Bug 23032 - [i845] xorg hang on video driver 2.8.0
Summary: [i845] xorg hang on video driver 2.8.0
Status: RESOLVED DUPLICATE of bug 26345
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/intel (show other bugs)
Version: 7.4 (2008.09)
Hardware: x86 (IA32) Linux (All)
: high critical
Assignee: Chris Wilson
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords: NEEDINFO
Depends on:
Blocks:
 
Reported: 2009-07-29 20:14 UTC by Bryce Harrington
Modified: 2010-03-18 13:41 UTC (History)
5 users (show)

See Also:
i915 platform:
i915 features:


Attachments
dri_debug.tgz (263.03 KB, application/x-compressed-tar)
2009-07-29 20:14 UTC, Bryce Harrington
no flags Details
Xorg.0.log.845 (18.74 KB, application/octet-stream)
2009-07-29 20:15 UTC, Bryce Harrington
no flags Details
Copied from dual boot (56.14 KB, application/octet-stream)
2009-08-13 08:31 UTC, Jerry Amos
no flags Details
dmesg from the hang (31.95 KB, application/octet-stream)
2009-08-13 08:31 UTC, Jerry Amos
no flags Details
.xsession-errors from the hang (1.23 KB, application/octet-stream)
2009-08-13 08:32 UTC, Jerry Amos
no flags Details
lspci from the hang pc (6.63 KB, application/octet-stream)
2009-08-13 08:33 UTC, Jerry Amos
no flags Details
same here, see intel_gpu_dump info (209 bytes, application/x-gzip)
2009-12-29 11:02 UTC, Elmo R
no flags Details
xorg crash with 2.6.32-15 kernel (202.38 KB, text/x-apport)
2010-03-04 06:11 UTC, Jerry Amos
no flags Details
Also got this crash report (7.18 KB, text/x-apport)
2010-03-04 06:14 UTC, Jerry Amos
no flags Details

Description Bryce Harrington 2009-07-29 20:14:07 UTC
Forwarding this bug from Ubuntu reporter jerrylamos:
http://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/406460

[Problem]
xorg hang on video driver 2.8.0 on i845 hardware.  Batchbuffer enclosed.

[Original Description]
i845 karmic 2.6.31-4 KMS intel driver 2.8.0 boots recovery mode, root prompt dhclient, then resume.  Black screen no keyboard or mouse response.

ssh logged in v-e-r-y slowly.  Key in response v-e-r-y slow.  Entered "ls" no response for the last several minutes.

Will power off and log on to dual boot to get logs which I'll enter as subsequent attachments.

i845 fails before login, earlier than i830 which fails after login see bug #403037.

Do note i845 runs O.K. for an alpha 3  using the same 2.6.31-4 however "nomodeset" and driver "vesa".

Here's a try at the batchbuffer dump.

 Booted recovery mode
 root prompt dhclient
 resume this time it got to login prompt
 ssh into test system
 logged in, this time it got to the brown screen with pointer then stopped. The pointer would  move but the whirligig didn't whirl.

Accidentally found out if I moved the mouse between each key entry I could get commands in.
Followed the batchbuffer dump directions. Copy a line, paste into the ssh, move the pointer, push enter, move the pointer. One shortcut was to type in the commands into text editor, then copy and paste into ssh.

Let me know if the batchbuffer dump has anything useful. I could try it on the 830 as well if it is worth it.

Jul 29 11:34:12 linux kernel: [  360.760037] INFO: task events/0:6 blocked for more than 120 seconds.
Jul 29 11:34:12 linux kernel: [  360.760047] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
Jul 29 11:34:12 linux kernel: [  360.760052] events/0      D c0810360     0     6      2 0x00000000
Jul 29 11:34:12 linux kernel: [  360.760061]  f7043f04 00000046 f7033efc c0810360 f7034168 c0810360 46c83c5d 0000002e
Jul 29 11:34:12 linux kernel: [  360.760072]  c0810360 c0810360 f7034168 c0810360 46c82505 0000002e c0810360 ef70dc00
Jul 29 11:34:12 linux kernel: [  360.760082]  f7033ed0 f73f0014 f73f0018 ffffffff f7043f30 c056a166 c073a180 f73f001c
Jul 29 11:34:12 linux kernel: [  360.760092] Call Trace:
Jul 29 11:34:12 linux kernel: [  360.760111]  [<c056a166>] __mutex_lock_slowpath+0xc6/0x130
Jul 29 11:34:12 linux kernel: [  360.760119]  [<c056a080>] mutex_lock+0x20/0x40
Jul 29 11:34:12 linux kernel: [  360.760159]  [<f81ae5ba>] i915_gem_retire_work_handler+0x2a/0x70 [i915]
Jul 29 11:34:12 linux kernel: [  360.760173]  [<c01520ae>] run_workqueue+0x6e/0x140
Jul 29 11:34:12 linux kernel: [  360.760195]  [<f81ae590>] ? i915_gem_retire_work_handler+0x0/0x70 [i915]
Jul 29 11:34:36 linux console-kit-daemon[2224]: WARNING: Couldn't read /proc/2222/environ: Failed to open file '/proc/2222/environ
': No such file or directory
Jul 29 11:40:12 linux kernel: [  360.760205]  [<c0152208>] worker_thread+0x88/0xe0
Jul 29 11:40:12 linux kernel: [  360.760212]  [<c0156880>] ? autoremove_wake_function+0x0/0x40
Jul 29 11:40:12 linux kernel: [  360.760219]  [<c0152180>] ? worker_thread+0x0/0xe0
Jul 29 11:40:12 linux kernel: [  360.760225]  [<c0156594>] kthread+0x74/0x80
Jul 29 11:40:12 linux kernel: [  360.760230]  [<c0156520>] ? kthread+0x0/0x80
Jul 29 11:40:12 linux kernel: [  360.760238]  [<c0103f17>] kernel_thread_helper+0x7/0x10
Comment 1 Bryce Harrington 2009-07-29 20:14:50 UTC
Created attachment 28180 [details]
dri_debug.tgz
Comment 2 Bryce Harrington 2009-07-29 20:15:50 UTC
Created attachment 28181 [details]
Xorg.0.log.845

