View previous topic :: View next topic |
Author |
Message |
bigfoot- n00b
Joined: 19 Jul 2003 Posts: 19 Location: Erlangen, Germany
|
Posted: Fri Sep 02, 2005 6:09 pm Post subject: device script works fine |
|
|
b1f1 wrote: | i had the same problem with kernel 2.6.13 nvidia 6000 series and udev.
i found out that /dev/nvidia* was gone, that was the problem.
for all of you who still have problems: run this script as root:
|
Thanks for debugging this, recreating the device nodes works fine. For udev users who have RC_DEVICE_TARBALL="no" in /etc/conf.d/rc it helps to run the provided script at boot time, for example from /etc/conf.d/local.start.
I also added this information to bug #104369. _________________ "We have seen the light at the end of the tunnel, and it's out." -- unknown |
|
Back to top |
|
|
gpierce Tux's lil' helper
Joined: 15 Dec 2004 Posts: 111
|
Posted: Fri Sep 02, 2005 8:22 pm Post subject: Puzzling |
|
|
I am completely perplexed. I just switched on my computer and X started without a hitch! I have altered nothing. I am still using the nvidia module and not the nv driver provided in the kernel. My computer had been off for several days I boot into 2.6.13, and it is af if there never was a problem. I am pleased, but what is going on here? |
|
Back to top |
|
|
skywarp n00b
Joined: 19 Jun 2004 Posts: 41 Location: Germany
|
Posted: Sat Sep 03, 2005 6:29 pm Post subject: |
|
|
Same situation here. All over sudden Xorg worked. Maybe an upgrade thing (udev-058 to 068)? But downgrading to 058 did not break Xorg again. So, I have no clue what made my box working all over sudden. |
|
Back to top |
|
|
bigfoot- n00b
Joined: 19 Jul 2003 Posts: 19 Location: Erlangen, Germany
|
Posted: Mon Sep 05, 2005 9:50 am Post subject: retry with RC_DEVICE_TARBALL="no" |
|
|
gpierce, skywarp: Maybe retry with RC_DEVICE_TARBALL="no" in your /etc/conf.d/rc? I guess these missing devices are just kept in this tarball from some earlier 2.6.12 boot... _________________ "We have seen the light at the end of the tunnel, and it's out." -- unknown |
|
Back to top |
|
|
skywarp n00b
Joined: 19 Jun 2004 Posts: 41 Location: Germany
|
Posted: Mon Sep 05, 2005 12:45 pm Post subject: tarball="yes" |
|
|
You are absolutely right about that. Thanx for the hint. |
|
Back to top |
|
|
gpierce Tux's lil' helper
Joined: 15 Dec 2004 Posts: 111
|
Posted: Mon Sep 05, 2005 10:01 pm Post subject: |
|
|
Will give it a try, thank you.
Greg |
|
Back to top |
|
|
piewie Guru
Joined: 26 Sep 2002 Posts: 489 Location: old Europe
|
Posted: Tue Sep 06, 2005 5:59 pm Post subject: |
|
|
my nvidia card isn't visible in /proc/interrupts either. |
|
Back to top |
|
|
bigfoot- n00b
Joined: 19 Jul 2003 Posts: 19 Location: Erlangen, Germany
|
Posted: Fri Sep 09, 2005 10:48 am Post subject: NVmakedevices.sh |
|
|
Note bug 104369, comment #18: There is already a script that comes with nvidia-kernel that re-creates the necessary devices: /sbin/NVmakedevices.sh
Though it essentially doesn't do anything else than b1f1's script (maybe he even quoted from there). _________________ "We have seen the light at the end of the tunnel, and it's out." -- unknown |
|
Back to top |
|
|
gpierce Tux's lil' helper
Joined: 15 Dec 2004 Posts: 111
|
Posted: Sun Oct 02, 2005 4:42 pm Post subject: |
|
|
Bigfoot,
Do you know why these device nodes weren't created to begin with. I just installed a new 2.6.13 kernel (from the ck branch) and had to look up this thread again to figure out what I had to do (I have such poor memory). What a pain! I am amazed that people even knew about this.
