Im having nightmares installing nvidia driver for FX 5500 in centos 5.3, i have checked plenty of threads on the forum but no use. i downloaded the driver from nvidia site 173.15.X

Problem No1:

following sequence of commands return with No precompiled kernel source found,

#init 3
#yum install kernel-devel (and yes im not running PAE kernel)
#chmod +X NVI...
#sh NVI...

now the installer returns with error, no precompiled kernel source found & exits. i have tried all the recommended parameters to include kernel source address. but it stays the same,

Then i tried

#init 3
# yum --enablerepo=rpmforge install nvidia-x11-drv
# reboot

everything went well the drivers were installed i rebooted the system but now the x server is not starting and returns with a Fatal error i tried #nvidia-xconfig but no use.

I can guess where it went wrong, but without going into all the details, I want you to get going with the installation of the Nvidia driver. Please follow these steps:

(1) First, uninstall the nvidia-x11-drv package you installed from rpmforge.


(2) Set up the ELRepo repository by following the instructions at http://elrepo.org. Just run:


(3) Install the driver from ELRepo by running:


(4) Reboot.

That should do it.

With the last 3 or 4 kernels I'm running the proprietary nVidia drivers from their site without any issues. (currently 2.6.18-164.el5 )
You need to make sure you have kernel, kernel-dev and kernel-headers all installed for your running kernel version.

The nVidia installer will not find a precompiled package but compile one on the fly.

# yum list installed | grep kernel
kernel.x86_64 2.6.18-128.1.16.el5 installed
kernel.x86_64 2.6.18-128.2.1.el5 installed
kernel.x86_64 2.6.18-128.4.1.el5 installed
kernel.x86_64 2.6.18-128.7.1.el5 installed
kernel.x86_64 2.6.18-164.el5 installed
kernel-devel.x86_64 2.6.18-128.1.16.el5 installed
kernel-devel.x86_64 2.6.18-128.2.1.el5 installed
kernel-devel.x86_64 2.6.18-128.4.1.el5 installed
kernel-devel.x86_64 2.6.18-128.7.1.el5 installed
kernel-devel.x86_64 2.6.18-164.el5 installed
kernel-headers.x86_64 2.6.18-164.el5 installed

After a kernel update you'll need to rerun the installer.

Quote:
This is exactly why the kmod package shines. It is kernel version independent and, unlike dkms, does not require development packages.

toracat,
Thanks, but I'm aware of the benefits of kmod packages. It's the reason I threw out dkms and dkms-fuse in favor of kmod-fuse.
But I wasn't aware that the nvidia-x11-drv package is built from the nvidia sources, as I thought these were proprietary.
Was I misinformed?

I've been running with the native CentOS driver for my card a long time, as the nvidia driver had issues with the non-graphics consoles. But since those were fixed I use the nVidia proprietary driver without issues. Recompiling the driver after an update is no more a hassle than rerunning vmware-config.pl, which is also required after an update. But it's all smooth and painless.

Having said that, I will use kmod packages when I can, for the very reason you mention, but I need to be sure that quality is equal or better then the original (like the kmod-r8168.x86_64 ElRepo driver is (much) better than the CentOS one, thanks to upstream), and I tend to avoid 'testing' repos.

Having explained my preferences, is the kmod driver stable and as good as the 'real thing'? I only needed that for the hardware video acceleration, which is dead slow with the CentOS native driver. And as I just found out, with the latest nVidia one, my notebook no longer locks up when awaking from standby (but maybe kernel 164 has something to do with that as well).

Quote:

No, that's correct. All the x11 libs are proprietary binaries that repositories may repackage and redistribute. There's also a kernel driver module that needs to be built against each kernel, hence the need to either rerun the proprietary nvidia installer, use a packaging method such as dkms that rebuilds against each kernel, or a kABI-tracking kmod that allows one build of the module to be weak linked against all compatible kernels.

