[fglrx] compiz will not launch with fglrx driver - falls back to metacity on xserver 1.4 (1.3 is ok)

[Edward Clark]

Binary package hint: linux-restricted-modules-2.6.22-14-generic

Ubuntu Hardy Heron
fglrx on compiz whitelist
Working with Miguel Martinez on this. http://ubuntuforums.org/showthread.php?t=623882&page=3

edward@smallville:~$ compiz --replace
Checking for Xgl: not present.
AIEEEEH, no Log file found
Keyboard Control:
  auto repeat: on key click percent: 0 LED mask: 00000000
  auto repeat delay: 500 repeat rate: 30
  auto repeating keys: 00ffffffdffffbbf
                        fadfffdfffdfe5ef
                        ffffffffffffffff
                        ffffffffffffffff
  bell percent: 50 bell pitch: 400 bell duration: 100
Pointer Control:
  acceleration: 2/1 threshold: 4
Screen Saver:
  prefer blanking: yes allow exposures: yes
  timeout: 0 cycle: 0
Colors:
  default colormap: 0x20 BlackPixel: 0 WhitePixel: 16777215
Font Path:
  /home/edward/.gnome2/share/cursor-fonts,/usr/share/fonts/X11/misc,/usr/share/fonts/X11/100dpi/:unscaled,/usr/share/fonts/X11/75dpi/:unscaled,/usr/share/fonts/X11/Type1,/usr/share/fonts/X11/100dpi,/usr/share/fonts/X11/75dpi,/var/lib/defoma/x-ttcidfont-conf.d/dirs/TrueType,/home/edward/.gnome2/share/fonts
Bug Mode: compatibility mode is disabled
DPMS (Energy Star):
  Standby: 0 Suspend: 0 Off: 0
  DPMS is Enabled
  Monitor is On
Detected PCI ID for VGA: 01:00.0 0300: 1002:4e50 (prog-if 00 [VGA])
        Flags: bus master, VGA palette snoop, 66MHz, medium devsel, latency 255, IRQ 11
Checking for texture_from_pixmap: not present.
Trying again with indirect rendering:
Checking for texture_from_pixmap: not present.
aborting and using fallback: /usr/bin/metacity

[Edward Clark (eeclark)]

Card
ATI 9600 Pro Turbo

[Edward Clark (eeclark)]

GLX_EXT_texture_from_pixmap is present but compiz says it is not (Checking for texture_from_pixmap: not present. )

edward@smallville:~$ glxinfo | grep GLX_EXT_texture
    GLX_ARB_multisample, GLX_EXT_import_context, GLX_EXT_texture_from_pixmap,
    GLX_SGIX_visual_select_group, GLX_EXT_texture_from_pixmap
edward@smallville:~$ compiz --replace
Checking for Xgl: not present.
AIEEEEH, no Log file found
Keyboard Control:
  auto repeat: on key click percent: 0 LED mask: 00000000
  auto repeat delay: 500 repeat rate: 30
  auto repeating keys: 00ffffffdffffbbf
                        fadfffdfffdfe5ef
                        ffffffffffffffff
                        ffffffffffffffff
  bell percent: 50 bell pitch: 400 bell duration: 100
Pointer Control:
  acceleration: 2/1 threshold: 4
Screen Saver:
  prefer blanking: yes allow exposures: yes
  timeout: 0 cycle: 0
Colors:
  default colormap: 0x20 BlackPixel: 0 WhitePixel: 16777215
Font Path:
  /home/edward/.gnome2/share/cursor-fonts,/usr/share/fonts/X11/misc,/usr/share/fonts/X11/100dpi/:unscaled,/usr/share/fonts/X11/75dpi/:unscaled,/usr/share/fonts/X11/Type1,/usr/share/fonts/X11/100dpi,/usr/share/fonts/X11/75dpi,/var/lib/defoma/x-ttcidfont-conf.d/dirs/TrueType,/home/edward/.gnome2/share/fonts
Bug Mode: compatibility mode is disabled
DPMS (Energy Star):
  Standby: 0 Suspend: 0 Off: 0
  DPMS is Enabled
  Monitor is On
Detected PCI ID for VGA: 01:00.0 0300: 1002:4e50 (prog-if 00 [VGA])
        Flags: bus master, VGA palette snoop, 66MHz, medium devsel, latency 255, IRQ 11
Checking for texture_from_pixmap: not present.
Trying again with indirect rendering:
Checking for texture_from_pixmap: not present.
aborting and using fallback: /usr/bin/metacity

