LinuxQuestions.org
Latest LQ Deal: Complete CCNA, CCNP & Red Hat Certification Training Bundle
Go Back   LinuxQuestions.org > Forums > Linux Forums > Linux - Newbie
User Name
Password
Linux - Newbie This Linux forum is for members that are new to Linux.
Just starting out and have a question? If it is not in the man pages or the how-to's this is the place!

Notices


Reply
  Search this Thread
Old 04-06-2014, 11:10 PM   #1
anthony000
LQ Newbie
 
Registered: Apr 2014
Location: RI
Distribution: RHEL 7.1
Posts: 15

Rep: Reputation: Disabled
Thumbs up Instructions on how to make nVidia GTX645 graphics card run using nVidia in RHEL 6.5


These are procedures for installing in Red Hat Enterprise Linux 6.5 the ElRepo RPM package 'kmod-nvidia', and for sucessfully using nVidia GTX645 OEM graphics card/GPU with nVidia's proprietary software . These procedures apply to RHEL 6.5 64-bit operating system. I'm sharing these instructions because they worked for me on my PC, and I realize the info out there is scattered and difficult to synthesize. I would expect these instructions should work for any nVidia card because all that's needed is for the appropriate kmod package from ElRepo repository - the instructions describe how to locate the proper package.

Summary of critical steps: 1) Install kmod-nvidia, 2) Uninstall xorg-x11-glamor, 3) Blacklist Nouveau (the default linux driver for the nVidia card) and recreate initramfs (initrd) w/o the Nouveau drivers, 4) Reboot. See procedures herein for details and precautions.

The details...

1. SOFTWARE PREPARATION: From within System>Administration>Add/Remove Software, check/install the software packages for the following: kernel-devel, kernel-headers, dkms, acpid, binutils, dracut (or mkinitrd, or update-initramfs) – dkms package is needed to enable the automatic recreating of initramfs w/o the nouveau driver, simulating Step 8 below after the next kernel upgrade). To enable video acceleration support for the player: vdpauinfo, libva-vdpau-driver, libva-utils, libvdpau, and possibly libvdpau-devel, libvdpau-docs (I did not include these last two in my installation). Uninstall the exiting package xorg-x11-glamor (see Step 6 below).

Note: If doing installations from the command prompt, you can use
]# yum intall 'package-name' instead, which may be useful for seeing the progress of the installation and any special notes.


2. EL REPO: Install the RHEL repository ElRepo from: http://elrepo.org/tiki/tiki-index.php


3. INSTALL NVIDIA-DETECT: Install the package nvidia-detect from within System>Administration>Add/Remove Software.


4. IDENTIFY NVIDIA DEVICE: From within the terminal, enter the command nvidia-detect.
Example:
[anthony@hp810135qe ~]$ nvidia-detect
Output is...
Probing for supported NVIDIA devices...
[10de:11c4] NVIDIA Corporation GK106 [GeForce GTX 645 OEM]
This device requires the current 331.49 NVIDIA driver kmod-nvidia


5. PRECAUTION: As a precaution for assuring access to the system files when rebooting, go into /etc/inittab and change the default runlevel (id:5:initdefault: ) to 3 (prevents X11 from starting but allows network devices to work).

5.1 Before doing so, save the original as follows
]# cp /etc/inittab /etc/inittab.original

5.2 Edit the inittab to runlevel 3 as follows
]# vi /etc/inittab
In the last line of the inittab file (should look like this id:5:initdefault: ) change the 5 to a 3 from within vi by using the following key-commands:
press <Insert>
scroll down and delete 5 and in its place enter a 3
press <Esc>
press and hold <Shift> and while holding press the <Z> key twice to save and exit vi.
Notes for later during boot-up: When booting up in runlevel 3 (from changing the 5 to a 3 as in step 5.2), you'll get prompts for login (username) and password. You can start X at this point by entering 'su –' (minus the tick marks) at the prompt and then entering the root password, then entering init 5 at the root prompt # will bring you into the graphical desktop. After rebooting (Step 9 below) and if everything is ok, change the run-level in /etc/inittab back to 5 (re-enables X11).


