Bug 10773

Summary: [855GM] LVDS flickers when using intel driver (i810 is OK)
Product: xorg Reporter: Jody Goldberg <jody>
Component: Driver/intelAssignee: Hong Liu <hong.liu>
Status: RESOLVED FIXED QA Contact: Xorg Project Team <xorg-team>
Severity: major    
Priority: medium CC: deejay1, Felix.Braun, michael.fu, noway, ProT3ch, shuber, subs, unggnu, zhenyu.z.wang
Version: unspecified   
Hardware: x86 (IA32)   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Bug Depends on:    
Bug Blocks: 13027, 13493    
Attachments:
Description Flags
Output from xrandr
none
ModeDebug run with git head
none
requested output
none
fix 8xx clock p2 value
none
DeeJay1's xorg.conf
none
xrandr --verbose, ddcprobe, xorg.conf, and modedebug run
none
ProTech's xorg.conf, xorg log, xrandr --verbose, ddcprobe
none
ProTech's regdump with 1.7.4
none
log none

Description Jody Goldberg 2007-04-26 13:24:07 UTC
sometimes it goes into an infinite loop when switching to a virtual console.
And makes everything somewhat green :-)
Debian is shipping '2.0.0'  which listsjjj

    commit 31bf269afed0a830e79cbbd9d4b1ee9843af326c
    Author: Keith Packard <keithp@neko.keithp.com>
    Date:   Thu Apr 19 20:03:41 2007 -0700

as the last entry.
The config and system details are attached to
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=420307

On the plus side, the driver is finding the correct resolution, and what seems like the correct physical width, but seems to scale the visual output by a factor 2.
Comment 1 Jody Goldberg 2007-06-17 18:13:29 UTC
git head as of June 16 still presents with the same issues.
Any hints on debugging ?
Comment 2 Jody Goldberg 2007-07-03 16:35:11 UTC
git head as of July 3 now sizes things correctly.

However, there is a significant flicker that still leaves it less useful than the old driver and 915resolution.  re-titling.
Comment 3 Lukasz Jernas 2007-07-04 06:42:51 UTC
Isn't it because of wrong refresh rate for VGA outputs? I can get only 1024x768@60Hz on my configuration and the list of availible resolutions looks smaller - previously it worked fine. Output from xrandr attached
Comment 4 Lukasz Jernas 2007-07-04 06:46:20 UTC
Created attachment 10584 [details]
Output from xrandr
Comment 5 Lukasz Jernas 2007-07-20 23:54:28 UTC
Issue still persists with git from July 20. Any hints how could we help to debug it?
Comment 6 Gordon Jin 2007-08-28 01:42:31 UTC
changing status from NEEDINFO to NEW, as I'm not seeing any query to reporter.
Comment 7 Keith Packard 2007-09-06 18:43:17 UTC
An X server log with the 'modedebug' option turned on might have some clues. Also, what kind of machine is this?
Comment 8 Jody Goldberg 2007-09-06 20:04:11 UTC
Created attachment 11459 [details]
ModeDebug run with git head

based on
commit 286f5df0b62f571cbb4dbf120679d3af029b8775
Author: Keith Packard <keithp@koto.keithp.com>
Date:   Thu Sep 6 14:31:55 2007 -0700
Comment 9 Jody Goldberg 2007-09-07 04:46:18 UTC
An older fujitsu P5020d 'lifebook'
Comment 10 ProTech 2007-09-08 14:03:40 UTC
I have the same problem, with a Fujitsu Siemens Lifebook E8010.

It looks like a CRT monitor running 60Hz. I never seen anything like this on an LCD monitor. It works with the old 1.7.x version, but not with 2.1.0 and 2.1.1.

Gentoo Linux, xorg-server 1.3. It's just got stabled today.
Comment 11 Stefan Huber 2007-09-12 00:15:20 UTC
Flicker occurs on my Asus Notebook A3827-NLP with 15" SXGA+ Display too. System:
   - Gentoo-Linux with xorg-7.2, xorg-server-1.3.0.0,
     xrandr-1.2.2, and randrproto-1.2.1
   - i810-1.7.4 works, i810-2.* does not
   - Same holds for xorg-server-1.2
Comment 12 Daniel Schwen 2007-10-04 08:53:22 UTC
Flicker (60Hz) occurs on my Fujitsu Lifebook 5010d at 1280x768 with a 

