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

Newegg Daily Deals

Closed Thread
 
Thread Tools
Old 03-25-04, 12:28 AM   #25
macemoneta
Registered User
 
macemoneta's Avatar
 
Join Date: Oct 2002
Location: NJ
Posts: 159
Default Re: Alternative approach to Linux Driver?

There's possibly another approach that Nvidia could use; a system service (init scripts). By adding a started service to the system (e.g., 'chkconfig --add nvidia') as part of the installation, the update/rebuild process could be automated and not interrupt the reboot (other than the time it takes to download/rebuild the module).

Here's how I see it working:

1. The init script gets control during startup
2. Check to see if the kernel has changed since the last boot
3. If kernel has changed, then:
-- a. Check that new kernel source to match running kernel available
-- b. Verify .config exists in kernel source tree
---- i. If not, copy appropriate (based on uname) .config from configs dir
---- ii. make oldconfig
-- c. Download or rebuild the module if the kernel has changed
4. Continue the boot

Each step is easy enough to do with the existing options on the installer. All that needs to be done is to create the small script used by the init scripts.

This would make upgrades and maintenance as painless as possible for users (brief delay when a new kernel is loaded), while still allowing for Nvidia to maintain its proprietary IP.

This should make the process completely transparent to users. Appropriate status and error messages will tell them why a delay is occuring, and they will boot into X as usual at the end of the process. There is really no change to the existing installer, just the addition of the init script installation.
macemoneta is offline  
Old 03-26-04, 09:21 PM   #26
thinkliberty
Registered User
 
Join Date: Dec 2003
Posts: 21
Default Re: Alternative approach to Linux Driver?

Quote:
Originally Posted by SuLinUX
I find that people who moan about nvidia opensourcing there driver seem to have problems or cannot do a simple install of it. Ati and nvidia drivers are not without there issues in Windows never mine Linux.
Thats funny. I find that people who moan about nvidia going open source with their graphics driver seem to have a problem with the bugs in the driver that causes the system to hang and cannot do anything to fix the situation.
thinkliberty is offline  
Old 03-26-04, 09:57 PM   #27
ejona
Registered User
 
Join Date: Mar 2004
Posts: 4
Default Re: Alternative approach to Linux Driver?

I definitely agree with you, however, keep in mind that everything on Linux doesn't work perfectly 100% of the time. Most of the time there is an easy install yet a mess with configuring. Although I would always hope that companies contribute their drivers as Open Source, I really don't complain much becuase lots of manufactures don't have drivers at all, so then some wonderful Linux developers get to reverse engineer the hardware and come up with a reasonable driver for it, with our without all of the supported options of the device. Plus, talk about a wait. I support NVidia's decision fully because if they are being nice to the Linux community let's not say that it isn't good enough and we want more.

Also, when a bug is found with a normal Linux driver (open source) or program what happens? You tell the developers and they fix it or you figure out how to fix it and then submit a patch. Both of these options are available to us. The part of the code that isn't open source already has little need to be, if it is indeed the cross-platform part, because the bugs we have do not relate to all OSs, but merely Linux.
ejona is offline  
Old 03-26-04, 11:11 PM   #28
etr
Registered User
 
Join Date: Aug 2002
Location: NC, US
Posts: 55
Default Re: Alternative approach to Linux Driver?

Well, to answer carpy's original question:

No, nVidia probably can't reasonably release an open source driver that would provide enough functionality for transparent kernel upgrades. First and foremost, this cannot be done because the GNU license used by the kernel forbids distribution of binaries linked to closed-source objects. This restriction of the GNU license is intended to prevent someone from exploiting a GNU project by making improvements or extensions but hiding them in a binary-only module so that they can't be adopted by others.

This means that anything that nVidia wants to use with the kernel must either be distributed separately from the kernel or must be entirely GNU compliant. So long as the nVidia kernel driver is distribute separately from the kernel, the upgrade difficulty will remain. However, nVidia seems to have code it needs to run in kernel space (judging by its implementation as a kernel driver) that it doesn't want to release to the rest of the world, so putting the code under a GNU-compliant license seems out of the question.

In theory, it might be possible to write a kernel package that wouldn't need proprietary code in a kernel module (without nVidia revealing information they prefer remain secret). However, this might have undesirable side effects, which is presumably why nVidia hasn't gone this route.

It should be noted that nVidia doesn't own all the code in its drivers. It licenses some of its code, and this code often comes with licenses that prevent it from being open-sourced (at least by nVidia). In short, even if nVidia decided to open-source its drivers, it would need to renegotiate with vendors and/or re-write portions of their driver to do so.

Quote:
Since XFREE 4.4 is no longer free software, a lot of distros are dropping it for other X implementations. This means......
I can't say I agree with this statement. Yes, XFree86 did have a license change, and yes, most distributions are ditching it as a result.

However, that doesn't mean that XFree86 isn't free. The change in the license is simply a new vanity/advertising clause. In principle, this clause isn't significantly more obnoxious than the copyright/license preservation clauses present in most OSS.

In practice, this clause is a potential legal pain. It's a lot easier to leave source files intact than to make sure you haven't missed one of a list of acknowledgments. I can see why distributions don't want to include a license it would be easy to run afoul of by sheer oversight. If they want to ditch XFree86 for that reason, they are within their rights and have a perfectly reasonable argument.

However, claiming that the clause disqualifies XFree86 as free software is disingenious our just plain dishonest...unless you intend to level the same charge at the GNU license. While code under the new XFree86 license can't be used in GNU projects, at least not without permission or a dual license scenario, the same hurdles exist for BSD projects that would like to use GNU code. I fail to see where one license is more "free" than the other.

Incidentally, the XFree86 license change only affects a relatively small portion of the code, and other portions of the code have long had even more problematic licenses--distros just haven't paid attention to that until now. If you want to argue that XFree86 4.4 isn't free, you also have to argue that prior versions weren't free. The thing to keep in mind here is that XFree86 doesn't have a single, monolithic license. The authors of given files are free to choose any license so long as it meets a couple of basic XFree86 objectives (which GNU, notably, does not). If the proposed license does not follow XFree86 guidelines, XFree86 simply does not incorporate the file into the project. XFree86 license guidelines have long allowed licenses like the new one...and some files have had such licenses for a long time.

None of this to say that the license change is a good thing...just that many of the arguments thrown at it are inaccurate.
etr is offline  
Old 03-27-04, 04:33 AM   #29
Thunderbird
 
Join Date: Jul 2002
Location: Netherlands, Europe
Posts: 2,105
Default Re: Alternative approach to Linux Driver?

Sorry to interrupt you guys again but in the end this thread is just another "nvidia release the source" thread. It has been argued dozens of times now that Nvidia can't /isn't allowed to open the source, so please stop these discussions as they are useless.
Thunderbird is offline  
Closed Thread


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
173.14.35 Linux Driver annb NVIDIA Legacy Graphics Cards 0 06-13-12 07:46 PM
GTX 460M, Xorg 1.12, nvidia driver 295.49, linux 3.3.5-1-ARCH and NVRM Xid 44 elsifaka NVIDIA Linux 0 05-14-12 09:48 AM
Best Linux driver for Quadro FX2700M dclement NVIDIA Linux 0 05-07-12 07:58 AM
302.07 (beta) for Linux x86/x86_64 released AaronP NVIDIA Linux 0 05-02-12 10:55 AM
download problem with linux NVIDIA driver packages Timo NVIDIA Linux 7 07-31-02 10:22 AM

All times are GMT -5. The time now is 04:19 PM.


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