Bug 19242

Created attachment 21414 [details]
verbose glxinfo

glxinfo -v output from normal user

Created attachment 21415 [details]
verbose lspci

lspci -vv as root

Created attachment 21417 [details]
verbose xrandr

xrandr --verbose output as normal user

Ben, please attach Xorg.0.log.

Stefan, is this using Intel Q3 release?

Yes. Novell intel-Q3 release

Created attachment 21450 [details]
Xorg log

copy of Xorg.0.log taken while the crashed X server was still running

Additional info: Mouse pointer unaffected, keyboard frozen including caps lock. all applications become unresponsive, screen does not refresh ( clock, etc ). shutting down the X server by stopping the XDM service does not work, must be killed via "kill -9". after it is killed the bottom half of the screen is garbage filled, and top half remains at last X displayed image. Unable to use computer locally now. Must be powercycled to even use a virtual terminal locally.

Forgot to mention: couldn't use "ModeDebug" "yes" even after removing quotes around ModeDebug. X claimed the setting was invalid and refused to start.

Parse error on line 153 of section Device in file /etc/X11/xorg.conf
	"ModeDebug" is not a valid keyword in this section.
(EE) Problem parsing the config file
(EE) Error parsing the config file

Fatal server error:
no screens found

In the device setion, /etc/X11/xorg.conf
set like this
Option "ModeDebug" "on"

By the way, please try to disable vblank to check if it works.

~/.drirc file
<driconf>
 <device screen="0" driver="i965">
    <application name="Default">
             <option name="vbalnk_mode" value="0" />
    </application>
 </device>
</driconf>

If it works, that is fine, just close this bug.

Created attachment 21486 [details]
Xorg log with ModeDebug

Xorg.0.log after graphics freeze initiated, with option "ModeDebug" "yes" in xorg.conf. ~/.drirc had the following added, and the system rebooted:

<driconf>
 <device screen="0" driver="i965">
    <application name="Default">
             <option name="vblank_mode" value="0" />
    </application>
 </device>
</driconf>

Just to clarify the attachment comment: I rebooted the system manually after modifying the file, it didn't just reset or powerdown on me by itself.

Haihao
Any progress for that?

Ben Cooksley
Would you like try option ExaNoComposite?

Add the line 
Option "ExaNoComposite" "on" 
in the device section. 

I have just tried this. It works and no crash. 

Thanks

Sorry, it failed terribly, I had to remove the option. While it did not ( get the chance to? ) hard lock the graphics system, I observed the following problems:

A) KDE 4.2 trunk system tray icons were completely messed up ( random image bits from memory, what was previously at that location, or totally black )
B) KWin, especially when the application was not full screen did not paint the window border at all. What was below it ( Plasma desktop ) displayed through instead.
C) Google Earth crashed after a small amount of time using it ( 30 seconds? ) it would normally have hardlocked within this period anyway, however.

Note that I do not have KWin 3D effects enabled what so ever.

I tested intel-q3 with kernel 2.6.27/2.6.28, and Google Earth works for me.
Could you try to start up X from a console then run Google Earth?

(In reply to comment #14)
> Sorry, it failed terribly, I had to remove the option. While it did not ( get
> the chance to? ) hard lock the graphics system, I observed the following
> problems:
> 
> A) KDE 4.2 trunk system tray icons were completely messed up ( random image
> bits from memory, what was previously at that location, or totally black )
> B) KWin, especially when the application was not full screen did not paint the
> window border at all. What was below it ( Plasma desktop ) displayed through
> instead.
> C) Google Earth crashed after a small amount of time using it ( 30 seconds? )
> it would normally have hardlocked within this period anyway, however.
> 
> Note that I do not have KWin 3D effects enabled what so ever.
> 

I guess you enable desktop effect in 965G. This option will slow many thing. I installed SLE11-RC1 for testing on 965GM. Before set the option, it will directly crash xserver. After setting, I keep google earth more than one hour and then close it by me. 

Please have a try with Desktop Effect disable.  Here are the step s
1) run simple-ccsm  command
2) uncheck enable desktop effect

And then try them 

Regards

I do not use compiz as my Window Manager, I use KWin. KWin 3D effects are disabled. 3D effects are disabled in Compiz anyway.

