[Gutsy, Hardy] Black screen with mouse pointer on i830, intel driver (PipeA quirk needed)

[Shaun Crampton]

On request, moved this bug report from bug 123283. Which covers very similar symptoms on ATI hardware.

Binary package hint: xserver-xorg-video-intel

Here's what happened on my laptop in Hardy:
 - Xubuntu Hardy beta live CD
 - Insert CD, boot to desktop
 - wait several minutes with no mouse activity
 - screen goes completely blank
 - move mouse
 - screen came back
 - wait few minutes
 - screen completely blank. While the screen is completely black, I can see feint detail on the screen (like the back-light is off but the screen still has stuff on it).
 - move mouse. When I move the mouse the back-light comes on but the screen is black apart from the mouse pointer.
 - mouse pointer appeared but screen still blank, mouse responds to movement and cursor changes when hovering over a text field for example, computer appears to respond to input (clicking etc.)
 - VT switch to terminal 1
 - VT switch to terminal 7
 - screen came back to normal.

In Gutsy, I cannot VT switch but if I ssh into the box and temporarily start a new X server and then kill it, that restores the original screen (possibly because it causes a VT switch).

[Bryce Harrington (bryce)]

Weird. But I've seen where doing a VT switch can bring a system back to sanity.

Thanks for attaching your Xorg.0.log. It looks like your hardware is:

Intel Corporation 82830 CGC [Chipset Graphics Controller] rev 3 [chip 8086,3577 card 1179,ff00]

No other obvious errors jump out though. But the 830 is a pretty old chipset. We have seen a lot of issues with i8xx chips when using the new -intel driver, and have worked with Intel to fix a lot of them, particularly i855; it wouldn't surprise me that there is additional work to be done for the old i830.

Fwiw, we're still including the "i810" driver in Hardy since we know some old chips like yours work better with it than with intel; however we plan to deprecate i810 in Intrepid so would like to help you solve the issue so you can use "intel" going forward.

The first step is to test some newer versions of "intel". I've built a handful of git snapshots from upstream for you to try out:

http://people.ubuntu.com/~bryce/bisect/

Try the newest version at the top of the page, and maybe one or two other picked at random. If you find it does not solve it, then the next step would be to file this bug upstream.

[Shaun Crampton (fasaxc)]

I have a Toshiba 3000-X4 laptop with i830 graphics card. If I leave my laptop for 10 minutes with no mouse activity then the screen goes blank. If I then move the mouse, the mouse pointer appears and responds to movement but the rest of the screen remains blank. If I VT switch (or ssh in and temporarily start a second X server and then kill it), the desktop is restored.

I have this problem in Ubuntu Gutsy and Hardy Beta as well as recent Git snapshots provided by the Ubuntu maintainers for me to test. The Ubuntu bug report is referenced in the URL field.

Specifically, I've tried git revisions r79a5585d and rac763634 with no luck as well as the current Hardy version and "2:2.1.1-0ubuntu9" (and previous versions) from Ubuntu Gutsy (I can't try versions in between due to another bug).

[Shaun Crampton (fasaxc)]

On the Hardy beta live CD:

First attempt with rac763634:
 - Installed the deb via dpkg -i, dpkg claimed it was a downgrade from 2.2.1-1ubuntu5 to 2.2.0+git.... (may just be the naming convention for pre-releases)
 - Logged out of the live CD and zapped X
 - X restarted but gave a blank screen
 - checked the X log from a VT, it was using the VESA driver rather than intel
 - added Driver "intel" to xorg.conf
 - restarted X
 - saw the mouse pointer for a couple of seconds then it went blank and I had to reboot (couldn't VT switch SysRq-S and SysRq-U seemed to work but SysRq-B hung with garbage onscreen)
I've had problems before where once the VESA driver has loaded, loading another driver crashes X.

Second attempt with rac763634:
 - installed
 - edited xorg.conf to add Driver "intel"
 - restarted X
 - 12:45 got desktop as expected
 - 12:59 noticed blank screen
 - moved mouse, got mouse pointer over black screen
 - VT switched, screen returned to normal.

