Page 12 of 33 FirstFirst ... 2101112131422 ... LastLast
Results 111 to 120 of 329

Thread: Bickering Continues About NVIDIA Using DMA-BUF

  1. #111
    Join Date
    Jul 2008
    Posts
    359

    Default

    as a GNU fanboy i've been called a communist before but never a nazi.....

    i'm shaking with rage here [H]old me back!!

  2. #112
    Join Date
    Oct 2009
    Posts
    62

    Default

    Please kindly stop using Nvidia hardware then. It's not for hypocrites like you all.

    Please move to Intel IGPs/GPUs on Sandy Bridge, Ivy Bridge, Haswell, Broadwell, Skylake, Skymont etc.

  3. #113
    Join Date
    Jun 2009
    Posts
    2,932

    Default

    Don't let facts get in the way of his quest!

    He is saving us from NAZIS! That's serious.

  4. #114
    Join Date
    Jul 2008
    Posts
    359

    Default

    Quote Originally Posted by ChrisXY View Post
    The GPL is a proper license before the law. If you don't follow the law the copyright holders can sue you. Does that mean anything to you?

    It's funny how in those instances there are always so many people who are suddenly the experts on the GPL.

    Guys, if highly experienced people who create GPL code (almost) every day say nvidia can't use dma-buf the way they wanted, then there is a good chance that's actually correct.
    and with so many developers ... so many infact that some have died, some have ended up in prison and some have gone to live on a desert island it would be almost impossible to get them all to agree to change the licence .... gawd bless the GPL

  5. #115
    Join Date
    Jul 2008
    Posts
    359

    Default

    Quote Originally Posted by GT220 View Post
    Please kindly stop using Nvidia hardware then. It's not for hypocrites like you all.

    Please move to Intel IGPs/GPUs on Sandy Bridge, Ivy Bridge, Haswell, Broadwell, Skylake, Skymont etc.

    I don't use nvidia hardware, the only time i come across it is when i pull some old 8800gt out of the garbage

    and even then they're usually shit overheated and ready to go back in the garbage

    infact i know of cottage industries that attempt to reflow them - to be fair sometimes it does work

  6. #116
    Join Date
    Jun 2009
    Posts
    2,932

    Default

    Quote Originally Posted by mememe View Post
    Well, license agreements can only govern whatever copyright applies to, as license agreements are terms of the contract between the copyright holder and the licensee. The question is whether the Nvidia blob would use DMA-BUF as an API or whether it would derive from it.
    The whole defence for Nvidia's driver (and the reason why Linus tolerates it, despite hating it) is that it is a driver which runs on several operating systems and as such it is not derivative of the Linux kernel, it merely interfaces with it. It does the same thing on all systems, so it can run just fine without Linux. It ran on Windows before it ran on Linux, and it can run just fine without Linux, so there is a good argument that the current blob is not derivative of Linux kernel code.

    However, if you start adding functionality which uses specialised internal functions of the GPLed kernel, which only exists in the Linux kernel, and for the express purpose of tight cooperation with other (partly GPL) drivers which only exist inside the Linux kernel, then it is hard to argue that these parts of your blob are not derivative work of the kernel. You can argue the X11 also works on BSDs, so it's not derivative of that, but DMA_BUF is Linux-only, and GPL to boot. You can't claim it's generic: without Linux, it cannot possibly work. It was especially written only to work with a special, Linux-only subsystem. That's as derivative as you can get. And if you GPL those parts as per licence requirements, then you have to GPL it all.

    Nvidia's strategy worked well as long as they didn't touch any part of Linux and did it all themselves. Optimus screwed them up -- you can't avoid GPL code, and you can't simply wish the licence away like they are trying to.
    Last edited by pingufunkybeat; 10-18-2012 at 07:28 PM.

  7. #117
    Join Date
    Jul 2008
    Posts
    359

    Default

    Quote Originally Posted by pingufunkybeat View Post
    The whole defence for Nvidia's driver (and the reason why Linus tolerates it, despite hating it) is that it is a driver which runs on several operating systems and as such it is not derivative of the Linux kernel, it merely interfaces with it. It does the same thing on all systems.

    However, if you start adding functionality which uses specialised internal functions of the GPLed kernel, which only exists in the Linux kernel, and for the express purpose of tight cooperation with other (partly GPL) drivers which only exist inside the Linux kernel, then it is hard to argue that these parts of your blob are not derivative work of the kernel. And if you GPL those, as per licence requirements, then you have to GPL it all.

    Nvidia's strategy worked well as long as they didn't touch any part of Linux and did it all themselves. Optimus screwed them up -- you can't avoid GPL code, and you can't simply wish the licence away like they are trying to.

    to be fair linus and all the other devs have to tolerate people using the nvidia blob. Its your system and you can put what the hell you like on it, The 'module taints kernel' messages and 'fuck you nvidia' is all they can do

  8. #118
    Join Date
    Jun 2009
    Posts
    2,932

    Default

    Of course, but it's a potential problem for distributors. Linus has specified that he (unlike Alan Cox) does not consider the current Nvidia blob to be in violation of the GPL, for the reasons I listed.

  9. #119
    Join Date
    Jun 2010
    Location
    ฿ 16LDJ6Hrd1oN3nCoFL7BypHSEYL84ca1JR
    Posts
    1,052

    Default

    Quote Originally Posted by GT220 View Post
    Please kindly stop using Nvidia hardware then. It's not for hypocrites like you all.

    Please move to Intel IGPs/GPUs on Sandy Bridge, Ivy Bridge, Haswell, Broadwell, Skylake, Skymont etc.
    Actually my new notebook has the ivy bridge intel gpu and a hd 7970m with amd's enduro. I'm looking forward to using it with dma-buf in the not so far future.


    <troll>U jelly because your beloved superior nvidia cards run on too proprietary drivers to use it?</troll>


    Theoretically, how many people directly hold copyright for code used for dma-buf? Are there considerations to get them all together and sign some document that they won't sue nvidia for using prime/dma-buf, of course totally unsupported? Do they not want it? Would nvidia not do it?
    Last edited by ChrisXY; 10-18-2012 at 07:46 PM.

  10. #120
    Join Date
    Jul 2008
    Posts
    359

    Default

    Quote Originally Posted by pingufunkybeat View Post
    Of course, but it's a potential problem for distributors. Linus has specified that he (unlike Alan Cox) does not consider the current Nvidia blob to be in violation of the GPL, for the reasons I listed.
    i know of at least two distros that ship amd/nvidia blobs on their iso.. xbmc live and openelec

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •