Go Back   nV News Forums > Linux Support Forums > NVIDIA Linux

Newegg Daily Deals

Reply
 
Thread Tools
Old 12-31-02, 09:57 AM   #1
eskimo
Registered User
 
Join Date: Dec 2002
Posts: 6
Default ????????????

Hello all: I appologize for repeating myself here. I am having troubles id'ing the needed driver for my system. I currently have the nVIDIA gforce4 mx 420 running with win2000. I am trying to learn the linux OS RH 7.2. After running the xconfigurator it appears that the drivers need to be installed. I am not sure of which drivers I need from those listed on the nVIDIA site. I'll be downloading them via win2000 and placing on disk for keeping. I can mount the disk and ls to see the files. I just do not know where to send them, ect...... Be easy on me as I have been a windows person for quite a while and this LINUX/UNIX stuff is quite a steep learning curve. In advance.... THNKS.
__________________
M/B: K7VZA
PROCESS: AMD T-BIRD 1.0 ghz
O/S's: win2000 and R.H 7.2
H/D's: 30gig samsung and 40gig WD.
eskimo is offline   Reply With Quote
Old 01-01-03, 12:27 PM   #2
volt
 
Join Date: Jul 2002
Location: /dev/null
Posts: 1,556
Default

First of all, your topic is not very informative so let me change it.
(Please do not do that again)

Old topic: ??????????????????????
New topic: Installing NVIDIA drivers on Redhat 7.2

There are 'x' number of threads that explain step by step how to install nvidia drivers on Redhat operating system. Because of my generosity I will go over it just one more time

Download http://download.nvidia.com/XFree86_4....0-3123.tar.gz

Then download http://download.nvidia.com/XFree86_4....0-3123.tar.gz

Once you place them on your Linux system put them into /root
I assume you logged in as root if not type:

su
<enter password>

Where ever you have the drivers put them into /root directory:
(let's assume you have them in /mnt/windows/downloads)

cp /mnt/windows/downloads/NVIDIA_* /root
(this will copy both files because of the * wildcard)

Now "unzip" them. With those drivers it's called untaring
Being in /root:
NOTE: Linux is case sensitive!!!

tar zxfv NVIDIA_G*
(will create a folder named NVIDIA_GLX-1.0-3123 with all the files that we extracted)

Now type:

tar zxfv NVIDIA_k*
(this will create a folder named NVIDIA_kernel-1.0-3123 with all the files that we extracted)

Now type:

cd NVIDIA_G*
(you will end up in NVIDIA_GLX-1.0-3123)

Type: make;make install
(wait till it compiles)

Type: cd ..
(yes a space after cd)

Let's go into NVIDIA kernel directory now:

cd NVIDIA_k*

Type: make;make install
(wait till it compiles, you should get NO errors only 1 warning about GPL license)

Now all you need to do now is configure your XF86Config in /etc/X11

Search our forums for word XF86Config and you will find plenty of examples that people attached.
__________________
[b]Optimization guidelines by Koji Ashida of NVIDIA:[/b][list][*]Use fx12 instructions whenever possible[*]Use lowest pixel shader version[/list][url=http://developer.nvidia.com/docs/IO/10878/ChinaJoy2004_OptimizationAndTools.pdf]source[/url]

[size=1]The politics are invading the technology. We don't really like to mess with politics because that kind of adversarial relationship has nothing to do with pure technical operations and the technical specifications of what we like to play with, the hardware![/size]
volt is offline   Reply With Quote
Old 01-01-03, 11:32 PM   #3
eskimo
Registered User
 
Join Date: Dec 2002
Posts: 6
Thumbs up Thyanks

Volt: Thanks a whole lot, I'll be attempting your instructions for the installing of the tar files. I have seen some examples of the XFree86Config files.
__________________
M/B: K7VZA
PROCESS: AMD T-BIRD 1.0 ghz
O/S's: win2000 and R.H 7.2
H/D's: 30gig samsung and 40gig WD.
eskimo is offline   Reply With Quote
Old 01-02-03, 11:06 AM   #4
eskimo
Registered User
 
Join Date: Dec 2002
Posts: 6
Question errors

Hello volt: I downloaded the files and installed them in root. All went well until I did the make;make install on the kernel. It looks like every line had an error in it. Here is the last two lines of the make;make install of the kernel. (The screen passes by so fast and I am not sure how to retrieve the info that is off the top of the screen. )


/usr/include/linux/modversions #error to build against the currently running kenel
make: ***[NV.0] error!

