gdm fails to start after hardy upgrade

Thank you for taking the time to report this bug and help make Ubuntu better. However, after having a quick look through your X server log files, I can't see anything drastically wrong with your X server. Could you please try to start GDM with your default configuration (when it fails), and then add the contents of /var/log/gdm in to a tar.gz archive and then attach it to this bug report.

Thanks

Thanks Chris,

I think I've found the problem.

Looks like GDM is trying to run X with the nvidia driver and failing:

(EE) Failed to initialize GLX extension (Compatible NVIDIA X driver not
found)

This is when using the xorg failsafe configuration.

So I tried using my nvidia xorg.

For each X guit/restart I see these messages in /var/log/messages:

Aug 19 09:00:38 aporia kernel: [167694.560423] agpgart: Found an AGP 2.0
compliant device at 0000:00:00.0.
Aug 19 09:00:38 aporia kernel: [167694.560458] agpgart: Putting AGP V2
device at 0000:00:00.0 into 2x mode
Aug 19 09:00:38 aporia kernel: [167694.560508] agpgart: Putting AGP V2
device at 0000:01:00.0 into 2x mode

I called the gdm and xorg logs .nvidia to differentiate them from the
failsafe tests.

The pause after running /etc/init.d/gdm stop is exceptionally long,
approximately 15 seconds.

To recap:

Seems when I run gdm with an xorg using the vesa or nv driver, GDM tries
to use nvidia anyhow and throws an error. Why startx uses the driver
specified in xorg.conf and gdm seems to always try and load the nvidia
driver is beyond me.

Even if I use the the nvidia driver in the xorg, GDM does not throw an
error (just complains about fonts), but does not start GDM either.

Seems to me a reasonable next step is to force GDM to use nv and not
load nvidia-glx and see if that works.

I added the nvidia driver using the "restricted drivers" GUI in startx
(as root). How can I remove the nvidia driver (withoutx) to see if I can
get nv to work with GDM?

Thanks for your efforts.

Chris Coulson wrote:
> Thank you for taking the time to report this bug and help make Ubuntu
> better. However, after having a quick look through your X server log
> files, I can't see anything drastically wrong with your X server. Could
> you please try to start GDM with your default configuration (when it
> fails), and then add the contents of /var/log/gdm in to a tar.gz archive
> and then attach it to this bug report.
>
> Thanks
>
> ** Changed in: gdm (Ubuntu)
> Status: New => Incomplete
>

If the NVIDIA driver is not explicitly enabled in your xorg.conf, then Xorg will not load it. That error you see is because there is no compatible NVIDIA driver found (because it isn't loaded). Your Xorg log file from the attempt when you ran with the NVIDIA driver shows that the driver loading ok but can't read the EDID for your display.

Your logs also show that the ACPI daemon on your machine isn't running, although I don't know what effect that would have or whether it is contributing to your problem:

(WW) Open ACPI failed (/var/run/acpid.socket) (No such file or directory)

Could you please try to run with the default xorg.conf. To recreate this, run the following in a terminal:

sudo dpkg-reconfigure xserver-xorg

This will create the default stripped down xorg.conf without all the modelines. Could you also attach your /var/log/syslog as well please?

Thanks

Thanks Chris,

Here are the logs from /etc/init.d/gdm start (after stop) using the
reconfigured xorg file.

I've run this machine for most of its life with ACPI disabled, as it
seems to cause crashes with nvidia drivers.

If the ACPI was the problem, then startx should not work either.

Maybe something stands out in these logs for you, as I'm stumped.

Thanks,
B. Bogart

Chris Coulson wrote:
> If the NVIDIA driver is not explicitly enabled in your xorg.conf, then
> Xorg will not load it. That error you see is because there is no
> compatible NVIDIA driver found (because it isn't loaded). Your Xorg log
> file from the attempt when you ran with the NVIDIA driver shows that the
> driver loading ok but can't read the EDID for your display.
>
> Your logs also show that the ACPI daemon on your machine isn't running,
> although I don't know what effect that would have or whether it is
> contributing to your problem:
>
> (WW) Open ACPI failed (/var/run/acpid.socket) (No such file or
> directory)
>
> Could you please try to run with the default xorg.conf. To recreate
> this, run the following in a terminal:
>
> sudo dpkg-reconfigure xserver-xorg
>
> This will create the default stripped down xorg.conf without all the
> modelines. Could you also attach your /var/log/syslog as well please?
>
> Thanks
>

Any other suggestions for debugging? I believe I've supplied enough
information for this to be no longer marked as incomplete.

