Announcement

Collapse
No announcement yet.

Kernel Mode-Setting With The Glint Driver

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

  • Kernel Mode-Setting With The Glint Driver

    Phoronix: Kernel Mode-Setting With The Glint Driver

    Earlier this month we reported on the ATI R300 GLSL compiler improvements being worked on as part of a Google Summer of Code (GSoC) project for X.Org, but how are their other 2010 projects progressing by these student developers? There's five GSoC X.Org projects in total this year and last night we received a bit of an update on the kernel mode-setting (KMS) efforts for porting the Glint driver...

    http://www.phoronix.com/vr.php?view=ODM0NQ

  • #2
    Aren't the lists now at freedesktop? I wonder if the message will be seen since it's not there.

    Comment


    • #3
      Originally posted by curaga View Post
      Aren't the lists now at freedesktop? I wonder if the message will be seen since it's not there.
      That was a screw-up on my part...

      Comment


      • #4
        Great, all drivers SHOULD have KMS

        There's no technical reason why every X.org driver can't have KMS. The logic and any reverse engineering has already been done before writing the userspace driver, and now it's just a matter of transferring that info to the kernel side.

        I hope this actually gets merged, unlike other Summer of Code projects like many listed under the EXA Status page that never made it into mainline for distro users to enjoy.

        Comment


        • #5
          Originally posted by stan View Post
          There's no technical reason why every X.org driver can't have KMS. The logic and any reverse engineering has already been done before writing the userspace driver, and now it's just a matter of transferring that info to the kernel side.

          I hope this actually gets merged, unlike other Summer of Code projects like many listed under the EXA Status page that never made it into mainline for distro users to enjoy.
          Nobody ever tested them stringently and actually backed them with a promise of maintenance. I'm working on the cards I do have, but I don't have everything.

          Comment


          • #6
            Originally posted by MostAwesomeDude View Post
            Originally posted by stan View Post
            There's no technical reason why every X.org driver can't have KMS. The logic and any reverse engineering has already been done before writing the userspace driver, and now it's just a matter of transferring that info to the kernel side.

            I hope this actually gets merged, unlike other Summer of Code projects like many listed under the EXA Status page that never made it into mainline for distro users to enjoy.
            Nobody ever tested them stringently and actually backed them with a promise of maintenance. I'm working on the cards I do have, but I don't have everything.
            I agree about a massive KMSification is a must, with a complete removing of fbdev and converting major projects into KMS system. I still hope to have KMS support for my i810 hardware, for example

            One of the killer points of Linux is the ability to make older hardware back into an usable state, this is quite useful for poorer people and as a matter to have cheap solutions for less powerful companies. I'm sad the corporativization of Linux is making it feel like other mainstream propietary operating systems in this aspect.

            About XOrg, I think the maintaining of source code is a must. Why waste efforts on GSoC or similar "student hiring" stuff when all that finally not goes into the project benefit? Seriously, Xorg Fund needs a proper management so being more efficient and faster evolving.

            Despite Permedia being considerated "old" hardware, it's still available on some not so old servers and workstations. This KMS support is quite good, I hope others like i810 and Matrox (both legacy and new) follow.

            Comment


            • #7
              Fbdev should only be removed after _most_ things using fb work in the KMS console. Currently this is not the case.

              On student summer projects, is it not a bit much to expect years of maintenance, for an agreed one-summer job & pay?

              Comment


              • #8
                Originally posted by timofonic View Post
                One of the killer points of Linux is the ability to make older hardware back into an usable state, this is quite useful for poorer people and as a matter to have cheap solutions for less powerful companies. I'm sad the corporativization of Linux is making it feel like other mainstream propietary operating systems in this aspect.
                In fairness, things like KMS were implemented on everything back to 10 year old chips, and a lot of that work was paid for by the "corporativization" folks.

                What I think you're seeing is vendors investing a bit more in supporting development, and that work *is* likely to be focused a bit more on relatively recent products, but that just means the community is free to shift a bit more of their work onto slightly older hardware. Look at all the great work marek and MostAwesomeDude are doing on 3xx-5xx Gallium3D, for example.

                Comment

                Working...
                X