew (peterbe)]

As of Fri Jan 8 11:00:45 PST 2010, there has been no apparent change in the behavior described in this bug report. The X screen is still split, and ttys still don't work.

[Peter Belew (peterbe)]

No improvement after updates as of Sat Jan 9 11:30:49 PST 2010

[Peter Belew (peterbe)]

As of Tue Jan 12 08:57:44 PST 2010, this seems to have been fixed. I updated the system with apt-get a few minutes ago, a large number of packages were updated. After rebooting, the screen now appears normally. I am not sure what package(s) were changed to fix the problem.

[Peter Belew (peterbe)]

This is not competely fixed. When I try to switch to tty screens, I get a multicolored background, and the ttys don't work. I can, however, switch back to the X display with alt-F7.

[Peter Belew (peterbe)]

Also note that this is dependent on the kernel version. The X display origin problem not happen with the lastest kernel, 2.6.32-10-generic #14-Ubuntu SMP, but it does happen with the previous kernel, 2.6.32-9. (See Bug #508456, describing another problem that has turned up in the last day).

The problem with switching to tty's still occurs.

[Peter Belew (peterbe)]

The TTY problem is still happening - switching to tty1...tty6 with ctl-alt-Fn results in a screen covered with random colored blocks, and alt-F6 switches back normally. The same colored screen shows while shutting down or rebooting.

[Bartosz Kosiorek (gang65)]

Hi.
Try apply this patch (i have added some comments) and attach Xorg.0.log file.
It will be very useful to resolve this bug.

[Peter Belew (peterbe)]

Hi Bartosz -

The result of doing this patch is ...

Xorg.0.log:(II) CHROME(0): Module DDC loaded
Xorg.0.log:(II) CHROME(0): Detected I2CBus1
Xorg.0.log:(II) CHROME(0): Not detected DDC2

I hope that information helps!

Thanks,

- Peter

On Wed, Jan 20, 2010 at 7:18 AM, Bartosz <email address hidden> wrote:
> Hi.
> Try apply this patch (i have added some comments) and attach Xorg.0.log file.
> It will be very useful to resolve this bug.
>
> ** Attachment added: "edid.patch"
>   http://launchpadlibrarian.net/38138351/edid.patch
>
> --
> [K8M800] Screen origin shifted to center, screen wraps, ttyN screens don't work, on Averatec 3280 running Lucid with openchrome driver
> https://bugs.launchpad.net/bugs/495553
> You received this bug notification because you are a direct subscriber
> of the bug.
>

[Peter Belew (peterbe)]

And here is the whole Xorg.0.log, after doing the patch above and rebooting:

[Bartosz Kosiorek (gang65)]

Hi
Apply this patch and attach your Xorg.0.log file after that.

[Peter Belew (peterbe)]

Should I apply that to the original via_driver.c file or on top of the
previous patch?

On Wed, Jan 20, 2010 at 2:40 PM, Bartosz <email address hidden> wrote:
> Hi
> Apply this patch and attach your Xorg.0.log file after that.
>
> ** Attachment added: "ddc2.patch"
>   http://launchpadlibrarian.net/38168035/ddc2.patch
>
> --
> [K8M800] Screen origin shifted to center, screen wraps, ttyN screens don't work, on Averatec 3280 running Lucid with openchrome driver
> https://bugs.launchpad.net/bugs/495553
> You received this bug notification because you are a direct subscriber
> of the bug.
>

--
It's completely normal that you lose data on "Windows" platforms.
That's why you have a UNIX or Linux or BSD server for backups.

SMAUG: http://scruz.org/
My Web:http://littlegreenmen.armory.com/~peterbe/
UBUNTU 10.04 LTS April 2010
  https://wiki.ubuntu.com/LucidLynx

[Peter Belew (peterbe)]

(I applied the patch to the original file)
Here's the Xorg.0.log file ...

[Peter Belew (peterbe)]

I'm not sure the last Xorg.0.log was valid ... I'm going to start with a new source tree later this evening and try again.

I ended up with X not starting the last couple of times I tried booting - I had to start in text mode and start daemons manually.

[Bartosz Kosiorek (gang65)]

Try to apply this patch and check if it resolve resolution problems.

[Peter Belew (peterbe)]

I was unable to patch:

peter@eros:~/openchrome/src$ patch -p0 < vbeEDID.patch
patching file src/via_driver.c
Hunk #1 FAILED at 1566.
Hunk #2 FAILED at 1620.
2 out of 2 hunks FAILED -- saving rejects to file src/via_driver.c.rej
patching file src/via_dri.c
Hunk #1 FAILED at 267.
1 out of 1 hunk FAILED -- saving rejects to file src/via_dri.c.rej

Perhaps an earlier patch was assumed? I had just checked out the sources, to make sure I was working with the original sources.

Also, when this does work, should I test without the xorg.conf file in place?

[Bartosz Kosiorek (gang65)]

Apply this on the latest and clear openchrome trunk revision.

Don't mix this patches.

[Peter Belew (peterbe)]

Trying the patch again resulted in some errors. I am attaching a .tgz file with the files produced in openchrome/src/src/

Next, I'll attach a .tgz file containing Xorg.0.log files.

[Peter Belew (peterbe)]

This contains Xorg.0.log, etc.

[Peter Belew (peterbe)]

When I try to boot with the driver I just created, the system hangs after the mouse cursor first appears, in a manner similar to that described in Bug #508456:

At the point at which the system freezes, the initial rotating cursor has appeared, and then has just turned into a pointer. This is slightly different from Bug #508456, in which it freezes about 2 seconds after the rotating cursor appears. Also, that problem hasn't occurred for the standard build of the system for about a week. So this may be a slightly different problem.

[Peter Belew (peterbe)]

I reinstalled the current binary driver, and now the system hangs before showing a cursor at all - I just get a blank screen.

However, in recovery mode, I can always start up fine, executing this script after logging in as root:

#!/bin/sh
echo start some services like the ssh daemon and apache
service ssh start
service apache2 start
echo start X
service gdm start

I always start the ssh daemon - in case the screen doesn't work, I can then ssh in over my LAN. It would be very nice if Ubuntu always gave the option of installing the ssh daemon in an installation - the FreeBSD installer has a checkbox for doing that - it's very useful when things go wrong!

[Peter Belew (peterbe)]

Note my comment #2 in Bug #508456 - that I think there is a problem with 'upstart' when the screen freezes on startup.

[Bartosz Kosiorek (gang65)]

Hi Peter.
So after apply patch "vbeEDID.patch" your system freeze during start?

Attach your /var/log/Xorg.0.log file (or /var/log/Xorg.0.log.old after reboot).

[Peter Belew (peterbe)]

Again, applying the patch gives the error messages:

peter@eros:~/openchrome/src$ patch -p0 < vbeEDID.patch
patching file src/via_driver.c
Hunk #1 FAILED at 1566.
Hunk #2 FAILED at 1620.
2 out of 2 hunks FAILED -- saving rejects to file src/via_driver.c.rej
patching file src/via_dri.c
Hunk #1 FAILED at 267.
1 out of 1 hunk FAILED -- saving rejects to file src/via_dri.c.rej

The files via_dri.c and via_driver.c did not change.

