[i965] X (and gnome session) crashes after suspend and resume

[Martin Pool]

Binary package hint: xserver-xorg

This seems to be a regression in the last few days of Jaunty updates.

I have a Thinkpad X61s. If I suspend and resume then the machine comes back up ok, except that more than 50% of the time my gnome session has disappeared and I'm back at the login screen. This has not yet happened when coming back from hibernate, but I've only tried a couple of times.

The bottom of my Xorg.0.log has

exaCopyDirty: Pending damage region empty!

and daemon.log has

Mar 10 14:30:58 lithe gdm[4126]: WARNING: gdm_slave_xioerror_handler: Fatal X error - Restarting :0

[Bryce Harrington (bryce)]

Martin, the exaCopyDirty warning is non-fatal I think.
Look in your /var/log/gdm/* files for better error messages.

Bryce

On Tue, Mar 10, 2009 at 04:44:54AM -0000, Martin Pool wrote:
> Public bug reported:
>
> Binary package hint: xserver-xorg
>
> This seems to be a regression in the last few days of Jaunty updates.
>
> I have a Thinkpad X61s. If I suspend and resume then the machine comes
> back up ok, except that more than 50% of the time my gnome session has
> disappeared and I'm back at the login screen. This has not yet happened
> when coming back from hibernate, but I've only tried a couple of times.
>
> The bottom of my Xorg.0.log has
>
> exaCopyDirty: Pending damage region empty!
>
> and daemon.log has
>
> Mar 10 14:30:58 lithe gdm[4126]: WARNING: gdm_slave_xioerror_handler:
> Fatal X error - Restarting :0
>
> ** Affects: xorg (Ubuntu)
> Importance: Undecided
> Status: New
>
> --
> gnome session crashes after suspend and resume
> https://bugs.launchpad.net/bugs/340328
> You received this bug notification because you are a member of Ubuntu-X,
> which is subscribed to xorg in ubuntu.

[Martin Pool (mbp)]

There's just the same message in there.

[Martin Pool (mbp)]

oh and also

I830PMEvent: Capability change

[Bryce Harrington (bryce)]

Aha, 'PM' stands for 'Power Management' there. So that's perhaps a good clue. But it's not associated with an abort or other termination logic, so isn't a smoking gun.

Looking at the code, the next thing that happens after that warning is it resets the screensaver, and then sets up the desired modes. Hard to say if it crashes at that point, or continues on and crashes later.

Since you say that the error began occurring within the last few days, could you review your /var/log/dpkg.log for things that changed at about that time, and try downgrading relevant packages (kernel, -intel, xserver, etc.) to see if you can identify which change resulted in the regression?

None of the recent changes in -intel look like they could trigger a problem on your hardware, especially not suspend/resume issues. So my first guess would be to look at the kernel or other system level things.

[Martin Pool (mbp)]

Here's my dpkg log. I'm at a sprint and can't try that this week but might next week. Thanks for the quick response.

[Martin Pool (mbp)]

I don't know if this is the same problem or a different one, but I've twice now woken up my machine and seen it strangely responsive to input, as if input was grabbed. I can't click on any of the panel applets, or type in the terminal that has focus, or switch focus using alt-tab. However, I can press ctrl-alt-d to show the desktop, and I can press Alt-f2 to bring up the run dialog, then start a new terminal, and keyboard input to that terminal does work properly. I tried, from that terminal, restarting metacity, gnome-panel and nautilus, and that did not fix input. When I restarted gdm, killing my session, everything worked ok.

[unggnu (unggnu)]

I am not sure but why don't you use Apport to get more information. I always get an Apport message if suspend doesn't work or X crashes.

[Biji (biji)]

