No announcement yet.

Ryzen 5 Mobile / Raven Ridge no direct rendering

  • Filter
  • Time
  • Show
Clear All
new posts

  • Ryzen 5 Mobile / Raven Ridge no direct rendering

    I recently purchased a Ryzen 5 mobile system with raven ridge gpu.
    In X I don´t get direct rendering, KDE is hardly usable that way.....

    Below I´d like to post some information about my system and pastebin some logs.
    I would be very happy, if anyone has some hints!

    lscpi detection
    VGA compatible controller [0300]: Advanced Micro Devices, Inc. [AMD/ATI] Device [1002:15dd] (rev c4)
            Subsystem: Acer Incorporated [ALI] Device [1025:1233]
            Kernel driver in use: amdgpu
    Kernel: 4.15.0 no options passed to amdgpu, amdgpu.dc=1 amdgpu.ngg=1 cause system to hang at boot

    /lib/firmware/amdgpu/raven* is on the system

    Output from dmesg | grep -i raven

    [    4.895388] [drm] initializing kernel modesetting (RAVEN 0x1002:0x15DD 0x1025:0x1233 0xC4).
    [    4.895640] ATOM BIOS: 113-RAVEN-102
    Some important lines from Xorg.0.log:

    (==) AMDGPU(0): DRI3 disabled
    [    17.956] (II) AIGLX: Screen 0 is not DRI2 capable
    [    17.958] (II) GLX: Initialized DRISWRAST GL provider for screen 0
    (II) glamor: EGL version 1.4 (DRI2):
    [    17.941] EGL_MESA_drm_image required.
    [    17.941] (EE) AMDGPU(0): glamor detected, failed to initialize EGL.
     (EE) AMDGPU(0): glamor detected, failed to initialize EGL.
    [    17.956] (EE) AIGLX: reverting to software rendering
    Output from LIBGL_DEBUG=verbose glxgears

    libGL: screen 0 does not appear to be DRI2 capable
    libGL: OpenDriver: trying /usr/lib64/dri/tls/
    libGL: OpenDriver: trying /usr/lib64/dri/
    3974 frames in 5.0 seconds = 794.729 FPS
    4112 frames in 5.0 seconds = 822.258 FPS
    XIO:  fatal IO error 11 (Resource temporarily unavailable) on X server ":0"
          after 26358 requests (26356 known processed) with 0 events remaining.
    Software versions used

    Kernel 4.15.0
    (libGL and libEGL also 17.1.6)
    opensuse 42.3

  • #2
    The libegl-nvidia0 and libegl1 packages (nvidias libglvnd_1.0.0 source package) cause this and removing it and a lot of qt5 apps that depends on it fixes the problem.


    [ 6.245] (EE) AMDGPU(0): eglGetDisplay() failed
    [ 6.245] (EE) AMDGPU(0): glamor detected, failed to initialize EGL.

    [ 6.452] (II) AMDGPU(0): Front buffer pitch: 7680 bytes
    [ 6.452] (==) AMDGPU(0): DRI3 disabled
    [ 6.452] (==) AMDGPU(0): Backing store enabled
    [ 6.452] (WW) AMDGPU(0): Direct rendering disabled
    [ 6.452] (II) AMDGPU(0): 2D and 3D acceleration disabled
    [ 6.452] (==) AMDGPU(0): DPMS enabled
    [ 6.452] (==) AMDGPU(0): Silken mouse enabled
    [ 6.453] (II) AMDGPU(0): RandR 1.2 enabled, ignore the following RandR disabled message.
    [ 6.489] (--) RandR disabled
    [ 6.491] (II) SELinux: Disabled on system
    [ 6.491] (II) AIGLX: Screen 0 is not DRI2 capable
    [ 6.491] (EE) AIGLX: reverting to software rendering
    [ 6.499] (II) IGLX: enabled GLX_MESA_copy_sub_buffer
    [ 6.500] (II) IGLX: Loaded and initialized swrast
    [ 6.500] (II) GLX: Initialized DRISWRAST GL provider for screen 0
    Last edited by debianxfce; 02-03-2018, 12:16 PM.


    • #3
      Thanks for this hint.

      I have these egl packages installed:

      ~$ rpm -qa | grep -i egl

      But, removing Mesa-libEGL would kill virtually my whole KDE setup.... RPM wants to remove more than 1000 packages, when trying to remove Mesa-libEGL, it does not look like good idea.....
      Do I understand correctly, EGL is the root of trouble here? It seems almost the complete KDE5 setup seem to depend on it?!


      • #4
        Originally posted by Fernseher View Post
        Thanks for this hint.
        Do I understand correctly, EGL is the root of trouble here? It seems almost the complete KDE5 setup seem to depend on it?!
        The libegl package does have nvidias libglvnd_1.0.0 source package that broke direct rendering. All QT5 packages depends of it. Arch Linux users had this bug in summer 2017, now it is in suse and debian testing. Thanks to package maintainers that test only with nvidia gpus. Amdgpu users does not not need libegl package so thanks to qt5 maintainers too. I made a 40GB partition and tested with win10 that my rx560 is not broken. Then I tested with Manjaro Linux Xfce live and installed it and using it now.

        [[email protected] ~]$ inxi -bM
        System:    Host: xfce-pc Kernel: 4.15.0-1-MANJARO x86_64 bits: 64
                   Desktop: N/A Distro: Manjaro Linux
        Machine:   Device: desktop Mobo: ASUSTeK model: PRIME B350M-K v: Rev X.0x serial: N/A
                   UEFI [Legacy]: American Megatrends v: 3401 date: 12/04/2017
        CPU:       6 core AMD Ryzen 5 1600 Six-Core (-MT-MCP-)
                   speed/max: 1371/3200 MHz
        Graphics:  Card: Advanced Micro Devices [AMD/ATI] Baffin [Radeon RX 560]
                   Display Server: x11 (X.Org 1.19.6 )
                   drivers: amdgpu (unloaded: modesetting)
                   Resolution: [email protected]
                   OpenGL: renderer: Radeon RX 560 Series (POLARIS11 / DRM 3.23.0 / 4.15.0-1-MANJARO, LLVM 5.0.1)
                   version: 4.5 Mesa 17.3.3
        Network:   Card: Realtek RTL8111/8168/8411 PCIE Gigabit Ethernet Controller
                   driver: r8168
        Drives:    HDD Total Size: 256.1GB (70.8% used)
        Info:      Processes: 292 Uptime: 3:05 Memory: 1446.3/7978.8MB


        • #5
          I fixed this bug by installing libegl-mesa0 from the experimental repository to Debian sid.

          Great, no need to change distribution and wipe my work partition.


          • #6
            Raven ridge requires latest ~agd5f/linux/log/?h=drm-next-4.17-wip kernel. For HVEC latest Mesa git too. If not using HVEC, my distribution might work:

            More Raven info: