NVidia upgrade from 325 to 331 stuck at tty1

tmoorman · 2013-11-18 23:08:20

I installed the upgrade from 325 to 331 (nvidia, nvidia-libgl and nvidia-utils). On reboot I get stuck at tty1. When I logon and manually startx, I get the x-windows demo. When I downgrade to 325 my system starts up right away.
This is odd because I have 331 running fine on my other Arch box. The main difference between them is the video card: GeForce GTS 450 (331 no) and GeForce 210 (331 yes). The GeForce 210 system has a custom /etc/X11/10-monitor.conf to force load an edid file for a wonky monitor. So it has a Screen section to load the edid file and a Device section where I load the nvidia driver. The GeForce GTS 450 system does not have any custom X11 config files.

I am running KDE. Here is my Xorg.0.log.old

[     2.734] (II) LoadModule: "glx"
[     2.737] (II) Loading /usr/lib/xorg/modules/extensions/libglx.so
[     2.890] (II) Module glx: vendor="NVIDIA Corporation"
[     2.890] 	compiled for 4.0.2, module version = 1.0.0
[     2.890] 	Module class: X.Org Server Extension
[     2.890] (II) NVIDIA GLX Module  331.20  Wed Oct 30 17:36:48 PDT 2013
[     2.890] Loading extension GLX
[     2.891] (==) Matched nouveau as autoconfigured driver 0
[     2.891] (==) Matched nvidia as autoconfigured driver 1
[     2.891] (==) Matched nv as autoconfigured driver 2
[     2.891] (==) Matched vesa as autoconfigured driver 3
[     2.891] (==) Matched modesetting as autoconfigured driver 4
[     2.891] (==) Matched fbdev as autoconfigured driver 5
[     2.891] (==) Assigned the driver to the xf86ConfigLayout
[     2.891] (II) LoadModule: "nouveau"
[     2.893] (WW) Warning, couldn't open module nouveau
[     2.893] (II) UnloadModule: "nouveau"
[     2.893] (II) Unloading nouveau
[     2.893] (EE) Failed to load module "nouveau" (module does not exist, 0)
[     2.893] (II) LoadModule: "nvidia"
[     2.894] (II) Loading /usr/lib/xorg/modules/drivers/nvidia_drv.so
[     2.907] (II) Module nvidia: vendor="NVIDIA Corporation"
[     2.907] 	compiled for 4.0.2, module version = 1.0.0
[     2.907] 	Module class: X.Org Video Driver
[     2.908] (II) LoadModule: "nv"
[     2.909] (WW) Warning, couldn't open module nv
[     2.909] (II) UnloadModule: "nv"
[     2.909] (II) Unloading nv
[     2.909] (EE) Failed to load module "nv" (module does not exist, 0)
[     2.909] (II) LoadModule: "vesa"
[     2.909] (WW) Warning, couldn't open module vesa
[     2.909] (II) UnloadModule: "vesa"
[     2.909] (II) Unloading vesa
[     2.909] (EE) Failed to load module "vesa" (module does not exist, 0)
[     2.909] (II) LoadModule: "modesetting"
[     2.909] (WW) Warning, couldn't open module modesetting
[     2.909] (II) UnloadModule: "modesetting"
[     2.909] (II) Unloading modesetting
[     2.909] (EE) Failed to load module "modesetting" (module does not exist, 0)
[     2.909] (II) LoadModule: "fbdev"
[     2.909] (WW) Warning, couldn't open module fbdev
[     2.909] (II) UnloadModule: "fbdev"
[     2.909] (II) Unloading fbdev
[     2.909] (EE) Failed to load module "fbdev" (module does not exist, 0)
[     2.909] (II) NVIDIA dlloader X Driver  331.20  Wed Oct 30 17:16:53 PDT 2013
[     2.909] (II) NVIDIA Unified Driver for all Supported NVIDIA GPUs
[     2.910] (++) using VT number 7