My version of Google Earth is quite old, I am going to try out the latest version.

Hi, quanxian
 I installed sled11-rc1 on my box and used xinit to start X server. Some simple 3D applications such as glxgears worked fine, but googleearth always caused the screen to freeze.  Then I tested intel-2008-q3 on this machine and googleearth worked well!!!. 

Are you sure sled11-rc1 use intel-2008-q3?  Why didn't I see some message such as 'Failed to initialize GEM.  Falling back to classic' with a NON-GEM kernel?

Oh? it is a little tricky. I ever built mesa from intel-q3 release by myself. Still the google earth hang. Also I can find the message with "Failed to initialize GEM.  Falling back to classic"

I will check with Novell for that. Thanks for your response.

Haiha(In reply to comment #18)
> Hi, quanxian
>  I installed sled11-rc1 on my box and used xinit to start X server. Some simple
> 3D applications such as glxgears worked fine, but googleearth always caused the
> screen to freeze.  Then I tested intel-2008-q3 on this machine and googleearth
> worked well!!!. 
> 
> Are you sure sled11-rc1 use intel-2008-q3?  Why didn't I see some message such
> as 'Failed to initialize GEM.  Falling back to classic' with a NON-GEM kernel?
> 

Haihao, do you use mouse to make the earth turn around? Try several times and make it turn to check what happens? I just patch intel-q3-release including Mesa and xf86-video-intel packages. It still hang there after sevral tries. 

I need your confirmation for this. Thanks. For the message you mentioned, I recalled generally Novell will not let user find such information. (Failed to initialize GEM.  Falling back to classic). They must comment this out.  I have send email to Novell to check this. Also please do the testing in your side to make sure the problem don't have.

I still make reopen until we make sure all thing are confirmed. 

Thanks for your understanding.

Of course I rotated it. My version of Google earth is 4.0.2091(beta)

Haihao, sometimes it will need some times for it. try it in different way.

Thanks

My version of googleearth is 4.3.7284.3916(beta). It should be the latest version of google earth. I just donwload it from official website.

I updated to the latest Google Earth, and was still able to reproduce the issue.

Hi, Ben Cooksley 
Would you like to get the latest packages from following website to have a try?

http://download.opensuse.org/repositories/X11:/XOrg/openSUSE_11.1/

Mesa-7.2
libdrm-2.4.1
xorg-x11-driver-video-7.4

I use openSUSE 11.1 kernel 2.6.27.7-9-pae, with kde 3.5.10 in my HP2230s
laptop, chipset Intel GM45. When I activate 3D in Sax then my laptop is hang
upon running google earth 4.3. The system become not responsive, screen is
freeze, cannot login from konsole, keyboard also freezed only mouse pointer can
move. I do not use any desktop effect, only enabling 3D on Sax. 

It does not happen when I disable 3D support.

I try to use Intel-Q3 but cannot help. I also try to upgrade the kernel to
2.6.28 from kernel.org but still the same, google earth only works if 3D is
disabled.

For the info compiz works if 3D is enabled.

I filled the same case in https://bugzilla.novell.com/show_bug.cgi?id=463454

Should I submit another info?

The last time I used that repository was on openSUSE 11.0, and I was burned badly by the experience. 

I only updated xorg-x11-driver-video previously, and it left me without a functioning X system ( because it didn't like a miniscule version change )

Are you sure that only those packages are required to be updated?

Hi, quanxian
   Does openSUSE 11.1 work well on GM45? I wonder it is a 965GM specific issue or not. 

(In reply to comment #28)
> Hi, quanxian
>    Does openSUSE 11.1 work well on GM45? I wonder it is a 965GM specific issue
> or not. 
> 

In my experience using GM45 (HP2330s) on openSUSE 11.1 everything works well; 3D, compiz (on KDE 3.5.10) and kwin (on KDE 4.1.2). 

The only problem I have so far is if I enabled 3D and use google earth then the machine will hang. But if I only use 2D then google earth can running well.

(In reply to comment #29)
> The only problem I have so far is if I enabled 3D and use google earth then the
> machine will hang. But if I only use 2D then google earth can running well.

So, it's the same problem...