On Sun, Jan 24, 2010 at 12:41 AM, Bartosz <email address hidden> wrote:
> Hi Peter.
> So after apply patch "vbeEDID.patch" your system freeze during start?
>
> Attach your /var/log/Xorg.0.log file (or /var/log/Xorg.0.log.old after
> reboot).
>
> --
> [K8M800] Screen origin shifted to center, screen wraps, ttyN screens don't work, on Averatec 3280 running Lucid with openchrome driver
> https://bugs.launchpad.net/bugs/495553
> You received this bug notification because you are a direct subscriber
> of the bug.
>

[Peter Belew (peterbe)]

Note that running 'lspci -vvk' tells me:

01:00.0 VGA compatible controller: VIA Technologies, Inc. K8M800/K8N800/K8N800A [S3 UniChrome Pro] (rev 01)
 Subsystem: TWINHEAD INTERNATIONAL Corp Device 0315
 Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
 Status: Cap+ 66MHz+ UDF- FastB2B- ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
 Latency: 64 (500ns min)
 Interrupt: pin A routed to IRQ 16
 Region 0: Memory at f4000000 (32-bit, prefetchable) [size=64M]
 Region 1: Memory at fd000000 (32-bit, non-prefetchable) [size=16M]
 Expansion ROM at fe2f0000 [disabled] [size=64K]
 Capabilities: <access denied>
 Kernel driver in use: viafb
 Kernel modules: viafb

Also, I can remove the 'Driver "openchrome"' line from xorg.conf and the display works as before, with no reference to openchrome.

Another thing I just noticed, having just done that but also having done various apt-get updates, that it is now Alt F8 that switches from tty consoles to X, instead of Alt F7.

The tty consoles still show random colors, and do not function.

[Peter Belew (peterbe)]

Note that booting in recovery mode and running "Xorg -configure" does produce an xorg.conf.new file that sets 'openchrome' to be the screen driver.

[Peter Belew (peterbe)]

Also see bug #508456 regarding upstart problems on this laptop, which may be related to these display driver problems.

Current issues remaining are:

  Unusable consoles - colored gibberish on the screen.
  Starting up in graphics mode results in wrong resolution,
    without an xorg.conf file. Having xorg.conf, or setting res.
    with System>Preferences>Display dialog fixes final
    Gnome res. - but xorg.conf needed to have right res. for
    login screen.
  Starting up in recovery (text) mode results in text
    display with too many lines - new lines are entered off
    (below) the visible screen.
  Unreliable start up (may be upstart bug)
  Totally blank screen for extended time during boot - no background.
  During shutdown, flashing colored screen (as with tty consoles)
    sometimes.

[Peter Belew (peterbe)]

SEVERE REGRESSION!!!! This has regressed to coming up in the split-screen mode - i.e. at 1024x768 the x-origin of the screen is in the middle.

I had just done an apt-get update/dist-upgrade in which the latest change was a new 'plymouth' with its associated library file.

Ubuntu version and kernel version
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=10.04
DISTRIB_CODENAME=lucid
DISTRIB_DESCRIPTION="Ubuntu lucid (development branch)"
Linux eros 2.6.32-12-generic #16-Ubuntu SMP Wed Jan 27 18:34:19 UTC 2010 i686 GNU/Linux
model name : Mobile AMD Sempron(tm) Processor 2800+

I'm attaching the output of apt-show-versions.

[Peter Belew (peterbe)]

I am attaching Xorg.0.log from before and after this update.

[Bartosz Kosiorek (gang65)]

Hello.
Try to set Virtual size in your xorg.conf file, as follows:

Section "Screen"
 Identifier "Default Screen"
 Monitor "Configured Monitor"
 Device "Configured Video Device"
 SubSection "Display"
  Virtual 1024 768
 EndSubSection
EndSection

[Peter Belew (peterbe)]

Hi - I had Virtual set that way already.

I just tried booting with the latest updates, a few minutes ago, with
the 2.6.32-12-generic kernel, and it failed in recovery mode as well
as in normal mode. But it does boot properly with the 2.6.32.11
kernel, at least in recovery mode, with the proper screen view. It
does not start reliable with that kernel in graphics mode, though it
did the first time I tried it a few minutes ago.

On Tue, Feb 2, 2010 at 1:46 PM, Bartosz <email address hidden> wrote:
> Hello.
> Try to set Virtual size in your xorg.conf file, as follows:
>
> Section "Screen"
>        Identifier      "Default Screen"
>        Monitor         "Configured Monitor"
>        Device          "Configured Video Device"
>        SubSection "Display"
>                Virtual 1024 768
>        EndSubSection
> EndSection
>
> --
> [K8M800] Screen origin shifted to center, screen wraps, ttyN screens don't work, on Averatec 3280 running Lucid with openchrome driver
> https://bugs.launchpad.net/bugs/495553
> You received this bug notification because you are a direct subscriber
> of the bug.
>

--
It's completely normal that you lose data on "Windows" platforms.
That's why you have a UNIX or Linux or BSD server for backups.

SMAUG: http://scruz.org/
My Web:http://littlegreenmen.armory.com/~peterbe/
UBUNTU 10.04 LTS April 2010
  https://wiki.ubuntu.com/LucidLynx

[Bartosz Kosiorek (gang65)]

To resolve this problem you must enable newModeSwitch.
Unfortunately it doesn't work properly on chipset K8M800.
Please try to enable the newModeSwitch on working openchrome - on kernel which works (I should know that this is not kernel problem).

To enable new mode switch, add to your xorg.conf file following line on the latest openchrome revision:
   Option "modeSwitch" "new"

The system hangs after restart, but please attach the Xorg.0.log file from it.
Attach your ~/.xsession-errors file

[Peter Belew (peterbe)]

I've had the 'Option "modeSwitch" "new"' in xorg.conf already.

What is happening with the 2.6.32-12-generic kernel is that the screen
starts getting dark (mostly dark gray) after selecting the recovery
mode, but I can type commands 'blind' to start up Xorg - first
selecting the root shell from the menu, then entering my root
password, then running a script which starts up some daemons - and
then the login screen will appear. So it doesn't really hang up in
recovery mode, I just can't see what I am typing.

The time it takes to get dark seems to vary. I think this is a viafb
driver problem - does that sound right? Up to a day or so ago, the
Xorg.0.log was showing a failure to find viafb when booting failed, a
number of times. Now I've been seeing situations where Xorg.0.log is
empty, instead.

I'm sending this from another computer - I'll log in to launchpad on
the laptop and attach the .xsession-errors and latest Xorg.0.log
files now.

On Tue, Feb 2, 2010 at 11:31 PM, Bartosz <email address hidden> wrote:
> To resolve this problem you must enable newModeSwitch.
> Unfortunately it doesn't work properly on chipset K8M800.
> Please try to enable the newModeSwitch on working openchrome - on kernel which works (I should know that this is not kernel problem).
>
> To enable new mode switch, add to your xorg.conf file following line on the latest openchrome revision:
>   Option "modeSwitch" "new"
>
> The system hangs after restart, but please attach the Xorg.0.log file from it.
> Attach your ~/.xsession-errors file
>
> --
> [K8M800] Screen origin shifted to center, screen wraps, ttyN screens don't work, on Averatec 3280 running Lucid with openchrome driver
> https://bugs.launchpad.net/bugs/495553
> You received this bug notification because you are a direct subscriber
> of the bug.
>

--
It's completely normal that you lose data on "Windows" platforms.
That's why you have a UNIX or Linux or BSD server for backups.