Greg |
|
Back to top |
|
|
Grunt n00b
Joined: 20 Dec 2002 Posts: 19
|
Posted: Mon Oct 03, 2005 5:13 pm Post subject: Nvida Problems |
|
|
Thanks to n1fb1 for the script. I've been fighting this since yesterday when I installed Gentoo on a system (the system had Gentoo several months ago but due to hardware, etc. it's been down). I had X starting with startx and running xfce4 but all black consoles when I exited xfce4. I have a GeForce 3 TI500 and the script worked. |
|
Back to top |
|
|
massheep Tux's lil' helper
Joined: 02 Jan 2005 Posts: 81
|
Posted: Wed Oct 12, 2005 9:06 pm Post subject: |
|
|
@raistlin
i know a guy at another forum who has just the same avatare as you do...
and wait! he even has the same signature! now that's what i call a coincidence. |
|
Back to top |
|
|
rhoegg n00b
Joined: 25 Nov 2002 Posts: 5
|
Posted: Fri Oct 14, 2005 1:22 pm Post subject: Isn't udev supposed to do this? |
|
|
I'm having this problem as well, with a GeForce4 Ti4400 and 2.6.13. I thought udev was responsible for putting the device files there with the right group ownership. My /etc/udev/rules.d/50-udev.rules looks like it wants to do it for me:
Code: |
# dri devices
KERNEL="card*", NAME="dri/card%n", GROUP="video"
KERNEL="nvidia*", NAME="%k", GROUP="video"
KERNEL="3dfx*", NAME="%k", GROUP="video"
|
I've never taken the time to figure out how udev works before, and now I'm going to be late for work. It looks like udev is supposed to create these device files when it hears that the kernel has a device named nvidia. hotplug is supposed to call udevsend when it finds out about any new device. I started looking into when hotplug is supposed to find out about the video card, and it looks like it comes from sysfs somehow. When I was looking around /sys/bus/pci, it looked like the nvidia driver wasn't mapped to the same address as the card itself (which was all zeroes).
I won't be able to work on this for a day or two, so I hope this gets someone started. Manually creating the device files didn't feel right to me.
good luck! |
|
Back to top |
|
|
Defiance n00b
Joined: 14 Feb 2005 Posts: 28
|
Posted: Fri Oct 14, 2005 2:06 pm Post subject: |
|
|
I upgraded udev to 068. Didn't work. I re-emerged the driver, didn't work. Finally I emerged nvidia-glx to the latest ~amd64 version and it worked. Can anyone explain what happened and how it started working? |
|
Back to top |
|
|
dmpogo Advocate
Joined: 02 Sep 2004 Posts: 3269 Location: Canada
|
Posted: Fri Oct 14, 2005 2:54 pm Post subject: Re: Isn't udev supposed to do this? |
|
|
rhoegg wrote: | I'm having this problem as well, with a GeForce4 Ti4400 and 2.6.13. I thought udev was responsible for putting the device files there with the right group ownership. My /etc/udev/rules.d/50-udev.rules looks like it wants to do it for me:
Code: |
# dri devices
KERNEL="card*", NAME="dri/card%n", GROUP="video"
KERNEL="nvidia*", NAME="%k", GROUP="video"
KERNEL="3dfx*", NAME="%k", GROUP="video"
|
I've never taken the time to figure out how udev works before, and now I'm going to be late for work. It looks like udev is supposed to create these device files when it hears that the kernel has a device named nvidia. hotplug is supposed to call udevsend when it finds out about any new device. I started looking into when hotplug is supposed to find out about the video card, and it looks like it comes from sysfs somehow. When I was looking around /sys/bus/pci, it looked like the nvidia driver wasn't mapped to the same address as the card itself (which was all zeroes).
I won't be able to work on this for a day or two, so I hope this gets someone started. Manually creating the device files didn't feel right to me.
good luck! |
See comments 15 and 17 in the following bug report
https://bugs.gentoo.org/show_bug.cgi?id=104369
However, the link in the comment 17 points to the changes to sysfs discussed over a year ago. Did they propogate to the main stream kernel only since 2.6.13 ?
I also hate the idea that after all those years, and switching with 2.6.13 to the brand modern udev way of managing /dev I now have to write scripts for manual device creation. |
|
Back to top |
|
|
Gotterdammerung l33t
Joined: 11 Feb 2004 Posts: 627 Location: Rio de Janeiro, Brazil
|
Posted: Fri Oct 14, 2005 8:20 pm Post subject: |
|
|
Same problem here. _________________ A mind that is stretched by a new experience can never go back to its old dimensions. - Oliver Wendell Holmes |
|
Back to top |
|
|
rhoegg n00b
Joined: 25 Nov 2002 Posts: 5
|
Posted: Sat Oct 15, 2005 5:03 pm Post subject: It's intentional, so let's just work around it |
|
|
I guess I should have read the bug report before I started digging in. So, we know it is a "feature" instead of a bug, because the udev developers decided not to support non-GPL devices. To paraphrase my comment on the bug report, we should make the devices for the user in another script in the boot runlevel:
/etc/init.d/nvidia-devices
Code: |
if [ ! -e /dev/nvidia0 ]; then
/sbin/NVmakedevices.sh
fi
|
command to make it run at boot
Code: | rc-update add nvidia-devices boot |
|
|
Back to top |
|
|
MadScientist n00b
Joined: 11 Apr 2003 Posts: 62
|
Posted: Sun Oct 16, 2005 5:45 am Post subject: |
|
|
Perhaps it's time to branch udev. There will be the 'pure' GPL-only udev that nobody uses and there will be the one that supports *all* drivers. This sounds a lot like XFree vs. xorg... |
|
Back to top |
|
|
dmpogo Advocate
Joined: 02 Sep 2004 Posts: 3269 Location: Canada
|
Posted: Sun Oct 16, 2005 5:04 pm Post subject: |
|
|
MadScientist wrote: | Perhaps it's time to branch udev. There will be the 'pure' GPL-only udev that nobody uses and there will be the one that supports *all* drivers. This sounds a lot like XFree vs. xorg... |
To my understanding, the problem is not with udev, but with sysfs, which is a part of the kernel.