If this is of any help here is my info on the kernel...
R.H7.2 kernel 2.4.7-10 on i686
__________________
M/B: K7VZA
PROCESS: AMD T-BIRD 1.0 ghz
O/S's: win2000 and R.H 7.2
H/D's: 30gig samsung and 40gig WD.
eskimo is offline   Reply With Quote
Old 01-04-03, 07:20 PM   #5
eskimo
Registered User
 
Join Date: Dec 2002
Posts: 6
Default

Hello volt: I cannot for the life of me figure why anyone would want to bang there head on the Linux wall. I have been doing just the same for the last few days. Reading post after post in a futile search to find an answer to my problem. It seems that I'll have to rebuild the entire o/s to get where I need to be. I've downloaded the latest kernel and attempted to install it (2.4.20.tar.gz) I have read some installing instructions at other sites but every sys, R.H, debian, ect.... is different. Do you know of a distro that will actually work with my video card without having to rebuild it? Frustration beyond need is just too much for this ecperience called linux?
__________________
M/B: K7VZA
PROCESS: AMD T-BIRD 1.0 ghz
O/S's: win2000 and R.H 7.2
H/D's: 30gig samsung and 40gig WD.
eskimo is offline   Reply With Quote
Old 01-04-03, 08:51 PM   #6
junkieclown
Helium Abuser
 
junkieclown's Avatar
 
Join Date: Nov 2002
Location: Summers in Rangoon, but mostly in my evil underground lair and Volcano laboratory. though
Posts: 45
Default

Eskimo- Sounds like your kernel and your kernel source are not matching version numbers.
# rpm -qa | grep kernel
will show all installed packages pertaining to the Linux kernel.
They need to match exactly down to the teeny version number, eg:
kernel-source-2.4.18-19.8.0
kernel-2.4.18-19.8.0
kernel-source should match the running version number you see by the command
'uname -a'
(unless you have compiled a kernel from source downloaded from kernel.org. Then your latest kernel-source rpm will not match the running kernel, which is OK)

It would be easy to have a mismatch if you had installed the kernel-source rpm only recently to build stuff like the nvidia module, and after having updated the kernel rpm one or more times. You might have grabbed the kernel-source rpm from your install disk, but the kernel package itself had been updated one or more times from Redhat updates repositories. That would produce a version mismatch.
In which case the thing to do would be to download a kernel-source rpm that matches the kernel you are currently running and replacing the old installed kernel source.

Another possible problem point: nvidia documentation these days says that the kernel module should be compiled with the same version of gcc that was used to compile your kernel.
Compiling the kernel using REdhat kernel-source rpm and the latest version of gcc for your version of redhat would eliminate that possible problem. I don't know that this is really a possibility -- I am just speculating that you could have a different version of gcc if Redhat had compiled the updated kernel rpms for your version of RH with a later version of gcc than is found on the rh install disks you have or the updates repository for your rh version. I don't know if that is something they'd ever do, but it's a logical possibility, and it's a possibility that can be removed completely from the question by compiling your own kernel. Just use the supplied default configuration after selecting processor family and whether to include SMP support. The other options (hundreds of them) are already setup the way redhat sets them up when making the kernel they shipped for your system. Then compile the nvidia module.

Then there's no doubt that your gcc, kernel and kernel source are all on the same page.
------------------------------------------------

The version of XFree86 on your distro almost certainly cannot automatically setup this card with your monitor. The card is newer than the utilities on your system used to configure it automatically. That means (I believe) you will need to run xf86config and create a config file by supplying details about your monitor. In that case, you will need the horizontal and vertical sync ranges that your monitor can perform within. Either that info will be found in the manual that came with your monitor or you'll need to find it on the manufacturer's website. Select a generic nvidia gforce4 identity for the video card.
There are very basic ranges that can selected but they'll result in poorer X performance than you can get by knowing what your monitor's sync values are.
After you have a working XF86Config you can modify it to use the nvidia binary driver.
It may be however, that even a generic gforce4 option is not listed when you run xf86config and look at the card database. I think it will still be possible to set up the card as a vesa driver.
But those are details that are far from certain.

In general with a new card like the gforce4 versions the newer your linux distro the better chance you'll have that the card is supported, or easily configurable.

Last edited by junkieclown; 01-04-03 at 09:03 PM.
junkieclown is offline   Reply With Quote
Old 01-05-03, 10:13 AM   #7
eskimo
Registered User
 
Join Date: Dec 2002
Posts: 6
Default

Hello junkieclown: Here is the current run down. I have downloaded the up to date kernel 2.4.20.tar.gz from kernel.org. I've cp'd it into root, tar'd it open and cd'd to do a make mrproper, when i do the make xconfig i get error messages starting at line 17.

17)apllication initialization failed: No display name and no $display environment var
18)error in startup script: invalid command name "button" while executing "button.ref" (file "scripts/kconfig.tk" line 51)
19 make: ***[xconfig] error1

I ran with what I had and did the make dep; make modules make modules_install; make install
I answered all the queries, most as highlited.

I cp arch/i686/boot/bzImage/boot/vmlinuz-2.4.20 here I got further errors and abandoned my attempt.
Can this be saved? What do I need to correct the xconfig errors?
Oh, one other Q. How do I save screen data to copy here for you all to look at? Thanks

Here I am again: I looked at the following and all is the same.
rpm -qa | grep kernel
kernel headers-2.4.7-10
kernel-2.4.7-10
and the uname -a is also 2.4.7-10

Ok, one other comment: I cd'd into the new kernel 2.4.20 and did a ls and then opened Kconfig.tk with vi and looked at line 51.
48)#
49)#Create a "reference" object to steal colors from
50)#
51)button .ref

there is a space between button and .ref, should i close up the space? or just comment out the line?