(In reply to comment #25)
> Hi, Ben Cooksley 
> Would you like to get the latest packages from following website to have a try?
> 
> http://download.opensuse.org/repositories/X11:/XOrg/openSUSE_11.1/

There are no Intel relevant changes against openSUSE 11.1 in this repo at the
moment. No need to update.

(In reply to comment #29)
> (In reply to comment #28)
> > Hi, quanxian
> >    Does openSUSE 11.1 work well on GM45? I wonder it is a 965GM specific issue
> > or not. 
> > 
> 
> In my experience using GM45 (HP2330s) on openSUSE 11.1 everything works well;
> 3D, compiz (on KDE 3.5.10) and kwin (on KDE 4.1.2). 
> 
> The only problem I have so far is if I enabled 3D and use google earth then the
> machine will hang. But if I only use 2D then google earth can running well.
> 

enabled 3D? Do you mean enabling 3D effects of your desktop? The original reporter said 3D effects were disabled, so maybe your problem is different from this bug.

(In reply to comment #32)
> 
> enabled 3D? Do you mean enabling 3D effects of your desktop? The original
> reporter said 3D effects were disabled, so maybe your problem is different from
> this bug.
> 

OK. I enabled 3D support in Sax2 configuration. But no desktop effect. 
In this condition if I run google earth then my pc completely hang.

We have done full test on GM45(Montevina, x86,x64), G45(McCreary,x86 x64). Xserver will hang after running google earth (vblank disabled, desktop effect disabled) for all.

It should be a general issue for all platform.

Same here for the Mobile GM45 Express (Cantiga) on the Lenovo x200s.

I confirm this bug on debian/sid x86_64 and G45 chipset with:

xserver-xorg-video-intel-2.5.1
mesa-7.2
drm-2.4.1
xorg-server-1.5.3

The screen freezes EVERY time an opengl application is used.
Is can take 1 minute or 1 hour, but it happens.

This bug seems due to /usr/lib/dri/i965_dri.so library from
the libgl1-mesa-dri package.

Tomorrow the mesa-7.3-beta1 will be released.
It would be useful if someone could compile and test if the bug is solved.

I can easily reproduce the machine freeze with GoogleEarth on GM45 (tested on Sony VAIO VGN-Z540). Setting EXANoComposite avoids the freeze.

(In reply to comment #38)
> I can easily reproduce the machine freeze with GoogleEarth on GM45 (tested on
> Sony VAIO VGN-Z540). Setting EXANoComposite avoids the freeze.
> 

Have you tried also with other applications (such as mplayer with opengl driver) to verify that the freeze really depends only on composite?

In my case disabling composite does not solve the issue.

Actually something changes: launching "Tile Racing" with EXANoComposite the main menu shows correctly, otherwise results in black screen.
But if I go further and try to play the game it immediately freezes, with and without composite.

I also found this strange dmesg output, just after rebooting:

