Posts: 146
Eperbab
Joined: 10 Dec 2012
#1
I have installed 686-pae kernel issuing 'sudo apt-get install linux-image-686 libc6-i686' and 'sudo update-grub'. After reboot to 686-pae, GUI doesn't start. After logging in manually, and running 'startx' the default desktop of slim is started (in my case, xfce). After running 'sudo lightdm' my default desktop of lightdm (Mate) is started. I would like to try out the 686-pae kernel without manually starting X.
Help, please.

BTW, the 686-pae kernel is 3.2.0-4-686-pae, while the 486 kernel is 3.5.0-antix.2-486-smp. So the 686-pae seems to be outdated. Is it normal?

* * *

Code: Select all

# dpkg-reconfigure slim
insserv: warning: current start runlevel(s) (2) of script `slim' overwrites defaults (2 5).
insserv: warning: current stop runlevel(s) (0 1 3 4 5 6) of script `slim' overwrites defaults (0 1 6).

Code: Select all

# dpkg-reconfigure lightdm
[ ok ] Reloading system message bus config...done.
***

Update:
- 686-pae debian kernel removed
- 3.7.4-antix.1-486-smp added
- 3.7.1-antix.2-486-smp added
- 3.7.0-4.dmz.1-liquorix-686 added
(all kernels with headers)

* attached menu.lst updated

Now kernel 3.5.0-antix.2-486-smp works propely, all other kernel fails. I see different behaviour after

Code: Select all

 
INIT: Entering runlevel 5
..
acpid: waiting for events: event logging is off
Here I get an 'acpid: client 2059[0:0] has disconnected' message on tty7(ctrl + alt + F7), and a 'no suitable virtualbox kernel moduls found' error on tty1. Also an error about 'cpufrequtils' with liquorix kernel. There is NO message about dbus start on tty1.

Any idea how to fix it? Wich additional informations would be helpful?
Last edited by Eperbab on 13 Feb 2013, 17:12, edited 3 times in total.
Posts: 146
Eperbab
Joined: 10 Dec 2012
#2
xorg.1.log from failed and succesful X starts plus latest xorg.conf attached. It seems to be a problem with radeon driver, but..

1.
After X failed, I have logged to tty1, moved to init 1, logged in as root, logged out from maintenance mode, logged in as normal user, run 'sudo lightdm', and X started propely with same user, config and driver. Also tried init 2 and init 3, but only init 1 helps.
After logging out from maintenance mode, 'startx' and 'sudo lightdm' both are working propely. Before going to init 1 both fails.

2. The original 3.5.0 kernel works out of the box. Same user, same config, same driver.

Help, please..

Critical part of the log:

Code: Select all

[   480.540] (II) [KMS] Kernel modesetting enabled.
[   480.540] (**) RADEON(0): Depth 16, (--) framebuffer bpp 16
[   480.540] (II) RADEON(0): Pixel depth = 16 bits stored in 2 bytes (16 bpp pixmaps)
[   480.540] (==) RADEON(0): Default visual is TrueColor
[   480.541] (==) RADEON(0): RGB weight 565
[   480.541] (II) RADEON(0): Using 6 bits per RGB (8 bit DAC)
[   480.541] (--) RADEON(0): Chipset:"ATI Radeon 9200SE 5964 (AGP)" (ChipID = 0x5964)
[   480.541] (II) RADEON(0): AGP card detected
[   480.541] drmOpenDevice: node name is /dev/dri/card0
[   480.541] drmOpenDevice: open result is 9, (OK)
[   480.541] drmOpenByBusid: Searching for BusID pci:0000:01:00.0
[   480.541] drmOpenDevice: node name is /dev/dri/card0
[   480.541] drmOpenDevice: open result is 9, (OK)
[   480.541] drmOpenByBusid: drmOpenMinor returns 9
[   480.541] drmOpenByBusid: Interface 1.4 failed, trying 1.1
[   480.541] drmOpenByBusid: drmGetBusid reports pci:0000:01:00.0
[   480.541] (EE) RADEON(0): [drm] failed to set drm interface version.
[   480.541] (EE) RADEON(0): Kernel modesetting setup failed
[   480.541] (II) UnloadModule:"radeon"
[   480.541] (EE) Screen(s) found, but none have a usable configuration.
[   480.541] 
Fatal server error:
[   480.541] no screens found
[   480.541] 
Posts: 1,308
BitJam
Joined: 31 Aug 2009
#3
Just a couple of shots in the dark.

You may need to install a different radeon driver for the newer kernels.

