[i845] xorg hang video driver 2.8.1-1ubuntu1

[jerrylamos]

apt-get update Aug 27 installed intel video driver 2.8.1-1ubuntu1 on karmic 2.6.31-7

Boot goes through automatic logon O.K., displays brown screen with faint stripes and ubuntu logo with bright highlight then cursor disappears and keyboard is dead except for Alt-SysRq-B which seems to screw up file mounts.

ssh in from another pc and followed http://wiki.ubuntu.com/X/troubleshooting as given in bug #382017

Rebooted with driver vesa to enter this bug.

Jerry

[lspci]
00:00.0 Host bridge [0600]: Intel Corporation 82845G/GL[Brookdale-G]/GE/PE DRAM Controller/Host-Hub Interface [8086:2560] (rev 01)
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]

[jerrylamos (jerrylamos)]

vesa does boot so get regdump

[jerrylamos (jerrylamos)]

Xorg.0.log.old

[jerrylamos (jerrylamos)]

/home/jerry/.xsession-errors

[jerrylamos (jerrylamos)]

I didn't have any luck with xandr after the hang from ssh; "Couldn't open display". So here's one from text console at recovery

[jerrylamos (jerrylamos)]

dmidecode

[jerrylamos (jerrylamos)]

karmic 2.6.31-7 i845 dmesg

[jerrylamos (jerrylamos)]

lspci -vvnn

[Nicola Ferralis (feranick)]

Thank you for taking the time to report this bug and helping to make Ubuntu better. This bug did not have a package associated with it, which is important for ensuring that it gets looked at by the proper developers. You can learn more about finding the right package at https://wiki.ubuntu.com/Bugs/FindRightPackage. I have classified this bug as a bug in xserver-xorg-video-intel.

[Bryce Harrington (bryce)]

Hi Jerry, thanks for including all the necessary info.

Btw, can you confirm that downgrading to the 2.8.0 driver solves it? If so, that'll help narrow down where to search for the cause of the regression.

[jerrylamos (jerrylamos)]

Bryce,

Only way 2.8.0 driver works is with ubuntu Alpha 2. The 2.8.0 driver was installed on Alpha 2 from a PPA you made available. As soon as Alpha 3 or anything later is used, driver 2.8.0 doesn't boot.

I haven't a clue how to narrow down which updates between Alpha 2 and Alpha 3 would have caused the problem. Any ideas?

There is an entry on Bugzilla bug 23032 which says "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." so it doesn't look like another intel_gpu_dump will help.

Thanks, Jerry

[Geir Ove Myhr (gomyhr)]

I have compiled three test kernels for finding the triggering commit that introduced these freezes on 845G into ubuntu. They can be downloaded from http://www.kvante.info/845Gfreeze/ . There is some more detailed information in the comments to bug 456902. If this bug is the same as bug 456902, the kernels with freezetest1 and freezetest8 should not cause the freeze, while the one with freezetest9 should cause a freeze. If this is not the case for you, we know that your bug is different from bug 456902 and we will have to do some additional tests.

I originally compiled the kernels on Jaunty, but tests by others seems to indicate that the tests can also be done in Karmic or Lucid. A problem with Karmic and Lucid is that the default kernel will cause a freeze while it doesn't in Jaunty.

[jerrylamos (jerrylamos)]

 Geir, I don't know enough command line linux to download the .deb files in your note. I tried using apt-get and sources.list but didn't get anywhere.

Thanks for looking at this bug.
Jerry

-----Original Message-----
From: Geir Ove Myhr <email address hidden>
To: <email address hidden>
Sent: Wed, Dec 30, 2009 2:18 pm
Subject: [Bug 420594] Re: [i845] xorg hang video driver 2.8.1-1ubuntu1

I have compiled three test kernels for finding the triggering commit
that introduced these freezes on 845G into ubuntu. They can be
downloaded from http://www.kvante.info/845Gfreeze/ . There is some more
detailed information in the comments to bug 456902. If this bug is the
same as bug 456902, the kernels with freezetest1 and freezetest8 should
not cause the freeze, while the one with freezetest9 should cause a
freeze. If this is not the case for you, we know that your bug is
different from bug 456902 and we will have to do some additional tests.

I originally compiled the kernels on Jaunty, but tests by others seems
to indicate that the tests can also be done in Karmic or Lucid. A
problem with Karmic and Lucid is that the default kernel will cause a
freeze while it doesn't in Jaunty.