[erdtmann (erdtmann)]

I'd like to confirm the bug on a ATI X700. If SKIP_CHECKS is used, compiz fails later (caused by the same problem):
(...)
SKIP_CHECKS is yes, so continuing despite problems.
Checking for nVidia: not present.
Checking for FBConfig: present.
Checking for Xgl: not present.
Starting gtk-window-decorator
/usr/bin/compiz.real (core) - Fatal: GLX_EXT_texture_from_pixmap is missing
/usr/bin/compiz.real (core) - Error: Failed to manage screen: 0
/usr/bin/compiz.real (core) - Fatal: No manageable screens found on display :0.0

[Joe Miller (joe-miller1079)]

ATI Radeon Xpress 200 Series

joe@joe-desktop:~$ glxinfo | grep GLX_EXT_texture
    GLX_ARB_multisample, GLX_EXT_import_context, GLX_EXT_texture_from_pixmap,
    GLX_SGIX_visual_select_group, GLX_EXT_texture_from_pixmap
joe@joe-desktop:~$ compiz --replace
Checking for Xgl: not present.
Detected PCI ID for VGA: Flags: bus master, VGA palette snoop, 66MHz, medium devsel, latency 64
01:05.0 0300: 1002:5954 (prog-if 00 [VGA])
Checking for texture_from_pixmap: not present.
Trying again with indirect rendering:
Checking for texture_from_pixmap: not present.
aborting and using fallback: /usr/bin/metacity

[Bernhard Sessler (burnhead)]

Same problem as above - compiz falls back to metacity upon launch, with exactly the same error messages.
I'm using a Radeon X2300 series card in my Samsung R60Plus Notebook.

doe@john:~$ glxinfo | grep GLX_EXT_texture
    GLX_ARB_multisample, GLX_EXT_import_context, GLX_EXT_texture_from_pixmap,
    GLX_SGIX_visual_select_group, GLX_EXT_texture_from_pixmap

doe@john:~$ compiz --replace
Checking for Xgl: not present.
Detected PCI ID for VGA: 01:00.0 0300: 1002:7188 (prog-if 00 [VGA])
Checking for texture_from_pixmap: not present.
Trying again with indirect rendering:
Checking for texture_from_pixmap: not present.
aborting and using fallback: /usr/bin/metacity

[Miguel Martinez (el-quark)]

I think this is due to an incompatibility in fglrx 8.42.3 and xserver 1.4 and later. I'm not really sure, but that's what it seems from reading posts from people in the ubuntu forums and in phoronix.

[Joachim Beckers (jbeckers)]

confirmed on hardy with fglrx 8.43.2 aka 7-11 and an ati x1250

[Joachim Beckers (jbeckers)]

bug still exists with hardy and fglrx 8.443.1 aka 7-12

[LordBernhard (mustermaxi)]

bug also exists on debian sid (fglrx 8.443.1) (manually installed because packages won't work for me because of missing dkms)

compiz --replace:

Checking for Xgl: not present.
Detected PCI ID for VGA: 03:00.0 0300: 1002:5653 (prog-if 00 [VGA controller])
Checking for texture_from_pixmap: not present.
Trying again with indirect rendering:
Checking for texture_from_pixmap: not present.
aborting and using fallback: /usr/bin/metacity
no /usr/bin/metacity found, exiting

glxinfo | grep texture_from_pixmap:
    GLX_ARB_multisample, GLX_EXT_import_context, GLX_EXT_texture_from_pixmap,
    GLX_SGIX_visual_select_group, GLX_EXT_texture_from_pixmap

[th (daimoni)]

Compiz works with Gutsy/ati after installing 'xserver-xgl' package...