00:00.0 Host bridge [0600]: Intel Corporation 82845G/GL[Brookdale-G]/GE/PE DRAM Controller/Host-Hub Interface [8086:2560] (rev 01)
	Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B- DisINTx-
00:02.0 VGA compatible controller [0300]: Intel Corporation 82845G/GL[Brookdale-G]/GE Chipset Integrated Graphics Device [8086:2562] (rev 01)
	Subsystem: IBM Device [1014:0267]
Comment 3 Wang Zhenyu 2009-08-06 01:23:00 UTC
Could you test my patches on bug #23082?
Comment 4 Jerry Amos 2009-08-06 12:31:36 UTC
(In reply to comment #3)
> Could you test my patches on bug #23082?
> I know enough linux to do some limited fix installation.  For example Bryce Harrington built a PPA:

deb http://ppa.launchpad.net/bryceharrington/blue/ubuntu karmic main
deb http://us.archive.ubuntu.com/ubuntu/ karmic-proposed restricted main multiverse universe
deb-src http://ppa.launchpad.net/bryceharrington/blue/ubuntu karmic main

which is an xserver-xorg-video-intel 2.8.0-0ubuntu2~bug382017

which still hangs on Ubuntu A3, namely kernels 2.6.31-4 and -5.  
It does run on Ubuntu A2 kernel 2.6.31-3.

Today's Ubuntu update resulted in 

2:2.8.0-0ubuntu2

which also hangs on karmic Alpha 3 kernel 2.6.31-5.

Jerry



Comment 5 Larry Reid 2009-08-13 07:35:46 UTC
I filed Ubuntu bug 412245 (https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/412245). I have the same symptoms reported here. In particular, when X is hung, I can still use an ssh session from another machine, but I have to move the mouse pointer on the hung machine in order to see any output in the ssh session.

I'm running Ubuntu 9.10 Alpha 3 on a Dell Inspiron 1100 with BIOS rev A32.
lspci -nn | grep VGA
00:02.0 VGA compatible controller [0300]: Intel Corporation 82845G/GL[Brookdale-G]/GE Chipset Integrated Graphics Device [8086:2562] (rev 03)

You can see more details and log files at the Ubuntu bug, or ask me to provide any further information you would find useful. I'd be happy to test any fixes.

I tried adding the PPA in comment 4 and update/upgrade, but there was no change.

If you really need me to try kernel patches as described in bug 23082 I could. It's been many years since I built kernels and I'd appreciate pointers to recent documentation about how you build kernels, if there is any.
Comment 6 Jerry Amos 2009-08-13 08:30:13 UTC
Today's update on karmic the cursor either won't move or else it is absent as I try CD Live 20090812 or hard disk update.  ssh won't work.  I boot in recovery mode, get ssh going, then resume.  brown screen with no cursor or else black screen fixed cursor. ssh doesn't work.  Will attach Xorg.0.log and dmesg as obtained by dual booting & copying the files.

If I boot in recovery, do nano /etc/X11/xorg.conf to change to Driver "vesa" it runs.  I'm using it now.  Default xserver-xorg-video-intel 2:2.8.0-0ubuntu2 is the one that hangs.