[     2.913] (II) Loading sub module "fb"
[     2.913] (II) LoadModule: "fb"
[     2.913] (II) Loading /usr/lib/xorg/modules/libfb.so
[     2.915] (II) Module fb: vendor="X.Org Foundation"
[     2.915] 	compiled for 1.14.4, module version = 1.0.0
[     2.915] 	ABI class: X.Org ANSI C Emulation, version 0.4
[     2.915] (WW) Unresolved symbol: fbGetGCPrivateKey
[     2.915] (II) Loading sub module "wfb"
[     2.915] (II) LoadModule: "wfb"
[     2.915] (II) Loading /usr/lib/xorg/modules/libwfb.so
[     2.917] (II) Module wfb: vendor="X.Org Foundation"
[     2.917] 	compiled for 1.14.4, module version = 1.0.0
[     2.917] 	ABI class: X.Org ANSI C Emulation, version 0.4
[     2.917] (II) Loading sub module "ramdac"
[     2.917] (II) LoadModule: "ramdac"
[     2.917] (II) Module "ramdac" already built-in
[     2.920] (II) NVIDIA(0): Creating default Display subsection in Screen section
	"Default Screen Section" for depth/fbbpp 24/32
[     2.920] (==) NVIDIA(0): Depth 24, (==) framebuffer bpp 32
[     2.920] (==) NVIDIA(0): RGB weight 888
[     2.920] (==) NVIDIA(0): Default visual is TrueColor
[     2.920] (==) NVIDIA(0): Using gamma correction (1.0, 1.0, 1.0)
[     2.920] (**) NVIDIA(0): Enabling 2D acceleration
[     2.922] (EE) NVIDIA(0): Failed to initialize the NVIDIA kernel module. Please see the
[     2.922] (EE) NVIDIA(0):     system's kernel log for additional error messages and
[     2.922] (EE) NVIDIA(0):     consult the NVIDIA README for details.
[     2.922] (EE) NVIDIA(0):  *** Aborting ***
[     2.922] (EE) NVIDIA(0): Failing initialization of X screen 0
[     2.922] (II) UnloadModule: "nvidia"
[     2.922] (II) UnloadSubModule: "wfb"
[     2.922] (II) UnloadSubModule: "fb"
[     2.922] (EE) Screen(s) found, but none have a usable configuration.
[     2.922] (EE) 
Fatal server error:
[     2.922] (EE) no screens found(EE)

Last edited by tmoorman (2013-11-18 23:39:49)

karol · 2013-11-18 23:21:30

Did you reboot after the update?
Try https://wiki.archlinux.org/index.php/Nv … figuration

When posting configs, code or command output, please use [ code ] tags https://bbs.archlinux.org/help.php#bbcode

like this

It makes the code more readable and - in case of longer listings - more convenient to scroll through.

tmoorman · 2013-11-18 23:45:36

Thanks for the tip karol. Updated the main post.
Yes I did reboot after the update.
I repeated it just to sure: upgraded to 331, reboot, stuck at tty1. Downgrade to 325, reboot, starts kde.
I will try a custom monitor config /etc/X11/xorg.conf.d/10-monitor.conf.

Section "Device"
    Identifier     "Device0"
    Driver         "nvidia"
#     Driver         "nv"
    VendorName     "NVIDIA Corporation"
    Option         "RenderAccel" "true"
    Option         "ConnectToAcpid" "Off"
    Option         "TripleBuffer" "True"
EndSection

This is the same config as on the GeForce 210 but excludes the Device section (because the GeForce GTS 450 has a new monitor).

tmoorman · 2013-11-19 00:44:36

Tried the monitor config and same result, stuck at tty1.
I added the correct busid to the device section.

# journalctl -b
shows this

...
kdm [399]: X server died during startup
kdm [399]: X server for display :0 cannot be started, session disabled
...

Scimmia · 2013-11-19 00:50:41

Is the kernel module loaded? The error specifically says it failed to initialize the kernel module and to check the kernel logs.

Last edited by Scimmia (2013-11-19 00:52:25)

tmoorman · 2013-11-19 02:10:00

Installed 331 and rebooted
This is what lsmod says:

nvidia              10573185  0 
drm                   238206  2 nvidia,vboxvideo
i2c_core               24164  3 drm,i2c_piix4,nvidia

What does the 0 in the third column mean?
On the system that has no problem with 331 there is a number > 0 in this column.

karol · 2013-11-19 02:14:30

The format is name, size, use count, list of referring modules.

brain0 · 2013-11-19 09:47:19

I have the same problem with 331. The X server starts too early and I did not have time to look into this. Since the nvidia driver behaves weird compared to others, I have trouble delaying kwin far enough.

tmoorman · 2013-11-19 15:52:04

brain0, it is interesting that you mention that the X server starts too early. Another difference between the two systems: the system partition on the 331 friendly box is on an old IDE spinning hard drive and on the 331 hostile system is on a SATA2 SSD. Both systems use systemd.

brain0 · 2013-11-19 16:01:37

My system is on SSD as well. It's weird how this worked fine with 325 and not with 331 - I wish we could wait with udev for /dev/nvidia*, but none of those devices properly registers with udev, so you cannot wait on them.

saty · 2013-11-19 17:18:27

I'm experiencing this very same issue too.

When I was trying to figure this on my own I found some "Your system is not currently configured to drive a VGA console on the primary VGA device. The NVIDIA Linux graphics driver requires the use of a text-mode VGA console." messages in the log files. They probably have been there forever but I haven't bothered with them since now.

So I went ahead and tried passing several different boot parameters for the kernel trying to get rid of this warning. During one of these so called tests everything just decided to work fine and I got all the way to the KDM screen. So this type of "early start" problem seems very likely at least in my mind after reading this thread.It's been like 50:50 for me whether I'll see some error or not.