[Edward Clark (eeclark)]

@daimoni
Is it actually using XGL? If so, it shouldn't be since this driver is supposed to use AIGLX.
What does your xorg.conf look like?

[Yann Papouin (yann-papouin)]

The compiz script is counting GLX_EXT_texture_from_pixmap in glxinfo and -gt means Greater Than : so glxinfo should display GLX_EXT_texture_from_pixmap 3 times

if [ $($GLXINFO 2>/dev/null | grep GLX_EXT_texture_from_pixmap -c) -gt 2 ] ; then
  verbose "present. \n"
  return 0;
 else
  verbose "not present. \n"

[Edward Clark (eeclark)]

edward@smallville:~$ compiz --replace
Checking for Xgl: not present.
Detected PCI ID for VGA: 01:00.0 0300: 1002:4e50 (prog-if 00 [VGA controller])
 Flags: bus master, VGA palette snoop, 66MHz, medium devsel, latency 255, IRQ 11
Checking for texture_from_pixmap: not present.
Trying again with indirect rendering:
Checking for texture_from_pixmap: not present.
aborting and using fallback: /usr/bin/metacity

edward@smallville:~$ glxinfo
name of display: :0.0
display: :0 screen: 0
direct rendering: Yes
server glx vendor string: SGI
server glx version string: 1.2
server glx extensions:
    GLX_ARB_multisample, GLX_EXT_import_context, GLX_EXT_texture_from_pixmap,
    GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_OML_swap_method,
    GLX_SGIS_multisample, GLX_SGIX_fbconfig, GLX_SGIX_visual_select_group
client glx vendor string: SGI
client glx version string: 1.4
client glx extensions:
    GLX_ARB_get_proc_address, GLX_ARB_multisample, GLX_EXT_import_context,
    GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_MESA_allocate_memory,
    GLX_MESA_copy_sub_buffer, GLX_MESA_swap_control,
    GLX_MESA_swap_frame_usage, GLX_OML_swap_method, GLX_OML_sync_control,
    GLX_SGI_make_current_read, GLX_SGI_swap_control, GLX_SGI_video_sync,
    GLX_SGIS_multisample, GLX_SGIX_fbconfig, GLX_SGIX_pbuffer,
    GLX_SGIX_visual_select_group, GLX_EXT_texture_from_pixmap
GLX version: 1.2
GLX extensions:
    GLX_ARB_get_proc_address, GLX_ARB_multisample, GLX_EXT_import_context,
    GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_OML_swap_method,
    GLX_SGIS_multisample, GLX_SGIX_fbconfig, GLX_SGIX_visual_select_group
OpenGL vendor string: ATI Technologies Inc.
OpenGL renderer string: ATI MOBILITY RADEON 9600/9700 Series
OpenGL version string: 2.1.7059 Release
OpenGL extensions:
    GL_AMD_performance_monitor, GL_ARB_depth_texture, GL_ARB_draw_buffers,
    GL_ARB_fragment_program, GL_ARB_fragment_shader, GL_ARB_multisample,
    GL_ARB_multitexture, GL_ARB_occlusion_query, GL_ARB_pixel_buffer_object,
    GL_ARB_point_parameters, GL_ARB_point_sprite, GL_ARB_shader_objects,
    GL_ARB_shading_language_100, GL_ARB_shadow, GL_ARB_shadow_ambient,
    GL_ARB_texture_border_clamp, GL_ARB_texture_compression,
    GL_ARB_texture_cube_map, GL_ARB_texture_env_add,
    GL_ARB_texture_env_combine, GL_ARB_texture_env_crossbar,
    GL_ARB_texture_env_dot3, GL_ARB_texture_float,
    GL_ARB_texture_mirrored_repeat, GL_ARB_texture_rectangle,
    GL_ARB_transpose_matrix, GL_ARB_vertex_buffer_object,
    GL_ARB_vertex_program, GL_ARB_vertex_shader, GL_ARB_window_pos,
    GL_ATI_draw_buffers, GL_ATI_envmap_bumpmap, GL_ATI_fragment_shader,
    GL_ATI_meminfo, GL_ATI_separate_stencil, GL_ATI_texture_env_combine3,
    GL_ATI_texture_float, GL_EXT_bgra, GL_EXT_blend_color,
    GL_EXT_blend_func_separate, GL_EXT_blend_minmax, GL_EXT_blend_subtract,
    GL_EXT_compiled_vertex_array, GL_EXT_copy_texture,
    GL_EXT_draw_range_elements, GL_EXT_fog_coord, GL_EXT_framebuffer_object,
    GL_EXT_gpu_program_parameters, GL_EXT_multi_draw_arrays,
    GL_EXT_packed_depth_stencil, GL_EXT_packed_pixels,
    GL_EXT_point_parameters, GL_EXT_rescale_normal, GL_EXT_secondary_c...