Jerry
Comment 7 Jerry Amos 2009-08-13 08:31:19 UTC
Created attachment 28593 [details]
Copied from dual boot
Comment 8 Jerry Amos 2009-08-13 08:31:54 UTC
Created attachment 28594 [details]
dmesg from the hang
Comment 9 Jerry Amos 2009-08-13 08:32:29 UTC
Created attachment 28595 [details]
.xsession-errors from the hang
Comment 10 Jerry Amos 2009-08-13 08:33:55 UTC
Created attachment 28596 [details]
lspci from the hang pc
Comment 11 Eric Anholt 2009-09-01 17:39:17 UTC
The "have to move the mouse to get ssh to show up" should be fixed in .31 release (it was rc4 or so that I got the fix merged).  The intel_gpu_dump command parsing fails on a bunch of these 8xx commands, so the dump attached ended up not showing the place where the GPU's hung.

I think we'll need to try disabling render acceleration for you, unfortunately it's not a xorg.conf option at this point, though.
Comment 12 Jerry Amos 2009-09-01 18:03:05 UTC
Agree the "have to move the mouse to get ssh to show up" is fixed in 2.6.31-9.

Boot still fails with xserver-xorg-video-intel 2:2.8.1-1ubuntu1.

Boot is successful with "vesa".  Runs fine (given some ubuntu Alpha bugs).

Is the "vesa" option going to be readily obvious to an ordinary linux Ubuntu user with i845 (and i830) video graphics?  Is there a grub boot option I could try?  There used to be a recovery mode option for this I think.

Thanks for looking at this bug.

Jerry
Comment 13 Jerry Amos 2009-09-03 18:36:48 UTC
Option "DRI" "0" in xorg.conf on default driver 2:2.8.1-1ubuntu1 does boot.

A few minutes later it hung.

With "DRI" "0" GtkPerf runs slowly three times longer than with driver "vesa".

So it's back to driver "vesa".  Could that be default for i845 and i830?

Jerry



Comment 14 Carl Worth 2009-09-11 14:02:45 UTC
Eric Anholt posted a kernel patch yesterday that fixes several hangs for
pre-9xx chipsets:

http://lists.freedesktop.org/archives/intel-gfx/2009-September/004122.html

Jerry, it would be very useful if you could test with that patch and report
here whether or not it fixes your bug.

Thanks,

-Carl
Comment 15 Jerry Amos 2009-09-12 15:00:39 UTC
Carl, wouldn't mind if I knew how to edit the driver source and re-compile?  Is that what is involved?  I haven't done patches.

About the earliest level I get into is when someone like Bryce Harrington makes a ppa or there's something to sudo apt-get install. 

Thanks for working on this.

Jerry
Comment 16 Gordon Jin 2009-11-01 23:10:05 UTC
Bryce, is the patch mentioned in comment#14 made into Ubuntu 9.10?
Comment 17 Jerry Amos 2009-11-02 16:42:29 UTC
Karmic release code now running on i845 and i830 pc's here, with
module version = 2.9.0 on ubuntu 2.6.31-14-generic 32 bit.

The i845 is getting occasional hangs however I don't have a good handle on whether it is Xorg or Firefox or kernel or Gnome as yet.

Jerry
Comment 18 Elmo R 2009-12-29 11:02:05 UTC
Created attachment 32355 [details]
same here, see intel_gpu_dump info
Comment 19 Chris Wilson 2010-03-02 07:59:43 UTC
There's not enough information here to classify this bug. It is most probably a dup of 26345. Perhaps if you try a recent kernel with "Record batch buffer at time of error" the i915_error_state would be enlightening.
Comment 20 Geir Ove Myhr 2010-03-02 12:46:36 UTC
(In reply to comment #19)
> There's not enough information here to classify this bug. It is most probably a
> dup of 26345. Perhaps if you try a recent kernel with "Record batch buffer at
> time of error" the i915_error_state would be enlightening.

Jerry, in order to do this with Ubuntu Lucid, you can install the newest kernel from http://kernel.ubuntu.com/~kernel-ppa/mainline/drm-intel-next/ (see https://wiki.ubuntu.com/KernelTeam/MainlineBuilds for information about how to install and test these kernels). Now, you only need to copy /sys/kernel/debug/dri/0/i915_error_state and not all the other files that you copied before.

Comment 21 Jerry Amos 2010-03-04 06:11:53 UTC
Created attachment 33753 [details]
xorg crash with 2.6.32-15 kernel

Will try to see how to do "Record batch buffer at time of error".

With latest kernel:

Linux version 2.6.32-15-generic (buildd@rothera) (gcc version 4.4.3 (Ubuntu 4.4.3-3ubuntu1) ) #22-Ubuntu SMP Tue Mar 2 02:24:17 UTC 2010

ran for several hours then crashed.  Will try to attach apport crash report.

Jerry
Comment 22 Jerry Amos 2010-03-04 06:14:03 UTC
Created attachment 33754 [details]
Also got this crash report

When the crash occurred, Firefox was up, just moving mouse over youtube.com page having previously played a video.

Jerry
Comment 23 Chris Wilson 2010-03-18 13:41:24 UTC
After spending more time banging my head against the Brookdale-G breakage, I can see the same GTT coherency issues in the GPU dumps.

*** This bug has been marked as a duplicate of bug 26345 ***


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.