agpgart-intel 0000:00:00.0: Intel G45/G43 Chipset
agpgart-intel 0000:00:00.0: detected 131068K stolen memory
agpgart-intel 0000:00:00.0: AGP aperture is 256M @ 0xd0000000
ACPI: Power Button (FF) [PWRF]
input: Power Button (CM) as /devices/LNXSYSTM:00/device:00/PNP0C0C:00/input/input4
ACPI: Power Button (CM) [PWRB]
Linux video capture interface: v2.00
iTCO_vendor_support: vendor-support=0
iTCO_wdt: Intel TCO WatchDog Timer Driver v1.04
iTCO_wdt: Found a ICH10R TCO device (Version=2, TCOBASE=0x0860)
iTCO_wdt: initialized. heartbeat=30 sec (nowayout=0)
i801_smbus 0000:00:1f.3: PCI INT C -> GSI 18 (level, low) -> IRQ 18
ACPI: I/O resource 0000:00:1f.3 [0x400-0x41f] conflicts with ACPI region SMRG [0x400-0x40f]
ACPI: Device needs an ACPI driver
...
...
...
HDA Intel 0000:00:1b.0: PCI INT A -> GSI 22 (level, low) -> IRQ 22
HDA Intel 0000:00:1b.0: setting latency timer to 64
hda_codec: Unknown model for ALC883, trying auto-probe from BIOS...
...
...
...
[drm] Initialized drm 1.1.0 20060810
pci 0000:00:02.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
pci 0000:00:02.0: setting latency timer to 64
pci 0000:00:02.0: irq 1275 for MSI/MSI-X
[drm] Initialized i915 1.6.0 20080730 on minor 0
resource map sanity check conflict: 0xd0000000 0xdfffffff 0xd0000000 0xd7feffff vesafb
------------[ cut here ]------------
WARNING: at arch/x86/mm/ioremap.c:226 __ioremap_caller+0x359/0x3a0()
Modules linked in: i915 drm binfmt_misc rfcomm l2cap bluetooth battery ppdev parport_pc lp parport cpufreq_conservative cpufreq_stats af_packet xfs fuse coretemp cpufreq_performance cpufreq_powersave cpufreq_ondemand acpi_cpufreq freq_table loop sg rtc_cmos rtc_core rtc_lib pcspkr snd_hda_intel sr_mod gspca_pac207 gspca_main i2c_i801 snd_pcm_oss i2c_core iTCO_wdt iTCO_vendor_support snd_mixer_oss snd_pcm cdrom snd_timer snd_page_alloc snd_hwdep compat_ioctl32 videodev v4l1_compat evdev snd soundcore intel_agp button ext3 jbd mbcache usbhid hid usb_storage ata_piix ata_generic firewire_ohci firewire_core crc_itu_t ohci1394 pata_acpi pata_marvell ieee1394 libata r8169 mii uhci_hcd ehci_hcd usbcore thermal processor fan sd_mod scsi_mod raid1 raid0 md_mod
Pid: 3603, comm: Xorg Not tainted 2.6.28-0.slh.8-sidux-amd64 #1
Call Trace:
 [<ffffffff8024d514>] warn_on_slowpath+0x64/0xa0
 [<ffffffff80253c68>] iomem_map_sanity_check+0x98/0xc0
 [<ffffffff802375d9>] __ioremap_caller+0x359/0x3a0
 [<ffffffffa04dcbd0>] i915_gem_entervt_ioctl+0x2d0/0x5a0 [i915]
 [<ffffffffa04dcbd0>] i915_gem_entervt_ioctl+0x2d0/0x5a0 [i915]
 [<ffffffffa04dc900>] i915_gem_entervt_ioctl+0x0/0x5a0 [i915]
 [<ffffffffa04bb9d2>] drm_ioctl+0x112/0x340 [drm]
 [<ffffffff802ec4f5>] vfs_ioctl+0x85/0xb0
 [<ffffffff802ec59c>] do_vfs_ioctl+0x7c/0x470
 [<ffffffff802eca31>] sys_ioctl+0xa1/0xb0
 [<ffffffff8021350a>] system_call_fastpath+0x16/0x1b
---[ end trace b2dbb73cbe703724 ]---
Xorg:3603 conflicting memory types d0000000-e0000000 write-combining<->uncached-minus
reserve_memtype failed 0xd0000000-0xe0000000, track write-combining, req write-combining
Xorg:3603 conflicting memory types d0000000-e0000000 write-combining<->uncached-minus
reserve_memtype failed 0xd0000000-0xe0000000, track write-combining, req write-combining
Xorg:3603 conflicting memory types d0000000-e0000000 write-combining<->uncached-minus
reserve_memtype failed 0xd0000000-0xe0000000, track write-combining, req write-combining
Xorg:3825 freeing invalid memtype d0000000-e0000000
Xorg:3603 conflicting memory types d0000000-e0000000 write-combining<->uncached-minus
reserve_memtype failed 0xd0000000-0xe0000000, track write-combining, req write-combining
Xorg:3829 freeing invalid memtype d0000000-e0000000
Xorg:3603 conflicting memory types d0000000-e0000000 write-combining<->uncached-minus
reserve_memtype failed 0xd0000000-0xe0000000, track write-combining, req write-combining
Xorg:3831 freeing invalid memtype d0000000-e0000000

quanxian,
   Could you try with mesa commits 59b2c2adbbece27ccf54e58b598ea29cb3a5aa85,
2fc9d671e0be77676818da51cd3b35b52af79421, c30eb2c104c19c758c0b2a2461ad7f38cfa747d0
?

OK. I will patch them into SLED11 package to if it works.

