Patching drivers for 2.5.38

[stefangorling]

Have anyone managed to patch the driver in order to make it work with kernel 2.5.38?

[pbrennen]

Bug the people in this forum to make a patch! You could aslo try this person: faasen@xs4all.nl http://thuis.zwanebloem.nl/nvidia/

Q: What is NVIDIA's policy towards development series Linux kernels?

A: NVIDIA does not officially support development series kernels.
However, all the kernel module source code that interfaces with the
Linux kernel is available in the NVIDIA_kernel package, and NVIDIA
encourages members of the Linux community to develop patches to these
source files to support development series kernels. A google search
will most likely yield several community supported patches.

[rol]

No, please, not again...

This should be read as :
NVIDIA strongly encourages people to do what *they*should be doing, without giving them access to the information they need to do it.

Thus, NVIDIA doesn't support the dev. kernel as long as they are called dev. kernels, and that's all.

Kernel developpers, could you please stop calling the 2.5.x branch a developpment one ?

More seriously, complain to linux-bugs but don't expect an anszer.
Offer your help to test new build/releases that could be compliant with 2.5.x, but don't expect an answer.

NVidia is working, sometimes releasing a new version, but is otherwise deaf, blind. Don't expect anything :-(

Rol

[stefangorling]

I modified an old patch and managed to get the module to load and work into the 2.5.38 kernel. As I did a few ugly things I can't guarantee that it's stable. If anyobody wants it, mail to nvidia at gorling.se and I'll send you a patch.

/Stefan

[pbrennen]

Thanks for your quick work Stefan. Here's another patch for 2.5.38.

http://groups.google.com/groups?q=nvidia+linux+2.5&hl=en&lr=&ie=UTF-8&scoring=d&selm=3D96CC7B.7060801%40drugphish.ch&rnum=2

You should probably post your patch to this thread in linux.kernel, and also to alt.comp.periphs.videocards.nvidia.

[mcelrath]

Q: What is NVIDIA's policy towards development series Linux kernels?

A: NVIDIA does not officially support development series kernels.
However, all the kernel module source code that interfaces with the
Linux kernel is available in the NVIDIA_kernel package, and NVIDIA
encourages members of the Linux community to develop patches to these
source files to support development series kernels. A google search
will most likely yield several community supported patches.

From the driver:

* Copyright 1999-2001 by NVIDIA Corporation. All rights reserved. All
* information contained herein is proprietary and confidential to NVIDIA
* Corporation. Any use, reproduction, or disclosure without the written
* permission of NVIDIA Corporation is prohibited.


I do not have written permission to do ANYTHING with the source. The partial source release is a red herring since we have no rights, NVIDIA retains them all.

Forget it. NVIDIA is trying to get something for nothing. Do your own damn work. There are many people in the community that would be more than happy to rework their driver for 2.5 (including me), but not under the above conditions. Exactly what incentive do we have? NVIDIA will just appropriate our changes without so much as a thank you.

This is why the GPL is good. Developers know that their improvements will be made available, they will receive credit, and they will receive future enhancements. THAT is why people work on linux. The situation with NVIDIA is more like: "Here do some unpaid work for us and we'll take it from you and make a bunch of money off it".

No thanks.

[Thunderbird]

The main reason there's no 2.5x support is because 2.5x is a dev kernel. That means the kernel is in heavy development, a lot of important things the module depends on change oftenly.

Nvidia wants to provide something stable. When 2.5x becomes 3.0 there will be a module very soon.

[bwkaz]

3.0? I thought it was going to become 2.6... of course, I don't read lkml, so I suppose Linux might have changed his mind since the last time I read through some of the archives...

Whatever. Not that it really matters...

[leahcim]

Nvidia want to provide something stable??

Erm, have you looked at the code?

Sequences like :-

allocate memory...
memset it to 0
then check if the pointer is null

grabs spinlocks....then later potentially error return with no unlock.

It's a mess, I'm no kernel developer, and I can pull bags of bugs, mostly obvious c coding bugs, with a quick glance...christ knows what the hidden parts are like or what a kernel developer who understands the implication of locking / memory allocation etc would find.

They could well be the best 3d graphics driver writers on the planet, but they could do with taking a long hard look at the kernel side - perhaps re-thinking the license and / or paying for some expertise.