[PlanetCCRMA] Need help getting realtime kernel working with Fedora 8 and nvidia

Fri May 9 08:11:39 PDT 2008

Some answers below:

On Fri, May 9, 2008 at 3:50 AM, Paul Vallee <pv.vallee at gmail.com> wrote:
> Hi Matt,
>
> Thanks for your help.
> I'm glad that your solution uses the livna-supplied drivers rather than the
> nvidia-supplied drivers because I understand from reading this article:
>
> Attention: Proprietary video driver users (ATI, Nvidia, etc.)
> https://www.redhat.com/archives/fedora-devel-list/2006-February/msg01178.html
>
> that the drivers supplied by Nvidia overwrite some OpenGL libraries with
> their own versions of those libraries. Very naughty.
>
> Before I can proceed I need ask you a few questions:
>
> 1. How do I find out if I have the plantetccrma kernel-devel package
> installed on my system and if it's not there, how do I get it?

to find out:

rpm -q kernel-rt-devel

if you want to find all the kernel packages installed:

rpm -qa | grep kernel

for more on the above commands, you can

man rpm

and do a search for "QUERY OPTIONS"

to install:

(as superuser)    yum install kernel-rt-devel

>
> 2. Do I have to remove the livna-supplied nvidia driver that I have already
> installed on my Fedora 2.6.24.5-85.fc8 kernel, or can it co-exist peacefully
> with the one I'm about to install on the planetccrma
> 2.6.24.3-1.rt1.2.fc8.ccrmart kernel?

You should be able to keep the livna-supplied driver, I think.  When
you run the rpm command to install the new one, make sure you use -ihv
rather than -Uhv, just to be safe (install rather than update).

>
> 3. Do I have to do the installation as root or should I do it as a regular
> user?

You'll need root privileges to install any rpm -- you can do it as
sudo if you have set up sudo privileges for your user(s).

>
> 4. Which 'package' do I need to download from livna:
> xorg-x11-drv-nvidia-169.12.1.lvn8.src.rpm,
> nvidia-kmod-169.12-7.lvn8.src.rpm, ...?
>

The nvidia-kmod package -- this is the kernel module that you have to
build to work with your specific kernel version.  If I'm not mistaken,
the other one is the driver components which are common to all nvidia
modules, as well as some documentation, the /dev devices, and some
config scripts.  (run

rpm -ql xorg-x11-drv-nvidia | less

to see what's there)

The beginning of the nvidia-kmod.spec looks like this for me:

----------------------------------------------------------------------------
# stuff to be implemented externally:
Source10: kmodtool
%define   kmodtool bash %{SOURCE10}
# end stuff to be ...

# hardcode for now:
%{!?kversion: %define kversion  2.6.24.3-1.rt1.2.fc7.ccrmart}

%define kmod_name nvidia
%define kverrel %(%{kmodtool} verrel %{?kversion} 2>/dev/null)

%define upvar ""
# kdump-devel missing
#define kdumpvar kdump
# xen uses a different version; ignore
#define xenvar xen
# don't build for debug kernels
#define debugvar debug PAE-debug
%{!?kvariants: %define kvariants %{?upvar}}

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

You can get the kernel version to put in the .spec file with

uname -r

again, notice I've deleted all references to any kvariants besides
upvar -- it's probably a bit nicer to comment out the defines rather
than delete them, but I like to get them out of my sight.  =o)

This has always worked for me -- anyone who is more of an rpm-master
who sees something terrible in the above should post a better
solution.

Hope this works for you, let me know.

Matt

PS.
This probably won't apply to you, but anyone who is using an old card
and can't get the nvidia drivers to work should try the
nvidia-96xx-kmod or (if it's REALLY old) nvidia-legacy-kmod drivers.

As I mentioned before, this method works great for livna's madwifi
package as well.  As of yet I can't get fglrx (ATI drivers) to work --
here's the error message:

FATAL: modpost: GPL-incompatible module fglrx.ko uses GPL-only symbol
'__rcu_read_lock'

I haven't found a workaround -- to anyone who knows the rt-kernel:  is
there likely to be one?