--
[i845] xorg hang video driver 2.8.1-1ubuntu1
https://bugs.launchpad.net/bugs/420594
You received this bug notification because you are a direct subscriber
of the bug.

[Geir Ove Myhr (gomyhr)]

On Thu, Dec 31, 2009 at 9:26 PM, jerrylamos wrote:
>  Geir, I don't know enough command line linux to download the .deb files in your note.  I tried using apt-get and sources.list but didn't get anywhere.

No problem, I'll be a bit more verbose. Let me know if you have more questions.

My preferred way of downloading the files would be to use firefox,
right click on the file name and select save as... But if you are on a
system which is freezeing and you don't have a workaround I understand
that this is problematic. You can also download the files with the
three commands
wget http://www.kvante.info/845Gfreeze/linux-image-2.6.30-freezetest1_git03347e2~gomyhr_i386.deb
wget http://www.kvante.info/845Gfreeze/linux-image-2.6.30-freezetest8_git6fd4693~gomyhr_i386.deb
wget http://www.kvante.info/845Gfreeze/linux-image-2.6.30-freezetest9_git02200d0~gomyhr_i386.deb
this will download the files to the current directory. If this
directory does not have othe deb-files, you can install them all with
dpkg -i *.deb

There are some more detailed instructions at bug 456902, comments #17 and #20.

Hope this helps. Happy New Year!

Geir Ove

[jerrylamos (jerrylamos)]

Thanks, Geir. That seemed to work, however now none of the images will do xorg, not even the 2.6.32-9 with "vesa". Black screen on all three "freeze" images with or without "vesa".

Ctrl-Alt-F1 doesn't work either after startx, however I can ssh in from a Thinkpad T40. It has ati video so Lucid is working on it except for losing the pointer from time to time.

Since the Lucid is broken altogether, I'm doing another install of Lucid Alpha 1 on another partition now....of course, the Lucid CD boots to black screen unless I do:

sudo nano /etc/X11/xorg.conf
Section "Device"
Identifier "Configured Video Device"
Driver "vesa"
EndSection

I don't quite understand why Ubuntu development doesn't have that as a boot choice. Maybe they do, and I just don't know how to invoke it. There have been a number of complaints on the Ubuntu forums about the lack of xorg.conf.

Thanks, Jerry

[Geir Ove Myhr (gomyhr)]

It looks like you may have problem which is different from the problem
that most other people have on 845G. Perhaps you could try some older
kernels from http://kernel.ubuntu.com/~kernel-ppa/mainline/ and see if
you can find one that doesn't have the problem. A 2.6.30, 2.6.29 or
2.6.28 kernel may work, or maybe changing the kernel doesn't work at
all.

Another thing you may try is to use the kernel option nomodeset (I
find that the easiest way to edit kernel options is to edit
/etc/default/grub, add it after "quiet splash", and then run
update-grub before reboot). It may be that the problems you have are
related to KMS, which is disabled with nomodeset.

Finally, maybe you can try the freezetest kernels in Jaunty? If Jaunty
works all right out of the box, it is possible that also freezetest1
and freezetest8 will work, but not freezetest9.

[jerrylamos (jerrylamos)]

i845G is working multi booted with Karmic, Jaunty, Debian, and Slax all running normally except for Lucid. Lucid originally worked for the first few weeks, then started getting sicker and sicker. I'll try downloading some mainline kernels as Geir suggests to see if they help. The Lucid just installed and updated today boots to black screen, but will run a command line from Crl-Alt-F1. I do try "out of the box", vesa, and nomodeset in combinations to try to get going, no luck right now.

Jerry

[jerrylamos (jerrylamos)]

Success!

v2.6.32-rc1 works with nomodeset, with or without "vesa". Other than that, all the rest of the updates as of 20100101 are on.

Hey, I'm running! Not with the current linux-image, though, and not without nomodeset.

Thanks, Jerry

[Geir Ove Myhr (gomyhr)]

> i845G is working multi booted with Karmic, Jaunty, Debian, and Slax all
> running normally except for Lucid.

Okay. So this is definitely a different bug from the freezes people
are experiencing on Karmic on 845G. I'm juggling several 845G freeze
bug reports right now, so I got a bit confused. With such a nice test
setup it would be nice if you also include in bug reports what other
versions you have where the problem does not exist. This is often very
helpful to know.

> v2.6.32-rc1  works with nomodeset, with or without "vesa".  Other than
> that, all the rest of the updates as of 20100101 are on.