I have packaged this 3 commits (Also based on the dependency, patch the following commit to make them works. mesa_1_c157cf.diff  mesa_2_59b2c2.diff mesa_3_59b2c2.diff  mesa_4_2fc9d6.diff  mesa_5_c30eb2.diff  mesa_5_7ce59750.diff)

I still find the hang problems. Also some times googleearth will crash(xserver will not).

We come across 2 cases.
1) After hang, wait a minute, the google earth crash. And then xserver works normally(not exit). (print the following debug info)

2) Sometimes googleearth will directly crash with the following info. xserver works normally

both 
----------------
Stacktrace from glibc:
  ./googleearth-bin [0x804f403]
  ./googleearth-bin [0x804f916]
  [0xffffe400]
  /lib/libc.so.6(abort+0x188) [0xb71e12c8]
  /usr/lib/libdrm_intel.so.1 [0xb4e0841d]
  /usr/lib/libdrm_intel.so.1 [0xb4e093a5]
  /usr/lib/libdrm_intel.so.1 [0xb4e096de]
  /usr/lib/libdrm_intel.so.1(drm_intel_bo_exec+0x2e) [0xb4e0734e]
  /usr/lib/dri/i965_dri.so(_intel_batchbuffer_flush+0x195) [0xac9dbe2e]
  /usr/lib/dri/i965_dri.so(intelCopyBuffer+0x8a9) [0xac9dcb5d]
  /usr/lib/dri/i965_dri.so(intelSwapBuffers+0x4de) [0xac9e02c1]
  /usr/lib/dri/i965_dri.so [0xac9d8a8b]
  /usr/lib/libGL.so.1(glXSwapBuffers+0x49) [0xb66bb169]
  /opt/google-earth/libIGGfx.so(_ZN3Gap3Gfx18igOglVisualContext19internalSwapBuffersEv+0x7d) [0xb4e7e1bd]
  /opt/google-earth/libIGGfx.so(_ZN3Gap3Gfx18igOglVisualContext7endDrawEv+0x3a) [0xb4e9333a]
  /opt/google-earth/libevll.so(_ZN5earth4evll13VisualContext8endFrameEv+0x28f) [0xb395d5ff]
  /opt/google-earth/libevll.so(_ZN5earth4evll13VisualContext4drawEbb+0x1d5) [0xb3962745]
  /opt/google-earth/libevll.so(_ZN5earth4evll17RenderContextImpl4drawEv+0x166) [0xb3aff2f6]
  ./librender.so(_ZN12RenderWidget10paintEventEP11QPaintEvent+0x2a) [0xb6835b6a]
  ./librender.so(_ZN5earth6render11RenderTimer4fireEv+0x1a) [0xb6842fba]
  ./libbase.so(_ZN5earth5Timer8dispatchEv+0x2d) [0xb718003d]
  ./libgoogleearth_lib.so(_ZN5earth6client11Application18CommandCustomEvent8dispatchEv+0x23) [0xb73be483]
  ./libgoogleearth_lib.so(_ZN5earth6client11Application11customEventEP6QEvent+0x3a) [0xb73afe4a]
  ./libQtCore.so.4(_ZN7QObject5eventEP6QEvent+0xed) [0xb7e929a1]
  ./libQtGui.so.4(_ZN19QApplicationPrivate13notify_helperEP7QObjectP6QEvent+0x1d3) [0xb788a4ab]
  ./libQtGui.so.4(_ZN12QApplication6notifyEP7QObjectP6QEvent+0xb4) [0xb7891020]
  ./libQtCore.so.4(_ZN16QCoreApplication14notifyInternalEP7QObjectP6QEvent+0x62) [0xb7e84eda]
  ./libQtCore.so.4(_ZN23QCoreApplicationPrivate16sendPostedEventsEP7QObjectiP11QThreadData+0x247) [0xb7e869db]
  ./libQtCore.so.4(_ZN16QCoreApplication16sendPostedEventsEP7QObjecti+0x23) [0xb7e86c4f]
  ./libQtCore.so.4 [0xb7ea4a9a]
  /usr/lib/libglib-2.0.so.0(g_main_context_dispatch+0x1e8) [0xb6cf29a8]
  /usr/lib/libglib-2.0.so.0 [0xb6cf6063]
  /usr/lib/libglib-2.0.so.0(g_main_context_iteration+0x71) [0xb6cf6221]
  ./libQtCore.so.4(_ZN20QEventDispatcherGlib13processEventsE6QFlagsIN10QEventLoop17ProcessEventsFlagEE+0x53) [0xb7ea461b]
  ./libQtGui.so.4 [0xb78f8cf2]
  ./libQtCore.so.4(_ZN10QEventLoop13processEventsE6QFlagsINS_17ProcessEventsFlagEE+0x2d) [0xb7e84861]
  ./libQtCore.so.4(_ZN10QEventLoop4execE6QFlagsINS_17ProcessEventsFlagEE+0xc7) [0xb7e84a07]
  ./libQtCore.so.4(_ZN16QCoreApplication4execEv+0x98) [0xb7e86cf0]
  ./libQtGui.so.4(_ZN12QApplication4execEv+0x25) [0xb7889f75]
  ./libgoogleearth_lib.so(_ZN5earth6client11Application3runEv+0x380) [0xb73bc570]
  ./googleearth-bin(main+0x2ba) [0x8050bea]
  /lib/libc.so.6(__libc_start_main+0xe5) [0xb71cb705]
  ./googleearth-bin [0x804f231]
----