6. INSTALL KMOD-NVIDIA: From within System>Administration>Add/Remove Software, install the appropriate kmod-nvidia package. Follow the installation instructions from within the installation dialogue prompts. When done, check /etc/X11/xorg.conf.d/xorg.conf for the “File” Section (see step 12 below). Note that for RHEL 6 there is no initial xorg.conf file and it is nvidia that will create one for its own use (nvidia uses nvidia-xconfig). If there is/may-be a conflict with 'glamor', see Step 13 below for removing xorg-x11-glamor and its dependencies from the previously installed packages. The recommendation (by nVidia in its notes during installation from within a terminal, i.e., from command prompt using 'yum install kmod-nvidia') is to uninstall xorg-x11-glamor right way – as commented by others it's of no use with nvidia and why later deal with bugs that can't be properly pinned down (see http://lists.elrepo.org/pipermail/el...er/002062.html) .


7. BLACKLIST & DISABLE NOUVEAU (which would conflict with kmod-nvidia):

7.1 To blacklist and disable the nouveau driver after the root device is mounted (i.e., /)

7.1.1 Create file /etc/modprobe.d/disable-nouveau.conf

7.1.2 Inside disable-nouveau.conf add the lines:
blacklist nouveau
options nouveau modeset=0

Note: The first line will prevent Nouveau's kernel module from loading automatically at boot, and it will not prevent the X server from loading the kernel module. The second line will prevent Nouveau from doing a kernel modeset. Without the kernel modeset, it is possible to unload Nouveau's kernel module, in the event it is accidentally or intentionally loaded. Blacklisting Nouveau per the above will only prevent it from being loaded automatically at boot. If an X server is started as part of the normal boot process, and that X server uses the Nouveau X driver, then the Nouveau kernel module will still be loaded. Should this happen, you will be able to unload Nouveau with 'modprobe -r nouveau' after stopping the X server (entering init 3 at the terminal root # prompt), as long as care was taken to prevent it from doing a kernel modeset. However, it is probably better to just make sure that X does not load Nouveau in the first place. See Step 11 below.

7.1.3 Another place these lines can be 'additionally' added (although it may get overridden at the next kernel update) is /etc/modprobe.d/blacklist.conf

7.1.4 NOTE: During my installation, nVidia automatically created a file called blacklist-nouveau.conf and placed it in /etc/modprobe.d/. It only included 'blacklist nouveau' and I manually edited it to add 'options nouveau modeset=0'.


7.2 OPTIONAL BELTS-N-SUSPENDERS - To blacklist and disable the nouveau driver during boot-up time: This step is having belts-n-suspenders with Step 8 below (one or the other should work; having both should provide extra assurance, especially since Step 7.1 can be undone by the next kernel upgrade, requiring manual intervention again):

7.2.1 Use vi from within a terminal to edit grub.conf. In my system (a UEFI system) this file is located in /boot/efi/EFI/redhat/. In BIOS-mode systems (i.e., Legacy mode), the this file may likely be found in /boot/grub/.

7.2.2 Identify the default kernel used to boot the system. Each kernel is specified in the grub.conf file with a group of lines that begins with title. The default kernel is specified by the default parameter near the start of the file. A value of 0 refers to the kernel described in the first group of lines, a value of 1 refers to the kernel described in the second group, and higher values refer to subsequent kernels in turn.

7.2.3 Edit the kernel line of the group to include the options (usually located immediately after the 'quiet' parameter): rdblacklist=nouveau nouveau.modeset=0 (if initrd does not support rdblacklist)

Note: Note that nouveau.modeset=0 will prevent a kernel modeset, but it may not prevent Nouveau from being loaded, so rebuilding the initrd (Step 7 below) or using rdblacklist may be more effective than using nouveau.modeset=0. What if the intial ramdisk (initrd, initramfs) contains Nouveau? This is the reason for rebuilding initramfs (Step 8 below). Rebuilding the initrd (initramfs) should pick up the module loader configuration files (/etc/modprobe.d/*.conf), including any which disable Nouveau, i.e., disable-nouveau.conf of Step 7.1 above. I used rdblacklist=nouveau only (no nouveau.modeset=0) and had no problem booting up.