SMAUG: http://scruz.org/
My Web:http://littlegreenmen.armory.com/~peterbe/
UBUNTU 10.04 LTS April 2010
  https://wiki.ubuntu.com/LucidLynx

[Peter Belew (peterbe)]

Here's the Xorg.0.log file.

In this case, I booted in recovery mode, and as the screen darkened, I was able to type commands to log in as root and start Xorg and other daemons even though I couldn't see what I was typing.

Part of xorg.conf:

Section "Device"
        Driver "openchrome"
        VendorName "VIA Tech"
        BoardName "openchrome"
        Identifier "Configured Video Device"
        Option "SWCursor" "true"
        Option "modeSwitch" "new"
        Option "PanelSize" "1024x768"
        #Option "PanelID" "0" #640x480, Single, Dithering
        # Option "EnableAGPDMA" "Off"
        # Option "NoAccel" "true"
EndSection

[Peter Belew (peterbe)]

Here is ~/.xsession-errors

[Peter Belew (peterbe)]

Note that I just updated another bug relating to tty screens on another system with a VIA chipset - Bug #448911.

[Peter Belew (peterbe)]

As of Thu Feb 4 14:57:56 PST 2010, after more updates, my screen is now positioned properly, and console TTYs work ok on this laptop.

I'm enclosing an archive of apt-show-versions output.

[Peter Belew (peterbe)]

But now, when I switch to a tty, I start getting kernel messages like these from /var/log/kern.log:

Feb 4 15:09:54 eros kernel: [ 1084.808079] phy0 -> rt2500pci_set_device_state: Error - Device failed to enter state 1 (-16).
Feb 4 15:09:57 eros kernel: [ 1087.924034] phy0 -> rt2500pci_set_device_state: Error - Device failed to enter state 1 (-16).
Feb 4 15:09:58 eros kernel: [ 1088.124034] phy0 -> rt2500pci_set_device_state: Error - Device failed to enter state 1 (-16).
Feb 4 15:10:05 eros kernel: [ 1095.904031] phy0 -> rt2500pci_set_device_state: Error - Device failed to enter state 1 (-16).
Feb 4 15:10:07 eros kernel: [ 1097.028038] phy0 -> rt2500pci_set_device_state: Error - Device failed to enter state 1 (-16).

[Peter Belew (peterbe)]

BTW I am actually using the rt2500pci wi-fi device at the moment, when getting the messages in the last comment.

[Peter Belew (peterbe)]

I don't think those messages are a bug. :)

[Leann Ogasawara (leannogasawara)]

Hi Peter,

From comment #76 is sounds like this issue you originally reported is now resolved. As a result I'm going to go ahead and close the linux task. For the messages you're seeing with your wifi card, could you open a new bug report for that. Thanks.

[Peter Belew (peterbe)]

Actually, not - read the previous comment.

On Thu, Feb 4, 2010 at 4:57 PM, Leann Ogasawara
<email address hidden> wrote:
> Hi Peter,
>
> >From comment #76 is sounds like this issue you originally reported is
> now resolved.  As a result I'm going to go ahead and close the linux
> task.  For the messages you're seeing with your wifi card, could you
> open a new bug report for that.  Thanks.
>
> ** Changed in: linux (Ubuntu)
>       Status: New => Fix Released
>
> --
> [K8M800] Screen origin shifted to center, screen wraps, ttyN screens don't work, on Averatec 3280 running Lucid with openchrome driver
> https://bugs.launchpad.net/bugs/495553
> You received this bug notification because you are a direct subscriber
> of the bug.
>

--
It's completely normal that you lose data on "Windows" platforms.
That's why you have a UNIX or Linux or BSD server for backups.

SMAUG: http://scruz.org/
My Web:http://littlegreenmen.armory.com/~peterbe/
UBUNTU 10.04 LTS April 2010
  https://wiki.ubuntu.com/LucidLynx

[Peter Belew (peterbe)]

Correction - the starting problems are bug #508456. Scott James Remnant feels that this is a screen driver problem, rather than an upstart problem, but I'm not sure about that.

Yes, the wifi card messages are a separate issue.

There is still an issue with the screen going into a strange mode with many colors in what looks like out-of-sync text graphics while shutting down. Aside from that this bug seems to fixed in other respects.

[Peter Belew (peterbe)]

This has just regressed severely. It is back to the situation where the x origin is in the middle of the screen in 1024x768 resolution.

Also I get a black, blank screen when shifting to a tty.

[Peter Belew (peterbe)]

This is an archive of apt-show-versions output after this morning's update

[Peter Belew (peterbe)]

Here is an archive of Xorg.0.log*

[Peter Belew (peterbe)]

This is still happening as of the latest updates this morning.
The screen x-origin is still offset, and the screen image wraps around, as before.
Selecting a text console with ctl-alt-F1 etc. results in a blank screen.
When the system is booting (after a selection is made in grub), the screen gradually gets obscured by a greyish, cloudy image; a graphics boot usually fails, and selecting recovery mode is difficult but possible due the screen being obscured - I have to type the mode selection, root password, and command to start daemons blindly.

I have attached apt-show-versions output, to show the current package versions on the system.

[Peter Belew (peterbe)]

Also see Bug #508456 - regarding startup.

[Peter Belew (peterbe)]

This is still happening on my Averatec 3280 laptop as of Mon Feb 15 7:53 AM PST

[Peter Belew (peterbe)]

This is still happening on my Averatec 3280 laptop as of Tue Feb 16 23:16:47 UTC 2010 (Tue Feb 16 15:17:07 PST 2010)

[Peter Belew (peterbe)]

Note that this problem was temporarily fixed earlier this month (see comment #76). I suspect examining the driver and other software versions installed at that point may lead to a solution to this bug.

[Peter Belew (peterbe)]

Still happening as of Fri Feb 19 17:49:20 UTC 2010 (Fri Feb 19 09:49:21 PST 2010), after update.

Screen origin is still offset (with wraparound of video), tty screens are still blank (black), screen is still obscured during boot with cloudy image.

I'm attaching list of packages as of this update.

[Peter Belew (peterbe)]

Also attaching Xorg.0.log

[Peter Belew (peterbe)]

This bug is still in effect up through immediately before release of Lucid alpha-3 yesterday. I tried alpha-3 (as a live boot) on 5 of my computers today (including this one), and submitted Bug #528767 regarding my results.

I haven't tried updating yet.

[Peter Belew (peterbe)]

The system has been updated to alpha-3 and updated since then, up to a few minutes ago. These problems are still happening.

A this point, I am close to give up on testing lucid, in which case I will install Karmic instead and check to see if any bugs have been fixed in that since the release. Hopefully someone will pay attention to this serious bug and others, and I can continue with Lucid.

[Peter Belew (peterbe)]

I have installed Karmic instead of the Lucid alpha-3 on this laptop, but will check the Lucid beta-1 live CD release when it appears.

There has been one minor improvement with this bug - the cloudy image which obscures boot messages while the system was booting disappeared with todays Lucid update.

