ASUS F3S laptop nvidia GeForce 8400M G fails to start x

Asked by Gerard

Hi all

I have a very similar problem to https://answers.launchpad.net/ubuntu/+question/7303, except using ASUS F3S laptop nvidia GeForce 8400M G.

The reason I went the nvidia driver route was it is too new for 7.04, that is to be expected.

However, I have followed all the instruction regarding installing the NVIDIA-Linux-x86-100.14.09-pkg1.run on my Kubuntu to no avail.

I did the purge removes, restricted disables, source downloads, telinit 3, step by step.

Yet the only xorg.conf setting that works is for vesa.

Looking for any new ideas, one I did see in the similar question 7303 was

    sudo apt-get install nvidia-glx-new=1.0.9755+2.6.20.5-16.28

Just wondering if somwthing like that would work for 100.14.09.

Oh, and I have lodged a bug report with nvidia and ubuntu, both have been very helpful - thank you to the people I have met there.

Regards

Gerard

Question information

Language:
English Edit question
Status:
Solved
For:
Ubuntu Edit question
Assignee:
No assignee Edit question
Solved by:
Gerard
Solved:
Last query:
Last reply:
Revision history for this message
Sitsofe Wheeler (sitsofe) said :
#1

The only way you're going to get a card as new as a GeForce 8400M going with binary drivers is by using drivers _newer_ than the ones that ship with Ubuntu Feisty (and just so you know Ubuntu won't be packaging newer binary drivers until Gusty comes out in October). The list cards supported by the Ubuntu shipped drivers can be found as link on https://help.ubuntu.com/community/BinaryDriverHowto/Nvidia . Just to reiterate, the 9755 driver (http://us.download.nvidia.com/XFree86/Linux-x86/1.0-9755/README/appendix-a.html ) is not compatible with the GeForce 8400M. It looks like the beta 100.14.09 driver might be though.

Gerard:
Tell me very carefully what you have attempted to do so far and in what order, what you have currently installed along with the output of
dpkg -l nvidia\* | grep ii
cat /etc/default/linux-restricted-modules-common
sudo modprobe nvidia
dmesg | tail
lspci | grep -i nv
ls -l /usr/lib/libGL*

Do not make any changes to your system at this stage. It is very easy to end up with a system no one will help you fix the moment you step away from .deb packages because there is no longer one system in charge of checking for conflicts.

Be warned! Installing the binary drivers "by hand" can leave your system in a state which may not be upgradable to Gusty. Further, EVERY time a kernel update comes out you and you switch to running the new kernel you will HAVE to recompile the kernel module part of the driver yourself. Also due to the way the libraries in the NVIDIA pkg install themselves when an update to Mesa is released the GL library part of the package will have to be reinstalled.

Revision history for this message
Gerard (gbendotti) said :
#2

I will try yo be brief, yet complete.

As it stands right now, since our last conversation I have started from scratch.
I have created a dual boot so I could manage the bios, I now have Vista on a 20 Gb partition, Kubuntu on the remaining 100 Gb.

I used the Kubuntu alternative CD as the Live CD does not work - off topic.
I followed the instructions from here:
   http://doc.gwos.org/index.php/Latest_nvidia_feisty#METHOD_2
with minor differences. I did this.
</instructions>
    sudo apt-get install linux-headers-generic build-essential gcc xserver-xorg-dev pkg-config
    sudo apt-get --purge remove nvidia-glx nvidia-settings nvidia-glx-legacy
    sudo /etc/init.d/kdm stop
    <my difference>
            [Alt+F1] to get a terminal as I was now looking at the blinking cursor that did nothing other than show what I typed in
    </my difference>
    sudo cp /etc/X11/xorg.conf /etc/X11/xorg.conf_backup
    cd /home/gerard/Downloads
    sudo sh NVIDIA-Linux-x86-1.0-9755-pkg1.run -n -s --x-prefix=/usr/lib/xorg/ --kernel-source-path=/usr/src/linux-headers-`uname -r`
    sudo nvidia-xconfig --no-composite
    sudo nano -w /etc/default/linux-restricted-modules-common
    edited so DISABLED_MODULES="nv"
    [Ctrl X]
    [Y]
    sudo /etc/init.d/kdm restart
</instructions>
This produced the nvidia splash screen.

But when it went away, it was still using vesa.

So I rebooted

Same symptoms as before.

The Kubuntu startup screen displays as normal, a status bar that fills in with a light blue indicator, right through to the end.

What should happen:
It would go black for a millisecond or two, then display the log in screen.

What does happen
It would go black for a millisecond or two, then displays the Kubuntu startup screen, with the status bar empty, as if starting from scratch.
This stays on the screen for maybe five seconds without altering.
It then goes away, leaving a black screen, with a white blinking cursor, as if in console mode. But one where all it does is display what you type in, similar to a text editor, no actions.