Correct me if I'm wrong.

B. Bogart

b wrote:
> Thanks Chris,
>
> Here are the logs from /etc/init.d/gdm start (after stop) using the
> reconfigured xorg file.
>
> I've run this machine for most of its life with ACPI disabled, as it
> seems to cause crashes with nvidia drivers.
>
> If the ACPI was the problem, then startx should not work either.
>
> Maybe something stands out in these logs for you, as I'm stumped.
>
> Thanks,
> B. Bogart
>
>
> Chris Coulson wrote:
>> If the NVIDIA driver is not explicitly enabled in your xorg.conf, then
>> Xorg will not load it. That error you see is because there is no
>> compatible NVIDIA driver found (because it isn't loaded). Your Xorg log
>> file from the attempt when you ran with the NVIDIA driver shows that the
>> driver loading ok but can't read the EDID for your display.
>>
>> Your logs also show that the ACPI daemon on your machine isn't running,
>> although I don't know what effect that would have or whether it is
>> contributing to your problem:
>>
>> (WW) Open ACPI failed (/var/run/acpid.socket) (No such file or
>> directory)
>>
>> Could you please try to run with the default xorg.conf. To recreate
>> this, run the following in a terminal:
>>
>> sudo dpkg-reconfigure xserver-xorg
>>
>> This will create the default stripped down xorg.conf without all the
>> modelines. Could you also attach your /var/log/syslog as well please?
>>
>> Thanks
>>
>
>
> ** Attachment added: "syslog"
> http://launchpadlibrarian.net/16939706/syslog
>
> ** Attachment added: "Xorg.0.log"
> http://launchpadlibrarian.net/16939708/Xorg.0.log
>
> ** Attachment added: ":0.log"
> http://launchpadlibrarian.net/16939710/%3A0.log
>

I realized that once GDM is started that I can log into the machine via
XDMCP. Things seem to work the same as when using startx.

The missing menu and crashing applets made me wonder so I ran:

dpkg-reconfigure `dpkg -l "*gnome*" | grep ^ii | cut -d " " -f 3`

to reconfigure all packages with gnome in them. dpkg-reconfigure
gnome-desktop-environment did not reconfigure the sub packages.

Even after doing so the GDM remote login does not change, many applet
crashes and still no menu.

See attached screenshot where the effected machine is in the Xnest. All
my custom application icons are missing from the panel, all menu ubuntu
menu buttons are also missing.

I mention these details in case they are a hint to what is going on.

Perhaps these gnome problems are another bug, but something is certainly
 messed after the upgrade. Is there a bug channel specifically for
upgrading issues? I suppose just the forums.

Thanks,
b.

Brian Murray wrote:
> ** Tags added: valid-xorg-conf
>

The machine has locked up solid twice in the last week.

ssh connections freeze and timeout, and the console keyboard does not
respond (no VT switching or any other feedback from keyboard input).

The logs show nothing, no errors. I checked the disk with smart, and it
reports no errors.

So I went back to the hardy up to date generic kernel (was running the
rt kernel before)

I removed the acpi=off noapm options from the kernel startup.

X still does not start via GDM, and I was running GDM at the time of
both failures.

I'll turn off GDM for a few days to see if the freeze still happens.

just keeping you posted.

B.

the issue doesn't really seem to be a gdm bug

I just updated the machine today.

Still having the same issue, GDM is unable to start X. Syslog says:

Gtk-WARNING: cannot open display: :0
WARNING: failsafe dialog failed (inhibitions: 0 0)

for each of the 6 gdm attempts.

For each one I see the cursor flash and then X dies.

happens with all xorg.conf configurations.

startx appears to work fine. (for some reason now startx runs xfce
rather than gnome)

attached is the whole syslog for reference.

Any suggestions for further debugging?