[Peter Belew (peterbe)]

I upgraded to the Lucid beta-1 starting one instance of Karmic on this laptop yesterday (this is a dual-boot system now, with Karmic and Lucid on it) and have installed some updates since.

The problem with the offset horizontal position of the screen (with wrapping around), as described above, remains.

Trying to switch to a text console with, for example, ctrl-alt-F2, results in a blank screen.

In addition, there seems to be a new problem. Even with /etc/X11/xorg.conf in place, the initial login screen is the wrong size - something like 1600x1200, so the login window is over toward the lower right corner of the screen. Once logged in, the screen background takes several seconds to resize to 1024x768.

I attaching a tar file with xorg.conf, dmesg, and /var/log/Xorg.0* files.

[Peter Belew (peterbe)]

I'd like to point out that the behavior of the driver looks to me like there is a 512-pixel delay or advance in the timing of horizontal sync. At 1024x768 this means the origin of the screen is shifted over half the screen width, and at 800x600 this means the origin is near the right side of the screen - 512/600 of the screen width. At the next lower resolution the origin shifts farther, and wraps over to near the left side of the screen.

[Peter Belew (peterbe)]

I was mistaken about one point in comment #96 - the login window IS centered when xorg.conf is in place.

[Peter Belew (peterbe)]

There is no impovement in this bug with the most recent build of the openchrome driver.

apt-show-versions xserver-xorg-video-openchrome
xserver-xorg-video-openchrome/lucid uptodate 1:0.2.904+svn827-1

[Peter Belew (peterbe)]

Please ignore the what I said in comment #96 about the login screen, and all of #98 - those comments apply to Karmic on the same computer.

Also, I was miscalculating in #97 regarding the 800x600 screen. The full screen is 246 mm wide, and the x-origin is shifted over to 226 mm. That is about 735 pixels.

[Peter Belew (peterbe)]

Should I report this directly to the openchrome project? Might that help?

[Georg Duffner (mcduff)]

I can confirm this behaviour. It only applies to kernels from the 2.6.32 series. Kernels from the 2.6.31 series are not affected: if I start this machine with 2.6.31-20 from Karmic (this machine has been upgraded from Karmic), the split-screen issue doesn’t appear.
My machine is a BenQ Joybook R22E with K8M800 chipset.

[Peter Belew (peterbe)]

This is still happening as of today. The screen is split, and wraps around, as described above, and switching to a text console results in a blank screen.

This does not happen under Karmic on this dual-boot system, so the kernel used is possibly an issue.

To Georg Duffner - when you use the 2.6.31-20 kernel, how much of Lucid are you using - is that kernel still in the /boot directory along with later Lucid kernels?

Also, today I installed Sun Virtualbox on this system, and was able to install Lucid beta-1 in a VM. The performance is extremely poor on this computer, as it only has 512 mb of RAM, which is split between the host (Karmic) and VM (Lucid under VirtualBox) systems - both systems are constantly thrashing - swapping a lot.

Is there actually a fix in progress - I see that at the top of the page - it hasn't appeared yet.

[Georg Duffner (mcduff)]

I have done a complete dist-upgrade to Lucid, so I should be using everything in its Lucid version. I only don’t have removed Karmic’s kernel from the install—it is still there where it was before, in /boot, along with Lucid kernels. I always keep old kernels for a while when I upgrade to alpha/beta versions.

[Peter Belew (peterbe)]

Georg Duffner - thank you. It's possible that this is a problem in the kernel framebuffer or in some other Lucid component relating to the newer kernel.

I still have kernel 2.6.31-20-generic in my Lucid /boot directory, and was able to boot into this. The login screen came up properly, and the Gnome screen is now properly aligned. Possible reasons this works:

1. The framebuffer is different between Karmic and Lucid kernels
2. Some component of Lucid wasn't started because of the old kernel - there were some screen messages during boot which might indicate that. Such a non-started component may have been the cause of the problem.

[Peter Belew (peterbe)]

I'm attaching some logs. The current kernel version - from Karmic - is given here:

Ubuntu version and kernel version
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=10.04
DISTRIB_CODENAME=lucid
DISTRIB_DESCRIPTION="Ubuntu lucid (development branch)"
Linux eros 2.6.31-20-generic #58-Ubuntu SMP Fri Mar 12 05:23:09 UTC 2010 i686 GNU/Linux
model name : Mobile AMD Sempron(tm) Processor 2800+

The kern.log file doesn't appear to show anything about running 2.6.31-20 - only from previous boots with Lucid kernels.

[Peter Belew (peterbe)]

When I boot with the 2.6.31-20-generic #58 kernel (see above) I get the following messages when I first boot:

mount: Mounting none on /dev failed: no such ...
chroot: Cannot execute /etc/apparmor/initramfs: No such file or directory
init: Plymouth main process (283) terminated with status 69
mountall: Could not connect to Plymouth

(I took those down by hand so there may be typos)

Looking around a bit at bugs like Bug #538292 on Plymouth I'm wondering if the problem when booting with Lucid kernels is due to a Plymouth problem.

Note that in addition to the split screen in Gnome and X and blank console screens, a couple of problems remain when booting with Lucid kernels:

- while the system is loading, a hazy image slowly covers the text messages on the screen
- without an xorg.conf file, and before the screen resolution has been set with the System>Preferences>Monitor dialog, the screen sizes at something like 1600x1200 - and the panel clips that to the upper-left 1024x768 pixels.

Note that using a Karmic kernel under Lucid also is a workaround for bug #538893 on a system I have with another model UniChrome display.

[Peter Belew (peterbe)]

The situation is unchanged as of this morning. Spit screen, hazy image while booting, no text consoles, etc.

[Peter Belew (peterbe)]

A major part of this bug was solved by disabling the viafb framebuffer driver - this was done by blacklisting viafb in /etc/modprobe.d/blacklist-framebuffer.conf (thanks to a suggestion by Ian Clark for bug #538893 regarding another system with a UniChrome display) - on the next update, the initrd was rebuilt.

The good news is that the split-screen problem ceased.

The bad news is that I can't still switch into consoles with ctl-alt-F1 through F6 - but instead of getting blank, black, screen, the resulting screen is multicolored. The patterns vary each time I try to switch into a console. This occurred before, as reported above in comments 31, 44, 46, 64, 66, and 82.

[Peter Belew (peterbe)]

So this problem was largely with the viafb driver, but the console problem is something different.

[Peter Belew (peterbe)]

The Beta-2 live cd will not boot properly.

First of all, the screen resolution is STILL set wrong initially, so the screen is cropped to the upper-left 1024x768 corner of a larger (1600x1200?) screen. I was able to get around this by changing the screen resolution after logging in.

Also, I still can't ctl-alt-F3 into a console, as described above.

The new problem is that I can't log in to Gnome - actually, what apparently happens is that the X server crashes immediately after logging in, so the screen goes blank, then a second or two later the splash screen reappears, with a login box. Logging in as 'ubuntu' with a blank password repeats this.

Not being able to use console screens, and not have a telnet or ssh server installed by default, makes this problem impossible to diagnose. It would be really useful to be able to log in over the LAN to diagnose such problems, as one can do optionally with FreeBSD.