[Shaun Crampton (fasaxc)]

With r79a5585d:
 - installed (maintained Driver "intel" in xorg.conf), zapped, waited
 - 14:10 got desktop
 - 14:21 noticed blank screen
 - moved mouse, only mouse pointer visible
 - VT switched, desktop restored
Then:
 - removed Driver "intel" from xorg.conf
 - got VESA driver with blank screen.

[Shaun Crampton (fasaxc)]

Filed upstream bug report at https://bugs.freedesktop.org/show_bug.cgi?id=15268.

[Gordon Jin (gordon-jin)]

Please attach your xorg.conf and Xorg.0.log here, prefereably with more info according to http://www.intellinuxgraphics.org/how_to_report_bug.html

Does this also happen with vesa driver?

[Shaun Crampton (fasaxc)]

Created an attachment (id=15607)
My Xorg.0.log covering the time of the issue.

[Shaun Crampton (fasaxc)]

Created an attachment (id=15608)
My xorg.conf (unmodified Ubuntu Hardy beta)

[Shaun Crampton (fasaxc)]

System Environment:

lspci -nn:
00:00.0 Host bridge [0600]: Intel Corporation 82830 830 Chipset Host Bridge [8086:3575] (rev 03)
00:02.0 VGA compatible controller [0300]: Intel Corporation 82830 CGC [Chipset Graphics Controller] [8086:3577] (rev 03)
00:02.1 Display controller [0380]: Intel Corporation 82830 CGC [Chipset Graphics Controller] [8086:3577]

uname -m:
i686

Package versions:
Long standing problem experienced in Ubuntu Gutsy and Hardy and git revisions detailed in previous post.

uname -r:
2.6.24-12-generic

Machine model:
Toshiba Satellite 3000-X4

Reproduction steps:
In previous post. 100% reproducible.

VESA Driver:
VESA driver (and i810) is broken on this machine (black screen). I haven't got around to filing those bug reports yet.

[Shaun Crampton (fasaxc)]

I thinks that's everything that applies from the page you referenced. Is there any more info that I can supply?

[Gordon Jin (gordon-jin)]

Jesse/Eric/Keith, are you able to reproduce this on 830m?

[Shaun Crampton (fasaxc)]

This is still a problem in Ubuntu Hardy Final with the following version:

$apt-cache policy xserver-xorg-video-intel
xserver-xorg-video-intel:
  Installed: 2:2.2.1-1ubuntu12
  Candidate: 2:2.2.1-1ubuntu12
  Version table:
 *** 2:2.2.1-1ubuntu12 0
        500 http://archive.ubuntu.com hardy/main Packages
        100 /var/lib/dpkg/status

[Jesse Barnes (jbarnes-virtuousgeek)]

I don't have an 830 to reproduce this problem with. But I'll attach a patch I've been working on to fix some mode register programming problems...

[Jesse Barnes (jbarnes-virtuousgeek)]

Created an attachment (id=16417)
Fixup ordering and wait for vblank

This patch *may* help... Can you also reproduce your problem using 'xset dpms force off' or any other way?

[Shaun Crampton (fasaxc)]

I tried "xset dpms force off" but it did not repro. the bug. Moving the mouse brought the screen back on. I'll see if I can work out how to build a patched driver...

[Shaun Crampton (fasaxc)]

Bryce, upstream have a patch that they claim may help. Please will you build a deb including that patch?

I've had a crack at building it myself but I haven't succeeded. I tried to follow the instructions at http://wiki.debian.org/XTips but the merge from the Debian repo to the upstream repo failed in a lot of places and I don't have the experience with this codebase to do a proper merge. I'd prefer to build a proper deb rather than just compile so that I don't break the machine I'm trying to sort out.