Quote:

The kmod package is based on the nvidia proprietary driver. It installs the binary X11 libs in the same manner as the proprietary installer, and also provides a kABI-tracking kmod for the kernel module. Assuming we've done a good job with the packaging, there shouldn't be any difference between the proprietary nvidia installer and the kmod other than the convenience of a one time install through yum (and the ability to pick up driver updates with yum).

The main reason it's still in the testing repo is that we've not had reports from many people on how well it works. It's been extensively tested by 3 or 4 people in elrepo, but we'd rather have real world feedback too from a wider set of users, especially as it's likely to be widely used by many people.

guys thanks for your early response,

Kindly pardon my ignorance, this is my first interaction with CentOS,

Okay toracat, as per your recommendations i did removed the previously installed driver and added Elrepo-testing and did the yum install in #init 3, but again no use xserver failed to start and reverted back to old configuration i can see the Nvidia utility in the systems menu, i did tried #nvidia-xconfig.

Couple of things id like to share with you guys, im doing a fresh installation and added the following packages when installing

-Server
-development
-xserver development

do i need to add anymore packages do i need to update the kernel first.

and r_hartman i did checked #yum list installed | grep kernel

but i have listing with file names ending in i686 and i386.

as im away from my machine but ill post the exact outputs ASAp.

Ok guys i hope i get some answers from you, i may be new to this arena but im willing to see how deep the rabbit hole goes.

I think their is some issue with the version of the driver the verson elrepo is installing is NVIDIA-185, but when i go to Nvidia.com and search for FX 5500 i get a driver version NVIDIA-173.

so i downloaded the driver , by the way i updated my kernel before going for the driver installation, and installed it in #init 3, every thing went well, i rebooted the PC and guess what this time i have no display at all,nothing, monitor stopped responding with amber light i could not see anything so i Hard rebooted the PC and again it reverted back to the old crappy nv driver.

I previously had no issues installing the driver in suse 10.3.

Sorry, I should have noticed earlier. As you've found out, the FX5xxx series requires the legacy release and the current version is 173.14.18. ELRepo does not have any legacy releases at the moment (being planned).

However, I would think that the Nvidia's driver you downloaded should work. Could you look at /var/log/Xorg.0.log and see if there is any hint? Is there any line that says Error ?

(EE) Failed to initialize GLX extension (Compatible NVIDIA X driver not found)

This is the failed line from /var/log/xorg.0.log, does that mean i am using a wrong driver.

Please tell us exactly what file you downloaded from the Nvidia site. When you ran the downloaded file, did it run through without any error? Also show us the output from:

ls -l /usr/lib/xorg/modules/extensions

(If your system is running 64-bit, replace the "lib" with "lib64")

Appologies for the late response..

No their were no errors during the installations..

The drive im trying to install is.

NVIDIA-Linux-x86-173.14.20-pkg1.run

the output of the command.

[root@dhcppc0 ~]# ls -l /usr/lib/xorg/modules/extensions
total 1908
-rwxr-xr-x 1 root root 16192 Jan 21 2009 libdbe.so
-rwxr-xr-x 1 root root 30832 Jan 21 2009 libdri.so
-rwxr-xr-x 1 root root 134176 Jan 21 2009 libextmod.so
lrwxrwxrwx 1 root root 19 Sep 27 18:43 libglx.so -> libglx.so.173.14.20
-rwxr-xr-x 1 root root 1269220 Sep 27 18:43 libglx.so.173.14.20
-rwxr-xr-x 1 root root 24900 Jan 21 2009 librecord.so
-rwxr-xr-x 1 root root 394388 Mar 25 2009 libvnc.so
-rwxr-xr-x 1 root root 36928 Jan 21 2009 libxtrap.so
[root@dhcppc0 ~]#

Not sure whether I should still post this here, but I just upgraded my notebook to 5.4 (thanks!) and decided to uninstall the native nvidia driver for my geForce 9800 card and install the one from elrepo-testing. I know the kernel wasn't updated but I expect one soon

I did an 'init 3' first, then 'nvidia-installer --uninstall' then 'yum --disablerepo=\* --enablerepo=elrepo-testing install kmod-nvidia.x86_64 nvidia-x11-drv.x86_64' and finally an 'init 5'. So far haven't noticed any differences with the one from nvidia.