8. RECREATE INITRD w/o NOUVEAU. Create a new initrd that does not have the nouveau drivers (the existing initrd still has the embedded nouveau drivers if this step is not performed):

8.1 Backup the existing initramfs nouveau image, as root from the terminal prompt
]# mv /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r).img.nouveau.bak

8.2 Create the new initramfs image, as root from the terminal prompt
]# dracut -v /boot/initramfs-$(uname -r).img $(uname -r)

Note: the -v option in step 8.2 shows the progress of the installation so that you can see all that's transpiring. If you didn't want to save a backup (i.e., step 8.1) then you would need to add the -f option, which ensures (f=forces) a complete overwrite of the original.


9. REBOOT

9.1 When rebooting for the first time it will go into command-line mode because of the init 3 setting in /etc/inittab (from Step 5). Enter your login name (user name) at the prompt. Enter your user name password.

9.2 If step 9.1 was successful and you have a command prompt, enter 'su -' minus the tick-marks and enter the root password.

9.3 At the root prompt ]# enter 'init 5' minus the tick-marks to start the graphical display (X11).

9.4 You should be at your normal graphical login screen at this time whereby you can login to your dektop as usual.

9.5 Go back into /etc/inittab and re-edit to init 5 (runlevel 5) as describe in Step 5.

9.6 Enjoy!



THE REMAINING ITEMS 10 thru 13 BELOW ARE FOR BACKGROUND AND ADDITIONAL INFORMATION ONLY.


10. RUNLEVELS & INIT

Tips: Runlevels may be entered from the command line as follows
]# init 3 (changes the current run level to X11 not running; init 3 is also full multiuser mode), or
]# init 5 (changes the run level to X11 running)
init 1 is single user mode, if needing to go into rescue mode (see RHEL 6 Chapter 36 Basic System Recovery in the Installation Guide , https://access.redhat.com/site/docum...#ap-rescuemode )


11. As a caution – How to prevent the X server from loading Nouveau:

Blacklisting Nouveau will only prevent it from being loaded automatically at boot. If an X server is started as part of the normal boot process, and that X server uses the Nouveau X driver, then the Nouveau kernel module will still be loaded. Should this happen, you will be able to unload Nouveau with `modprobe -r nouveau` after stopping the X server, as long as you have taken care to prevent it from doing a kernel modeset; however, it is probably better to just make sure that X does not load Nouveau in the first place.
If your system is not configured to start an X server at boot, then you can simply run the NVIDIA driver installer after rebooting. Otherwise, the easiest thing to do is to edit your X server's configuration file so that your X server uses a non-modesetting driver that is compatible with your card, such as the vesa driver. You can then stop X and and install the driver as usual. Please consult your X server's documentation to determine where your X server configuration file is located.
This tip is from http://us.download.nvidia.com/XFree8...nproblems.html

12. Check /etc/X11/xorg.conf.d/xorg.conf for the “File” Section. This tip is old and may not apply; it is found at http://elrepo.org/bugs/view.php?id=98
Look closely at the xorg.conf file. Check the nvidia docs in /usr/share/docs/nvidia-x11-drv-%{version} - there is lots of useful information regarding xorg.conf settings (e.g, see section 6B and the appendix).

Our package requires (and adds) the following in xorg.conf:

Section "Files"
********ModulePath "/usr/lib64/xorg/modules/extensions/nvidia"
********ModulePath "/usr/lib64/xorg/modules"
EndSection

Adjust the paths as appropriate for 32-bit systems. If you are creating your own xorg.conf, either ensure these are present or ensure there is at least a "Files" section and allow our driver to add the ModulePath entries. Either way, these entries were *missing* from the xorg.conf that you posted.


