Bug 2603 - [Intel/i855] i810 periodically crashes system hard
Summary: [Intel/i855] i810 periodically crashes system hard
Status: CLOSED NOTOURBUG
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/intel (show other bugs)
Version: 6.8.2
Hardware: x86 (IA32) Linux (All)
: high major
Assignee: Alan Hourihane
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2005-02-23 10:01 UTC by Owen Williams
Modified: 2005-03-09 15:48 UTC (History)
0 users

See Also:
i915 platform:
i915 features:


Attachments
Xorg.0.log (62.55 KB, text/plain)
2005-02-23 10:02 UTC, Owen Williams
no flags Details
my xorg.conf (6.14 KB, text/plain)
2005-02-23 10:03 UTC, Owen Williams
no flags Details
Small screenshot of crashed computer (20.79 KB, image/jpeg)
2005-02-23 10:04 UTC, Owen Williams
no flags Details

Description Owen Williams 2005-02-23 10:01:35 UTC
Hardware: alienware sentia 223 with intel extreme graphics 2.  My screen is
1280x800 at 24bpp.  I have had this problem with stock fedora core 3 xorg
(updated all the way) as well as rawhide 6.8.2.

Symptoms:
periodically the system will suddenly crash hard.  At the time of crashing the
screen image corrupts with vertical dancing lines and other flashing dots and
artifacts.  There is no warning when the crash will happen. I will be working,
and suddenly the image will go corrupt.  At this point the entire system is
non-responsive.  There are no kernel-panic blinking lights.

When:
This seems to occur under heavy cpu / network load, but it's not really
predictable.  It is almost guaranteed to happen if I leave the machine on
overnight.  

I will attach an xorg.log and xorg.conf file, as well as an image of what the
crash looks like.  I also managed to capture a video, here:
http://ywwg.com/videos/p2190009.mov (4.8M, works in mplayer).
Comment 1 Owen Williams 2005-02-23 10:02:35 UTC
Created attachment 1961 [details]
Xorg.0.log

My Xorg.0.log
Comment 2 Owen Williams 2005-02-23 10:03:33 UTC
Created attachment 1962 [details]
my xorg.conf

There are a lot of setups, but it crashes with any one of them, whether on just
the internal screen or with xinerama.
Comment 3 Owen Williams 2005-02-23 10:04:53 UTC
Created attachment 1963 [details]
Small screenshot of crashed computer
Comment 4 Alan Hourihane 2005-03-01 09:58:48 UTC
Can you try the driver at

http://www.fairlite.demon.co.uk/intel.html
Comment 5 Alan Hourihane 2005-03-01 09:59:34 UTC
Assigning to me.
Comment 6 Owen Williams 2005-03-01 18:03:58 UTC
I tried that driver, and first indications were very promising, but my machine
has crashed twice.  However, both crashes were after the system had been
suspended and resumed with swsusp.  Since that's known to do wacky things, take
it with a grain of salt.  If the machine crashes again from a clean boot with
that driver, I'll post to that effect.

(btw, where did this driver come from?  All of the other visual artifacts wrt
XVideo are gone!  yay!)
Comment 7 Owen Williams 2005-03-02 04:40:36 UTC
Crashed twice yesterday on clean boot with new driver.  The only pattern is that
it seems to happen more often if Azureus is running.
Comment 8 Alan Hourihane 2005-03-02 04:43:49 UTC
Is it the same crash as shown in the picture (i.e. with lines running down the
screen ??)
Comment 9 Owen Williams 2005-03-02 05:58:52 UTC
It's always the same.  The picture smears a bit horizontally and vertical and
often also horizontal flickering lines dance around.  During one of the crashes
I was watching a video in totem, and larger rectangular sections than normal
were flashing colors in the xvideo area.
Comment 10 Alan Hourihane 2005-03-02 06:50:49 UTC
Does it crash at all when not running 3D or Xvideo applications ?
Comment 11 Alan Hourihane 2005-03-02 06:53:56 UTC
I've just uploaded a new driver v1.5.19 - try that.
Comment 12 Owen Williams 2005-03-03 05:46:40 UTC
So far so good, but it can go days between crashes.

It can crash at any time, including when the lid is down and the screen is
locked (blank screensaver).  The occasion last week when it crashed while
playing a video was the first time it had done that.
Comment 13 Owen Williams 2005-03-04 04:53:55 UTC
Crashed last night, after nearly 14 hours of uptime.
Comment 14 Alan Hourihane 2005-03-04 05:00:40 UTC
Presumably it was using xscreensaver during last night ??

Do you use 3D regularly when these crashses occur ?
Comment 15 Alan Hourihane 2005-03-04 05:01:33 UTC
Oh, and can you ssh in from the network when the crash occurs ??

If so, can you check the output of 'dmesg' when it next occurs.
Comment 16 Owen Williams 2005-03-04 08:30:33 UTC
Yes, xscreensaver was on.  I never use 3D.  

It just crashed 10 minutes ago while k3b was creating an image to burn to a DVD,
and I wasn't even at the machine.  It had been up for four whole minutes.  I
could tell by looking at the screen that the processor was doing basically
nothing, and the network was doing nothing.  The fans were low.  I looked at the
screen carefully, and it appeared that the pixels had spread horizontally a bit
by primary color, so you could see separate red, green, and blue.  Also the left
side of the screen was mostly static (a little flickering here and there) but
the right two or three inches of the screen was flashing more energetically. 
This time there were none of the vertical lines I usually see.

The machine is never reachable through the network when it crashes.  The machine
just stops cold.

Could this be a bad hardware issue?  I sent the laptop off to Alienware service
once (for another issue) and they checked it out and claimed it was ok, but I
wonder about it.  I hate this computer.
Comment 17 Owen Williams 2005-03-04 08:31:39 UTC
to clarify, I meant the left half of the screen was "static" as in "not moving."
Comment 18 Alan Hourihane 2005-03-04 08:39:43 UTC
Could be. Usually X can detect that the graphics chip has hung and cleanly
aborts the Xserver. This is obviously a severe crash issue which could be
attributed to hardware. I'd suspect your RAM first of all. And is it spec'ed
correctly for the machine ?

How much memory is installed, and in how many sticks ?

If possible, try reducing the RAM, and then swapping the sticks around to see if
it re-occurs. But check the speed of it too. I know some systems, although they
have 3 sockets, they can only use 2 of them when using certain higher speeds of
memory. 
Comment 19 Owen Williams 2005-03-08 04:44:33 UTC
memtest86 didn't find anything wrong even after 2 cycles.  However, I tried
swapping the RAM modules, and four days later I haven't had a crash.  At this
point, even if it does crash again, I would say that it's not the fault of the
driver and is more likely due to bad hardware.  I would suggest this bug be
closed and marked PEBKAC.
Comment 20 Alan Hourihane 2005-03-10 10:48:33 UTC
Marking NOTOURBUG as it looks like a hardware issue.


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.