I experience this also,

daemon.log:
WARNING: gdm_slave_xioerror_handler: Fatal X error - Restarting :0

Xorg.log:
(II) AIGLX: Suspending AIGLX clients for VT switch

pm-suspend.log
Apr 18 22:00:48 onyx kernel: [10600.450445] ------------[ cut here ]------------
Apr 18 22:00:48 onyx kernel: [10600.450447] WARNING: at /build/buildd/linux-2.6.28/kernel/power/main.c:177 suspend_test_finish+0x7c/0x80()
Apr 18 22:00:48 onyx kernel: [10600.450448] Component: resume devices
Apr 18 22:00:48 onyx kernel: [10600.450449] Modules linked in: tun pppoe pppox aes_x86_64 aes_generic ecb binfmt_misc i915 drm bridge stp b
nep vboxnetflt vboxdrv input_polldev nls_iso8859_1 nls_cp437 vfat fat joydev lp parport psmouse pcspkr serio_raw usbhid iTCO_wdt iTCO_vendor_support snd_hda_intel snd_pcm_oss snd_mixer_oss snd_pcm snd_seq_dummy ieee80211_crypt_tkip snd_seq_oss snd_seq_midi snd_rawmidi wl(P) ieee80211_crypt snd_seq_midi_event snd_seq snd_timer snd_seq_device snd soundcore snd_page_alloc intel_agp video output usb_storage tg3 fbcon tileblit font bitblit softcursor
Apr 18 22:00:48 onyx kernel: [10600.450482] Pid: 15779, comm: pm-suspend Tainted: P 2.6.28-11-generic #42-Ubuntu
Apr 18 22:00:48 onyx kernel: [10600.450484] Call Trace:
Apr 18 22:00:48 onyx kernel: [10600.450490] [<ffffffff80250927>] warn_slowpath+0xb7/0xf0
Apr 18 22:00:48 onyx kernel: [10600.450493] [<ffffffff8026d478>] ? down_trylock+0x38/0x50
Apr 18 22:00:48 onyx kernel: [10600.450496] [<ffffffff80251040>] ? try_acquire_console_sem+0x10/0x40
Apr 18 22:00:48 onyx kernel: [10600.450500] [<ffffffff8025c676>] ? lock_timer_base+0x36/0x70
Apr 18 22:00:48 onyx kernel: [10600.450505] [<ffffffff8069bf36>] ? printk+0x67/0x69
Apr 18 22:00:48 onyx kernel: [10600.450509] [<ffffffff80418877>] ? kobject_put+0x27/0x60
Apr 18 22:00:48 onyx kernel: [10600.450513] [<ffffffff804b6045>] ? put_device+0x15/0x20
Apr 18 22:00:48 onyx kernel: [10600.450517] [<ffffffff804be06a>] ? dpm_complete+0x18a/0x1a0
Apr 18 22:00:48 onyx kernel: [10600.450519] [<ffffffff8028003c>] suspend_test_finish+0x7c/0x80
Apr 18 22:00:48 onyx kernel: [10600.450522] [<ffffffff80280124>] suspend_devices_and_enter+0xe4/0x180
Apr 18 22:00:48 onyx kernel: [10600.450525] [<ffffffff802803d9>] enter_state+0xe9/0x120
Apr 18 22:00:48 onyx kernel: [10600.450527] [<ffffffff802804ca>] state_store+0xba/0x100
Apr 18 22:00:48 onyx kernel: [10600.450529] [<ffffffff80418717>] kobj_attr_store+0x17/0x20
Apr 18 22:00:48 onyx kernel: [10600.450533] [<ffffffff80347675>] sysfs_write_file+0xc5/0x140
Apr 18 22:00:48 onyx kernel: [10600.450536] [<ffffffff802e7eeb>] vfs_write+0xcb/0x130
Apr 18 22:00:48 onyx kernel: [10600.450539] [<ffffffff802e8040>] sys_write+0x50/0x90
Apr 18 22:00:48 onyx kernel: [10600.450542] [<ffffffff8021253a>] system_call_fastpath+0x16/0x1b
Apr 18 22:00:48 onyx kernel: [10600.450544] ---[ end trace 79f4c7b1e8f64ec5 ]---
Apr 18 22:00:48 onyx kernel: [10600.450580] PM: Finishing wakeup.
Apr 18 22:00:48 onyx kernel: [10600.450581] Restarting tasks ... done.
Apr 18 22:00:48 onyx kernel: [10600.636051] ggl-gtk[15027]: segfault at 7f12d0dab2c0 ip 00007f12dabd1034 sp 00007fffe5b68a80 error 4 in libggadget-1.0.so.0.0.0[7f12da903000+...

[Tobias Wolf (towolf)]

I also experience this on intel (Thinkpad X301).

My gdm logs all terminate with this line:

X: ../../src/i830_batchbuffer.h:78: intel_batch_emit_dword: Assertion `pI830->batch_ptr != ((void *)0)' failed.

[Matthew Weier O'Phinney (mweierophinney)]

I experience this on a Thinkpad T61 running Jaunty, and my GDM log terminates exactly as that reported by Tobias Wolf (intel_batch_emit_dword failure).

[Tobias Wolf (towolf)]

Matthew, I moved to use the drivers from the Xorg edgers PPA. I don’t get that problem anymore. This PPA is aptly named, but it can’t do harm to try it and revert if graphics become unstable.

https://edge.launchpad.net/~xorg-edgers/+archive/ppa

[Matthew Weier O'Phinney (mweierophinney)]

Tobias, I'm using that PPA already -- though I need to verify that I'm using the xserver-xorg-video-intel driver (I assume I am, as I now have compositing, but need to verify).

Can you give your xorg.conf? I'm wondering if there are settings that are causing the issues.

[Tobias Wolf (towolf)]

Perhaps it is "Tiling"? Not sure anymore. But not having this in was problematic in some sense. I have nothing else in there.

$ cat /etc/X11/xorg.conf
Section "Device"
  Identifier "Configured Video Device"
  Option "Tiling" "true"
EndSection

Section "Monitor"
 Identifier "Configured Monitor"
EndSection

Section "Screen"
 Identifier "Default Screen"
 Monitor "Configured Monitor"
 Device "Configured Video Device"
EndSection

[Matthew Weier O'Phinney (mweierophinney)]

That was indeed the key -- I had some other garbage per the ubuntu forums post. Removing it and enabling tiling now has my ThinkPad correctly resuming.

So, I can confirm that the Xorg edgers PPA kernel and xorg packages resolve these issues for me.

[Bryce Harrington (bryce)]

Martin, could you re-test after booting a karmic livecd of alpha-2 or newer? If the issue can be reproduced there, we can send it upstream.

[Martin Pool (mbp)]

I upgraded this laptop to current Karmic on the weekend and didn't see
this problem in a few suspend/resume cycles.

--
Martin <http://launchpad.net/~mbp/>

[Bryce Harrington (bryce)]

Thanks for letting us know the issue is resolved in karmic.