With the native nvidia one I always had to enter runlevel 3 to be able to install; not sure if that limitation is applicable to the elrepo one. Hope not, as it would sort of negate the benefits of being able to just yum update.

An important consideration to take this step is that it was mentioned here that this driver was still in elrepo-testing because of little feedback on it, but should be stable. I hope to be able to help moving it out of testing by using is and reporting back on my experience. Please let me know whether this thread would be the proper place for that.

For completeness here's my lspci output:
bash(1):0# lspci
00:00.0 Host bridge: Intel Corporation Mobile 4 Series Chipset Memory Controller Hub (rev 07)
00:01.0 PCI bridge: Intel Corporation Mobile 4 Series Chipset PCI Express Graphics Port (rev 07)
00:1a.0 USB Controller: Intel Corporation 82801I (ICH9 Family) USB UHCI Controller #4 (rev 03)
00:1a.1 USB Controller: Intel Corporation 82801I (ICH9 Family) USB UHCI Controller #5 (rev 03)
00:1a.2 USB Controller: Intel Corporation 82801I (ICH9 Family) USB UHCI Controller #6 (rev 03)
00:1a.7 USB Controller: Intel Corporation 82801I (ICH9 Family) USB2 EHCI Controller #2 (rev 03)
00:1b.0 Audio device: Intel Corporation 82801I (ICH9 Family) HD Audio Controller (rev 03)
00:1c.0 PCI bridge: Intel Corporation 82801I (ICH9 Family) PCI Express Port 1 (rev 03)
00:1c.1 PCI bridge: Intel Corporation 82801I (ICH9 Family) PCI Express Port 2 (rev 03)
00:1c.2 PCI bridge: Intel Corporation 82801I (ICH9 Family) PCI Express Port 3 (rev 03)
00:1c.4 PCI bridge: Intel Corporation 82801I (ICH9 Family) PCI Express Port 5 (rev 03)
00:1c.5 PCI bridge: Intel Corporation 82801I (ICH9 Family) PCI Express Port 6 (rev 03)
00:1d.0 USB Controller: Intel Corporation 82801I (ICH9 Family) USB UHCI Controller #1 (rev 03)
00:1d.1 USB Controller: Intel Corporation 82801I (ICH9 Family) USB UHCI Controller #2 (rev 03)
00:1d.2 USB Controller: Intel Corporation 82801I (ICH9 Family) USB UHCI Controller #3 (rev 03)
00:1d.7 USB Controller: Intel Corporation 82801I (ICH9 Family) USB2 EHCI Controller #1 (rev 03)
00:1e.0 PCI bridge: Intel Corporation 82801 Mobile PCI Bridge (rev 93)
00:1f.0 ISA bridge: Intel Corporation ICH9M LPC Interface Controller (rev 03)
00:1f.2 SATA controller: Intel Corporation ICH9M/M-E SATA AHCI Controller (rev 03)
00:1f.3 SMBus: Intel Corporation 82801I (ICH9 Family) SMBus Controller (rev 03)
01:00.0 VGA compatible controller: nVidia Corporation G94 [GeForce 9800M GTS] (rev a1)
02:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168B PCI Express Gigabit Ethernet controller (rev 02)
04:00.0 FireWire (IEEE 1394): JMicron Technology Corp. IEEE 1394 Host Controller
04:00.1 System peripheral: JMicron Technology Corp. SD/MMC Host Controller
04:00.2 SD Host controller: JMicron Technology Corp. Standard SD Host Controller
04:00.3 System peripheral: JMicron Technology Corp. MS Host Controller
07:00.0 Network controller: Intel Corporation PRO/Wireless 5300 AGN [Shiloh] Network Connection

