[SOLVED] X start failure after upgrade

replabrobin · 2014-06-17 10:08:25

After a recent pacman -Syu my 32 bit system cannot seem to start X. I see this in the log

[    31.221] (II) intel(0): Output VGA1 connected
[    31.221] (II) intel(0): Using exact sizes for initial modes
[    31.221] (II) intel(0): Output VGA1 using initial mode 1920x1200
[    31.221] (II) intel(0): Using default gamma of (1.0, 1.0, 1.0) unless otherwise stated.
[    31.221] (II) intel(0): Kernel page flipping support detected, enabling
[    31.221] (==) intel(0): DPI set to (96, 96)
[    31.221] (II) Loading sub module "fb"
[    31.221] (II) LoadModule: "fb"
[    31.221] (II) Loading /usr/lib/xorg/modules/libfb.so
[    31.222] (II) Module fb: vendor="X.Org Foundation"
[    31.222]    compiled for 1.15.1, module version = 1.0.0
[    31.222]    ABI class: X.Org ANSI C Emulation, version 0.4
[    31.222] (II) Loading sub module "dri2"
[    31.222] (II) LoadModule: "dri2"
[    31.222] (II) Module "dri2" already built-in
[    31.222] (II) Loading sub module "dri3"
[    31.222] (II) LoadModule: "dri3"
[    31.222] (WW) Warning, couldn't open module dri3
[    31.222] (II) UnloadModule: "dri3"
[    31.222] (II) Unloading dri3
[    31.222] (EE) intel: Failed to load module "dri3" (module does not exist, 0)
[    31.222] (==) Depth 24 pixmap format is 32 bpp
[    31.222] (II) intel(0): Allocated new frame buffer 1920x1200 stride 8192, tiled
[    31.222] (II) UXA(0): Driver registered support for the following operations:
[    31.223] (II)         solid
[    31.223] (II)         copy
[    31.223] (II)         composite (RENDER acceleration)
[    31.223] (II)         put_image
[    31.223] (II)         get_image
[    31.223] (II) intel(0): [DRI2] Setup complete
[    31.223] (II) intel(0): [DRI2]   DRI driver: i915
[    31.223] (II) intel(0): [DRI2]   VDPAU driver: i915
[    31.223] (==) intel(0): Backing store enabled
[    31.223] (==) intel(0): Silken mouse enabled
[    31.223] (II) intel(0): Initializing HW Cursor
[    31.223] (II) intel(0): Use legacy UXA acceleration.
[    31.223] (II) intel(0): RandR 1.2 enabled, ignore the following RandR disabled message.
[    31.223] (==) intel(0): DPMS enabled
[    31.223] (==) intel(0): Intel XvMC decoder disabled
[    31.223] (II) intel(0): Set up overlay video
[    31.223] (II) intel(0): DRI2: Enabled
[    31.223] (II) intel(0): DRI3: Disabled
[    31.223] (==) intel(0): hotplug detection: "enabled"
[    31.250] (--) RandR disabled
[    31.270] (II) AIGLX: enabled GLX_MESA_copy_sub_buffer
[    31.271] (II) AIGLX: enabled GLX_ARB_create_context
[    31.271] (II) AIGLX: enabled GLX_ARB_create_context_profile
[    31.271] (II) AIGLX: enabled GLX_EXT_create_context_es2_profile
[    31.271] (II) AIGLX: enabled GLX_INTEL_swap_event
[    31.271] (II) AIGLX: enabled GLX_SGI_swap_control and GLX_MESA_swap_control
[    31.271] (II) AIGLX: enabled GLX_EXT_framebuffer_sRGB
[    31.271] (II) AIGLX: enabled GLX_ARB_fbconfig_float
[    31.271] (II) AIGLX: GLX_EXT_texture_from_pixmap backed by buffer objects
[    31.271] (II) AIGLX: Loaded and initialized i915
[    31.271] (II) GLX: Initialized DRI2 GL provider for screen 0
[    31.272] (EE)
Fatal server error:
[    31.273] (EE) failed to create screen resources(EE)
[    31.273] (EE)
Please consult the The X.Org Foundation support
     at http://wiki.x.org
 for help.
[    31.275] (EE) Please also check the log file at "/var/log/Xorg.0.log" for additional information.
[    31.275] (EE)
[    31.275] (II) AIGLX: Suspending AIGLX clients for VT switch
[    31.310] (EE) Server terminated with error (1). Closing log file.

Looking at older logs I don't see this AIGLX thing. Is this new?

Edit: upstream fix is present in xf86-video-intel-git available from AUR
Edit: this is certainly now fixed in official package xf86-video-intel-2.99.914-1

Last edited by replabrobin (2014-07-30 08:52:30)

jasonwryan · 2014-06-17 10:15:58

Not a Sysadmin issue, moving to NC...