[Peter Belew (peterbe)]

Incidentally, this system boots properly with the correct screen resolution from the live CD for the Cuban Nova Linux 30012009 V1 (BAIRE-1.1.2-30012009-V1) which I downloaded today. This is a Gentoo-based distro. It appears to use an openchrome driver for this Averatec system. I've attached a tar file with some logs and xorg.conf

[Peter Belew (peterbe)]

The problem with text consoles is still happening as of 15 Apr 2010. For example, ctrl-alt-F2 results in an unintelligible colored screen. The same for F1 through F6. alt-F7 switches back to Gnome.

I'll try the live-cd booting problem on the RC release, unless someone suggests trying a daily build.

[Bartosz Kosiorek (gang65)]

Attach the Xorg.0.log file before and after switch to console.

[Peter Belew (peterbe)]

I am adding Xorg.0.log before switching to a text console, after switching to a text console, and after switching back to the GUI Gnome screen.

Xorg.0.log - before
Xorg.0.log.console - after switching to (unusable) console
Xorg.0.log.x - after switching back

[Peter Belew (peterbe)]

The above Xorg.0.log files were obtained after updating the system around 14:30 UTC on Apr 16.

The files were saved while logged in from another machine using ssh. Again, I advocate installation of an ssh server an option on live CDs and during installation (I believe the Sabayon Linux DVD allows this; also it provides an option to choose screen resolution manually via one of the function-key options, another thing which would save a lot of grief on some computers).

[Peter Belew (peterbe)]

The booting problem described in #111 is still happening with the 17 April '10 daily build. It's impossible to log in to the X/Gnome display, since as soon as the system logs in, X fails and the GDM login screen reappears after several seconds.

After some experimentation, I was able to log on in single-user mode as follows: Boot, type ctrl-alt-F1 to get a screen which allows one to edit the kernel command line (after choosing the language). I edited the command line to end in " single nosplash nofb --". A list of choices appears, and I chose to log in as root with networking enabled. The system was able to connect using dhcp, and I determined the IP address for the system.

One problem at this point is that the logical screen had more text lines than would fit on the screen, so I was typing blind.

I then set a password for root, and used apt-get to install openssh-server, so I could ssh in and log in from another system. I then tried to start the X server, using 'X', 'runlevel 3' , etc. I was able to get a blank X screen with a cursor. I saved /var/log/Xorg.0.log 3 times during this process. I have attached those.

[Peter Belew (peterbe)]

A clarification to #111 and #117 and the problem description - the live CDs for beta-2 and the april 17 build do boot up, but it's impossible to log in once the live CD has booted.

However if Lucid has been installed previously, and is updated to date, the principal remaining problem is that it is impossible to use text consoles.

Also, there is the remaining issue of determining the screen size on boot - without having set the res. down to 1024x768, and without an xorg.conf file, the system will start trying to use 1600x1200 resolution.

[Bartosz Kosiorek (gang65)]

Hi Peter.
Try to disable AIGLX.
Do you use the latest Lucid Lynx kernel (it solves several problems with VIA chipsets)?

[Peter Belew (peterbe)]

Hi Bartosz -

I'm using the 2.6.32-21-generic #32-Ubuntu SMP kernel, dated Fri Apr 16 08:10:02 UTC - I update with apt-get more or less daily. On the live CD, it's whatever came on the Apr 17 live CD - probably the same, but I didn't check.

I looked at Bug #471646, where some similar boot or login problems with KDE are mentioned, and in which you made a similar suggestion regarding AIGLX. I'm not sure which problem this was supposed to help, though.

I haven't tried any upstream kernel - that would be a bit difficult for the live CD login problem, might be appropriate for other problems.

[Peter Belew (peterbe)]

I downloaded the 19 April daily build live CD, and it still won't allow logging in.

[Bartosz Kosiorek (gang65)]

Try to install Lucid on hard drive and install openchrome package from Karmic Koala (or you can go back using instruction from https://help.ubuntu.com/community/OpenChrome).

I want to make sure is this problem exists in openchrome.

[Peter Belew (peterbe)]

I checked out the driver source specifying "-r 758". I get a fatal error on trying to make the driver - the header file xf86Resources.h is missing, also there are some warnings about formats not matching variables, in via_accel.c, function viaExaUploadToScratch()

[Peter Belew (peterbe)]

The latest Karmic driver for this is

  xserver-xorg-video-openchrome/karmic 1:0.2.903+svn758-0ubuntu1

How, exactly, can I install that in Lucid?

[Bartosz Kosiorek (gang65)]

Try to install package from that (just double click on that):
https://launchpad.net/ubuntu/+archive/primary/+files/xserver-xorg-video-via_0.2.903+svn758-0ubuntu1_i386.deb

If it not work try to get source code from trunk, and just go back to previous version.
You could do this in following way:

svn update -r758

[Peter Belew (peterbe)]

The xserver-xorg-video-via driver installed ok, but it doesn't seem to
make any difference. Should I change or remove xorg.conf?

On Thu, Apr 22, 2010 at 11:53 PM, Bartosz <email address hidden> wrote:
> Try to install package from that (just double click on that):
> https://launchpad.net/ubuntu/+archive/primary/+files/xserver-xorg-video-via_0.2.903+svn758-0ubuntu1_i386.deb
>
> If it not work try to get source code from trunk, and just go back to previous version.
> You could do this in following way:
>
> svn update -r758
>
> --
> [K8M800] Beta-2 live-CD login failure, Screen origin shifted to center, screen wraps, ttyN screens don't work, on Averatec 3280 running Lucid
> https://bugs.launchpad.net/bugs/495553
> You received this bug notification because you are a direct subscriber
> of the bug.
>

--
It's completely normal that you lose data on "Windows" platforms.
That's why you have a UNIX or Linux or BSD server for backups.

SMAUG: http://scruz.org/
My Web:http://littlegreenmen.armory.com/~peterbe/
UBUNTU 10.04 LTS April 2010
  https://wiki.ubuntu.com/LucidLynx

[Bartosz Kosiorek (gang65)]

Sorry I send you wrong package.

Check this package:
http://launchpadlibrarian.net/29586443/xserver-xorg-video-openchrome_0.2.903%2Bsvn758-0ubuntu1_i386.deb

Or just go back in svn by using "svn update -r758" command.

[Peter Belew (peterbe)]

I tried to install that; here are the errors that are reported:

# dpkg -i xserver-xorg-video-openchrome_0.2.903+svn758-0ubuntu1_i386.deb
dpkg: warning: downgrading xserver-xorg-video-openchrome from
1:0.2.904+svn827-1 to 1:0.2.903+svn758-0ubuntu1.
dpkg: regarding
xserver-xorg-video-openchrome_0.2.903+svn758-0ubuntu1_i386.deb
containing xserver-xorg-video-openchrome:
 xserver-xorg-core conflicts with xserver-xorg-video-5
  xserver-xorg-video-openchrome provides xserver-xorg-video-5 and is
to be installed.
dpkg: error processing
xserver-xorg-video-openchrome_0.2.903+svn758-0ubuntu1_i386.deb
(--install):
 conflicting packages - not installing xserver-xorg-video-openchrome
Errors were encountered while processing:
 xserver-xorg-video-openchrome_0.2.903+svn758-0ubuntu1_i386.deb

On Sat, Apr 24, 2010 at 3:42 AM, Bartosz <email address hidden> wrote:
> Sorry I send you wrong package.
>
> Check this package:
> http://launchpadlibrarian.net/29586443/xserver-xorg-video-openchrome_0.2.903%2Bsvn758-0ubuntu1_i386.deb
>
> Or just go back in svn by using "svn update -r758" command.
>
> --
> [K8M800] Beta-2 live-CD login failure, Screen origin shifted to center, screen wraps, ttyN screens don't work, on Averatec 3280 running Lucid
> https://bugs.launchpad.net/bugs/495553
> You received this bug notification because you are a direct subscriber
> of the bug.
>

--
It's completely normal that you lose data on "Windows" platforms.
That's why you have a UNIX or Linux or BSD server for backups.

