2

I'm having some problems with the graphical configuration of a fresh Debian 8 install: things like HTML Canvas, Videos, Games etc. are insanely slow.

I spent some time trying to fix things in the xorg.conf file but it hasn't changed anything atm .

Here is the Device section of my xorg.conf file:

Section "Device"
    Identifier  "Intel Graphics"
    Driver      "modesetting"
    Option      "AccelMethod" "glamor"
    BusID       "PCI:0:2:0"
EndSection

I get this in glxinfo:

OpenGL vendor string: VMware, Inc.
OpenGL renderer string: Gallium 0.4 on llvmpipe (LLVM 3.5, 256 bits)
OpenGL version string: 3.0 Mesa 10.3.2
OpenGL shading language version string: 1.30

And the informations I find relevant in the Xorg.log file:

[     1.689] (II) Module ABI versions:
[     1.689]    X.Org ANSI C Emulation: 0.4
[     1.689]    X.Org Video Driver: 18.0
[     1.689]    X.Org XInput driver : 21.0
[     1.689]    X.Org Server Extension : 8.0
[     1.689] (II) xfree86: Adding drm device (/dev/dri/card0)
[     2.427] (--) PCI:*(0:0:2:0) 8086:5912:1458:d000 rev 4, Mem @ 0xee000000/16777216, 0xd0000000/268435456, I/O @ 0x0000f000/64, BIOS @ 0x????????/131072
[     2.428] (II) "glx" will be loaded. This was enabled by default and also specified in the config file.
[     2.428] (II) LoadModule: "glx"
[     2.430] (II) Loading /usr/lib/xorg/modules/extensions/libglx.so
[     2.436] (II) Module glx: vendor="X.Org Foundation"
[     2.436]    compiled for 1.16.4, module version = 1.0.0
[     2.436]    ABI class: X.Org Server Extension, version 8.0
[     2.436] (==) AIGLX enabled
[     2.436] (II) LoadModule: "dri2"
[     2.436] (II) Module "dri2" already built-in
[     2.436] (II) LoadModule: "glamoregl"
[     2.436] (II) Loading /usr/lib/xorg/modules/libglamoregl.so
[     2.445] (II) Module glamoregl: vendor="X.Org Foundation"
[     2.445]    compiled for 1.16.4, module version = 1.0.0
[     2.445]    ABI class: X.Org ANSI C Emulation, version 0.4
[     2.445] (II) LoadModule: "modesetting"
[     2.445] (II) Loading /usr/lib/xorg/modules/drivers/modesetting_drv.so
[     2.446] (II) Module modesetting: vendor="X.Org Foundation"
[     2.446]    compiled for 1.16.4, module version = 0.9.0
[     2.446]    Module class: X.Org Video Driver
[     2.446]    ABI class: X.Org Video Driver, version 18.0
[     2.446] (II) modesetting: Driver for Modesetting Kernel Drivers: kms
...
[     3.916] (==) modesetting(0): Backing store enabled
[     3.916] (==) modesetting(0): Silken mouse enabled
[     3.916] (II) modesetting(0): RandR 1.2 enabled, ignore the following RandR disabled message.
[     3.977] (==) modesetting(0): DPMS enabled
[     3.977] (WW) modesetting(0): Option "AccelMethod" is not used
[     4.077] (--) RandR disabled
[     4.087] (II) SELinux: Disabled on system
[     4.088] (II) AIGLX: Screen 0 is not DRI2 capable
[     4.088] (EE) AIGLX: reverting to software rendering
[     4.150] (II) AIGLX: Loaded and initialized swrast
[     4.150] (II) GLX: Initialized DRISWRAST GL provider for screen 0

Here is my kernel version : 4.11.0-trunk-amd64 #1 SMP Debian 4.11-1~exp2 (2017-05-05) x86_64 GNU/Linux

Thanks for your help, feel free to answer more details if you want.

Edit : Updating to Debian 9 fixed the problem

Julien Dutriaux
  • 31
  • 8
  • 1
    Obviously your kernel is new enough, but I suspect you’re missing the firmware (you’re running with software 3D, no acceleration). – Stephen Kitt May 12 '17 at 12:25
  • Thanks for taking time to answer Stephen, I tried the solution you gave on the other question but it didn't changed anything. Intel firmware is worst as I get tons of artifacts and glitches. – Julien Dutriaux May 12 '17 at 12:29
  • Could you try with the 4.9 kernel from Jessie backports? That gives good results with the packaged firmware on Kaby Lake systems. In any case you can’t get 3D acceleration on modern Intel GPUs without the firmware. – Stephen Kitt May 12 '17 at 12:31
  • I should mention that it was the first kernel I tried, sorry. I upgraded to the 4.11 because I read that there was improvements on Kaby Lake support in 4.10 for the power management and 4.11 for the audio. Audio is working well now but I doesn't changed anything for graphics. Are you sure that my question is a duplicate of the one you gave? – Julien Dutriaux May 12 '17 at 12:36
  • For graphics support, yes, I’m pretty sure the answer to the other question is relevant. I would expect the 4.9 kernel with associated firmware to behave correctly on Kaby Lake, as far as graphics go at least. Later kernels may need newer firmware than what’s available in the packages. – Stephen Kitt May 12 '17 at 12:42
  • I should add that I’m perfectly willing to re-open the question if you think that’s appropriate. – Stephen Kitt May 12 '17 at 12:42
  • Okay, I will try again with the 4.9 kernel. What makes me think that it's not a duplicate is because I don't get errors like "no screens found". Thanks again for taking time to answer – Julien Dutriaux May 12 '17 at 12:44
  • Hi, I have exactly the same problem with 4.9 Kernel :(. – Julien Dutriaux May 15 '17 at 10:59
  • With the firmware installed? – Stephen Kitt May 15 '17 at 11:10
  • If you *do* have the firmware installed, you could try deleting the `Device` section of your `xorg.conf`, and perhaps upgrading Mesa (start with `libgl1-mesa-dri`, version 13 is available in Jessie backports). – Stephen Kitt May 15 '17 at 11:14
  • Thanks for your help but unfortunately I still have the problem :(. When I deleted the Device section, Xorg took back the Intel driver which still show so much graphical artifacts. Uninstalling `xserver-xorg-video-intel` made Xorg use modesetting but the problem is still here. My `Xorg.log` is the same as before except that there is no line saying `Option "AccelMethod" is not used`. – Julien Dutriaux May 15 '17 at 14:27
  • You will find the full xorg.log file [here](https://gist.github.com/MrHalfman/8f2996e97b0c4edefaf8ff919ff89ae4) (My last test was made on 4.11, problem is the same on 4.9) – Julien Dutriaux May 15 '17 at 14:29
  • You’re still getting the “AIGLX: Screen 0 is not DRI2 capable” message, which suggests the firmware isn’t being loaded. Are there any `i915` firmware-related messages in `dmesg`? – Stephen Kitt May 15 '17 at 15:11
  • I think that you are right, [here](https://gist.github.com/MrHalfman/8b85364035c46beac7029ff064e548dc) is what I get from a `dmesg | grep i915`. It seems like the driver is not fully loaded. – Julien Dutriaux May 16 '17 at 08:22

0 Answers0