Everything was fine for me with 325 series of drivers too. I'm running SSD and a gtx 570 fermi card.

Thanks!

brain0 · 2013-11-19 18:22:27

saty wrote:

When I was trying to figure this on my own I found some "Your system is not currently configured to drive a VGA console on the primary VGA device. The NVIDIA Linux graphics driver requires the use of a text-mode VGA console." messages in the log files. They probably have been there forever but I haven't bothered with them since now.

That is normal and you should remember it when you get actual instabilities. I run nvidia together with efifb and experience no problems so far.

Anyway, I seem to have solved the problem for me, just create two files:

# /etc/udev/rules.d/99-make-udev-drm-aware.rules

SUBSYSTEM=="drm", TAG+="systemd"

# /etc/systemd/system/kdm.service.d/10-wait-for-card.conf

[Unit]
Wants=dev-dri-card0.device
After=dev-dri-card0.device

If you have more than one nvidia card, also plug in card1 etc.

cju · 2013-11-19 19:02:56

The same: https://bbs.archlinux.org/viewtopic.php?id=172991

kokoko3k · 2013-11-19 19:24:02

brain0 wrote:

Anyway, I seem to have solved the problem for me, just create two files:
# /etc/udev/rules.d/99-make-udev-drm-aware.rules

SUBSYSTEM=="drm", TAG+="systemd"
# /etc/systemd/system/kdm.service.d/10-wait-for-card.conf

[Unit]
Wants=dev-dri-card0.device
After=dev-dri-card0.device
If you have more than one nvidia card, also plug in card1 etc.

Could you explain what the udev rule does and how is related to the unit, please?

mangus · 2013-11-19 19:38:18

I had the same problem with kdm failed to start after upgrade to 331 and kernel 3.12, I had to startx to have a proper X environment.
Then suddenly It's 3 boots in a row thatI haven't this problem , and I didn't touch anything.
It seems really a timing issue
I have an SSD disk.

Last edited by mangus (2013-11-19 19:39:16)

Thaodan · 2013-11-19 21:27:52

kokoko3k wrote:

brain0 wrote:
Anyway, I seem to have solved the problem for me, just create two files:
# /etc/udev/rules.d/99-make-udev-drm-aware.rules

SUBSYSTEM=="drm", TAG+="systemd"
# /etc/systemd/system/kdm.service.d/10-wait-for-card.conf

[Unit]
Wants=dev-dri-card0.device
After=dev-dri-card0.device
If you have more than one nvidia card, also plug in card1 etc.
Could you explain what the udev rule does and how is related to the unit, please?

As far as I unterstand this tells systemd to start kdm.service only if the depense dev-dri-card0.device is up.

tmoorman · 2013-11-19 21:40:03

Thaodan,
Your fix seems to work for me too.

Nov 19 16:27:17 aria kernel: [drm] Initialized nvidia-drm 0.0.0 20130102 for 0000:01:00.0 on minor 0
Nov 19 16:27:17 aria kernel: NVRM: loading NVIDIA UNIX x86_64 Kernel Module  331.20  Wed Oct 30 17:43:35 PDT 2013
Nov 19 16:27:17 aria systemd[1]: Found device /dev/dri/card0.
Nov 19 16:27:17 aria systemd[1]: Starting K Display Manager...
Nov 19 16:27:17 aria systemd[1]: Started K Display Manager.
Nov 19 16:27:17 aria kernel: nvidia 0000:01:00.0: irq 50 for MSI/MSI-X

When I installed 331, I also got an update to libdrm (libdrm-2.4.48-1). I haven't checked but could it be the new drm?
Anyway, your fix should not break future updates to nvidia. AFAIK, it just forces systemd to pause the kdm service until the nvidia driver has the video card ready.

So, this is not really [SOLVED] yet. This is more of a work around than a fix.

brain0 · 2013-11-19 22:17:24

kokoko3k wrote:

brain0 wrote:
Anyway, I seem to have solved the problem for me, just create two files:
# /etc/udev/rules.d/99-make-udev-drm-aware.rules

SUBSYSTEM=="drm", TAG+="systemd"
# /etc/systemd/system/kdm.service.d/10-wait-for-card.conf

[Unit]
Wants=dev-dri-card0.device
After=dev-dri-card0.device
If you have more than one nvidia card, also plug in card1 etc.
Could you explain what the udev rule does and how is related to the unit, please?

The udev rule instructs systemd to add a .device unit for the dri device that nvidia creates.