So v2.6.32-rc1 works with nomodeset, but not without nomodeset? And
v2.6.32-rc2 does not work at all? That narrows it down pretty well.
Did you try if any older kernel would work without the nomodeset
option?

[jerrylamos (jerrylamos)]

Geir, karmic 2.6.31-16 runs without nomodeset, i.e. KMS on. Karmic runs normally with default intel xorg and default grub.cfg. Now earlier karmics from late June thru sometime in August also would fail, but karmic RC does not have that problem.

I can try v2.6.32-rc2 etc. to see what fails on lucid. When it fails I can do ubuntu-bug xorg from the command line & save the apport file and whatever else might be useful. I don't know if apport includes Xorg.0.log and .xsession-errors.

I also did get failures with lucid after a while the screen locks up doing something like making a launchpad entry. Seems to me that's a different bug, I don't remember the bug # at the moment. karmic and jaunty don't get those on this pc. At the moment not booting at all is more important to me.

Thanks, Jerry

[Geir Ove Myhr (gomyhr)]

On Fri, Jan 1, 2010 at 4:55 PM, jerrylamos wrote:
> Geir, karmic 2.6.31-16 runs without nomodeset, i.e. KMS on.

We're trying to identify the changes that trigger the problems. Since
the released Karmic works all right, and current Lucid doesn't, we
want to identify either a upgrade in Karmic that introduces the
problems you see in Lucid, or a downgrade in Lucid that removes the
problems. It may be that the kernels do not influence this, but it is
the easiest thing to test because of the pre-compiled mainline
kernels, so I suggest we do that first.

> I can try v2.6.32-rc2 etc. to see what fails on lucid.  When it fails I
> can do ubuntu-bug xorg from the command line & save the apport file and
> whatever else might be useful.  I don't know if apport includes
> Xorg.0.log and .xsession-errors.

Yes, it would be nice if you could test v2.6.32-rc2 and newer and see
which one is the first to fail. We may then want to do a more
fine-grained search between the last rcX that works and the first rcX
that doesn't. We don't need the logs at this point. I don't think they
will be more useful than the ones you already uploaded.

> I also did get failures with lucid after a while the screen locks up
> doing something like making a launchpad entry.  Seems to me that's a
> different bug, I don't remember the bug # at the moment.  karmic and
> jaunty don't get those on this pc.  At the moment not booting at all is
> more important to me.

Sure. For a while I thought your bug was the same as the freezes
reported on Karmic, but that it manifested itself a little differently
in Lucid. Since Karmic runs fine, that cannot be the case.

[jerrylamos (jerrylamos)]

Just tried 2.6.33rc2 from ppa/mainline (see Geir's note above) and it boots O.K. with "quiet splash" most of the time. It seems to like to boot from a full shutdown/power on sequence however that may be superstition; it failed once after a reboot.

linux-image-2.6.33-020633rc2-generic_2.6.33-020633rc2_i386.deb

Let me run it for a while to see if it gets the freeze after a successful boot, which I think is another bug.

Thanks much, Jerry

[jerrylamos (jerrylamos)]

Yep, froze after about an hour or so. Tried batchbuffer dump see attachment to subsequent post.
This is 2.6.33rc2 on i845G.

Preliminary steps:

sudo apt-get install intel-gpu-tools
sudo INTEL_DEBUG=batch /etc/init.d/gdm restart

then got the hang.

I tried this:

sudo mount -t debugfs debugfs /sys/kernel/debug
datestr=$(date +%Y%m%d)
mkdir dri_debug-$datestr
sudo cp -r /sys/kernel/debug/dri/0/i915* dri_debug-$datestr
sudo intel_gpu_dump > dri_debug-$datestr/intel_gpu_dump.txt
dmesg > dri_debug-$datestr/dmesg.txt
cp /var/log/Xorg.0.log dri_debug-$datestr/
sudo cp /var/log/gdm/\:0.log dri_debug-$datestr/gdm.log
sudo tar czf dri_debug-$datestr.tgz dri_debug-$datestr/

and got:

mount: debugfs already mounted or /sys/kernel/debug busy
mount: according to mtab, none is already mounted on /sys/kernel/debug

but it did create:

jerry@linux:~$ ls -l dri*
-rw-r--r-- 1 root root 388052 2010-01-02 13:39 dri_debug-20100102.tgz

Of any use? Anything else needed for next hang?

to be attached on reboot of hung machine since that's where the tgz is.

Can't get gdm going from ssh or from Crtl-Alt-F1 will cycle power.