Trilby · 2014-06-17 10:45:32

There's really nothing to go on here. You need to provide some information:

What was updated? What WM/DE do you use? Do you use a compositor? How do you start X? What's in your xinitrc (or equivalent file)?

replabrobin · 2014-06-17 12:50:16

These packages were created in /var/cache/pkg on the day in question

binutils-2.24-6-i686.pkg.tar.xz
isl-0.13-1-i686.pkg.tar.xz
cloog-0.18.1-3-i686.pkg.tar.xz
libdbus-1.8.4-1-i686.pkg.tar.xz
dbus-1.8.4-1-i686.pkg.tar.xz
gcc-libs-4.9.0-4-i686.pkg.tar.xz
gcc-4.9.0-4-i686.pkg.tar.xz
grep-2.20-1-i686.pkg.tar.xz
libsystemd-213-9-i686.pkg.tar.xz
perl-5.20.0-4-i686.pkg.tar.xz
man-pages-3.69-1-any.pkg.tar.xz
systemd-213-9-i686.pkg.tar.xz
systemd-sysvcompat-213-9-i686.pkg.tar.xz
tzdata-2014e-1-any.pkg.tar.xz
cifs-utils-6.3-2-i686.pkg.tar.xz
cmake-3.0.0-2-i686.pkg.tar.xz
ffmpeg-compat-1:0.10.13-1-i686.pkg.tar.xz
giflib-5.1.0-1-i686.pkg.tar.xz
fontforge-20140101-3-i686.pkg.tar.xz
gc-7.4.2-1-i686.pkg.tar.xz
gparted-0.19.0-1-i686.pkg.tar.xz
imlib2-1.4.6-2-i686.pkg.tar.xz
jre7-openjdk-headless-7.u60_2.5.0-2-i686.pkg.tar.xz
jre7-openjdk-7.u60_2.5.0-2-i686.pkg.tar.xz
jdk7-openjdk-7.u60_2.5.0-2-i686.pkg.tar.xz
librevenge-0.0.1-2-i686.pkg.tar.xz
libwpd-0.10.0-1-i686.pkg.tar.xz
libcdr-0.1.0-1-i686.pkg.tar.xz
libetonyek-0.1.1-1-i686.pkg.tar.xz
libmspub-0.1.0-1-i686.pkg.tar.xz
libodfgen-0.1.1-1-i686.pkg.tar.xz
libproxy-0.4.11-4-i686.pkg.tar.xz
libreoffice-common-4.2.4-2-i686.pkg.tar.xz
libreoffice-base-4.2.4-2-i686.pkg.tar.xz
libreoffice-calc-4.2.4-2-i686.pkg.tar.xz
libreoffice-draw-4.2.4-2-i686.pkg.tar.xz
libreoffice-extension-nlpsolver-4.2.4-2-i686.pkg.tar.xz
libreoffice-extension-wiki-publisher-4.2.4-2-i686.pkg.tar.xz
libreoffice-gnome-4.2.4-2-i686.pkg.tar.xz
libreoffice-impress-4.2.4-2-i686.pkg.tar.xz
libreoffice-math-4.2.4-2-i686.pkg.tar.xz
libreoffice-sdk-4.2.4-2-i686.pkg.tar.xz
libreoffice-sdk-doc-4.2.4-2-i686.pkg.tar.xz
libreoffice-writer-4.2.4-2-i686.pkg.tar.xz
libwpg-0.3.0-1-i686.pkg.tar.xz
libvisio-0.1.0-1-i686.pkg.tar.xz
libwebp-0.4.0-2-i686.pkg.tar.xz
libwps-0.3.0-1-i686.pkg.tar.xz
sbcl-1.2.0-1-i686.pkg.tar.xz
maxima-5.33.0-4-i686.pkg.tar.xz
mplayer-37224-1-i686.pkg.tar.xz
python-setuptools-1:5.1-1-any.pkg.tar.xz
python2-mako-1.0.0-1-any.pkg.tar.xz
python2-setuptools-1:5.1-1-any.pkg.tar.xz
serf-1.3.6-1-i686.pkg.tar.xz
xf86-video-intel-2.99.912-1-i686.pkg.tar.xz
wireshark-cli-1.10.8-1-i686.pkg.tar.xz
wireshark-gtk-1.10.8-1-i686.pkg.tar.xz

of those I guess xf86-video-intel-2.99.912-1-i686.pkg.tar.xz might be the one causing the damage. I downgraded to the previous version xf86-video-intel-2.99.911-2-i686.pkg.tar.xz and now everything works fine. I suppose the latest code has something that disagrees with my systems.

I am using the slim 1.3.6-4 DM with WM openbox 3.5.2-6. Fr compositor I am using xcompmgr-dana 20091025-4

I am starting x using the slim.service; my xinit starts openbox which starts xcompmgr etc etc.