Thanks,
B. Bogart

Sebastien Bacher wrote:
> ** Changed in: xorg-server (Ubuntu)
> Sourcepackagename: gdm => xorg-server
>

So on a hunch I tried installing ol XDM.

xdm does appear to work, it starts X fine, provides a login window, and allows a login.

I seem only to be able to log into xfce and not gnome.

So something is different about what GDM attempts to do and what XDM does.

Thanks,
B. Bogart

I've installed the recent updates.

No change in GDM failure.

I've now configured XDM for gnome, and gnome seems to be working fine.

What is the best way to reset GDM configuration settings? I tried sudo
dpkg-reconfigure gdm, but gdm will still not start X. (XDM works fine).

.b.

b wrote:
> So on a hunch I tried installing ol XDM.
>
> xdm does appear to work, it starts X fine, provides a login window, and
> allows a login.
>
> I seem only to be able to log into xfce and not gnome.
>
> So something is different about what GDM attempts to do and what XDM
> does.
>
> Thanks,
> B. Bogart
>

[This is an automated message]

Hi ben-ekran,

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

Hi Bryce,

I gave gdm another try today, still the same behaviour. This is how I test:

sudo /etc/init.d/xdm stop
sudo dpkg-reconfigure gdm # Select gdm as default display manager
sudo /etc/init.d/gdm start # gdm fails to start X

Thanks,
B.

Bryce Harrington wrote:
> [This is an automated message]
>
> Hi ben-ekran,
>
>
>
> Please attach the output of `lspci -vvnn` too.
>
>
> ** Changed in: xorg-server (Ubuntu)
> Status: New => Incomplete
>

I upgraded to intrepid, and the same GDM issue exists.

Still happily using xdm for now.

I've attached the logs for this version.

.b.

b wrote:
> Hi Bryce,
>
> I gave gdm another try today, still the same behaviour. This is how I
> test:
>
> sudo /etc/init.d/xdm stop
> sudo dpkg-reconfigure gdm # Select gdm as default display manager
> sudo /etc/init.d/gdm start # gdm fails to start X
>
> Thanks,
> B.
>
> Bryce Harrington wrote:
>> [This is an automated message]
>>
>> Hi ben-ekran,
>>
>>
>>
>> Please attach the output of `lspci -vvnn` too.
>>
>>
>> ** Changed in: xorg-server (Ubuntu)
>> Status: New => Incomplete
>>
>
>
> ** Attachment added: "lspci.txt"
> http://launchpadlibrarian.net/20578826/lspci.txt
>
> ** Changed in: xorg-server (Ubuntu)
> Status: Incomplete => New
>

Hmm, looking at all the various logs, there still doesn't seem to be any error message indicating that X is doing something wrong. So not sure that this can be troubleshot from the X angle... I know this was rejected as not being a gdm issue, but given that xdm is working, it makes me wonder if there is something GNOME-ish broken.

gdm is configured with the /etc/gdm/gdm.conf file. Can you experiment with removing or changing some of the parameters in that file, particularly as relates to anything mentioning "failsafe"?

It would also be helpful if you could try booting a Jaunty LiveCD (ISOs available at http://cdimages.ubuntu.com/releases/jaunty/) and seeing if you can reproduce the issue there.

Attached is my gdm.conf

How can I confirm (overwrite) this file to make sure its the proper
fresh installed config? I don't recall making any changes to it.

Thanks,
.b.

Bryce Harrington wrote:
> Hmm, looking at all the various logs, there still doesn't seem to be any
> error message indicating that X is doing something wrong. So not sure
> that this can be troubleshot from the X angle... I know this was
> rejected as not being a gdm issue, but given that xdm is working, it
> makes me wonder if there is something GNOME-ish broken.
>
> gdm is configured with the /etc/gdm/gdm.conf file. Can you experiment
> with removing or changing some of the parameters in that file,
> particularly as relates to anything mentioning "failsafe"?
>
> It would also be helpful if you could try booting a Jaunty LiveCD (ISOs
> available at http://cdimages.ubuntu.com/releases/jaunty/) and seeing if
> you can reproduce the issue there.
>
> ** Changed in: xorg-server (Ubuntu)
> Status: Confirmed => Incomplete
>

Well, even simpler, please do this:

1. switch to a tty (ctrl-alt-f3) and login
2. sudo /etc/init.d/gdm stop
3. pkill X

At this point, you should have no gdm or X processes running. Doublecheck this by running 'ps aux | grep X' and 'ps aux | grep gdm'.