I hope to bang this out and get this up and running, learning a whole lot doing this.

Whoops: another lol... Just where do I go to check out the error reported on line 17 (above)??? Once again thanks....

__________________
M/B: K7VZA
PROCESS: AMD T-BIRD 1.0 ghz
O/S's: win2000 and R.H 7.2
H/D's: 30gig samsung and 40gig WD.

Last edited by eskimo; 01-05-03 at 10:56 AM.
eskimo is offline   Reply With Quote
Old 01-05-03, 11:24 AM   #8
junkieclown
Helium Abuser
 
junkieclown's Avatar
 
Join Date: Nov 2002
Location: Summers in Rangoon, but mostly in my evil underground lair and Volcano laboratory. though
Posts: 45
Default

Instead of make xconfig try 'make menuconfig' or 'make config'
Sometimes xconfig is unable to correctly parse the most recent kernel source.

Some terminal programs don't handle menuconfig's curses interface very well. If the terminal emulator you're using is showing garbage characters, you can use xterm which will run menuconfig cleanly.

You'll want to make sure that ext3 is compiled directly into the kernel, not as a module, (otherwise / will be mounted as ext2 ) and that agpgart is compiled as a module (i believe --that way it can't get in the way of possibly using nvidia's builtin agp support. I'm not sure if that really matters. ) Obviously, there are a gazillion possible compilation options and oopses with 2.4 kernel sources.
I think it's also better not to compile any framebuffer support, since as the nvidia docs note, this constitutes having two drivers touching the hardware directly.
DRM / DRI I guess can likewise be left out.
(Anybody who knows more definitely what bears on the nvidia modules operation feel free to speak up.)

Good Luck.

(P.S. I am not sure that having the most recent kernel is really necessary to getting the nvidia module compiled and inserted. Unless, there was some note on the nvidia driver package that said "works best with 2.4.18 kernels or later) Depending on your hardware, a later kernel like 2.4.20 _might_ support your motherboards' agp bus better but I don't think it should have any bearing on whether you're able to compile the Nvidia module or not.
It could be that many people who're having problems compiling the nvidia module would have better luck if they rpm -e the 'kernel-headers' and rpm -i installed the full 'kernel-source' package that corresponds to their running kernel. That's just a hunch, though. Of course, if there's any validity to that hunch, compiling a latest stable kernel from kernel.org sources would accomplish the same thing.
One more possible sticking point: rpm installed kernels put /module-info-2.4xxx and /System.map-2.4xxx files into the /boot directory. But they also tend to install a _symlink_ without any version numbers into /boot which points to the latest of each of these files ; eg: /boot/System.map -> /boot/System.map-2.4.18-19.8.0

I wonder if the symlink to the module-info might screw nvidia module compilation up. I always rearrange these symlinks when I install a new kernel from source. The symlink file 'module-info' I delete, leaving the versioned module-info files around but without a default one. Then I copy 'System.map ' from /usr/src/linux-2.4.20-ac2 (the kernel I intend to use) to /boot , with its version number appended to it, (System.map-2.4.20-ac2) and then make a create a symlink for it in /boot , System.map -> System.map-2.4.20-ac2 )
Maybe an Nvidia person could comment on the interaction, if any, between the Nvidia module build and these files in a redhat /boot directory --or whether it's necessary to have full kernel-sources installed versus just kernel-headers.)

--------------------------

Checking around this forum, it's become clear to me that you must have full sources installed. ( that is, a kernel-source rpm that matches your rpm installed kernel exactly - or a kernel source tree downloaded from kernel.org from which you'll build a custom kernel)
Redhat's 'kernel-headers' won't cut it. rpm -e kernel-headers-xxxxxx and install the kernel-source rpm that matches your updated kernel rpm, and you should be able to build the nvidia module easily.

Last edited by junkieclown; 01-06-03 at 09:33 PM.
junkieclown is offline   Reply With Quote

Reply


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


Similar Threads
Thread Thread Starter Forum Replies Last Post
monitor not recognised after installing nvidia drivers Horaciusl NVIDIA Linux 3 08-10-12 10:59 PM
Need Help Installing NVIDIA Tesla M2070Q in Linux RHEL5 Ferianto85 NVIDIA Linux 0 05-18-12 08:35 PM
Radeon 9700 not all that? sancheuz Other Desktop Graphics Cards 200 10-12-02 09:31 PM
NVidia Drivers HowTo for RedHat 8.0 needed eduardp NVIDIA Linux 10 10-04-02 03:59 AM
Redhat 7.3 + Nvidia drivers 3123 = troubles e-r-a-n NVIDIA Linux 4 09-17-02 12:26 PM

All times are GMT -5. The time now is 10:46 AM.


Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright 1998 - 2014, nV News.