13. glamor provides xorg-x11 acceleration using the OpenGL driver. This tip applies to ElRepo nVidia release 304.xx for RHEL 6.5 and may not apply to 331.49 releases, but here it is anyway (it can be found at http://lists.elrepo.org/pipermail/el...er/002058.html )

# yum remove xorg-x11-glamor (or remove from within Add/Remove Software)

Don't panic, this will also remove “xorg-x11-drivers” which is only a meta-package requiring all the drivers). Also removed may be xorg-x11-drv-ati. See http://lists.elrepo.org/pipermail/el...er/002061.html
 
Old 04-07-2014, 03:17 AM   #2
Drakeo
Senior Member
 
Registered: Jan 2008
Location: Urbana IL
Distribution: Slackware, Slacko,
Posts: 3,092
Blog Entries: 3

Rep: Reputation: 324Reputation: 324Reputation: 324Reputation: 324
wonder where I have seen this before. welcome to LQ.
 
Old 04-07-2014, 09:24 AM   #3
anthony000
LQ Newbie
 
Registered: Apr 2014
Location: RI
Distribution: RHEL 7.1
Posts: 15

Original Poster
Rep: Reputation: Disabled
Thanks Drakeo. The next question to answer is: If DKMS is not active or being used, what do you do when the kernel upgrades? The answer depends on whether nVidia's ./*.run installer is used, or whether El Repo's kmod-nvidia installer is used. In either case, without DKMS being setup (i.e., not having the nVidia module registered with DKMS), then it seems a remove-n-reinstall of the module may likley be needed, and this would require its own process to ensure your system's display doesn't crash or become inoperable. So, to be continued...
 
Old 04-07-2014, 11:55 AM   #4
John VV
LQ Muse
 
Registered: Aug 2005
Location: A2 area Mi.
Posts: 17,015

Rep: Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454
this is redhat ( and NOT cent)
so you do have access to the redhat Knowledge base

https://access.redhat.com/search/browse/search/
the full instructions are in there


but you will have to login to redhat first to read the search answer

use the kmod-nvidia


BUT if you MUST use the nvidia,run driver
you WILL!!! have to manually REINSTALL it for EVERY kernel,xorg,mesa update .

you will need to FIRST install
gcc and autotools
and the kernel source
Code:
su -
yum groupinstall "Development Tools"
yum install kernel-devel

Last edited by John VV; 04-07-2014 at 11:57 AM.
 
Old 04-07-2014, 03:11 PM   #5
anthony000
LQ Newbie
 
Registered: Apr 2014
Location: RI
Distribution: RHEL 7.1
Posts: 15

Original Poster
Rep: Reputation: Disabled
Thanks John.

Based on what I've read, and as reinforced by your advisement, I think I'll stick with kmod-nvidia. I do have an account with Red Hat and therefore I have access to their knowledge-base. I initially had anticipated being able to use dkms to auto-install the kmod-nvidia modules for future kernel upgrades, but it seems this is not possible using El REpo's kmod-nvidia? Before I ask too many more questions I'll check out the RHEL docs as you suggested. Thanks again.

Anthony
 
Old 04-07-2014, 03:58 PM   #6
John VV
LQ Muse
 
Registered: Aug 2005
Location: A2 area Mi.
Posts: 17,015

Rep: Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454
i use the .run driver , but that is out of habit .I have used it for YEARS

back in the fedora4,5,6 days the kernel would be updated ,then 3 to 7 days later the nvidia driver would be

this would cause problems

downgrade the kernel until the matching kmod-nvidia.rpm was released
or
use a text ONLY install for a week until the kmod was ready

that forced most to use the nvidia binary blob


so fast forward many years
it is now habit

some say the NVIDIA.run is a pain to use , but after years of using it , it really is not

you just have to manually reinstall it ( takes about 1 min. ) after some updates

and decide IF you need the NVIDIA version of "Gl.h" or mesa's DIFFERENT version of "Gl.h"
the two are incompatible

the .run uses the nvidia gl.h ( and MATCHING!!!! *.so "libnvidia.so")
the kmod uses mesa's gl.h

most of the time this dose NOT matter
but
but if it dose then it is a big deal .

Last edited by John VV; 04-07-2014 at 04:02 PM.
 
