This may be a dumb question, but why can't Redhat distribute NVIDIA binary drivers?
In NVIDIA's licence (http://www.nvidia.com/object/nv_swlicense.html) it says:
"2.1.2 Linux Exception. Notwithstanding the foregoing terms of Section
2.1.1, SOFTWARE designed exclusively for use on the Linux operating system may be copied and redistributed, provided that the binary files thereof are not modified in any way (except for unzipping of compressed files)."
So, what's keeping RedHat from putting the drivers in the distribution? If it's a GPL thing, would it be easy to just download it during installation or at least give the option to the user?
Thanks,
-- David St.Clair dstclair@cs.wcu.edu
On Friday 25 July 2003 08:09, David St.Clair wrote:
So, what's keeping RedHat from putting the drivers in the distribution? If it's a GPL thing, would it be easy to just download it during installation or at least give the option to the user?
It's a closed source thing. Red Hat doesn't distribute any closed source software anymore (that I know of). It's impossible to support, and it taints your kernel.
As I see it, the best way would be for a third-party project like Fedora to package the binary drivers, but with redhat-config-xfree86 being able to modify the XF86Config file. That way, you'd just
1) apt-get install nvidia-glx
2) redhat-config-xfree86
3) Check the "Enable 3d acceleration" box.
4) Restart X.
Unbelievably simple. This feature would make RHL look really good in a Distro Review on osnews.com :-)
/Peter
On Friday 25 July 2003 17.31, Jesse Keating wrote:
On Friday 25 July 2003 08:09, David St.Clair wrote:
So, what's keeping RedHat from putting the drivers in the distribution? If it's a GPL thing, would it be easy to just download it during installation or at least give the option to the user?
It's a closed source thing. Red Hat doesn't distribute any closed source software anymore (that I know of). It's impossible to support, and it taints your kernel.
On Fri, 2003-07-25 at 18:47, Peter Backlund wrote:
As I see it, the best way would be for a third-party project like Fedora to package the binary drivers,
See https://bugzilla.fedora.us/show_bug.cgi?id=402 - will be there once the package passes QA.
- Panu -
On Saturday 26 July 2003 12:44, Panu Matilainen Panu Matilainen pmatilai@welho.com wrote:
On Fri, 2003-07-25 at 18:47, Peter Backlund wrote:
As I see it, the best way would be for a third-party project like Fedora to package the binary drivers,
See https://bugzilla.fedora.us/show_bug.cgi?id=402 - will be there once the package passes QA.
- Panu -
I've had a quick browse of the above link, and I downloaded the MD5SUM and the src package. A couple of questions: (I'm running *only* Severn on my machine). Can I go ahead and install this on my system, _AND_ do I need to re-edit my XConfig file (as previously, removing DRI, and changing "nv" to 'nvidia", switch to init 3, etc. or will the install do the necessary)?
Second question: how do I install the src package: rpmrebuild -rebuild package, then rpm-ivh package? (sorry, I confess I'm being a bit lazy ... or maybe it's Sunday morning... so go ahead and rap my knuckles! <G>).
TIA,
Elton ;-)
Can I go ahead and install this on my system, _AND_ do I need to re-edit my XConfig file (as previously, removing DRI, and changing "nv" to 'nvidia", switch to init 3, etc. or will the install do the necessary)?
The package is in QA, and therefore not quite ready for general use. A better way would be to use Nvidia's installer.
http://www.nvidia.com/object/linux.html
If you're not sure, use the IA-32 driver. Btw, read the readme. Yes, all your problems will be solved if you follow the instructions there :-).
Second question: how do I install the src package: rpmrebuild -rebuild package, then rpm-ivh package? (sorry, I confess I'm being a bit lazy ... or maybe it's Sunday morning... so go ahead and rap my knuckles! <G>).
man rpmbuild
On Sunday 03 August 2003 12:11, Peter Backlund Peter Backlund peter.backlund@home.se wrote:
The package is in QA, and therefore not quite ready for general use. A better way would be to use Nvidia's installer.
