Page: 1, 2  Next

sgfxi on siduction 13.1 "firestarter"
julian516
Status: Interested
Joined: 18 Aug 2009
Posts: 43
Location: Loveland, CO
Reply Quote
I posted this on the siduction forum:

After a fresh 13.1 installation I immediatiately did a dist-upgrade for new packages. I then installed the smxi suite and I used sgfxi to install a driver for my nvidia 698M card. In the past sgfxi has been quite reliable with siduction so I anticipated no trouble.

sgfxi attempted to install nvidia 331.20. This failed repeatedly. The effort stalls with a line reading:
rv (rv) 169.16.143.241:21 ---

the only thing I could do was to restart the system wth shutdown -r now.

Everything on the system works flawlesly but I am stubborn so I attempted to use the procedure in the siduction manual to install an nvidia driver. All went well until I attempted "apt-get install ... nvidia-kernel-dkms". apt reported that it could not find this file. I did apt-cache search on nvidia and I found quite a lot of material but I was not confident enough to try installing things with no sense of what I was doing.

What has worked is sgfxi -o 319.72. I used that module because I saw it referenced in the apt-cache search referred to above. So if anyone else attempts to use sigfixi to install an nvidia driver perhaps this experience will be helpful.

inxi reports (edited for brevity):

System: Host: Home03 Kernel: 3.12-1.towo-siduction-amd64 x86_64 (64 bit)
Desktop: KDE 4.11.3 Distro: siduction 13.1.0 Firestarter - kde - (201305211844)
Machine: System: System76 product: Pangolin Performance version: panp4n
Mobo: CLEVO model: M740TU(N)/M760TU(N) Bios: Phoenix version: 1.02.18 date: 02/25/2009
CPU: Dual core Intel Core2 Duo CPU T6400 (-MCP-) cache: 2048 KB flags: (lm nx sse sse2 sse3 sse4_1 ssse3)
Clock Speeds: 1: 1200.00 MHz 2: 1200.00 MHz
Graphics: Card: NVIDIA G98M [GeForce 9300M GS]
X.org: 1.14.3 drivers: vesa,nouveau (unloaded: fbdev) tty size: 183x64 : 1000 Mbps duplex: full mac: 00:90:f5:89:f9:fd
Card-2: Intel WiFi Link 5100 driver: iwlwifi
IF: wlan0 state: down mac: 00:21:5d:9f:da:0e
Drives: HDD Total Size: 252.1GB (44.4% used) 1: id: /dev/sda model: Hitachi_HTS72322 size: 250.1GB
2: USB id: /dev/sdb model: STORE_N_GO size: 2.0GB
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4075
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
100 to 1 your issue is not related at all to nvidia drivers.

A standard stable nvidia driver install failing plus a standard apt install failing points to some type of file system failure, have you checked diskspace, for example?

inxi -pl is useful in this circumstance, also, a fsck [root partition] would be useful to run.

what I would not spend much time looking at is sgfxi, or nvidia. A full system crash, well, a near crash since the shutdown command works, means something went majorly wrong, and that something was almost certainly not nvidia.
Back to top
julian516
Status: Interested
Joined: 18 Aug 2009
Posts: 43
Location: Loveland, CO
Reply Quote
Much appreciated. Full backup has been done and we shall see what happens. Disk space should be ample.

Devil tells me that the first line in the siduction manual for nvidia installation is wrong. What he says is:

:: Quote ::
The manual needs an update, as nvidia changed the name of nvidia-kernel-dkms to nvidia-driver (at least for your card.


Well that would bolix the apt install.

sgfxi in my experience is dead-bang reliable, which is why I was surprised. This is a high-mileage 5 year old laptop. You may well be right. If you are I will know it shortly.
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4075
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
apt-cache search nvidia-kernel-dkms

gives packages. Always check first when apt says something doesn't exist. Could be a virtual package, but even those exist.

:: Code ::
apt-get install -s nvidia-kernel-dkms
NOTE: This is only a simulation!
      apt-get needs root privileges for real execution.
      Keep also in mind that locking is deactivated,
      so don't depend on the relevance to the real current situation!
Reading package lists... Done
Building dependency tree       
Reading state information... Done
The following extra packages will be installed:
  dkms glx-alternative-mesa glx-alternative-nvidia glx-diversions nvidia-alternative nvidia-installer-cleanup nvidia-kernel-common
Suggested packages:
  nvidia-driver
The following NEW packages will be installed:
  dkms glx-alternative-mesa glx-alternative-nvidia glx-diversions nvidia-alternative nvidia-installer-cleanup nvidia-kernel-common nvidia-kernel-dkms


Unless the package was just removed a few days ago, of course. nvidia-driver is a metapackage, that's all. If you failed to update apt, apt-get update, then of course, apt could request an older version of the package, which will give a package doesn't exist error, that's about the only way you could get that error if your system is working.