SMAUG: http://scruz.org/
My Web:http://littlegreenmen.armory.com/~peterbe/
UBUNTU 10.04 LTS April 2010
  https://wiki.ubuntu.com/LucidLynx

[Bartosz Kosiorek (gang65)]

try to compile the source code by using instruction:
https://help.ubuntu.com/community/OpenChrome

After checkout type in console:
"svn update -r758"

then proceed normal configure/build/install procedure.

[Peter Belew (peterbe)]

This still does not work. The make suggests some changes to
configure.ac and to Makefile.am - I did the first and the second
refers to something that's in Makefile.in (I think) but isn't clear.
In any event, the make fails.

On Sun, Apr 25, 2010 at 9:22 AM, Bartosz <email address hidden> wrote:
> try to compile the source code by using instruction:
> https://help.ubuntu.com/community/OpenChrome
>
> After checkout type in console:
> "svn update -r758"
>
> then proceed normal configure/build/install procedure.
>
> --
> [K8M800] Beta-2 live-CD login failure, Screen origin shifted to center, screen wraps, ttyN screens don't work, on Averatec 3280 running Lucid
> https://bugs.launchpad.net/bugs/495553
> You received this bug notification because you are a direct subscriber
> of the bug.
>

--
It's completely normal that you lose data on "Windows" platforms.
That's why you have a UNIX or Linux or BSD server for backups.

SMAUG: http://scruz.org/
My Web:http://littlegreenmen.armory.com/~peterbe/
UBUNTU 10.04 LTS April 2010
  https://wiki.ubuntu.com/LucidLynx

[Peter Belew (peterbe)]

If I boot with Karmic kernel:

2.6.31-20-generic #58-Ubuntu SMP Fri Mar 12 05:23:09 UTC 2010 i686 GNU/Linux

Then some Lucid 'features' are suppressed during startup, and I can switch into text consoles with ctrl-alt-F1 through ...F6 properly. The display driver is still

xserver-xorg-video-openchrome/lucid uptodate 1:0.2.904+svn827-1

I'm attaching a tar file with

apt-show-versions.txt kern.log Xorg.0.log xorg.conf
dmesg versioninfo.txt Xorg.0.log.old

[Peter Belew (peterbe)]

Early after booting (as in #131) I see the message

chroot: cannot execute /etc/apparmor/initramfs

This might be relevant or not.

[Peter Belew (peterbe)]

Apparmor is not the problem.

The Lucid final-release live cd still will not boot on this system, that is, it's impossible to log in.

It's still impossible to use the console text screens when Lucid is installed on this system.

Contrary to the 'Fix released' status at the top, this is NOT fixed.

[Peter Belew (peterbe)]

After looking at a suggestion made for bug #455954, I installed the latest kernel from
http://kernel.ubuntu.com/~kernel-ppa/mainline/daily/current/

namely 2.6.34-999-generic. With that running, I can switch in and out of text consoles with ctrl-alt-F1 etc. (However the system complains about library dependencies while booting, so I suspect this kernel isn't safe to use).

[Peter Belew (peterbe)]

There is no change with the lastest release kernel, as of today:

2.6.32-22-generic #33-Ubuntu SMP Wed Apr 28 13:27:30 UTC 2010 i686 GNU/Linux

I still can't switch into text consoles - I just get garbage on the screen, as before.

Also, the first time I tried to boot after today's updates, the system hung up without getting to the gdm login screen - it just showed the progress dots, which stopped changing. I was able at this point to log in via ssh, however.

[Aeb1barfo (aeb1barfo)]

I fixed this problem in an earlier release. Sorry, I don't have notes on how I did it. Silly me, I upgraded and lost the Linux side of a dual boot Averatec 3280 system and I use the XPpro side until this bug is fixed.

[Peter Belew (peterbe)]

Aeb1barfo: This bug report has gone way too complicated - some of the problems were fixed a long time ago, others persist.

Update: I've run the 10.10 Maverick Meerkat live CD on this Averatec 3280 system, and the remaining problems are:

 - I can't switch to a text console. Also that has gotten worse - on trying to switch back, I end up with a screen background but no top or bottom panels.

- On booting, X starts up in 1600x1200 resolution, so it's necessary to change that after starting up. Adding an xorg.conf file allows for setting the proper startup resolution (1024x768).

[Peter Belew (peterbe)]

As of today, this bug is still in effect. The most important issue is that text consoles can't be used.

Booting and screen sizing issues can be worked around, especially for the Gnome version, since the System>Preferences>Monitor dialog is accessible from top-panel menus; that's more difficult in the case of KDE.

Note that an upgraded kernel was found to solve the text console problem, but it introduced other problems.

The problems seem to be the same (or worse) booting the 10.10 alpha-1 live CD.

Can this be fixed for 10.04.1?

[Peter Belew (peterbe)]

The situation with Lucid on this Averatec laptop is unchanged. However today I tried booting the Fedora 13 i686 live CD. As before, with Fedora and Ubuntu, Linux started up in a 1600x1200 mode, with only the upper-left 1024x768 corner of the screen showing, but I was able to use the System>Preferences>Monitors dialog to select the desired 1024x768 mode.

What is important is that I was able to switch in and out of text consoles - ctrl-alt F2 through ctrl-alt-F6, with alt-F1 switching back to Gnome. I'll attach some files which may show up some module versions that are different from the ones on Lucid.

I've added an archive containing lspci -vvxx output, /var/log/Xorg.0.log, and the output of rpm -qa, which lists installed modules and versions.

This comment was created running the Fedora 13 system.

[Peter Belew (peterbe)]

Kernel information while running Fedora 13:

[liveuser@localhost bug555]$ uname -a
Linux localhost.localdomain 2.6.33.3-85.fc13.i686 #1 SMP Thu May 6 18:44:12 UTC 2010 i686 i686 i386 GNU/Linux

[Peter Belew (peterbe)]

I have totally given up on running Ubuntu on this laptop. I have installed first Fedora 13, and now Debian Lenny. Once properly installed, the screen issues no longer occur, except that both of those systems try to run the screen in 1600x1200 mode, so that installing a custom xorg.conf is also necessary under those distros.

Some distros, such as AntiX, have a boot-time option to select the initial screen resolution. IMHO ALL distros should do this.

I *may* try future betas on this laptop, but don't intend to install Ubuntu on it again.