Old 04-07-2014, 04:30 PM   #7
anthony000
LQ Newbie
 
Registered: Apr 2014
Location: RI
Distribution: RHEL 7.1
Posts: 15

Original Poster
Rep: Reputation: Disabled
Hi John

I checked Red Hat's knowledge-base and they don't provide support for nvidia and the community involvement in the forums didn't address my question. It looks like I'll have to manually update El Repo's kmod-nvidia at each kernel update. Here's a draft procedure I'll be trying out (I developed this using my judgement based on what I think is logical according to the procedure I repeated earlier in the post):

POST-NVIDIA KERNEL UPGRADES

After nVidia was previously installed (nouveau was disabled), subsequent kernel updates would require another process
for installing the kmod-nvidia module in the new kernel.

DRAFT PROCEDURE: When the kernel updates, do the following ASAP:

1. Go into a Terminal

2. ]$ sudo vi /etc/inittab

3. While in inittab, change the 5 runlevel to a 3

4. Use File Manager to inspect the latest kernel in /boot and note the version exactly,
which will be used in Step 15.
For example: initramfs-2.6.32-431.11.2.el6.x86_64.img

5. ]$ locate nvidia (note the output and whether nvidia got picked up by the new
kernel)

6. ]$ nvidia-detect (note the output and what version of kmod-nvidia is required)

7. Go into System>Administration>Add/Remove Software, and identify the two existing
kmod-nvidia packages that were previously installed. In my case, one would be
kmod-nvidia-331.49 and the other nvidia-x11-drv-331.49 , or their latest installed versions.

8. ]$ init 3

9. ]$ sudo yum remove kmod-nvidia-331.49 (or whatever it is from step 6)

10. ]$ sudo yum remove nvidia-x11-drv-331.49 (or whatever it is from step 6)

11. ]$ sudo yum install new-kmod-nvidia

12. ]$ sudo yum install new-kmod-x11-drv (if not already installed as a dependency)

13. ]$ sudo vi /etc/modprobe.d/blacklist-nouveau.conf
(ensure the two lines are there: blacklist nouveau ; options nouveau modeset=0)

14. ]$ sudo vi /boot/efi/EFI/redhat/grub.conf (add rdblacklist=nouveau after 'quiet'
in the default kernel stanza)

15. ]$ sudo dracut -fv /boot/initramfs-2.6.32-431.11.2.el6.x86_64.img initramfs-2.6.32-431.11.2.el6.x86_64.img
NOT SURE ABOUT THIS STEP - newkernelversion substituted for $(uname -r).img
Get kernel name from Step 4. The -f option is to force an overwrite to initramfs (-v for verbose to see what's going on).

16. ]$ exit

17. REBOOT to init 3 and then after login and password enter ]$ sudo init 5

I'LL HAVE TO TEST THIS EVENTUALLY.
 
Old 04-07-2014, 07:37 PM   #8
John VV
LQ Muse
 
Registered: Aug 2005
Location: A2 area Mi.
Posts: 17,015

Rep: Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454
Quote:

2. ]$ sudo vi /etc/inittab

3. While in inittab, change the 5 runlevel to a 3
no need to do that

during the 3 second count down during the early boot
hit < e > for edit on the current of the 3 bootlines
then use the arrow button to move down one
to the kernel line

add a blank space then a < 3 >
and hit < enter>
the end of the boot line should look something like
Code:
crashkernel=128M rhgb quiet 3
that 3 passes boot into text ONLY mode

so no need to manually edit inittab


also unless YOU have already set up " sudo"
NOT configured BY DEFAULT and NOT used by default
use "su -"
then the root password ,
just once at the start

and by booting into TEXT only mode you CAN LOGIN as root
( logging in as ROOT in text only mode to install the driver IS a good idea and recommended )

also
step 15
using the dracut command
that ONLY!!!!!!!!!
repet ONLY!!!!
needs to be done ONE TIME ONLY !!!!!!
and only if moving from the Nouveau driver to the nvidia driver

and ONLY the VERY FIRST TIME

now if you are going to do that much work ( the above 18 steps) -- way to much work

you might as well use the .run driver

takes 1 min. to install ( well maybe 2 min. )
Code:
1) download and save to " / " the current .run driver for your card 
2) make sure that gcc and autotools and the kernel source are installed 
3) boot into text only mode and login as root
4) cd /
5) sh *.run 
( say "yes / accept" to everything in the on screen instructions)
6)when DONE  type in " reboot " and hit enter

or fallow this
http://elrepo.org/tiki/kmod-nvidia

Last edited by John VV; 04-07-2014 at 07:43 PM.
 
