Announcement

Collapse
No announcement yet.

dri2 and/or kms status?

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

  • #16
    Originally posted by EmbraceUnity View Post
    In the recent kernel logs and IRC convos I have definitely seen some things about radeon-specific enhancements in .30

    This wouldn't be KMS by any chance? That would be quite surprising.

    http://kernel.org/pub/linux/kernel/v...6.29-git10.log

    There is specific stuff about r600 in there (ctrl f)

    Why is r600 mentioned at all in the kernel logs if not KMS?
    2.6.30 adds r6xx/r7xx drm support for 2D and video acceleration (EXA and Xv). There is no radeon KMS support in 2.6.30.

    Comment


    • #17
      Ok, so I have got all the git branches compiled and running, with the exception of the DDX driver. For some reason it won't let me compile it.

      Code:
      radeon_textured_video.c: In function ‘RADEONPutImageTextured’:
      radeon_textured_video.c:505: error: ‘struct _ScrnInfoRec’ has no member named ‘fboffset’
      radeon_textured_video.c:510: error: ‘struct _ScrnInfoRec’ has no member named ‘fboffset’
      radeon_textured_video.c:572: error: ‘struct _ScrnInfoRec’ has no member named ‘fboffset’
      make[1]: *** [radeon_textured_video.lo] Error 1
      make[1]: Leaving directory `/usr/src/xf86-video-ati-airlied/xf86-video-ati/src'
      make: *** [install-recursive] Error 1
      Any tips? I have R500
      Last edited by EmbraceUnity; 04-04-2009, 09:14 PM.

      Comment


      • #18
        Originally posted by EmbraceUnity View Post
        Ok, so I have got all the git branches compiled and running, with the exception of the DDX driver. For some reason it won't let me compile it.

        Code:
        radeon_textured_video.c: In function ‘RADEONPutImageTextured’:
        radeon_textured_video.c:505: error: ‘struct _ScrnInfoRec’ has no member named ‘fboffset’
        radeon_textured_video.c:510: error: ‘struct _ScrnInfoRec’ has no member named ‘fboffset’
        radeon_textured_video.c:572: error: ‘struct _ScrnInfoRec’ has no member named ‘fboffset’
        make[1]: *** [radeon_textured_video.lo] Error 1
        make[1]: Leaving directory `/usr/src/xf86-video-ati-airlied/xf86-video-ati/src'
        make: *** [install-recursive] Error 1
        Any tips?
        I ran into this as well.

        I just messaged airlied on #dri-devel and he said,

        <airlied> mattst88|laptop: oops I forgot to push the fix
        It's fixed as of http://cgit.freedesktop.org/~airlied...bd8b4173b9547a
        Last edited by mattst88; 04-04-2009, 09:12 PM.

        Comment


        • #19
          thanks, I just did a git pull and it worked... guess that explains why.

          good work

          now, how do you turn DRI2 on? special xorg.conf setting?
          Last edited by EmbraceUnity; 04-04-2009, 10:57 PM.

          Comment


          • #20
            Well, the new stack works slightly better

            http://embraceunity.com/wp-content/u...ind-radeon.png

            too bad it doesn't go past the title screens.

            Anybody know how to test gallium?
            Last edited by EmbraceUnity; 04-05-2009, 07:38 PM.

            Comment


            • #21
              I just tried this again, last time running glxgears locked my machine up, and my clumsy attempts to add the relevant work from the drm-rawhide kernel tree into a 2.6.28 went nowhere (unsurprisingly), but everything (well, almost) is working very well now.

              KMS works perfectly, the only hangup is xv video, it's like watching something in stop-motion, and lags the entire desktop, is this a known issue?
              Anyways, -vo gl in mplayer works fine, at least 'till this issue is resolved, and apart from that everything works beautifully, so I'm going to stick with it unless something goes wrong.

              I've made a diff of the drm-rawhide kernel branch and the 2.6.29-rc8 kernel, and the resulting patch applies cleanly to 2.6.29.1, anyone interested can get the patch here (the earlier one is the one I'm currently using, the later one includes three commits from the last 24 hours which I haven't booted yet, but it does apply and build just fine).

              Thanks to agd5f for pointing me in the right direction, and all the devs involved for the great work.

              Comment


              • #22
                What's the situation with the Framebuffer drivers?

                I use the VesaFB kernel drivers, but this doesn't work when I compile with KMS, so I get nothing until Xorg starts. I assume this is the whole "if you don't have a kms userspace then everyone you love will suffer a wasting disease" thing, but presumably the kernel FB drivers will require a rewrite to be KMS aware?

                Comment


                • #23
                  Originally posted by DuSTman View Post
                  What's the situation with the Framebuffer drivers?
                  ...but presumably the kernel FB drivers will require a rewrite to be KMS aware?
                  Nope, the kernel can set the console to whatever mode you want using KMS. IOW the console works on top of KMS, making the "old" FB drivers obsolete (at least on HW supported by KMS).

                  Comment


                  • #24
                    Originally posted by DuSTman View Post
                    What's the situation with the Framebuffer drivers?

                    I use the VesaFB kernel drivers, but this doesn't work when I compile with KMS, so I get nothing until Xorg starts. I assume this is the whole "if you don't have a kms userspace then everyone you love will suffer a wasting disease" thing, but presumably the kernel FB drivers will require a rewrite to be KMS aware?
                    The kms drm drivers provide a kernel fb device when they are loaded. You'll need to load fbcon to use it as a console however. They should be considered a replacement for existing kernel fb drivers.

                    Comment

                    Working...
                    X