[DarkMageZ (darkmagez)]

i suffer this problem as well. it's because fglrx does support aiglx on xorg-core 1.3 branch but not on 1.4 (which is what we've been using for a long time in hardy).
fglrx needs to be fixed OR xorg-core needs to be reverted. (do not revert this yourself, wait for the ubuntu devs to do it) OR the ubuntu dev's need to throw some kind of serious hack in =D

[Miguel Martinez (el-quark)]

There's some interesting stuff in the compiz script, apart from the check pointed out by Yann Papouin. If I count the number of times that "texture from pixmap" appears on glxinfo while using radeon (the FOSS driver), I get:

$ glxinfo | grep texture_from_pixmap
    GLX_ARB_multisample, GLX_EXT_import_context, GLX_EXT_texture_from_pixmap,
    GLX_SGIX_visual_select_group, GLX_EXT_texture_from_pixmap
$ glxinfo | grep texture_from_pixmap -c
2

I've also grepped Edward Clark's full glxinfo report and "texture_from_pixmap" also appears just two times, even though he is using fglrx and I am using radeon. Please note that my card is a Mobility Radeon 9600. But of course compiz loads with radeon[1]. I see the hardy version is still 7.11 (I haven't tried it yet), but as reverting xorg-core won't happen (I'm 99% sure of that), we'll have to wait for a newer fglrx release. Does any of you know if 8.01 works with xserver 1.4?

[1] After dist-upgrading to hardy, compiz refused to start, but that's because the script was pointing to /usr/local/ instead of /usr/ for both compiz.real binary and plugin libraries.

[harrydb (harrydeboer)]

I figured out that the first (direct rendering) test reveals a count of 2 instead of 3 and thus fails, but the indirect test fails for another reason. When I run

LIBGL_ALWAYS_INDIRECT=1 glxinfo

the output is as follows.

name of display: :0.0
X Error of failed request: GLXBadContext
  Major opcode of failed request: 143 (GLX)
  Minor opcode of failed request: 5 (X_GLXMakeCurrent)
  Serial number of failed request: 16
  Current serial number in output stream: 16

[Joachim Beckers (jbeckers)]

still an issue with 8.45.4

[Bryce Harrington (bryce)]

Thanks for confirming it remains an issue with the latest 8-01 version of the driver. Could someone please attach their /etc/X11/xorg.conf, ~/.xsession-errors, and /var/log/Xorg.0.log files after booting with this driver and attempting to start compiz?

[harrydb (harrydeboer)]

Xorg.0.log after trying to enable compiz

[harrydb (harrydeboer)]

.xsession-errors after trying to enable compiz

[Daniel Swarbrick (pressureman)]

I'm running a Mobility Radeon 9700 with fglrx, and also cannot get compiz to start. I notice that by default, the restricted drivers manager adds the following to xorg.conf when installing the fglrx driver:

Section "Extensions"
 Option "Composite" "0"
EndSection

Is this relevant? Shouldn't "Composite" be set to "1" or "True"?

[harrydb (harrydeboer)]

@daniel
The flgrx version shipped with feisty (and gutsy I think) does not (officially) support compositing, so it is disabled. Are you using one of them? If so you could try setting Composite to 1 and installing xserver-xgl (no aiglx in feisty/gutsy as far as I know), but no guarantees...