00:02.0 VGA compatible controller: Intel Corporation 82852/855GM Integrated Graphics Device (rev 02)

So far neither adding and selecting new gtf modelines (85Hz) with xrandr, nor the patch by Eric Anholt allows me to select a higher refresh rate for my local flat panel. Needless to say it was flicker-free with the i810 driver plus videobios patching.
Comment 13 Hong Liu 2007-12-12 01:26:17 UTC
From the xorg.log:
1. we can't probe the EDID data, so we change to use the probe the BIOS programmed mode. It seems this mode is also not correct, the dotclock is too low, and the refresh rate is only 24KHz. 
(Modeline "1280x768"x0.0   40.54  1280 1328 1440 1688  768 771 777 806 (24.0 kHz))

2. Also we can't get a working mode from VBT table.
Modeline "54x0"x0.0  287.20  54 54 852 54  0 23 70 0 (5318.5 kHz)

So our driver can't find a good mode for the panel. Can you post the output of
1. xrandr --verbose
2. ddcprobe
Can you switch to other 1024x768 modes if there are some ?

thanks,
Hong 
Comment 14 ProTech 2007-12-16 01:40:32 UTC
Here are the requested outputs. This is with xorg-server 1.3, xrandr 1.2.2, intel driver 1.7.4. The video BIOS patched to support the 1400x1050 mode I needed. It's working fine with this combination. If you need the outputs with newer driver version please let me know, and I install the new driver again.

hu3cs41c ~ # ddcprobe 
VESA 3.0 detected.
OEM Name: Intel(r)852GM/852GME/855GM/855GME Graphics Chip Accelerated VGA BIOS
Vendor Name: Intel Corporation
Product Name: Intel(r)852GM/852GME/855GM/855GME Graphics Controller
Product Revision: Hardware Version 0.0
Memory installed = 192 * 64k blocks = 12288kb
Supported standard modes:
        1280x1024x256
        1280x1024x64k
        1280x1024x16m
        1024x768x256
        1024x768x64k
        1024x768x16m
        640x480x16m
        800x600x64k
        800x600x16m
        640x480x256
        800x600x256
        640x480x64k
EDID read failed. (No DDC-capable monitor attached?)
hu3cs41c ~ # logout
andrewjr@hu3cs41c ~ $ xrandr --verbose
Screen 0: minimum 1280 x 1024, current 1400 x 1050, maximum 1400 x 1050
default connected 1400x1050+0+0 (0x54) normal (normal left inverted right) 0mm x 0mm
        Identifier: 0x53
        Timestamp:  -498900606
        Subpixel:   unknown
        Clones:    
        CRTC:       0
        CRTCs:      0
  1400x1050 (0x54)   88.2MHz
        h: width  1400 start    0 end    0 total 1400 skew    0 clock   63.0KHz
        v: height 1050 start    0 end    0 total 1050           clock   60.0Hz
  1280x1024 (0x55)   78.6MHz
        h: width  1280 start    0 end    0 total 1280 skew    0 clock   61.4KHz
        v: height 1024 start    0 end    0 total 1024           clock   60.0Hz
Comment 15 Jody Goldberg 2007-12-16 16:10:04 UTC
Created attachment 13150 [details]
requested output

Using the 2.2 release

commit 177924e879564b7e9e70fd607141978bfd053fff
Author: Jesse Barnes <jesse.barnes@intel.com>
Date:   Thu Nov 8 18:16:48 2007 -0800
Comment 16 Hong Liu 2008-01-21 21:15:41 UTC
Created attachment 13849 [details] [review]
fix 8xx clock p2 value
Comment 17 Hong Liu 2008-01-21 21:23:29 UTC
sorry for the long latency...

Would you please try the attached patch to see if anything changed?

For those who have i810-1.7.4 working on your laptops, would you please try to dump the registers when your laptop is working? You can use the intel_reg_dump tool which is in the new driver package (xf86-video-intel/src/reg_dump).

With this kind of info, we may compare the difference between intel-driver and vbios when programming the display registers.

Thanks,
Hong
Comment 18 Lukasz Jernas 2008-01-22 08:16:01 UTC
Everything looks fine for me after removing the monitor sections from xorg.conf (1.4.0.90 + intel 2.2-branch)
Comment 19 Michael Fu 2008-01-22 18:10:38 UTC
Lukasz Jernas,

could you post your xorg.conf file here? Did you apply Hong's patch?