> Thanks
>
>
> On 5/8/08, Matt Barber <brbrofsvl at gmail.com> wrote:
>>
>> Hello,
>>
>> I highly suggest you use the RPM source -- it keeps things much
>> cleaner, and you get to use the pretty nifty livna-config-display
>> scripts (one of which is nvidia-config-display).
>>
>> I posted some directions on this here:
>>
>>
>> http://ccrma-mail.stanford.edu/pipermail/planetccrma/2008-April/014640.html
>>
>>
>> I think everything there is correct -- please let me know if it works
>> or if you have any questions -- if you want I can post what the
>> beginning of the final .spec file looks like when I do it.
>>
>> (hopefully I'M doing it right!...  )
>>
>> Thanks,
>>
>> Matt
>>
>>
>>
>>
>>
>> > Date: Wed, 7 May 2008 19:11:58 -0700
>> > From: "Paul Vallee" <pv.vallee at gmail.com>
>> > Subject: [PlanetCCRMA] Need help getting realtime kernel working with
>> >        Fedora  8 and nvidia
>> > To: planetccrma at ccrma.Stanford.EDU
>> > Message-ID:
>> >        <830daa640805071911r6708a09and2d0b8d323c895f6 at mail.gmail.com>
>> > Content-Type: text/plain; charset="utf-8"
>> >
>> > Hi all,
>> >
>> > I've just installed Fedora Core 8 on my laptop and am very anxious to
>> > get
>> > the planetccrma realtime kernel and all of those wonderful applications
>> > running on it. But I'm having a problem getting the rt kernel working.
>> >
>> > First, my vital statistics:
>> >
>> > Computer: HP Pavilion dv9608ca laptop, AMD Athlon 64/x2 cpu, 2GHz, 2GB
>> > memory
>> > video card: nvidia MCP67M (GeForce 7150M / nForce 630M compatible)
>> > Installed kernel that is working properly with the (livna-supplied)
>> > nvidia
>> > driver: 2.6.24.5-85.fc8
>> > Installed ccrma kernel: 2.6.24.3-1.rt1.2.fc8.ccrmart (NOT WORKING)
>> >
>> > Here's my story:
>> >
>> > I've successfully installed Fedora Core 8 on my laptop and now have a
>> > dual-boot system with Windows Vista as the other OS and GRUB as my
>> > bootloader. Everything working fine at this point. Then I installed the
>> > nvidia driver from livna to get full support for my display. Everything
>> > working superfine at this point. Played around with it for a couple of
>> > days,
>> > no problems.
>> >
>> > Then I followed the instructions at planetccrma and installed the
>> > realtime
>> > kernel listed above. No problems yet.
>> >
>> > When I rebooted into the ccrma kernel I noticed some error messages
>> > involving the nvidia driver. Couldn't find nvidia.ro or something like
>> > that.
>> > I didn't see the nvidia splash screen during boot either, so it
>> > definitely
>> > wasn't initializing properly. When I got to the login screen I was able
>> > to
>> > login to the GUI but it was unstable and eventually just froze; no
>> > response
>> > from mouse or keyboard. I've read many of the posts on various forums
>> > about
>> > similar problems being experienced by others and their solutions, but
>> > it's
>> > all a bit confusing and I don't want to do something out of desperation
>> > and
>> > loose my whole system in the process.
>> >
>> > Here's what I THINK the problem is:
>> >
>> > Because I installed the nvidia driver while running under the
>> > 2.6.24.5-85.fc8 kernel (and before installing the planetccrma realtime
>> > kernel) it only works with that kernel, and does not work with the ccrma
>> > kernel. After viewing the livna repository files I noted that there
>> > aren't
>> > any driver rpm's for the ccrma version of the kernel
>> > (2.6.24.3-1.rt1.2.fc8.ccrmart). Many people talked about this in the
>> > forums
>> > and suggested downloading the livna sources and re-compiling the driver
>> > for
>> > the ccrma kernel. Is this what I need to do?
>> >
>> > Here's what I THINK I have to do:
>> >
>> > 1. Install the kernel-rt-devel package for the ccrma realtime kernel (do
>> > I
>> > need the kernel-headers too?)
>> > 2. Download the source rpm's for the livna driver
>> > 3. Install the source rpm's for the livna driver
>> > 4. Compile the livna source rpm's and install the resulting binary rpm's
>> > while running under the ccrma realtime kernel
>> > 5. Reboot and I will be in ccrma heaven
>> >
>> > Is this correct?
>> > If so, could someone please give me, or point me to, detailed
>> > instructions
>> > on how to accomplish this. I'm far from a Linux expert and very new to
>> > Fedora.
>> >
>> > If it's not correct, could someone please illuminate me.
>> >
>> >
>> > --
>> > Kind Regards,
>> >
>> > Paul Vallee
>> > pv.vallee at gmail.com
>> >
>> > .
>> > -------------- next part --------------
>> > An HTML attachment was scrubbed...
>> > URL:
>> > http://ccrma-mail.stanford.edu/pipermail/planetccrma/attachments/20080507/4fb32244/attachment-0001.html
>
>
>
> --
> Kind Regards,
>
> Paul Vallee
> pv.vallee at gmail.com
>
> .