[Jesse Barnes (jbarnes-virtuousgeek)]

Shaun, any update? If you need help with building, you could try contacting Ubuntu's X maintainer, Bryce Harrington, I think he's been working on making it easier to build & distribute patched driver packages.

[Shaun Crampton (fasaxc)]

Jesse, sorry about that. I think I missed your follow up in my email. I tried to build your patch against the Ubuntu tree but the codebases were too divergent (and I don't have enough experience of the code to manually merge). I've asked Bryce if he'd mind building the patch for me -- I'm just waiting on his response.

[Shaun Crampton (fasaxc)]

I tried using setterm to prevent any kind of console blanking but the screen still blanks after 10 mins. The command I used (after stopping X and switching to VT1) was the following:

$setterm -blank 0 -powersave off -powerdown 0 > /dev/console

[Bryce Harrington (bryce)]

I'm kind of tied up with 8.04.1 work at the moment, might be a few days. Doesn't look like the patch builds trivially against hardy's -intel.

[Shaun Crampton (fasaxc)]

Bryce says it'll be a few days before he has time to build the patch against Hardy's driver. I'll keep you posted.

[Shaun Crampton (fasaxc)]

Thanks for the follow-up, let me know when you have something for me to try.

[Jesse Barnes (jbarnes-virtuousgeek)]

you could also try the "ForceEnablePipeA" option, since this bug seems similar to some others where that works around the problem.

[Shaun Crampton (fasaxc)]

Jesse, that workaround seems to have done the trick. Thanks a lot.

[Shaun Crampton (fasaxc)]

As suggested on the freedesktop bugzilla, I added

Option "ForceEnablePipeA" "true"

to the device section of xorg.conf and that seems to have worked around the problem.

[Jesse Barnes (jbarnes-virtuousgeek)]

Can you attach a log file with the 'modedebug' option enabled? Looks like LVDS is getting routed to pipe A, which would explain why the ForceEnablePipeA option is working for you...

[Shaun Crampton (fasaxc)]

Created an attachment (id=17467)
X.org.log with modedebug enabled

[Shaun Crampton (fasaxc)]

I've attached the X log with modedebug enabled.

I just noticed that when I turn on ForceEnablePipeA I can no longer use the VGA output from the laptop (although it works around the flat panel issue recorded here). I simply get no output from the VGA port to my projector. Is this expected as a consequence of the ForceEnable option or should I file another bug?

[Bryce Harrington (bryce)]

Aha, then it sounds like we could potentially put in a Pipe-A quirk to force it to work this way. However, the fact that this makes VGA out not work is somewhat troubling in terms of SRUing it to Hardy.

Can you confirm that the issue doesn't exist for intrepid?

[Shaun Crampton (fasaxc)]

Sorry, can't seem to get the laptop to boot from CD at the moment (think the drive is faulty). I'll see if I can work out a way to boot it from disk (don't want to upgrade just yet since intrepid is still alpha).

[Jesse Barnes (jbarnes-virtuousgeek)]

No, that's not expected, so it would be another bug. Some of our code must have some "pipe b" assumptions that don't apply in your case since LVDS is attached by DVO.

It almost sounds like your BIOS is doing the screen blanking for you, maybe there's an option to disable it?

Gordon, do we have any i830 machines in QA? We could probably reproduce this problem by hacking the driver to route things like the register dumps on Shaun's machine indicate; it would probably be easy to fix after doing that, but I don't have an i830...

[Gordon Jin (gordon-jin)]

(In reply to comment #19)
> Gordon, do we have any i830 machines in QA?
No. We've never had i830 in Shanghai.

[Elod VALKAI (elod)]

I've upgraded to Intrepid 2 days ago. I'm seeing the same symptoms (black screen with "x" cursor). I can log in (through the Display Manager), and the mouse pointer changes to normal, but the black screen remains...

Pipe-A does help, but I've got massive screen corruption after a while. Tested with and without compiz, it's all the same. It's something with this new intel driver.