$ systemctl status dev-dri-card0.device 
dev-dri-card0.device - /dev/dri/card0
   Follow: unit currently follows state of sys-devices-pci0000:00-0000:00:01.0-0000:01:00.0-drm-card0.device
   Loaded: loaded
   Active: active (plugged) since Di 2013-11-19 19:13:53 CET; 4h 0min ago
   Device: /sys/devices/pci0000:00/0000:00:01.0/0000:01:00.0/drm/card0

The unit file drop-in ensures that kdm.service waits until the device is actually there and thus solves the problem that kdm starts "too early".

firekage · 2013-11-19 22:27:40

I would like to ask for one thing: how to check which card, in fact, is used? I'm asking about "dev-dri-card0.device". How to list this name from terminal in order to find out it? I would like to know how to check it, and why it is named like that (want to learn more about Arch, that's why i asked).

brain0 · 2013-11-20 00:52:11

The name itself is meaningless, it's just that if you have one GPU, you only have the /dev/dri/card0 device.

kokoko3k · 2013-11-20 07:31:09

Thanks brain0,
Any hope this will be included in the arch systemd package or better to be reported to systemd mantainers?

cju · 2013-11-20 18:24:16

tmoorman wrote:

Anyway, your fix should not break future updates to nvidia. AFAIK, it just forces systemd to pause the kdm service until the nvidia driver has the video card ready.
So, this is not really [SOLVED] yet. This is more of a work around than a fix.

This.

saty · 2013-11-20 22:53:15

I can quickly add that the proposed fix/workaround works just fine for me. It's really non intrusive so I'm very happy with it. Thanks @brain0 for the quick work on it.

mangus · 2013-11-21 09:33:55

brain0 wrote:

My system is on SSD as well. It's weird how this worked fine with 325 and not with 331 - I wish we could wait with udev for /dev/nvidia*, but none of those devices properly registers with udev, so you cannot wait on them.

This is probably my current problem. I have a desktop pc with integrated intel graphic chip (disabled) and gts460 nvidia card.
Can we insert some grace time somewhere else during boot ? I know it's horrible but until it's broken....

kokoko3k · 2013-11-21 12:39:06

Workaround is here:
https://bbs.archlinux.org/viewtopic.php … 0#p1351600

Last edited by kokoko3k (2013-11-21 12:39:40)

Arch Linux

#1 2013-11-18 23:08:20

NVidia upgrade from 325 to 331 stuck at tty1

#2 2013-11-18 23:21:30

Re: NVidia upgrade from 325 to 331 stuck at tty1

#3 2013-11-18 23:45:36

Re: NVidia upgrade from 325 to 331 stuck at tty1

#4 2013-11-19 00:44:36

Re: NVidia upgrade from 325 to 331 stuck at tty1

#5 2013-11-19 00:50:41

Re: NVidia upgrade from 325 to 331 stuck at tty1

#6 2013-11-19 02:10:00

Re: NVidia upgrade from 325 to 331 stuck at tty1

#7 2013-11-19 02:14:30

Re: NVidia upgrade from 325 to 331 stuck at tty1

#8 2013-11-19 09:47:19

Re: NVidia upgrade from 325 to 331 stuck at tty1

#9 2013-11-19 15:52:04

Re: NVidia upgrade from 325 to 331 stuck at tty1

#10 2013-11-19 16:01:37

Re: NVidia upgrade from 325 to 331 stuck at tty1

#11 2013-11-19 17:18:27

Re: NVidia upgrade from 325 to 331 stuck at tty1

#12 2013-11-19 18:22:27

Re: NVidia upgrade from 325 to 331 stuck at tty1

#13 2013-11-19 19:02:56

Re: NVidia upgrade from 325 to 331 stuck at tty1

#14 2013-11-19 19:24:02

Re: NVidia upgrade from 325 to 331 stuck at tty1

#15 2013-11-19 19:38:18

Re: NVidia upgrade from 325 to 331 stuck at tty1

#16 2013-11-19 21:27:52

Re: NVidia upgrade from 325 to 331 stuck at tty1

#17 2013-11-19 21:40:03

Re: NVidia upgrade from 325 to 331 stuck at tty1

#18 2013-11-19 22:17:24

Re: NVidia upgrade from 325 to 331 stuck at tty1

#19 2013-11-19 22:27:40

Re: NVidia upgrade from 325 to 331 stuck at tty1

#20 2013-11-20 00:52:11

Re: NVidia upgrade from 325 to 331 stuck at tty1

#21 2013-11-20 07:31:09

Re: NVidia upgrade from 325 to 331 stuck at tty1

#22 2013-11-20 18:24:16

Re: NVidia upgrade from 325 to 331 stuck at tty1

#23 2013-11-20 22:53:15

Re: NVidia upgrade from 325 to 331 stuck at tty1

#24 2013-11-21 09:33:55

Re: NVidia upgrade from 325 to 331 stuck at tty1

#25 2013-11-21 12:39:06

Re: NVidia upgrade from 325 to 331 stuck at tty1

Board footer