I don't have an access to the box running 2.6.13 right now, but could anybody check if an entry in /sys corresponding to NVIDIA card is
created at all in 2.6.13 ? I suspect it may not, if non-GPL explanation is right.
On my 2.6.11 kernel the entry is a subdirectory
/sys/devices/pci0000:04/0000:04:01.0/0000:05:00.0
where 0000:04:01.0 is pci address for AGP bridge (which sits on 0000:04 controller) and 0000:05:00.0 is NVIDIA controller.
You can get those numbers with "lspci"
# ls /sys/devices/pci0000:04/0000:04:01.0/0000:05:00.0 |
|
Back to top |
|
|
gpierce Tux's lil' helper
Joined: 15 Dec 2004 Posts: 111
|
Posted: Sat Oct 29, 2005 5:29 pm Post subject: Same problems with 2.6.14 |
|
|
I am having the same problem again with the latest 2.6.14 kernel release except this time when I run /sbin/NVdevice... X still won't start. I am wondering if anyone else has encountered this problem? |
|
Back to top |
|
|
Freesnake n00b
Joined: 12 May 2005 Posts: 35
|
Posted: Sun Nov 13, 2005 12:16 pm Post subject: |
|
|
Unexpectably that even now nvidia driver becomes so buggy¡¡& ATI drivers are comfortable now¡¡ |
|
Back to top |
|
|
fil1411 n00b
Joined: 13 Nov 2005 Posts: 4
|
Posted: Sun Nov 13, 2005 7:04 pm Post subject: nvidia module cannot load even after creating nodes |
|
|
Hello
The nvidia module still cannot load on my gentoo box with kernel 2.6.13, even after I tried all the advices, links and suggestions of the thread :
I compiled the module with the last nvidia driver NVIDIA-Linux-x86-1.0-7676-pkg1.run (my card is geforce4, supported by this version of the driver)
'modprobe nvidia' returns :
"FATAL: Error inserting nvidia (/lib/modules/2.6.13-gentoo-r5/video/nvidia.ko): No such device"
''dmesg' returns this :
" NVRM: The NVIDIA probe routine was not called for 1 device(s).
NVRM: This can occur when a driver such as rivafb or rivatv was
NVRM: loaded and obtained ownership of the NVIDIA device(s).
NVRM: Try unloading the rivafb (and/or the rivatv) kernel module
NVRM: (or reconfigure your kernel without rivafb support), then
NVRM: try loading the NVIDIA kernel module again.
NVRM: No NVIDIA graphics adapter probed!"
(the rivafb support is not activated in the kernel, nor as a module)
and nvidia devices look like this :
crw-rw-rw- 1 root root 195, 0 Nov 13 12:45 /dev/nvidia0
crw-rw-rw- 1 root root 195, 1 Nov 13 12:45 /dev/nvidia1
crw-rw-rw- 1 root root 195, 2 Nov 13 12:45 /dev/nvidia2
crw-rw-rw- 1 root root 195, 3 Nov 13 12:45 /dev/nvidia3
crw-rw-rw- 1 root root 195, 4 Nov 13 12:45 /dev/nvidia4
crw-rw-rw- 1 root root 195, 5 Nov 13 12:45 /dev/nvidia5
crw-rw-rw- 1 root root 195, 6 Nov 13 12:45 /dev/nvidia6
crw-rw-rw- 1 root root 195, 7 Nov 13 12:45 /dev/nvidia7
crw-rw-rw- 1 root root 195, 255 Nov 13 12:45 /dev/nvidiactl
Does anyone have an idea of what is happening ?
thanks |
|
Back to top |
|
|
mariley n00b
Joined: 29 Dec 2004 Posts: 2
|
Posted: Mon Nov 14, 2005 1:20 am Post subject: nvidia: module license 'NVIDIA' taints kernel. |
|
|
I have had the same problems.