Just tested with mesa-7.3b1, libdrm-2.4.3, and intel-2.5.99.2 but the bug is still alive: Tile Racer completely freezes the system

Moreover osd is now corrupted in mplayer (gl driver).

I have observed this bug as well with my intel x4500hd video card,
and found the following kernel messages:
Jan 12 22:55:41 linux-w1pw kernel: [drm:i915_getparam] *ERROR* Unknown parameter 5
Jan 12 22:55:48 linux-w1pw kernel: [drm:i915_wait_irq] *ERROR* EBUSY -- rec: 29122 emitted: 29125

Created attachment 21963 [details] [review]
bug patch

This is the patch based on Q3 release.
(the patch is coming from commit id 59b2c2adbbece27ccf54e58b598ea29cb3a5aa85,
2fc9d671e0be77676818da51cd3b35b52af79421,
c30eb2c104c19c758c0b2a2461ad7f38cfa747d0
)
I have tested it. It works fine on G965.
You can have a try. 

Don't need to set ExaNoComposite

GoogleEarth still hangs the Xserver with the patch applied (EXANoComposite not set). Works fine with patch applied and EXANoComposite set.

*** Bug 18433 has been marked as a duplicate of this bug. ***

*** Bug 19162 has been marked as a duplicate of this bug. ***

(In reply to comment #49)
> GoogleEarth still hangs the Xserver with the patch applied (EXANoComposite not
> set). Works fine with patch applied and EXANoComposite set.

I was wrong. It works fine with this patch with EXANoComposite *not* set. Looks like I forgot to replace 32bit Mesa libs before. Closing as FIXED. The patch will be in SLE11-RC3.

(In reply to comment #52)
> I was wrong. It works fine with this patch with EXANoComposite *not* set. Looks
> like I forgot to replace 32bit Mesa libs before. Closing as FIXED. The patch
> will be in SLE11-RC3.
> 

Stephan, would it also be put in http://download.opensuse.org/repositories/X11:/XOrg/openSUSE_11.1/

or should I patch it manually?

(In reply to comment #53)
> Stefan, would it also be put in
> http://download.opensuse.org/repositories/X11:/XOrg/openSUSE_11.1/

The patch is already included.

I update Mesa, libdrm, xorg from http://download.opensuse.org/repositories/X11:/XOrg/openSUSE_11.1/

In my case (GM45, openSUSE 11.1, KDE 3.5.10), I can run google earth smoothly if I disable compiz (desktop effect). 

If I enable compiz then google earth is running (not crash) but with worst graphics closed to not viewable condition, even with ExaNoComposite is set. X Server is not crash.

Should I open a new bug report? 

shouldn't the distribution-specific topics be posted elsewere?

(In reply to comment #56)
> shouldn't the distribution-specific topics be posted elsewere?
> 

That's why I ask ;-)
But in my understanding openSUSE just use the same version with the upstream version, CMIIW, and somehow they will also forward the related problem to here. 