and lsusb:
bash(1):0# lsusb
Bus 007 Device 001: ID 0000:0000
Bus 006 Device 001: ID 0000:0000
Bus 003 Device 001: ID 0000:0000
Bus 003 Device 002: ID 046d:c521 Logitech, Inc. MX620 Laser Cordless Mouse
Bus 004 Device 001: ID 0000:0000
Bus 005 Device 001: ID 0000:0000
Bus 005 Device 002: ID 147e:2016 Upek Biometric Touchchip/Touchstrip Fingerprint Sensor
Bus 002 Device 001: ID 0000:0000
Bus 008 Device 001: ID 0000:0000
Bus 001 Device 001: ID 0000:0000
Bus 001 Device 003: ID 1164:1f08 YUAN High-Tech Development Co., Ltd
Bus 001 Device 004: ID 5986:0303 Acer, Inc

and uname -a:
bash(1):0# uname -a
Linux hacnb08 2.6.18-164.el5 #1 SMP Thu Sep 3 03:28:30 EDT 2009 x86_64 x86_64 x86_64 GNU/Linux

Quote:

Thanks for reporting your successful installation of the nvidia packages from ELRepo. I believe it is now time to move it out of testing. Ned?

Hi All,

I'd like to add that I also switched a few weeks ago from the NVidia supplied version to the elrepo-testing version and have had no problems with it. I haven't tried it across a kernel update yet, but I am assuming that you can just keep the machine in run-level 5 when you reboot into a new kernel and it will work.
Cheers,

Quote:
Yes, it should work that way. Please keep us posted.

Quote:

What do other users feel about this? Any objections to moving it from testing to the main elrepo respository?

I just switched from the dkms nvidia package (rpmforge) to the kmod/driver packages provided by ELrepo. Seems to work properly, but I got a strange warning during installation:
something about nvidia.ko not being found in a ...../weak-updates/...... directory under the .128 (!) kernel tree (have to cite from memory). The system is running a fully updated CentOS 5.4, with official build 164 kernel.

Any idea why the install script is looking for the nvidia module in the older kernel? May this have to do with the (removed) dkms package? (I don't have a system available for testing which has not yet "seen" the dkms version...)

Oliver

Oliver,

Please post the output returned by the following --

Alan,

ok, here we go...

(~): uname -rmi
2.6.18-164.el5 x86_64 x86_64

(~): rpm -qa kernel\* | sort
kernel-2.6.18-128.el5.x86_64
kernel-2.6.18-164.el5.x86_64
kernel-devel-2.6.18-164.el5.x86_64
kernel-headers-2.6.18-164.el5.x86_64

(~): find /lib/modules -name nvidia.ko
/lib/modules/2.6.18-164.el5/weak-updates/nvidia/nvidia.ko
/lib/modules/2.6.18-128.el5/weak-updates/nvidia.ko
/lib/modules/2.6.18-128.el5/extra/nvidia/nvidia.ko

(~): ls -l $(find /lib/modules -name nvidia.ko)
-rw-r--r-- 1 root root 13363717 Sep 13 02:42 /lib/modules/2.6.18-128.el5/extra/nvidia/nvidia.ko
lrwxrwxrwx 1 root root 43 Oct 26 11:16 /lib/modules/2.6.18-128.el5/weak-updates/nvidia.ko -> /lib/modules/2.6.18-164.el5/extra/nvidia.ko
lrwxrwxrwx 1 root root 50 Oct 26 11:24 /lib/modules/2.6.18-164.el5/weak-updates/nvidia/nvidia.ko -> /lib/modules/2.6.18-128.el5/extra/nvidia/nvidia.ko

Ooooops.... These cross-links between modules of different kernels look VERY suspicious They got created today, during/after installation of the kmod package!

Hope this helps
Oliver

EDIT: Just checked with another system, which has been running with the dkms version all the time. It shows a similar cross-referencing. Maybe this is not as unusual as it appears at first sight?