You could try using the"nomodeset" boot parameter.
anticapitalista
Posts: 5,955
Site Admin
Joined: 11 Sep 2007
#4
Did you use smxi/sgfxi to install the driver?

As BJ says plus old radeon is no longer supported in newer kernels (The issue is according to h2 of the smxi scripts is with amd not the kernel devs)
Posts: 146
Eperbab
Joined: 10 Dec 2012
#5
I tried .. and I'm tired..

@ Anti: I have never used smxi/sgfxi. Only inxi. Driver installed with apt-get install.

- boot options: init 3, init 5 without boot params, nomodeset, nouveau.modeset=0, nouveau.modeset=1, vga=771 .. no change
- removed virtualbox: now X fails without error message on tty1 about vbox
- reinstalled virtualbox: only tries to build for 3.5.0 and 3.7.4 486 kernels, doesn't see others, build failed on 3.7.4
- dist upgrade to testing repo: no change
- update lightdm to experimental
- apt-get install -t experimental firmware-linux-nonfree libgl1-mesa-dri xserver-xorg-video-radeon xserver-xorg-video-ati

I'm playing games on wine, so I need 3D acceleration with my Radeon 9200 SE card. Wich driver-kernel combinations may work?
At the moment, the original 3.5.0-antix.2-486-smp works with the packages above, everything else failed. So I set it default.
Posts: 1,308
BitJam
Joined: 31 Aug 2009
#6
Eperbab wrote:@ Anti: I have never used smxi/sgfxi. Only inxi. Driver installed with apt-get install.
You really should use smxi or sgfxi to re-install the video driver. The purpose of these scripts is to help people easily solve the sort of problems you are encountering. If"apt-get install" worked in these cases then there would be no need for these scripts.
Posts: 146
Eperbab
Joined: 10 Dec 2012
#7
Yesterday it worked after using smxi, but today it doesn't work again. I have checked what smxi did, and repeated the steps without smxi on another kernel. Today all new kernels failed again, including the two kernels configured with smxi. Here is what I did yesterday:

Code: Select all

apt-get update
apt-get install firmware-linux-nonfree libgl1-mesa-dri xserver-xorg-video-radeon xserver-xorg-video-ati 
(re)boot with the kernel that has to be fixed:

Code: Select all

* Ctrl + Alt + F1 to tty1, login &
$ su
$ killall Xorg
$ apt-get install --reinstall xserver-xorg-core libgl1-mesa-glx xserver-xorg-video-ati
Now we can start the display manager. (lightdm, or slim, kdm, xdm, gdm whatsoever..)


Today I found out, that X starts before the Radeon driver is fully loaded. I see a blank screen for a few second, before X starts on working kernels. There's no blanking with failing kernels. After X failed, now I did the following:

Code: Select all

* switch to tty1, login
$ su
$ killall Xorg
$ lightdm
X starts.

It seems, that switching to init 1 killed Xorg. Smxi did the same, and before reinstalling the drivers I also typed 'killall Xorg'.
So how I can force Xorg to wait until the Radeon driver is loaded?
Posts: 146
Eperbab
Joined: 10 Dec 2012
#8
After reinstalling radeon with and without smxi for 2 weeks, I lost my patience and started to make random changes to the system, including
- modification of etc/modprobe.d/radeon-kms.conf (options radeon modeset=0,1, commented out)
- sudo apt-get install xvba-va-driver libva-glx1 libva-egl1 vainfo fglrx-driver libgles2-mesa driconf
- etc/modules-load.d/modules.conf -> adding/removing radeon
- apt-get install --reinstall firmware-linux-nonfree libgl1-mesa-dri xserver-xorg-video-radeon xserver-xorg-video-ati within and outside of X
- apt-get install --reinstall xserver-xorg-core libgl1-mesa-glx xserver-xorg-video-ati
- dpkg-reconfigure xserver-xorg
- modprobe radeon
- smxi -kwidt
- Xorg -configure
...

I lost OpenGL support for 2 days, but now all kernels are up and running with openGL. Current settings:
- radeon-kms.conf: options radeon modeset=1 #1 (0 breaks openGL support, 1 (original value) breaked X, 1#1 gives error,but .. __{{emoticon}}__
- Xorg.conf: radeon & glx modules are loaded
- menu.lst: #defoptions=vga=791 quiet nosplash nouveau.modeset=0

Managed to break the 1200 FPS barrier with glxgears at 10x10 pixel resolution, running X on 3.7.5.antix.1-486-smp kernel.
I will post the remaining problem in other thread.