Is this enough. Should I mark this as solved or is it just a bug in my setup versus the latest xf86-video-intel-2.99.912-1-i686.pkg.tar.xz?

Trilby · 2014-06-17 12:59:50

Take all this with a large grain of salt - I don't really know anything about these composite managers - but AIGLX is only for composite managers. It would not surprise me if a change in the intel video driver required a revision in composite managers that run under it. Xcompmgr-dana (while I happily used it for some time) seems not to have been updated in quite a long time (several years it seems).

I suspect that xcompmgr-dana may not be compatible with the new video driver.

Downgrading the video driver supports but doesn't confirm this. Further helpful tests would be to reinstall the new video driver and do either or both of the following: 1) disable/remove the xcompmgr-dana, or 2) replace xcompmgr-dana with an actively updated compositor.

These steps could be temporary changes for diagnostics only. If neither of the above avoids the problem, then we can rule out an incompatibility between the video driver and xcompgr-dana. But if it is an incompatibility, no new versions of the intel driver would work for you in future - permenantly ignoring a package upgrade would not be a long term solution.

rebootl · 2014-06-17 13:23:52

Hm, had a similar issue recently, but that was on a quiet old GPU. What is your GPU (lspci|grep VGA)?

In addition to what Trilby suggested (try to disable xcompmgr-dana first), you may want to try to disable "uxa", I think you must have had enabled it at some point.

I'm not sure for the AIGLX...

Evtl. see my bug report for more info: https://bugs.freedesktop.org/show_bug.cgi?id=80088

Last edited by rebootl (2014-06-17 13:24:19)

replabrobin · 2014-06-17 14:26:43

I don't think this can be the compositor as that isn't even started until slim opens openbox. I use compton-git elsewhere as that is quite actively developed and will try that. I suspect this is a hardware issue as my hardware is quite old

$ lspci|grep VGA
00:02.0 VGA compatible controller: Intel Corporation 82865G Integrated Graphics Controller (rev 02)

It seems that uxa is the problem. If I remove this file from /etc/X11/xorg.conf.d/ then xf86-video-intel-2.99.912-1-i686.pkg.tar.xz gives me a booting system.

Section "Device"
    Identifier "Intel Graphics"
    Driver "intel"
    Option "AccelMethod" "uxa"
EndSection

However, it has the same problems as reported here
https://bbs.archlinux.org/viewtopic.php?id=167900 and here https://bbs.archlinux.org/viewtopic.php?pid=1310215 where the [SOLVED] comes from using UXA. I tried disabling AIGLX, but that doesn't help either. Looking at rebootl's bug seems there may be some fix coming. I will downgrade and hope.

rebootl · 2014-06-17 15:29:56

I will downgrade and hope.

Not sure, since your GPU seems newer, it was on a 830M ... but I'm not an expert for GPU's...

You can try with the xf86-video-intel-git package from AUR, it contains the fix!

(SNA is the default now UXA is older. Using UXA with the -git package should also work now, I think.)

replabrobin · 2014-06-17 15:45:35

Well the -git package does work for me. Presumably at some point the change that fixes this will become mainstream and I should be able to go back to the standard package. Should we consider this [SOLVED].

rebootl · 2014-06-17 17:30:58

Cool.

Should we consider this [SOLVED].

Up to you to decide. As you mentioned at some point you should be able to go back to the standard package. That's certainly a good idea then, since the -git version might be a bit glitchy dependending on what they're working on...
I don't see really much left to do atm.

Regards

Trilby · 2014-06-17 18:08:27

It does seem solved - although I think a "[WORKAROUND]" label in the title could be suitable as well provided you'd come back when the main repo package is updated to change it to "[SOLVED]" if the relevant parts of the -git package are implemented in the main repo package.

Arch Linux

#1 2014-06-17 10:08:25

[SOLVED] X start failure after upgrade

#2 2014-06-17 10:15:58

Re: [SOLVED] X start failure after upgrade

#3 2014-06-17 10:45:32

Re: [SOLVED] X start failure after upgrade

#4 2014-06-17 12:50:16

Re: [SOLVED] X start failure after upgrade

#5 2014-06-17 12:59:50

Re: [SOLVED] X start failure after upgrade

#6 2014-06-17 13:23:52

Re: [SOLVED] X start failure after upgrade

#7 2014-06-17 14:26:43

Re: [SOLVED] X start failure after upgrade

#8 2014-06-17 15:29:56

Re: [SOLVED] X start failure after upgrade

#9 2014-06-17 15:45:35

Re: [SOLVED] X start failure after upgrade

#10 2014-06-17 17:30:58

Re: [SOLVED] X start failure after upgrade

#11 2014-06-17 18:08:27

Re: [SOLVED] X start failure after upgrade

Board footer