Old 04-07-2014, 09:19 PM   #9
anthony000
LQ Newbie
 
Registered: Apr 2014
Location: RI
Distribution: RHEL 7.1
Posts: 15

Original Poster
Rep: Reputation: Disabled
Hi John

Thanks for the big help here. As you can tell, I'm new at this and doing things the hard way I guess.
Questions:
1. What is a nvidia binary blob?
2. Will it cause a problem if re-running dracut if not needed (like after the first time to chnage from Nouveau to nVidia drivers?
3. Why does dracut not have to be re-run whenever there's a kernel upgrade? Will the initrd/initramfs.img be automatically loaded with the now-existing nvidia modules (I have blacklisted nouveau)?
4. When the kernel upgrades to a newer version, say 2.6.32-431.11-2.el6.x86_64 upgrades to 2.6.32-432.el6.x86_64, is it necessary to find a matching kmod-nvidia package? And if yes, will I need to uninstall the previous packages first?

I like the option of be able to use nVidia's *.run and the way you explain it seems straightforward, and definitely much easier than my million steps.
5. If going with *.run, would I be able to take advantage of its dkms capability (the nVidia help section says the opportunity will be given during installation to answer yes to register nvidia with dkms.
6. If going with *.run will I need to first uninstall the kmod-nvidia and nvidia-x11-drv packages from El Repo?

I search all over (it seems) for info but you've been the most helpful by far. The elrepo tiki on kmod-nvidia doesn't go into this level of detail and Red hat as you know won't for a proprietary driver.

Thanks agin,
Anthony
 
Old 04-07-2014, 09:29 PM   #10
anthony000
LQ Newbie
 
Registered: Apr 2014
Location: RI
Distribution: RHEL 7.1
Posts: 15

Original Poster
Rep: Reputation: Disabled
John

I did assign myself sudo permission and find it convenient now that I'm used to it.

Best,
Anthony
 
Old 04-07-2014, 10:03 PM   #11
John VV
LQ Muse
 
Registered: Aug 2005
Location: A2 area Mi.
Posts: 17,015

Rep: Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454Reputation: 2454
Quote:
1. What is a nvidia binary blob?
in order to run on every distro nvidia includes it's own libraries in the .run " blob"
blobs can overwrite system files , and the .run will replace PARTS of mesa opengl

Quote:
Will it cause a problem if re-running dracut if not needed (like after the first time to chnage from Nouveau to nVidia drivers?
that set of commands rebuilds the boot image "initramfs-????"

the Nouveau is installed and used by default AND included in the boot image
by blacklisting or uninstalling Novueau that boot image MUST get rebuilt with nouveau NOT USED

then and only then can the nvidia driver be installed and used

i posted a short script to run here:
( i am a VERY bad speller and make a ton of typos so scripts help)
http://scientificlinuxforum.org/inde...showtopic=2258
it ONLY needs to be ran one time and only ONE time

to rebuild the boot image WITHOUT the nouveau driver

Quote:
3. Why does dracut not have to be re-run whenever there's a kernel upgrade? Will the initrd/initramfs.img be automatically loaded with the now-existing nvidia modules (I have blacklisted nouveau)?
yum takes care of running scripts that need running on a update
this is automatically done

Quote:
4. When the kernel upgrades to a newer version, say 2.6.32-431.11-2.el6.x86_64 upgrades to 2.6.32-432.el6.x86_64, is it necessary to find a matching kmod-nvidia package? And if yes, will I need to uninstall the previous packages first?
"yum" should auto take care of that ,that is what it is for .