Jody, 

any update from you?
Comment 20 Lukasz Jernas 2008-01-23 08:15:56 UTC
Created attachment 13878 [details]
DeeJay1's xorg.conf

It looks ugly (most of it can and should be stripped away) but works for me.
Comment 21 Lukasz Jernas 2008-01-23 09:04:01 UTC
It was without Hong's patch.
Comment 22 Jody Goldberg 2008-01-23 17:17:00 UTC
Created attachment 13891 [details]
xrandr --verbose, ddcprobe,  xorg.conf, and modedebug run

Unchanged with patch applied to git HEAD.

commit db0a7c569e383436a2725e1e74f35fb426da1196
Author: Andreas Stawinoga <a.stawinoga@gmx.de>
Date:   Thu Jan 24 08:51:09 2008 +0800
Comment 23 Hong Liu 2008-01-23 23:18:08 UTC
(In reply to comment #22)
> Created an attachment (id=13891) [details]
> xrandr --verbose, ddcprobe,  xorg.conf, and modedebug run
> 
> Unchanged with patch applied to git HEAD.
> 
> commit db0a7c569e383436a2725e1e74f35fb426da1196
> Author: Andreas Stawinoga <a.stawinoga@gmx.de>
> Date:   Thu Jan 24 08:51:09 2008 +0800
> 

It's weird that your log doesn't change with patch applied.

(II) intel(0): DPLL_B: 0x90046000 (enabled, non-dvo, spread spectrum clock, LVDS mode, p1 = 3, p2 = 14)
              p2 should be 7 after patch applied

(II) intel(0): pipe A dot 50307 n 2 m1 18 m2 7 p1 13 p2 4
(II) intel(0): pipe B dot 122857 n 3 m1 22 m2 7 p1 3 p2 7
               the dot is wrong because we are using the 96M/100M ref clock which only available on i9XX platform. After patch, we should use 48M/66M ref clock.

(II) intel(0): Modeline "1280x768"x0.0   40.54  1280 1328 1440 1688  768 771 777 806 (24.0 kHz)
 the dotclock should be changed to 80.09MHz.

Would you please recheck the patch is applied, thanks.

Can you try the following mode by xrandr to see if it works?
Modeline "1280x768"x0.0   80.09  1280 1328 1440 1688  768 771 777 806

Thanks,
Hong
Comment 24 ProTech 2008-01-25 08:07:09 UTC
Created attachment 13948 [details]
ProTech's xorg.conf, xorg log, xrandr --verbose, ddcprobe

I used the following Gentoo packages:

xorg-server-1.3.0.0
x11-drivers/xf86-video-i810-2.2.1_pre20080123 + patch from this bug

The display is still flickering.
Comment 25 ProTech 2008-01-25 08:24:31 UTC
I tried to compile this reg_dumper program, but it failed. All I did was:
# ./autogen.sh
# cd src/reg_dumper/
# make

main.c:32:23: error: pciaccess.h: No such file or directory

Is this because I'm running xorg-server 1.3 or I made a mistake somewhere?
Comment 26 Julien Cristau 2008-01-25 13:21:16 UTC
(In reply to comment #25)
> I tried to compile this reg_dumper program, but it failed. All I did was:
> # ./autogen.sh
> # cd src/reg_dumper/
> # make
> 
> main.c:32:23: error: pciaccess.h: No such file or directory
> 
> Is this because I'm running xorg-server 1.3 or I made a mistake somewhere?
> 
reg_dumper needs libpciaccess. You can get it from
git://anongit.freedesktop.org/git/xorg/lib/libpciaccess
Comment 27 ProTech 2008-01-26 00:10:09 UTC
Created attachment 13952 [details]
ProTech's regdump with 1.7.4

After installing libpciaccess I was able to compile the reg_dumper utility.

Here is the output with the 1.7.4 driver, which is working correctly for me.
Comment 28 Michael Fu 2008-01-26 18:33:50 UTC
(In reply to comment #23)
> (In reply to comment #22)

> Would you please recheck the patch is applied, thanks.
> 
> Can you try the following mode by xrandr to see if it works?
> Modeline "1280x768"x0.0   80.09  1280 1328 1440 1688  768 771 777 806
> 
> Thanks,
> Hong
> 

Jody, Need your response...