[Alt + F1] brings up a working login console prompt.
I logged in, ran sudo nvidia-bug-report.sh, so the machine is operational.

I have experienced the pain of hand building then upgrading, my move from Dapper to Edgy was... a learning experience. I am prepared for the problems associated with kernel updates, I have been there before with USB TV under Edgy.
I am prepared at this stage to keep on reinstalling Kubuntu in a effort to keep things as vanilla as possible, I hope to be able to pass something useful on to others, and to get a working version for myself of course.
In the spirit of this, if you wanted this to be test machine for Gutsy, I would be more than happy to assist, because that is where I would like to see this going.

Right, back to your request:
I ran these using vesa so I had a GUI.

:~$dpkg -l nvidia\* | grep ii
ii nvidia-kernel-common 20051028+1ubuntu7 NVIDIA binary kernel module common files

:~$cat /etc/default/linux-restricted-modules-common
# This file is sourced from the linux-restricted-modules-common init
# script and is used to disable the link-on-boot feature, one module
# at a time. This can be useful if you want to use hand-compiled
# versions of one or more modules, but keep linux-restricted-modules
# installed on your system, or just to disable modules you don't use
# and speed up your boot process by a second or two.
#
# Use a space-separated list of modules you wish to not have linked
# on boot. The following example shows a (condensed) list of all
# modules shipped in the linux-restricted-modules packages:
#
# DISABLED_MODULES="ath_hal fc fglrx ltm nv"
#
# Note that disabling "fc" disables all fcdsl drivers, "ltm" disables
# ltmodem and ltserial, and "nv" disables both the nvidia drivers.
# You can also name each module individually, if you prefer a subset.

DISABLED_MODULES="nv"

:~$sudo modprobe nvidia
:~$Not loading nvidia module; not used in /etc/X11/xorg.conf

<note>
 Seeing this I then
  :~$sudo nano -w /etc/X11/xorg.conf
 changed
  driver = "nvidia"
 :~$sudo modprobe nvidia
 :~$
        Being ever hopeful I left it like that, when I rebooted, no change.
</note>

:~$ dmesg | tail
[ 37.116000] Bluetooth: HCI socket layer initialized
[ 37.144000] Bluetooth: L2CAP ver 2.8
[ 37.144000] Bluetooth: L2CAP socket layer initialized
[ 37.352000] Bluetooth: RFCOMM socket layer initialized
[ 37.352000] Bluetooth: RFCOMM TTY layer initialized
[ 37.352000] Bluetooth: RFCOMM ver 1.8
[ 474.296000] nvidia: module license 'NVIDIA' taints kernel.
[ 474.556000] ACPI: PCI Interrupt 0000:01:00.0[A] -> GSI 16 (level, low) -> IRQ 16
[ 474.556000] PCI: Setting latency timer of device 0000:01:00.0 to 64
[ 474.556000] NVRM: loading NVIDIA UNIX x86 Kernel Module 100.14.09 Sat May 26 00:47:07 PDT 2007

:~$ lspci | grep -i nv
01:00.0 VGA compatible controller: nVidia Corporation Unknown device 0428 (rev a1)

:~$ ls -l /usr/lib/libGL*
lrwxrwxrwx 1 root root 22 2007-06-14 06:53 /usr/lib/libGLcore.so.1 -> libGLcore.so.100.14.09
-rwxr-xr-x 1 root root 10080488 2007-06-14 06:53 /usr/lib/libGLcore.so.100.14.09
lrwxrwxrwx 1 root root 16 2007-06-13 16:11 /usr/lib/libGLEW.so.1.3 -> libGLEW.so.1.3.4
-rw-r--r-- 1 root root 199848 2007-03-05 14:16 /usr/lib/libGLEW.so.1.3.4
-rw-r--r-- 1 root root 659 2007-06-14 06:53 /usr/lib/libGL.la
lrwxrwxrwx 1 root root 10 2007-06-14 06:53 /usr/lib/libGL.so -> libGL.so.1
lrwxrwxrwx 1 root root 18 2007-06-14 06:53 /usr/lib/libGL.so.1 -> libGL.so.100.14.09
-rwxr-xr-x 1 root root 607632 2007-06-14 06:53 /usr/lib/libGL.so.100.14.09
lrwxrwxrwx 1 root root 20 2007-06-12 23:38 /usr/lib/libGLU.so.1 -> libGLU.so.1.3.060502
-rw-r--r-- 1 root root 521000 2007-03-31 07:07 /usr/lib/libGLU.so.1.3.060502

Revision history for this message
Shashank Chintalagiri (shashank-chintalagiri) said :
#3

I use the same drivers for my GeForce 8500 GT. The driver comes with its share of issues... so far I have had to reinstall the drivers (simply by running the Nvidia install file (the .run file) as root) each time i reboot. If you've installed the drivers using the script, as you have seem to have done seeing the line -
sudo sh NVIDIA-Linux-x86-1.0-9755-pkg1.run -n -s --x-prefix=/usr/lib/xorg/ --kernel-source-path=/usr/src/linux-headers-`uname -r`

then at the end of the install process after compiling and installing your kernel module, it would have asked you if you wanted it to modify your xorg.conf. In case you said yes there, it would have automatically set your xorg.conf to use to nvidia driver. I presume the nvidia-xconfig command you run does somewaht the same function. You could try letting the script modify the xorg.conf on its own and see if it does a better job for some reason.