Harry

[Gilad F (giladfr)]

Having the same problem.
Thinkpad T42, Mobility Radeon 9600 with the new 8.01 fglrx driver.
Compiz wont start, compositing and AIGLX is enabled in xorg.conf.
Compiz.real output - "Fatal: GLX_EXT_texture_from_pixmap is missing"
Other Open-GL programs work without problems.

[DarkMageZ (darkmagez)]

xgl is not a good solution to this problem as it introduces even worse problems than not having compiz running in the first place.

the core of this problem is in fglrx's lack of aiglx support for xorg server 1.4.
hopefully if amd gets poked enough about this then they will fix the issue.

Changing:

Section "Extensions"
 Option "Composite" "0"
EndSection

to

Section "Extensions"
 Option "Composite" "1"
EndSection

Doesn't resolve the problem. Tho when it does get resolved it'd be nice if this was changed to "1" by default.

[Gilad F (giladfr)]

This is strange since I experience this bug, but using gutsy and xorg server - 1.3.
Compiz with AIGLX worked flawlessly (Ignoring all the "glitches" that were reported in the documentation of the driver) with the last two releases of fglrx...
For some reason i can't find anyone complaining about this problem in phoronix forums, maybe it affects only a fraction of the supported cards of this driver.

[DarkMageZ (darkmagez)]

the initial report was on hardy after the transition from xorg server 1.3 to 1.4.
the initial reporter's first problem then would be the lack of working aiglx support by the fglrx drivers on xorg server 1.4.

gutsy has xorg server 1.3.
you are using the latest fglrx drivers which supports aiglx on xorg server 1.3.
i believe your problem is a separate issue/bug.

[kevinrong (kevinrong)]

My xorg.conf

[kevinrong (kevinrong)]

My xsession-errors

[Miguel Martinez (el-quark)]

Dear fellows,

ATi has just released fglrx 8.02 (13 february). It seems that compiz is
working on xserver 1.4 now. I've downloaded the drivers, but haven't tested
them. Let's hope these work.

OT: Suspend and hibernate (using oss radeon) work for me on hardy with a
100% success rate so far. I do fear installing fglrx will break the current
homeostasis I'm in.

[Rocko (rockorequin)]

I just tested the new driver (Catalyst 8-2, fglrx 8.45.5) in xserver 1.4 (Hardy alpha 4), and compiz now works.

[Edward Clark (eeclark)]

So how long before this is included in the Hardy Repo?

[Launchpad Janitor (janitor)]

This bug was fixed in the package linux-restricted-modules-2.6.24 - 2.6.24.9-8.20

---------------
linux-restricted-modules-2.6.24 (2.6.24.9-8.20) hardy; urgency=low

  [ Timo Aaltonen ]
  * nvidia-glx*:
    - When using the nvidia-settings binary for storing
      values in xorg.conf it clutters up the file with unneeded settings,
      and some are just plain wrong nowadays (the mouse device for
      instance). So, don't ship the version included in the drivers, instead
      Suggest nvidia-settings -package (again), and fix the issues there.
    - Don't ship nvidia-settings.desktop, only the .png file.
      Nvidia-settings will include the desktop-file.
  * rules:
    - Enable fglrx for -rt (LP: #182242)
  * rules, control.stub.in:
    - Enable -server flavor. (LP: #153011)
  * dvb:
    - Add dvb-usb-bluebird-01.fw (LP: #90723)
    - Add dvb-fe-tda10046.fw (LP: #107090)
  * ati: Update fglrx to 8-02. (LP: #173663, #189772)
  * {nvidia-glx,nvidia-glx-dev,xorg-driver-fglrx}.postrm.in
    - Check that the diverted file actually belongs to the current package
      before trying to remove the diversion. Thanks to Kirk Turner and
      Alberto Milone for patches, and Michael Vogt for review!
      (LP: #36625, #130208, #175789)
  * nvidia-glx.{preinst,postrm}.in, rules:
    - Fix the path of libglx.so.
  * rules:
    - Unexport CFLAGS and LDFLAGS to cope with the new dpkg-buildpackage.

  [ Alberto Milone ]
  * debian/lrm-video: Add support for EnvyNG.

 -- Timo Aaltonen <email address hidden> Thu, 14 Feb 2008 17:57:44 +0200