Announcement

Collapse
No announcement yet.

An Updated ATI Kernel Mode-Setting Driver

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

  • #16
    Originally posted by bridgman View Post
    AFAIK the primary reason X needed root privileges was that the X drivers accessed hardware directly. Once that stops (KMS moves the hardware accesses into the kernel) then in principle you don't need root privileges for X any more. There are probably a few other things that need to be tweaked but airlied first ran non-root X over a year ago with early KMS.
    Following the message chain, the "how to do it properly" is still somewhat of a work in progress. This post http://lists.x.org/archives/xorg-dev...ly/001296.html is imo one of the most interesting in the chain.

    Comment


    • #17
      Just wondering if any of this pulls have something to improve (KMS/GEM) AGP (PCIE Rialto Chipset) situation?

      As of yesterday, I still couldn't even boot using it.

      Monitor enters on save energy mode and the system freezes.

      AGP X1600pro (Karmic Koala)

      Anyone?

      Comment


      • #18
        About implementing KMS in Catalyst:
        Originally posted by mtippett View Post
        There are no plans at this stage. Most people I have spoken to don't have an interest in seeing kernel messages appear on their screen. Last time I saw that in a production system was on old Sparc X Terminals.

        Beyond that (which can already be achieved with a kernel based chvt + message to that console (aka the Linux BSOD)), it is about avoiding slow switches during boot and fast chvt's. We haven't seen any direct requirements for these. THe intel chvt takes a second or two, under fglrx takes about half a second.

        The GEM/TTM/other parts of what people call KMS we have already had since the R6xx family was enabled.

        Regards,

        Matthew

        Comment


        • #19
          Originally posted by NeoBrain View Post
          About implementing KMS in Catalyst:
          Then, I would have answered "it is possible in the future using radeon.ko" and so, catalyst would have switched to TTM and would have used radeon.ko for modesetting. But I don't really know if the main difficulty is in the kernel or in the driver :s
          the main dificulty is of legal nature.

          the way proprietary ati and nvidia drivers are right now - small interface compiled against the kernel which gets linked with closed source blob to make a kernel module - is still debatable to many people. it's already been discussed and remains a headache, because apparently every person has a different opinion on the subject (that's why there are distributions that don't even ship with firmware).

          if the drivers would move to use kms that would mean that they would have to interface with the kernel even more than they do now. which might ignite the discussion yet again.

          but that's a highly technical topic anyway. and i'm no expert.

          of course it should be the way to go, since it would allow for root-less X server with proprietary drivers.

          Comment


          • #20
            Originally posted by yoshi314 View Post
            of course it should be the way to go, since it would allow for root-less X server with proprietary drivers.
            I'm not sure the internals of the radeon DRM are compatible between the free and the closed driver. In fact, I'm pretty sure it is not.
            So, we'll have to wait for AMD :s. But mtippett said r6xx and following are already using GEM, so, the transition to KMS should be less painful than what happened with the free driver.

            Comment


            • #21
              Originally posted by MPF View Post
              Last time I tried KMS, it was already almost stable. So, with theses patchs, I guess this should be stable enough to be used by users who know what they are doing
              Also, it should be available in the 2.6.32 that should be out within 3 months.

              Concerning root-less X server, my bet that it is already stable but not in good-enough shape to enter the X-server mainline
              Originally posted by Louise View Post
              Can KMS be used together with the Catalyst drivers?



              So it doesn't require a major rewrite of X to make it root-less?
              1. As far as I know, Catalyst won't in the foreseeable future use KMS (I assume it's because they want as much code the be the same for all OSes).

              2. From what I've read, rootless X-server is almost trivial to do once KMS is stable (like removing a sudo prefix, only a tiny bit more complicated). On any account, it won't take major X.org rewrites to make it run rootless.