Understood. But I'm running Seven only. So my system's beta anyway.
http://www.nvidia.com/object/linux.html
If you're not sure, use the IA-32 driver. Btw, read the readme. Yes, all your problems will be solved if you follow the instructions there :-).
I've already installed *that* driver with previous kernels, but it now refuses to create the module with the Severn (RH 9.0.93) kernel-2.4.21-20.1.2024.2.1.nptl.
I'm quite willing to use their stock rpm (as previously, and to the usual init3, editing of Xcongfig, etc. However, I need some "hand-holding" (i.e. dummy / newbie level instructions on the exact procedure of how I pass the option of "ignore kernel version" ... sorry, I KNOW it's been discussed before, but I didn't pay much attention, .. and the instructions assumed a bit more than 'newbieness' .
Furthermore I didn't *need to do that* at the time. 8-(
Thanks again,
Elton ;-)
P.S. and I *shall* read "man rpmbuild"!
On Sunday 03 August 2003 12:56, Elton Woo Elton Woo elwoo@videotron.ca wrote:
However, I need some "hand-holding" (i.e. dummy / newbie level instructions on the exact procedure of how I pass the option of "ignore kernel version" ... sorry, I KNOW it's been discussed before, but I didn't pay much attention, .. and the instructions assumed a bit more than 'newbieness' .
... to be more specific: how do I do this while in init 3 as root and running the NVIDIA~.bin file:
On Thursday 24 July 2003 21:45, Elliot Peele Elliot Peele ebpeele2@pams.ncsu.edu wrote:
to do. Either set the IGNORE_CC_MISMATCH environment variable or set the CC environment variable to the compiler used to compile the kernel. Personally I do the first (ie. export IGNORE_CC_MISMATCH=1).
Elliot
... sorry, but I guess I'm being a bit dense here... :-( So, dummy instructions, please?
Elton ;-)
On Sun, 2003-08-03 at 12:56, Elton Woo wrote:
I've already installed *that* driver with previous kernels, but it now refuses to create the module with the Severn (RH 9.0.93) kernel-2.4.21-20.1.2024.2.1.nptl.
Did you export IGNORE_CC_MISMATCH=1 before running the driver installation program? Aside from needing to do that, the new drivers run fine on my Severn box.
Phil
On Sunday 03 August 2003 13:19, Phillip Compton Phillip Compton pcompton@proteinmedia.com wrote:
Did you export IGNORE_CC_MISMATCH=1 before running the driver installation program? Aside from needing to do that, the new drivers run fine on my Severn box.
No. I didn't. Let's see if I have this right:: Is the *correct syntax* ---> "export IGNORE_CC_MISMATCH=1"? or is the command all upper case? After that, I do sh NVIDIA~.bin. Am I on track, now?
Elton ;-)
On Sun, 2003-08-03 at 14:05, Elton Woo wrote:
No. I didn't. Let's see if I have this right:: Is the *correct syntax* ---> "export IGNORE_CC_MISMATCH=1"? or is the command all upper case? After that, I do sh NVIDIA~.bin. Am I on track, now?
yes
$ export IGNORE_CC_MISMATCH=1 $ ./NVIDIA*.bin
and edit the XF86Config as usual
the problem is that the kernel is compiled with gcc3.2
Phil
On Sunday 03 August 2003 14:23, Phillip Compton Phillip Compton pcompton@proteinmedia.com wrote:
$ export IGNORE_CC_MISMATCH=1 $ ./NVIDIA*.bin
and edit the XF86Config as usual
Thank you very much!
Elton ;-)
On Friday 25 July 2003 08:09, David St.Clair wrote: It's a closed source thing. Red Hat doesn't distribute any closed source software anymore (that I know of). It's impossible to support, and it
taints
your kernel.
