Announcement

Collapse
No announcement yet.

Radeon R600 Tiling Patches Are Ready

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

  • #46
    Originally posted by glisse View Post
    I pushed a fix to ddx upstream it should be ok now.
    I'm able to log in with startx now, but with ColorTiling2D just starting X or loading any compressed textures (with ColorTiling2D disabled), I get a kernel crash with a BUG on the memcpy in radeon_cs_update_pages on 3.3-rc2 with my 6950.

    Comment


    • #47
      I have applied the two 0001-drm* patches (can you please provide the second patch in less comfortable way than in a mail to a mailinglist complete with headers with the patch attached encoded in base64?)
      Is it normal that git apply doesn't have any output whatsoever?

      But it seems good:
      Code:
      [    13.302] (**) RADEON(0): Option "ColorTiling2D" "on"
      I have only tested xonotic on "high" so far. Looks ok, only some minor problems with reflections in the water.

      HD 6550M, 3.3rc2

      edit:
      No, it doesn't:
      Code:
      xrandr --output HDMI-0 --rotate right --mode 1600x1200 --right-of LVDS --output LVDS --mode 1600x900
      Result=

      Right screen looks good, left screen is very broken. When I move windows on the right screen the broken graphics on the left screen move as well. Screenshot with scrot looks normal for both screens.
      Happens also with no 2d color tiling enabled. Happens only when the right screen is rotated left OR right. With rotate normal it looks ok.
      It also looks good with xrandr --output HDMI-0 --rotate right --mode 1280x1024 --right-of LVDS --output LVDS --mode 1024x768.
      Maybe both need to be the same aspect ratio for rotation?
      The problem are not the kernel patches. Just installed mainline 3.3rc2 and it still happens. My uneducated guess would be the ddx.
      Last edited by ChrisXY; 02-08-2012, 04:56 AM.

      Comment


      • #48
        Hey, if stuff is in git master, couldn't I just install xorg-edgers on ubuntu and recompile the ubuntu kernel with patch?
        If I want to test this that is.

        Comment


        • #49
          Originally posted by PuckPoltergeist View Post
          Video playback with xv still leads to a xserver crash.
          Seems to be fixed by commit f63262e2f112a348c45f0dcecd891c8b6d9c5ee8 from Michel Dänzer.

          Comment


          • #50
            Originally posted by Wilfred View Post
            Hey, if stuff is in git master, couldn't I just install xorg-edgers on ubuntu and recompile the ubuntu kernel with patch?
            If I want to test this that is.
            No, because xorg-edgers probably won't provide a built kernel with the kernel patches.

            The kernel patches are absolutely critical. You can't "skip" them. And you have to use a 3.3-rc or later kernel.

            Comment


            • #51
              Originally posted by ChrisXY View Post
              I have applied the two 0001-drm* patches (can you please provide the second patch in less comfortable way than in a mail to a mailinglist complete with headers with the patch attached encoded in base64?)
              Is it normal that git apply doesn't have any output whatsoever?
              Yes, that's normal. No output is good; complaining is bad.

              Comment


              • #52
                Originally posted by PuckPoltergeist View Post
                Partly ok. I get a short flickering with garbled screen with one message in dmesg:
                Code:
                radeon 0000:01:00.0: evergreen_surface_value_conv_check:329 invalid array mode 5
                radeon 0000:01:00.0: evergreen_packet3_check:1918 invalid cmd stream 454
                [drm:radeon_cs_ib_chunk] *ERROR* Invalid command stream !
                After this it works.
                I have to correct this. The Invalid command stream error came from the xv crash. The garbled screen on X startup doesn't leave any errors in the logs.

                Comment


                • #53
                  Originally posted by allquixotic View Post
                  And you have to use a 3.3-rc or later kernel.
                  hmm running kernel 3.2.5 here. One doesn't have to use the 3.3-rc, 3.2.x + drm patches from 3.3 + 2d patches works too.

                  Comment


                  • #54
                    Nice, the boost in Doom3 seems to be around 37%. (Unfortunately since Mesa has serious problems with this particular game it only means that we go from 19 to 26 fps, but still..)

