ubuntu graphic card error
Hello,
when booting with ubuntu 12.04, I sometimes get a huge error that prevents me from getting to the log in screen. The error window I get is this one (though the image isn't mine, I'm not running a VM: http://i.imgur.com/QaP9d.png). I think this is due to my graphic card (nvidia 680 gtx). I have the recommended (proprietary) driver. The log files can be found here http://speedy.sh/nvNuB/failsafeX-backup.tar.gz. Can you help me figure out what's wrong please ? |
Hi, the open source nvidia module is loaded early in the start process.
You may have to blacklist it and edit (/etc/X11/xorg.conf)then reboot before you can get it to not load before X /etc/modprobe.d/blacklist-mga.conf Code:
blacklist nouveau |
Thanks for your answer.
I'm sorry I really don't understand what you're talking about, I'm new to linux and I have no idea what you're talking about. Could you give further details, in precise steps, as to what I am supposed to attempt, and the possible associated risks ? Thank you. |
What GlennsPref is trying to explain is that your system is for whatever reason configured to use two drivers for the videocard at the same time (the proprietary driver and the free driver), which, as you can see, will lead to problems. The solution is to prevent the free driver from loading. To do that create a new file blacklist-nouveau.conf in the directory /etc/modprobe.d (or use an already existing blacklist file, as suggested by GlennsPref) and add the line
Code:
blacklist nouveau Code:
sudo gedit /etc/modprobe.d/blacklist-nouveau.conf Code:
sudo nvidia-xconfig |
Ok thanks for this explanation, it was easier to understand.
So I did just that, and I added the line blacklist nouveau to the file /etc/modprobe.d/blacklist-nouveau.conf, which I had to create. I then wrote 'sudo nvidia-xconfig'. It said that no previous configuration was found, and thus a new one had to be created. Then I rebooted. But the problem is still there. After rebooting once again, I was able to go straight to the login screen. But this was already the case before the above-mentioned manipulation, my problem was that *only sometimes* did I have problems booting. Given your explanations, this randomness could be due to the fact that one driver is picked up over the other in some cases, making it impossible to boot. So what else can I do from here ? Thanks |
Hi, It's seems strange,
but I generally have to reboot twice too, and when I get the login screen I find I need to once again confirm that the nouveau driver is not loaded and that xorg.conf has not been reconfigured by an auto script someplace (dri/drm/dkms). I think you'll find this the last hurdle. run Code:
sudo nvidia-xconfig or edit the file manually, vi /etc/X11/xorg.conf find the listing for nouveau and edit it to say nv or nvidia as you have been trying. Now startx, you should be off the loop. p.s. thankyou to Tobi for explaining what I neglected. |
so... sorry for the delay, I don't restart the computer often, specially because I'm afraid it will stop working one day due to this issue...
in any case, the problem is still there. Here are the new logs: http://speedy.sh/HH8YK/failsafeX-bac...0417214340.tar |
hi, can't extract archive.
Quote:
|
The problem in this case is not that the free driver is loaded, as it seemed to be the problem before, but that the proprietary driver module could not be found. Do you have made any changes to the system before this happened?
|
GlennsPref: I think tobiSGD was able to do so, so please try again.
tobiSGD: No, not that I'm aware of. I recall having this problem on a fresh ubuntu installation. |
That is weird, sometimes the computer is finding the nvidia driver, sometimes it is not. Honestly, I have no clue what could cause such a behavior.
|
I tried to dl and open with ark 3 separate times, says unrecognised ....
Too much trouble to fix blindfolded. I have most of the standard extractors, tar gzip bzip2. wanna post the info in code tags instead? |
do you have "nokmsboot" on your kernel boot cli?
|
I have just used tar to extract the archive, but anyways, I have re-packaged the log-files, you can download them here: http://slackeeevps.no-ip.org/public/...-shakira19.tar
|
Thank you, that worked. (???)
have you tried kernel cli param, Code:
video=vesa |
A few things I see in the logs and configs.
1. nvidia-xconfig: version 295.40 is very old, may cause an ABI mismatch during compilation, update to 313.30 asap. 2. remove "quiet splash" from kernel cli, so you can see errors.... 3. append to the end of the line "nokmsboot vga=791 3" without the quotes. legend, nokmsboot(kernel module loading service, not for proprietary ati and nvidia driver modules. vga=791, (must remove splash and quiet, text mode 1024x768x16bit) 3 is boot to runlevel 3 (don't try to load the gui-xserver) 4. Quote:
Now, you got all that? try again from the beginning, Oh yeah, make sure you have kernel-devel installed for your running kernel version. |
Quote:
|
OK, Thanks tobi(I'm using Mageia) then it should be 1, for runlevel 1. (0,1,5,6)
also, diff... xorg does not contain this section (and is missing fonts) Load "glx" Code:
Section "Module" regards glenn |
So in the end, after this expert conversation, could you please give step by step newbie details as to what I am to attempt ?
|
Yes, but it will take a little while to write. all the info is above, brb...
|
check for and if not, download and install kenel-latest
check for and if not, download and install kernel-devel-latest download nvidia-313.30 > gfx 680 gt (for later) set blacklist nouveau reboot pause at the grub menue screen, F3, then ESC to drop out of that dialogue. use the arrow keys to navigate along that (current) command-line remove any terms, "silent", "quiet", "splash", "vga=..." (without quotes (""), grub uses a space separated list) add these terms to the line, "nokmsboot" "vga=794" "1" now boot. this should get you to runlevel 1 on your .deb system. command interface shell. now login as root or if you have sudo setup, login as usuall. don't try to start X! yet. type, lsmod | grep nouveau (should have no answer), but if it is there, reboot with ctrl-alt-del. repeat the boot process as above. and check again for the module. type, lsmod | grep nouveau (should have no answer), if the module is still loaded, we'll try one more term in the grub cli, video=vesa (maybe vesafb on some systems) once again, check if the module is loaded Hopefully it's gone by now. ;-) cd to the directory you saved the nvidia .run file to...type, sh NVIDIA-Linux-x86_64-313.30.run --ui=none -a legend, --ui=none (plain text installer) -a (answer yes to licence) reboot using the same grub line as was used last. back at the cli, login and check if the module is loaded. type lsmod | grep nvidia if so, check your xorg.conf file for the nvidia listing in the driver section. if all goood, or you edit it, save and (logged in as normal user) and type startx vga=794 (for cli text) Code:
640x480 800x600 1024x768 1280x1024 1600x1200 Ask user at boot see how you go. regards Glenn ps, you may or may not require this section in your xorg.conf. (but I may be wrong) https://www.linuxquestions.org/quest...ml#post4934462 check with a gui, type Code:
glxinfo | grep render |
I've been reading this last post on and off for almost two weeks now, but there's no way I'm going to be able to do all that on my own. I'm just a regular user, and even though I quite like linux, I'm not even close to understanding half of what is written there.
I'd just like to go baby steps at a time to make sure I'm not missing anything. So first step: What is kernel-latest and kernel-dev-latest ? If I do apt-get install these, I don't get anything. |
I'm sorry I am not more help with your distro commands, as I use Mageia.
you may find some usefull info here. |
All times are GMT -5. The time now is 09:51 PM. |