No closed source and considering yum or apt-get?
Why am I expecting three figures dressed as cardinals to burst in? "Nobody expects Debian!" "Three of our chief weapons are fear, surprise, ruthless packaging rules and an almost fanatical devotion to the GPL."
On Friday 25 July 2003 08:53, John Hearns wrote:
No closed source and considering yum or apt-get?
Why am I expecting three figures dressed as cardinals to burst in? "Nobody expects Debian!" "Three of our chief weapons are fear, surprise, ruthless packaging rules and an almost fanatical devotion to the GPL."
This has little to do with GPL, and everything to do with RHL developers being unable to view/modify if necessary the nvidia driver source. Once it's "included" in the distro, then steps have to be made to make sure that it'll work with all other newer software as it comes out, such as new kernels, new XFree86, all that fun stuff. Not being able to modify the nvidia source puts a huge damper on getting such updates out and available.
Please note that there is _plenty_ of software in Red Hat that isn't GPL. XFree86 for one...
This has little to do with GPL,
I agree. It was just a joke.
and everything to do with RHL developers
being unable to view/modify if necessary the nvidia driver source. Once it's "included" in the distro, then steps have to be made to make sure that it'll work with all other newer software as it comes out, such as new kernels, new XFree86, all that fun stuff.
Absolutely. I've heard Alan Cox speak on this matter much more eloquently than I can put it.
On Fri, 25 Jul 2003 16:53:52 +0100, "John Hearns" John.Hearns@micromuse.com wrote:
[...]
No closed source and considering yum or apt-get?
[...]
What has yum to do with Debian?
regards
Why not download it during installation (so no "distribution") making it clear that it is a 3rd party driver and is not supported? That's what SUSE does. That way the user gets to choose. I could care less if it "taints" my kernel. I just want a 3d driver that works. I know it's not hard to install, but it would just be a nice feature that would make redhat seem more polished.
-- David St.Clair dstclair@cs.wcu.edu
On Friday 25 July 2003 08:09, David St.Clair wrote:
So, what's keeping RedHat from putting the drivers in the distribution?
If
it's a GPL thing, would it be easy to just download it during installation or at least give the option to the user?
It's a closed source thing. Red Hat doesn't distribute any closed source software anymore (that I know of). It's impossible to support, and it
taints
your kernel.
-- Jesse Keating RHCE MCSE http://geek.j2solutions.net Mondo DevTeam (www.mondorescue.org)
On Friday 25 July 2003 09:13, David St.Clair wrote:
Why not download it during installation (so no "distribution") making it clear that it is a 3rd party driver and is not supported? That's what SUSE does. That way the user gets to choose. I could care less if it "taints" my kernel. I just want a 3d driver that works. I know it's not hard to install, but it would just be a nice feature that would make redhat seem more polished.
And then Red Hat will have to create some mechanism to re-download or rebuild the driver every time you install a new kernel or an XFree86 update. Not exactly a trivial thing to automate.
As Havoc mentioned on this mailing list yesterday, we should be talking about the bigger picture here. I think a better discussion would be how can we make the redhat configuration tools aware of third party, possibly closed source, drivers, not just for video cards, but networks cards, printers, and anything and everything else. Not only that, but is it possible and is it desirable. Maybe like devising some sort of scheme where an RPM for a driver can have a driverinfo file that the redhat tools can use to figure out how to set up and configure devices using that driver.
Geoff Reedy
desirable. Maybe like devising some sort of scheme where an RPM for a driver can have a driverinfo file that the redhat tools can use to figure out how to set up and configure devices using that driver.
Whats generally worked well here is things like the .desktop files, where the app scans a directory of files owned by various packages to build the menus.
-
Alan Cox -
David St.Clair -
Elton Woo -
Geoff Reedy -
Jesse Keating -
John Hearns -
Marco Ermini -
Panu Matilainen -
Peter Backlund -
Phillip Compton