                    However, after I run any game in Wine the desktop (GNOME fallback so no GL compositing) fails to redraw and the following is logged:

                    Code:
                    [  150.686126] radeon 0000:01:00.0: evergreen_surface_check_linear_aligned:213 cb pitch 1728 invalid must be aligned with 128
                    [  150.686130] radeon 0000:01:00.0: evergreen_cs_track_validate_cb:426 cb[0] invalid (0x000000d7 0x00006f5f 0x00000890 0x01088168)
                    [  150.686133] radeon 0000:01:00.0: evergreen_packet3_check:1918 invalid cmd stream 465
                    [  150.686135] [drm:radeon_cs_ib_chunk] *ERROR* Invalid command stream !
                    (This seems to be an odd place to track these problems, should we file bugs on bugs.freedesktop.org instead?)

                    Comment


                    • #55
                      Originally posted by whizse View Post
                      Nice, the boost in Doom3 seems to be around 37%. (Unfortunately since Mesa has serious problems with this particular game it only means that we go from 19 to 26 fps, but still..)

                      However, after I run any game in Wine the desktop (GNOME fallback so no GL compositing) fails to redraw and the following is logged:

                      Code:
                      [  150.686126] radeon 0000:01:00.0: evergreen_surface_check_linear_aligned:213 cb pitch 1728 invalid must be aligned with 128
                      [  150.686130] radeon 0000:01:00.0: evergreen_cs_track_validate_cb:426 cb[0] invalid (0x000000d7 0x00006f5f 0x00000890 0x01088168)
                      [  150.686133] radeon 0000:01:00.0: evergreen_packet3_check:1918 invalid cmd stream 465
                      [  150.686135] [drm:radeon_cs_ib_chunk] *ERROR* Invalid command stream !
                      (This seems to be an odd place to track these problems, should we file bugs on bugs.freedesktop.org instead?)
                      Yeah better to open bug but first make sure you are using lastest ddx from git few fixes went in.

                      Comment


                      • #56
                        Hi-Z patches for R600 has been updated if someone is interested: https://bugs.freedesktop.org/show_bug.cgi?id=36602
                        ## VGA ##
                        AMD: X1950XTX, HD3870, HD5870
                        Intel: GMA45, HD3000 (Core i5 2500K)

                        Comment


                        • #57
                          Yes! It's finally all coming together w00t w00t!

                          Comment


                          • #58
                            Originally posted by glisse View Post
                            Yeah better to open bug but first make sure you are using lastest ddx from git few fixes went in.
                            Already using latest git. Bug is filed at
                            https://bugs.freedesktop.org/show_bug.cgi?id=45827

                            I also noticed that if ColorTiling2D is not turned on I get a corruption in X after running a game using Mesa with the tiling patches applied. Is this another bug?

                            Comment


                            • #59
                              After rebuild from git everything works.
                              Must have done somethig wrong when patching.

                              Comment


                              • #60
                                my display card is 6950

                                i apply the patch to kernel 3.3-rc3, when glxgear

                                xorg will crash, anyone has similar problem?

