Announcement

Collapse
No announcement yet.

Whoops, There's A Big Problem For Wayland GTK+

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • RussianNeuroMancer
    replied
    Originally posted by elanthis View Post
    They also require being able to access the graphics drivers and GL calls from outside of the GLX-based libGL API and via EGL.
    I do not believe that the binary drivers offer any of that right now.
    Originally posted by smitty3268 View Post
    I thought every GL ES driver had EGL support in it, but i guess i could be wrong. EGL support is about more than just Wayland or linux, though, so i definitely think the binary drivers will want to support that eventually if they don't yet.
    http://phoronix.com/forums/showthrea...869#post254869

    Leave a comment:


  • smitty3268
    replied
    Originally posted by elanthis View Post
    They also require being able to access the graphics drivers and GL calls from outside of the GLX-based libGL API and via EGL.

    I do not believe that the binary drivers offer any of that right now.
    I thought every GL ES driver had EGL support in it, but i guess i could be wrong. EGL support is about more than just Wayland or linux, though, so i definitely think the binary drivers will want to support that eventually if they don't yet.

    Leave a comment:


  • elanthis
    replied
    Originally posted by smitty3268 View Post
    That means those drivers just have to patch Wayland to interact with their own kernel drivers instead of the OSS APIs. That should be very simple to do, as Wayland handles that code very cleanly. Their drivers already have to heavily patch X to get it to work, so they just have to do the same thing to Wayland, but it will be much easier.
    They also require being able to access the graphics drivers and GL calls from outside of the GLX-based libGL API and via EGL.

    I do not believe that the binary drivers offer any of that right now.

    Leave a comment:


  • smitty3268
    replied
    Wayland requires kernel modesetting, or kms. It DOES NOT require KMS. Notice the difference in capitalization.

    The nvidia and fglrx blobs already contain working kms systems in their drivers - they just use a different API than the open source drivers do (KMS).

    That means those drivers just have to patch Wayland to interact with their own kernel drivers instead of the OSS APIs. That should be very simple to do, as Wayland handles that code very cleanly. Their drivers already have to heavily patch X to get it to work, so they just have to do the same thing to Wayland, but it will be much easier.

    Leave a comment:


  • chithanh
    replied
    Originally posted by asdx
    Originally posted by garegin View Post
    i've reported it to almost every linux forum, including this. kms is broken on many macs including the mac mini and macbook air. this is a known bug.
    So file a bug if there isn't one already.
    Actually there are several bugs already, just search https://bugs.freedesktop.org/.

    One of the more notorious ones: Radeon KMS on Macs with EFI boot. The workaround is to modify the kernel with a patch that was rejected by the radeon developers, and make it load the VGA BIOS from a dump you created in bootcamp.

    Leave a comment:


  • RussianNeuroMancer
    replied


    Originally posted by RealNC View Post
    In general, KMS in the blobs doesn't sound impossible. They would need glue in form of a kernel module with the actual functionality being provided by the blob.
    Originally posted by Serafean View Post
    KMS won't be easy for nvidia to implement in their binary blob (anyone remember the mess when OSS drivers became KMS enabled?). Another hard requirement for Wayland is EGL support, which I don't know if the binary blobs support.
    Really? nVidia already support EGL and KMS in proprietary driver for Tegra. AMD already support EGL in driver for embedded hardware.
    EGL is not strict dependecy for Wayland by the way (check Wayland FAQ).

    Leave a comment:


  • mrugiero
    replied
    Originally posted by garegin View Post
    i've reported it to almost every linux forum, including this. kms is broken on many macs including the mac mini and macbook air. this is a known bug.
    That's not reporting a bug. You report it to bugzilla (or any other bug filing framework kernel devs use) or to their mailing list. Not all devs are in forums, you know?

    This can help you. I didn't read it yet because I've never found a mainstream bug (well, as a matter of fact, I've found a few, but they were already reported and fixed when I did figured out they were kernel bugs), and the ones with Openchrome were reported directly to their devlist.

    EDIT: Whoops, forgot to put the link http://kernelnewbies.org/FoundBug.

    Leave a comment:


  • kraftman
    replied
    Originally posted by garegin View Post
    i've reported it to almost every linux forum, including this. kms is broken on many macs including the mac mini and macbook air. this is a known bug.
    So, it seems you're using broken hardware. I never had any problems with KMS on my PC.

    Leave a comment:


  • mrugiero
    replied
    Originally posted by asdx
    And what have you done to help with that? Have you tried reporting bugs?

    I never had problems with KMS, what driver/GPU were you using?
    Idem. The only problem I had with KMS was in my Unichrome box, because it's not ready yet.

    Leave a comment:


  • garegin
    replied
    i've reported it to almost every linux forum, including this. kms is broken on many macs including the mac mini and macbook air. this is a known bug.

    Leave a comment:

Working...
X