[Peter Belew (peterbe)]

The 10.10 Maverick Meerkat alpha and beta live CDs will not boot fully on this Averatec 3280. I saw this most recently with the 10.10 Beta and the 10.10 daily build for 19 Sept 2010. What happens is that X login fails.

When I get to the screen which asks whether to boot live or to install the new system, I select the live boot. After a few seconds, it tries to log in automatically (presumably with user 'ubuntu' and password ''). This fails: a login dialog appears after several seconds. If I enter the above username and password, it seems to accept the password, but soon fails again, presenting me with the login dialog.

Note that the login dialog is off center, towards the lower right of the screen. Apparently (as many times previously), a 1600x1200 mode has been chosen.

I suspect that if I were able to log in in text mode, I could copy the xorg.conf from the Debian Lenny installation on the system (which works) to the ramdisk /etc/X11 directory, and possibly enter X.

I have not been able to verify the problems with text displays (F1..F6) reported above.

[Peter Belew (peterbe)]

(See comment #142 for current version and boot problem information). I am able to boot from the live CD in text mode, and eventually get X to start, as follows:

Boot live CD
Hit ESC when the first welcome screen appears
When a screen with boot choices appears, first select the language (English) and then hit F6
Hit ESC to cancel the menu which pops up, and the boot command line appears.
Use backspace to remove "silent splash --" from the end of the line, then enter " -- text", and hit return.
The system will (usually) boot, showing a login prompt after a while.

At this point, I create passwords for 'ubuntu' and 'root' (for convenience) and use apt-get to install 'openssh-server', so I can access the system over my Ethernet if necessary (and I found it to be necessary)

I then copied an xorg.conf file from the Debian Lenny installation on this machine, and ran 'gdm start' as root.

PROBLEM: using ctl-alt-F3 (for example) to get a text screen results in a screen which is virtually unusable, with huge text characters. It is possible to use F7 to restore the X display.

I have attached the xorg.conf file and Xorg.0.log* .

I am actually sending this from Firefox on the booted 10.10 system.

Summary at this point: Booting from the live CD is very difficult but possible by an expert, but F1..F6 text mode selection does not work.

[Peter Belew (peterbe)]

This problem persists with the post-RC 07-oct-2010 build.
Linux distro/version and kernel version
Ubuntu 10.10
Linux ubuntu 2.6.35-22-generic #33-Ubuntu SMP Sun Sep 19 20:34:50 UTC 2010 i686 GNU/Linux
model name : Mobile AMD Sempron(tm) Processor 2800+
cpu MHz : 800.000

If I just boot the live CD, without trying to boot into text mode, eventually the screen for choosing 'trying Ubuntu' versus installing it shows, in 1600x1200 mode - so as before this dialog appears in the lower left-hand corner of the 1024x768 screen. If I then click on the 'trying' button, eventually the screen comes up in an unusual state - somewhat similar to an early 'split-screen' bug (shown in an early comment in this bug report) but completely unusable - nothing but purple horizontal lines, with a clear dividing line down the center.

However, if I hit ESC right after booting, and use F6 followed by ESC to get an editable command line, and end that with "-- text" (as in a couple of comments back), I eventually end up booting in text mode. At that point, I can create passwords for root and ubuntu, install openssh-server, copy an xorg.conf file from my Debian installation into /etc/X11, and start X with 'gdm start'. THEN the Gnome screen will eventually appear in 1024x768 mode. (The xorg.conf file is attached a few comments back).

(Note that Debian Lenny works fine on this computer, and the xorg.conf file is one I use under that distro).

However, trying to switch into text mode with ctrl-alt-F2 (for example) results in a screen with what looks like huge text characters. Switching back into X seems to require F8 rather than the usual F7, which is odd.

I have attached a tar file with (a) Xorg.0.log and Xorg.0.log.old before trying to switch into text mode, and (b) after trying this. (Having the openssh-server installed was helpful for doing this, so I could log in from another computer).

[Peter Belew (peterbe)]

Ubuntu 11.04 'natty narwhal' seems to work well on this system, which is pleasantly surprising!

I downloaded the ISO 2 days ago, and burnt a DVD from the image. As before, the live CD was unable to determine the actual resolution of the screen (1024x768, Unichrome controller) and set a higher resolution. However the System>Preferences>Monitors dialog was accessible, so I was able to set the proper resolution, and the copy in the xorg.conf file I have used previously, log out, then log in again in 1024x768 resolution. I was able to switch to text consoles and back without any problems. So I decided to install the 11.04 system (instead of Debian Lenny, which I had been using). On rebooting, setting 1024x768 resolution, and installing the xorg.conf file, the installed system worked properly.

I note that Unity does not work with this display and the Openchrome driver, but Gnome works (with a nagging popup reminding me about that!). After some confusion about the popup, I finally found that I was able to use the System>Administration>Login Screen dialog to set the legacy Gnome UI as the default. I left a comment about that on the Ubuntu forum.

After that, installation of Apache2, MySQL, PHP5, openssh-server, and other software I use frequently went well. I'll keep updating 11.04 until release.

[Bartosz Kosiorek (gang65)]

Hi Peter.
Attach the Xorg.0.log file from Ubuntu 11.04 'natty narwhal'. Make sure that you don't have set the xorg.conf file.

[Peter Belew (peterbe)]

Hi Bartosz - I removed the xorg.conf file and rebooted. The login background and login dialog appear in the 1600x1200 mode (note that the grub2 screen appears properly, using a custom background I installed).

I've attached the Xorg.0.log file.

[Peter Belew (peterbe)]

After that, I proceeded to log in. The screen came up in 1024x76 mode, because I had previously set that resolution with the System>Preferences>Monitors dialog.

So I used System>Preferences>Monitors to set the resolution to 1600x1200, and rebooted. After I logged in, the screen came up in 1600x1200, but the top panel didn't appear. I then retrieved the Xorg.0.log file and have attached it (as Xorg.0.log2).

[Peter Belew (peterbe)]

On rebooting another time, and logging in, the screen came up 'properly' in 1600x1200 mode, with the top panel showing correctly. I've attached Xorg.0.log3 after that.

[Peter Belew (peterbe)]

I then set the res to 1024x768, using the System>Preferences/Monitor dialog, and logged out and logged in (without rebooting the OS). This time the screen came up in 1024x768 without the top panel, but showing the bottom panel. I've attached yet another Xorg.0.log (-.log4) file.

THEN I rebooted completely. The login screen showed as 1600x1200, but the final screen showed as 1024x768.

[Peter Belew (peterbe)]

Note that the tests of Natty on my Averatec - last few comments - were using the Ubuntu Classic Session - Gnome. Unity will not work on the Unichrome display hardware.

[Peter Belew (peterbe)]

A new bug appeared yesterday: frequently the top and bottom panels don't appear after login. I just filed this bug separately as #689377. This happens most of the time in Classic Mode (Gnome), but not in Gnome Safe Mode.

[Peter Belew (peterbe)]

Note that further comments were added to Bug #689377 today, as further problems appeared. I don't think these problem were due to the display driver, but they do affect testing of display problems somewhat, and vice versa. For example, if the xorg.conf file is removed, it's not possible to go into safe mode, because the selection of safe and other modes is in a bottom bar, which is off the screen when the system tries to start in 1600x1200 mode.