              Comment


              • #22
                Originally posted by MPF View Post
                I'm not sure the internals of the radeon DRM are compatible between the free and the closed driver. In fact, I'm pretty sure it is not.
                So, we'll have to wait for AMD :s. But mtippett said r6xx and following are already using GEM, so, the transition to KMS should be less painful than what happened with the free driver.
                There is very little compatibility between the open and Catalyst driver internals. The Catalyst drivers have had a high performance kernel memory manager for a number of years, with similar capabilities to GEM/TTM plus a few years of performance tuning, and bringing that memory manager into fglrx was one part of the new 3D stack we released back in September 07.

                The memory manager is tightly integrated with both the 3D drivers and the low-level GPU management across all OSes, so switching to GEM/TTM is not a practical option.
                Last edited by bridgman; 07-29-2009, 02:15 PM.

                Comment


                • #23
                  Originally posted by bridgman View Post
                  There is very little compatibility between the open and Catalyst driver internals. The Catalyst drivers have had a high performance kernel memory manager for a number of years, with similar capabilities to GEM/TTM plus a few years of performance tuning, and bringing that memory manager into fglrx was one part of the new 3D stack we released back in September 07.

                  The memory manager is tightly integrated with both the 3D drivers and the low-level GPU management across all OSes, so switching to GEM/TTM is not a practical option.
                  So, what about KMS with catalyst ?
                  I understand the little interest for it for professionals.
                  About us, by the time KMS will be used, oss drivers for r100 to r700 should be able to perform fairly for desktop usage (with KMS/DRI2 and gallium3D).

                  Comment


                  • #24
                    Realistically we would need to implement KMS inside the fglrx kernel driver and have fglrx's kernel module come up at boot time rather than the open source drm. Any other approach would seem to result in losing things like MultiView (accelerated 3D across multiple screens and multiple GPUs) and Crossfire.

                    Comment


                    • #25
                      Pulled: http://git.kernel.org/?p=linux/kerne...026bad630cd3da

                      Comment


                      • #26
                        Oh yeah !! Well done Linus !

                        I'll wait for the rc5, I've had some troubles with my filesystem using the 2.6.31. I don't know if it is because my computer was crashing (KMS) or if it was a real problem of the ext4 driver.

                        Comment


                        • #27
                          Originally posted by MPF View Post
                          Oh yeah !! Well done Linus !

                          I'll wait for the rc5, I've had some troubles with my filesystem using the 2.6.31. I don't know if it is because my computer was crashing (KMS) or if it was a real problem of the ext4 driver.
                          I had the same problem - my filesystem crashed twice on 2.6.31-rc2 with KMS enabled and I stopped experimenting (but I had ext3).

                          Comment


                          • #28
                            Should compiz run on TTM on radeons ( r200 ) with decent performance?

                            I've made quite a mess of my system pulling in various unstable branches of things, and I've got X reporting DRI2 is initialising, and I can run glxgears ( woohoo! ), but when I start ecomorph ( compiz on e17 ), it runs at about 1 frame every 5 seconds

                            Comment


                            • #29
                              Originally posted by dkasak View Post
                              Should compiz run on TTM on radeons ( r200 ) with decent performance?

                              I've made quite a mess of my system pulling in various unstable branches of things, and I've got X reporting DRI2 is initialising, and I can run glxgears ( woohoo! ), but when I start ecomorph ( compiz on e17 ), it runs at about 1 frame every 5 seconds
                              Performances with KMS/DRI2 (I assume that's what you mean) are quite bad when using a compositor. Otherwise, it is already pretty fast, I was even able to play flash videos in fullscreen on my r300 (XPress 200M) on my fullHD screen. Something that even my new HD4770 can't do at the moment with fglrx.

                              Nevertheless, it was not _that_ slow on my computer. A lot of work is done for the r300, so, I assume It already had a bit of optimisation.

                              Comment


                              • #30
                                Originally posted by MPF View Post
                                Performances with KMS/DRI2 (I assume that's what you mean) are quite bad when using a compositor.
                                Really? Yes well that's how I'd describe performance at this point I suppose, so I'm not surprised. Disappointed. Not surprised. I've already had to disable AGP and AccelDFS to get rid of graphics corruption under EXA ( and DRI1 ). It's hard to avoid the feeling that I'm going backwards.

                                Comment

                                Working...
                                X