Next, start up X by hand:

4. startx

Possibly this will work fine. Or maybe it will exit back to the command line with or without an error message. Or maybe something else.

If you do see it come back to the command line with some sort of error, run this to capture it:

5. startx > /var/log/startx.log 2>&1

And then attach startx.log to this bug report.

If it does something else, please describe what you experience in as much detail as you can.

Hi Bryce,

This is what I did for the intitial report (on gutsy):
https://bugs.launchpad.net/ubuntu/+source/xorg-server/+bug/259156/comments/4

startx works fine, XDM works fine, GDM will not start X.

I don't see any significant differences between those results and these. Attached.

I agree that it must be a GDM config error somewhere, but where? I'll try a liveCD when I get the chance.

Hmm, well the startx log has this error:

(EE) Logitech USB Receiver: Read error: No such device

Not sure... probably next step is for this bug to go upstream.

"Logitech USB Receiver" is my wireless mouse.

It just means I was switched away (on the KVM) from this machine when X
started. I also tried it without switching away, the same thing happens.

I have intrepid on my powerbook now also, but it was working fine with
GDM. I think its likely a configuration issue somehow.

This machine has not had a fresh install in about 7 years.

GDM was working in all other releases before intrepid.

Could also be that these old machines are not tested often. Its a Duron
800 KT7.

.b.

Bryce Harrington wrote:
> Hmm, well the startx log has this error:
>
> (EE) Logitech USB Receiver: Read error: No such device
>
> Not sure... probably next step is for this bug to go upstream.
>
> ** Changed in: xorg-server (Ubuntu)
> Importance: Medium => High
>

Resolved by Jaunty upgrade.

I just re-enabled gdm with dpkg-reconfigure after upgrade and gdm
suddenly starts. I made no config changes.

Thanks.
.b.

Bryce Harrington wrote:
> ** Changed in: xorg-server (Ubuntu)
> Status: Confirmed => Triaged
>

[This is an automatic notification.]

Hi b,

This bug was reported against an earlier version of Ubuntu, can you
test if it still occurs on Lucid?

Please note we also provide technical support for older versions of
Ubuntu, but not in the bug tracker. Instead, to raise the issue through
normal support channels, please see:

    http://www.ubuntu.com/support

If you are the original reporter and can still reproduce the issue on
Lucid, please run the following command to refresh the report:

  apport-collect 259156

If you are not the original reporter, please file a new bug report, so
we can work with you as the original reporter instead (you can reference
bug 259156 in your report if you think it may be related):

  ubuntu-bug xorg

If by chance you can no longer reproduce the issue on Lucid or if you
feel it is no longer relevant, please mark the bug report 'Fix Released'
or 'Invalid' as appropriate, at the following URL:

  https://bugs.launchpad.net/ubuntu/+bug/259156

This bug has been resolved since Jaunty.

Working fine in Karmic and now Lucid.

.b.

On 10-05-04 01:24 PM, Bryce Harrington wrote:
> [This is an automatic notification.]
>
> Hi b,
>
> This bug was reported against an earlier version of Ubuntu, can you
> test if it still occurs on Lucid?
>
> Please note we also provide technical support for older versions of
> Ubuntu, but not in the bug tracker. Instead, to raise the issue through
> normal support channels, please see:
>
> http://www.ubuntu.com/support
>
>
> If you are the original reporter and can still reproduce the issue on
> Lucid, please run the following command to refresh the report:
>
> apport-collect 259156
>
> If you are not the original reporter, please file a new bug report, so
> we can work with you as the original reporter instead (you can reference
> bug 259156 in your report if you think it may be related):
>
> ubuntu-bug xorg
>
> If by chance you can no longer reproduce the issue on Lucid or if you
> feel it is no longer relevant, please mark the bug report 'Fix Released'
> or 'Invalid' as appropriate, at the following URL:
>
> https://bugs.launchpad.net/ubuntu/+bug/259156
>
>
> ** Changed in: xorg-server (Ubuntu)
> Status: Triaged => Incomplete
>
> ** Tags added: needs-retested-on-lucid-by-june
>