maybe related to bug#26053

Can you try again with mesa master?

Probably a dupe.

*** This bug has been marked as a duplicate of bug 26053 ***

This issue still exists on all platform. It seems this is not a dupe of bug#26053.

You ran with the same bits as Shuang He from bug 26053?  gnome-session works ok for me, even with compiz enabled.

(In reply to comment #5)
> You ran with the same bits as Shuang He from bug 26053?  

Yes.

If you were running with an old kernel, this may have been fixed by

commit 6610bcbac51c9ac970128012f0d4566d8cfba000
Author: Jesse Barnes <jbarnes@virtuousgeek.org>
Date:   Fri Feb 5 17:18:50 2010 -0800

    DRI2: only use version 4 APIs if kernel support exists

(In reply to comment #7)
> If you were running with an old kernel, this may have been fixed by
> commit 6610bcbac51c9ac970128012f0d4566d8cfba000
> Author: Jesse Barnes <jbarnes@virtuousgeek.org>
> Date:   Fri Feb 5 17:18:50 2010 -0800
>     DRI2: only use version 4 APIs if kernel support exists

We tested with newest code on drm-intel-next. And I also updated our driver to the newest(6610bcbac51c9ac970128012f0d4566d8cfba000 included), but it still fails to start gnome with compiz on. 
Libdrm:         (master)1802e1a4e747b5906d3af10c4a53fd457eddcbb4
Mesa:           (master)5a743d9c513e6a36612872f270ea97b269451280
Xserver:                (master)db687f718f760ba254ab51994769db101dc9ca3a
Xf86_video_intel:       (master)41784e15d39c29af5a87fe8d0982bd1b0be3562d
Kernel: (drm-intel-next)a9c8bea64706f86a9107ab41e7a3e284d1508f9c

and when it stucks it has the following backtrace: 
(nm-applet:7282): GLib-GObject-CRITICAL **: g_object_unref: assertion `G_IS_OBJECT (object)' failed
Acceleration key: disabled
XSETTINGS manager support is explicitly disabled.
Traceback (most recent call last):
  File "/usr/bin/sealert", line 108, in <module>
    from setroubleshoot.analyze import *
  File "/usr/lib/python2.5/site-packages/setroubleshoot/analyze.py", line 43, in <module>
    from setroubleshoot.avc_audit import *
  File "/usr/lib/python2.5/site-packages/setroubleshoot/avc_audit.py", line 54, in <module>
    my_context = AvcContext(selinux.getcon()[1])
OSError: [Errno 22] Invalid argument
compiz (video) - Warn: No 8 bit GLX pixmap format, disabling YV12 image format
[mi] EQ overflowing. The server is probably stuck in an infinite loop.

Backtrace:
0: X (xorg_backtrace+0x28) [0x466f98]
1: X (mieqEnqueue+0x1ae) [0x4a2fee]
2: X (xf86PostMotionEventP+0xd8) [0x471188]
3: /opt/X11R7/lib/xorg/modules/input/evdev_drv.so (0x7f7f4d284000+0x50ee) [0x7f7f4d2890ee]
4: X (0x400000+0x72307) [0x472307]
5: X (0x400000+0x1111c7) [0x5111c7]
6: /lib64/libc.so.6 (0x3b3ae00000+0x32f90) [0x3b3ae32f90]
7: /lib64/libc.so.6 (ioctl+0x7) [0x3b3aede037]
8: /opt/X11R7/lib/libdrm_intel.so.1 (0x7f7f4d68f000+0x5326) [0x7f7f4d694326]
9: /opt/X11R7/lib/xorg/modules/drivers/intel_drv.so (0x7f7f4d899000+0x124f8) [0x7f7f4d8ab4f8]
10: /opt/X11R7/lib/xorg/modules/drivers/intel_drv.so (0x7f7f4d899000+0x1420d) [0x7f7f4d8ad20d]
11: X (0x400000+0xd38a8) [0x4d38a8]
12: X (0x400000+0x124d09) [0x524d09]
13: X (BlockHandler+0x50) [0x439380]
14: X (WaitForSomething+0x161) [0x46c571]
15: X (0x400000+0x29018) [0x429018]
16: X (0x400000+0x216bd) [0x4216bd]
17: /lib64/libc.so.6 (__libc_start_main+0xe6) [0x3b3ae1e576]
18: X (0x400000+0x21269) [0x421269]

I wonder if this is the same as 26064; please check the kernel patch mentioned in that bug.

(In reply to comment #10)
> I wonder if this is the same as 26064; please check the kernel patch mentioned
> in that bug.

Is that patch will work for 945GM and other platforms other than Ironlake? I
have patched it to drm-intel-next on 945GM, and it still fails to start gnome-session with compiz on. 

Arg, I don't know why it works for me and not for you.  I've even got the new bits running on Ironlake now with no issues.  This is probably a dupe of #26064, but that doesn't make any sense to me either.

With current code, I find it works fine on some machines with 64-bit fedora operating system (Piketon,G45). It still fails on 945gme and calpella(os is 32-bit). I make a 32-bit OS and try it on Piketon, G45, it also fails. Maybe it related to 32 and 64-bit OS?

With bug#26814 fixed, it works well on i965 platform(G45,piketon,calpella,965gm and gm45). But it still fails on i915
platform(945gm and apsire1).

Mark as a dup, if not, reopen. 

*** This bug has been marked as a duplicate of bug 26064 ***

not dup. bug#26064 has gone around March 6th, while this one still exists.

If we disable the pageflip, then the compiz or other full screen applications work well. 

Did you test with the patch from Li Peng that landed today?

(In reply to comment #18)
> Did you test with the patch from Li Peng that landed today?

No, I haven't yet. But I will update 2D driver and have a test today. 

With the newest code on master, it works well now. But with mesa on 7.8 branch, compiz will crash X. As bug #27258.