                                Feb 10 08:12:40 earth kernel: [ 55.982929] radeon 0000:01:00.0: GPU lockup CP stall for more than 10020msec
                                Feb 10 08:12:40 earth kernel: [ 55.982931] GPU lockup (waiting for 0x000001B6 last fence id 0x000001B2)
                                Feb 10 08:12:40 earth kernel: [ 55.984481] radeon 0000:01:00.0: GPU softreset
                                Feb 10 08:12:40 earth kernel: [ 55.984482] radeon 0000:01:00.0: GRBM_STATUS=0xE55009A0
                                Feb 10 08:12:40 earth kernel: [ 55.984484] radeon 0000:01:00.0: GRBM_STATUS_SE0=0xE4000001
                                Feb 10 08:12:40 earth kernel: [ 55.984485] radeon 0000:01:00.0: GRBM_STATUS_SE1=0xEC000001
                                Feb 10 08:12:40 earth kernel: [ 55.984486] radeon 0000:01:00.0: SRBM_STATUS=0x20020FC0
                                Feb 10 08:12:40 earth kernel: [ 55.984487] radeon 0000:01:00.0: VM_CONTEXT0_PROTECTION_FAULT_ADDR 0x00000000
                                Feb 10 08:12:40 earth kernel: [ 55.984489] radeon 0000:01:00.0: VM_CONTEXT0_PROTECTION_FAULT_STATUS 0x00000000
                                Feb 10 08:12:40 earth kernel: [ 55.984490] radeon 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x00000817
                                Feb 10 08:12:40 earth kernel: [ 55.984492] radeon 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x03020014
                                Feb 10 08:12:41 earth kernel: [ 56.117912] radeon 0000:01:00.0: Wait for MC idle timedout !
                                Feb 10 08:12:41 earth kernel: [ 56.117913] radeon 0000:01:00.0: GRBM_SOFT_RESET=0x0000DF7B
                                Feb 10 08:12:41 earth kernel: [ 56.118015] radeon 0000:01:00.0: GRBM_STATUS=0x00003828
                                Feb 10 08:12:41 earth kernel: [ 56.118016] radeon 0000:01:00.0: GRBM_STATUS_SE0=0x00000007
                                Feb 10 08:12:41 earth kernel: [ 56.118017] radeon 0000:01:00.0: GRBM_STATUS_SE1=0x00000007
                                Feb 10 08:12:41 earth kernel: [ 56.118018] radeon 0000:01:00.0: SRBM_STATUS=0x20020FC0
                                Feb 10 08:12:41 earth kernel: [ 56.119020] radeon 0000:01:00.0: GPU reset succeed
                                Feb 10 08:12:41 earth kernel: [ 56.285111] radeon 0000:01:00.0: Wait for MC idle timedout !
                                Feb 10 08:12:41 earth kernel: [ 56.428836] radeon 0000:01:00.0: Wait for MC idle timedout !
                                Feb 10 08:12:41 earth kernel: [ 56.430355] [drm] PCIE GART of 512M enabled (table at 0x0000000000040000).
                                Feb 10 08:12:41 earth kernel: [ 56.430425] radeon 0000:01:00.0: WB enabled
                                Feb 10 08:12:41 earth kernel: [ 56.430426] [drm] fence driver on ring 0 use gpu addr 0x80000c00 and cpu addr 0xffff880427beec00
                                Feb 10 08:12:41 earth kernel: [ 56.430427] [drm] fence driver on ring 1 use gpu addr 0x80000c04 and cpu addr 0xffff880427beec04
                                Feb 10 08:12:41 earth kernel: [ 56.430428] [drm] fence driver on ring 2 use gpu addr 0x80000c08 and cpu addr 0xffff880427beec08
                                Feb 10 08:12:41 earth kernel: [ 56.598995] [drm:r600_ring_test] *ERROR* radeon: ring 0 test failed (scratch(0x8500)=0xCAFEDEAD)
                                Feb 10 08:12:41 earth kernel: [ 56.598996] [drm:cayman_resume] *ERROR* cayman startup failed on resume
                                Feb 10 08:12:44 earth kernel: [ 56.602042] BUG: unable to handle kernel paging request at ffffc90007986900
                                Feb 10 08:12:44 earth kernel: [ 56.602064] IP: [<ffffffff812b1aab>] memcpy+0xb/0x120
                                Feb 10 08:12:44 earth kernel: [ 56.602080] PGD 42e80f067 PUD 42e820067 PMD 42750d067 PTE 0
                                Feb 10 08:12:44 earth kernel: [ 56.602097] Oops: 0002 [#1] PREEMPT SMP
                                Feb 10 08:12:44 earth kernel: [ 56.602117] CPU 1
                                Feb 10 08:12:44 earth kernel: [ 56.602123] Modules linked in:
                                Feb 10 08:12:44 earth kernel: [ 56.602131]
                                Feb 10 08:12:44 earth kernel: [ 56.602136] Pid: 1102, comm: Xorg Not tainted 3.3.0-rc3-custom2 #1 System manufacturer System Product Name/P8Z68-V LE
                                Feb 10 08:12:44 earth kernel: [ 56.602162] RIP: 0010:[<ffffffff812b1aab>] [<ffffffff812b1aab>] memcpy+0xb/0x120