[Shaun Crampton (fasaxc)]

I've been runnig with the Pipe-A workaround for a long time. Just had to reinstall my laptop and the bug seems to have been fixed (I'm no longer running with the Pipe-A workaround but it doesn't repro). Thanks!

xserver-xorg-video-intel:
  Installed: 2:2.2.1-1ubuntu13.7
  Candidate: 2:2.2.1-1ubuntu13.7
  Version table:
 *** 2:2.2.1-1ubuntu13.7 0
        500 http://ubuntu.virginmedia.com hardy-proposed/main Packages
        100 /var/lib/dpkg/status
     2:2.2.1-1ubuntu13.6 0
        500 http://ubuntu.virginmedia.com hardy-updates/main Packages
     2:2.2.1-1ubuntu12 0
        500 http://ubuntu.virginmedia.com hardy/main Packages

[Shaun Crampton (fasaxc)]

Unfortunately, the CDROM and HD in my laptop are now both broken.

However, I've wired it up as an LTSP thin client and I can't repro the bug so I think it is fixed. Until I reinstalled as a thin client, I was running with the Pipe A workaround so I don't know exactly when the fix arrived.

FYI, I'm still running Hardy on my LTSP server.

xserver-xorg-video-intel:
  Installed: 2:2.2.1-1ubuntu13.7
  Candidate: 2:2.2.1-1ubuntu13.7
  Version table:
 *** 2:2.2.1-1ubuntu13.7 0
        500 http://ubuntu.virginmedia.com hardy-proposed/main Packages
        100 /var/lib/dpkg/status
     2:2.2.1-1ubuntu13.6 0
        500 http://ubuntu.virginmedia.com hardy-updates/main Packages
     2:2.2.1-1ubuntu12 0
        500 http://ubuntu.virginmedia.com hardy/main Packages

[Jesse Barnes (jbarnes-virtuousgeek)]

Ok, guess we can close this then, thanks for the update!

[Elod VALKAI (elod)]

This upgrade came through today: xserver-xorg-video-intel, version 2:2.4.1-1ubuntu6.

It solves the Pipe-A problem (setting not required in xorg anymore).

I've managed to solve screen corruption with the folowing options (found in some earlier bugreports on this issue):
 Option "AccelMethod" "exa"
 Option "MigrationHeuristic" "greedy"
 Option "ExaNoComposite" "true"

[riny (rinyfeller)]

I have installed Intrepid yesterday. Also got a black screen after login. My xorg.conf was almost empty. Is that normal? I installed the daily build.

My xorg.conf:
Section "Device"
 Identifier "Configured Video Device"
 Driver "Vesa"

EndSection

Section "Monitor"
 Identifier "Configured Monitor"
EndSection

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

EndSection

It seems that I have a Intel 830M Chipset
I Have a onboard Intel video on a Intel board.

The Driver "Vesa" I did put in myself. Now I can use the system without problems. Only Compiz doesn't work anymore.

On older versions of Ubuntu the video worked perfectly, with Compiz enabled.

[Guy Stone (stoneguy3)]

My comments and attachmments for bug 277699 really belong here. I didn't notice Markus' original was for an ATI board. Mine's for an integrated 865G

[Guy Stone (stoneguy3)]

Fedora 9 Live seems to be somewhat similar to Intrepid in the X department. And it boots OK on that machine. There's no xorg.conf anywhere I can find (huh?), but the attached logfile will show what combination of things works. Hope it's helpful. Can't use Intrepid on this box as things stand.

[Steve Langasek (vorlon)]

Based on comments from the original submitter, I'm marking this as fixed for intrepid. If you have anything other than the chipset of the original submitter
  Intel Corporation 82830 CGC [Chipset Graphics Controller] rev 3 [chip 8086,3577 card 1179,ff00]
then please file a separate bug report and provide an Xorg.log file showing the problem on intrepid.

[Rolf Leggewie (r0lf)]

Hardy has seen the end of its life and is no longer receiving any updates. Marking the Hardy task for this ticket as "Won't Fix".