But anyway, I really thanks for the latest patch from quanxian, I can live with that :-)

I think 3D application performance with intel driver has always been bad when compiz has been enabled.

openSUSE 11.1 is just using 2008 Q3 release. So feel free to replace openSUSE 11.1 with Intel 2008 Q3 release. :-)

Updating to the repository provided now allows me to use Google Earth without problems. Thank you for finding the cause of the issue.

Stefan: Any hope of this getting pushed as an update for openSUSE 11.1?

It's also working here.

Thanks Haihao's help for this problem. He provides the patches. I just make a little adjustment based Intel Q3 release. 

Thanks Haihao for your great support.

Upgraded to mesa-7.3b3 and intel-2.6.0, but I still have Xserver freezes.

A good test comes from TileRacer, an opengl game.

With composite enabled it's impossible to play because the screen gets dark just after launching the game and have to esc pressing ESCAPE.
Without composite the game menu appears but after pressing "play" the system freezes. This should not happen in any case!

Do I have to open another bug? The topic seems the same...

R.

Do you try the new patch?
http://download.opensuse.org/repositories/X11:/XOrg/openSUSE_11.1/

Find the mesa packages, and install it. Try it

I've been testing this since 15 january and it's really working fine.
But with this update I've got the problem described in http://bugs.freedesktop.org/show_bug.cgi?id=14427

(In reply to comment #63)
> Do you try the new patch?
> http://download.opensuse.org/repositories/X11:/XOrg/openSUSE_11.1/
> 
> Find the mesa packages, and install it. Try it
> 

I'm not on suse, but debian.

isn't the patch included in mesa7.3rc3??

is tile racer working on your system?

i bet it isn't...

I've tried TileRacer here, but I've got a segmentation fault from it.
And I don't think this problem is related to the X server.
Everything else is working fine.

after upgrading xserver to 1.6beta Tile Racer doesn't freeze the system anymore, but exits ad you sad. The game works properly on the same system with a GeForce 7600 and proprietary drivers.

But I also encounter freezes EVERY time I use smplayer with gl video output.
Please follow these steps:

1) open a video with smplayer
2) enter the settings and set gl for the video output
3) try to seek moving the seek bar.

In my case it freezes the screen.

I also found out that smplayer freezes the screen if used after suspend-to-ram, even if xv output is used! This didn't happen before updating.

Should I open a new bug?

(In reply to comment #69)
> I also found out that smplayer freezes the screen if used after suspend-to-ram,
> even if xv output is used! This didn't happen before updating.
> 
> Should I open a new bug?

Try with the latest drivers. Please file a new bug if this issue still exists and provide the necessary info according to http://intellinuxgraphics.org/how_to_report_bug.html.
 

Created attachment 22966 [details]
Xorg logfile

My Xserver still hangs at the startup of google-earth. I hope, that you can use my Xorg logfile to find the error. I'm running OpenSuSE 11.1 with updates in 
http://download.opensuse.org/repositories/X11:/XOrg/openSUSE_11.1/
I'm using the package xorg-x11-driver-video-7.4-47.6 with my intel x4500hd video card.

Apparently it's not fixed yet. I hope you don't mind I reopened it.

I can confirm that X still hangs with some openGL screensavers (from standerd kde4.2 package)

We have done the testing on GM45(Montevina-X200) G45(McCreary-Offical),
965GM(Santa-T61), 945GM(x60) including x86 and x86_64 bits. GoogleEarth works
fine. 

The release is used by SLED11-RC4. 

I ever tried it on SLED11-RC1 release with latest graphics packages.
Googleearth crashes. :(. Seems google earth doesn't work with the environment.

Therefore I think the crash is not related with driver. I suggest you try
latest packages on SLED11-RC4. It should be OK.

Mark this bug as fixed, (see commit #74).

(In reply to comment #74)
> We have done the testing on GM45(Montevina-X200) G45(McCreary-Offical),
> 965GM(Santa-T61), 945GM(x60) including x86 and x86_64 bits. GoogleEarth works
> fine. 

The Bug title says "3D apps"; the fact that GoogleEarth is working doesn't mean that other 3D apps can not crash the server. For example I couldn't run TileRacer without freezing the screen. Does it work now?

Please try the latest drivers. If it still exists, please file a new bug to track your specific issue.