                                Feb 10 08:12:44 earth kernel: [ 56.602187] RSP: 0018:ffff8804261c7b30 EFLAGS: 00010202
                                Feb 10 08:12:44 earth kernel: [ 56.602199] RAX: ffffc90007986900 RBX: ffff880428dce680 RCX: 00000000000000eb
                                Feb 10 08:12:44 earth kernel: [ 56.602214] RDX: 0000000000000004 RSI: ffff880425c5c000 RDI: ffffc90007986900
                                Feb 10 08:12:44 earth kernel: [ 56.602229] RBP: ffff8804261c7b88 R08: 0000000000000000 R09: ffff88041193a800
                                Feb 10 08:12:44 earth kernel: [ 56.602244] R10: ffff88042e84ee28 R11: ffff88042e84edf8 R12: ffff8804261c7cc8
                                Feb 10 08:12:44 earth kernel: [ 56.602259] R13: 0000000000000000 R14: 0000000000000001 R15: 0000000000000000
                                Feb 10 08:12:44 earth kernel: [ 56.602274] FS: 00007fb4d86d88a0(0000) GS:ffff88043f440000(0000) knlGS:0000000000000000
                                Feb 10 08:12:44 earth kernel: [ 56.602291] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
                                Feb 10 08:12:44 earth kernel: [ 56.602304] CR2: ffffc90007986900 CR3: 0000000425e88000 CR4: 00000000000406e0
                                Feb 10 08:12:44 earth kernel: [ 56.602319] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
                                Feb 10 08:12:44 earth kernel: [ 56.602334] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
                                Feb 10 08:12:44 earth kernel: [ 56.602349] Process Xorg (pid: 1102, threadinfo ffff8804261c6000, task ffff8804259f4440)
                                Feb 10 08:12:44 earth kernel: [ 56.602366] Stack:
                                Feb 10 08:12:44 earth kernel: [ 56.602371] ffffffff813dc02b ffff880428dbc720 0000000000000003 000000000000075c
                                Feb 10 08:12:44 earth kernel: [ 56.602390] 0000000000000001 0000000000003286 ffff880428dce680 ffff8804261c7cc8
                                Feb 10 08:12:44 earth kernel: [ 56.602409] 0000000000000000 ffff88042e84f578 00000000fffffff4 ffff8804261c7bb8
                                Feb 10 08:12:44 earth kernel: [ 56.602428] Call Trace:
                                Feb 10 08:12:44 earth kernel: [ 56.602436] [<ffffffff813dc02b>] ? radeon_cs_update_pages+0xfb/0x130
                                Feb 10 08:12:44 earth kernel: [ 56.602452] [<ffffffff813d9883>] radeon_get_ib_value+0x53/0xd0
                                Feb 10 08:12:44 earth kernel: [ 56.602465] [<ffffffff8140f543>] evergreen_cs_packet_parse+0x43/0x160
                                Feb 10 08:12:44 earth kernel: [ 56.602480] [<ffffffff814131ae>] evergreen_cs_parse+0x2e/0x710
                                Feb 10 08:12:44 earth kernel: [ 56.602494] [<ffffffff813d9b95>] ? radeon_ib_get+0x1b5/0x2f0
                                Feb 10 08:12:44 earth kernel: [ 56.602506] [<ffffffff813dbd80>] radeon_cs_ioctl+0x350/0x500
                                Feb 10 08:12:44 earth kernel: [ 56.602520] [<ffffffff8137a4b4>] drm_ioctl+0x434/0x500
                                Feb 10 08:12:44 earth kernel: [ 56.602532] [<ffffffff813dba30>] ? radeon_cs_finish_pages+0xb0/0xb0
                                Feb 10 08:12:44 earth kernel: [ 56.602547] [<ffffffff817adfac>] ? do_page_fault+0x2ac/0x520
                                Feb 10 08:12:44 earth kernel: [ 56.602561] [<ffffffff81137f5e>] do_vfs_ioctl+0x8e/0x500
                                Feb 10 08:12:44 earth kernel: [ 56.602574] [<ffffffff810fd3ec>] ? sys_mmap_pgoff+0xdc/0x220
                                Feb 10 08:12:44 earth kernel: [ 56.602587] [<ffffffff81138461>] sys_ioctl+0x91/0xa0
                                Feb 10 08:12:44 earth kernel: [ 56.602599] [<ffffffff817b0622>] system_call_fastpath+0x16/0x1b
                                Feb 10 08:12:44 earth kernel: [ 56.602612] Code: 43 4e 48 83 c4 08 5b 5d c3 0f 1f 80 00 00 00 00 e8 8b fb ff ff eb e1 90 90 90 90 90 90 90 90 90 48 89 f8 89 d1 c1 e9 03 83 e2 07 <f3> 48 a5 89 d1 f3 a4 c3 20 48 83 ea 20 4c 8b 06 4c 8b 4e 08 4c
                                Feb 10 08:12:44 earth kernel: [ 56.602722] RIP [<ffffffff812b1aab>] memcpy+0xb/0x120
                                Feb 10 08:12:44 earth kernel: [ 56.602735] RSP <ffff8804261c7b30>
                                Feb 10 08:12:44 earth kernel: [ 56.602743] CR2: ffffc90007986900
                                Feb 10 08:12:44 earth kernel: [ 56.608340] ---[ end trace 284d340da3940897 ]---

                                Comment

                                Working...
                                X