I have a GeForce4 Mx420 and I have tried everything on the web related to this issue and so far nothing has fixed it.
I have read several posts where people are just having trouble with Xwindows but I am unable to load the nvidia module at boot time.
I have compiled nvidia mod support into my kernel and disabled the riva modules. Prior to this I was just getting an error stating nvidia.ko: no such device
Code: |
mythtv tmp # modprobe nvidia
FATAL: Error inserting nvidia (/lib/modules/2.6.13-gentoo-r5/kernel/drivers/video/nvidia.ko): No such device
|
Code: |
mythtv tmp # lspci
0000:00:00.0 Host bridge: VIA Technologies, Inc. VT8375 [KM266/KL266] Host Bridge
0000:00:01.0 PCI bridge: VIA Technologies, Inc. VT8633 [Apollo Pro266 AGP]
0000:00:08.0 Multimedia video controller: Internext Compression Inc iTVC16 (CX23416) MPEG-2 Encoder (rev 01)
0000:00:09.0 Multimedia video controller: Internext Compression Inc iTVC15 MPEG-2 Encoder (rev 01)
0000:00:0a.0 Ethernet controller: 3Com Corporation 3c905 100BaseTX [Boomerang]
0000:00:10.0 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller (rev 80)
0000:00:10.1 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller (rev 80)
0000:00:10.2 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller (rev 80)
0000:00:10.3 USB Controller: VIA Technologies, Inc. USB 2.0 (rev 82)
0000:00:11.0 ISA bridge: VIA Technologies, Inc. VT8235 ISA Bridge
0000:00:11.1 IDE interface: VIA Technologies, Inc. VT82C586A/B/VT82C686/A/B/VT823x/A/C PIPC Bus Master IDE (rev 06)
0000:00:11.5 Multimedia audio controller: VIA Technologies, Inc. VT8233/A/8235/8237 AC97 Audio Controller (rev 50)
0000:01:00.0 VGA compatible controller: nVidia Corporation NV17 [GeForce4 MX 420] (rev a3)
|
Code: |
#dmesg
nvidia: module license 'NVIDIA' taints kernel.
NVRM: The NVIDIA probe routine was not called for 1 device(s).
NVRM: This can occur when a driver such as rivafb or rivatv was
NVRM: loaded and obtained ownership of the NVIDIA device(s).
NVRM: Try unloading the rivafb (and/or the rivatv) kernel module
NVRM: (or reconfigure your kernel without rivafb support), then
NVRM: try loading the NVIDIA kernel module again.
NVRM: No NVIDIA graphics adapter probed!
NVRM: The NVIDIA probe routine was not called for 1 device(s).
NVRM: This can occur when a driver such as rivafb or rivatv was
NVRM: loaded and obtained ownership of the NVIDIA device(s).
NVRM: Try unloading the rivafb (and/or the rivatv) kernel module
NVRM: (or reconfigure your kernel without rivafb support), then
NVRM: try loading the NVIDIA kernel module again.
NVRM: No NVIDIA graphics adapter probed!
NVRM: The NVIDIA probe routine was not called for 1 device(s).
NVRM: This can occur when a driver such as rivafb or rivatv was
NVRM: loaded and obtained ownership of the NVIDIA device(s).
NVRM: Try unloading the rivafb (and/or the rivatv) kernel module
NVRM: (or reconfigure your kernel without rivafb support), then
NVRM: try loading the NVIDIA kernel module again.
NVRM: No NVIDIA graphics adapter probed!
|
(???????????!!!!!!!!!!!!!!!???????????)
please relpy with details such as commands and code snips b/c I am just a casual linux user (it usualy works so I don't have to muck with it much!) |
|
Back to top |
|
|
fil1411 n00b
Joined: 13 Nov 2005 Posts: 4
|
Posted: Mon Nov 14, 2005 9:10 pm Post subject: |
|
|
Mariley,
I solved my problem by following the instructions of krasnik on this thread :
https://forums.gentoo.org/viewtopic-t-394817-start-0-postdays-0-postorder-asc-highlight-xorg+nvidia.html
name = 'XOrg can't load NVidia'
+ I recompiled my 2.6.13 kernel with option 'nVidia Framebuffer Support FB_NVIDIA' as a module (previously it was built-in)
this option is found in '/ device drivers / graphics support'
I hope it will help you too to make this d4mn module load as it should!
Ciao
F. |
|
Back to top |
|
|
|