Jerry

[jerrylamos (jerrylamos)]

Hmmm. 2.6.33rc2 now boots to black screen after having hung.

Booting 2.6.32rc7 O.K., attached dri_debug created from instructions in

https://wiki.ubuntu.com/X/Troubleshooting/Freeze#How

Anything else?

Jerry

[jerrylamos (jerrylamos)]

2.6.33rc2 running now, after this progression:

1. 2.6.33rc2 boot to black screen after hang see previous post.
2. 2.6.32rc7 boots O.K.
3. 2.6.33rc2 boots to brown screen with top and bottom white panels, no icons.
4. 2.6.32rc7 boots O.K.
5. 2.6.33rc2 booted and ran for hours. Best results I've had with Lucid for months.
6. Boots next day O.K. I've got ssh in from another pc just in case....

Thanks all, Jerry

[jerrylamos (jerrylamos)]

Ran about 9 hours until desktop froze. It had been playing YouTube video. Ctrl-Alt-F1 worked but couldn't get the desktop to restart.

Linux version 2.6.33-020633rc2-generic (root@zinc) (gcc version 4.2.3 (Ubuntu 4.2.3-2ubuntu7)) #020633rc2 SMP Fri Dec 25 10:56:21 UTC 2009

Anything else?

Jerry

[jerrylamos (jerrylamos)]

i845G with Lucid 2.6.32-10 boots to black screen unless at grub I do down arrow, e, end, nomodeset, Crl-x.

Then runs a few minutes with Gnome, then FREEZE. ssh doesn't even work. Dead altogether. Can't do batchbuffer dump if ssh doesn't work since lucid is dead stopped altogether.

With LXDE runs tens of minutes then FREEZE. ssh doesn't even work. Dead altogether.

Culprit from /var/log/Xorg.0.log is:

 2.634776] (II) LoadModule: "evdev"
[ 2.635818] (II) Loading /usr/lib/xorg/modules/input/evdev_drv.so
[ 2.636237] (II) Module evdev: vendor="X.Org Foundation"
 compiled for 1.7.2, module version = 2.3.2
 Module class: X.Org XInput Driver
 ABI class: X.Org XInput driver, version 7.0

Namely intel driver 2.3.2. which just came in on a deadly update.

How do I back level xorg intel driver to one that works, example karmic level?

Thanks much, jerry

[Brian Rogers (brian-rogers)]

Due to an ABI change in Xorg 1.7, the older binary packages built against 1.6 won't work, so new packages have to be recompiled for 1.7. And unfortunately I wasn't able to easily compile the older driver versions on Lucid.

I didn't investigate yet how much work it'd take to work around this problem, since I'm focused on the Karmic freeze issue right now, but the better solution may in fact be to roll back the X server itself to 1.6, or reinstall Karmic for testing purposes.

When you say it boots to a black screen with KMS, do you mean even before X starts the system freezes? Have you tried leaving KMS on, but removing 'splash' from the command line? Does it hang right away, or when it starts X? And if it hangs starting X, what happens if you disable auto-login? Can you then get to the login screen?

If you can reproduce a hang even before X starts, then we can rule out the X intel driver. It'd mean it's definitely a kernel bug.

BTW, "module version = 2.3.2" is referring to the input device driver, evdev. The current intel driver version in Lucid is 2.9.1.

[jerrylamos (jerrylamos)]

Brian,

with:

nomodeset

vesa

2.3.32-10 boots and has run for several minutes!

quiet splash (I don't see a splash, just sits there blank until the login screen comes up).

Default, boots to a black screen and can't even ssh in. Ctrl-Alt-F1 won't even work.

I do have Karmic, Debian, and Slax running O.K.

I really do prefer Karmic over any of the linux's I've run over the years. I just gave a brief overview class comparing Karmic, openSUSE, fedora, slax, debianLXDE.

Thanks, Jerry

Jerry

[jerrylamos (jerrylamos)]

Progress....

A2 CD live won't boot, initial boot screen freezes no matter what is tried on the linux boot line.

A1 updated to A2 level won't run any time at all unless "nomodeset" and "vesa" are used.

A2 Alternate installed O.K., has been running! (careful! Don't update it, it's running!)

Linux version 2.6.32-10-generic (buildd@palmer) (gcc version 4.4.3 20100105 (prerelease) (Ubuntu 4.4.2-8ubuntu2) ) #14-Ubuntu SMP Thu Jan 7 17:38:40 UTC 2010
jerry@linux:~$

So on this machine the fix is the A2 alternate install.

Jerry