Comment 29 Hong Liu 2008-01-28 23:14:59 UTC
(In reply to comment #27)
> Created an attachment (id=13952) [details]
> ProTech's regdump with 1.7.4
> 
> After installing libpciaccess I was able to compile the reg_dumper utility.
> 
> Here is the output with the 1.7.4 driver, which is working correctly for me.
> 

Hi, ProTech
Thanks for providing the log, and would you please try the following mode to see if it works?
Modeline "1400x1050_test"   106.86  1400 1416 1528 1688  1050 1051 1054 1066

Thanks,
Hong
Comment 30 ProTech 2008-01-29 04:28:39 UTC
I made a mistake last time I checked the patch. It's working now with or without setting the Modeline.

The mistake: I used the Gentoo ebuild unpack method to extract the source, then applied the patch manually and then executed the ebuild install, which installed it under /var/tmp/portage/something. I missed the last step ebuild merge, which installs it to the real system. I never used these ebuild commands before...

Thanks for the fix.
Comment 31 Jody Goldberg 2008-01-30 16:55:45 UTC
Created attachment 14044 [details]
log

My apologies.  I had recompiled and installed, but not checked to see if the new module was actually being used.  The changes appear to fix things for me.
Comment 32 Hong Liu 2008-01-31 22:39:18 UTC
*** Bug 11917 has been marked as a duplicate of this bug. ***
Comment 33 Hong Liu 2008-02-04 17:53:52 UTC
Patch is commited.
Comment 34 Waldgeist_dI 2008-02-23 05:49:57 UTC
(In reply to comment #33)
> Patch is commited.
> 

I am using a Fujitsu-Siemens Lifebook E4010 with an Intel 855GM chip. According to the folks at ubuntu the current Alpha 5 of Hardy Heron includes the patch/fix but it's not working for me (see also Incorrect refresh rate probed from bad BIOS tables in 855GM; https://bugs.launchpad.net/bugs/157213). I do get the right resolution (1400x1050) but still with the annoying flicker. The x.org.log reports driver version 2:2.2.0.90-2ubuntu7 (hardy) (http://launchpadlibrarian.net/12151169/Xorg.0.log) which is supposed to solve the problem.

I'm stuck. What else information do you need?
Comment 35 Felix Braun 2008-02-24 00:22:13 UTC
I looked at the sources distributed with ubuntu hardy alpha xserver-xorg-2.2.90-2ubuntu7 and they seem to include some version of the patch albeit not the exact same version attached to this bug. And I can confirm that the issue has regressed for me too. Moreover, the patch does no longer apply :-( 
Comment 36 Störm Poorun 2008-02-24 04:08:17 UTC
I can confirm Comment #34: 

I am using an acer C111Tci (Acer C100 series) running Kubuntu 8.04 (Hardy Heron) Alpha 5.

The graphics processor is 855GM, resolution 1024x768, @ 60Hz

I have to fall back on the i810 driver, which works with a previous (fully configured0 xorg.conf, but lacks the auto-configure functionality, and does not deal well with external screens/monitors.

The refresh rate problem with the Intel driver has persisted for me, even with the latest release I have installed - xserver-xorg-2.2.90-2ubuntu7. 

The resolution has been set correctly by the driver, but the refresh rate is still wrong - 130Hz! I am still suffereing flickering and an unusable display as a result. (I cannot change the refresh rate with randr.)


Comment 37 Hong Liu 2008-02-24 17:25:46 UTC
Would you please try the latest git version of our driver?

We've made a mistake when committing the patch. And the fix has been committed (commit id cdc5cb7c3d24f36a30fb931b43ec6df8294709e2), I am not sure whether your ubuntu package (xserver-xorg-2.2.90-2ubuntu7) has the this patch.

Thanks,
Hong
Comment 38 Waldgeist_dI 2008-02-24 21:44:13 UTC
(In reply to comment #37)
> Would you please try the latest git version of our driver?
>
> ...
> 
> Thanks,
> Hong

I'd like to give it a try, but ...

Is there a manual for creating this driver from the sources? I just stumbled through the git manual but wasn't able to find a hint to this task. I am new to this.
Comment 39 Gordon Jin 2008-02-24 21:59:38 UTC
(In reply to comment #38)
> Is there a manual for creating this driver from the sources? 
http://www.intellinuxgraphics.org/install.html
Comment 40 Waldgeist_dI 2008-02-27 02:10:05 UTC
Today's upgrade of Hardy Heron did the trick! I am very happy with a flicker-free 855GM-based lifebook :D! Thank you very much!!!

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.