only if you are MANUALLY installing the .run driver ( and NOT!!! the kmod-nvidia.rpm) do you need to manually reinstall the .run driver on a kernel update

that is the advantage of using the kmod-nvidia.rpm

a kernel update will AUTO grab the kmod-nvidia driver

Quote:
5. If going with *.run, would I be able to take advantage of its dkms capability (the nVidia help section says the opportunity will be given during installation to answer yes to register nvidia with dkms.
see this ScientificLinux6 forum post
http://scientificlinuxforum.org/inde...showtopic=2173

Quote:
6. If going with *.run will I need to first uninstall the kmod-nvidia and nvidia-x11-drv packages from El Repo?
just the kmod-nvidia
you can leave "nvidia-x11-drv" installed
 
Old 04-08-2014, 11:45 AM   #12
anthony000
LQ Newbie
 
Registered: Apr 2014
Location: RI
Distribution: RHEL 7.1
Posts: 15

Original Poster
Rep: Reputation: Disabled
John

Thanks a lot for answering all my questions - you took away alot of the mystery for me. I'll check out the links to your scripts later too. It seems all the work for kmod-nvidia centers around its initial installation to replace the nouveau set-up. What's key for my situation (having already gone the El Repo kmod route) is the fact that the advantage of using the kmod-nvidia.rpm is that a kernel update will AUTO grab the kmod-nvidia driver. By this action it seems there's nothing for me to do for subsequent kernel upgrades once kmod-nvidia is installed - YUM should take care of it all. I could use nVidia's *.run method as a backup method.

Thanks again. I really appreciate your guidance on this.

Best,
Anthony
 
Old 04-11-2014, 11:42 AM   #13
anthony000
LQ Newbie
 
Registered: Apr 2014
Location: RI
Distribution: RHEL 7.1
Posts: 15

Original Poster
Rep: Reputation: Disabled
UPDATE 11 April 2014: I just went through my first automatic update using PackageKit (YUM) from El Repo's kmod-nvidia-331.49 to kmod-nvidia-331.67 and it performed without any need for manual intervention. All is normal.

Looking back at the whole process and thanks to John VV's help, all that's needed for using the nVidia graphics card with kmod-nvidia can be summarized as follows (see details of the steps earlier in this thread):
1. Install El Repo repository.
2. Install nvidia-detect; install dracut package.
3. Install the appropriate kmod-nvidia package identified from running nvidia-detect in Terminal mode.
4. Blacklist Nouveau in /etc/modprobe.d by creating the file blacklist-nouveau.conf (or blacklist whatever other default video/graphicsGPU module you were using. In the file, add the lines:
blacklist nouveau
options nouveau modeset=0
5. In Terminal recreate initramfs using dracut. ]# dracut -f -v /boot/initramfs-$(uname -r).img $(uname -r)
6. Reboot.
7. Done. Let updates occur automatically with YUM and/or PackageKit.
 
  


Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

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 Off
HTML code is Off



Similar Threads
Thread Thread Starter Forum Replies Last Post
[SOLVED] Bad screen resolution when switch from NVIDIA graphics card to motherboard graphics crs17 Linux - Hardware 9 11-15-2012 09:06 AM
No display on new instal of RHEL 4A 64 with Nvidia Quadro NVS 290 graphics eric_h Linux - Newbie 1 06-03-2009 03:13 PM
nvidia graphics card telefox Linux - Newbie 2 11-21-2007 05:44 AM
Nvidia graphics card discala Linux - Hardware 6 01-04-2007 08:30 AM
which nvidia graphics card John5788 Linux - Hardware 2 05-05-2004 08:26 AM


All times are GMT -5. The time now is 08:36 PM.

Main Menu
Advertisement
My LQ
Write for LQ
LinuxQuestions.org is looking for people interested in writing Editorials, Articles, Reviews, and more. If you'd like to contribute content, let us know.
Main Menu
Syndicate
RSS1  Latest Threads
RSS1  LQ News
Twitter: @linuxquestions
Facebook: linuxquestions Google+: linuxquestions
Open Source Consulting | Domain Registration