So THIS bug is very much alive in the sense that being able to get a useable screen mode at initial startup is important.

On another system, poor resolution detection at startup resulted in a 800x600 screen on a 1024x768 monitor - this was acceptable, since adding an xorg.conf file meant that only the login screen was at the lower resolution, and that was quite useable. But this is not. The situation would be much worse with KDE, with the panel at the bottom of the screen.

[Peter Belew (peterbe)]

Actually, the remaining problem regarding the initial screen resolution is covered by Bug #186103 - other issues on THIS bug seem to have been resolved.

[Peter Belew (peterbe)]

There has been a severe regression TODAY in 11.04 - now the tty screens don't work. Ctrl-alt-F1 etc switch to a random screen which doesn't sync.

[Peter Belew (peterbe)]

I just reinstalled the current openchrome driver from the repository, rebooted, and now switching to text modes works properly, though I have to have xorg.conf installed to get the right resolution. I'll try rebuilding the driver according to Bug #186103 and see if that helps this issue.

[Bartosz Kosiorek (gang65)]

Thanks Peter for your investigation

So this is regression in openchrome. Could you please go back to the previous openchrome trunk versions by following command

svn update -r887

and check where exactly this error appear?

Does this bug appear in openchrome revision 888?

[Bartosz Kosiorek (gang65)]

Check also if this bug exists in the latest openchrome revision (r891):
http://openchrome.org/trac/changeset/891/trunk

[Peter Belew (peterbe)]

I'll try both 887 and 891 (and 888?) in about 24 hours - I'm out of town today.

On Sun, Dec 19, 2010 at 11:58 AM, Bartosz <email address hidden> wrote:
> Check also if this bug exists in the latest openchrome revision (r891):
> http://openchrome.org/trac/changeset/891/trunk
>
> --
> You received this bug notification because you are a direct subscriber
> of the bug.
> https://bugs.launchpad.net/bugs/495553
>
> Title:
>  [K8M800] Maverick 10.10 live-CD login failure,  ttyN screens don't work, on Averatec 3280 running Lucid
>

--
It's completely normal that you lose data on "Windows" platforms.
That's why you have a UNIX or Linux or BSD server for backups.

SMAUG: http://scruz.org/
UBUNTU 10.10 October 2010
  https://wiki.ubuntu.com/LucidLynx

[Peter Belew (peterbe)]

See http://www.openchrome.org/trac/ticket/366 - The initial screen res seems to be set at 1024x768, and switching to/from text screens works too, with r891 of the OpenChrome driver.

Thanks, Bartosz!

[Peter Belew (peterbe)]

A serious problem has just arisen after an update today. I did the update while the system was started, but before logging in (I switched to a text console while the login screen was appearing). Once I logged in, in safe mode, the 2 icons I'd created on the desktop were bouncing up and down, and neither top nor bottom panel showed.

Then I rebooted. The login dialog was bouncing up and down so I could not log in. I was still able to switch to a text console and collect some data.

By the way, the screen resolution seemed to have been set properly, but I'm not completely sure about that.

I've attached a .tgz archive with Xorg.0.log and the output of 'apt-show-versions'.

[Peter Belew (peterbe)]

A side issue - the system was unable to support USB devices that were attached - a keyboard and mouse. Console 1 reported errors about that.

[Peter Belew (peterbe)]

The behavior described above was with the Openchrome driver build from source a couple of days ago. I then reinstalled the driver from the repository, and the only difference in behavior was that the initial resolution after booting turns out to be 1600x1200; and that can be changed by installing an xorg.conf file.

It's not clear if this behavior arises from this particular display driver; possibly not.

[Peter Belew (peterbe)]

Bartosz: What I would really like to happen is to have these fixes backported to 10.04.1 LTS, so that they might be released by 10.04.2 update. Is that reasonable? 11.04 seems very unstable, and is not useable at the moment. Perhaps I should repartition the HD and set up a dual boot between 10.04 and 11.04.

[Peter Belew (peterbe)]

As of the last couple of days, it appears that the screen resolution of 1024x768 is set automatically when X starts.

Also, switching to/from text modes seems to work. There is an oddity about switching back from a text console to X: using Alt-F7 just switches to a blank screen with "[ OK ]" in the upper-right corner, while Alt-F8 switches back to X. This is new behavior (to me).

[Bryce Harrington (bryce)]

According to upstream bug report and confirmation by comments on this bug report, the issue has been fixed in recent Ubuntu releases.

If you can identify the exact revision number that provides the fix, it can be considered for backporting to 10.04 under SRU policies. Provide the revision number for the patch and reopen the bug in this case.

[Peter Belew (peterbe)]

Reopen for possible backporting of 11.04 version of this driver to 10.04.3

[Peter Belew (peterbe)]

The version of the openchrome driver which is working properly on this system under Natty 11.04 is:

peter@cupid:~$ apt-show-versions xserver-xorg-video-openchrome
xserver-xorg-video-openchrome/natty uptodate 1:0.2.904+svn916-1build1

Will there be daily builds of 10.04.3 available prior to that release?

[Peter Belew (peterbe)]

I downloaded the 16 July 2011 build of 10.04.3, and apparently the repaired driver has not been backported to this release.

Two issues:

1. the screen comes up in 1600x1200 or something like that, so the bottom right of that screen image does not show on the 1024x768 panel, as before.

2. A login dialog shows up, in the lower right of the screen (center of 1600x1200 image), which means the system tried to start gdm, failed, and showed a login dialog. Trying to log in with 'ubuntu' and a blank password results in the same thing happening again after several seconds.

So this RC of 10.04.3 is unusable on this system.

I wasn't able to access any files generated during system startup.

This system has 11.04 installed, and works with the 11.10 alpha-2 live cd, also with the 2011-07-05 daily build of Kubuntu 11.10. BTW this 11.10 Kubuntu is the first Kubuntu I've tried on this laptop, since the earlier screen-size problem results in Kubuntu's panel being off-screen on the bottom.

[Peter Belew (peterbe)]

I was able to get the 2011-07-20 build of 10.04.3 to boot, using roughly the same tricks that I used for another system described in bug#707755, on a Compaq system.

For the live-CD boot line, I hit F6 and selected 'nomode' set. For the rest of the command line, I removed 'quiet' and 'splash' and added 'text' and 'vga=771'. 'text' brings the system up in a shell, so I am able to add an xorg.conf file, and 'vga=771' gives me a text display which fits on the screen (the default goes off the bottom of the screen).

Then I used scp to copy a basic xorg.conf file, specifying admissible screen modes, into /etc/X11. Having done that, I ran gdm to start the X display. It came up properly.

I've attached xorg.conf, Xorg.0.log, and Xorg.0.log.old in a tar file.

This bug was fixed in 11.04 and 11.10 releases - it should have been backported, as suggested in #166.

[Peter Belew (peterbe)]

I'm not sure that the 'nomode' boot line setting was necessary for this Openchrome display - I had used that (see previous comment) for a system with an nVidia display.

[Peter Belew (peterbe)]

At any rate, even if the repaired driver isn't added to the .3 or .4 release of 10.04 LTS, this provides a workaround to allow someone to install this LTS version in a system like this. It's a little beyond what a newbie can be expected to do, though!