NVIDIA 304.37 Linux Driver Brings 41 Official Changes

[RealNC]

This is such a silly argument. Nvidia's blob and nouveau cannot coexist, so it only makes sense to offer an option for disabling nouveau. This option simply blacklists nouveau, so what's the big deal?

It's simply that some people lack common sense.

[pingufunkybeat]

This is such a silly argument. Nvidia's blob and nouveau cannot coexist, so it only makes sense to offer an option for disabling nouveau. This option simply blacklists nouveau, so what's the big deal?

The command-line option is a small deal. The big deal is that they should release documentation for their hardware, which would improve out-of-the-box experience for millions of Linux users out there. They can still offer their driver to whoever wants it.

The fact that their only contribution to the Linux driver for Nvidia hardware has been a command-line switch to blacklist it makes this a very good place to point this out -- again.

That's all. Don't get distracted by RealNC's circular schizo-arguments with himself. I only posted a one-post-rant before he came here with his merry-go-round lala-world.

[RealNC]

Your argument is one-sided. Because people's experience would also get better if Linux would be more friendly to proprietary drivers.

[pingufunkybeat]

Your argument is one-sided. Because people's experience would also get better if Linux would be more friendly to proprietary drivers.

They have the kernel source, they have the permission to link their blobs with the GPL-ed kernel, they have all the information about the hardware (they made the hardware), what else do they need? Their blob is fine, at least according to you.

The Linux kernel people, on the other hand, don't have any access to documentation so people are effectively not able to run open-source drivers on their Nvidia hardware (though the nouveau guys are really changing this, and fast).

The situation is one-sided by nature.

[RealNC]

NVidia requested permission to use DMABUFF in order to implement Optimus on Linux. The kernel devs said no. It's not one-sided.

[pingufunkybeat]

dmabuf is not even finished yet and the licensing has not been fully decided, AFAICR

[brent]

The big deal is that they should release documentation for their hardware, which would improve out-of-the-box experience for millions of Linux users out there.

That is a completely different issue.

[AJSB]

I don't suppose anyone knows if this includes the new-fangled fixes from the Valve visit?

Don't think so but the switch --dusable-nouveau might be related

[ninez]

dmabuf is not even finished yet and the licensing has not been fully decided, AFAICR

I've been reading this thread and am curious (not arguing the above) but, i was under the impression it has already been decided that dmabuf will use EXPORT_SYMBOL_GPL which means nvidia cannot use the interface ~ have i missed something here?

also, taking a quick look at the code for kernel 3.5; http://lxr.free-electrons.com/source...base/dma-buf.c

they are using EXPORT_SYMBOL_GPL, which to me says (or would appear) that the decision has been made and Nvidia won't be permitted to use that.

anyway, i am curious that if it hasn't been decided, then how come all of the code appears this way?

We went through something similar with the RT-patch, where he had to initially change these symbols in the kernel in a few places, before a new patch surfaced that got around it. ~ So my question would be (if i am not missing anything) how is nvidia going to use dmubuf when these symbols are EXPORT_SYMBOL_GPL???

it would seem to me to be a 'showstopper' as far as i can tell.

[pingufunkybeat]

I haven't followed it closely, has anything happened since this? It was undecided then.