Since the last line of your dmesg is the one talking about loading the Nvidia module, then you dont seem to have issues with the module itself.

The nVidia corporation unknown device in the lspci seems to happen with all the geForce8 GPUs.

Revision history for this message
Sitsofe Wheeler (sitsofe) said :
#4

Gerard:
In /etc/default/linux-restricted-modules-common can you change
DISABLED_MODULES="nv"
to
DISABLED_MODULES="nv nvidia_new"
. Reboot. Can you then add the output of doing
sudo modprobe nvidia
dmesg | tail

Edit xorg.conf to use the nvidia driver and try to start X by using
/etc/init.d/kdm restart
When X crashes copy the file
/var/log/Xorg.0.log
to your home directory. Change xorg.conf back to vesa and attach (using the launchpad web interface) the copy of the /var/log/Xorg.0.log file. Please also attach the file that nvidia-bug-report.sh made.

Revision history for this message
Gerard (gbendotti) said :
#5

With /etc/default/linux-restricted-modules-common DISABLED_MODULES="nv nvidia"
and /etc/X11/xorg.conf Driver "vesa"
reboot

:~$ sudo modprobe nvidia
Password:
Not loading nvidia module; not used in /etc/X11/xorg.conf

:~$ dmesg | tail
[ 28.548000] Bluetooth: Core ver 2.11
[ 28.548000] NET: Registered protocol family 31
[ 28.548000] Bluetooth: HCI device and connection manager initialized
[ 28.548000] Bluetooth: HCI socket layer initialized
[ 28.580000] Bluetooth: L2CAP ver 2.8
[ 28.580000] Bluetooth: L2CAP socket layer initialized
[ 28.748000] Bluetooth: RFCOMM socket layer initialized
[ 28.748000] Bluetooth: RFCOMM TTY layer initialized
[ 28.748000] Bluetooth: RFCOMM ver 1.8
[ 31.248000] eth0: no IPv6 routers present

:~$ sudo /etc/init.d/kdm restart

I do not see a way of attaching files here, I am a new user of Answers so I could be missing something.

I have placed the files here:

http://members.iinet.com.au/~gbendotti/logs/

But this is unsatisfactory as this link will be dead at some point, creating a 404 for some poor lost soul.

If there is a way to attach, please let me know, I will do that.

Cheers

Gerard

Revision history for this message
Gerard (gbendotti) said :
#6

On re-reading this, I must confirm it was in fact /etc/default/linux-restricted-modules-common DISABLED_MODULES="nv nvidia_new"

NOT what I typed 7 seconds ago.

Revision history for this message
Gerard (gbendotti) said :
#7

Gees I really should read my posts more closely.
I need to also confirm I am trying to use NVIDIA-Linux-x86-100.14.09-pkg1, that was the other change - important one - granted, I made from the METHOD 2 instructions.

Revision history for this message
Best Gerard (gbendotti) said :
#8

OK - Played with it some more: now it works, not sure what I did that was different.

This time I did this, even though they did not exist.
rm -f /etc/init.d/nvidia-glx
rm -f /etc/init.d/nvidia-kernel
rm -f /lib/linux-restricted-modules/.nvidia_new_installed

I also added automake, autoconf.

I made sure /etc/default/linux-restricted-modules-common DISABLED_MODULES="nv nvidia_new" was done before I rebooted.

And for good measure I added /etc/default/linux-restricted-modules-common DISABLED_MODULES="nv nvidia_new nvidia" inspired by my earlier error in this forum.

I followed the normal script deal, telinit 3 etc through to the end.

This time I sudo modprobe nvidia before restarting.

I did not even notice a change in screen quality as it was still using 1024x768, I thought it was still using vesa & I had missed a step.

The only hint was the big nvidia splash screen on startup.

I went to the display panel (sorry not at home, using windows box, forgotten correct terminology) and under hardware is still said vesa, checked xorg.conf, definitely says nvidia...

I then noticed the nvidia settings icon in the K menu, click it, and lo and behold, all the setting were there.

I changed resolution to 1440x900, beautiful.

I sit here somewhat mystified.

Shutdown, bootup. restarted. still there. Solved?

I can not honestly click the button below [I Solved the Problem on My Own] as it wasn't down to me.

Thank you Lonni J Friedman from nvidia and a big thank you to Sitsofe Wheeler both showed more patience than I had any right to expect.

Ciao

Gerard