Installation lost after reboot (sgfxi)
goe
Status: Curious
Joined: 28 Apr 2014
Posts: 5
Reply Quote
I am using 'sgfxi' script to install nVidia drivers, and it works perfectly. In the end of the installation, the script asks me to reboot, and after this reboot the drivers are working perfectly. But, after next reboot, the graphic card is not found by the system, and xorg gives this error:

:: Code ::

Failed to intialize the NVIDIA kernel module. Please see the system's kernel log for additional error messages and consult the NVIDIA README
for details.
No devices detected.
no secreens found(EE)


Any idea why this is happening?

Thanks.
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 3986
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
Show: inxi -bxx

I have an idea, so I will wait to see that output to see if my idea is right.

Also, run:

sgfxi --debugger

and either put the file on a file uploading site like dropbox, or use ftp to drag it here: ftp://techpatterns.com/incoming

those two things let me skip hours of debugging headaches, so I won't proceed without them.
Back to top
goe
Status: Curious
Joined: 28 Apr 2014
Posts: 5
Reply Quote
Hi,
I have left the package into the ftp server with the name: 'sgfxi-2014-04-29-184449-ju'.

The output for inxi is:
:: Code ::

System:    Host: ju Kernel: 3.13-1-amd64 x86_64 (64 bit gcc: 4.8.2)
           Console: tty 1 dm: lightdm Distro: Debian GNU/Linux jessie/sid
Machine:   System: Gigabyte product: Z87X-UD3H Chassis: type: 3
           Mobo: Gigabyte model: Z87X-UD3H-CF v: x.x
           Bios: American Megatrends v: F5 date: 05/16/2013
CPU:       Quad core Intel Core i7-4770K CPU (-HT-MCP-) clocked at 2908 MHz
Graphics:  Card: NVIDIA GK104 [GeForce GTX 770]
           bus-ID: 01:00.0 chip-ID: 10de:1184
           Display Server: X.org 1.15.0.901 driver: nvidia
           tty size: 80x25 Advanced Data: N/A out of X
Network:   Card: Intel Ethernet Connection I217-V
           driver: e1000e v: 2.3.2-k port: f040
           bus-ID: 00:19.0 chip-ID: 8086:153b
Drives:    HDD Total Size: 2120.4GB (26.5% used)
Info:      Processes: 136 Uptime: 8 min Memory: 227.6/16038.1MB
           Init: SysVinit v: 2.88 runlevel: 2 default: 2 Gcc sys: 4.8.2
           Client: Shell (bash 4.3.81 running in tty 1) inxi: 2.1.16


Does this confirm your thoughts?
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 3986
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
No, I am pleased to say that it dis-confirms my thoughts, so to speak, which means, you get support.

Thank you for the data, it helps remove many possible explanations.

I can see from what the debugger faileid to collect that I will need to expand it a little bit, but it's enough for now.

My first guess is that you have your system set up incorrectly, that's easy to confirm or unconfirm:

do you have a separate /boot partition?

Is it a multiboot system? When I have seen this type of issue in the past, it's almost always been because of system setup, ie, the user installs a new kernel, believes that this new kernel is running the system, but it's actually not, because of some unusual /boot setup.

I don't remember the details, but tell me exactly the setup of this system, and we can start from there.

The nvidia kernel module believes it installed to your kernel, but on true restart (reboot keeps some stuff in ram, so the best test is always: shutdown, wait 60 seconds, then start system again), the kernel you believe is running the system says it does not have the nvidia module installed.

I've updated sgfxi to use the xiin.py uploader module from inxi debugger, so now to run it, all you have to do is: sgfxi --debugger
and it will create then upload the file, assuming you have an internet connection of course.
Back to top
goe
Status: Curious
Joined: 28 Apr 2014
Posts: 5
Reply Quote
Ok, yes, my boot configuration is a mess. I was having problems with uefi boot configuration, and I am using the bios setup to change os boot.
My debian os is starting from the MBR of the second hdd.
But well, the module is loaded in the active kernel, and I can see 'nvidia' module in the 'lsmod' command output.

Also, if I do:
:: Code ::
sudo rmmod nvidia
sudo modprove nvidia
sudo lightdm start

the system works again.

By the way, I am downloading the script from 'smxi.org/smxi.zip', but I dont know if this is the last version, because is not trying to upload the files to the net. Am I using the wrong script?

Thanks for your help.
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 3986
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
smxi/sgfxi always update themselves to latest version if out of date, so it doesn't matter which you first install.

smxi taught me that issues like this, almost always actually, are caused by improperly setup /boot stuff.

If I were to guess, I'd guess that initrdimg is going to the wrong place, but that's certainly the cause of your probllems.

there's several sections, one, is grub blacklisting the nouveau module. It is, by the debugger data. Two, are blacklists going to /etc/modules, and yes, they are.

So that's not the issue.

Xorg.0.log unambiguously shows that nvidia module is not present, that's without any question.

My guess, if I had to guess, and I haven't seen these issues in a while, is that, you got a new debian kernel, but the version number is the same, but the new one, or the old one, is not actually the one that is booting your system, even though you believe otherwise.

That's the only way I know of where nvidia will fail to load.

Because debugging broken /boot scenarios is far outside of sgfxi/smxi support, you'll have to fix the issues yourself, all I can tell you is areas to look into.

For example, try the liquorix or siduction kernels, those do not suffer from same version number but actually different kernel problems that debian kernels suffer from, that will show you at least that:

A: sgfxi installs nvidia to that kernel
B: the kernel that boots is the kernel you thought you had installed: uname -r shows that, so does inxi.

Try that, it should help clarify your boot issues.

Once you can see that the kernel that sgfxi installed nvidia to is the same as booted, then you are well on your way to fixing it, since most likely it will not be that case.

If it's something else, I can't say. Also there are potential uefi boot scenarios that I have not seen yet aa well. inxi output has been updated to show that data as well, though older inxi's do not show it. Always update inxi first when creating debugger output, it should be at least version 2.0, current in debian testing/sid is 2.1.16, still out of date, but not in ways that matter here.
Back to top
goe
Status: Curious
Joined: 28 Apr 2014
Posts: 5
Reply Quote
I think that the first thing I have do is fix the boot problem, and then I will use the script again to reinstall the drivers in a proper environment.
Thank you for your help, and I will answer when I achieve it.
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 3986
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
I strongly suggest you try installing the liquorix or siduction kernel, both available via smxi advanced kernel install options

Doing this takes you about 5 minutes, then you will know what is going on since you can easily check which kernel booted your system since it's obviosu if it's the right one or not.

Rather than 'reboot', use 'halt' then wait, then start, to avoid resident data in ram still being active on reboot.
Back to top
goe
Status: Curious
Joined: 28 Apr 2014
Posts: 5
Reply Quote
Hi,

I finally solved the problem reinstalling all the system. After the installation, with the correct booting configuration, the script works perfect, and the driver is installed.

Thank for your help.
Back to top
sedonix
Status: Contributor
Joined: 09 Sep 2008
Posts: 56
Location: Sedona, AZ
Reply Quote
Now that nails the problem down, doesn't it!
Back to top
Display posts from previous:   

All times are GMT - 8 Hours