I have however just remembered bug reports I got about specifically the towo/siduction kernel configuration causing a failure to install 331.20, the 331 series specifically that is, but nobody followed up on it and 331.20 works fine on debian stable apparently. Actually not just siduction, a few other distros reported problems with their 331.xx plus 3.12 kernels, but I was unable to track down any causes, but the failure should have resulted in sgfxi exiting with a handled error, showing error, not a weird few lines of something. All online error reports are totally incoherent and unusable at this point so I was not able to determine anything.

The 319 driver works in user tests on 3.12 with the siduction configuration. I know an override flag I can use on the driver install but I am reluctant to use it until it's clear why 331 drivers work on vanilla debian kernels but not on some others.

If you find out more, let me know.
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4075
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
also, post the last 100 or so lines of /var/log/sgfxi/sgfxi.log
Back to top
anticapitalista
Status: Contributor
Joined: 13 Jun 2008
Posts: 202
Location: Greece
Reply Quote
The 'failure' to build 330 nvidia kernel on siduction (with the unified memory error) is due to this being set to y in the latest siduction kernel.

# CONFIG_UIDGID_STRICT_TYPE_CHECKS is not set

If it is not set, as with the latest 3.12.1 antiX kernel, then sgfxi builds the 330 nvidia module just fine.
Back to top
julian516
Status: Interested
Joined: 18 Aug 2009
Posts: 43
Location: Loveland, CO
Reply Quote
My thanks to each of you! I will be following up. It is starting to sound to me as though we have some sort of towo kernel issue?

I am not able to get to my Linux box at the moment but I will follow up as soon as I am able, which might be tomorrow.

Again my thanks.
Back to top
drb
Status: Contributor
Joined: 09 Jul 2009
Posts: 130
Reply Quote
I had the same problem here on a fresh install of siduction 13.1. You can install nouveau using sgfxi to get to your new install. I managed to get 331.20 to install from the siduction experimental repository.

:: Code ::
inxi -v3
System:    Host: siductionbox Kernel: 3.12-1.towo-siduction-amd64 x86_64 (64 bit, gcc: 4.8.2)
           Desktop: KDE 4.11.3 (Qt 4.8.6) Distro: siduction 13.1.0 Firestarter - kde - (201305211844)
Machine:   Mobo: Gigabyte model: M57SLI-S4 Bios: Award version: F14 date: 03/07/2008
CPU:       Dual core AMD Athlon 64 X2 5600+ (-MCP-) cache: 2048 KB flags: (lm nx sse sse2 sse3 svm) bmips: 11257.2
           Clock Speeds: 1: 2800.00 MHz 2: 2800.00 MHz
Graphics:  Card: NVIDIA GT218 [GeForce 210] bus-ID: 02:00.0 X.Org: 1.14.3 driver: nvidia Resolution: 1920x1080@60.0hz
           GLX Renderer: GeForce 210/PCIe/SSE2 GLX Version: 3.3.0 NVIDIA 331.20 Direct Rendering: Yes
Network:   Card: NVIDIA MCP55 Ethernet driver: forcedeth port: e400 bus-ID: 00:08.0
           IF: eth0 state: up speed: 100 Mbps duplex: full mac: 00:16:e6:84:ec:51
Drives:    HDD Total Size: 4961.0GB (35.3% used) 1: model: WDC_WD3200AAKS
           2: model: WDC_WD6400AAKS 3: model: ST2000DM001 4: model: ST2000DM001
Info:      Processes: 173 Uptime: 4:17 Memory: 1040.4/3960.0MB Runlevel: 5 Gcc sys: 4.8.2
           Client: Shell (bash 4.2.45) inxi: 1.9.16

Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4075
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
I can do a custom towo kernel config adjustment, what is being done there is the uvm flag is set to false in the nvidia installer, tha't's probably why the repo driver works.

anticap, nice going, you actually figured it out, good.

I will see if there is a special test I can run for those specific kernels that have that flag set in the compile stuff.
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4075
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
finding that config flag will be easy, so I'll be able to force all kernels that were created with it into using the override flag on the nvidia installer, I'll do that a bit later today or tomorrow.

What the flag does doesn't interest me that much as long as it's easy to detect and handle.
Back to top
Display posts from previous:   
Page: 1, 2  Next
All times are GMT - 8 Hours